Fundamentos y Aplicaciones de Micro y Nanofibras

Uploaded by:
Ebooks Chemical Engineering

https://www.facebook.com/pages/Ebooks-Chemical-Engineering/238197077030
For More Books, softwares & tutorials Related to Chemical Engineering

Join Us
@facebook: https://www.facebook.com/pages/Ebooks-ChemicalEngineering/238197077030
@facebook: https://www.facebook.com/AllAboutChemcalEngineering
@facebook: https://www.facebook.com/groups/10436265147/
ADMIN:
I.W
<< If you like this Book, than support the author and BuY it >>
C:/ITOOLS/WMS/CUP-NEW/4597349/WORKINGFOLDER/YARA/9781107060296HTL.3D
i [1--2] 10.12.2013 5:29PM
Fundamentals and Applications of Micro- and Nanobers

A comprehensive exposition of micro- and nanober forming, this text provides a unied
framework of all these processes (melt- and solution blowing, electrospinning, etc.)
and describes their foundations, development and applications. It provides an up-to-date,
in-depth physical and mathematical treatment, and discusses a wide variety of applications
in different elds, including nonwovens, energy, healthcare and the military. It further
highlights the challenges and outstanding issues from the perspective of an interdisciplinary
basic science and technology, incorporating both fundamentals and applications.
Ideal for researchers, engineers and graduate students interested in formation of microand nanobers and their use in functional smart materials.
Alexander L. Yarin is a Professor of Mechanical Engineering at the University of Illinois
at Chicago and concurrently a Professor of the College of Engineering at Korea
University in Seoul, South Korea.
Behnam Pourdeyhimi is a Distinguished Chaired Professor of Materials in the College of
Textiles and the Executive Director of the Nonwovens Institute.
Seeram Ramakrishna is a Professor of Materials Engineering and Director of the Center
for Nanobers and Nanotechnology at the National University of Singapore.
C:/ITOOLS/WMS/CUP-NEW/4597349/WORKINGFOLDER/YARA/9781107060296HTL.3D
ii [1--2] 10.12.2013 5:29PM
C:/ITOOLS/WMS/CUP-NEW/4597349/WORKINGFOLDER/YARA/9781107060296TTL.3D
iii [3--3] 10.12.2013 5:33PM
Fundamentals and Applications

of Micro- and Nanobers
ALEXANDER L. YARIN
University of Illinois, Chicago
BEHNAM POURDEYHIMI
North Carolina State University
SEERAM RAMAKRISHNA
National University of Singapore
C:/ITOOLS/WMS/CUP-NEW/4597349/WORKINGFOLDER/YARA/9781107060296IMP.3D
iv
[4--4] 10.12.2013 5:36PM
University Printing House, Cambridge CB2 8BS, United Kingdom

Published in the United States of America by Cambridge University Press, New York
Cambridge University Press is part of the University of Cambridge.
It furthers the Universitys mission by disseminating knowledge in the pursuit of
education, learning and research at the highest international levels of excellence.
www.cambridge.org
Information on this title: www.cambridge.org/9781107060296
Alexander Yarin, Benham Pourdeyhimi and Seeram Ramakrishna 2014
This publication is in copyright. Subject to statutory exception
and to the provisions of relevant collective licensing agreements,
no reproduction of any part may take place without the written
permission of Cambridge University Press.
First published 2014
Printed in the United Kingdom by MPG Printgroup Ltd, Cambridge
A catalogue record for this publication is available from the British Library
Library of Congress Cataloguing in Publication data
ISBN 978-1-107-06029-6 Hardback
Additional resources for this publication at www.cambridge.org/9781107060296
Cambridge University Press has no responsibility for the persistence or accuracy of
URLs for external or third-party internet websites referred to in this publication,
and does not guarantee that any content on such websites is, or will remain,
accurate or appropriate.
C:/ITOOLS/WMS/CUP-NEW/4601975/WORKINGFOLDER/YARA/9781107060296TOC.3D
v [5--8] 11.12.2013 3:13PM
Contents
Preface
1
Introduction
1.1
1.2
1.3
1.4
1.5
1.6
Polymer physics and rheology

2.1
2.2
2.3
2.4
2.5
2.6
2.7
Polymer structure, macromolecular chains, Kuhn segment,

persistence length
Elongational and shear rheometry
Rheological constitutive equations
Micromechanics of polymer solutions and melts
Solidication
Crystallization
References
General quasi-one-dimensional equations of dynamics of free liquid jets,

capillary and bending instability
3.1
3.2
3.3
3.4
3.5
3.6
History and outlook

Melt spinning
Dry spinning
Wet or solvent spinning, gel spinning
Spunbonding
References
Mass, momentum and moment-of-momentum balance equations

Closure relations
Capillary instability of free liquid jets
Bending perturbations of Newtonian liquid jets moving in air with high
speed
Buckling of liquid jets impinging on a wall
References
page ix
1
1
4
17
18
18
23
25
25
25
35
45
48
50
59
63
63
65
67
78
83
85
Melt- and solution blowing
89
4.1
4.2
90
94
Meltblowing process
Turbulence of surrounding gas jet
vi
4.6
4.7
4.8
4.9
4.10
Bending and apping of exible solid threadlines in a gas jet

Aerodynamically driven stretching of polymer jets
Aerodynamically driven bending instability of polymer
jets: linearized theory
Meltblowing of a single planar polymer jet
Fully three-dimensional blowing of single and multiple polymer jets
Subsonic and supersonic solution blowing of monolithic and coreshell
bers
Blowing of natural biopolymer bers
References
Electrospinning of micro- and nanobers

5.1
5.2
5.3
5.4
5.5
5.6
5.7
5.8
5.9
5.10
5.11
[5--8] 11.12.2013 3:13PM
Contents
4.3
4.4
4.5
vi
Electrospinning of polymer solutions

Leaky dielectrics
Taylor cone and jet initiation
Straight part of the jet
Electrically driven bending instability: experimental observations
Electrically driven bending instability: theory
Branching, garlands, multineedle and needleless electrospinning
Co-electrospinning and emulsion spinning of coreshell bers
Alignment of electrospun nanober mats
Electrospinning of polymer melts
References
Additional methods and materials used to form micro- and nanobers

Island-in-the-sea multicomponent bers and nanobers
Fibers from melt fracture in meltblowing processes
Fibers from ash spinning processes
Fibers from polymer solutions in Couette ow
Centrifugal spinning, forcespinning
Electrospinning of liquid crystals, conducting polymers, biopolymers
and denatured proteins
6.7 Nanobers containing nanoparticles and nanotubes
6.8 Drawing of optical microbers
6.9 Polarization-maintaining optical microbers and multilobal bers
6.10 References
6.1
6.2
6.3
6.4
6.5
6.6
102
109
112
117
124
160
165
174
179
179
180
183
197
209
216
231
240
249
252
254
262
262
262
264
264
266
266
269
272
278
294
Tensile properties of micro- and nanobers
297
7.1
7.2
7.3
297
303
Tensile tests on individual nanobers

Tensile tests on nanober mats
Phenomenological model of stressstrain dependence of
nanober mats
304
vii
[5--8] 11.12.2013 3:13PM
Contents
7.4
7.5
8
11
306
317
Post-processing
319
8.1
8.2
8.3
8.4
319
320
331
336
Carbonization, sol-gel transformation, calcinations and metallization

Chemical cross-linking
Physical cross-linking
References
Applications of micro- and nanobers
337
9.1
9.2
337
9.3
9.4
9.5
10
Micromechanical model of stressstrain dependence of nanober mats

References
vii
Filters and membranes

Electrodes for fuel cells, batteries, supercapacitors and electrochemical
reactions
Thorny devil nanobers: enhancement of spray cooling and pool boiling
Nanouidics
References
338
344
353
357
Military applications of micro- and nanobers
359
10.1
10.2
10.3
10.4
10.5
10.6
10.7
360
364
370
374
375
375
377
Nanobers and chemical decontamination

Nanobers for biowarfare decontamination
Functionalization of nanobers for protective applications
Sensors
Nanober decontamination wipes
Respirator masks
References
Applications of micro- and nanobers, and micro- and nanoparticles:

healthcare, nutrition, drug delivery and personal care
380
11.1
11.2
11.3
11.4
11.5
11.6
11.7
11.8
381
391
393
407
417
418
419
423
Nanobrous scaffolds for tissue regeneration

Drug delivery
Desorption as drug-delivery mechanism
Modulation of drug release rate
Health suppliments (vitamin-loaded nanober mats)
Cosmetic facial masks
Electrosprayed nanoparticulate drug-delivery systems
References
Subject Index
432
viii
[5--8] 11.12.2013 3:13PM
C:/ITOOLS/WMS/CUP-NEW/4598294/WORKINGFOLDER/YARA/9781107060296PRF.3D
ix [9--12] 10.12.2013 9:55PM
Preface
Fiber-forming processes and the resulting bers have become a key element in many
modern technologies. Today, practically everyone is directly or indirectly using these
bers. Manmade macroscopic bers are widely used in our garments and many other
items of everyday life. On the other hand, much smaller microscopic and, especially,
nanobers are only beginning their path to prominence. The chemical, physical and
technological aspects of manufacturing of such bers are still weakly linked and not fully
understood. Two main processes associated with formation of micro- and nanobers
are melt- or solution blowing and electrospinning. They require concerted interaction of
synthetic chemistry, responsible for polymers used as raw materials, polymer physics,
providing a link to their viscoelastic behavior, rheological characterization of ow properties, non-Newtonian hydrodynamics of polymer solutions and melts, aerodynamics, associated with gas blowing, and electrohydrodynamics, in the case of electrospinning. The
key element of the ber-forming processes is a thin jet of polymer solution or melt, which
rapidly changes its three-dimensional conguration under the action of the aerodynamic
or electric forces applied to its surface and the internal viscous and elastic stresses. There is
a denite and imperative need to interpret and rationalize these phenomena, which requires
acquisition of extensive experimental data and establishment of an appropriate theoretical
framework as an essential element in the further technological design and optimization.
In addition to the above-mentioned broad spectrum of disciplines, this involves different
aspects associated with materials science, such as the methods developed in polymer
crystallography, and elasticity and plasticity theory. Although many aspects of berforming processes can today be considered as uncovered and well described, either
experimentally or theoretically/numerically, numerous important details are still to be
explored. The importance of this subject is attested by an exponential increase in scientic
publications devoted to microscopic and nanobers and a broad involvement of the
industries associated with ber media, nonwovens, nano-textured materials, novel
biomedical and healthcare products and optical bers, as well as defense applications.
The idea of writing this book was motivated by the need for a comprehensive exposition
of different aspects of ber-forming processes including the fundamental polymer science
facts, rheology, non-Newtonian hydrodynamics and electrohydrodynamics, applied mathematics, materials science, process development and applications. Numerous recent experimental and theoretical achievements on this subject can now be tied in an integrated text
covering signicant advances in our understanding of the micro- and nanober-forming
x [9--12] 10.12.2013 9:55PM
Preface
processes, which are radically different from those well documented for macroscopic
bers. There is still no other book in the eld of micro- and nanobers that exposes the
subject with the breadth and depth of the seminal book by A. Ziabicki, Fundamentals of
Fibre Formation, published 50 years ago and devoted to macroscopic bers. The present
book aims at charting the domain of our state-of-the-art knowledge in the eld of microand nanobers, and also highlighting the not yet fully understood challenges and outstanding issues from the perspective of interdisciplinary basic science and technology,
incorporating both fundamentals and applications. We have endeavoured to contribute
to a wide audience of researchers, engineers and post-graduate students from various
disciplines, i.e. engineering, applied chemistry and physics and materials science, as well
as technology and process development, interested in the formation of micro- and nanobers and their use in functional smart materials, such as novel lter media, nonwovens,
membranes, biomedical and healthcare products, uffy electrodes for fuel cells and
batteries, polarization-maintaining optical bers, etc.
The book is a monograph signicantly based on the results published by the authors in
the peer-reviewed journals over the last 12 years. These works covered a wide range of
the inter-related topics and in part inspired the idea to write a comprehensive monograph
encompassing the scattered mosaic of our own journal publications and the related
important results of the other groups. The present book is the culmination of these efforts.
The structure of the book is rooted in its goals. The introductory Chapter 1 exposes the
history of articial macroscopic ber technology and some basic aspects of the existing
technology and its foundations. Chapter 2 contains the basic facts from the eld of
polymer physics and rheology needed for the understanding and description of ows
of polymer solutions and melts, their solidication and crystallization. The fundamentals
of the hydrodynamics of free liquid jets moving in air, i.e. the quasi-one-dimensional
equations of such jets and basic instability phenomena are described in Chapter 3. These
equations are applied to the analysis of polymer melt- and solution blowing in Chapter 4.
In Chapter 5 these equations are supplemented by elements of electrohydrodynamics and
similarly applied to the analysis of electrospinning of polymer nanobers. Several other
methods of forming of polymer nanobers and optical glass microbers are summarized
in Chapter 6. Polymer bers and their nonwovens are frequently subjected to postprocessing, aimed at improving their properties, which is discussed in Chapter 7. The
tensile properties and strength of the individual nanobers and nanober mats are
described in Chapter 8. Chapter 9 introduces a range of applications of nanobers and
their mats as lters and membranes, catalyst supports, uffy electrodes, nanotextured
coatings that facilitate heat removal from high heat-ux surfaces, and in nanouidics.
Military applications of nanober mats for decontamination, and protection from
nuclear, biological and chemical warfare, as well as nanober-based sensors are summarized in Chapter 10. Numerous applications of micro- and nanobers and nanoparticles for healthcare and drug delivery, as well as the physical mechanisms
involved, are discussed in Chapter 11. All references are combined in the end of the
book chapters in a strictly alphabetic order. References in the text with coinciding names
of the rst author and the publication year are distinguished by an additional sufx added
to the year, e.g. Smith et al. (2011a) and Smith et al. (2012b). In the list of references
xi [9--12] 10.12.2013 9:55PM
Preface
xi
these two works can be separated by several others due to the alphabetic order of
the initials or/and the second etc. co-authors names. Moreover, in the list of references
Smith et al. (2012b) can even precede Smith et al. (2011a) if the former is Smith A.B.,
and the latter is Smith C.D.
The book allows for selective reading and Chapters 1, 2, 611 can be read stand-alone.
On the other hand, reading about the modeling aspects of melt- and solution blowing in
Chapter 4 and electrospinning in Chapter 5 imply understanding of the general quasione-dimensional equations described in Chapter 3. The book contains a wide range of
references to the relevant existing literature, albeit the description of all the subjects
treated in the book is practically self-contained, covered in depth and in sufcient detail.
This book is written for the benet of senior-year undergraduate students, graduate
students (as a text book), researchers, engineers, and consultants and practitioners in
industry (as a reference book). The scope of the book is related to the growing number of
specialists in non-Newtonian uid mechanics, rheology, electrohydrodynamics and
applied mathematics, materials scientists and engineers, textile and nonwoven engineers,
nanotechnologists, micro- and nanoscale engineers, design engineers, sustainability
engineers, energy engineers, chemical engineers, biotechnologists, bioengineers, biomedical engineers, environmental scientists and engineers, life scientists, physicists,
chemists, food scientists and engineers, etc. Readers with basic knowledge of materials
science and engineering, physics, chemistry and mathematics will be able follow the
contents of the book.
Special thanks are directed to our families, Liliya, Naomi, Shirley and Leonid Yarin,
Atefeh, Roxana and Neda Pourdeyhimi, and Sridhar, Sundar and Susithra. Without their
encouragement and help this book could not have appeared.
xii [9--12] 10.12.2013 9:55PM
C:/ITOOLS/WMS/CUP-NEW/4597349/WORKINGFOLDER/YARA/9781107060296C01.3D
1 [1--24] 10.12.2013 3:58PM
Introduction
The rst chapter is devoted to the traditional methods of ber forming, which are used to
produce macroscopic bers. Since the novel methods used to form micro- and nanobers
described in this monograph have branched from the traditional methods, an introduction
into the history of manmade bers is instructive and fully appropriate (Section 1.1).
There is a brief discussion of such traditional extrusion methods of ber forming as melt
spinning (Section 1.2), dry spinning (Section 1.3), wet spinning (Section 1.4) and the
integrated process of spunbonding, which is used to form nonwoven ber webs
(Section 1.5). Melt and dry spinning are closely related to the electrospinning used to
produce nanobers, so the discussion of these traditional methods allows a rst glimpse
of electrospinning, covered in Chapter 5. One of the key elements of spunbonding
is pulling polymer laments by fast co-owing air, which is known as meltblowing.
Meltblowing, and its offshoot solution blowing, are also used to form micro- and
nanobers, as detailed in Chapter 4. In a sense, Section 1.5 serves as an introduction to
the nonwoven nanober mats discussed later. Section 1.2 also contains some elements of
quasi-one-dimensional theory; namely, its application to the draw resonance instability
of melt spinning. In its more involved form a similar quasi-one-dimensional approach
is applied in Chapters 36 to describe processes characteristic of melt- and solution
blowing and electrospinning used to form micro- and nanobers.
1.1
History and outlook

The term ber originates from the French word bre, from Latin bra a ber, lament,
of uncertain origin, perhaps related to Latin lum thread, or from the root ndere to
split (Online Etimology Dictionary 2013). For centuries, the use of bers was limited to
natural materials such as cotton and linen, which had inherent problems with wrinkling.
Silk was difcult to produce and was often too delicate. Wool was strong and abundant,
but would shrink and was irritating next to the skin, and would not last long, as it was a
food source for moths.
The idea of forming manmade bers dates back to Robert Hooke and was expressed in
1664. In 1713, Ren Antoine de Raumur produced the rst spun glass bers and in 1735
suggested forming bers from liquid varnish. The initial progress paced in quanta of
about 100 years, and in 1883 Sir Joseph Swan issued a solution of nitrocellulose in acetic
2 [1--24] 10.12.2013 3:58PM
1 Introduction
acid into a bath lled with alcohol, and thus realized the rst wet spinning process, which
formed long continuous bers (Lewin 2007).
The invention of rayon extends back to 1855 in England, when Georges Audemars, a
Swiss chemist, discovered how to make cellulose nitrate. In 1884 Count Hilaire
de Chardonnet invented a method of forming bers from regenerated cellulose. In
1889, the introduction of fabrics made of articial silk at the Paris Exhibition received
a lot of attention, and in 1891 Chardonnet established the rst company in Besanon,
France, producing the so-called Chardonnet silk bers. After it was found that cellulose is
soluble in aqueous solutions containing copper and ammonia, mass production of
cuprammonium rayon bers was started in Germany in 1899. The rst rayon ber was
introduced as articial silk partly because of its luster and its continuous lament
nature. Viscose rayon bers were introduced by Ch. F. Cross, E.J. Bevan and C. Beadle in
1893 and commercialized in England in 1905. The American Viscose Company, formed
by S. Courtaulds and Co., Ltd., began production of rayon in 1910 in the USA.
The discovery of the origins of cellulose acetate is attributed to A.D. Little of Boston in
1893. Acetate was rst introduced during 19041910, by two brothers, Camille and
Henri Dreyfus in Basel, Switzerland (Morris 1989), making acetate motion picture lm.
The rst commercial textile uses for acetate in ber form are attributed to the Celanese
Company in 1924. Manmade cellulosics are a major player in the ber market today and
are expected to continue due to their unique properties in terms of strength, exibility and
absorbency.
Nylon bers were the rst truly synthetic bers that were industrially produced in
1939, thanks to the group led by W.H. Carothers. In 1931 Carothers reported on
research at the DuPont Company on a polymer macromolecule called nylon 6,6. By
1938, P. Schlack of the I.G. Farben Company in Germany, polymerized caprolactam and
created a different form of the polymer, identied simply as nylon 6. Nylon was the rst
commercially successful synthetic polymer. As the rst synthetic ber, nylon was
designed to replace articial silk. Nylon led to the global synthetic ber revolution.
Unlike rayon and acetate, which were derived from renewable cellulose stock, nylon was
synthesized completely from petrochemicals. This rst discovery led to the eld of
macromolecules and the new world of synthetic bers. Nylon consists of repeating
units linked by amide bonds and is frequently referred to as polyamide (PA). It is a
thermoplastic, silky material, rst used commercially in a nylon-bristled toothbrush
(1938), and then for ladys stockings (nylons; 1940), after being introduced as a fabric
at the 1939 New York Worlds Fair. Nylon stockings were shown in February 1939 at the
San Francisco Exposition. The USA entered World War II in December 1941 and all
production of nylon was dedicated for military use; nylon replaced silk in parachutes and
ak vests, and found many other military uses.
That was the origin of all the modern manmade macroscopic synthetic bers and
modern textile industry. Polyesters commercialization in 1953 was accompanied by the
introduction of triacetate. Today, polyester is the king of all synthetic bers and is found
in almost all apparel and many other applications. Polyesters have been developed with
special shapes, ber nishes, dyes and pigments, and consequently, offer the greatest
level of control over the performance attributes important to the industries they serve.
3 [1--24] 10.12.2013 3:58PM
1.1 History and outlook
The other important types of bers and ber products include berboard, made out of
wood bers (dating back to 1897), berglass (1937) and ber optics (1956).
Today, ber spinning is a commercial process to produce thin polymeric laments
that are used by a myriad of industries. Filaments can be produced from synthetic,
manmade or natural polymers, usually by the process of extrusion. Extrusion is the process
of forcing the raw materials in their liquid state through tiny orices and solidifying them to
form bers. In their initial state the raw materials are solid. If they are thermoplastic
polymers they are heated or melted, while if they are nonthermoplastic, they are dissolved
in a suitable solvent. While we now take these processes for granted, it is interesting to note
that their history only extends back about a century, and rayon was the rst manmade
ber introduced only 160 years ago. Today, manmade bers are found in almost every
application, ranging from apparel and home furnishings, to automotive industry and
medicine. The introduction of manmade and synthetic bers has led to the introduction
of many high-performance products touching many different industries. It is hard to
imagine what we would have used for these applications today without access to these
innovations.
The birth of nanobers is related to the patent by Formhals (1934), in which electrospinning of cellulose acetate bers was proposed. Electrically driven jets were in focus
much earlier (Zeleny 1914, 1917), however, these were jets of inelastic Newtonian
liquids, which are prone to capillary instability and cannot be used to form long
cylindrical laments. Only the presence of viscoelasticity in the solutions used by
Formhals allowed him to form bers. Moreover, these were nanobers, since the
presence of the electric forces results in dramatic reduction of the ber cross-sectional
diameter due to the so-called electrically driven bending instability found much later by
Reneker et al. (2000). The nanoscale of the bers was actually considered to be a
drawback in the time of Formhals, since they could not be used in the textile industry.
As a result, they did not stir up too much interest, and only occasional publications related
to the electrically driven jets of polymer solutions and melts, and the bers formed from
them appeared in the 60 years after Fromhals work (Baumgarten 1971, Larrondo and
Manley 1981a, 1981b, 1981c). However, the situation had radically changed after the
work of D.H. Renekers group in the 1990s (Doshi and Reneker 1995, Reneker and Chun
1996). This was the time of nanotechnology, new applications of nanobers were
immediately recognized and the number of publications devoted to nanober forming
started to increase exponentially. This process continues today and is described in the
technical sections of this book.
The works on traditional and novel methods of ber forming encompass
synthetic chemistry, polymer physics, non-Newtonian uid mechanics, electrohydrodynamics, applied mathematics and materials science, and require the concerted
efforts of specialists from distant elds. The need for a comprehensive monograph
encompassing different aspects of ber-forming processes materialized rst in the
seminal monograph by Ziabicki (1976). With interest in nanobers growing, several
monographs exclusively devoted to electrospinning were published. Ramakrishna et al.
(2005) covered the rapidly widening biomedical applications of electrospun nanober
mats. Filatov et al. (2007) described the work of Petryanov-Sokolovs group in the Soviet
4 [1--24] 10.12.2013 3:58PM
1 Introduction
Union, which resulted in electrospun lters for protection from radioactive aerosols.
Wendorff et al. (2012) discussed in depth the aspects of electrospinning related to
materials science.
The existing and rapidly extending processes for forming micro- and nanobers include
meltblowing, electrospinning, solution blowing and several other methods. The bers are
formed from petroleum-derived and biopolymers. The scientic foundations of berforming processes and their practical implementations are rooted in polymer physics,
rheology, non-Newtonian hydrodynamics, electrohydrodynamics, aerodynamics and
applied mathematics, while their applications extend to lters, membranes, electrodes,
coatings, nanouidics, communications (optical bers), sensors, biomedical scaffolds and
drug delivery, as well as various military-oriented aspects. The present monograph aims
for a comprehensive in-depth description of all these aspects.
1.2
Melt spinning
The basic principle of ber extrusion involves feeding pellets or granules of the solid
polymer into an extruder. The pellets are compressed, heated and melted by an extrusion
screw, then fed to a spinning pump and into the spinneret. The polymer is passed through
the extruder and then a lter, to a manifold, and is distributed to one or more spinning
positions (Hensen 1997).
The spinneret is the main component in determining ber shape and size after
extrusion. It may contain one to several hundreds of capillaries for lament spinning.
In the case of spunbond systems, discussed in more detail in Section 1.5, there are
as many as 6000 capillaries per meter. These tiny openings are very sensitive to
impurities, damage and corrosion. When warranted, the spinneret can be made from
very expensive, corrosion-resistant metals for example, for extruding uoropolymers
such as peruoroalkoxy polymer resin (PFA) and polyvinylidene diuoride (PVDF), as
well as other exotic polymers such as polyphenylene sulde (PPS). The polymer liquid
feeding them must be carefully ltered, and should not leave residue on the face of the
spinneret, as this would lead to breaks and drips. Most polymers have lubricants, antioxidants and other additives compounded into them to overcome challenges due to
polymer degradation, and spinning breaks and drips. Maintenance is also critical, and
spinnerets must be removed and cleaned on a regular basis to prevent clogging.
Thus, the term extrusion in the ber industry refers to the process of forming polymeric
laments by forcing the uid through a spinneret, and spinning is the collective term used
for the extrusion and solidication of the laments produced.
An important element is ber drawing following the extrusion. Drawing results in the
desired properties in the nal product and in a decreased ber diameter, increased
molecular orientation, increased tensile properties and a reduction in strain to failure.
The extent to which bers can be drawn depends on the properties of the materials being
extruded. Fibers are drawn as much as eight times their original length to form bers with
the desired properties.
5 [1--24] 10.12.2013 3:58PM
1.2 Melt spinning
It is interesting to note also that some polymers such as polyester (PET) require a high
ber spinning speed to form crystallinity. At about 3200 m min1 PET starts to show signs
of crystal orientation. Fibers spun at low speed shrink extensively if exposed to heat. For
PET, therefore, bers are extruded at much higher speeds 4000 to 10 000 m min1 to
overcome the issues with shrinkage.
There are several methods for forming bers from molten state or from solution: melt,
dry, wet and gel spinning. These are briey described in this and the following sections.
In melt spinning, the ber-forming polymer is melted and extruded through the spinneret,
stretched and directly solidied by cooling (Figure 1.1) and then drawn to achieve higher
degrees of orientation and crystallization. Examples are polypropylene, polyester and
nylon, among others. Melt spinning is by far the most widespread system globally.
Continuous laments, as well as discontinuous crimped bers (also referred to as staple
bers), are globally available. The process for forming continuous laments is somewhat
different from those for producing staple bers. Staple bers are produced in continuous
Polymer
Chips
Feed
Hopper
d0
Cold
Air
Melter/
Extruder
Melt
Spinning
d0, T0, V0
d, T, V
Bobbin
Twisting
and
Winding
Stretching
dL, TL, VL
3
Melt Spinning Polymer from Chip

(a)
Figure 1.1
(b)
(a) Melt spinning (Fiber Source, 2013). (b) Schematic of an individual molten threadline in
melt spinning process: 1 spinneret, 2 molten threadline, which cools down due to convective
heat transfer to the surrounding gas and solidies, 3 winding bobbin. Ziabicki (1976). Courtesy
of John Wiley and Sons.
6 [1--24] 10.12.2013 3:58PM
1 Introduction
form in large tows, which are then crimped, heat set and cut into the desired staple
lengths. These staple bers are then blended with other bers (natural as well as
manmade or synthetic bers) and are formed into a yarn. Most textile yarns are made
from blends of various staple bers; the yarn spinning technology is quite well developed
and produces incredibly interesting and desirable textures and properties. Filaments are
sometimes textured to form bulk or stretch in post-processing. Many of the facilities have
integrated polymer synthesis and ber formation that is, the most widespread processes
for staple (discontinuous) ber production couple synthesis and ber extrusion, and
thereby control costs.
Molten threadlines in the melt spinning process are free liquid jets pulled by a winding
bobbin to form solidied bers (Figure 1.1b). These molten threadlines are subjected to
several instabilities and perturbation-amplication phenomena, which can make the
resulting bers nonuniform. One of these instabilities, the so-called draw resonance,
was discovered in the seminal works of Matovich and Pearson (1969) and Pearson and
Matovich (1969), and below we discuss the elementary theory of this phenomenon.
Consider an isothermal straight liquid threadline that is issued from a spinneret hole
of radius a0 with velocity V0. The longitudinal axis along the threadline axis is denoted x.
It is reckoned from the spinneret hole where x = 0. The threadline has a circular crosssection and tapers due to the pulling force transmitted from a winding bobbin located at
x = L. The winding velocity imposed by the bobbin on the threadline at x = L is V1;
however, the cross-sectional radius at that point is to be determined.
To formulate the mass and momentum balance and derive the corresponding
quasi-one-dimensional equations, we consider an innitesimally short slice of the threadline of length dx located close to cross-section x. The liquid mass currently contained in
this slice is equal to a2dx, where is the liquid density and a(x,t) is the cross-sectional
radius, which depends on x and time t. During the time interval dt, this mass can change,
due to the liquid inux through the cross-section x, which is a2 Vjx dt, and the outow
through the cross-section x+dx, which is a2 Vjxdx dt, where V(x,t) is the longitudinal
velocity in the threadline. The mass balance reads

D a2 dx a2 V jx dt a2 V jxdx dt
1:1
Using the Taylor series, we see that

a2 V
a2 V jx a2 V jxdx
dx
x
1:2
and thus Eq. (1.1) reduces to the following differential mass balance, or following the
uid mechanical terminology, continuity equation:
a2 Va2
0
t
x
1:3
In the momentum balance, we neglect inertial forces, surface tension and gravity, and
account for only the internal normal stresses acting in the threadline cross-sections and
assumed to be dominant. Denote normal stress as xx, and thus the corresponding force
7 [1--24] 10.12.2013 3:58PM
1.2 Melt spinning
acting in cross-section x of the innitesimal threadline slice under consideration would

be xx a2 jx . Accounting for the force acting at the cross-section x+dx of the slice, we
form the momentum balance in the following form
xx a2 jx xx a2 jxdx 0
1:4
xx a2
0
x
1:5
which means that
In the elementary model we are dealing with, we assume that liquid in the threadline is
Newtonian uid and is characterized by a single rheological parameter, viscosity (see
Loitsyanskii 1966, Landau and Lifshitz 1987, Batchelor 2002). Polymer melts used in
melt spinning, as well as polymer solutions used in dry and wet spinning, can hardly be
treated as viscous Newtonian liquids, since they develop signicant and even dominant
elastic stresses in strong elongational ows. Such uids are viscoelastic. Viscoelasticity
is introduced in Chapter 2 and accounted for when considering different types of polymer
jet ows and ber-forming processes relevant to manufacturing of micro- and nanobers
in Chapters 35. The simplied rheological model of Newtonian uids employed here
would be directly relevant to the formation of optical bers (see Sections 6.6 and 6.7 in
Chapter 6). Molten glasses are Newtonian liquids, albeit their viscosities are strong
functions of temperature. Here, for simplicity, we consider an isothermal case and thus
viscosity = const.
For the incompressible Newtonian uids
xx p xx
1:6
yy p yy
1:7
where p is pressure, yy is the radial normal stress in the threadline cross-section, and xx
and yy are the normal deviatoric stresses corresponding to xx and yy, respectively.
Since the outer surface of the threadline can practically always be considered as
unloaded because all the tractions imposed by air or surface tension effects are negligibly
small in comparison with the inner stresses in the liquid, yy = 0 practically everywhere in
the cross-section (Yarin 1993), and thus Eq. (1.7) yields p = yy. The latter allows
transformation of Eq. (1.6) to the following form
xx xx yy
1:8
For Newtonian uids,

xx 2
V
;
x
yy
V
x
1:9
and thus using Eq. (1.8) we arrive at

xx 3
V
x
1:10
8 [1--24] 10.12.2013 3:58PM
1 Introduction
where the factor 3 is called the Trouton viscosity (Yarin 1993).

Substituting Eq. (1.10) into Eq. (1.5), we transform the momentum balance to the
following form

2 V
a
0
1:11
x
x
The continuity and momentum balance equations (1.3) and (1.11) form a closed
system of the two quasi-one-dimensional equations required to determine two unknown
functions, the radius and velocity distributions a(x,t) and V(x,t). These equations represent the simplest version of the quasi-one-dimensional equations of the dynamics of free
liquid jets (3.1) and (3.2) discussed in Section 3.1 in Chapter 3.
Render Eqs. (1.3) and (1.11) using the following scales: L for x, a0E1/2 for a and V1
for V. Here E= V1/V0 denotes the draw ratio, which is the governing parameter of
this problem. The continuity and momentum balance equations (1.3) and (1.11) in the
dimensionless form do not change.
In the case of melt spinning, solutions of the dimensionless system of Eqs. (1.3) and
(1.11) are subjected to the following dimensionless boundary conditions:
x 0 : a E1=2 ; V E1
1:12
x 1: V 1
1:13
In steady state, the time derivative in Eq. (1.3) vanishes and the steady-state solutions as
and Vs depend only on x:
as E1x=2 ; Vs Ex1
1:14
Since the draw ratio E is always larger than 1, Eqs. (1.14) describe tapering of the
threadline from the cross-sectional radius a0 to a smaller value a1= a0E1/2 at the winding
bobbin, whereas velocity is increasing from V0 to V1. To form smaller bers, one is
interested in increasing the draw ratio E. This, however, is subject to a severe limitation
related to the instability of the steady-state solution (1.14). Indeed, consider small
perturbations <<1 and <<1 of these solutions, i.e. take
a as x 1 x; t; V Vs x 1 x; t
1:15
Substituting these distributions into Eqs. (1.3) and (1.11), and linearizing the resulting
equations, we arrive at the following system of equations for and
Vs
Vs
0
t
x 2 x

1 2
2
0
x x ln E x2
1:16
1:17
Since the conditions at the spinneret hole and the winding bobbin are assumed to be xed,
the radius and velocity perturbations and are subject to the following boundary
conditions
9 [1--24] 10.12.2013 3:58PM
1.2 Melt spinning
x 0: 0; 0
1:18
x 1: 0
1:19
Excluding from the system of equations (1.16) and (1.17), we reduce it to a single
equation for :
2
2
0
E1x
2
x
xt
1:20
with the corresponding equation relating to .

Solving Eq. (1.20) for , one can nd the corresponding as
2
2E1x
2
t
ln E t ln E x
1:21
The solutions for and are subject to the boundary conditions (1.19). As a result, we
nd the following distribution of the radius perturbation:
1
x
E
exp t Fx; Fx exp
d
1:22
ln E
0
where is the dummy variable, and the eigenvalue satises the following characteristic
equation:
F1
1
1:23
1
E Fd
0
Solutions of Eq. (1.23) are sought on the complex plane.

The investigation of the characteristic equation (1.23) shows that in the range
1 < E < 20.22 it possesses only the solutions with a negative real part r. Therefore, the
perturbations (1.22) decay in time as expjr jt, and the steady-state solution for the
molten threadline (1.14) appears to be stable. At E = 20.22 the nondecaying small
oscillations set in, which corresponds to the solution of Eq. (1.23), = 0.693i, where i
is the imaginary unit. Mathematically speaking, this is a classical Hopf bifurcation. At
20.22 < E < 49.98 the solutions of Eq. (1.23) represent a pair of complex conjugate
solutions with a positive real part r. The corresponding linear perturbations grow in time
as exp(rt), and the draw resonance sets in (Pearson and Matovich 1969, Pearson 1985).
Some of the relevant eigenvalues found from Eq. (1.23) are listed in Table 1.1, where i
denotes the imaginary part of .
At E > 20.22, the instability of the steady-state solution (1.14) results in a bifurcation to
a new solution. Under the xed boundary conditions (1.18) and (1.19), the nonlinear
solutions of the continuity and momentum balance equations (1.3) and (1.11) can be
found either by the asymptotic method of multiple scales or numerically (Yarin 1993,
Yarin et al. 1999). The solutions represent the self-sustained oscillations (the so-called
limit-cycle solution) illustrated by the numerical results depicted in Figure 1.2. The ber
radius at the winding bobbin a(1,t) becomes a periodic function of time, with the
10
10 [1--24] 10.12.2013 3:58PM
1 Introduction
Table 1.1 Spectrum (1.23) of the linear stability problem. Reprinted with
permission from Yarin et al. (1999). Copyright 1999, AIP Publishing LLC.
E
19
20.22
21
25
28
30
35
0.0111
0
0.00495
0.01918
0.0258
0.029
0.0349
0.731
0.693
0.6723
0.5869
0.5369
0.508
0.4497
amplitude variation in the range 0.5 to 7.5 (see Figure 1.2b). Therefore, formation of
uniform bers at E > 20.22 is impossible, and the draw resonance instability severely
restricts the increase of the winding speed, the other parameters being xed.
Donnelly and Weinberger (1975) conducted model experiments with isothermal
spinning of Newtonian liquids, intended to verify experimentally the existence of the
draw resonance instability. This work represents signicant fundamental interest,
since it excluded all complicating effects related to the heat transfer at the threadline
surface, the elasticity of the polymer melt and solidication. In these experiments a
highly viscous silicon oil threadline was drawn by a winding bobbin, from which it
was immediately scraped by a doctor blade located close to the bobbin surface. The
experiments by Donnelly and Weinberger (1975) showed that the draw resonance sets
in as soon as the draw ratio E reaches a value close to 20, in full agreement with the
theory. Similar data were obtained by Ishihara and Kase (1976), who employed another
liquid, polyethylene terephthalate (PET), with approximately Newtonian behavior. In
their case the threadline solidied on contact with the winding bobbin, which was
submerged in cold water.
In the general case, perturbations can be also introduced at both ends of molten
threadlines due to oscillations of metering pumps, equipment vibratitons and blowing
of air for ber cooling. Then, the question of the threadline sensitivity to the imposed
perturbations arises. To illustrate that, consider the case where only perturbations of the
initial threadline radius are present, for example, due to the die swell affected by the
oscillations of the metering pump. In this case x = 0 approximately corresponds to the end
of the die swell (the length of the die swell is negligible compared to the length of
the threadline). Then, the linearized boundary conditions that replace the boundary
conditions (1.18) and (1.19) become
x 0: 0 sin t; 0
1:24
x 1: 0
1:25
where 0 and are the dimensionless perturbation amplitude and frequency,

respectively.
11 [1--24] 10.12.2013 3:58PM
1.2 Melt spinning
11
0.5
0.4
3
0.3
1
0.2
0.1
0
0
(a)
10
a
7
6
a (1,t)
5
4
3
2
1
0
2900 2920
(b)
Figure 1.2
2940
2960
2980
t
3000
3020
3040 3060
The phase plane describing the threadline midlength x = 0.5. The limit-cycle behavior (draw
resonance) at: (1) E = 28, (2) E = 50 and (3) E = 100. (b) The ber radius at the winding bobbin
a(1,t) as a function of time, corresponding to E = 100. Reprinted with permission from Yarin et al.
(1999). Copyright 1999, AIP Publishing LLC.
To nd the response of the threadline radius at the winding bobbin at x = 1 to the

perturbations of the initial radius at x = 0, in the linearized approximation one should
again solve the linearized problem (1.20) for the radius perturbation and calculate the
corresponding from Eq. (1.21). After that, the boundary conditions (1.24) and (1.25)
are applied. The resulting distribution of the perturbation of the threadline cross-sectional
radius reads
12
12 [1--24] 10.12.2013 3:58PM
1 Introduction

)
1 E B1 Six B2 Cix
2

x; t 0 sin t 1
B1 B22 ln E
(

)
1 E B1 Cix B2 Six
2

0 cos t
B1 B22 ln E
where
1:26

E
Si
ln E
ln E

E
Ci Ci
Ci
ln E
ln E

E
cos cos
cos
ln E
ln E

E
sin sin
sin
ln E
ln E
Si Si
1:27
1:28
1:29
1:30
In the above equations, Si(.) and Ci(.) denote the sine and cosine integral functions
(Abramowitz and Stegun 1972). In addition, the following notations are used
Si
ln E
Ci
B2 sin Si
ln E
1x

E
E
Ci
Cix Ci
ln E
ln E
1x

E
E
Six Si
Si
ln E
ln E
B1 cos Ci
1:31
1:32
1:33
1:34
The amplitude-frequency characteristic of the threadline is of interest in the applications.

It is dened as
2=
A 2
1; t2 dt
1:35
0 0
It is calculated using Eq. (1.26) and the result is shown in Figure 1.3 by curve 2, which
corresponds to the isothermal case considered in detail in this section. The result
corresponds to the sub-critical case of E < 16, where the draw resonance instability
does not appear, and the dominant role is played by the externally excited perturbations.
In the nonisothermal case, threadline cooling results in a signicant increase in liquid
viscosity (see Section 6.6 in Chapter 6), which suppresses radius uctuations at the
winding bobbin, as is shown by curve 1 in Figure 1.3.
The fully nonlinear analysis of the dynamics of molten threadlines subjected to
external perturbations at E > 20.22 reveals a nontrivial interplay of the self-sustained
13 [1--24] 10.12.2013 3:58PM
1.2 Melt spinning
13
200
2
100
1
0
0.5
Figure 1.3
1.5
2.5
The amplitude-frequency characteristic of a threadline with perturbations of the cross-sectional

radius at the die exit in the subcritical case of E = 16. (1) Non-isothermal case, (2) isothermal case,
which corresponds to Eq. (1.26). Yarin (1993). Courtesy of Pearson Education.
oscillations corresponding to the draw resonance instability with the imposed excitation.
The boundary conditions that replace (1.12) and (1.13) read
x 0: a E1=2 1 cos t; V E1
1:36
x 1: V 1
1:37
with the dimensionless parameter being responsible for the perturbation amplitude
(0=E1/2).
Yarin et al. (1999) solved Eqs. (1.3) and (1.11) numerically and showed that at the
draw ratio E 30 the nonlinear interplay between the propensity to the draw resonance
instability and the external excitation in the form of Eq. (1.36) results in quasi-periodic
or chaotic variation of the ber cross-section at the winding bobbin. The route to chaos
may be smooth, via period doubling, or explosive, via abrupt disappearance of quasiperiodic solutions; as a result a strange attractor sets in. The irregular variation of the
ber diameter corresponding to a strange attractor is illustrated in Figures 1.4 and 1.5.
These results point at the possibility that optical bers drawn from molten glass may be
perturbed nonperiodically. The experimental data of Tyushkevich et al. (1970) and
references therein show that this, indeed, may take place in glass ber drawing. They
state, . . .that a change of thickness of glass ber length-wise is a random process . . .
and the corresponding correlation function decays exponentially. Draw resonance,
being a periodic process, certainly cannot be responsible for this picture. A possible
source for the random unevenness of drawn bers may be random noise entering a
threadline. This, however, should lead to random perturbations at any winding velocity
(at any draw ratio E), which is not the case; they are seen at the bers drawn only at
high enough speed (Burgman 1970), which agrees with the predictions of the theory
outlined above.
Note that melt spinning of hollow bers is of interest in relation to fabrication
of optical bers and glass capillaries. Stability and sensitivity study of spinning of
14
14 [1--24] 10.12.2013 3:58PM
1 Introduction
0.4
0.3
V 0.2
0.1
0
1.0
2.0
3.0
a
(a)
4.0
5.0
3.5
3.0
2.5
a (1,t)
2.0
1.5
1.0
0.5
0
10000
(b)
Figure 1.4
12000
14000
16000
18000
20000
Irregular ber shape at = 0.13, = 1 and E = 30. (a) The Poincar map sampled at x = 0.5. (b) The
ber cross-sectional radius a(1,t) at the winding bobbin corresponding to a strange attractor.
Reprinted with permission from Yarin et al. (1999). Copyright 1999, AIP Publishing LLC.
15 [1--24] 10.12.2013 3:58PM
1.2 Melt spinning
15
0.6
0.4
V
0.2
0
0
2.0
(a)
4.0
a
6.0
8.0
4.0
3.5
3.0
a (1,t)
2.5
2.0
1.5
1.0
0.5
0
10000
(b)
Figure 1.5
12000
14000
16000
18000
20000
Irregular ber shape at = 0.13, = 1 and E = 30. (a) The Poincar map sampled at x = 0.5. (b) The
ber cross-sectional radius a(1,t) at the winding bobbin corresponding to a strange attractor.
isothermal drawing of hollow bers was published by Yarin et al. (1994). The results
revealed that drawing of hollow bers is also subjected to the draw resonance instability
and the threshold value of the draw ratio is the same as the one found in the present
section, E = 20.22.
16
16 [1--24] 10.12.2013 3:58PM
1 Introduction
Heat transfer with the surrounding gas is a factor in stabilizing draw resonance (Petrie
and Denn 1976, Yarin 1993 and references therein). The corresponding experimental and
theoretical results show that when heat removal is sufciently rapid and the ber
solidies before reaching the winding bobbin, the draw resonance instability can be
fully suppressed. This probably explains the fact that it is possible to exclude this type of
instability in the industrial melt-spinning processes where the values of the draw ratio E
are on the scale of several hundred, and V1 is of the order of 104 m min1. In the case of
nonisothermal drawing of optical microcapillaries, an analysis of the draw resonance
instability is given by Gospodinov and Yarin (1997).
Given the fact that melt spinning is applied to polymer melts, the effect of the
elastic stresses on the draw resonance instability is of signicant importance. In
isothermal conditions for a number of polymers, threshold values of E < 20 were
found (sometimes as small as E 2) for destabilization due to the elastic stresses
(see Weinberger et al. 1976). On the other hand, Chang and Denn (1979) observed
suppression of the draw resonance instability by elastic stresses. It is emphasized
that the theoretical investigation of the draw resonance of viscoelastic polymeric
threadlines initiated by the seminal work of Fisher and Denn (1976) (see also Hyun
1978) is still far from over.
An additional type of instability that affects melt spinning in practice is the
so-called melt fracture (Koopmans et al. 2010). Initially it manifests itself in the
form of roughness, with multiple ripples appearing at the threadline and ber surface,
called sharkskin melt fracture. A more developed instability of this type leads to the
appearance of large-amplitude bending perturbations and undulations on the threadline
right after it leaves the spinneret hole (Figure 1.6). In the case of a fully developed
melt fracture, the extrudate transforms into a succession of chunks that practically
are not connected to each other. Melt fracture sets in above a certain threshold of the
pressure drop on a spinneret hole, i.e. above a certain extrusion velocity (Figure 1.6).
The threshold value depends on liquid properties. The experimental data summarized
by Petrie and Denn (1976) shows that melt fracture is never observed in the
extrusion of Newtonian liquids. Therefore, this type of instability is directly related
to viscoelasticity of polymer melts. This phenomenon severely restricts the rate of melt
a)
b)
c)
2d
Figure 1.6
Melt fracture: (a) for the lowest extrusion speed the surface is smooth, (b) for the intermediate speed
some roughness appears, (c) at the highest speed a more developed melt fracture is observed.
Reprinted with permission from Bertola et al. (2003). Copyright 2003, American Physical Society.
17 [1--24] 10.12.2013 3:58PM
1.3 Dry spinning
17
spinning. A fully acceptable theory and explanation of melt fracture is still unavailable.
However, all signs point to the emergence of self-sustained oscillations in polymer melt
ow inside the die (Drda and Wang 1995, Wang and Drda 1996, Yarin and Graham
1998, Bertola et al. 2003, Koopmans et al. 2010).
1.3
Dry spinning
Dry spinning of sufciently concentrated polymer solutions is used in cases where
polymers are not thermoplastics and cannot be effectively melted. If solvents are volatile
enough, threadline precipitation and solidication is possible on the way from the
spinneret hole to the winding bobbin, and this method is called dry spinning
(Figure 1.7). Sometimes evaporation is facilitated with a stream of air or inert gas
blown at threadline. The list of solvents appropriate for dry spinning includes alcohols,
acetone, ether solvents and tetrahydrofuran (THF). Because there is no precipitating
liquid involved (see Section 1.4), the ber does not need to be dried and the solvent is
more easily recovered. Examples are acetate, triacetate, acrylic, modacrylic, spandex,
etc. Dry spinning is typically conducted in a vertical connement, where polymer
solution jets are surrounded by a stream of a high-temperature gas (e.g. air) that facilitates
solvent evaporation. Chardonnet silk bers, discussed in Section 1.1, were spun from
nitrocellulose solutions in a blend of alcohol and ether using dry spinning, which makes
the latter the ancestor of melt spinning described in Section 1.2, since the ber-forming
and pulling systems are alike. On the other hand, the physical processes which take place
in and around the liquid threadline in dry spinning of macroscopic bers are similar to
1
5
8
Figure 1.7
Schematic of dry spinning: (1) dosing pump, (2) spinneret, (3) liquid threadlines, (4) drying section,
(5)(7) receiving pulleys, (8) hot air inlet, (9) hot air outlet. Ziabicki (1976). Courtesy of John
Wiley & Sons.
18
18 [1--24] 10.12.2013 3:58PM
1 Introduction
1
10
7
6
Figure 1.8
Schematic of horizontal wet-spinning process. (1) polymer supply line, (2) spinneret, (3)
threadlines, (4) precipitation bath, (5) false twist grooved rolls, (6) inlet for the nonsolvent, outlet
of the bath content (mixture of solvent and nonsolvent), (8) plastication bath, (9) and (10) dry-heat
setting cans. Ziabicki (1976). Courtesy of John Wiley & Sons.
those in solution blowing and electrospinning of polymer micro- and nanobers,

discussed in detail, respectively, in Sections 4.8 and 4.9 in Chapter 4, and in several
sections in Chapter 5.
1.4
Wet or solvent spinning, gel spinning

Wet spinning is used to form bers from polymers that are not thermoplastics and
thus cannot be spun as melts, and can be dissolved only in nonvolatile solvents.
Then, polymer solutions are issued as jets into a precipitation bath lled with a
nonsolvent for a particular polymer. As a result of the binary diffusion of solvent
into the bath, and nonsolvent into the threadlines, polymer precipitates and solidied
laments are pulled out by the receiving pulleys (Figure 1.8). Wet spinning is applied
to polyamides (nylon 6 and nylon 6,6), as well as to process rigid-rod polymers, such as
Kevlar brand Aramid bers. In the latter case, sulfuric acid (H2SO4) is used as the
solvent, and water in the precipitation bath as the nonsolvent. The other examples are
acrylic, rayon, etc.
Combinations of wet and dry spinning are also possible, with part of the threadline being
in air, and another part in the precipitation bath. The part in air can be a polymer gel, which
is partially liquid. The nal solidication of the bers is achieved after solvent evaporation
in air followed by precipitation in the bath. Such a process is called gel spinning.
1.5
Spunbonding
The spunbond process is an integrated system producing nonwoven webs of continuous
laments that are then bonded normally in a single step to form a strong and exible
fabric. While attempts have been made to produce laments from solution, the current
processes in place use a thermoplastic polymer resin to form the laments/fabrics. In its
standard form, the spunbond process results in relatively large bers, more than 20 m in
diameter.
19 [1--24] 10.12.2013 3:58PM
1.5 Spunbonding
19
In the spunbond processes, when a polymer other than a polyolen is used, it is

often necessary to have drying and/or crystallizing (in the case of PET) of the
polymer chips. The polymer chips are delivered by extruders designed to process
polymers with a wide range of characteristics. The extruder barrel is heated and the
high degree of compression of the polymer chips during processing also contributes
to the generation of heat that results in melting of the polymer chips. The current
standard processes can reach a maximum of 350 C this is sufcient for most
ordinary polymers used in the formation of nonwoven fabrics today, ranging from
polyolen to polyester.
The molten polymer exiting the extruder goes through a lter pack to remove solid
particulates and contamination with screen lters. It then ows into a gear pump that
meters and delivers precise amounts of polymer to the spinpack. The spinpack is
contained in the spinbeam that in todays technology would cover the entire width of
the machine. At the bottom of the spin beam resides a spinneret plate, normally with up to
6000 holes m1. The molten polymer ows through the spinneret holes and forms
laments upon solidication brought about by quenching. In current (modern) systems,
the laments form a curtain, owing into an air gap where fast moving air provides a drag
force that pulls the laments away from the spinneret and reduces the ber diameter, but
more importantly leads to a much higher degree of molecular orientation in the structure.
Upon drawing, the laments may be traveling at velocities exceeding 8000 m min1.
Pulling polymer bers by high-speed co-owing air is characteristic of the meltblowing
process considered in detail in Chapter 4.
As the laments approach a moving collection belt, they slow down, and due to the
resulting bending instabilities, the laments intertwine and are laid on the moving belt.
The lay-down is a critical element in the formation of the nal structure, as it controls the
degree of isotropy in the structure. The web is then normally bonded by two heated rolls
(calendar), or sometimes other forms of bonding, such as hydroentangling or chemical
bonding, may also be employed.
The features of the process that give spunbond fabrics their uniqueness lie in the
technologies used to extrude the laments and form the web. These are
described below, in a historical context; most of them are still in use in their
original or somewhat modied form. The spunbond process emerged from lament
spinning and the early systems developed in the 1950s and 1960s relied on using
multiple lament extrusion units placed in a row to form continuous webs. Each
extrusion unit was equipped with spinnerets containing a few hundred holes. The
bers, instead of being wound on a bobbin to form continuous lament yarn, were
fed into a lay-down system to separate and place the laments on a moving belt.
The early systems suffered from incomplete lament separation, causing rope
formation and spatial mass nonuniformity, as well as streaks caused by using
multiple side-by-side systems to form a web. A number of techniques were
developed to deal with these inadequacies these included the use of lament
distributors in various forms, and also the use of electrostatic charge to cause
lament separation. A good review of these early developments is found in the
book by Batra and Pourdeyhimi (2012).
20
20 [1--24] 10.12.2013 3:58PM
1 Introduction
10
11
15
14
14b
14a
l
20a
13
16
12
Figure 1.9
Schematic of the full-width slot attenuator-based system (Matsuki et al. 1974).
The origin of the modern systems stems from a few key developments. The rst is
the introduction of curtain spinning using a single spinneret block by Matsuki et al.
(1974), assignee Asahi. To improve the formation (web uniformity) they chose to use
high capacity extruders, which could supply polymer to a single spinneret block
that extended across the collecting belt. Figure 1.9 shows a schematic of the equipment.
The rectangular spinneret die would deliver a two-dimensional array of laments
continuously across the width of the machine. The so-called open systems offered
today by Hills, Neumag and others are all based on this early development.
The next most signicant advance was due to a development by Kimberly Clark that
combined the quenching and lament draw functions in a full-width slot attenuator, a
patent granted in 1982. Figure 1.10 shows a quench and draw box integrated with
the spinneret section into a single enclosure. This led to the development of technologies
known today as closed systems. Such a system is offered by Reicol (Reifenhauser),
the global leader in supplying such systems. The rst Reifenhauser system was introduced in 1989 and modied the full-width slot concept of Kimberly Clarks technology.
The schematic of the system is shown in Figure 1.11. It integrates the spinneret block
with the quenching and draw system into a single closed unit.
The next major development relates to the use of bicomponent bers in the spunbond
process. The earliest patent discussing the manufacture of a bicomponent ber may well
have been granted to DuPont in 1934. A series of subsequent developments followed.
Hills (1992) patented the multicomponent spin pack in 1992.
In the area of spunbond today, there are numerous commercial systems installed
around the globe. These naturally employ two dryer hoppers, each feeding a different
polymer to two different extruders, followed by lter-packs, gear pumps and onto the
spin block, containing the spinnerets. The spin packs used here are of special design; one
21 [1--24] 10.12.2013 3:58PM
1.5 Spunbonding
21
18
26
28
10
46
29
28
54
24
20
25
40
34
36
38
Figure 1.10
Integrated spinneret, quenching and slot attenuator (Appel and Mormon 1982).
is shown schematically in Figure 1.12. The designs are quite complex, but permit the
extrusion of a variety of cross-sectional arrangements of two or more polymers in the
lament. The rest of the process is essentially the same as a homocomponent system,
with the usual draw and collection systems. The initial motivation for the use of
bicomponent bers was essentially due to the fact that bicomponent laments facilitate
thermal bonding. For example, a lower melting-point sheath allows thermal bonding at a
lower temperature without melting the core, potentially resulting in improved performance at a lower cost. This arrangement allows the exploitation of a higher-strength second
component at a lower energy cost during bonding. However, the rationale for using
bicomponent laments in spunbond fabrics goes far beyond simply thermal bonding.
Bicomponent structures are enablers that can help develop new generations of nonwovens with improvements in strength and softness, with ultrane bers, improvement
in loft (crimped bers) in the fabric, etc. and even nonwovens with stretch and recovery.
22 [1--24] 10.12.2013 3:58PM
22
1 Introduction
3
14
12
10
13
y1
11
17
19
4 35
a
16
35
15
18
5
6
19
18
19
yn
xn
y2
x2
18
20
9
II
1
y1
Figure 1.11
x1
The Reifenhauser system (Balk 1989).

11
16
16
10
20
18
16
17 19
23
12
26
28
25
29
33
32
27
33
32
38
33
33
32
38
16
36
38
14
32
30
22
13
36
36
36
36
36
41
41
38
41
15
Figure 1.12
Spin pack for bicomponent bers (Hills 1992).
16
32
41
40
23 [1--24] 10.12.2013 3:58PM
1.6 References
1.6
23
References
Abramowitz, M., Stegun, I. A. (Editors), 1972. Handbook of Mathematical Functions. Dover,
New York
Appel, D. W., Mormon, M. T., 1982. Method for forming nonwoven webs. US Patent No. 4340563.
Balk, H., 1989. Apparatus for making a spun eece from endless synthetic-resin lament. US
Patent No. 4812112.
Batchelor, G. K., 2002. An Introduction to Fluid Dynamics. Cambridge University Press,
Cambridge.
Batra, S. K., Pourdeyhimi, B., 2012. Introduction to Nonwovens, DEStech Publishing, Lancaster.
Baumgarten, P. K., 1971. Electrostatic spinning of acrylic microbers. J. Colloid. Interface Sci. 36,
7179.
Bertola, V., Meulenbroek, B., Wagner, C., Storm, C., Morozov, A., van Saarloos, W., Bonn, D., 2003.
Experimental evidence for an intrinsic route to polymer melt fracture phenomena: A nonlinear
instability of viscoelastic poiseuille ow. Phys. Rev. E 90, 114502.
Burgman, J. A., 1970. Liquid glass jets in the forming of continuous glass bers. Glass Technol. 11,
110116.
Chang, J. C., Denn, M. M., 1979. An experimental study of isothermal spinning of a Newtonian
and viscoelastic liquid. J. Non-Newton. Fluid Mech. 5, 369385.
Donnelly, R. J., Weinberger, C. B., 1975. Stability of isothermal ber spinning of a Newtonian
uid. Ind. Eng. Chem. Fundam. 14, 334337.
Doshi, J., Reneker, D. H., 1995. Electrospinning process and applications of electrospun bers.
J. Electrostatics 35, 151160.
Drda, P. P., Wang, S. Q., 1995. Stick-slip transition of polymer melt/solid interfaces. Phys. Rev. Lett.
75, 26982701.
Fiber Source, 2013. Available at http://www.bersource.com/f-tutor/techpag.htm. Accessed July
27, 2013.
Filatov, Y., Budyka, A., Kirichenko, V., 2007. Electrospinning of Micro- and Nanobers.
Fundamentals and Applications in Separation and Filtration Processes. Begell House,
New York.
Fisher, R. J., Denn, M. M., 1976. A theory of isothermal melt spinning and draw resonance.
AIChE J. 22, 236246.
Formhals, A., 1934. Process and apparatus for preparing articial threads. US Patent No. 1975504.
Gospodinov, P., Yarin, A. L., 1997. Draw resonance of optical micro-capillaries in non isothermal
drawing. Int. J. Multiphase Flow 23, 967976.
Hensen, F. (Editor), 1997. Plastic Extrusion Technology. C.Hanser, Munich.
Hills, W. H., 1992. Method of making plural component bers. US Patent No. 5162074.
Hyun, J. C., 1978. Theory of draw resonance. Part II: Power-law and Maxwell uids. AIChE J. 24,
423426.
Ishihara, H., Kase, S., 1976. Studies on melt spinning. VI. Simulation of draw resonance using
Newtonian and power law viscosities. J. Appl. Polym. Sci. 20, 169191.
Koopmans, R., Den Doelder, J., Molenaar, J., 2010. Polymer Melt Fracture. CRC Press, Boca
Raton, FL.
Landau, L. D., Lifshitz, E. M., 1987. Fluid Mechanics. Pergamon Press, New York.
Larrondo, L., Manley, R. S. J., 1981a. Electrostatic ber spinning from polymer melts.
I. Experimental observations on ber formation and properties. J. Polym. Sci., Polym. Phys.
Ed. 19, 909920.
24
24 [1--24] 10.12.2013 3:58PM
1 Introduction
Larrondo, L., Manley, R. S. J., 1981b. Electrostatic ber spinning from polymer melts. II.
Examination of the ow eld in an electrically driven jet. J. Polym. Sci., Polym. Phys. Ed. 19,
921932.
Larrondo, L., Manley, R. S. J., 1981c. Electrostatic ber spinning from polymer melts. III.
Electrostatic deformation of a pendant drop of polymer melt. J. Polym. Sci., Polym. Phys. Ed.
19, 933940.
Lewin, M. (Editor), 2007. Handbook of Fiber Chemistry. 3rd Edition. CRC Press, Boca Raton, FL.
Loitsyanskii, L. G., 1966. Mechanics of Liquids and Gases. Pergamon Press, Oxford (the English
translation of the 2nd Russian edition), and the 3rd Russian edition published by Nauka,
Moscow, 1970.
Matovich, M. A., Pearson, J. R. A., 1969. Spinning a molten threadline. Steady-state viscous ows.
Ind. and Eng. Chem. Fundam. 8, 512520.
Matsuki, M., Nishimura, S., Goto, M., 1974. Apparatus for producing non-woven eeces. US
Patent No. 3802817.
Morris, P. J. T., 1989. The American Synthetic Rubber Research Program. Pennsylvania Press,
Philadelphia, PA.
Online Etimology Dictionary, 2013. Available at http://www.etymonline.com/index.php?term=
ber. Accessed July 27, 2013.
Pearson, J. R. A., 1985. Mechanics of Polymer Processing. Elsevier, London.
Pearson, J. R. A., Matovich, M., 1969. Spinning a molten threadline. Stability. Ind. Eng. Chem.
Fundam. 8, 605609.
Petrie, C. J. S., Denn, M. M., 1976. Instabilities in polymer processing. AIChE J. 22, 209236.
Ramakrishna, S., Fujihara K., Teo, W. E., Lim, T. C., Ma, Z., 2005. An Introduction to
Electrospinning and Nanobers. World Scientic, Singapore.
Reneker, D. H., Chun, I., 1996. Nanometer diameter bers of polymer, produced by
electrospinning. Nanotechnol. 7, 216223.
Reneker, D. H., Yarin, A. L., Fong, H., Koombhongse, S., 2000. Bending instability of electrically
charged liquid jets of polymer solutions in electospinning. J. Appl. Phys. 87, 45314547.
Tyushkevich, N. I., Krasko, A. S., Chepurkin, A. A., Shiman, O. P., Kozello, T. O., Ananich, N. A.,
1970. Study of the unevenness of glas ber by statistical methods. Glass Ceram. 27, 9597.
Wang, S. Q., Drda, P. P., 1996. Superuid-like stick-slip transition in capillary ow of linear
polyethylene melts. I. General features. Macromol. 29, 26272631.
Weinberger, C. B., Cruz-Saenz, G. F., Donnelly, G. J., 1976. Onset of draw resonance during isothermal
melt spinning: a comparison between measurements and predictions. AIChE J. 22, 441448.
Wendorff, J. H., Agarwal, S., Greiner, A., 2012. Electrospinning. Wiley-VCH, Weinheim.
Yarin, A. L., 1993. Free Liquid Jets and Films: Hydrodynamics and Rheology. Longman Scientic
and Technical and John Wiley & Sons, Harlow, NY.
Yarin, A. L., Gospodinov, P., Gottlieb, O., Graham, M. D., 1999. Newtonian glass ber drawing:
Chaotic variation of the cross-sectional radius. Phys. Fluids 11, 32013208.
Yarin, A. L., Gospodinov, P., Roussinov, V., 1994. Stability loss and sensitivity in hollow ber
drawing. Phys. Fluids 6, 14541463.
Yarin, A. L., Graham, M. D., 1998. A model for slip at polymer/solid interfaces. J. Rheol. 42,
14911504.
Zeleny, J., 1914. The electrical discharge from liquid points and a hydrostatic method of measuring
the electric intensity at their surfaces. Phys. Rev. 3, 6991.
Zeleny, J., 1917. Instability of electried liquid surfaces. Phys. Rev. 10, 16.
Ziabicki, A. 1976. Fundamentals of Fibre Formation. John Wiley & Sons, London.
25 [25--62] 10.12.2013 4:42PM
Polymer physics and rheology
Several physical concepts that are of the utmost importance in ber-forming processes
are described in this chapter. The basic physical model of a exible polymer macromolecule as a random walk is outlined in Section 2.1. The elongational and shear
rheometry of polymer solutions and melts, which elucidate the stress relation with strains
and strain rate, as well as stress relaxation is described in Section 2.2. The phenomenological rheological constitutive equations appropriate for the description of viscoelastic
polymer solutions and melts are introduced in Section 2.3. The micromechanical foundations of the entropic elasticity responsible for viscoelasticity of polymer solutions and
melts are sketched out in Section 2.4. Solidication and crystallization are discussed in
Sections 2.5 and 2.6, respectively.
2.1
Polymer structure, macromolecular chains, Kuhn segment,

persistence length
A linear polymer macromolecule can be represented as a succession of identical rigid
segments connected at arbitrary angles, i.e. freely jointed with each other (Flory 1969,
de Gennes 1979, Doi and Edwards (1986). Such a macromolecule is comprised of
N segments, each of length b. The total length of a fully stretched macromolecule is
then L = Nb. The rigid segments are called Kuhn segments. A real macromolecular chain
consisting of n monomers is idealized as a random walk of N Kuhn segments, which are
not monomers, nor is N identical to the degree of polymerization n. If the number of
Kuhn segments in a macromolecule is not large, i.e. N is close to 1, it is rather inexible,
almost rod-like. On the other hand, if N >> 1, the macromolecule is very exible, and on
length scales that are signicant compared to b, but much smaller than L, it can be viewed
as a exible string. Persistence length is another length scale that characterizes the
resistance of segments of macromolecular chains to bending. It is of the same order of
magnitude as the length of the Kuhn segments.
2.2
Elongational and shear rheometry

Rheological characterization of viscoelastic polymer solutions and melts used in berforming processes should include elongational and shear rheometry. In particular, not
26
26 [25--62] 10.12.2013 4:42PM
2 Polymer physics and rheology
Solenoid
High-speed CCD
Light source
Figure 2.1
Schematic of an elongational rheometer based on self-thinning threads of polymer solutions.

Dotted lines below the solenoid show the initial uid conguration before the movement of the top
plate. Tiwari et al. (2009), with kind permission of Springer Science-Business Media.
only zero-shear viscosity and ow curves in simple shear ow should be measured to

characterize rheological behavior of polymeric liquids, but also their elastic relaxation
time, since all such liquids are viscoelastic (see Section 2.3). Moreover, it is highly
desirable to measure the viscoelastic properties in the uniaxial elongational ows at
high strain rates, i.e. in the situations resembling those in ber-forming processes. It
should be emphasized that rheological characterization of polymeric liquids in both
simple shear and uniaxial elongational ows encompasses the most important types of
ow kinematics, which are mixed in all other types of ows. In this context, elongational rheometers based on uniaxial elongational ow resulting in self-thinning threads
(discussed in Yarin 1993, Stelter et al. 1999, 2000, 2002, McKinley and Tripathi 2000,
Wunderlich et al. 2000, Yarin et al. 2004, Reneker et al. 2007 and Tiwari et al. 2009)
can be used to characterize spinnability (i.e. the ability to form bers) and to measure
the rheological parameters of polymer solutions, depending on the polymer type and
concentration. An elongational rheometer of this type consists of a stationary lower
plate and a moveable upper plate driven by a solenoid, a continuous light source and a
high-speed video camera (see Figure 2.1). A droplet of the polymer solution of interest
is placed in between the plates. Then the upper plate (spindle) is retracted upward
rapidly with the help of the solenoid, forming a cylindrical liquid thread. After the plate
motion has ceased, the thread exhibits a uniaxial elongational ow driven by surface
tension that results in self-thinning of the thread. A high-speed digital camera (for
example, Redlake MotionPro) equipped with a 185-mm macro-lens records the
thread self-thinning. An example of a thread-thinning video clip is shown in
Figure 2.2. This variant of data acquisition is the most effective way to analyze selfthinning threads with sufcient accuracy.
It should be emphasized that ow in the thread is directed from the center toward the
two end regions. This ow is driven by surface tension, since the capillary pressure in the
thread is much higher than in the end regions. A detailed theory of such ow is given as
an example in Section 2.3 (see Eqs. 2.312.39 there). The nal result, which is relevant
for measuring the elastic relaxation time reads

t
d d0 exp
2:1
3
27 [25--62] 10.12.2013 4:42PM
2.2 Elongational and shear rheometry
Figure 2.2
27
Self-thinning thread of 1000 ppm aqueous solution of polyacrylamide Praestol 2540. Several
snapshots illustrate how the thread diameter decreases as time increases. Reprinted with permission
from Stelter et al. (2000). Copyright 2000, The Society of Rheology.
where d = 2a is the cross-sectional diameter in the thread (a is the radius), with the initial
value d0 at t = 0; t is time. Correspondingly, the elongational viscosity el, which is the
ratio of the normal axial stress to the rate of elongation, exponentially increases in time as

3
t
el
exp
2:2
d0
3
where is the surface tension.
An example of data acquired using an elongational rheometer with a self-thinning
thread is shown in Figure 2.3. Fitting Eq. (2.1) to the data, as shown in that gure, yields
the values of the elastic relaxation time . For polymer solutions relevant in forming
nanobers using solution blowing and electrospinning (see Chapters 4 and 5, respectively), such an approach is demonstrated in Theron et al. (2004). The polymer solutions
they studied are listed in Table 2.1, along with the rheological parameters measured.
These include, zero shear viscosities and ow curves of these polymer solutions in
simple shear ow. For example, Figure 2.4 shows that poly(ethylene oxide), PEO, solutions revealed pronounced shear thinning. Relaxation times are in the range 1360 ms.
Relaxation times of polycaprolactone, PCL, solutions could not be measured because of
the high evaporation rates of the solvents, acetone and methylene chloride (MC). Note,
also that surface tension measurements conducted with a pulsating bubble surfactometer
28
28 [25--62] 10.12.2013 4:42PM
Table 2.1 Rheological properties of several polymer solutions: poly(ethylene oxide), PEO; poly(acrylic acid), PAA;
poly(vinyl alcohol), PVA; polyurethane, PU; polycaprolactone, PCL. Solvents: tetrahydrofuran, THF; methylene chloride
(dichloromethane, MC); dimethylformamide, DMF. Molecular weight is denoted Mw, polymer weight concentration,
C, zero-shear viscosity, , and the elastic relaxation time, .
Polymer
Mw [Da]
Solvents
C[%]
[P] = [101 Pa.s]
[ms]
PEO
6 105
Ethanol/water (40/60)
PEO
106
PEO
4 106
PEO
Water
PVA
PU
106
4 106
2.5 105
4.5 105
104
Tecoex
THF/ethanol (50/50)
PCL
8 104
Acetone
PCL
8 104
MC/DMF (75/25)
MC/DMF (40/60)
2
3
4
6
2
3
1
2
3
2
1
6
5
6
6
8
8
10
14
10
10
2.85
12
30
432
15.9
96
42.5
900
3350
5.7
26
4.55
2.55
3.55
0.25
82
107
165
400
670
950
21
25
28
33
142
183
217
298
359
128
48.1
22.75
29.6
1.77
PAA
1
Measured
d [mm]
Fitted
0.1
250 ppm
62.5
0.01
Figure 2.3
500 ppm
125 ppm
200
400
600
800
t [ms]
1000
1000 ppm
1200
1400
1600
Diameter decrease of a liquid thread of an aqueous Praestol 2540 solution for the different
concentrations listed in the graph. Reprinted with permission from Stelter et al. (2000).
Copyright 2000, The Society of Rheology.
29 [25--62] 10.12.2013 4:42PM
29
3.5
C = 2%
3%
4%
Viscosity [Pa.s]
3.0
2.5
2.0
1.5
4%
1.0
3%
0.5
0
Figure 2.4
2%
0
500
1000
Shear rate [1/s]
1500
Flow curves: shear viscosity versus shear rate. Plots for solutions of PEO (Mw = 6 105 Da)
in ethanol/water (40/60) at different weight concentrations. Reprinted from Theron et al. (2004),
with permission from Elsevier.
Light sensor
Stretching
arm
Polymer fluid
column
Linear
source
of light
High-speed
camera
Figure 2.5
Polymer
drop
Sketch of the elongational rheometer used by Xu et al. (2003) and Reneker et al. (2007). Reprinted
from Reneker et al. (2007), with permission from Elsevier.
showed that surface tension is mainly a function of the solvent in the solutions and tends to
be less sensitive to variation in the polymer concentration. Therefore, the values of surface
tension can be taken as those of the solvents.
Xu et al. (2003) designed an elongational rheometer, shown in Figure 2.5. The
rheometer was able to generate extensional ows mechanically and was applied to
polymer solutions used in electrospinning. In their experiments, approximately 0.2 ml
of polymer solution was placed in a reservoir located on the bottom plate of the
rheometer. A cylindrical tip mounted on a horizontal arm was dipped in the polymer
solution initially. The arm could move vertically at a constant speed for a certain distance.
30
Figure 2.6
30 [25--62] 10.12.2013 4:42PM
t = 0 ms
t = 280 ms
600 micron
600 micron
Reections of two linear light sources from the lateral surface of the polymer solution thread.
Reprinted from Reneker et al. (2007), with permission from Elsevier.
The motion stopped after a distance, chosen by the experimenter, where the horizontal
arm ran in between the infrared emitter and the sensor pair and blocked the infrared
emission. The tip picked up a portion of the polymer solution and moved 21 mm upward
at a constant speed of 350 mm s1. A polymer-solution thread was created between
the bottom plate and the tip. Self-thinning due to gravity of the thread then started. The
threads were much longer than those in the rheometer sketched in Figure 2.1, therefore
gravity rather than surface tension determined the self-thinning. A high-speed camera
monitored the decreasing diameter of this self-thinning thread. Two linear halogen lights
were adjusted to provide proper illumination to the lament and a dark background. The
liquid thread was outlined by the specular reection of two linear lights from its lateral
surface. The contour of the thread was seen as two bright lines on a dark background.
Figure 2.6 shows the reections of the two linear light sources from the lateral surface of
the liquid thread. The initial thread diameter d0 was recorded when the probe tip reached
its highest position at time t = 0. Polymer-solution threads with diameters as small as
80 m could be accurately measured. The rheometer was mounted on a vibrationdamped imaging bench. A special thread holder was designed to minimize its vibration
during the self-thinning process.
In the rheometer of Xu et al. (2003) shown in Figure 2.5, polymer threads thinned
mostly due to gravity, which resulted in the following expressions (Reneker et al. 2007):

t
d d0 exp
2:3
2
with the corresponding elongational viscosity being

t
el g0 exp
2:4
where is the solution density, 0 the initial lament length, and g gravity acceleration.
Figure 2.7 shows the entire thread during the stretching and self-thinning stages.
The cylindrical shape of the thread reects that a uniform elongational ow was
31 [25--62] 10.12.2013 4:42PM
Entire jet,
during stretching
Figure 2.7
31
Entire jet,
self thinning
The entire thread (jet) during stretching and self-thinning stages. Reprinted from
Reneker et al. (2007), with permission from Elsevier.
produced. At the middle part of the thread a pure extensional deformation was sustained.
The diameter decrease was monitored at this position. The large strain was produced by
the long residence time of the thread during the thinning process. The time evolution of the
thread diameter was studied for several aqueous solutions of PEO of different concentrations. The results shown in Figure 2.8 reveal that the rate of thread thinning during the
self-thinning process was higher for solutions with lower polymer concentrations.
The relaxation time found by tting Eq. (2.3) to the experimental data is shown in
Figure 2.9. It can be seen that the relaxation time window suitable for electrospinning
nanobers of PEO solutions is in the range 20 ms to 80 ms, which agrees with the data for
PEO in Table 2.1 obtained using another rheometer. The logarithm of the relaxation time
decreases linearly with polymer concentration. The corresponding elongational viscosities of the aqueous solutions of PEO found using Eq. (2.4) are shown in Figure 2.10.
They exponentially increase in time during the self-thinning stage. The gure also shows
that a low-concentration solution has a lower initial elongational viscosity. However, in
such solutions the viscosity increases at a much faster rate than in the high-concentration
solutions. As a result, the low-concentration PEO solutions can reveal a higher elongational viscosity than the high-concentration solutions at the latter stages of thinning. Note
that the threads were birefringent, which points to a high order of alignment of polymer
macromolecules along the self-thinning threads (Xu et al. 2003).
Han et al. (2008) introduced two additional approaches to elongational rheometry.
They worked with polymer solution jets stretched by electric forces, as in electrospinning
used to form polymer nanobers (see Chapter 5). A digital video camera attached to an
32
32 [25--62] 10.12.2013 4:42PM
700
Diameter (micron)
600
500
400
300
200
10%
100
80
4%
2%
6%
8%
0
0
50
100
150
200
250
300
350
400
450
Time (ms)
Figure 2.8
Time evolution of PEO threads (Mw is of the order of 6 105 Da). The weight concentrations
are shown on the graph. Reprinted from Reneker et al. (2007), with permission from Elsevier.
0.2
0.18
Relaxation time(s)
0.16
y = 0.0076e0.308x
0.14
0.12
0.1
0.08
0.06
0.048
0.04
Concentration range
suitable for electrospinning
0.02
0
0
Figure 2.9
6
Concentration (%)
10
Relaxation time of aqueous PEO solutions. Reprinted from Reneker et al. (2007), with permission
from Elsevier.
optical microscope was used to record the prole of the transition zone between the
polymer droplet at the exit of a capillary with an inserted electrode and the beginning of
the jet pulled towards a grounded counter electrode (see the sketch in Figure 2.11 and the
image in Figure 5.11 in Chapter 5). The experiment was conducted as follows. Polymer
solutions were held in a glass pipette that had a tip several centimeters long with an
800 m inner diameter. A copper wire was immersed in the solution and connected to a
33 [25--62] 10.12.2013 4:42PM
33
450
50
Elongational viscosity (pa*s)
400
10%
10%
8%
8%
40
30
350
6%
20
300
4%
10
2%
250
50
200
100
6%
150
100
50
0
0
Figure 2.10
50
100
150
200
250
Time (ms)
300
350
400
450
Elongational viscosity of aqueous PEO solutions. Reprinted from Reneker et al. (2007),
(1)
(5)
(2)
(6)
(7)
x
(4)
A
(3)
Figure 2.11
Schematic drawing of an elongational rheometer based on measuring radius and velocity

distributions, and evaluating the stretching rate: (1) pipette, (2) high-voltage DC power supply, (3)
ammeter, (4) grounded collector, (5) digital video camera, (6) optical microscope, (7) illumination
light. Reprinted from Han et al. (2008), with permission from Elsevier.
high-voltage power supply which generated DC voltage up to 13 kV. A grounded

horizontal plate was placed below the pipette tip to serve as a collector electrode. The
distance between the pipette tip and the grounded plate could be adjusted from 0.1 to
30 cm. An ammeter was connected between the collector plate and ground to measure the
current carried by the electrically driven jet.
34
34 [25--62] 10.12.2013 4:43PM
(6)
(1)
(2)
(8)
(5)
W(x)
x
(7)
(4)
A
(3)
Figure 2.12
Schematic drawing of an elongational rheometer for measuring normal axial stresses in electrically
stretched jets of polymer solutions: (1) pipette, (2) high-voltage DC power supply, (3) ammeter,
(4) grounded collector, (5) loudspeaker, (6) polystyrene bar, (7) high-speed camera, (8)
illumination light. The viewing direction of the camera is perpendicular to both the displacement
direction and the jet axis. Reprinted from Han et al. (2008), with permission from Elsevier.
A detailed theory of this device is given in Section 5.4 in Chapter 5. It shows that the
measured distribution of the jet cross-sectional radius a(x) along the straight jet axis can
be converted into the normal axial stress and the elastic relaxation time in the polymer
solution undergoing strong uniaxial stretching.
Another elongational rheometer introduced by Han et al. (2008) and applied to
electrically stretched jets of polymer solutions is sketched in Figure 2.12. A bar of
foamed polystyrene with high stiffness and low density was glued onto a loudspeaker
that produced a single lateral displacement pulse when a suitable electrical pulse was
applied. The bar was positioned as shown in Figure 2.12, so that the displacement pulse
from the loudspeaker imparted a pulse-like lateral motion to the tip of the pipette issuing
an electried jet of polymer solution. The amplitude of the lateral motion of the tip was
about 2 mm. The time from the beginning to the end of the pulse was about 45 ms. As a
result of the tip motion, a lateral displacement was imposed on the electried jet moving
toward the grounded counter electrode.
An impact of the polystyrene bar imposed a bending perturbation of the jet sketched in
Figure 2.13. It produces a bending pulse moving with the jet, as well as propagating and
widening relative to it, as shown in the images in Figure 2.14.
As is shown in Han et al. (2008), the widening of the pulse in time W between t = 0
and t =t, which can be measured using the snapshots in Figure 2.14, reveals the normal
axial stress xx in the jet to be
35 [25--62] 10.12.2013 4:43PM
2.3 Rheological constitutive equations
35
H (t, x)
Figure 2.13
0 ms
Sketch of a tapered straight jet, and the lateral bending perturbation of its axis. The detailed
shape of the perturbation pulse is not critical, as long as the locations of two or more
recognizable features, such as extreme points or zero crossings, can be followed. Reprinted
from Han et al. (2008), with permission from Elsevier.
0.5 ms
1 ms
Width
2.5 ms
Figure 2.14
3 ms
Position
3.5 ms
1.5 ms
2 ms
4 ms
4.5 ms
Propagation and widening of a single lateral displacement pulse. The interelectrode distance
L = 5.5 cm, potential difference of U = 3 kVand the electric current I = 100 nA. The jet was straight
before the lateral displacement pulse was applied. Reprinted from Han et al. (2008), with
permission from Elsevier.
xx
2t
2
2:5
It should be emphasized that the stress values measured in Han et al. (2008) (see
Section 5.4 in Chapter 5) were tremendous, which shows that polymer solutions are
capable of developing very signicant forces of elastic origin (see Sections 2.3 and 2.4)
in strong elongational ows.
2.3
Rheological constitutive equations

Below we discuss several generic phenomenological rheological constitutive equations
(RCEs) of incompressible non-Newtonian uids important for the description of elongational ows of polymeric liquids (solutions and melts) in ber-forming processes. The
rst indication that a liquid is non-Newtonian can be typically obtained from experiments
with steady-state simple shear ows. Such ows are traditionally used for measuring
36
36 [25--62] 10.12.2013 4:43PM
uid viscosity, which is dened as the ratio of the shear stress xy to an imposed shear
:
:
rate . For a number of liquids and gases the ratio xy = appears to be constant over a
:
wide range of . That permits to be considered as a basic rheological parameter of such
uids and these uids are called Newtonian. For them the NewtonStokes tensorial
rheological constitutive equation is taken in the form
2D
2:6
pI
2:7
where is the stress tensor, p is the pressure, I is the unit tensor, is the deviatoric
stress tensor, and D rv rvT =2 is the rate-of-stress tensor, with v being the tensor
gradient of velocity. The split of the total stress into the isotropic pressure-related part pI
and the deviatoric part is general for all uids, Newtonian and non-Newtonian. It should
be emphasized that for most uids, Newtonian and non-Newtonian (except the liquid
crystals), the total and deviatoric stress tensors and are symmetric (Loitsyanskii 1966).
The phenomenological RCE of a Newtonian liquid (Eq. 2.6) is constructed in such a
way that it reproduces the relation following from the corresponding simple shear
:
experiments, namely xy .
Many liquids, in particular, polymeric liquids do not follow Eq. (2.6) when strain rates
:
vary over a sufciently wide range. It was found that for such liquids the ratio xy = varies
:
with the shear rate , even in simple shear ow. The data for simple shear ows of
:
polymeric liquids can be approximated over a certain range of as
:
xy Kn
2:8
where two basic rheological parameters appear: the consistency and ow behavior
indexes, K and n, respectively.
:
Then the shear viscosity sh xy = is not a constant anymore, but rather depends on
the rate of strain
:
sh Kn1
2:9
Such uids are known as non-Newtonian uids. Following the phenomenological

approach, a tensorial RCE for the deviatoric stress tensor of such uids can be constructed
to reproduce Eq. (2.9) in simple shear ows, namely,
n1=2
2K 2tr D2
D
2:10
and the stress tensor given by Eq. (2.7) (Astarita and Marrucci 1974).
Equation (2.10) is called the Ostwaldde Waele RCE, or the power-law uid RCE.
In the particular case of pseudoplastic or shear-thinning uids, the exponent n < 1, and
:
the shear viscosity decreases as the shear rate increases, as Eq. (2.9) shows. In the
case of dilatants or shear-thickening uids the exponent n > 1, and the shear viscosity
:
increases as increases. The pseudoplastic behavior is much more frequent than the
dilatant one. If n = 1 and K = , RCE of the power law uids (Eq. 2.10) reduces to the
Newtonian RCE (Eq. 2.6).
37 [25--62] 10.12.2013 4:43PM
37
The tensorial RCEs for power-law uids can describe relatively accurately ow
curves measured in steady-state simple shear ow, which allows determination of
the basic rheological parameters K and n. However, beyond that, these equations are
quite limited at describing any other ow of the same uid with the same set of the
rheological parameters. For example, many polymer solutions and melts with exible
macromolecules demonstrate shear-thinning ow curves in steady-state simple shear
ow, i.e. seemingly correspond to the power-law RCE (Eq. 2.10) with n < 1. However,
in strong enough uniaxial elongational ows, which are the most relevant case for
the ber-forming processes, the effective elongational viscosity does not follow the
shear-thinning pattern and can increase with the stretching rate. The latter formally
corresponds to shear-thickening with n > 1 for the same uid. This contradiction
shows that the power-law RCE (Eq. 2.10) is basically incapable of describing the
rheological behavior of polymer solutions and melts. It should be emphasized that the
simple shear and uniaxial elongational ows represent the utmost limits of ow
kinematics. All the other possible ows can be represented locally by some mix of
these two limiting ow types. It is tempting to assume that if a uid can be described
by the same rheological constitutive equation with the same set of basic rheological
parameters in such distant ows as simple shear and uniaxial elongation ows,
this RCE can be considered as a universally valid one. Then most probably, such
an RCE can be successfully applied to any other ow of this uid with the rheological
parameters measured in any one of these two limiting cases. Newtonian uids
have passed this test. In particular, those of them that are viscous enough to be used
in uniaxial elongation experiments as free liquid threads follow the predictions of
the NewtonStokes RCE (Eq. 2.6) and reveal the same value of viscosity as that
measured in the simple shear experiments (McKinley and Tripathy 2000). There are
only a few recorded examples of uniform validity of the power-law RCE (Eq. 2.10)
with approximately the same rheological parameters K and n for the same uid
in simple shear and uniaxial elongational ows (Yarin 1993, Yarin et al. 2004,
Sinha-Ray et al. 2011). The inability of an RCE to describe the ow curves of the
same uid with the same set of rheological parameters in steady-state and uniaxial
elongation experiments shows that such an RCE is not a true RCE, even though it
seemingly performs well for simple shear.
Beginning with the early experiments (see, for example, Bird et al. 1987) it was
realized that polymeric liquids possess elasticity, i.e. represent themselves viscoelastic
liquids. Therefore, the inelastic power-law (Eq. 2.10) inherently misses the most important physical feature in the rheological behavior of polymeric liquids. The physical nature
of the elasticity of polymer solutions and melts is explained by their micromechanical
models developed in the framework of statistical physics of liquid polymers developed
by de Gennes (1979) and Doi and Edwards (1986), and is outlined in Section 2.4.
Since the elastic effects are related to a uid memory their description requires
kinematic tensors that operate between a reference and the current state of the material.
The only kinematic tensor used for Newtonian and power-law uids, the rate-of-strain
tensor D, is instantaneous, i.e. inappropriate for description of the memory effect, since
it is fully determined by the current state of the uid (i.e. by its current velocity eld).
38
38 [25--62] 10.12.2013 4:43PM
To introduce additional kinematic tensors to describe nonlinear kinematics of continua

possessing shape memory, consider an innitesimally small material element dX() in a
reference conguration at time moment . After deformation (not necessarily small)
during the time interval t , this material element in a new conguration at the moment t
becomes dX(t). The reference and current congurations are related via the gradient-ofdeformation tensor F(t), introduced as
dXt F t dX
2:11
The gradient-of-deformation tensor is asymmetric and as such cannot be used directly

to describe a symmetric deviatoric stress tensor . A symmetric kinematic tensor, the
so-called Green tensor B(t), can be formed using the gradient-of-deformation tensor and
its transpose as
B t F t F t T
2:12
It is easy to see that F B I.

The material time derivative of F(t) can be found using Eq. (2.11) (Astarita and
Marrucci 1974) as
DF t
rvt F t
Dt
2:13
with v(t) being the tensor gradient of velocity at time t. Then, Eqs. (2.12) and (2.13)
yield the material time derivative of the Green tensor in the following form
DB t
rvt B t B t rvt T
Dt
2:14
where D/Dt denotes material time differentiation.

As is shown in Astarita and Marrucci (1974), the Green tensor is associated with
transformation of lengths of linear material elements or areas of material platelets from a
reference to a current material conguration. Therefore, the elastic stresses can be
phenomenologically related to such transformations, and a nonlinear rheological constitutive equation of an elastic neo-Hookean body can be formulated as
GB t I
2:15
where G is the elastic shear modulus.

Consider, for example, an axisymmetric cylindrical thread of a neo-Hookean
material (Eq. 2.15), which undergoes uniaxial elongation along its axis Ox with the
:
rate of stretching . From Eq. (2.14) we nd the normal axial and radial components
Bxx and Byy of the Green tensor as
:
:
Bxx exp2t ; Byy expt
2:16
:
For small strains, xx t 551. Then, Eqs. (2.16) can be linearized and reduced to
the form
39 [25--62] 10.12.2013 4:43PM
Bxx 1 2xx ; Byy 1 xx
39
2:17
The corresponding axial and radial stresses are found using Eqs. (2.7), (2.15) and (2.17) as
xx p 2Gxx ; yy p Gxx
2:18
The unloaded surface of a thin thread (or a liquid threadline in most ber-forming
processes) implies that yy = p Gxx = 0 in the thread cross-section, and thus one
can nd pressure as p = Gxx; therefore, xx = 3Gxx. On the other hand, the elastic
shear modulus G = E/[2(1+ p)], where E is Youngs modulus and p is Poissons
ratio (Landau and Lifshitz 1970). For incompressible materials, e.g. polymeric liquids,
p = 1/2. Then, G = E/3 and xx = Exx which recovers Hookes law as a linear limit of the
nonlinear RCE of an elastic neo-Hookean body (Eq. 2.15).
A stretched thread of a neo-Hookean material, like an elastic spring, remembers forever
that it has been stretched. Such a thread has an innite memory and a stress xx should be
permanently applied at its ends, otherwise it will contract. Polymeric liquids, in general, and
their threads, jets, threadlines and bers, in particular, behave differently, namely, their
memory fades and they practically forget that they have been stretched. As a result, after a
while they cease to resist stretching (Astarita and Marrucci 1974, Bird et al. 1987, Yarin
1993). Therefore, after a certain period of time the stretching force can be removed and the
previously stretched polymeric thread still remains stretched indenitely.
The fading memory of polymeric solutions and melts can be accounted for in their
rheological constitutive equation phenomenologically. Namely, it is natural to assume
that any reference conguration of material at time is forgotten and becomes practically immaterial in the current state at time t, if the time span between and t is long
enough. This means that if the ratio (t )/, with being a physical parameter of the
material (its relaxation time) is large enough, the elastic stresses associated with the
deformation from one conguration to another are effectively forgotten. The simplest
expression would be the exponential fading memory, exp[(t )/]. Using it as a
weight function, a succession of elastic stresses corresponding to the neo-Hookean
body (Eq. 2.15) would be modied in the following form
t
t

G B t I expt =d
t
2:19
expt =d
where is a dummy variable.

Fluid viscosity can be introduced as the product = G, which shows that we are
inevitably dealing with a viscoelastic material instead of the original elastic neo-Hookean
one. Then, Eq. (2.19) can be rewritten as the following integral RCE:
t 2
t

B t I expt =d
2:20
40
40 [25--62] 10.12.2013 4:43PM
Here the integral is evaluated over the entire lifetime of the material from to the
current moment t.
The integral viscoelastic RCE (Eq. 2.20) describes the deformation history of individual
material elements. The deviatoric stress tensor in each of them is described as a function of
time (t), i.e. one is tracking the material elements in the Lagrangian sense, according to
uid mechanical terminology (Lamb 1959, Loitsyanskii 1966, Landau and Lifshitz 1987).
In most cases, tracking individual material elements is computationally time-consuming.
Moreover, the entire deformation history of each individual element expressed by its Green
tensor B(t) for t, should be permanently remembered and updated as time t. This
makes numerical solutions of ows of viscoelastic liquids in the framework of the integral
RCE prohibitively time- and computer-memory-consuming. However, for this particular
RCE (Eq. 2.20) a convenient alternative was established by Lodge (1964). By differentiating the integral RCE (Eq. 2.20) by time t and using Eq. (2.14) for the material time
derivative of the Green tensor, we obtain an equivalent differential RCE
Dt
2
t
rvt t t rvt T Dt
Dt
2:21
where all the terms on the right-hand side depend only on the current time moment t.
Since, as usual, the material time derivative can be expressed as the sum of the
temporal and spatial terms
Dt t
v r t
Dt
t
2:22
the differential RCEs (Eqs. 2.21 and 2.22) can be solved in the Eulerian sense (Lamb
1959, Loitsyanskii 1966, Landau and Lifshitz (1987) by direct time-marching without
following individual material elements and carrying over the entire history of their
deformation.
Note that the equivalence of the RCEs (Eqs. 2.20 and 2.21) is slightly disturbed by the
fact that when solving these equations, an initially fully unloaded state of material is
typically assumed, i.e. = 0 at t = 0 when the computations begin, whereas in the integral
Eq. (2.20) it might be that 0 at t = 0.
Equation (2.21) is called the upper-convected Maxwell model, UCM. The origin of
this term is mostly historic and is discussed in several monographs devoted to rheology,
e.g. Astarita and Marrucci (1974) and Bird et al. (1987).
The strength of any deformation, e.g. of the uniaxial elongation characteristic of
ber-forming processes, is compared to the inverse relaxation time 1, which is the
:
natural measure of the rate of stretching . As a result, the dimensionless Deborah number
De is introduced as a measure of the deformation strength in the competition between
stretching and relaxation of polymeric liquids (Reiner 1969)
:
De
2:23
The integral RCE (Eq. 2.20) can be rendered dimensionless to make transparent the
effect of the Deborah number in the limiting cases of very small and very large De. As the
41 [25--62] 10.12.2013 4:43PM
41
stress scale, it is natural to use G = /, and as the time scale for t, and s = t,
: 1
the characteristic hydrodynamic timescale represents a natural scale. Then, the
RCE (Eq. 2.20) takes the following dimensionless form:

1
t
B t I exps=Deds
2:24
De 0
In this expression, bars over dimensionless variables are omitted for brevity.
In the case of small Deborah numbers, De 0, the integral in Eq. (2.24) can be
expanded as a series in De using the Laplace asymptotic method (Nayfeh 1981). Then,
the two leading-order terms can be found in the following dimensional form:

DDt
T
t 2Dt 2
2:25
rvt Dt Dt rvt
Dt
The rst term on the right-hand side reproduces the viscous Newtonian RCE (Eq. 2.6),
whereas the second includes an additional elastic effect. The latter represents a small
correction, since the case of small Deborah numbers is considered. Flows with De << 1
are called weak ows. Viscoelastic uids which behave according to RCE (Eq. 2.25)
are called second-order uids (the name stems from the asymptotic expansion in De
used). In this RCE the elastic memory effects (the effect of the previous time) are
incorporated, in particular, through the material time derivative of the rate-of-strain
tensor D. This means that the deviatoric stress is, in fact, determined not only by the
instantaneous velocity eld, but also by its most recent history. In the other words, only
the short-term memory plays role at small values of the Deborah numbers when the rate
:
of relaxation 1 is very high compared to the rate of deformation , i.e. the ow is weak
compared to the stress relaxation.
In the opposite limit of extremely strong ows, when De , Eq. (2.24) reduces to
that of the neo-Hookean body (Eq. 2.15) with G = /. Overall, for extremely weak ows
with De 0, the integral RCE (Eq. 2.20), which is equivalent to the differential Upper
Convected Maxwell model (Eq. 2.21), reduces to Newtonian viscous uids, whereas for
extremely strong ows with De it yields purely elastic neo-Hookean materials. This
is a desirable from the phenomenological point of view, and the physically expected
situation: viscoelastic liquids approach purely Newtonian behavior in extremely weak
ows, and purely elastic behavior in extremely strong ows.
One may expect then, that the viscoelastic RCEs (Eq. 2.20), or equivalently (Eq. 2.21),
should behave well in the intermediary viscoelastic range of De, where both viscosity and
elasticity are equally important. Unfortunately, this is not necessarily true. For example,
polymeric liquids typically reveal shear-thinning behavior, which formally corresponds
to the power-law uids of Eq. (2.10) with n < 1. However, RCEs (Eqs. 2.20 and 2.21)
are incapable of reproducing this behavior in steady-state shear ows, which is, in fact,
the simplest case. They produce a purely Newtonian response with constant viscosity and
n = 1. In the framework of the phenomenological approach, the rheological parameters
or , or both of them, can be taken as functions of the second invariant of the deviatoric
stress tensor, or of one of the kinematic tensors (including D). Then, these modied
42
42 [25--62] 10.12.2013 4:43PM
RCEs can be t to the data from simple shear ow. Denitely, the clear physical spirit of
the previously outlined steps is lost in such a posteriori ne tuning. In addition, whether
this remedy helps in the description of the other types of ows in general remains an open
question until the corrected REC is tested.
It should be emphasized that the upper convected Maxwell model is supported by
micromechanical rheological models in ows with strong uniaxial elongation and can be
used in such ows without modications (Yarin 1993). In particular, it represents an
attractive model for ows associated with ber-forming processes, and will be employed
in Chapters 35 of the present book. Two examples of the performance of the upper
convected Maxwell model in ows with uniaxial elongation are considered below.
:
In the rst case, consider uniaxial stretching, with a constant rate , of a thin uniform
axisymmetric viscoelastic liquid thread in the direction of its axis Ox (Chang and Lodge
1972). Then, Eq. (2.21) yields the following ordinary differential equations describing
longitudinal and radial normal deviatoric stresses as

dxx
2 :
: 1
2 xx
2:26
dt

dyy
:
: 1
xx
2:27
dt
Assuming the unloaded state at t = 0 and solving these equations, we nd the normal
stresses in the following form:
:
h : 1
i
2
2 t 1 ; when 2: 1 0

xx :
e
2:28
2 1
:
2
:
t; when 2 1 0
xx
2:29
:
h
i
:
1 t

yy :
1

e
2:30
1
Equation (2.28) shows that ows with De > 1/2 are already so strong that the elastic
stresses accumulate, since relaxation cannot fully relax them, and the normal stress in the
axial direction xx grows exponentially with time. In the borderline case of De = 1/2
corresponding to Eq. (2.29), xx still grows, but linearly, whereas in the case of weak
ows with De < 1/2, xx stops growing and saturates. Equation (2.30) shows that the
radial stress yy will always be negligibly small compared to xx in ows with strong
uniaxial elongation.
Another example of the application of the upper-convected Maxwell model to uniaxial
elongation ow is related to threads of dilute and semi-dilute polymer solutions undergoing self-thinning under the action of surface tension, as depicted in Figure 2.2 (Stelter
et al. 2000). In this case the thread closely resembles a circular cylinder (see the sketches
in Figure 2.15, as well as the images in Figure 2.2). The capillary pressure in the
cylindrical thread is equal to /a, with being surface tension, and a being the crosssectional radius. In the end regions in Figure 2.15 the capillary pressure is of the order of
43 [25--62] 10.12.2013 4:43PM
(a)
43
(b)
x
End region
V
O
2a
Uniaxial
elongational flow
Liquid
thread
Plates
End regions
End region
R
Figure 2.15
Self-thinning thread of polymer solution. (a) The case of fully wettable plates. Tiwari et al.
(2009), with kind permission of Springer Science+Business Media. (b) The case of
poorly wettable plates. Reprinted with permission from Stelter et al. (2000). Copyright 2000,
The Society of Rheology.
/R, where R is the effective cross-sectional radius. Since R >> a, the capillary pressure
in the thread is much larger than that in the end regions. Then, the gradient of capillary
pressure drives the ow in two directions from the thread center to the end regions, as
shown by the arrows in Figure 2.15a.
Denote, as before, the axial coordinate x, and the radial one y. From Eq. (2.7) stresses
are found as
xx p xx ; yy p yy
2:31
At the free surface the radial stress yy = /a, i.e. to the negative capillary pressure,
where is the surface tension and a = a(t) is the cross-sectional radius. In a thin thread,
the stress yy is approximately equal to the value at the surface, and thus the second
Eq. (2.31) yields p = /a yy, whereas the rst Eq. (2.31) takes the form
xx xx yy
a
2:32
(cf. Eq. 1.8 in Chapter 1 where surface tension was neglected compared to the internal
stresses in polymer melts).
The rheological constitutive equation of the upper-convected Maxwell model (2.21)
yields
dxx
: xx
2xx
dt
2:33
44
44 [25--62] 10.12.2013 4:43PM
dyy
: yy
yy
dt
2:34
In these equations the effect of viscosity is neglected in anticipation of the dominant

elastic effects. The rate of elongation in the thread is related to the rate of thinning due to
mass conservation
2 da
:

a dt
2:35
Then, Eqs. (2.33) and (2.34) can be integrated as

xx
C1
expt=; yy C2 a2 expt=
a4
2:36
where C1 and C2 are the constants of integration. It can be seen that, as before, xx >> yy,
and the latter can be neglected.
In inertialess ow in a uniform thread, xx is constant along the thread and is approximately equal to /a (Stelter et al. 2000). Then, Eq. (2.32) and the rst Eq. (2.36) yield
C1
2
expt=
a4
a
2:37
The initial thread radius at t = 0 is a = a0, which determines the constant of integration as
C1 2a30 . Then, Eq. (2.37) yields

t
2:38
a a0 exp
3
i.e. is identical to Eq. (2.1), which is central to elongational rheology.
Using Eq. (2.38), we nd the rate of stretching and the elongational viscosity
:
el =a= as

2
3
t
:
exp
2:39
; el
3
2a0
3
In such ows elongational viscosity is exponentially growing in time, which determines
the long life of the thread in Figure 2.15.
Two additional comments on the phenomenological viscoelastic equation of the
upper-convected model are required:
(1) The phenomenological approach to viscoelasticity outlined in the present section
can be also criticized because its result is not unique. Indeed the Green tensor is not the
only tensor which is associated with transformation of lengths of linear material elements
or areas of material platelets from a reference to an actual material conguration. The
Cauchy tensor
C t F t T F t
2:40
45 [25--62] 10.12.2013 4:43PM
2.4 Micromechanics of polymer solutions
45
can also be associated with such transformations. Therefore, the entire theory of
viscoelasticity, in particular, Eqs. (2.15), (2.20), (2.21) and (2.25), can be recast using
the Cauchy tensor C(t), which will yield RCEs completely different from the previous
ones. They all will be formally admissible, and the phenomenological approach is
incapable, in principle, of discriminating between these two different sets of RCEs.
Only a supportive micromechanical model can substantiate a phenomenological RCE.
A better comparison with the available experimental data is also used as a factor in
favor of a certain phenomenological model. Several other kinematic tensors and their
invariants can be involved as building blocks of some other non-Newtonian RCEs and,
in particular, of viscoelastic RCEs. A review of these equations is available in the
rheological literature, e.g. in Astarita and Marrucci (1974), Bird et al. (1987), Larson
(1988) and Macosco (1994).
(2) The momentum balance equation of uid mechanics supplemented with the
viscoelastic RCE of the upper-convected Maxwell model (Eqs. 2.7 and 2.25) becomes
hyperbolic in ows with a strong elongational component as the Deborah number
De > 1/2 (cf. Eq. 2.28), as distinct from the elliptic equation for the viscous Newtonian
uids. This has dramatic implications on the numerical solutions for strong ows of
viscoelastic uids (Joseph 1990).
2.4
Micromechanics of polymer solutions and melts

Consider a micromechanical model of a exible polymer macromolecule as a random
walk of N freely jointed Kuhn segments of length b (see Section 2.1), where N >> 1.
The i-th Kuhn segment is given by vector ai = bu0i, where u0i is a randomly oriented
unit vector. The tail-to-head vector corresponding to the macromolecule is equal to
N
P
R
bu0i . Since each Kuhn segment experiences thermally driven Brownian motion
i1
in time t, the tail-to-head vector R depends on time.

The radius of gyration Rg is introduced using the scalar product R2g hR Ri. The
averaging denoted as is understood either as being conducted over different realizations of conguration of an individual macromolecule at different time moments, or as
being an instantaneous averaging over congurations of different noninteracting macromolecules (which corresponds to ensemble averaging in the case of dilute polymer
solution). The replacement of time average by ensemble average corresponds to the
ergodic hypothesis and does not change the result.
The radius of gyration Rg is given by
R2g b2
N X
N D
E
X
u0i u0j
2:41
i1 j1
For a Markovian process corresponding to the random walk, u0i . u0j = ij where ij is
Kroneckers delta. Therefore, Eq. (2.41) yields
46
46 [25--62] 10.12.2013 4:43PM
Rg bN1=2
2:42
Since N >> 1, the gyration radius is much less than the full length of macromolecule
L = bN. Random walk corresponds to the most chaotic macromolecular conguration,
i.e. represents the most probable shape without any external forces applied, thus it is
expected to correspond to an equilibrium. That means that the equilibrium conguration
of an isolated exible macromolecule is a coil of size comparable to Rg. Solvent ow in
the case of a polymer solution, or strong interactions with the other macromolecules in a
polymer melt can impose viscous drag or intermolecular forces, which are different at
different sections of such a macromolecular coil. Thus, macromolecular coils can be
stretched quite signicantly in the range from Rg to L, even though they are comprised of
absolutely nonstretchable Kuhn segments. Being isolated from any external forces,
macromolecules will tend to coil again, which is a thermodynamically determined
process dictated by the tendency to increase entropy, i.e. directly by the second law of
thermodynamics.
If one considers a random walk over a straight coordinate line Ox staring from x = 0,
with equal steps of length b randomly done back and forth, the probability density
function W(x, Nx) of the head being about a certain location x in Nx steps is given by
the following FokkerPlanck equation
W b2 2 W
2 x2
Nx
2:43
which is to be solved on x subject to the initial condition W(x,0) = (x),

where (x) is the delta function.
The solution reads

1
x2
Wx; Nx p exp 2
2:44
2b Nx
2b2 Nx
If random walk is conducted in three-dimensional space, with steps done only on the
Cartesian coordinate axes x, y and z, the probability density functions W(y,Ny) and
W(z,Nz) are given by expressions similar to Eq. (2.44), which holds for the part of
the steps done in the x-direction. Indeed, stepping from one axis to the other and choosing
the direction are done randomly, as a Markovian process, therefore there is no correlation
between partial random walks over different axes. The probability density function for
the walk head being close to a location x,y and z in space is given by the product W(x,Nx)
W(y,Ny)W(z,Nz) with Nx = Ny = Nz = N/3, and N being the total number of steps.
This yields

3=2
3
1
3r2
2:45
Wr; N
exp

2
2Nb2
b3 N3=2
where r2 = x2 + y2 + z2. This result can be obtained without stepping only on the
coordinate axes, which is considered a useful simplication.
47 [25--62] 10.12.2013 4:43PM
2.4 Micromechanics of polymer solutions
47
According to Boltzmanns entropy formula, the entropy S = kBlnW, where kB is

Boltzmanns constant. Therefore, from Eq. (2.45), we nd
Sr const
3kB r2
2Nb2
2:46
The Helmholtz free energy F = U TS, where U is the internal energy and T is
temperature. For polymer macromolecules the internal energy practically does not
depend on r, as distinct from crystalline elastic solids. Therefore, Eq. (2.46) yields
Fr const
3kB Tr2
2Nb2
2:47
Macromolecular coils tend to minimize their Helmholtz free energy, achieved by increasing
the entropy S, which is equivalent to decreasing r, according to Eq. (2.47). The latter
corresponds to shrinkage to the size of the macromolecular
coilRg. Note that the probability

density function (Eq. 2.45) can be seen as W~ exp 3r2 =2R2g , which shows that the most
probable macromolecular conformations in equilibrium are of the order of r ~ Rg.
The force f with which a macromolecule resists its stretching relative to the equilibrium
coil is f = F / r, which yields
f r
3kB T
r
Nb2
2:48
This force is linearly proportional to the stretching distance, i.e. closely resembles the
linear elasticity. However, as is explained above, this force has an entropic origin, and is
related to the tendency to minimize the internal energy. Therefore, it is called the entopicelasticity force (Yarin 1993). It should be emphasized that one can consider polymer
macromolecules in dilute or even semi-dilute polymer solutions as elastic dumb-bells,
i.e. two beads connected with an elastic spring that obeys Eq. (2.48). The beads are
subject to viscous drag forces that tend to stretch the spring, whereas the entropic-elastic
force tends to contract it. Calculating the contribution of the elastic forces from individual
dumb-bells to the macroscopic stresses in polymer solutions, one can arrive at the rheological constitutive equation of the upper-convected Maxwell model (Eq. 2.21) with the
viscosity and relaxation time expressed through the molecular characteristics
Nb2 c
Nb2
;
6
6kB T
2:49
In these expressions, c is the number of macromolecules per unit volume, and the friction
factor = 6sa, with s being the solvent viscosity and the bead size a = bN1/2.
The drawback of the above model of macromolecular chains is related to the fact that
random walk is assumed to be non-self-avoiding, i.e. can intersect itself. This is a
phantom-like view of a macromolecule, whereas in reality physical macromolecules
cannot intersect themselves, and there should be an excluded volume into which random
walk cannot enter. Then, macromolecular coils should be expected to swell compared
to the radius of gyration predicted in Eq. (2.42). An elegant description of such swelling
was given by Flory (1969). To nd the radius of gyration Rg corresponding to a swollen
48
48 [25--62] 10.12.2013 4:43PM
macromolecular chain, he considered the average Kuhn segment concentration inside

the coil cinternal N=R3g . Since they avoid each other, there should be total energy of
repulsion in the volume of a macromolecular coil R3g
D E
1
1
1
N2
Frepulsion kB TVb c2b R3g kB TVb c2internal R3g kB TVb 3
2
2
2
Rg
2:50
with Vb being of the order of the Kuhn segment volume b3, and cb being the real Kuhn
segment concentration in the coil.
On the other hand, according to Eq. (2.47), the coil also possesses elastic energy
effectively corresponding to the mutual attraction of the Kuhn segments
Fattraction
3kB TR2g
2Nb2
2:51
Then, the attractionrepulsion energy in the coil can be presented as

3R2g
Frepulsion Fattraction 1 N2
Vb 3
kB T
2 Rg 2Nb2
2:52
Omitting all the dimensionless factors of the order of one on the right-hand side, one nds
the radius of gyration of a swollen chain R = Rgs, as the radius that minimizes the energy
(Eq. 2.52). The result reads

1=5 3=5
Rgs Vb b2
N bN3=5
2:53
Comparing the Rgs of Eq. (2.53) with the Rg of Eq. (2.42), one can nd that the gyration
radius of a swollen chain can be signicantly larger in dilute polymer solutions than that of
a phantom chain, as N3/5 >> N1/2 when N >> 1. However, in concentrated polymer
solutions and melts multiple neighboring macromolecules can repel a given chain in all
possible directions, which effectively nullies the net repulsion. Then, chains in equilibrium are closely resembled by phantom chains with the gyration radius given by
Eq. (2.42) (de Gennes 1979, Doi and Edwards 1986). The micromechanical model
of polymer chains in concentrated systems can be recast into macroscopic rheological
constitutive equations. For weak deformations, where the relaxation mechanism is associated with reputational motion of macromolecules, this results in the DoiEdwards
constitutive equation, signicantly different from the upper-convected Maxwell model
(Eq. 2.21). On the other hand, for strong uniaxial elongation characteristics of berforming processes, the micromechanically based constitutive equation closely resembles
the upper-convected Maxwell model (Eq. 2.21) (Yarin 1990, 1993).
2.5
Solidification
Solidication is an important element in ber spinning, and for processes like spunbonding and meltblowing (Section 1.5 in Chapter 1 and Section 4.1 in Chapter 4, respectively)
49 [25--62] 10.12.2013 4:43PM
2.5 Solidification
49
a number of groups have investigated the role of solidication (Chen et al. 1983, Hajji et al.
1992, Kikutani et al. 1996, Beyreuther and Brnig 1997, Brnig et al. 1999). It was shown
that an increase in the air temperature delays solidication, resulting in a decrease in the
ber diameter and spinline stresses. Kikutani et al. (1996) studied sheathcore bicomponent bers, where polypropylene, PP (melt ow ratio, MFR 40), was used as the sheath and
polyethylene terephthalate, PET (inherent viscosity IV 0.65), was used as the core. The
predicted increase in the solidication stresses for PET and the decrease in the solidication
stresses for PP in PP/PET systems were in agreement with experimental data for molecular
orientation of these components in a bicomponent form. Similar results were obtained by
Mistra et al. (1993). In reality, when two polymers are co-spun together, it is likely that they
solidify at different temperatures. Therefore, while one component can solidify, the other
may be still in the semi-molten state. Thus, the elongational stresses experienced by the
latter would vanish instantly as further deformation of the spinline cannot occur beyond
this position, leading to a possible stress relaxation in the molten component (Fedorova
2006). The predictions of Kikutani et al. (1996) indeed show the existence of signicant
stress relaxation for PP after the solidication of PET. Yoshimura et al. (2003) experimentally obtained results similar to those of Kikutani et al. (1996). They demonstrated that
in bicomponent ber spinning, the component having higher viscosity would solidify
faster and had higher crystallinity and molecular orientation than the component with
lower viscosity. Choi and Kim (1999) found similar results.
Variation of the zero-shear viscosity of polymer melts with temperature T is
sufciently abrupt and can be mimicked by the Arrhenius law if the solidied material
is practically amorphous (Yarin 1992, 1993)

U
0 exp
2:54
RT
where 0 and U are the pre-exponential factor and the activation energy, respectively,
and R is the absolute gas constant.
A modied version of this expression was used by Yarin et al. (2010, 2011) and
Sinha-Ray et al. (2013b)

U 1
1
0 exp
2:55

R T T0
where 0 corresponds to temperature T0.
The corresponding dependence of the relaxation time on temperature reads
(Yarin et al. 2010, 2011, Sinha-Ray et al. 2013b)

U 1
1
T0
U 1
1
; 0 exp
2:56
0 exp

T
R T T0
R T T0
In the light of Eq. (2.54), solidication is understood to be an enormous increase
of viscosity, which effectively prevents further deformation. When the degree of crystallinity c (the material fraction that has crystallized) is signicant, Eq. (2.54) is
generalized and takes the following form (Yarin 1992, 1993):
50
50 [25--62] 10.12.2013 4:43PM

U
exp Bbc
0 exp
RT
2:57
where B and b are two dimensionless material constants.

The degree of crystallinity c is described in more detail in Section 2.6.
Solidication of polymer solutions results from solvent evaporation followed by
solute precipitation. To account for these phenomena, Yarin et al. (2001) used the
following expression inspired by Ziabicki (1976)
0 10Kc=c0
2:58
where K and m are two dimensionless material constants, and c and c0 are the current and
initial polymer concentrations in solution, respectively.
Due to solvent evaporation, c > c0, and the exponent in Eq. (2.58) increases signicantly, while the viscosity values become so large that in the high-concentration domain
the solution abruptly solidies and deformation is effectively ceased. In parallel, the
relaxation time increases in the course of solvent evaporation as
0 c0
2:59
where 0 is the relaxation time corresponding to the initial polymer concentration c0.
2.6
Crystallization
The degree of crystallinity is determined by Avramis law (Avrami 1939, Ziabicki 1976,
Jena and Chaturvedi 1992), which in its simplest form reads
c 1 exp Pts
2:60
where t is time and P and s are two dimensionless parameters. In the general case, the
expression for the exponent in Avramis law is determined by the rate of nucleation and
growth of crystal nuclei and should be calculated simultaneously with the temperature
eld (Yarin 1992, 1993).
The majority of polymers used to form macroscopic textile bers have enough time to
crystallize and are wound onto the bobbin at a certain degree of crystallinity. Polymer
macromolecules that were crystallized in the elongational ow in the threadline are
oriented not chaotically, but rather with a certain degree of orientation along the ber
axis. It should be emphasized that when macroscopic bers are formed using standard
melt, dry and wet spinning, it is practically impossible to reach signicant orientation
in the as-spun bers, and an additional stage, cold drawing in the solid state, is required.
In electrospun nanobers from polymer solutions, the degree of crystallinity is reduced
compared to macroscopic bers due to rapid evaporation of solvent and solidication
of polymer (Srikar et al. 2009).
51 [25--62] 10.12.2013 4:43PM
2.6 Crystallization
51
Still, it is possible to affect the crystalline structure, and thus the physical properties
(e.g. tensile strength) of the as-spun bers by purely mechanical or electromechanical
(hydrodynamical or electrohydrodynamical) means. If polymers are crystallizing under
conditions of strong elongational ow, where macromolecules experience a coilstretch
transition and are oriented along the threadline, such a crystallization process results not
in the ordinary lamellar crystals, but rather in axially oriented brillar crystals called
shish-kebab structures. Scanning electron microscopy images of such brillar crystals
were published by A.J. Pennings group (for example, see Chapter III in Ciferri and Ward
(1979) and Section 6.2 in Chapter 6 in the present book). This type of crystallization is
called ow-induced crystallization.
In shish-kebab structures the central brillar part represents sections of macromolecules that were crystallized fully stretched. A number of lamellar kebab-like crystals
that were crystallized in a fold-like manner are hafted on the brillar central part.
Tensile tests of the ordinary solid bers reveal, as a rule, relatively low strength at break
of about b = 0.2 GPa. On the other hand, bers containing axially oriented brillar shishkebab structures possess b = 35 GPa when they are formed from polymer solutions or
gels (Ciferri and Ward 1979, Smook and Pennings 1983). Note that Smook and Pennings
(1983) formed bers with axially oriented brillar shish-kebab structures using a method
that is very similar to dry spinning, discussed in Section 1.3 in Chapter 1. They explained
the formation of ultra-high strength bers by gel spinning in their work by stretching of
macromolecular chains due to direct entanglement with the neighboring chains.
Macroscopic ber forming under conditions of strong uniaxial stretching of macromolecular coils allows ow-induced crystallization of stretched chains and forms brillar
shish-kebab crystals, which possess a lower number of defects and a signicantly higher
tensile strength than the ordinary lamellar crystals. When bers are formed from polymer
solutions or gels, in addition to an ordinary ber spinning onto a winding bobbin, it is also
possible to employ the elongational ow domains of the TaylorCouette and Poiseuille
ows (see Chapter III in Ciferri and Ward 1979, and Section 6.2 in Chapter 6 in the
present monograph).
Scanning electron microscopy (SEM) of electrospun nanobers collected on a rotating
wheel by Zussman et al. (2003) uncovered brillar crystals similar to those found by
Pennings group in macroscopic bers. In particular, SEM revealed multiple necking
patterns at certain places along the nanobers, as shown in Figures. 2.16 and 2.17. In
Figure 2.16a, nanobers electrospun from 7% poly(ethylene oxide) (PEO) solution are
shown. The ber diameter contracted from 320 nm to nearly 60 nm in the neck regions.
The necks were not evenly spaced along the nanobers, and the distance between them
varied between 1 and 7 m. Figure 2.16b also shows nanobers electrospun from 7%
PEO solution, with thin brillar structures visible in the neck regions. The brils are only
weakly bonded together and in some places are separated. Figures 2.16c and d show
nanobers electrospun from 4% PEO solution, with the distance between the necks in the
range 46 m. In these cases brillar structures are also visible in the neck regions. This is
similar to the crazing often observed in cracking of polymers.
In Figure 2.17, several examples of brillar structures of PEO nanobers are shown in
more detail. In Figure 2.17a, a failed single nanober electrospun from 7% PEO solution,
52
52 [25--62] 10.12.2013 4:43PM
2 m
(a)
1 m
(b)
500 nm
(c)
Figure 2.16
500 nm
(d)
Electron micrographs of multiple neck formation in electrospun nanobers. (a) and (b) 7% PEO
electrospun nanobers, (c) and (d) 4% PEO electrospun nanobers. Reprinted with permission
from Zussman et al. (2003). Copyright 2003, AIP Publishing LLC.
is shown. Several brils are sticking out from the nanober cross-section. The estimated
diameter of the nanober is of about 550 nm, whereas a single bril diameter is about 20
nm. Figure 2.17a also seemingly shows that the nanober has a compact shell and a
rareed core with brils. Such a structure could appear as a result of a fast solvent
evaporation (see Chapter 5). In Figure 2.17b, a nanober electrospun from 7% PEO
solution is shown. The nanober diameter is in the range of 8001000 nm, whereas its
bril diameters are of about 80 nm. Figures 2.17c and d depict electrospun nanobers
formed from 4% PEO solution with clearly separated brils. In this case, the nanober
diameter is about 250 nm, which is smaller than that of the 7% PEO solution nanobers.
The bril diameters here are about 20 nm. Note that this width is reminiscent of the
brillar structure observed in Jaeger et al. (1996), Buchko et al. (1999) and Fong and
Reneker (1999). It is also very close to the thickness of the extended chain crystals of
PEO, which can be folded, revealed in Cheng et al. (1988). Small-angle X-ray scattering
could not reveal any additional information. Some of the nanobers appear to have large
beads, formed by capillary instability (see Section 3.3 in Chapter 3) or due to poor control
of the polymer ow rate during the electrospinning process.
The atomic force microscope (AFM) images in Zussman et al. (2003) unambiguously
showed genuine necking of electrospun nanober and excluded the possibility of
53 [25--62] 10.12.2013 4:43PM
2.6 Crystallization
500 nm
500 nm
(b)
(a)
500 nm
500 nm
(c)
Figure 2.17
53
(d)
Electron micrographs of brillar structures. (a) Failed 7% PEO electrospun nanober,

(b) brillar structure in 7% PEO electrospun nanober, (c) and (d) brillar structures in 4 %
PEO electrospun nanobers. Reprinted with permission from Zussman et al. (2003).
Copyright 2003, AIP Publishing LLC.
nanober disruption or thinning by the electron beams during SEM observations.

Figure 2.18 shows a wide-angle X-ray diffraction (WAXD) pattern obtained for a
microrope of PEO nanobers. The pattern reveals six diffraction arcs with a high degree
of orientation (texture). Analysis of this pattern points to a monoclinic crystalline
structure of PEO with helical molecular conformation (Takahashi and Tadokoro 1973).
Electrospinning generally results in birefringent bers, indicating that the molecular
chain axis was aligned parallel to the ber axis, which agrees with the prediction of a coilstretched transition of macromolecules during electrospinning (Reneker et al. 2000),
albeit the local crystalline perfection that produces sharp diffraction spots was not
completely developed in rapidly dried nanobers. As discussed in Reneker and Yarin
(2008), the crystallographic precision of the molecular positions and conformations for
such polymers as poly(meta-phenylene isophthalimide), poly(vinylidene uoride),
54 [25--62] 10.12.2013 4:43PM
54
Figure 2.18
Typical X-ray pattern of the oriented nanober rope made of 6% PEO (MW = 600 kDa)
aqueous solution with 40% ethanol. Reprinted with permission from Zussman et al. (2003).
Copyright [2003], AIP Publishing LLC.
polyaniline and poly(meta-phenylene isophthalamide) is improved by annealing at a

temperature lower than the crystallographic melting temperature. In particular, nanobers of poly(meta-phenylene isophthalamide), a high-performance, temperatureresistant polyamide, were electrospun at room temperature from a solution of the
polymer in N,N-dimethylacetamide by Liu et al. (2000). Optical microscopy with
polarized light showed that each nanober was strongly birefringent, indicating that
the polymer chains in the as-spun bers were aligned parallel to the axis. A wide-angle
X-ray diffraction pattern from a loosely twisted yarn of the as-spun nanobers is shown
in Figure 2.19. The diffuse equatorial diffraction spots indicate that the polymer crystallites
were small and imperfect, but aligned in the general direction of the axis of the yarn.
The wide-angle X-ray diffraction pattern of the poly(meta-phenylene isophalamide)
yarn after a 4 min annealing in air at 320 C is shown in Figure 2.20. As usual, sharp
diffraction spots are observed at angles that correspond to the Bragg diffraction from
crystallographic planes of triclinic crystals. Figure 2.20 shows that the thermal annealing
improved the irregular packing of the as-spun nanobers, presumably by allowing many
cooperative picometer-scale movements of molecular segments, which eliminated small
voids, established hydrogen bonds, adjusted conformations by small rotations around
bonds and thereby improved the t of molecules into the crystals, while preserving the
orientation of the polymer chains along the axis of the nanober (Reneker and Yarin 2008).
For some crystalline polymers that do not possess hydrogen bonds, sharp diffraction
peaks were observed immediately after nanober solidication without annealing.
In Section 4.8 of Chapter 4 the electrically assisted supersonic solution blowing of
nylon 6 nanobers in the range 2050 nm is discussed. Sinha-Ray et al. (2013a) showed
55 [25--62] 10.12.2013 4:43PM
2.6 Crystallization
55
Figure 2.19
A wide angle X-ray diffraction pattern from a yarn of twisted as-spun poly(meta-phenylene
isophthalamide) nanobers. Liu et al. (2000), Reneker and Yarin (2008). Reprinted from Reneker
and Yarin (2008), with permission from Elsevier.
Figure 2.20
X-ray diffraction pattern of annealed poly(meta-phenylene isophalamide) nanobers.

Liu et al. (2000), Reneker and Yarin (2008). Reprinted from Reneker and Yarin (2008), with
56
56 [25--62] 10.12.2013 4:43PM
that these nanobers possess a peculiar novel crystalline phase of nylon 6, which is
distinct from those previously observed. The results of the X-ray analysis and modeling
(Holmes et al. 1955, Li and Goddard 2002) revealed two different stable crystal structures of nylon 6: - and -phases. The monoclinic -phase, with hydrogen-bonded
anti-parallel fully extended chains, appears when the nylon 6 melt gradually cools. On
the other hand, the -phase, with hydrogen-bonded parallel pleated chains, appears when
the nylon 6 melt is rapidly quenched. The -phase is thermodynamically preferable,
albeit the formation of the -phase can be facilitated more easily than the -phase by an
appropriate process route. Two intermediate crystal forms, - and -phases, emerge as
the -phase transforms into the -phase and vice versa. The - and -phases are unstable
and have either different H-bond patterns or different chain conformations from the
stable - and -phases.
TEM images of a single 50 nm nanober from the experiments of Sinha-Ray et al.
(2013a) are shown in Figures 2.21a and c, and the corresponding X-ray diffraction patterns
in Figures 2.21b and d, respectively. The diffraction patterns reveal several important
features: a single sharp ring indicating no -phase in the nanober, which corresponds to
doublet rings indexed as (200) and (002)/(202) reections (Holmes et al. 1955). The
diffraction patterns in Figures 2.21b and d do not correspond to the -phase, which
would be indicated by doublet rings indexed as (020) and (001) (Holmes et al. 1955).
(a)
(b)
20 nm
30 cm
(c)
(d)
20 nm
Figure 2.21
30 cm
TEM images of two different nanobers are shown in panels (a) and (c), and the corresponding
small-area electron diffraction (SAED) pattern in panels (b) and (d). In panels (b) and (d), the
arrows point to the brightest spot on the rings. In panels (b) and (d), the written value (30 cm)
indicates the camera distance. Sinha-Ray et al. (2013). Reproduced with permission from The
Royal Society of Chemistry.
57 [25--62] 10.12.2013 4:43PM
2.6 Crystallization
57
For comparison, in electrospun nylon 6 bers, equatorial reections assigned to the

-phase lattice planes associated with (200) and (011) reection planes and meridional
reections associated with the (020) reection plane were observed (Zhao et al. 2007),
as distinct from the single ring in Figures 2.21b and d. The existence of a single ring
shows that these nanobers did not contain the -phase observed in nylon oligomers
(Dersch et al. 2003) or the -phase (Sikorski and Atkins 2001). From Figures 2.21b and
d, the d-spacing is 0.156 nm, which is much smaller than those for the -phase, which
are 0.432 nm and 0.385 nm, and that for the -phase, which is 0.413 nm (Russell and
Beaumont 1980). Therefore, the supersonically solution-blown nanobers do not
consist of the -, -, -, and -phases. CC bonds of length 0.154 nm are located in
the sheet planes, whereas the d-spacing of 0.156 nm signies the intersheet distance.
This corresponds to the fact that there is no ordering in the longitudinal direction,
because otherwise a sharp ring corresponding to a d-spacing larger than the CC bond
length would have appeared in Figures 2.21b and d.
The existence of a sharp ring reveals that the ber in Figure 2.21 is not amorphous; if
it were, a diffuse ring would appear. It is still possible that the ber is comprised of the
intermediary metastable -phase (Li and Goddard 2002, Zhao et al. 2007). The typical
diffraction pattern (Figures 2.21b and d) shows that the ring is the sharpest in the
direction approximately perpendicular to the direction of ber orientation, which
means that the macromolecular sheets are aligned along the ber.
Sinha-Ray et al. (2013a) applied differential scanning calorimetry (DSC) to the supersonically solution-blown nanobers to compare them with the parent polymer pellet
(Figures 2.22ad). The thermograms of the original pellet and the resulting nanobers
were signicantly different. It can be seen that during the rst stage of heating
(Figure 2.22b), the nanobers undergo glass transition in the temperature range
45125 C, whereas the pellet undergoes glass transition in the temperature range 165
205 C. The fact that the nanobers have a signicantly lower glass transition temperature
range than the corresponding bulk polymer is veried by Srikar et al. (2009). For the -phase,
an exothermal recrystallisation peak at 55 C was reported for nylon 6 in Penel-Pieron et al.
(2001), where cast lms from the -crystal-dominated samples were studied. The absence of
any exothermal peak in Figure 2.22 shows that neither the parent pellet nor the nanobers are
comprised of the -phase in the present case. The absence of any exothermal peak for the
nanobers at 55 C and the appearance of an endothermal peak instead are very instructive
and unusual. In Figure 2.22b, a broad exothermal rise is seen at 125185 C for nanober
samples that underwent the rst cycle of heating, originating from a complete or partial
restructuring of its phase into a more stable phase (most probably, the -phase). Indeed, the
DSC thermogram of nanobers that underwent the second heating cycle (Figure 2.22c)
shows features visible in both DSC thermograms in panels (a) and (b), i.e. already a mixture
of two phases, the novel - and -phases. The co-existence of these two crystalline phases can
also be traced in the lower temperature range (42125 C; Figure 2.22d), where the DSC
thermogram of the nanobers that underwent the second heating cycle lies in between the
data for the pellet and for the nanobers that underwent the rst heating cycle. In summary,
the results shown in Figure 2.22 exclude the last previously known -phase in nanobers.
58
58 [25--62] 10.12.2013 4:43PM
170 180 190 200

Temperature (C)
250
100
150
205C
200
227C
250
1 W/g
100
150
200
Temperature (C)
250
100
80
Temperature (C)
120
60
(d)
14 880
45 500
185C 227C
47C
42
Temperature (C)
(c)
50
(b)
Heat flow (W/g)
1 W/g
Glass transition zone
50
Exothermic
region
125C
227C
200
100
150
Temperature (C)
50
(a)
Glass
transition
zone
0.15 W/g
160
Heat flow (W/g)
Heat flow (W/g)
0.04 W/g
Heat flow (W/g)
1 W/g
Heat flow (W/g)
2924.8
2903.2
3225.3
Intensity (a.u.)
Intensity (a.u.)
39 000
32 500
26 000
3297.1
3226.9
19 500
2 800
(e)
Figure 2.22
2 940
3 080
3 220
Wave number (cm1)
14 570
14 260
13 950
13 640
3 360
2 760
(f)
3 000 3 120
2 880
Wave number (cm1)
3 240
DSC thermograms of (a) nylon 6 pellet, (b) nanober mat undergoing the 1st cycle of heating,
(c) nanober mat undergoing the 2nd cycle of heating, and (d) zoomed-in view of panels (a)-(c) at a
low temperature range with the line colors corresponding to the respective DSC plots in panels
(a)-(c). In panel (a) in the inset, a zoomed-in view of the glass transition zone is shown. After the 1st
heating cycle, the nanober sample was cooled down and then subjected to the 2nd heating cycle.
Panels (e) and (f) show the micro-Raman spectroscopy results of the nylon 6 pellet and the 50 nm
nanober mat, respectively. The blue values indicate the wavenumbers corresponding to CH2
stretching and the green values indicate the wavenumber corresponding to NH stretching.
Sinha-Ray et al. (2013). Reproduced with permission from The Royal Society of Chemistry.
59 [25--62] 10.12.2013 4:43PM
2.7 References
59
Therefore, a novel -phase of nylon 6 was formed in the supersonically solution-blown 20

50 nm nanobers in Sinha-Ray et al. (2013a).
Figures 2.22ac also show that both nanobers and pellet melt at almost the same
temperature of 227 C. It is known that bigger electrospun nanobers have a signicantly
lower melt temperature than the parent bulk polymers (Srikar et al. 2009). Then, the data
in Figure 2.22 imply that the crystal phase in the supersonically solution-blown 2050 nm
nanobers has a melt temperature signicantly higher than the -phase in the pellet
(Tm = 227 C).
The nature of H-bonds in the crystal phases was explored in Sinha-Ray et al. (2013a)
by micro-Raman spectroscopy of a pellet (Figure 2.22e) and nanobers (Figure 2.22f).
For nylon 6, the range 28532920 cm1 corresponds to CH2 stretching, and 3300 cm1
corresponds to N-H stretching (Maddams and Royaud 1991). The peak corresponding
to CH2 stretching shifted from 2903.2 cm1 for the pellet to 2924.8 cm1 for the
nanobers. In Figure 2.22e, in addition to the peak at 3297.1 cm1 there is another
small peak at 3226.3 cm1. This additional peak corresponds to a shift in the N-H
stretch caused by pressing at the manufacturing stage (Reynolds and Sternstein 1964).
However, in the nanobers, the N-H stretch peak shifts to 3225.3 cm1 (Figure 2.22f).
The ber diameter is reduced on the millisecond range from ~1 mm at the needle to a
nal value of about 50 nm at a stretching rate of the order of 1010 s1 (see Section 4.8
in Chapter 4) and by squeezing of macromolecular sheets in the radial direction.
The distance between the electronegative atoms N and O in the bond structure of
C = O H N decreases, and the O H bond becomes stronger. This weakens the
NH bond, as well as shifts the NH stretch toward a lower wavenumber. Then, a CH2
group of one chain is forced to move closer to a CH2 group of a neighboring chain,
with the corresponding decrease in the mean CH bond distance and the increase in
the HCH angle. The increased angle leads to a shift of the CH2 stretching mode
and a decrease in the axisymmetric stretching mode. Figures 2.22b and d reveal that the
d-spacing is signicantly lower than those of the standard crystal phases of nylon 6.
In addition to the hydrogen bonds within the same macromolecular sheet, intersheet
hydrogen bonds can also form in the present case of the supersonically solution-blown
2050 nm bers. The existence of stronger O H bonds supports this conjecture. The
stronger O H bond should manifest as unusually stiff nanobers, which is, indeed,
supported by the results discussed in Section 7.1 in Chapter 7.
2.7
References
Astarita, G., Marrucci, G., 1974. Principles of Non-Newtonian Fluid Mechanics. McGraw-Hill,
New York.
Avrami, M., 1939. Kinetics of phase change. I. General theory. J Chem. Phys. 7, 11031112.
Beyreuther, R., Brnig H., 1997. High lament velocities in the underpressure spunbonding
nonwoven process. Int. Fiber J. December, 129134.
Bird, R. B., Curtiss, C. F., Armstrong, R. C., Hassager, O., 1987. Dynamics of Polymeric Liquids,
John Wiley & Sons, New York.
60
60 [25--62] 10.12.2013 4:43PM
Brnig, H., Beyreuther, R., Hoffman, H., 1999. The inuence of quench air on ber formation and
properties in the melt spinning process. Int. Fiber J. April, 104107.
Buchko, C. J., Chen, L. C., Shen, Y., Martin, D. C., 1999. Processing and microstructural characterization of porous biocompatible protein polymer thin lms Polymer 40, 73977407.
Chang, H., Lodge, A. S., 1972. Comparison of rubberlike-liquid theory with stress-growth data for
elongation of a low-density branched polyethylene melt. Rheol. Acta 11, 127129.
Chen, C. H., White, J. L., Spruiell, J. E., Goswami, B. C., 1983. Dynamics, air drag and orientation
development in the spunbonding process for nonwoven fabric. Tex. Res. J. January, 4451.
Cheng, S. Z. D., Bu, H. S., Wunderlich, B., 1988. Double lamellae of low-molecular-mass fractions
of poly(ethylene oxide) crystallized from the melt. Polymer 29, 579583.
Choi, Y. B., Kim, S. Y., 1999. Effects of interface on the dynamic mechanical properties of PET/
Nylon 6 bicomponent bers. J. Appl. Polym. Sci. 74, 20832093 (1999).
Ciferri, A., Ward, I. M., 1979. Ultra-high Modulus Polymers. Appl. Sci. Publ., London.
de Gennes, P. G., 1979. Scaling Concepts in Polymer Physics. Ithaca, Cornell Univ. Press.
Dersch, R., Liu, T., Schaper, A. K., Greiner, A., Wendorff, J. H., 2003. Electrospun nanobers: internal structure and intrinsic orientation. J. Polym. Sci.: Part A: Polym. Chem. 41,
545553.
Doi, M., Edwards, S. F., 1986. The Theory of Polymer Dynamics. Clarendon Press, Oxford.
Fedorova, N., 2006. Investigation of the utility of islands in the sea bicomponent ber technology
in the spunbond process. PhD Thesis, NC State University,.
Flory, P., 1969. Statistics of Chain Molecules. Interscience Publishers, New York.
Fong, H., Reneker, D. H., 1999. Elastomeric nanobers of styrene-butadiene-styrene triblock
copolymer. J. Polym. Sci., Polym. Phys. Ed. 37, 34883493.
Hajji, B., Spruiell, J. E., Lu, F. M., Malkan, S., Richardson, G. C., 1992. Modeling of the Reicol
spunbonding process. INDA Journal of Nonwovens Reasearch 4, 1621.
Han, T., Yarin, A. L., Reneker, D. H., 2008. Viscoelastic electrospun jets: initial stresses and
elongational rheometry. Polymer 49, 16511658.
Holmes, D. R., Bunn, C. W., Smith, D. J., 1955. The crystal structure of polycaproamide: Nylon 6.
J. Polym. Sci. 17, 159177.
Jaeger, R., Schonherr, H., Vancso, G. J., 1996. Chain packing in electro-spun poly(ethylene oxide)
visualized by atomic force microscopy. Macromolecules 29, 76347636.
Jena, A. K., Chaturvedi, M. C., 1992. Phase Transformations in Materials. Prentice Hall,
Englewood Cliffs.
Joseph, D. D., 1990. Fluid Dynamics of Viscoelastic Liquids. Springer, New York.
Kikutani, T., Radhakrishnan, J., Arikawa, S., Takaku, A., Okui, N., Jin, X., Niwa, F., Kudo, Y.,
1996. High-speed melt spinning of bicomponent bers; mechanism of ber structure development in poly(ethylene terephthalate)/polypropylene system. J. Appl. Polym. Sci. 62,
19131924.
Lamb, H., 1959. Hydrodynamics. Cambridge University Press, Cambridge.
Landau, L. D., Lifshitz, E. M., 1970. Theory of Elasticity. Pergamon Press, Oxford.
Larson, R., 1988. Constitutive Equations for Polymer Melts and Solutions. Buttersworths, New York.
Li, Y., Goddard III W. A., 2002. Nylon 6 crystal structures, folds, and lamellae from theory.
Macromolecules 35, 84408455.
Liu, W., Wu, Z., Reneker, D. H., 2000. Structure and morphology of poly(metaphenylene
isophthalamide) nanobers produced by electrospinning. Polymer Reprints 41 11931194.
Lodge, A., 1964. Elastic Liquids. Academic Press, London.
61 [25--62] 10.12.2013 4:43PM
2.7 References
61
Moscow, 1970.
Macosco, C. W. 1994. Rheology Principles, Measurements and Applications. Wiley and Sons,
New York.
Maddams, W. F., Royaud, I. A. M., 1991. The application of Fourier transform Raman spectroscopy to the identication and characterisation of polyamidesII. Double- number. Spectrochim.
Acta: Mol. Spectr. 47A, 13271333.
McKinley, G. H., Tripathi, A., 2000. How to extract the Newtonian viscosity from capillary
breakup measurements in a lament rheometer. J. Rheol. 44, 653669.
Mistra, S., Spruiell, J. E., Richeson, G. C., 1993. Investigation of the spunbonding process via
mathematical modeling. INDA J. Nonwovens Res. 5, 1319.
Nayfeh, A. H., 1981. Introduction to Perturbation Techniques. John Wiley & Sons, New York.
Penel-Pieron, L., Depecker, C., Seguela, R., Lefebvre, J. M., 2001. Structural and mechanical
behavior of Nylon 6 lms. Part 1. Identication and stability of the crystalline phases. J. Polym.
Sci.: Part B: Polym. Phys. 39, 484495.
Reiner, M., 1969. Deformation, Strain and Flow: An Elementary Introduction to Rheology. HK
Lewis, London.
Reneker, D. H., Yarin, A. L., 2008. Electrospinning jets and polymer nanobers. Polymer 49,
23872425.
Reneker, D. H., Yarin, A. L., Zussman, E., Xu, H., 2007. Electrospinning of nanobers from
polymer solutions and melts. Adv. Appl. Mech. 41, 43195.
Reynolds, J., Sternstein, S. S., 1964. Effect of pressure on the infrared spectra of some hydrogen
bonded solids. J. Chem. Phys. 41, 4751.
Russell, D. P., Beaumont, P. W. R., 1980. Structure and properties of injection-molded nylon-6:
Part 1. Structure and morphology of nylon 6. J. Mater. Sci. 15, 197207.
Sikorski, P., Atkins, E. D. T., 2001. The three-dimensional structure of monodisperse 5-amide
nylon 6 crystals in the lambda-phase. Macromolecules 34, 47884794.
Sinha-Ray, S., Lee, M. W., Sinha-Ray, S., An, S., Pourdeyhimi, B., Yoon, S. S., Yarin A. L., 2013a.
Supersonic nanoblowing: A new ulta-stiff phase of nylon 6 in 2050 nm connement. J. Mater.
Chem. C 1, 34913498.
Sinha-Ray, S., Srikar, R., Lee, C. C., Li, A., Yarin, A. L., 2011. Shear and elongational rheology of
gypsum slurries. Applied Rheology 21, 63071.
Sinha-Ray, S., Yarin, A. L., Pourdeyhimi, B., 2013b. Prediction of angular and mass distribution in
meltblown polymer laydown. Polymer 54, 860872.
Smook, J., Pennings, A. J., 1983. Preparation of ultra-high strength polyethylene bers by
gel-spinning/hot-drawing at high spinning rates. Polym. Bull. 9, 7580.
Srikar, R., Gambaryan-Roisman, T., Steffes, C., Stephan, P., Tropea, C., Yarin, A. L., 2009.
Nanober coating of surfaces for intensication of spray or drop impact cooling. Int. J. Heat
and Mass Transf. 52, 58145826.
Stelter, M., Wunderlich, J., Rath, S. K., Brenn, G., Yarin, A. L., Singh, R. P., Durst, F., 1999.
Shear and extensional investigations in solutions of grafted/ungrafted amylopectin and
polyacrylamide. J. Appl. Polym. Sci. 74, 27732782.
Stelter, M., Brenn, G., Yarin, A. L., Singh, R. P., Durst, F., 2000. Validation and application of a
novel elongational device for polymer solutions. J. Rheol. 44, 595616.
62
62 [25--62] 10.12.2013 4:43PM
Stelter, M, Brenn, G., Yarin, A. L., Singh, R. P., Durst, F., 2002. Investigation of the elongational
behavior of polymer solutions by means of an elongational rheometer. J. Rheol. 46, 507527.
Takahashi, Y., Tadokoro, H., 1973. Structural studies of polyethers (-(CH2)M-O-)N.10. crystalstructure of poly(ethylene oxide) Macromolecules 6, 672675.
Theron, S. A., Zussman, E., Yarin, A. L., 2004. Experimental investigation of the governing
parameters in the electrospinning of polymer solutions. Polymer 45, 20172030.
Tiwari, M. K., Bazilevsky, A. V., Yarin, A. L., Megaridis, C. M., 2009 Elongational and shear
rheology of carbon nanotube suspensions-uids with yield stress. Rheologica Acta 48, 597609.
Wunderlich, T., Stelter, M., Tripathy, T., Nayak, B. R., Brenn, G., Yarin, A. L., Singh R. P.,
Brunn, P. O., Durst, F., 2000. Shear and extensional rheological investigations in solutions of
grafted and ungrafted polysaccharides. J. Appl. Polym. Sci. 77, 3200 3209.
Xu, H., Yarin, A. L., Reneker, D. H., 2003. Characterization of uid ow in jets during electrospinning. Polymer Preprints 44, 5152.
Yarin, A. L., 1990. Strong ows of polymeric liquids: 1. Rheological behavior. J. Non- -Newton.
Fluid Mech. 37, 113 138.
Yarin, A. L., 1992. Flow-induced on-line crystallization of rodlike molecules in bre spinning.
J. Applied Polymer Sci. 46, 873878.
Yarin, A. L., 1993. Free Liquid Jets and Films: Hydrodynamics and Rheology. Longman
Scientic & Technical and John Wiley & Sons, Harlow, New York.
Yarin, A. L., Koombhongse, S., Reneker, D. H., 2001. Bending instability in electrospinning of
nanobers. J. Appl. Phys. 89, 30183026.
Yarin, A. L., Sinha-Ray, S., Pourdeyhimi, B., 2010. Meltblowing: II-Linear and nonlinear waves on
viscoelastic polymer jets. J. Appl. Phys. 108, 034913.
Yarin, A. L., Sinha-Ray, S., Pourdeyhimi, B., 2011. Meltblowing: Multiple jets and ber-size
distribution and lay-down patterns. Polymer 52, 29292938.
Yarin, A. L., Zussman, E., Theron, A., Rahimi, S., Sobe, Z., Hassan, D., 2004. Elongational
behavior of gelled propellant simulants. J. Rheol. 48, 101116.
Yoshimura, M., Iohara, K., Nagai, H., Takahashi, T., Koyama, K., 2003. Structure formation of
blend and sheath/core conjugated bers in high-speed spinning of PET, including a small amount
of PMMA. J. Macr. Sci.: Part BPhysics B42, 325339.
Zhao, Z., Zheng, W., Tian, H., Yu, W., Han, D., Li, B., 2007. Crystallization behaviors of
secondarily quenched nylon 6. Mater. Lett. 61, 925930.
Zussman, E., Rittel, D., Yarin, A. L., 2003. Failure modes of electrospun nanobers. Appl. Phys.
Lett. 82, 39583960.
63 [63--88] 10.12.2013 5:06PM
General quasi-one-dimensional
equations of dynamics of free liquid
jets, capillary and bending instability
This chapter introduces the fundamental general equations of the dynamics of free liquid
jets in Sections 3.1 and 3.2. The applications of these equations encompass practically
all ber-forming processes from melt spinning, as was sketched out in Section 1.2
in Chapter 2 to melt- and solution blowing, and electrospinning of polymer microand nanobers discussed in detail in Chapters 4 and 5, or drawing of optical bers
outlined in Section 6.5 in Chapter 6. They also form the framework for description of
several types of instabilities characteristic of the hydrodynamics of free liquid jets in
general and of ber-forming processes in particular. These include capillary instability
(Section 3.3), aerodynamically driven bending instability (Section 3.4) and buckling of
liquid impinging onto a wall (Section 3.5).
3.1
Mass, momentum and moment-of-momentum balance equations

The dynamics of free liquid jets moving in air, which are characteristic of ber-forming
processes, involve growth of various perturbations. The most notable are driven by surface
tension and the dynamic interaction with the surrounding air, as well as electrically
driven effects. Theoretical/numerical description of the jet evolution in general, and of
perturbed jets in particular, is hindered by the fact that such problems typically involve a
three-dimensional, time-dependent evolution of ows with free surfaces, the locations in
time of which should also be established. Solving such problems in the framework of
the rigorous equations of uid mechanics, say the NavierStokes equations, in most
cases would be prohibitively time-consuming, even using super-computers. Additional
complicating factors arise due to the rheological complexity of polymer solutions used in
ber forming, as well as the temperature-dependent variation of material properties in
nonisothermal situations. Accounting for all these factors together in the framework of the
rigorous equations of non-Newtonian uid mechanics would be tremendously difcult.
However, these difculties can be relatively easily overcome in the framework of a quasione-dimensional description of liquid motion in the bending jets. In the works of Entov and
Yarin (1980, 1984a) and Yarin (1983, 1993) the general quasi-one-dimensional equations
of the straight and bending jets were derived from the integral balances of mass,
momentum and moment of momentum, as well as by averaging the three-dimensional
equations of hydrodynamics over the jet cross-section. In the quasi-one-dimensional
approximation, a jet is considered as a directed continuum, i.e. a long linear object
64
64 [63--88] 10.12.2013 5:06PM
3 General quasi-one-dimensional equations
with no lateral dimensions, which is characterized by a number of such integral parameters

as the cross-sectional radius, ow rate, etc. distributed along the jet axis. This approach is
kindred to the hydraulic theory of ows in long pipes (especially in pipes with exible
elastic walls), and its natural range of validity is linked to ows slowly varying along the jet
axis (Loitsyanskii 1966). The relations between three-dimensional uid mechanics and the
quasi-one-dimensional approximation are also similar to those between the theory of
elasticity and the theory of bar bending (Landau and Lifshitz 1970), or between realism
and surrealism in art. It is always benecial to sacrice the excessive details, and to build on
a simplied vision of a specic object (in the present case, an elongated thin jet with no
signicant tractions at the free surface), which is the best way to achieve the simplest
physically relevant understanding and description of such an object.
The quasi-one-dimensional continuity and momentum equations, as well as the
moment of momentum equation for the general case of motion of a thin liquid jet in air
derived by Yarin et al. read
f Wf
0;
t
s
f a2
f V Wf V 1
P Q Ff q
t
s
s
j2 kU j1 j1 WK1 j3 V
t
s
1 M
Q m kj1 F
3:1
3:2
3:3
These equations are a close hydrodynamic analog of the equations of Euler

Bernoulli beam theory (Landau and Lifshitz 1970). They are related to the threedimensional equations of hydrodynamics (the NavierStokes equations in the case
of viscous Newtonian liquids) exactly as the equations of the EulerBernoulli beam
theory are related to the three-dimensional equations of the theory of elasticity. The
derivation of such equations in the simplest case of a liquid threadline in melt
spinning is sketched in Section 1.2 in Chapter 1. In the framework of quasi-onedimensional equations, the hydrodynamics of thin liquid jets is reduced to nding
the evolution in time t over a spatial coordinate s of the integral parameters the
cross-sectional area f, the velocity of the center of mass of a liquid cross-section V
and the angular velocity of this cross-section . Equation (3.1) is the mass balance
(the continuity equation). Equation (3.2), the momentum equation, represents the
balance of forces acting on a jet element, namely the inertial, internal and
external forces. Equation (3.3) is the moment-of-momentum balance, in particular,
its left-hand side expresses the rate of change of the moment of the inertial forces.
The following notation is used in Eqs. (3.1)(3.3):
K x xdS kj1 V; K1 x xdS

3:4
D
65 [63--88] 10.12.2013 5:06PM
3.2 Closure relations
65
j1
yxdS; j2
D
x x x dS
3:5

R
j2 x x x dS;
s
D

R
1 1 V
kVn
; WV U ;
t
2 s
3:6
3:7
Here R and x denote the position vectors of the jet axis, and of a point in the jet
cross-section, respectively (x is reckoned from the center of mass of a jet crosssection D(s, t) and belongs to its plane); F is the body force per jet element of unit
volume; q and m are the distributed force and moment of force imposed on the jet
by the environment or self-induced, for example, electric forces; Q is the shearing
force acting in the jet cross-section, which is determined using Eq. (3.3); k is the jet
axis curvature and is liquid density. Here and below, n, b and denote the
principal unit normal, unit binormal and unit tangent to the jet axis, respectively.
The angular velocity of the trihedron n, b and associated with the jet axis is
denoted . Subscripts n, b and denote projections on the principal normal,
binormal and tangent to the jet axis, respectively and y is the coordinate reckoned
along the principal normal.
3.2
Closure relations
In the case of Newtonian viscous liquids the magnitude of the longitudinal force in the jet
cross-section P and the moment of the internal stresses M are related to the kinematic
parameters in a cross-section of radius a by the following expressions:

1 V
P 3
kVn G f P
3:8
s
"
#1=2 "
#3=2

1
1 a 2
1 a 2
1 1 a
G 1 2
1 2
a
s s
s
s
#1=2
1 a 2
P 2a 1 2
s
3:9
"
1 n
k b
Mn 3I
s
3:10

3:11
66
66 [63--88] 10.12.2013 5:06PM

1 b
3 k V 3 2
n
k Vn
Mb 3I
s
2 s
2
"
2 #3=2
1
1 a
k 1 2
a
s

2 k Vb
a4
M I
kVn kn ; I
4
s
s
3:12
3:13
Here is the geometric torsion of the jet axis, and are the viscosity and surface tension
coefcient, respectively, and I is the moment of inertia of the jet cross-section.
The system of equations (3.1)(3.7) with the closure expressions (3.8)(3.13) describes
both the axisymmetric capillary instability of straight jets driven by surface tension, and
bending instability of liquid jets driven by the dynamic interaction with air or self-induced
electric forces. It is closed if the jet cross-sections possess double symmetry (in particular, it
is closed for jets with circular cross-sections). This system of equations was derived by
Yarin et al., assuming the jet slenderness and the absence of signicant shear tractions at its
surface. These assumptions are sufciently accurate in the case of highly viscous jets
moving in air. In the cases of short-wavelength perturbations and large axis curvatures, as
well as the case of liquid jets propagating in a liquid medium of comparable viscosity, the
quasi-one-dimensional description, strictly speaking, is inappropriate. The assumptions
of the jet slenderness and of the absence of signicant shear tractions at the jet surface
lead to the following additional restrictions on the internal kinematics in the jet, namely
n
1 Vb
1 Vn
Vn ; b
Vb V
s
s
3:14
The kinematic equation, which should be added to the system of equations (3.1)(3.7)
and (3.8)(3.14), determines the location of the jet axis in space in accordance with the
velocity eld in it
R
V V i
t
3:15
Equation (3.15) is written here for the simplest case where the tangent to the jet axis is
inclined at any point to a certain straight line O1 by an acute angle, and it is possible to
introduce a Cartesian coordinate system O1 with the corresponding unit vectors i, j
and k, and describe the jet axis using the following equations
s; s; t; s; t; R i j k
3:16
In the other cases the jet axis parameter s can be chosen differently, which leads to
changes in the expression for R=t.
Note that equations similar to Eqs. (3.1)(3.7) and (3.8)(3.16) introduced by Entov
and Yarin (1980, 1984a) and Yarin (1983, 1993) were also later employed by Ribe (2002)
and Marheineke and Wegener (2009).
67 [63--88] 10.12.2013 5:06PM
3.3 Capillary instability of free liquid jets
67
The distributed force and moment of force imposed on the jet, q and m, should be
specied separately. In particular, in the case of small spatial perturbations of the jet axis
when it rapidly moves in air, inviscid ow theory (see Loitsyanskii 1966, Landau and
Lifshitz 1987, Batchelor 2002) yields the following expressions (Yarin 1993):
2

2
2
q g U0 f 0 j 2 k 2 ; m 0; f a2
3:17
s
s
where g is gas (air) density, and U0 is the velocity of liquid jet relative to air. Equation
(3.17) for q means that it is directed along the normal (as n).
For nite bending perturbations, the drag force imposed by a relative air ow should be
accounted for in addition. Then, for example, for planar jet bending Eqs. (3.17) are
generalized by the following expressions (Yarin 1993):
9
8
"
2 #5=2
2 "
2 #1
=
< 2
n;
q g U20 f 2 1
a
1
sin
: s
s
s
s
s ;
m0
3:18
These expressions close the system of general quasi-one-dimensional equations of free

liquid jets moving in air with arbitrary speeds. In the context of ber forming, the bending
force arising due to the dynamic interaction with the surrounding air is of the utmost
importance for melt- and solution blowing (see Chapter 4).
In the context of the electried jets in electrospinning (see Chapter 5), the distributed
force q originates from the Coulomb repulsion of different parts of the jet surface and is
given by the following expression (Reneker et al. 2000, 2007, Yarin et al. 2001):

L
2
q e n
kn
3:19
a
where e is the electric charge at the jet surface per unit jet length and L is a cutoff length
along the jet axis. Comparison of Eqs. (3.17) and (3.19) shows that both the aerodynamic
and electric bending forces are directed along the normal (as n) and should result in a
very similar aerodynamically or electrically driven bending instability, as discussed
below in Chapters 4 and 5, respectively.
3.3
Capillary instability of free liquid jets

Straight thin liquid jets moving in vacuum, or moving slowly enough in air, are prone to
capillary instability driven by surface tension, which tends to minimization of the surface
energy and thus the surface area. This tendency results in the growth of axially symmetric
perturbations, which ultimately break the jet up into a succession of droplets. The
operational conditions common in ber-forming processes are typically chosen that
way that such capillary instability is prevented, either due to signicant viscoelastic
68
68 [63--88] 10.12.2013 5:06PM
stresses in polymer solutions and melts, or sufciently low temperatures, and thus, high
viscosity. Then, the intact and sufciently uniform bers can be collected. Nevertheless,
in some ber-forming processes polymer solutions can appear to be too dilute, or
polymer molecular weight too low, which strongly diminishes the stabilizing elastic
stresses, as well as the temperature being too high, thus strongly reducing the damping
viscosity, and capillary instability sets in.
The tendency of small perturbations of thin liquid jets of Newtonian liquids to grow,
and the growth rate, can be predicted using rigorous equations of uid mechanics. The
foundations of stability theory for free liquid jets were laid down in the classical work of
Lord Rayleigh (1878). This theory implies that liquid jets break up into droplets due to
the instability of an innite liquid thread of circular cross-section a0 under the action of
surface tension. Rayleighs result can be obtained as a solution of the continuity and
NavierStokes equations with zero viscosity, when the latter degenerate into the Euler
equations (Lamb 1959, Loitsyanskii 1966, Landau and Lifshitz 1987, Batchelor 2002).
In the case of small perturbations of a straight liquid jet, the Euler equations written in
cylindrical coordinates r, and x (the radial, azimuthal and axial coordinates), as well as
the boundary conditions at the jet surface, can be linearized. Then, it is possible to
show that on an innite thread small perturbations inevitably depend on time t and
the azimuthal and axial coordinates and z as exp(t ikx + is). The predicted
dependence of the perturbation increment (growth rate) on the perturbation wavelength,
called the characteristic equation, has the form
s
Is

1 s2 2
3:20
a30 Is
where 2a0 = is the dimensionless wavenumber, with being the wavelength, s the
azimuthal wavenumber, and Is() the modied Bessel functions of s-th order.
In the case of axisymmetric perturbations, s = 0, and the radicand in Eq. (3.20) is positive
for < 1. Thus, one of the values of the perturbation increment is real and positive. This
means that a solution for perturbations exponentially increasing in time exists. Therefore,
liquid threads are unstable to axisymmetric perturbations with wavelength > 2a0 (equivalent to < 1). The unstable axisymmetric perturbations represent a system of periodic
waves standing on a thread with the amplitudes growing exponentially with time. Rayleighs
result means that the thread (and jet) conguration, which does not correspond to a
minimum of the surface energy, is unstable. As a result of growth of the longwave ( < 1)
perturbations, the surface area and energy diminish, which results in the capillary breakup of
a jet into droplets (this explanation of the capillary breakup was given by Plateau 1873).
Note that the absolute minimum of surface energy would correspond to a liquid sphere,
whereas a local minimum is given by a succession of liquid spheres (droplets) emerging
from a cylinder (jet/thread). According to Eq. (3.20), the increment maximum * corresponds to the wavenumber 0:698. Rayleigh assumed that jet breakup is driven by the
perturbation with the corresponding wavelength 2a0 = 9a0 , as in the case of an
innite liquid thread with the same cross-sectional diameter. The comparison of Rayleighs
spectrum (Eq. 3.20) with the experimental data shown in Figure 3.1 is perfect.
69 [63--88] 10.12.2013 5:06PM
69
g/a/(rao3)
0.4
0.2
0
0
Figure 3.1
0.5
1.0
Dimensionless growth rate versus wavenumber. Rayleighs result (Eq. 3.20) is shown by the
curve ( in the graph denotes surface tension and is equivalent to in the text). Open symbols
show the data of Donnelly and Glaberson (1966), lled symbols, the data of Cline and Anthony
(1978). Yarin (1993). Courtesy of Pearson Education.
In the case of nonaxisymmetric perturbations (s 1), the increment determined by Eq.

(3.20) will be imaginary for all wavelengths and the liquid thread (jet) remains stable. In
this case the perturbations represent traveling waves.
The capillary instability of straight thin jets with a nonzero viscosity can still be studied
based on the NavierStokes equations, as in the inviscid case described above, albeit the
calculations appear to be rather involved (Weber 1931). One of the most natural ways
of asymptotic description of the dynamics of jets is the quasi-one-dimensional approach
outlined in Sections 3.1 and 3.2. For thin straight jets the mass and longitudinal
momentum balance Eqs. (3.1) and (3.2) with the closure relations (Eqs. 3.83.10) reduce
to the following system of two equations:
f Vf
0; f a2
t
x
2
3

V
V
xx f
6
2a
7
V
4q5
f
t
x
x
x
2
1 a=x
3:21
3:22
In Eqs. (3.21) and (3.22), t is time, x is the longitudinal coordinate reckoned along the jet axis,
f the cross-sectional area (a is the cross-sectional radius), V and xx are the longitudinal
velocity and normal stress in the jet cross-section, and is the surface tension coefcient. For
viscous Newtonian liquids the stress is given by Eq. (3.23) following from Eqs. (3.8) and (3.9)
9
8
>
>
=
<
2
2
V
1
a=x
3:23
xx 3
q h
i
3=2 >
>
x
;
:a 1 a=x2
1 a=x2
This expression accounts for the fact that the jet surface is subjected to negligible
tractions and the deformation of the jet element corresponds in the rst approximation
to uniaxial elongation/compression. The viscous term on the right-hand side in Eq. (3.23)
thus naturally involves the Trouton elongational viscosity 3, with being liquid
viscosity; the second term corresponds to the capillary pressure proportional to the
local double mean curvature of the jet surface.
70
70 [63--88] 10.12.2013 5:06PM
Equations (3.21)(3.23) represent a slightly modied version of the equations used

in Kase and Matsuo (1965), Matovich and Pearson (1969) and Lee (1974). The
modication introduced in Yarin (1993), and references therein, involves the exact (not
the asymptotic) expressions for the capillary force at the jet surface and capillary pressure
in the jet cross-section, which allows for description of the capillary breakup until
formation of drops. A detailed derivation of such equations based on the abovementioned physical assumptions and the integral mass and momentum balances can be
found in the monograph by Yarin (1993) (see also the later reviews in Eggers 1997,
Eggers and Villermaux 2008, Ashgriz and Yarin 2011 and Yarin 2011). Note that there is
a number of works where similar equations are constructed as a regular asymptotic
expansion of the solutions of the three-dimensional differential equations of uid
mechanics (which were, in their turn, derived from the integral mass and momentum
balances!; Bechtel et al. 1992). The latter method is much more involved and yields the
same equations.
The slower the cross-sectional radius and longitudinal velocity vary along the jet, the
more accurate becomes the quasi-one-dimensional description. Therefore, the linear
theory of the temporal stability of a slightly perturbed innite cylindrical thread based
on Eqs. (3.21)(3.23) yields the long-wave limit of the expression for the perturbation
growth rate found in classical RayleighWeber theory based on the three-dimensional
NavierStokes equations:
r
a0
3
9 2 4 Oh 2
2
Oh
Oh
1 2
3:24
2
4
2
h
pi1=2
According to Eq. (3.24), the fastest growth rate corresponds to 2 1 3 Oh=2
with the Ohnesorge number Oh = 2 / (a0). It should be emphasized that Eq. (3.24)
with = 0 (in the inviscid case) is identical to the limit of Eq. (3.20) as ! 0, i.e. in the
longwave limit of the Rayleigh result. The same is true in the case of a nite viscosity
where Eq. (3.24) accurately approximates in the longwave limit ! 0 the exact
spectrum found by Weber (1931) using the NavierStokes equations.
Summarizing, in the linear stability theory of capillary breakup of thin free liquid jets
the quasi-one-dimensional approach allows for a simple and straightforward derivation
of the results almost exactly coinciding with those obtained in the framework of a rather
tedious analysis of the three-dimensional equations of uid mechanics. This serves as an
important argument for further applications of quasi-one-dimensional equations to more
complex problems that do not allow, or almost do not allow, exact solutions, in particular
to the nonlinear stages of the capillary breakup of straight thin liquid jets, and polymer
solutions and melts relevant in the context of ber forming. The quasi-one-dimensional
equations are also popular in studies of melt- and dry-spinning processes (Chapter 1), as
well as melt- and solution blowing (Chapter 4) and electrospinning (Chapter 5). In these
processes surface tension effects are typically negligible, whereas viscous forces, and
jet cooling and solidication are signicant (Yarin 1993). The quasi-one-dimensional
equations also represent the most important tool in the studies of capillary breakup of
non-Newtonian and rheologically complex jets (Yarin 1993).
71 [63--88] 10.12.2013 5:06PM
71
The linear RayleighWeber theory of capillary breakup developed for small perturbations allows for a rather accurate prediction of the breakup time and length of capillary
jets of low-viscosity liquids when it is extrapolated to the nonlinear stage up to the jet
breakup. However, it does not allow prediction of the jet proles at the late nonlinear
stage of their evolution. As a result, the linear theory is incapable of predicting the sizes of
all droplets originating from the jet breakup. In the experiments, even in cases with an
imposed monochromatic excitation, as a rule, small satellite droplets are formed from
liquid threads between the main drops (Goedde and Yuen 1970, Lafrance and Ritter
1977, Pimbley and Lee 1977, Scarlett and Parkin 1977). The linear theory predicts only
the main drops with the volume equal to the volume corresponding to the fastest growing
wavelength. The asymptotic theories of the nonlinear phenomena characteristic of
capillary breakup of liquid jets lead to rather cumbersome expressions, even for the
simplest case of an inviscid liquid. Therefore, direct numerical simulation of the rigorous
NavierStokes equations or the quasi-one-dimensional equations of the dynamics of thin
liquid jets attracted signicant attention in the literature. For example, the numerical
solution of the quasi-one-dimensional equations in the inviscid case allowed modeling
of formation of both main and satellite drops (Lee 1974). The investigation of the
nonlinear stage of the capillary breakup of low-viscosity liquids in the framework of
the quasi-one-dimensional approach is not fully legitimate, since at a later stage a
signicant nonuniformity of ow should arise in the cross-sections of the growing
main drop. The best results for low-viscosity liquids can be obtained using direct
numerical simulations of the rigorous three-dimensional NavierStokes equations
(Mansour and Lundgren 1990, Ashgriz and Mashayek 1995, Ambravaneswaran et al.
2000, Basaran and Suryo 2007, Ashgriz and Yarin 2011).
On the other hand, capillary breakup of sufciently viscous liquid jets is a longwave
phenomenon, and its description in the framework of the quasi-one-dimensional equations
of the dynamics of liquid jets is sufciently accurate. The effect of the viscosity on
the capillary breakup of highly viscous liquid jets was studied numerically by Yarin
(1993). The initial perturbation of the jet surface was imposed as a harmonic function of
small amplitude, with the wavelength corresponding to the fastest-growing mode
according to RayleighWeber theory. The dimensional group characterizing viscous
effects is the Ohnesorge number, Oh 2 =a0 where , and are the viscosity, density
and surface tension of liquid, respectively, and a0 is the unperturbed cross-sectional jet
radius. As the Ohnesorge number increases, the main drop at the moment of breakup
becomes more and more elongated (spindle-like), whereas the volume corresponding to the
forming satellite droplet decreases (Figures 3.2 and 3.3). The predicted breakup time is
in good agreement with the experimental data. The results shown in Figures 3.2 and 3.3
correspond to innite, periodically perturbed jets (threads). Capillary breakup of
semi-innite jets issued from a nozzle was numerically modeled in Bogy et al. (1980)
using the quasi-one-dimensional equations of the dynamics of liquid jets.
At the late stages of capillary breakup near the jet cross-section where the breakup
will eventually occur, liquid ow completely forgets the initial conditions. Then, the
ow is dominated by the local ow conditions and becomes self-similar. The numerical
description of the latest stages of capillary breakup is unreliable near where the
72
72 [63--88] 10.12.2013 5:06PM
3
4
a
2
3
2
(a)
1
0
0
0.25
x 0.5
(b)
1a
0 10 20 30
Figure 3.2
Capillary breakup of a glycerine jet (Oh = 1.755). (a) Jet prole corresponding to one half of the
perturbation wavelength. The cross-sectional radius a is rendered dimensionless by a0 = 6 105 m,
the axial coordinate x, by * = 1.04 103m. The dimensionless time t = 10 for curve 1, 20 for
curve 2, 30 for curve 3 and 30.81 for curve 4 (the timescale is T = a0/ = 0.848 104s). (b) The jet
shape at the moment of breakup. In the insert, the radius evolution in the cross-section where
the breakup takes place is shown. Yarin (1993). Courtesy of Pearson Education.
3
a
2
1
1
(a)
0
0
0.25
x 0.05
(b)
1a
0
Figure 3.3
10
20
Capillary breakup of a castor oil jet (Oh = 370.3). (a) Jet prole corresponding to one half of the
perturbation wavelength. The cross-sectional radius a is rendered dimensionless by a0 = 6 105 m,
the axial coordinate x, by * = 3.45 103 m. The dimensionless time t = 10 for curve 1, 20 for
curve 2 and 22.4 for curve 3 (the time scale is T = a0/ = 1.52 103 s). (b) The jet shape at the
moment of breakup. In the insert, the radius evolution in the cross-section where the breakup
takes place is shown. Yarin (1993). Courtesy of Pearson Education.
73 [63--88] 10.12.2013 5:06PM
73
cross-sectional radius tends to zero. Theoretical description of such self-similar nal jet
pinching is given in Eggers (1993, 1995), Brenner et al. (1994, 1996, 1997) and
Papageorgiou (1995), assuming either inertia- or viscosity-dominated ows in the tiny
threads and, in particular, using quasi-one-dimensional equations.
Polymer solutions and melts used in ber forming demonstrate memory effects, i.e.
viscoelasticity. Among the experiments devoted to capillary breakup of dilute polymer
solution jets, the seminal work of Goldin et al. (1969) played an important role. In these
experiments, the axisymmetric capillary perturbations began to increase similarly to their
growth in corresponding jets of pure solvents. However, at the later stages, capillary jets
of dilute polymer solutions formed the beads-on-a-string structure (see photographic
image 2 in Figure 3.4). This structure appears to be amazingly long-lived. As a result, the
jet length to breakup sharply increases compared to corresponding jets of pure solvents
(with practically the same values of density, surface tension, initial cross-sectional radius
and velocity). An increase in the polymer concentration results in suppression of the
initial fast perturbation growth after a jet is issued from the nozzle. In this case such a jet
has an appearance of a smooth cylinder. Only at a signicant distance from the nozzle do
the visible axisymmetric perturbations become apparent on the jet. This pattern was fully
corroborated by later experiments by Rubin and Wharshavsky (1972) and Gordon et al.
(1973), and in numerous consequent works.
One of the reasons for the enhanced stability of viscoelastic jets is related to the fact
that they can develop signicant longitudinal stresses during their ow inside the nozzle
or in the transitional zone of the jet formation beyond the nozzle exit, which do not
have enough time to relax during capillary breakup of the free jet. The physical reason for
the emergence of signicant longitudinal stresses is in the coilstretch transition of
polymer macromolecular coils in the converging part of ow near the nozzle entrance
or exit. The presence of longitudinal stresses in polymer liquid jets was experimentally
demonstrated by Goren and Gavis (1961) and Han et al. (2008), using observations of
bending perturbations imposed on straight jets (see some additional detail below).
It should be emphasized that this is a non-trivial phenomenon. Indeed, free nonelectried
1
Figure 3.4
Capillary breakup of a thin jet of 0.02 wt% aqueous solution of poly(oxyethylene) WSR-301. The
unperturbed cross-sectional jet radius a0 = 2.74 104m. (1) The results of the numerical
simulations. (2) The experimental image. (3) The analytical asymptotic solution by Yarin (1993)
and Entov and Yarin (1984b). Yarin (1993). Courtesy of Pearson Education.
74
74 [63--88] 10.12.2013 5:06PM
jets are not pulled at their leading edge. Therefore, a jet can continue to be stretched
only if the deformation from the unloaded leading edge does not have enough time to
reach the nozzle exit (Joseph 1990, Entov and Kestenboim 1987). This can happen if
the jet velocity is higher than the speed of propagation of the waves of elastic stresses
(the elastic sound), i.e. the jet is supersonic, in a sense. On the other hand, if a
jet propagates slowly, it will be subsonic and completely unloaded. Then, after being
issued from a nozzle, a viscoelastic jet abruptly swells, a phenomenon called die swell in
ber-spinning technology (Ziabicki 1976).
The appearance of signicant longitudinal elastic stresses represents only one of the
manifestations of stretching of macromolecular coils in jets of polymer solutions.
Sometimes the measured stresses are so large (e.g. of the order of 10 kPa to 100 kPa;
Han et al. 2008) that one can speak about an almost complete stretching of macromolecules along the jet. However, this alone does not fully explain the mechanism of
stabilization of jets by macromolecular additives. Some additional mechanisms are
discussed below.
Formation of the beads-on-a-string structure at the late stages of capillary breakup of dilute
polymer solution jets is also directly related to the coilstretch transition of macromolecular
coils in such jets. The very fact that the emerging tiny threads between the growing drops are
long-lived (see Figure 3.4) demonstrates an unusual resistance of the liquid in them to any
further deformation. A thin thread is squeezed radially by the capillary pressure /a.
Therefore, liquid in the thread ows axially towards the two ends attached to the neighboring
drops. The effective longitudinal stress is also of the order of /a. In a viscous Newtonian
liquid, such longitudinal stress will result in the following rate of elongation
: 2 da
a dt
3a
3:25
Therefore, measurements of the rate of thread self-thinning da/(adt) reveal the value of
the effective viscosity of liquid . The results of such experiments with dilute polymer
solution jets and threads (Reneker et al. 2007, Bazilevsky et al. 1981, Stelter et al. 2000)
revealed the values of the effective viscosity, which are larger than the solvent viscosity
or the shear viscosity of the same solution by 34 orders of magnitude. For example, for
dilute 0.02 wt% aqueous solution of poly(oxyethylene) WSR-301, the elongational
viscosity was found to be el = 7 Pss, whereas its shear viscosity sh = 3 103 Pas.
This approach resulted in development of a new type of elongational rheometer for dilute
polymer solutions (Bazilevsky et al. 1981, Schmmer and Tebel 1983, Stelter et al. 2000,
Anna and McKinley 2001, Reneker et al. 2007, see also Chapter 2).
The theoretical studies of the capillary breakup of viscoelastic jets were seemingly in
disagreement with the experimental results for quite some time. Linear stability analysis
of the effect of the elastic stresses on the capillary instability of polymer jets always
(irrespective of the viscoelastic constitutive equation used) leads to a conclusion
that small axisymmetric perturbations in viscoelastic liquid jets grow faster than in
corresponding Newtonian jets of the same viscosity Goldin et al. 1969, Kroesser
and Middleman 1969, Rubin 1971, Sagiv et al. 1973, Sagiv and Takserman-Krozer
75 [63--88] 10.12.2013 5:06PM
75
1975, Lee et al. 1981, Goren and Gottlieb 1982). The reason for this is quite elementary:
in the framework of the linearized small perturbation theory, when the unperturbed state
corresponds to an unloaded liquid at rest, the elastic stresses are of the order of the square
of the perturbation amplitude, and thus, are negligibly small. As a result, the role of
elasticity is reduced to a decrease in the effective viscosity, which makes the jet more
unstable. In other words, a Newtonian liquid is rigid, whereas its linearized viscoelastic
counterpart is elastic, i.e. weaker than the rigid originator, and thus is prone to a faster
perturbation growth rate. The resolution of the above-mentioned disagreement can be
achieved if a jet with signicant longitudinal elastic stresses is taken as an unperturbed
state (Goren and Gottlieb 1982, Entov and Yarin 1984b, Yarin 1993). According to the
results of these works, it is possible to state that a fast growth of the axisymmetric
perturbations of capillary jets of viscoelastic liquids is possible only after a full relaxation
of such longitudinal stresses, i.e. with a delay of the order of the relaxation time of
liquid . Perhaps that is the reason for a long absence of visible perturbations on the jets of
semi-dilute polymer solutions in Goldin et al. (1969).
Yarin (1993) and Entov and Yarin (1984b) gave a theory of the capillary breakup
of thin jets of dilute polymer solutions and formation of the bead-on-the-string
structure (some additional later results can be found in Bouseld et al. 1986). The
basic quasi-one-dimensional equations of capillary jets (Eqs. 3.21 and 3.22) are supplemented with an appropriate viscoelastic model for the longitudinal stress. Yarin (1993)
and Entov and Yarin (1984b) used the Hinch (1977) rheological constitutive model,
which yields the following expression:
9
8
>
>
=
<
2
2

V
1
a=x
3
q
xx 3 1 5cr K1

h
i
3=2 >
>
x
;
:a 1 a=x2
1 a=x2

cK2 Lxx Lyy
3:26
The last term on the right-hand side in Eq. (3.26) expresses the elastic stress through the
components of the orientationdeformation tensor L. The evolution of this tensor in the
jet ow is described by the following equations, accounting for macromolecular stretching and relaxation (similar to the upper-convected Maxwell model of Chapter 2):

Lxx
Lxx
V
r2
V
2K1 Lxx
3:27
K2 K3 Lxx
t
x
3
x

Lyy
Lyy
V
r2
V
K1 Lyy
3:28
K2 K3 Lyy
t
x
3
x
2Lyy Lxx
Nb
p
; K2
K1 2
3:29
3r 2Lyy Lxx
Nb 2Lyy Lxx
2r
K3 p
2Lyy Lxx
3:30
76
76 [63--88] 10.12.2013 5:06PM
In Eqs. (3.26)(3.30), denotes solvent viscosity, c is the number of macromolecules in a

unit volume, b is the length of Kuhn segments in the macromolecule, N is the number of
Kuhn segments in the macromolecule, r = bN1/2 is of the order of the equilibrium
macromolecular coil size (see Section 2.1), = 6r/ is the relaxation time and the
elasticity of macromolecular coils. According to Eq. (3.29) macromolecules cannot be
stretched beyond their fully expended length Nb.
The results of the numerical calculations based on Eqs. (3.21), (3.22) and (3.26)(3.30),
depicted in Figure 3.5, reveal the evolution in time of a jet segment corresponding to
one wavelength of perturbation. The relevant dimensionless groups for viscoelastic
jets are the volume fraction of the macromolecular coils in solution in equilibrium =
4cr3/3, the inverse Deborah number De1 = a0 / () = a0 / (6r) and L0, discussed
below. The inverse Deborah number represents the ratio of the characteristic time of the
capillary breakup affected by viscous forces to the elastic relaxation time. It can also be
interpreted as the ratio of the initial modulus of elasticity of macromolecular coils to
capillary pressure. The value L0 represents the initial value of the longitudinal component
of the orientationdeformation tensor L, Lxx, which characterizes the initial axial elongation of macromolecular coils. The ratio of the initial elastic energy to the surface energy
of the jet E0 can be expressed as E0 = 9De1L0/8. The results in Figure 3.5 correspond
to 0.03 wt% aqueous solution of poly(oxyethylene) WSR-301. In the case of L0 = 2.5, the
value of E0 = 2.5 103, which shows that the effect of the elastic stresses at the early
stages of perturbation growth is small. However, the elongational ow in the emerging
thin threads connecting growing drops results in the axial orientation and elongation of
macromolecular coils. Such an elongation can reach one third of the fully stretched
macromolecular length. The macromolecular stretching at this level appears to be
sufcient for such a signicant reinforcement of liquid in the threads that the outow
from them into drops abruptly decreases, as the jet evolution decelerates as a whole. As a
(a)
(b)
(c)
Figure 3.5
The emergence of the beads-on-a-string structure during capillary breakup of a thin viscoelastic
jet. The values of the dimensionless groups are: = 0.419, De1 = 0.21 102, L0 = 2.5. (a) t = 0, (b) t =
0.37 103s, (c) t = 0.56 1033.04 102s. Yarin (1993) and Entov and Yarin (1984b), with
kind permission of Springer Science+Business Media.
77 [63--88] 10.12.2013 5:06PM
77
result, the beads-on-a-string structure forms. It should be emphasized that a signicant

stretching of macromolecular coils in the jet after some delay emerges almost abruptly. In
the numerical calculations the beads-on-a-string structure can be observed almost
unchanged up to times vefold the liquid relaxation time ( = 6.15 103s). At this
moment the calculations were terminated. If the initial axial elongation of macromolecular coils is increased up to L0 = 100 (E0 = 0.1) with the other parameters being
unchanged, the initial elastic stresses begin to affect the initial stage of perturbation
evolution. This results in a delay of the order of 2.5, during which there is no visible
perturbation growth, in correspondence with the predictions of linear stability theory.
After a partial relaxation of the longitudinal stress, a relatively slow perturbation growth
begins. It results in an increase in the longitudinal stresses in the jet contractions, the
stretching of macromolecular coils and the emergence of the beads-on-a-string structure.
An increase in the solvent viscosity leads to an increase in the relaxation time, as well as
in the delay time. In general, the numerical results conrm the stabilizing effect of the
elastic stresses at both the initial stage of evolution of capillary viscoelastic jets, and at the
stage of formation of the beads-on-a-string structure. The following numerical works also
predicted formation of the beads-on-a-string structure in capillary viscoelastic jets using
quasi-one-dimensional or three-dimensional (axisymmetric) equations and different
viscoelastic rheological constitutive equations: Bouseld et al. (1986), Chang et al.
(1999), and Li and Fontelos (2003).
Since the beads-on-a-string structure evolves very slowly, it can be considered as a
succession of nontrivial (noncylindrical) quasi-equilibrium shapes of an elastic material
subjected to the action of surface tension and a very slow viscoelastic relaxation. An
asymptotic analysis of such jet shapes under the assumption that the material represents
a neo-Hookean body was conducted by Yarin (1993) and Entov and Yarin (1984b). It
revealed that the nontrivial quasi-equilibrium jet shapes represent a succession of
spherical drops of radius a0 3=4a0 1=3 (with being perturbation wavelength)
that are fully relaxed and practically do not possess any elastic stresses, with strongly
stressed thin uniform threads connecting the drops. The cross-sectional radius of these
threads is a0 =21=3 , where = 2Ga0/ is the ratio of the modulus of elasticity to capillary
pressure. This analytical solution was compared to the experimental data and the results
of the numerical simulations in Figure 3.4 (curve 3). The analytical solution for the
beads-on-a-string structure was later conrmed by Clasen et al. (2006). The thread
stability can be sustained if the gradual thinning is accompanied by an increase in the
longitudinal elastic stress that is faster than the corresponding decrease of the
surface tension force a0. The initial cylindrical shape will always be stable if the liquid
stiffness is sufciently high, i.e. xx t 0 a0 = >> 1. In the latter case there are no
nontrivial (noncylindrical) jet shapes.
It should be emphasized that in experiments, even a beads-on-a-string structure
ultimately breaks up. Yarin (1991, 1993) attributed the weakening and breakup of the
beads-on-a-string structure to mechanical degradation of polymer macromolecules in
strong elongational ows in the tiny threads between drops. Another possible mechanism
of weakening of the threads is related to viscoelastic relaxation leading to a partial
unloading (Chang et al. 1999). As a result, new cycles of capillary instability appear in
78
78 [63--88] 10.12.2013 5:06PM
a cascade-like manner and new drops are formed between the original beads in the
structure. These iterative instabilities were revealed in the experiments of Oliveira and
McKinley (2005). Note that the duration of the uniform stretching in thin threads of dilute
polymer solutions is very long. The uniform self-thinning in many cases proceeds to such
sizes that the iterative instabilities cannot be observed optically. Such threads are a key
element of the elongational rheometers for dilute polymer solutions (Bazilevsky et al.
1981, Schmmer and Tebel 1983, Stelter et al. 2000, Reneker et al. 2007), where the
threads emerge either between the drops in a jet or between small solid plates. The
dynamics of the uniform thread self-thinning under the action of capillary pressure was
elucidated in Bazilevsky et al. (1981), Renardy (1994, 1995), Stelter et al. (2000),
Bazilevsky et al. (2001), Oliveira and McKinley (2005), and Reneker et al. (2007).
This theory predicts the existence of an intermediate universal regime of thread thinning
according to the following exponential law:

t
a a0 exp
3:31
3
where a0 is the cross-sectional thread radius at t = 0.
Comparing Eq. (3.31) with the experimental data allows for measurements of the
viscoelastic relaxation time of dilute and semi-dilute polymer solutions (see Chapter 2).
3.4
Bending perturbations of Newtonian liquid jets moving in air

with high speed
Thin liquid jets demonstrate not only capillary breakup, but some other regular longwave
forms of instability and breakup, e.g. bending instability of jets moving in air with
relatively high speed U. A theoretical investigation of the dynamics of the bending
instability due to small perturbations of liquid jets rapidly moving in air was begun by
Weber (1931) and Debye and Daen (1959). This led to a coupled problem of the dynamic
interaction of air ow with the liquid jet. The linear stability analysis of the bending
perturbations on an innite inviscid jet can be carried using the rigorous continuity
equation and the linearized version of the Euler equations, to which the NavierStokes
equations reduce when viscosity = 0. Debye and Daen (1959) assumed the bending
perturbations to be planar and obtained an expression for the perturbation growth rate.
Neglecting several minor terms, this expression takes the form
s
g U2 K1 I1
I
20
3:32
3 1
a0 K1 I1 a0 I1
The gas density is denoted by g, and I1 and K1 denote the modied Bessel functions. In
the case of bending perturbations, the surface tension is a stabilizing factor, since
0
0
bending results in an increase in the jet surface area ( I1 > 0; K1 50 for any ).
Above a threshold value of the relative air velocity U0, the rst (positive) term in the
79 [63--88] 10.12.2013 5:06PM
3.4 Bending Perturbations
79
radicand in Eq. (3.32) acquires a larger magnitude than the second term, which
corresponds to the onset of the bending instability and an exponential growth in the
bending perturbations. The bending instability is determined by the pressure distribution in the gas over the jet surface: the gas pressure on convex surface elements is lower
than on the concave ones.
The theory of Debye and Daen (1959) does not account for a number of important
factors. The most important of them is the effect of liquid viscosity, which should
counteract the perturbation growth. In addition, the experiments show that the
growing bending perturbations are three-dimensional rather than planar. Also, in the
case of low-viscosity liquid jets, bending perturbations grow together with the axisymmetric capillary perturbations, which signicantly changes the cross-sectional sizes and
shapes during bending (Grant and Middleman 1966). Moreover, all these factors are
dominant in reality, and especially in the context of ber forming. Therefore, the analysis
of the dynamics of bending perturbations in the framework of an inviscid liquid model
is intrinsically contradictory. Accounting for these factors in the framework of the
NavierStokes equations in the problems of bending perturbations of liquid jets is
tremendously difcult. However, these difculties can be relatively easily overcome
in the framework of the quasi-one-dimensional description of liquid motion in the
bending jets.
In the case of highly viscous jets, bending perturbations are expected to have long
waves and the quasi-one-dimensional equations of the jet dynamics can be employed
(Yarin 1993). Below in this section we discuss in brief the applications of Eqs. (3.1)(3.7)
and (3.8)(3.18) to the aerodynamically driven bending instability of uncharged
liquid jets rapidly moving in air, following the works of Entov and Yarin (1980,
1984a) and Yarin (1993). The characteristic equation for the growth rate of small bending
perturbations of highly viscous slender liquid jets moving in air in the case of temporal
instability derived from the linearized version of these equations reads
!
2
4
U
3
g
0
2
2 2 0
3:33
4 a20
a0
a30
According to Eq. (3.33), any planar or three-dimensional (helical), small bending
perturbations increase with the same growth rate if the velocity of liquid gas relative to
gas (air) is large enough
q
U0 > =g a0
3:34
when the dynamic action of air can overcome the resistance of surface tension to growth
of bending perturbations.
Solving Eq. (3.33), one can nd the wavenumber and the growth rate of the
fastest-growing bending perturbation

1=6
8 a20
2

U
3:35
9 2 g 0 a0
80
80 [63--88] 10.12.2013 5:06PM
h

i2=3
a0 g U20
3a40
1=3
3:36
Comparing the fastest growth rates predicted by Eqs. (3.24) and (3.33), we nd that the
growth rate of the axisymmetric capillary perturbations is much smaller than that of the
bending perturbations for sufciently viscous liquids when the inequality
a20 g U20
>>1
3:37
holds. In this case deformations of the jet due to the capillary RayleighWeber instability
can be neglected during bending.
It is worth noting that at = 0, Eq. (3.33) coincides with the longwave limit ( ! 0)
of Eq. (3.32).
In the works of Yarin (1982a, 1982b, 1993) and Entov and Yarin (1984a), the aerodynamically driven nonlinear bending instability of thin jets of highly viscous liquids rapidly
moving in air was studied numerically by solving Eqs. (3.1)(3.7) and (3.8)(3.18). It was
shown that the nonlinear effects, and in particular the most important of them the viscous
stresses originating from the elongation of the bending jet axis decelerate growth of
bending perturbations. However, for the estimates of the perturbation amplitudes and timing
up to the amplitudes of the order of (24)a0 one may extrapolate the predictions of the linear
theory with sufcient accuracy. The presence in Eq. (3.18) of the quadratic drag force leads
to a slow sweep of bending perturbations down the gas ow, in addition to growth of their
amplitude (the latter is due to the lift component of the aerodynamic force q). The
congurations of the jet axis corresponding to one length of the bending perturbation at
several consecutive time moments denoted by numerals next to the curves are shown in
Figure 3.6. It can be seen that the jet axis at the end takes the form of a cliff, which leads to
an overturning. At this moment the amplitude of the bending perturbation is of the order
of 4a0. Figure 3.7 depicts the corresponding jet section at the moment of overturning
The dynamics of bending perturbations of high-speed viscoelastic jets of uncharged
polymer solutions and melts, as well as of concentrated micellar solutions was studied by
Yarin (1986, 1993). The closure expressions for the Newtonian jets of Section 3.2 should
be replaced by the corresponding ones, or the method of energy balance developed in Yarin
(1986, 1993) should be applied. Using the latter approach it was shown that the growth rate
of small bending perturbations is determined by the following characteristic equation:
!
g U20 0
3
4
2

2 2 2 0
3:38
4 a20 1
a0
a0
a30
which generalizes Eq. (3.33) to the case of viscoelastic liquid jets (the Newtonian case is
recovered with a relaxation time = 0). In Eq. (3.38) the initial longitudinal stress in the
jet can either be absent (0 = 0) or present and frozen 0 const 0:
If 0 = 0, Eq. (3.38) predicts an accelerated growth of small bending perturbations of
viscoelastic liquids compared to the corresponding Newtonian liquid (with the same
81 [63--88] 10.12.2013 5:06PM
3.4 Bending Perturbations
0.5
H
1
34
81
0.5
0
Figure 3.6
0.5
1.5
Jet evolution in the case of the bending perturbations of nite amplitude affected by air drag force.
All the parameters in the plot are dimensionless. As a length scale, the wavelength of the fastesth

i1=6
is chosen.
growing bending mode in the linear approximation 2 9=82 a40 = g U20
Time denoted by the numerals near the curves is rendered dimensionless by the characteristic time
h

i1=3
. Yarin (1982b), with kind permission of
of small bending perturbations, T a20 = 2g U40
Springer Science+Business Media (see also Entov and Yarin 1984a and Yarin 1993).
Figure 3.7
Predicted instantaneous shape of a jet with a large-scale aerodynamically driven bending instability
corresponding to the jet axis conguration at t = 7 in Figure 3.6. The cross-sectional radius ranges
from 65% to 80% of its initial value. Yarin (1982b), with kind permission of Springer Science
+Business Media (see also Entov and Yarin 1984a and Yarin 1993).
values of , , , a0 and U0) due to a decrease in the effective viscosity eff = (1 + ).

The initial stress 0 > 0 is a stabilizing factor that diminishes the growth rate or can even
prevent bending instability setting in if (0 + /a0) > gU20 . The following dimensionless
groups govern the bending perturbations of viscoelastic jets:
1
g
g U20
; 2 2
;
;
=
a0 g U20
0
=a0
4
; 5
g U20
g U20
3:39
Figure 3.8 depicts the growth rates predicted by Eq. (3.38) for two jets of upper-convected
Maxwell liquids, which are shown by curves 1 and 3. They correspond to different values
82
82 [63--88] 10.12.2013 5:06PM
, sec1
100
1
2
50
3
4
0
0.5
Growth rate of small bending perturbations of viscoelastic jets of the upper-convected Maxwell
liquid. For all curves 1 = 103, 4 = 5 = 0. For curves 1 and 2, 2 = 0.156 104; for curves 3 and
4, 2 = 0.4 104. For curves 2 and 4, 3 = 0; for curves 1 and 3, 3 = 0.64 and 0.25, respectively.
Yarin (1986), with kind permission of Springer Science+Business Media (see also Yarin 1993).
Figure 3.8
1
Y
4.5
4.5
3
8
0
(a)
Figure 3.9
14
21
28
14
21
28
(b)
(a) The amplitude Y of the bending perturbations of a jet of upper-convected Maxwell liquid is
shown by curve 1. The values of the dimensionless groups are: 1 = 103, 2 = 0.156 104,
3 = 0.64, 4 = 5 = 0. Curve 2 depicts the amplitude of the corresponding jet of Newtonian liquid
(3 = 0). (b) Bending instability of a jet with initial tension. Curve 1 corresponds to the case of
an upper-convected Maxwell liquid with frozen initial tension: 4 = 0.91, with the values of
13 and 5 being the same as in panel (a); for curve 2 the values of 15 are the same as
for curve 1, but the initial tension relaxes along the jet; curve 3 shows the corresponding
Newtonian jet. Yarin (1986), with kind permission of Springer Science+Business Media (see
also Yarin 1993).
of the relative gas velocity (U0 for curve 1 is higher than for curve 3). Curves 2 and 4 show
the results for the corresponding Newtonian liquid jets, with curve 2 corresponding to curve
1, and curve 4 to curve 3. It should be emphasized that the growth rates of the small
perturbations of Newtonian viscous jets in Figure 3.8 correspond to the initial stages of
further nonlinear evolution similar to that depicted in Figures 3.6 and 3.7.
The nonlinear dynamics of the bending perturbations of high-speed viscoelastic
uncharged jets moving in air, reveal the phenomenon found by Yarin (1986, 1993).
Figure 3.9a shows that at the nonlinear stage, the growth of the perturbation amplitude is
not only drastically slowed down due to the longitudinal stresses resulting from jet
elongation (similar to Newtonian jets, as mentioned above), but the amplitude can also
83 [63--88] 10.12.2013 5:07PM
3.5 Buckling of liquid jets
83
decrease and oscillate. These latter phenomena result from competition between inertial
and elastic forces. A jet element undergoing bending misses its equilibrium position
due to its inertia and the jet axis becomes overstretched, which produces an extra
longitudinal elastic stress. This stress tends to contract the jet element. However, during
the contraction stage (when the bending perturbation amplitude decreases), the jet
element once again misses its equilibrium position due to its inertia and becomes
overcompressed. This initiates a new cycle of oscillations. Viscous stresses gradually
dissipate the energy of these oscillations.
In addition, Figure 3.9b shows that the pre-existing longitudinal stress in the jet corresponding to a sufciently high value of 4 0 =g U20 > 0 can signicantly delay growth
of the bending perturbations (curve 1) compared to the initially unloaded case (curve 2).
3.5
Buckling of liquid jets impinging on a wall

In his seminal works, G.I. Taylor (1969a, 1969b) discovered a new instability mode of
highly viscous jets, namely buckling of slowly moving jets impinging onto a wall when
they are subjected to a longitudinal compressive force, the situation similar to microand nanober collection in melt- and solution blowing, and electrospinning (Chapters 3
and 4). In one of Taylors experiments, compressive forces resulted from squeezing the
ends of a liquid thread oating on a mercury surface. In another experiment, highly
viscous jets were either moving vertically downward in a liquid and passing into a
lower layer of higher density, which created a sudden increase in the compressive
buoyancy force, or impinging on a wall. The latter case was experimentally studied in
detail by Cruickshank and Munson (1981) and Munson (1982). These works showed
that buckling occurs only in very slowly moving highly viscous jets. Jets with values of
the Reynolds number Re exceeding the critical threshold of Recr 1.2 were stable and
straight (Re = V0d0/, where V0 is the jet normal velocity relative to the wall and d0 is
the nozzle diameter). Therefore, in this case the buckling instability emerges when the
Reynolds number decreases. On the other hand, there is also a restriction on the jet
length L from nozzle to wall. If L is less than a certain critical value Lcr, there
is no buckling. Immediately after the onset of the buckling instability (at Re < Recr
and L/d0 Lcr/d0 ), a two-dimensional buckling (folding) sets in, and the jet is deposited
on the wall as folds. However, with a further increase in the ratio L/d0, bucking
perturbations become spiral-like. The jet axis becomes three-dimensional, which
signies the bifurcation from folding to coiling, and the jet is deposited on the wall
as coils. In the experiments of Cruickshank and Munson (1981) the dependences of Lcr
and the folding and coiling frequencies on the liquid jet viscosity, its velocity and the
nozzle diameter were established. Buckling was also observed in horizontal jets
moving over the free surface of a denser liquid (such jets widen beginning from the
nozzle, unlike gravity-driven jets, which initially become thinner and begin to widen
only close to the wall onto which they impinge). All the observations conrmed the
idea of G.I. Taylor (1969a) that buckling of liquid jets is determined by the presence of
a longitudinal compressive force acting in the jet and in this sense is a direct analog of
84
84 [63--88] 10.12.2013 5:07PM
the elastic buckling of bars and columns studied by Euler (Landau and Lifshitz 1970).
A detailed theory of the onset of buckling instability (folding of highly viscous liquid
jets and lms planar jets impinging on a wall) was given by Yarin (1993),
Tchavdarov et al. (1993) and Yarin and Tchavdarov (1996) based on the general
quasi-one-dimensional equations of jet dynamics (Eqs. 3.13.7 and 3.83.18) (see
also the later efforts directed on a nonlinear buckling theory in Skorobogatiy and
Mahadevan 2000).
15.0 m
15.0 mm
3b
15.0 m
15.0 mm
(a)
15.0 m
15.0 m
15.0 mm
5b
15.0 mm
(c)
5e
(d)
15.0 m
15.0 m
15.0 mm
4e
(e)
Figure 3.10
4a
(b)
15.0 mm
5a
(f)
(a)(f): Comparison of the buckled patterns created by electried jets of aqueous solution of
polyethylene oxide (PEO) collected on glass slides in Han et al. (2007), to patterns produced by the
buckling of the uncharged gravity-driven syrup jets in Chiu-Webster and Lister (2006). Note that
the gravity-driven syrup jets and their buckling patterns are about 1000 times larger than those of
the electried PEO solution jets. The upper panel in each pair depicts the results for the electried
PEO jets in Han et al. (2007). The lower panels show the similar patterns produced by the syrup jets
in Chiu-Webster and Lister (2006). The symbols in the lower right corner of each panel are the
gure number found in Chiu-Webster and Lister (2006). Reprinted from Han et al. (2007), with
85 [63--88] 10.12.2013 5:07PM
3.6 References
85
Recently jet buckling on laterally moving solid surfaces nearly perpendicular to the jet
axis was reported (Chiu-Webster and Lister 2006, Ribe et al.2006, Han et al. 2007),
which is of interest in relation to melt- and solution blowing, electrospinning and writing
by short straight electrically driven jets (so-called near-eld electrospinning). The
stability analysis in the case of buckling was also based on the quasi-one-dimensional
equations of the dynamics of liquid jets (Eqs. 3.13.7 and 3.83.16). It revealed that the
characteristic frequencies of buckling are practically unaffected by the lateral motion of
the wall and stay the same as in the case of liquid jet impingement on a stationary hard at
surface (Ribe et al. 2006, Han et al. 2007). Moreover, the deposit morphology at the wall
is practically unaffected by the method of jet initiation (gravity-driven jets, Chiu-Webster
and Lister 2006, versus electrically driven jets, Han et al. 2007), as Figure 3.10 demonstrates. The buckling frequency predicted by the linear stability theory of Yarin (1993)
and Tchavdarov et al. (1993)
!

d0
Q
log
0:0194 log
0:2582
3:40
V0
g d40
is in a reasonable agreement with the experimental data for evaluated from images
similar to those in Figure 3.10 (d0 denotes the initial cross-sectional jet diameter
approximately equal to that of the nozzle, Q is the volumetric ow rate in the jet and g
is the gravity acceleration).
3.6
References
Ambravaneswaran, B., Phillips, S. D., Basaran, O. A., 2000. Theoretical analysis of dripping
faucet. Phys. Rev. Lett. 85, 53325335.
Anna, S. L., McKinley, G. H., 2001. Elasto-capillary thinning and breakup of model elastic liquids.
J. Rheol. 45, 115138.
Ashgriz, N., Mashayek, F., 1995. Temporal analysis of capillary jet breakup. J. Fluid Mech. 291,
163190.
Ashgriz, N., Yarin, A. L. 2011. Chapter 1. Capillary instability of free liquid jets. Springer
Handbook of Atomization and Sprays, pp. 353, Springer, Heidelberg.
Basaran, O. A., Suryo, R., 2007. The invisible jet. Nature Physics 3, 67968.
Cambridge.
Bazilevsky, A. V., Entov, V. M., Rozhkov, A. N., 2001. Breakup of an Oldroyd liquid bridge as a
method for testing the rheological properties of polymer solutions. Polym. Sci. Ser. A 43,
716726.
Bazilevsky, A. V., Voronkov, S. I., Entov, V. M., Rozhkov, A. N., 1981. Orientational effects in
capillary breakup of jets and threads of dilute polymer solutions. Sov. Phys. Doklady 257,
336339 (the English version in Vol. 26).
Bechtel, S. E., Cao, J. Z., Forest, M. G., 1992. Practical application of a higher-order perturbationtheory for slender viscoelastic jets and bers. J. Non-Newton. Fluid Mech. 41, 201273.
Bogy, D. B., Shine, S. J., Talke, F. E., 1980. Finite difference solution of the Cosserat uid jet
equations. J. Comput. Phys. 38, 294326.
86
86 [63--88] 10.12.2013 5:07PM
Bouseld, D. W., Keunings, R., Marrucci, G. Denn, M. M., 1986. Nonlinear analysis of the surface
tension driven breakup of viscoelastic laments. J. Non-Newtonian Fluid Mech. 21, 7997.
Brenner, M. P., Shi, X. D., Nagel, S. R., 1994. Iterated instabilities during droplet ssion. Phys. Rev.
Lett. 73, 33913394.
Brenner, M. P., Lister, J., Stone, H. A., 1996. Pinching threads, singularities and the number 0.0304.
Phys. Fluids 8, 28272836.
Brenner, M. P., Eggers, J., Joseph, K., Nagel, S. R., Shi, X. D., 1997. Breakdown of scaling in
droplet ssion at high Reynolds number. Phys. Fluids 9, 15731590.
Chang, H. C., Demekhin, E. A., Kalaidin, E., 1999. Iterated stretching of viscoelastic jets. Phys.
Fluids 11, 17171737.
Chiu-Webster, S., Lister, J. R., 2006. The fall of a viscous thread onto a moving surface: a
uid-mechanical sewing machine. J. Fluid Mech. 569, 89111.
Clasen, C., Eggers, J., Fontelos, M. A., Li, J., McKinley, G. H., 2006. The beads-on-string structure
of viscoelastic threads. J. Fluid Mech. Vol. 556, 283308.
Cline, H. E., Anthony, T. R., 1978. The effect of harmonics on the capillary instability of liquid jets.
J. Appl. Phys. 49, 32033208.
Cruickshank, J. O, Munson, B. R., 1981. Viscous uid buckling of plane and axisymmetric jets.
J. Fluid Mech. 113, 221239.
Debye, P., Daen, J., 1959. Stability considerations of nonviscous jets exhibiting surface or body
tension. Phys. Fluids 2, 416421.
Donnelly, R. J., Glaberson, W., 1966. Experiments on the capillary instability of a liquid jet. Proc.
Roy. Soc. London A 290, 547556.
Eggers, J., 1993. Universal pinching of three-dimensional axisymmetric free-surface ow. Phys.
Rev. Lett. 72, 34583460.
Eggers, J., 1995. Theory of drop formation. Phys. Fluids 7, 941953.
Eggers, J., 1997. Nonlinear dynamics and breakup of free-surface ows. Rev. Mod. Phys. 69,
865929.
Eggers, J., Villermaux, E., 2008. Physics of liquid jets. Reports on Progress in Physics 71, 036601.
Entov, V. M., Kestenboim, Kh. S., 1987. Mechanics of ber formation. Fluid Dynamics 22, 677686.
Entov, V. M., Yarin, A. L., 1980. Dynamical equations for a liquid jet. Fluid Dynamics 15, 644649.
Entov, V. M., Yarin, A. L., 1984a. The dynamics of thin liquid jets in air. J. Fluid Mech. 140, 91111.
Entov, V. M., Yarin, A. L., 1984b. Inuence of elastic stresses on the capillary breakup of jets of
dilute polymer solutions. Fluid Dynamics 19, 2129.
Goedde E. F., Yuen. M. C., 1970. Experiments on liquid jet instability. J. Fluid Mech. 40, 495511.
Goldin, M., Yerushalmi, J., Pfeffer, R., Shinnar, R., 1969. Breakup of a laminar capillary jet of
viscoelastic uid. J. Fluid Mech. 38, 689711.
Gordon, M., Yerushalmi, J., Shinnar, R., 1973. Instability of jets of non-Newtonian uids. Trans.
Soc. Rheol. 17, 303324.
Goren, S., Gavis, J., 1961. Transverse wave motion on a thin capillary jet of a viscoelastic liquid.
Goren, S. L., Gottlieb, M., 1982. Surface-tension-driven breakup of viscoelastic liquid threads.
J. Fluid Mech. 120, 245266.
Grant, R. P., Middleman, S., 1966. Newtonian jet stability. AIChE Journal 12, 669678.
Han, T., Reneker, D. H., Yarin, A. L., 2007. Buckling of jets in electrospinning. Polymer 48,
60646076.
87 [63--88] 10.12.2013 5:07PM
3.6 References
87
Hinch, E. J., 1977. Mechanical models of dilute polymer solutions in strong ows. Phys. Fluids
20, 2230.
Kase, S., Matsuo, T., 1965. Studies on melt spinning. I. Fundamental equations on the dynamics of
melt spinning. J. Polym. Sci. A, 3, 25412554.
Kroesser, F. W., Middleman, S., 1969. Viscoelastic jet stability. AIChE Journal 15, 383386.
Lafrance, P., Ritter, R. C., 1977. Capillary breakup of a liquid jet with a random initial perturbation.
Trans ASME: J. Appl. Mech. 44, 385388.
Lamb, H., 1959. Hydrodynamics. Cambridge University Press, Cambridge.
Lee, H. C., 1974. Drop formation in a liquid jet. IBM J. Res. Dev. 18, 364369.
Lee, W. K., Yu, K. L., Flumerfelt, R. W., 1981. Instability of stationary and uniformly moving
cylindrical uid bodies. Int. J. Multiphase Flow 7, 385400.
Li, J., Fontelos, M. A., 2003. Drop dynamics on the beads-on-string structure for viscoelastic jets:
A numerical study. Phys. Fluids 15, 922937.
Moscow, 1970.
Lord Rayleigh, On the instability of jets. Proc. Lond. Math. Soc. 10, 413 (1878).
Mansour, N. N., Lundgren, T. S., 1990. Satellite formation in capillary jet breakup. Phys. Fluids A
2, 11411144.
Marheineke, N., Wegener, R., 2009. Asymptotic model for the dynamics of curved viscous bers
with surface tension. J. Fluid Mech. 622, 345369.
Matovich, M. A., Pearson, J. R. A., 1969. Spinning a molten threadline. Steady-state viscous ows.
Ind. and Eng. Chem. Fundam. 8, 512520.
Munson, B. R., 1982. Viscous buckling of slender horizontal jets. Phys. Fluids 24, 17801783.
Oliveira, M. S. N., McKinley, G. H., 2005. Iterated stretching and multiple beads-on-a-string
phenomena in dilute solutions of highly extensible exible polymers. Phys. Fluids 17, 071704.
Papageorgiou, D. T., 1995. On the breakup of viscous liquid threads. Phys. Fluids 7, 15291544.
Pimbley, W. T., Lee, H. C., 1977. Satellite droplet formation in a liquid jet. IBM J. Res. Dev. 21,
2130.
Plateau, J. Statique Experimentale et Theorique des Liquides Soumis aux Seules Forces
Moleculaires. Gauthier Villars, Paris, 1873.
Renardy, M., 1994. Some comments on the surface-tension driven break-up (or lack of it) of
viscoelastic jets. J. Non-Newton. Fluid Mech. 51, 97107.
Renardy, M., 1995. A numerical study of the asymptotic evolution and breakup of Newtonian and
viscoelastic jets. J. Non-Newton. Fluid Mech. 59, 267282.
Ribe, N. M., 2002. A general theory for the dynamics of thin viscous sheets. J. Fluid Mech. 457,
255283.
Ribe, N. M., Lister, J. R., Chiu-Webster, S., 2006. Stability of a dragged viscous thread: Onset of
stitching in a uid-mechanical sewing machine. Phys. Fluids 18, 124105.
Rubin, H., 1971. Breakup of viscoelastic liquid jet. Isr. J. Technol. 9, 579581.
88
88 [63--88] 10.12.2013 5:07PM
Rubin, H., Wharshavsky, M., 1972. A note on the breakup of viscoelastic liquid jets. Isr. J. Technol.
8, 285288.
Sagiv, A., Rubin, H., Takserman-Krozer, R., 1973. On the breakup of cylindrical liquid jets. Isr.
J. Technol. 11, 349354.
Sagiv, A., Takserman-Krozer, R., 1975. Capillary breakup of viscoelastic liquid jet of variable
cross-section. Rheol. Acta 14, 420426.
Scarlett, B., Parkin, C. S., 1977. Droplet production on controlled jet breakup. Chem. Eng. J. 13,
127141.
Schmmer, P., Tebel, K. H., 1983. A new elongational rheometer for polymer solutions.
J. Non-Newton. Fluid Mech. 12, 331347.
Skorobogatiy, M., Mahadevan, L., 2000. Folding of viscous sheets and laments. Europhys. Lett.
52, 532538.
Stelter, M., Brenn, G., Yarin, A. L., Singh, R. P., Durst, F., 2000. Validation and application of
a novel elongational device for polymer solutions. J. Rheol. 44, 595616.
Taylor, G. I., 1969a. Instability of jets, threads and sheets of viscous uid. In: Proceedings of the
12th International Congress on Applied Mechanics, Stanford, 1968. Stanford, p. 382388.
Taylor, G. I., 1969b. Electrically driven jets. Proc. Roy. Soc. London A 313, 453475.
Tchavdarov, B., Yarin, A. L., Radev, S., 1993. Buckling of thin liquid jets. J. Fluid Mech.
253, 593615.
Weber, C., 1931. Zum Zerfall eines Flussigkeitsstrahles. Z. Angew. Math. und Mech. 11, 136154.
Yarin, A. L., 1982a. Dynamics of bending disturbances of nonlinear viscous liquid jets in air.
J. Applied Mechanics and Technical Physics 23, No. 1, 3943.
Yarin, A. L., 1982b. A numerical investigation of the bending instability of thin jets of liquid,
J. Applied Mechanics and Technical Physics 23, No. 4, 498502.
Yarin, A. L., 1983. On the dynamical equations for liquid jets. Fluid Dynamics 18, 134136.
Yarin, A. L., 1986. Flexural perturbations of free jets of Maxwell and Doi-Edwards liquids.
J. Applied Mechanics and Technical Physics 27, 828836.
Yarin, A. L., 1991. Strong ows of polymeric liquids: 2. Mechanical degradation of macromolecules. J. Non-Newton. Fluid Mech. 38, 127136.
Yarin, A. L., 2011. Chapter 2. Bending and buckling instabilities of free liquid jets: experiments
and general quasi-one-dimensional model. Springer Handbook of Atomization and Sprays,
pp. 5573, Springer, Heidelberg.
Yarin, A. L., Tchavdarov, B., 1996. Onset of folding in plane liquid lms. J. Fluid Mech.
307, 8599.
89 [89--178] 10.12.2013 9:59PM
Melt- and solution blowing
This chapter describes the machinery, mechanism and signicant experimental and
theoretical aspects of melt- and solution blowing. Meltblowing is a popular method
of producing polymer micro- and nanobers en masse in the form of nonwovens via
aerodynamic blowing of polymer melt jets (Section 4.1). Its physical aspects were
revisited recently. The process involves a complex interplay of the aerodynamics of
turbulent gas jets with strong elongational ows of polymer melts, only recently uncovered and explained.
The role of turbulent pulsations (produced by turbulent eddies in the gas jet) in
meltblowing is discussed rst in Section 4.2 in the framework of a model experimental
situation where solid exible sewing threadlines are used to probe a parallel high-speed
gas jet. After that, in Section 4.3, the dynamics of bending and apping of exible
threadlines in a gas jet is considered. In Section 4.4 the aerodynamically driven stretching
of a straight polymer jet is considered. In Section 4.5 it is shown how a severe bending
instability leading to strong stretching and thinning of polymer jets can arise. This is done
in the framework of a linearized version of the governing equations in the case of small
bending perturbations of a single threadline or polymer jet in meltblowing. Then, in
Section 4.6 the fully nonlinear case of large-amplitude planar bending perturbations of a
single polymer jet is discussed. Both isothermal and non-isothermal cases are considered.
In particular, it is shown how the cooling of the surrounding gas jet results in cooling of
the polymer jet inside, and to the arrest of the bending perturbation growth due to melt
solidication. Section 4.7 is devoted to predictions of three-dimensional congurations
of polymer jets in meltblowing from die exit to deposition screen. Not only a single
polymer jet, but multiple polymer jets are modeled simultaneously, as well as deposition
on a screen moving normally to the principal jet direction being accounted for. The
results include prediction of the ber deposition patterns in lay-down and ber-size
distributions in the resulting nonwovens. The angular distributions in lay-down nonwovens are also predicted. Comparisons with the experimental data suggest that the
model captures main trends rather accurately.
Meltblowing results in formation of microbers, whereas solution blowing produces
polymer nanobers. The experimental aspects of subsonic and supersonic solution
blowing of monolithic and coreshell nanobers are discussed in Section 4.8. Solution
blowing can be benecial when forming nanobers from polymers with low dielectric
permittivity, such as Teon, and biodegradable biopolymers. Solution blowing of
biodegradable biopolymers is treated in Section 4.9, concluding this chapter.
90 [89--178] 10.12.2013 9:59PM
90
4 Melt- and Solution Blowing
4.1
Meltblowing process
Meltblowing is the most important technology developed for the nonwovens and afliated
industries. Meltblown products dominate medical and hygiene markets, but are also found
in many other applications ranging from wipes to oil adsorbents and lters, etc.
Meltblowing offers structures that are lightweight and thin, with bers much smaller than
is the norm with other processes the result is a structure that can act as a barrier, a lter.
The structures are inherently weak, and therefore are most often sandwiched in between
two other spunbonded fabrics (discussed later in this book) to form a composite referred to
commonly as SMS (spunbond, meltblown, spunbond). Note however, that SMS very likely
possesses more than three layers. SMMSSMS, for example, is a seven-layer medical fabric
that is generically referred to as SMS. The higher number of layers leads to an averaging out
of the potential nonuniformities in the structure. More importantly, however, a seven-beam
(a beam refers to each of the meltblowing or spunbonding units placed in series) machine
produces signicantly higher volumes and the unit cost of production becomes lower. The
capital costs are not scalar and the total capital investment is also lower per unit. Consider
that a 12 g m2 SMS medical fabric may consist of seven layers and that each beam is laying
down ~1 to 2 g m2 per meter. Given that the throughput of a meltblown per meter is
~90 100 kg, this means that the machine collection speed can be 1000 m min1.
Considering that the meltblowing beam is producing a web with 25 m bers, this is a
tremendous engineering marvel. It is equally possible to produce bers that are 1 m or less.
Moreover, some new developments in this area now allow the formation of sub-micron
bers without adversely affecting throughput this is discussed below.
Today, several companies offer meltblowing equipment. However, the most successful
in offering large-scale integrated solutions is the Reifenhauser Group in Troisdorf,
Germany. Their Reicol division is the global leader in this segment, dedicated to
meltblowing and spunbonding, and offer integrated systems globally.
The concept of meltblowing extends back to 1954, to Van A. Wente (Wente 1954,
1956) of the Naval Research Laboratories, who was interested in developing ne bers
to collect radioactive particles in the upper atmosphere to monitor worldwide testing of
nuclear weapons. In this process an extruder forced a molten polymer through a row of
ne orices directly into two converging high-velocity streams of heated air or other gas.
The abstract of his paper states:
Fibers 0.1 to 1 in diam. were made from polyamides, poly(ethylene terephthalates), poly(methyl
methacrylates), polystyrene, and poly(triuorochloroethylene). A ram extruder forces a hot thermoplastic melt through a row of ne orices into two converging streams of high velocity heated
gas, usually air. A screen moving past the discharge collects a mat of ne bers. Fiber diam. can be
regulated by control of air pressure and temp., nozzle temp., and polymer feed rate. Fiber mats so
made had high ltration efciency when tested with an aerosol of 0.3- liquid droplets.
During the late 1960s and early 1970s, according to McCulloch (1999), Wentes work
inspired Exxons engineers (Prentice, Buntin and many others) to make the process
operational and to improve it further. Their motivation was rooted in the need to develop
a market for Exxons newly commercialized polypropylene-ber resin. The US patent
91 [89--178] 10.12.2013 9:59PM
4.1 Meltblowing processes
91
led in 1969 sketches out the machine developed by them, whereas the patent by Buntin
(1976) gives its construction in more detail.
Exxon Research launched the rst semi-works line, licensing the technology and
coining a name for the new process: meltblowing. Thus, Exxon became the rst to
demonstrate, patent, publicize and license the use of Wentes concept as a very practical
one-step process to produce unique types of nonwoven webs. Early successful licensees
included Kimberly-Clark, Johnson & Johnson, James River, Web Dynamics and Ergon
Nonwovens, followed by many other companies, including 3M (McCulloch 1999).
The technology is suitable for thermoplastic polymers and can process, in principle, a
broad variety of polymers. However, the process requires extremely low viscosity. Thus,
a principal requirement for the material is that its average molecular weight be adjusted to
yield the desired low viscosity at the melting temperatures in the die. Polypropylene (PP)
is most widely used polymer for meltblowing, due to its ease of processibility, availability, lower cost and its availability in high values of melt-ow index (MFI). PP also
does not require drying like many other polymers, such as nylons and polyethylene
terephthalates (PETs). Typically, specially designed PP resins used for meltblowing
range from 1000 to 1500 in their MFI values. While it is possible to meltblow structures
from polymers in the range of 100 to 400 MFI, the process yields larger bers and the
advantage of the process in forming ne bers is lost.
The process is well described in Figure 4.1. The polymer is fed to the die head by an
extruder. Upon exiting the die, the bers are attenuated by high-velocity air and collected
on a forming surface. The die assembly is the most critical part of this technology. The die
and the associated air knives for the die are depicted in Figure 4.2, which shows the crosssection of the die head. Plates 20 and 21 form chambers 22 and 23. The gap and the recess
of the air knives (identied as 11 and 12 the air knives control the angle and the gap
for the air) can be adjusted to control the air attenuation. The polymer melt ows into the
distribution reservoir and, via a well-insulated pipe attached to the inlet (17), is distributed to the capillaries (19). Compressed hot air, free of any particulate matter, at the
desired temperature is pumped into the chamber (22, 23) at a considerable but constant
pressure. The compressed hot air exits through the slits on both sides of the die tip and all
1
Hot Air
13
Drive
4
Die Head
3
Mat
10
Fibers
7
11
6
Extruder
5
12
Hot Air
Figure 4.1
14
Schematic of a typical meltblowing apparatus. Buntin (1976) (Assignee Prentice/Exxon).
92
92 [89--178] 10.12.2013 9:59PM
24
15
3
20
22
18
19
11
6
17
12
5
23
21
16
25
Figure 4.2
The die head cross-section. Buntin (1976) (Assignee Prentice/Exxon).
Figure 4.3
A high-speed photograph of meltblown bers. Courtesy of C.Y. Chang.
along its length at very high speed, which may reach close to the speed of sound, i.e. a
Mach number close to 1. The two air streams meet at a point on the axis of each orice,
resulting in an air stream moving at a reasonably high speed along the axis. This air
stream picks up the polymer melt stream from each orice and attenuates it into a very
ne ber, as it aps, cools and solidies in ambient air (Figure 4.3). Collectively, a thin
93 [89--178] 10.12.2013 9:59PM
4.1 Meltblowing processes
93
layer of bers impinges on a collection surface. As the bers approach the collecting
surface, they slow down, go through bending instability and buckling, touch one another
and form a bond, since they are still not fully solidied. Todays systems may comprise a
secondary air quench, or may even use a water mist to quench the bers. In the case of
hydrophobic polymer webs, sometimes hydrophilic agents are sprayed to incorporate
sorptive properties into the nal fabric. The process parameters in the system are:
temperature, pressure in the die, temperature of hot air, ow rate of hot air, offset of
the die tip and die-tip-to-collector distance. The material parameters are: molecular
weight of the polymer and its melting temperature. By controlling the process and
material parameters the average ber diameter can be controlled, typically in the range
210 m, and average ber lengths between 520 cm (in rare cases up to 50 cm) may
be obtained. The average ber diameter and length have a profound inuence on the
different properties of the resulting fabrics. The ber size depends on the MFI of the
polymer and throughput rates. Lower throughput rates, higher MFI or higher barrel
temperatures yield ner bers and softer fabrics. Throughput rates of 0.30.8 g hole 1
min1 are typical for higher-quality fabrics for battery separators, ltration media, etc.
Throughput rates of 0.83.0 g hole 1 min1 are typical for less-critical applications
such as oil sorbents, wipes, etc.
Accurate Products Co. was the rst to successfully build a 40 inch meltblowing die
(Buehning 1991) and Mitsui Petrochemical Industries obtained patents on the use of
capillary meltblowing dies (Mende and Sakai 1991, 1992). Many improvements/modications soon followed. For example, Kimberly-Clark patented a slot die for meltblowing to
minimize orice plugging (Appel et al. 1988). In the same year, Kimberly-Clark also
patented a process to produce an in-line perturbation of the attenuation air to form crimped
or uncrimped bers at reduced energy costs (Lau and Haynes 1998). Eckerd Schwartz of
Biax FiberFilm designed meltblowing equipment that uses multiple rows of orices with
co-centric air supply to provide higher productivity (Schwartz 1983, 1995). Chisso
Corporation developed the equipment to produce conjugate meltblown islands-in-the-sea
(Nishioi et al. 1994) and side-by-side (S/S) (Terakawa and Nakajima 1996) web types. The
side-by-side is commercially possible, but the islands-in-the-sea has not materialized. The
use of a modular die to produce mixtures of bers in the range 0.51 m was also patented
(Fabbricante et al. 2000). Reifenhauser (Reicol) remains the leader for signicantly
improving the meltblowing die design and the process (Rubhausen and Roock 1993).
Bicomponent meltblowing, while interesting, remains unexplored. Meltblowing is
primarily used in applications requiring bers that are small (25 m generally at high
throughputs and sub-micron at moderate throughputs). The bicomponent system does
not add much to this capability given the limited number of low-viscosity polymers
available on the market for meltblowing.
The initial applications of meltblown fabrics were focused on those needing smaller
bers and smaller pore sizes, such as battery separators, facemasks, respirators, cigarette
lters, and various air and liquid ltration media (Prentice 1978, Komatsu et al. 1988,
Kanno et al. 1992, Howard and Young 1993, Berger 1996, Shipp and Vogt 1987, Pall
(1996), Mozelack et al. 2003, Midkiff 2001). More recently, the meltblowing process has
also been adapted for forming meltblown adhesive webs intended for laminating various
94
94 [89--178] 10.12.2013 9:59PM
substrates (Kwok et al. 1999). Even though there is an ever-increasing body of patent
literature dealing with the meltblowing process and products, the basic principles are still
based on the original system commercialized by Exxon.
The work of Bresee and Ko (2003 and references therein) demonstrate that polymer jet
congurations in meltblowing are extremely transient and are already not straight several
centimeters from the hole exits in the die nosepiece, the jetjet interactions are signicant
and merging of neighboring jets is quite possible. Thermal regimes and polymer rheology have strong effects. Also, turbulence of the surrounding gas ow can have a very
signicant effect on the nal characteristics of meltblown nonwovens. However, it is
improbable that polymer jet characteristics in meltblowing are determined by turbulent
eddies alone. Indeed, the results of the experiments (Bresee and Ko 2003) show that the
initial sections of the jets in meltblowing (of the order of 4 cm) are straight, even though
the turbulent pulsations are the strongest in the surrounding high-speed gas ow. The jet
bending pattern is thus inconsistent with one driven solely by turbulent eddies. Therefore,
the rheological behavior of the polymeric liquid is expected to have a signicant effect on
jet dynamics. Also, the interaction with the surrounding gas ow cannot be reduced to
only the effect of turbulent eddies, and the interaction with the mean ow can be very
signicant. In addition, a realistic description of the dynamics of bending perturbations
should account for the interplay of the above-mentioned factors, which determine the
pattern of bending perturbation propagation over polymer jets.
A number of defects in meltblown nonwovens, and drawbacks of the technology in
general, are rooted in the process uid- and aeromechanics. Currently, so-called roping
(formation of streak-like structures in the webs and mats) is typically attributed to
uncontrolled turbulence. Such drawbacks of the process as so-called y (contamination
of the surrounding gas by short and extremely thin bers that do not connect to the ber
network) are generally attributed to blowing conditions being too violent. The appearance of so-called shots most probably results from excessively high temperatures, when
capillary breakup sets in.
4.2
Turbulence of surrounding gas jet

The turbulence structure in the submerged gas jet surrounding the polymer jet in
meltblowing can be elucidated by studying the dynamics of a exible solid threadline
suspended near the nozzle exit, as was done by Sinha-Ray et al. (2010a). A sketch of the
experimental setup is shown in Figure 4.4. It consists of a compressed gas (nitrogen at
2540 bar) cylinder used to create a turbulent blowing gas jet. High pressure tubing
leads to the nozzle (500 m inner diameter) issuing the gas jet. A sewing threadline
(mean diameter 110 m, straightened by ironing before the experiments) was attached to
the nozzle (Figure 4.4). Blowing was directed downward and the nozzle positioning was
adjusted using a moving stage with a ruler relative to the CCD camera, which was xed
(Figure 4.4). An electronic CCD camera (MotionScope Redlake Imaging Corporation)
was used for high-speed imaging. It was operated at 500 fps using a shutter speed of
1/85001/10000 s. When gas was blowing and the threadline was apping, high-speed
95 [89--178] 10.12.2013 9:59PM
95
4.2 Turbulence of surrounding gas jet
Ruler
High-pressure
tubing
Moving
stage
Light
source
Threadline
Gas
cylinder
CCD
camera
Figure 4.4
PC
Schematic of the experimental setup used to measure turbulent pulsations in a turbulent

submerged gas jet interacting with a exible threadline. Reprinted with permission from
Sinha-Ray et al. (2010a). Copyright 2010, AIP Publishing LLC.
0.3
Lateral displacement (cm)
0.6
0.4
0.2
0
0.2
0.4
0.2
0.1
0
0.1
0.2
0.3
0.4
0.6
0
(a)
Figure 4.5
0.5
1
Time (second)
0.5
1.5
(b)
0.5
1
Time (second)
1.5
Lateral displacements of the threadline at two locations: (a) x = 10.4 cm and at (b) x = 14.82 cm
from the nozzle. Reprinted with permission from Sinha-Ray et al. (2010a). Copyright [2010],
AIP Publishing LLC.
imaging was conducted for a number of segments of the threadline. At any new elevation
of the threadline suspension point, a delay of 5 min (sufcient for the transients to fade)
was adopted before data acquisition at that location. The images recorded with the highspeed CCD camera were processed using a MATLAB-R2007A interface. In addition, an
electronic DSLR Nikon D70s camera was used at a very high shutter speed (1/41/6 s) to
record the whole envelope of the apping threadline in a single still image.
As a basic variant, an L = 19.3 cm-long threadline was used with gas blowing at 35 bar
(263.9 m s1). The threadline was vigorously apping under such conditions. The time
series for the lateral threadline displacements at two locations are depicted in Figure 4.5.
96
96 [89--178] 10.12.2013 9:59PM
Two specic positions were chosen for the measurements: Figure 4.5a shows the
evolution at x = 10.4 cm and Figure 4.5b shows the evolution at x = 14.82 cm, with x
being the Cartesian vertical coordinate reckoned from the nozzle. It should be emphasized that the dc part of the lateral displacement, caused by a slight uncontrolled tilt of the
gas jet axis relative to the vertical, was subtracted from the data in Figure 4.5 using fast
Fourier transformation (FFT) of the recorded signal.
FFT of the time series for the lateral displacements of the threadline corresponding
to time intervals of 3 s (two shorter sections of the data of about 1.5 s are shown in
Figure 4.5) was done using the corresponding MATLAB routine. The absolute values
of the Fourier coefcients obtained are plotted against frequency in Figures 4.6a and
b. It can be seen that the spectral band covered was up to 250 Hz, with the spectrum
being continuous. Note that the recorded signal obtained from the CCD operated at
500 fps, as well as its processing with FFT, involving frequencies up to 250 Hz,
precluded probing frequencies higher than those in Figures 4.6a and b (up to 250 Hz).
However, a very signicant and gradual decrease in the signal amplitudes for the
high-frequency harmonics allows one to assume that there are no signicant higher
frequency modes. Indeed, Figures 4.6a and b show that above a frequency of about
167 Hz, the Fourier coefcients become negligible compared to those for frequencies
less than 167 Hz, and thus the importance of the entire high-frequency range in the
compound signal is expected to be small. This is illustrated in Figures 4.6c and d,
where the Fourier series for a truncated spectrum with frequencies less than 167 Hz
(and a cut-off of the spectral band with frequencies above 167 Hz) and the corresponding Fourier coefcients from Figures 4.6a and b are plotted versus experimental
data. It can be clearly seen that the experimental data are reproduced fairly well, in
spite of the truncated spectrum.
It should be emphasized that the unperturbed threadline is vertical in the experimental
setup shown in Figure 4.4, i.e. potentially represents a kind p
of
a pendulum. The
eigenfrequency of the pendulum-like oscillations p 21 g=L, with g being
gravity acceleration and the factor (2)1 included for comparison with the frequency
in Figures 4.6a and b. Taking the pendulum length within the range L 120 cm (since
any part of the threadline can be excited), we nd p 5 HZ. These frequencies are too
low to cause resonances with the oscillations in Figures 4.5 and 4.6, and thus can be
excluded from consideration.
The character of the time series for lateral displacements of the threadline depicted
in Figure 4.5 can also be elucidated using the corresponding autocorrelation functions.
The latter were calculated using standard a MATLAB routine and are plotted in
Figure 4.7. It can be clearly seen that the autocorrelation function rapidly decays and
approaches zero a clear sign of a chaotic process.
The threadline is subjected to turbulent eddies of the surrounding turbulent gas jet.
If we denote the exit velocity of the gas jet by Ug(0), the nozzle diameter by d0 and the
kinematic viscosity of gas by g, and dene the Reynolds number as Red = Ug(0)d0/g,
then the turbulence Taylor microscale in the axisymmetric gas jet, according
1=2
to Antonia et al. (1980,) is equal to = 1.23Red x. If we take as an estimate
1
Ug(0)=230 m s and d0= 0.05 cm then, for air, Red = 7667. Therefore, at x = 1 cm,
97 [89--178] 10.12.2013 9:59PM
97
Absolute value of coefficients (cm)
Absolute value of coefficients (cm)
101
102
103
104
0
100
Frequency (Hz)
(a)
0.15
0.1
0.15
0.2
0.25
(c)
Figure 4.6
0.1
0.15
Time (second)
200
0.3
0.05
0.05
100
Frequency (Hz)
(b)
104
0.05
0.3
103
105
200
Calculated signal
Experimental
0.1
102
0.2
0.1
0
0.1
0.2
0.3
0.4
0.2
(d)
Calculated signal
Experimental
0.05
0.1
0.15
Time (second)
0.2
Absolute values of the Fourier coefcients corresponding to the lateral displacement of the
threadline at (a) x = 10.4 cm (Figure 4.5a) and (b) x =14.82 cm (Figure 4.5b). (c) Lateral
displacement at x =10.4 cm reconstructed using the Fourier series with a truncated spectrum.
The calculated displacement is shown by symbols, the experimental data by a solid line. (d) Same
as (c) but at x =14.82. Reprinted with permission from Sinha-Ray et al. (2010a). Copyright 2010,
AIP Publishing LLC.
turbulent eddy scales are of the order of = 0.014 cm, and at x = 10 cm, they are of the
order of = 0.14 cm. The corresponding frequencies are of the order of = Ug(0)/ (Pope
2000), which is 105106 Hz, well above the frequency band characteristic of the threadline oscillations (Figure 4.6). On the other hand, the large eddies in the system gas jet/
threadline are of the order of L = 10 cm, and their frequencies = Ug(0)/L are 103 Hz.
This is still signicantly higher than the frequency band in Figure 4.6.
Local interactions of random, high-frequency (compared to those of the threadline)
turbulent eddies in gas with a massive threadline resembles those responsible for
Brownian motion. In the latter case, multiple simultaneous tiny molecular impacts in
98
98 [89--178] 10.12.2013 10:00PM
1
0.9
0.8
Autocorrelation
Autocorrelation
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0.1
0.2
(a)
Figure 4.7
102
101
Time lag (Second)
100
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0.1
0.2
0.3
(b)
102
101
Time lag (Second)
100
Autocorrelation function for the time series for the lateral displacements of the threadline
in Figures 4.5a and b at (a) x = 10.4 cm and (b) x =14.82 cm. Reprinted with permission from
different directions produce macroscopic displacement of massive (sub-micron) particles

on timescales much larger than those of the individual impacts. In the turbulence context,
this idea was developed by G.I. Taylor. In particular, a lateral averaged displacement in
time t produced by turbulent eddies is <A> = [2<v2>t]1/2 (Antonia et al. 1980, Frost
and Moulden 1977), where <v2> is the correlation of the lateral velocity pulsation and
the characteristic timescale of large eddies. In turbulent jets, all the pulsation velocity
components, in particular, v (lateral) and u (longitudinal) are of the same order.
Thus, <v2> = <uv>. The characteristic timescale of large eddies is u=y1 ,
where u is the mean longitudinal velocity component and y is the lateral coordinate.
Therefore, 5v02 > 5u0 v0 >= u=y t , where t is the kinematic eddy viscosity,
which here plays the role of the diffusion coefcient. In axisymmetric turbulent jets the
kinematic eddy viscosity is constant over the jet and with good accuracy is given by t =
0.015Ug(0)d0 (Yarin 2007). Turbulent eddies can push a threadline element at a certain
location only for a time of the order of the characteristic time r of propagation of
a bending perturbation over the threadline (which plays the role of relaxation time in
this case). After that, the excursion leaves the previous location and propagates along
the threadline. Therefore, we can estimate the amplitude of the threadline lateral displacements as <Athreadline> = (2tr)1/2. The speed of propagation of bending perturbations
along a stretched threadline is of the order of [P/(S)]1/2, where P is the tensile force in the
threadline cross-section, S is the cross-sectional area (xx = P/S is the longitudinal stress
in the threadline cross-section) and the threadline density. Therefore, the characteristic
time of propagation of a bending perturbation over a threadline is r = L[P/(S)]1/2. The
tensile force P is evaluated as qL, where q is the longitudinal air drag imposed on
the unit length of the threadline. An appropriate expression for the drag force is given
by (Ziabicki 1976, Ziabicki and Kawai 1985)
99 [89--178] 10.12.2013 10:00PM
q 0:65a0 g U20

2U0 a0 0:81
g
99
4:1
where g and g are the gas density and kinematic viscosity, respectively, U0 is the gas
velocity relative to the unperturbed threadline and a0 = d0/2 is the cross-sectional radius
of the threadline.
Taking for the estimate L = 10 cm, S = S d 20 =4 1:96 103 cm2 (for d0 =
0.05 cm), U0 = Ug(0) = 230 m s1 and = 0.786 g cm3, we nd for air t = 0.015Ug(0)d0 =
17.25 cm2 s1, q= 23.5 g s2 and r = 0.0256 s (the value is in very good agreement with
39 Hz).
the main threadline frequencies in Figure 4.6, which are of the order of 1
r
Therefore, <Athreadline> = 0.94 cm, in reasonable agreement with the experimental data in
Figure 4.5. This result conrms that the lateral threadline oscillations are imposed by
multiple impacts of large turbulent eddies and are relaxed by propagation of bending
perturbations along the threadline.
Another parameter of interest is the length of the threadline where apping is the most
vigorous. This requires a denition of vigorous apping in the experiment. Two different
denitions and methods based on them can be considered. In the rst, the images of the
free end of the threadline are analyzed using CCD records. The threadline cross-sections
from which large lateral displacements of the free end were visible are denoted by arrows
in the images (Figure 4.8).
The apping region depicted in Figure 4.8 shows that propagation of bending perturbations along the threadline serves as a relaxation mechanism for lateral oscillations
imposed by turbulent eddies only in the strongly stretched part at the beginning of the
threadline, which supports practically the entire air drag force. On the other hand, the free
end of the threadline is almost unloaded and imposes weak restrictions on the lateral
excursions driven by turbulent eddies. Moreover, bending perturbations entering from
the upper part of the threadline can amplify with practically no restriction, delivering their
entire energy parcel into such lateral motion. This is one of the reasons for the strong
amplication of bending perturbations close to the free end, which results in apping.
Another reason for strong amplication will be the appearance of a distributed lift force,
which also diminishes the restricting effect of the threadline tension and is discussed in
detail in Section 4.3.
Another method of determining a vigorously apping part of the threadline can be
based on the threadline envelope photographed by a DSLR Nikon D-70s camera
(Figure 4.9). This method is capable of observing almost the entire threadline length
(as opposed to the rst method based on CCD images), which allows easy recognition of
the apping part of the threadline (Figure 4.9).
Comparison between the results of the two methods for measuring the apping part
reveal the following. In an experiment with a 18.5 cm-long threadline subjected to
blowing at 35 bar (263.9 m s1), the apping length found by averaging the data recorded
by the CCD camera was 1.82 cm. On the other hand, application of the method shown in
Figure 4.9, based on the data recorded by a DSLR camera revealed the length of the
apping end of the threadline to be 1.76 cm. Therefore, the results obtained by the two
100 [89--178] 10.12.2013 10:00PM
(a)
(b)
1 cm
1 cm
(c)
1 cm
Figure 4.8
Flapping of a 19.3 cm-long threadline, which was subjected to a gas jet issued at 35 bar,
recorded by CCD. The arrows show the beginning of the apping part of the threadline.
Panels (a)(c): three different snapshots from the same experiment are shown to illustrate the
different congurations recorded. Reprinted with permission from Sinha-Ray et al. (2010a).
A
Flapping Length
1 cm
Figure 4.9
Determination of the apping part of the threadline. The beginning of the apping section is
indicated by the inclined arrows. Experiment with a 19 cm-long threadline subjected to a gas jet
issued at 28 bar. Reprinted with permission from Sinha-Ray et al. (2010a). Copyright 2010, AIP
Publishing LLC.
101 [89--178] 10.12.2013 10:00PM
101
different methods are sufciently close. The slight difference in the values by the two
methods can be attributed to the following reasons: (i) the pixel resolution of the images
obtained by the DSLR camera is not sharp enough; (ii) due to the three-dimensional
nature of apping, the actual beginning cross-section of the apping part might be off by
a few pixels from the one in the images; (iii) in spite of the fact that the threadlines were
ironed and straightened prior to the observations, they always possessed some local nonzero curvatures along their length; (d) a certain inhomogeneity of the threadlines.
The method illustrated in Figure 4.9 was applied to determine apping lengths of
threadlines of different lengths subjected to the same blowing speed. The experiment
began with a 21 cm-long threadline subjected to a parallel gas jet ow issued at 35 bar
(263.9 m s1). After the transient effects had faded, the corresponding apping length was
measured as in Figure 4.9. Blowing was then ceased and the threadline was cut shorter.
Blowing was turned on once again and the corresponding apping length was measured.
After that, the procedure was repeated until a threadline length of 10.1 cm was reached.
The results are shown in Figure 4.10a. It can be seen that the experimental points are
clearly subdivided into two groups one for shorter threadlines (10 to 16 cm long) and
the other for longer ones (16 to 21 cm long). In the shorter group the data suggest some
variation of the apping length Lf versus threadline length L, albeit that might be caused
by data scatter. Note that the shorter threadlines might be signicantly affected by such
end effects as shedding vortices forming a von Karman street, as suggested in Argentina
and Mahadevan (2005). On the other hand, in the longer group the apping length is
practically independent of the threadline length L.
Using the well-known analytical solution for axisymmetric turbulent gas jets
(Abramovich 1963, Yarin 2007), it is possible evaluate the ratio ugA/Ug(x), where ugA
is the gas velocity value at points A and A in Figure 4.9, and Ug(x) is the local axial
velocity of the gas jet. The results are shown in Figure 4.10b. They demonstrate that
the part of the threadline preceding the apping part is located within the core of the
surrounding gas jet corresponding to values of ugA/Ug(x) between 0.78 and 0.87, i.e. to
gas velocity values of 7887% of the local axial gas jet velocity in cross-section AA.
0.95
Group 2
0.9
1.6
1.4
ugA/Ug(x)
Flapping length (cm)
2
1.8
Group 1
1.2
0.85
0.8
0.75
1
0.8
8
(a)
Figure 4.10
10
12
14
16
18
Length of thread (cm)
20
0.7
22
8
(b)
10
12
14
16
18
Length of thread (cm)
20
22
(a) Flapping length versus threadline length. (b) Gas velocity at points A and A in Fig 4.9 relative
to the local axial velocity value Ug(x) for different threadlines. Reprinted with permission from
0.95
0.45
0.4
y = 2.4246x 5.4943
0.9
0.35
0.3
0.25
ugA/Ug(x)
Log (flapping length)
102
102 [89--178] 10.12.2013 10:00PM
y = 1.0566x 2.2663
0.2
1.15
0.1
0.05
0
2.25
(a)
Figure 4.11.
0.85
0.8
0.75
2.3
2.35
Log (velocity)
2.4
0.7
150
2.45
(b)
170
190
210
230
250
270
290
Nozzle exit velocity (m/s)
(a) Flapping length versus the gas jet velocity at the nozzle exit. (b) Gas velocity at points A and A
in Figure 4.2.6 relative to the local axial velocity value Ug(x) for different blowing velocities.
Reprinted with permission from Sinha-Ray et al. (2010a). Copyright [2010], AIP Publishing LLC.
In an additional experiment a 19 cm-long threadline was subjected to gas jets issued

at different pressures from 24 to 40 bar with a step of 2 bar (blowing velocities
Ug(0) =181.6274.1 m s1) and the apping length was measured using the method of
Figure 4.9. The results are depicted in Figure 4.11a on logarithmic scales. They reveal
that the scaling Lf ~ Ug(0) at Ug(0) 230 m s1. At Ug(0) > 230 m s1 the scaling changes
to Lf ~ Ug(0)2.42. The values of the ratio ugA/Ug(x) in the cross-sections corresponding to
points A and A in Figure 4.9 for all different blowing velocities are shown in
Figure 4.11b. They demonstrate that the part of the threadline preceding the apping
part is located within the core of the surrounding gas jet corresponding to ugA/Ug(x) in the
range 0.75 to 0.95, i.e. to gas velocity values of 7595% of the local axial gas jet velocity
in cross-section AA.
4.3
Bending and flapping of flexible solid threadlines in a gas jet

Consider a solid exible threadline subjected to parallel high-speed gas ow. The
threadline is assumed to be non-stretchable and with negligible bending stiffness. Its
cross-sectional radius a0 is constant. In the unperturbed state the threadline is straight
and aligned in the direction of ow, and does not move. The force balance in the
threadline in this state reads:
dP
q 0
dx
4:2
where P = xxa20 is the longitudinal force acting in the threadline cross-section (with xx
being the normal stress) and q is the distributed aerodynamic drag force acting on a unit
length of the threadline. An appropriate expression for the drag force is given by Eq. (4.1)
in Section 4.2. The longitudinal axis Ox is directed along the unperturbed threadline,
with x = 0 corresponding to a certain initial cross-section, which is discussed below.
The normal stresses in the threadline cross-sections are of elastic origin. Since the
threadline stretchability is assumed to be very low (which means that its Youngs
103 [89--178] 10.12.2013 10:00PM
4.3 Bending and flapping
103
modulus is sufciently high), the static (and, further on, dynamic) threadline elongation
is assumed to be negligible, even though the stress xx might be large.
The threadline length is L, and at the free end x = L the normal stress vanishes,
i.e. xx jxL 0. Then, Eq. (4.2) is integrated to yield the normal stress distribution along
the unperturbed threadline
xx
q L x
a20
4:3
In particular, the largest stress is achieved in the initial cross-section of the threadline
at x = 0, which is xx0 = xx(0) = qL/(a20 ), since it supports practically the entire
aerodynamic drag applied to the threadline.
The solid exible threadline is subject to bending perturbations, which can result in
lateral threadline motions due to the appearance of the normal component qn of the
aerodynamic force per unit threadline length related to the mean ow (see Eqs. 3.17 and
3.18 in Section 3.2; Weber 1931, Debye and Daen 1959, Yarin 1993). This component is
of inviscid origin and resembles a distributed lift force, which originates from the
unsteady Bernoulli equation for gas motion (Loitsyanskii 1966, Landau and Lifshitz
1987, Batchelor 2002). Obviously, it is not alone, since the unsteady Bernoulli equation
also produces a force component associated with the added mass. However, in highspeed ows the lift force is dominant compared to the added-mass effects, since only it is
proportional to U20 , whereas the additional force components are linear in U0. It should
be emphasized that we do not include in this force the effect of the turbulent eddies in the
gas ow, which will be accounted for separately.
The distributed lift force is proportional to the local curvature of the threadline axis, and
in the linear approximation (for small bending perturbations) is given by (cf. Eq. 3.17):
qn g U20 a20
2 H
x2
4:4
where = (x,t) represents the curved threadline axis. Note that planar bending is
considered here for simplicity, while generalization to three-dimensional bending is
straightforward (cf. Eq. 3.17).
It is emphasized that the distributed lift force of Eq. (4.4) is responsible for the bending
instability of liquid jets having sufciently high relative velocity to the surrounding gas
discussed in Section 3.4. The distributed drag and lift forces acting on a perturbed
threadline subjected to a gas ow are shown schematically in Figure 4.12. Random
forces imposed by turbulent eddies are also shown for completeness.
The linearized lateral momentum balance for the perturbed threadline in a high-speed
gas ow reads
a20
Vn
kP qn
t
4:5
where is the threadline density, V n H=t is the lateral local velocity of the threadline
and k 2 H=x2 is the local curvature.
104
104 [89--178] 10.12.2013 10:00PM
Lift
Lift
Lift
Drag
Drag
Drag
Drag
Lift
Figure 4.12
Drag
Lift
Drag
Random
force from
Lift large eddies
Distributed forces acting on perturbed threadlines (and polymer jets) subjected to high-speed gas ow.
Reprinted with permission from Sinha-Ray et al. (2010a). Copyright 2010, AIP Publishing LLC.
Accounting for Eq. (4.4), Eq.(4.5) takes the following form:

h
i
2
2
U
x
xx
g
0
H
2 H
0
2
t
x2
4:6
where xx(x) is given by Eq. (4.3).

Equation (4.6) is rather peculiar. It shows that with regard to the lateral motion of
the threadline, the distributed lift force diminishes the effect of stretching imposed
by the aerodynamic drag. Correspondingly, if xx0 > gU20 , this equation is hyperbolic
at 0 x x*, where xx > gU20 , and elliptic at x* x L, where xx < gU20 , since,
according to Eq. (4.3), the normal stress decreases along the threadline. The transition
cross-section x*, according to Eq. (4.3), is determined by the condition
g U20
q L x
0
a20
which yields the length of the elliptic (apping) part of the threadline Lf as

a0 2U0 a0 0:81
Lf L x
0:65
g
4:7
4:8
Equation (4.8) shows that the length of the elliptic part of the threadline does not
depend on its length L. On the other hand, it increases with blowing speed as U0:81
0 .
The fact that Eq. (4.8) changes type at x = x* makes it similar to the Tricomi equation
familiar in gas dynamics (Courant and Friedrichs 1948), albeit different from it. The
general solution of Eq. (4.8), with xx given by Eq. (4.3), is readily found by the method
of characteristics as
2x
3
dx
Hx; t 4
t5
q L x=a20 g U20 =
0
2
3
4:9
x
dx
t5
F4
q L x=a20 og U20 =
0
where and F are arbitrary functions.
105 [89--178] 10.12.2013 10:00PM
105
Perturbations on the threadline are imposed by turbulent eddies in the surrounding

turbulent gas jet. The experimental data discussed in Section 4.2 show that signicant perturbations of the threadline conguration are visible at some distance from
the suspended end of the threadline. In the rst approximation, it is assumed that at a
certain relatively small distance from the suspended end perturbations of frequencies
and amplitudes corresponding to the local turbulent pulsation structure of the mixing
layer in the gas jet are imposed on the threadline, whereas down the threadline no
new perturbations are imposed, since the gas jet widens. This cross-section of
the threadline where bending perturbations are imposed is considered to be the
initial one, and x = 0 there. In this cross-section it is assumed that the threadline
is subjected to small lateral perturbations generated in the mixing layer of the gas
jet, so that
Hjx0 H0 expit;
H=xjx0 0
4:10
where 0 (which is a complex number in the general case) and real are the corresponding mode amplitude and frequency, and i is the imaginary unit.
The boundary conditions (Eq. 4.10) fully determine the functions and F in the
hyperbolic part and reduce Eq. (4.9) to the following threadline conguration:
Hx; t H0 expit cosIx
4:11
The real function I(x) results from the evaluation of the integrals in Eq. (4.9) as
(
1=2
1=2 )
2a20
q L
q L x
2
2
Ix
g U0 =

g U0 =
4:12
q
a20
a20
Note that
2a20
I0 0; Ix
q

1=2
q L
2
g U0 =
a20
4:13
The physically relevant solution for the threadline shape in the hyperbolic part is given
by the real part of Eq. (4.11).
On the other hand, in the elliptic part the characteristics are imaginary, and Eq. (4.9)
reduces to
Hx; t expitfA expiIx expJx B expiIx expJxg
4:14
where A and B are constants and
"
#

2a20 q x q L= a20 g U20

1=2
Jx
q
a20
Note that
4:15
106
106 [89--178] 10.12.2013 10:00PM
0.16
0
0.14
0.005
0.12
0.01
0.015
0.08
0.1
0.06
0.02
0.04
0.025
0.02
0.03
0.035
0
(a)
Figure 4.13
0.1
0.2
0.3
0.4
x
0.5
0.6
0.7
0.1
0.2
0.3
0.4
0.5
0.6
0.7
(b)
(a) and (b) Two snapshots of the threadline shape corresponding to the case where
R = g/ = 0.00122, L/a0 = 100, the Reynolds number Re = 2U0a0/g = 40, H0 = 0.01
and the dimensionless perturbation frequency = L/U0 = 0.01 (for L = 10 cm and
U0 = 104 cm1, = 10 Hz) at two different time moments. The perturbation amplitude H
and the longitudinal coordinate x are rendered dimensionless by L. Reprinted with permission
from Sinha-Ray et al. (2010a). Copyright 2010, AIP Publishing LLC.
2a20 g U20
Jx 0; JL
q
!1=2
4:16
The constants are found via matching of Eq. (4.14) to Eq. (4.11) at x = x*, which yields
A B H0 =2. Then, the threadline shape in the elliptic part is given by
Hx; t H0 expitfcoshJxcosIx i sinhJxsinIx g
4:17
As with the hyperbolic part, the physically relevant result for the elliptic part is
given by the real part of Eq. (4.17). It should be emphasized that since
dI=dxjxx dJ=dxjxx , the transition from the hyperbolic part to the elliptic
part occurs in the cross-section x = x*, where dH=dxjxx , i.e. the threadline is
oriented perpendicularly to the gas ow.
Four dimensionless snapshots of the threadline shape calculated using Eqs. (4.11),
(4.12), (4.15) and (4.17) are plotted in Figure 4.13. It can be seen that the wavelength
of the standing perturbation wave in the hyperbolic part becomes shorter as it
approaches the transition cross-section x = x* (= 0.695 in the present case). In the
elliptic part, after the transition, the perturbation amplitude increases very rapidly,
which rapidly violates the assumption of small perturbations. A fully realistic prediction
of the threadline shape in this part can be achieved only in the framework of fully
nonlinear theory. However, a practical approximation is probably one where the calculation is truncated at a cross-section x > x*, where the perturbed threadline length reaches
a value of the order of L (i.e. 1 in the dimensionless case).
107 [89--178] 10.12.2013 10:00PM
107
The explosive behavior of the solution in the elliptic part is a clear manifestation
of the fact that the elliptic problem is solved as an initial value problem. This results
from the matching condition imposed only at one side of the interval at the transition
from the hyperbolic part at x = x*. That means that in the elliptic part we are dealing
with a Hadamard-like ill-posed problem, which is inevitably prone to fast perturbation
growth (Tikhonov and Arsenin 1977). Situations where physical problems reduce to
ill-posed Hadamard-like problems are not unique. For example, the well-known capillary Rayleigh instability, if treated in the quasi-one-dimensional, linearized approximation for small perturbations (Yarin 1993) immediately reveals an Hadamard-like
initial-value problem for the Laplace equation with surface-tension-related regularization for the short waves.
Strong amplication of bending perturbation waves as they pass from the hyperbolic
(stretched) to the elliptic (almost unstretched) part of the threadline is partially related
to the fact that the wave brings with it a parcel of mechanical energy delivered to it by
turbulent eddies in the initial cross-section x = 0. In the almost unloaded part, this
energy is released as kinetic energy of lateral excursions unrestricted by longitudinal
stretching and manifesting themselves as apping. Also, the distributed lift force strongly
amplies bending perturbations in the unloaded part of the threadline. This effect also
signicantly contributes to apping.
In reality, turbulent pulsations can affect the threadline not only at the initial crosssection, but over a certain length. Considering the effect of distributed turbulent pulsations
on threadlines separately from the effect of the distributed lift force, we apply the formula
for the oscillation magnitude <Athreadline> derived in Section 4.2, accounting for the
fact that the restrictive longitudinal tension diminishes in the x-direction. Therefore,
<Athreadline(x)> = [2tr(x)]1/2, where t = 0.015U0d0, r(x) = L[xx(x)/)]1/2 and xx(x)
is given by Eqs. (4.1) and (4.3). As a result, we nd the threadline envelope as
!1=4

U0 d0 0:2025 d0 1=4 p
1
5Athreadline x> 0:16
d0 L
4:18
g
L
g
1 x1=4
where x is rendered dimensionless by L. The result shows that the shape of the threadline
envelope should follow the dependence <Athreadline> ~ (1x)1/4 if only turbulent pulsations
are accounted for, whereas distributed aerodynamic lift force is disregarded.
Consider comparison of the theoretical predictions of this section with the experimental data. The results for longer threadlines in Figure 4.10a in Section 4.2 show that the
length of the apping region Lf does not vary with the threadline length L. This is in
agreement with the theoretical prediction of Eq. (4.8).
The experimental data in Figure 4.11 in Section 4.2 reveal the scaling Lf ~ Ug(0) at
Ug(0) 230 m s1, which is close to the scaling Lf ~ Ug(0)0.81 expected from Eq. (4.8). On
the other hand, at Ug(0) > 230 m s1 the scaling changes to Lf ~ Ug(0)2.42, which deviates
signicantly from the predictions of Eq. (4.8). A much stronger dependence of the
aerodynamic drag on the gas jet velocity at such high values of Ug(0) most probably
corresponds to a strong increase in the turbulent eddy viscosity at higher blowing rates.
The experiments described in the book edited by Ziabicki and Kawai (1985) that
108
108 [89--178] 10.12.2013 10:00PM
Calculated profile
1 cm
X
Figure 4.14
The envelope observed in the experiment with a 19 cm-long threadline subjected to a gas jet
issued at 28 bar versus the prediction accounting for only the effect of turbulent eddies, and
disregarding distributed aerodynamic lift force. Reprinted with permission from Sinha-Ray et al.
(2010a). Copyright 2010, AIP Publishing LLC.
established the expression for the aerodynamic drag (Eq. 4.2) used to derive Eq. (4.8)
were conducted with threadlines moving in stagnant air, which does not involve the high
levels of turbulent eddy viscosity found in blowing jets. Therefore, deviations from
Eqs. (4.1) and (4.8) are expected at higher blowing velocities, and it is remarkable that
these equations work rather accurately up to 230 m s1.
The comparison of the prediction of the effect of the distributed turbulent pulsations
to the experimentally observed threadline envelope shown in Figure 4.14 reveals the
following. In the upper, hyperbolic part of the threadline, where bending is strongly
restricted by longitudinal stretching, and the aerodynamic lift cannot be large, the
envelope prole is rather accurately described by the predicted dependence of
Eq. (4.18) <Athreadline> ~ (1x)1/4. On the other hand, in the lower, elliptic part of
the threadline the effect of turbulent eddies alone is insufcient to describe the observed
apping amplitudes, which are mostly determined by the distributed aerodynamic lift force.
In summary, the experimental and theoretical study of threadline blowing in Section 4.2
and this section revealed the following process mechanism. Large turbulent eddies in the
high-speed gas jet with eddy frequencies of the order of 103 Hz impact the threadlines,
introducing bending perturbations. It takes multiple rapid eddy impacts on a massive
threadline to cause a visible lateral excursion. The bending amplitude achieved locally
by the eddy bombardment is controlled by a much slower (with frequencies of about
10100 Hz) propagation of bending perturbations along a stretched (by the aerodynamic
drag) threadline. Bending perturbations propagate along the threadline and reach lower
regions that are practically unstretched/unloaded, since they do not support the entire
aerodynamic drag imposed on the threadline. There, the restrictions on large-amplitude
bending imposed by strong stretching are removed. Then, turbulence energy stored in
the bending perturbations of the threadline can be released as the kinetic energy of a further
bending. Moreover, bending perturbations of the unloaded threadlines are strongly
109 [89--178] 10.12.2013 10:00PM
4.4 Aerodynamically driven stretching
109
amplied by the distributed aerodynamic lift force. As a result, the entire region close to the
free unloaded threadline end experiences apping of large amplitude. Blown threadlines
are capable of simultaneously extracting kinetic energy from both the mean ow and the
turbulent pulsations in it.
4.4
Aerodynamically driven stretching of polymer jets

In the case of a polymeric liquid jet stretched by a surrounding parallel high-speed gas jet,
ow in the former should be calculated differently from the case of a solid exible
threadline considered in Section 4.3. Steady-state ow in an unperturbed straight polymer jet is governed by the quasi-one-dimensional continuity and momentum balance
equations following from the general equations for liquid jets (Eqs. 3.1 and 3.2)
dfV
0; f a2
dx
4:19
dfV2 dxx f
q
dx
dx
4:20
where f(x) and a(x) are the cross-sectional area and radius, respectively, and V(x) is the
unperturbed absolute axial velocity of the polymeric liquid in the jet. In Eq. (4.20) surface
tension and gravity effects are neglected as insignicant compared to the dominating
inertial and internal (rheological) stresses (xx is the longitudinal stress) and the aerodynamic drag q (per unit jet length). As in Section 4.3, the effect of turbulent pulsations in
gas is considered to be localized at the initial cross-section where they are assumed to
impose bending perturbations (in the perturbed case considered in Sections 4.5 and 4.6).
The stretching aerodynamic drag force q depends on the relative velocity between the
gas stream and the polymer jet (Ug V). Therefore, similarly to Eq. (4.1), the following
expression is used in this case to calculate the local value of the drag force q(x)
0:81

2 2 Ug V a
q 0:65ag Ug V
4:21
g
where a is the cross-sectional radius, g and g are the gas density and kinematic
viscosity, respectively.
Equation (4.19) is readily integrated, which yields
a2 V a20 V0
4:22
where subscript zero designates the cross-sectional radius and longitudinal velocity
values at the initial cross-section. This expression allows one to exclude the crosssectional radius from consideration, as is done below.
In thin liquid jets the normal stress in the cross-section can be always presented as a
difference between the normal and radial deviatoric stresses, xx and yy, respectively,
i.e. as xx = xx yy (see Sections 2.2 and 2.3). The deviatoric stresses are related to the
110
110 [89--178] 10.12.2013 10:00PM
ow kinematics via a rheological constitutive equation (RCE). For polymeric liquids

experiencing strong uniaxial stretching as, for example, in meltblowing, an appropriate
RCE is the viscoelastic upper-convected Maxwell model (UCM), which is substantiated
by direct statistical consideration of macromolecular stretching and the corresponding
entropic elasticity (Sections 2.3 and 2.4; also Yarin 1993). In the present case, the RCE of
the Maxwell model reduces to the following axial and radial (lateral) projections:
V
dxx
dV
2 dV xx
2
xx

dx
dx
dx
4:23
dyy
dV
dV yy

yy

dx
dx
dx
4:24
where and are the liquid viscosity and relaxation time, respectively.
Combining Eqs. (4.20)(4.24), we can transform them into the following system of
dimensionless equations:

Exx yy =DeV2 q
dV

4:25
dx
1 Exx 2yy 3=V2

dxx
1
dV
dV xx
xx 2

2
dx
dx
dx De
V
4:26

dyy
1
dV
dV yy
yy

dx
dx
dx De
V
4:27
h
i 0:81
q 0:65R= V 1=2 Ug V 2 ReUg V = V1=2
4:28
where
The equations are rendered dimensionless by the following scales: V0 for V and Ug, the
distance between the initial cross-section and deposition screen L for x, a0 for a, / for
xx and yy, and further on, L/ V0 for time t. The primary dimensionless groups involved
in Eqs. (4.25)(4.28) are given by
R
g
L
2V0 a0
V0
; ; Re
; De
g
L
a0
4:29
with Re and De being the Reynolds and Deborah numbers, respectively; the secondary
dimensionless groups are
E
g
2R
; M
De Re M
In Eq. (4.30) g denotes gas viscosity.
4:30
111 [89--178] 10.12.2013 10:00PM
4.4 Aerodynamically driven stretching
111
The system of three ordinary differential equations (Eqs. 4.254.27) is subject to the
following dimensional conditions at the initial cross-section of the polymer jet:
x 0: V 1; xx xx0 ; yy 0
4:31
The fact that all boundary conditions for Eqs. (4.25)(4.27) can be imposed at x = 0 stems
from the hyperbolicity of this system of equations, which holds if the dimensional initial
velocity V0 is larger than the dimensional speed of the elastic sound (xx/)1/2 (Entov
and Kestenboim 1987, Joseph 1990, Yarin 1993). Accounting for Eq. (4.31), the latter
corresponds to the following dimensionless condition
1 > Exx0
4:32
This means that even though polymeric liquids can develop rather signicant longitudinal deviatoric stresses in the ow inside the die, and carry a signicant part of it as xx0 to
the initial cross-section (Han et al. 2008), the convective effects in the polymeric jet are
initially stronger than propagation of the elastic sound. Therefore, the information in
such a jet is convected downstream, even though the elastic sound can propagate not
only down, but also upstream (i.e. against the ow, but swept by it). In such cases all
boundary conditions are imposed at the beginning of the polymer jet, as in Eq. (4.31).
The system of equations (Eqs. 4.254.27) subject to the conditions (Eq. 4.31)
was solved numerically using the KuttaMerson method by Yarin et al. (2010). The
solution obtained for the unperturbed jet is illustrated in Figures 4.15 and 4.16 for
the following values of the parameters: M = 0.001, R = 0.00122, = 83000, De = 0.01,
Re = 40, the dimensionless velocity of the gas ow, assumed to be constant in the present
10
9
8
7
VTAU a
6
5
4
3
2
1
0
Figure 4.15
0.25
0.5
x
0.75
The unperturbed distributions of the longitudinal velocity (solid line) and cross-sectional radius
(dashed line) along a polymer jet stretched by high-speed gas jet. Reprinted with permission from
Yarin et al. (2010). Copyright 2010, AIP Publishing LLC.
112
112 [89--178] 10.12.2013 10:00PM
10
0.3
(a)
(b)
0.25
0.2
0.15
0.1
6
TAUyy
TAUxx/1000
5
4
0.05
0
0.05
0.1
0.15
0.2
0.25
0
0
Figure 4.16
0.25
0.5
x
0.75
0.3
0.25
0.5
x
0.75
The unperturbed distributions of the longitudinal (a), and lateral (b) deviatoric stresses
(xx and yy, respectively) along polymer jet stretched by high-speed gas jet. Reprinted with
case, is Ug = Ug(0) = 10 and xx0 = 104. These values of the dimensionless groups correspond
to the model values of the physical parameters partially taken from Marheineke and Wegener
(2007): a0= 3 103, L = 250 cm, g = 1.22 103 g cm3, g = 0.15 cm2/s, V0 = 103cm s1,
the dimensional Ug(0) = 104 cm s1 and = 0.25 102 s. A more realistic set of parameters
will be used in Section 4.7.
Figure 4.15 depicts the unperturbed longitudinal velocity V(x) and radius a(x)
distributions, and Figures 4.16a and b the unperturbed distributions of the longitudinal
and lateral deviatoric stresses, xx(x) and yy(x), respectively. It can be seen that the
polymer jet is gradually accelerated by the aerodynamic drag imposed by the gas stream,
and simultaneously thins (Figure 4.15). The longitudinal deviatoric stress xx, which is
rather high at the die exit due to the prior strong stretching in the die channel, decreases
along the jet because the elongation rate due to gas ow is insufciently high to overcome
the viscoelastic relaxation (so-called weak ow; see Figure 4.16a). Comparison of
Figures 4.16a and b shows that the lateral deviatoric stress yy is negligibly small
compared to the lateral, xx, as is the case in the uniaxial elongational ows.
4.5
Aerodynamically driven bending instability of polymer

jets: linearized theory
Bending perturbations of polymer jets stretched by a high-speed gas jet are studied using
the quasi-one-dimensional equations of the dynamics of free liquid jets (see Section 3.1).
In the momentless approximation (neglecting the bending stiffness and thus, the entire
Eq. 3.3) and assuming small bending perturbations (linearizing), one can obtain from
Eqs. (3.1) and (3.2) the normal projection of the momentum balance equation in the
following dimensional form:
113 [89--178] 10.12.2013 10:00PM
4.5 Aerodynamically driven bending
"
#
g Ug V 2 xx 2 H
2 H
2 H
2
V
2V
0
xt
x2
t2
113
4:33
where is the bending perturbation amplitude.

Rendering this equation dimensionless using the scales listed above in Section 4.4, the
following dimensionless equation is obtained:
i 2
2
2 h 2
2
2V
RU

V

E
0
xx
t2
xt
x2
4:34
where L is also used as a scale for . Equations (4.33) and (4.34) are similar to Eq. (4.6)
in Section 4.3 for the bending perturbations of a exible solid threadline.
It should be emphasized that in the linear approximation, perturbations of the
longitudinal ow do not affect small bending perturbations, i.e. the latter are completely uncoupled from the former, since coupling could happen only via nonlinear
terms. Therefore, in Eq. (4.34) the factors multiplying the derivatives in the second
and third terms on the left depend on the unperturbed distributions of V(x) and
xx(x) = xx(x) yy(x), which are found from Eqs. (4.25)(4.27) in Section 4.4.
The general solution of Eq. (4.34) is found using the characteristics
h
i1=2
dx
V Exx RUg V 2
4:35
dt
Equation (4.34) is hyperbolic if Exx R(Ug V)2 > 0 and the characteristics are real,
and elliptic if Exx R(Ug V)2 < 0 and the characteristics are complex. This means that
at x = 0, if the inequality Exx0 > R[Ug(0)1]2 holds, the initial part of the jet is
hyperbolic. Given Eq. (4.32) in Section 4.4, the conditions that the initial part of the
jet is hyperbolic in both unperturbed and perturbed states are
1 > Exx0 > RUg 0 12
4:36
The transition cross-section x = x* is found from the following equation

Kx Exx x RUg x V x2 0
The general solution of Eq. (4.34) is given by
3
2
x
dx
7
6
q t5
Hx; t 4
2
E xx RUg V
0 V
2
3
x
dx
6
q t7
F4
5
E xx RUg V 2
0 V
where and F are arbitrary functions.
4:37
4:38
114
114 [89--178] 10.12.2013 10:00PM
Assume that the inequalities (Eq. 4.36) hold. Then, the initial part of the polymeric
jet 0 x x* is hyperbolic, whereas the following part x* x 1 is elliptic.
Here again, as in the case of a exible solid threadline (see Section 4.3), we apply
the conditions for the perturbation wave at the initial cross-section of the polymer jet in
the following form:
jx0 0 expit; =xjx0 0
4:39
which corresponds to the overall effect of the turbulent pulsations being combined; H0
and are the amplitude and frequency of a perturbation mode, respectively.
Then, one can nd the functions and F and reduce Eq. (4.38) to the following
solution for the hyperbolic part at 0 x x*:
x; t
0
expitf expiI1 x expiI2 xg
1
4:40
where the two real functions I1(x) and I2(x) are given by
x
I1 x
0
x
I2 x
0
dx
q
V x Exx x RUg x V x2
4:41
dx
q
V x Exx x RUg x V x2
4:42
and

dI2 =dx
dI1 =dxx0
4:43
The corresponding solution for the elliptic (with regard to bending perturbations) part
of the polymer jet at x* x 1 is also obtained from the general solution (Eq. 4.38). After
it is matched to the hyperbolic solution at the transition point x = x*, it reads
x; t
0
expfit J1 xg
1
4:44
f expiI1 x expJ2 x expiI2 x expJ2 xg
In Eq. (4.44) the two additional real functions J1(x) and J2(x) are dened as
x
J1 x
V x
V2 x
x
x
J2 x
x
RUg x V x2 Exx x
q
RUg x V x2 Exx x
V2 x RUg x V x2 Exx x
dx
4:45
dx
4:46
115 [89--178] 10.12.2013 10:00PM
4.5 Aerodynamically driven bending
115
0.6
0.5
0.4
0.3
0.2
0.1
0
0.1
0
Figure 4.17
0.25
0.5
x
0.75
Distribution of K(x). Reprinted with permission from Yarin et al. (2010). Copyright 2010, AIP
Publishing LLC.
Note that the dimensional gas stream velocity at the outer boundary of the boundary layer
near the liquid jet surface, Ug (x), can be evaluated using the theory of axisymmetric
turbulent gas jets (Abramovich 1963, Yarin (2007):
Ug x Ug 0
2:4d0
x 2:4d0
4:47
where d0 2a0 is the diameter of the coaxial gas-jet nozzle, and 2.4d0 is the polar distance
of the jet.
The corresponding dimensionless expression becomes
Ug x Ug 0
4:8=
x 4:8=
4:48
where Ug(x) and Ug(0) are rendered dimensionless by V0.

Figure 4.17 depicts the function K(x) of Eq. (4.37), which is obtained from the solution
for the unperturbed straight jet in Section 4.4. It can be seen that K(x) = 0 at x = x* =
0.146. This means that at 0 x 0.146 bending perturbations are hyperbolic, whereas
at 0.146 x 1 they are elliptic. Two predicted snapshots of the bending jet congurations corresponding to two different time moments are shown in Figure 4.18 (with the
parameters listed in Section 4.4, and in addition, H0 = 0.01, and the dimensionless
perturbation frequency = L/V0 = 1500 with = 6 103 Hz). It is seen that in the
hyperbolic part described by Eq. (4.40) that the traveling perturbation wave has an
amplitude of the order of 0. After the transition to the elliptic part, the amplitude of
the traveling perturbation wave described by Eq. (4.44) decreases, but after the
116
116 [89--178] 10.12.2013 10:00PM
0.2
0.15
0.1
0.05
0
0.05
0.1
0.15
0.2
0
0.25
0.5
0.75
X
Figure 4.18
Two snapshots (one shown by a solid line, another by a dashed line) of the predicted traveling
wave of bending perturbations of a polymer jet enhanced by the distributed aerodynamic lift force.
bottleneck seen in Figure 4.18 at about x = 0.35, the perturbation amplitude rapidly
increases, as is expected for an elliptic problem solved as an initial-value problem. This is
the area where the perturbed jet will be drastically stretched and thinned. An accurate
description of such thinning can be achieved only in the framework of a fully nonlinear
description in the Sections 4.64.8.
The pattern of the perturbation waves predicted for meltblowing in Figure 4.18 is
rather peculiar. It shows that the initial part of the jet attached to the die can be almost
straight, since it is stabilized by sufciently large longitudinal stresses generated in the
die that have still not fully vanished. On the other hand, when relaxation of the
longitudinal stresses signicantly reduces their level, bending perturbations grow and
begin to release the stored energy delivered by the initial perturbations (turbulent
pulsations) and are also enhanced by the distributed lift force. A renewed signicant
liquid stretching is expected because of strong bending, which will be also accompanied
by drastic thinning of the polymer jet. This behavior is quite similar to the patterns
characteristic of electrospun polymer jets considered in Chapter 5. In the latter case the
bending force related to Coulombic repulsion is formally similar to the distributed
aerodynamic lift force in the present case (both are proportional to the local curvature
of the jet axis, as can be seen from Eqs. 3.17 and 3.19). The presence of signicant
longitudinal viscoelastic stresses in the initial part of electrospun jets stabilizes them and
they stay almost straight. Later on, the stresses fade due to viscoelastic relaxation, and
strong bending begins (Reneker et al. 2000, 2007, Yarin et al. 2001, Han et al. 2008).
Correspondingly, meltblown jets can stay initially almost straight due to stabilization by
117 [89--178] 10.12.2013 10:00PM
4.6 Meltblowing of a single polymer jet
117
high longitudinal stress. Only after it fades due to the dominant relaxation, does vigorous
bending leading to jet elongation and thinning begin.
4.6
Meltblowing of a single planar polymer jet

In the quasi-one-dimensional theory of liquid jets, two types of approaches to describing
jet bending (with and without accounting for the bending stiffness) are available. In the
momentless approximation, the bending stiffness of very thin liquid jets is neglected
compared to the other internal forces affecting bending, since it depends on the crosssectional jet radius as a4 and the other forces as a2 (which is much larger as a tends to
zero). Correspondingly, the continuity and momentum balance equations (Eqs. 3.1 and
3.2) contain terms of the order of a2 (except the shearing force Q), whereas the terms of
the moment-of-momentum equation (Eq. 3.3), which determines Q, are of the order of a4.
In this present and the following section we adopt the momentless approximation, and
neglect the shearing force in the jet cross-section and the entire moment-of-momentum
equation (Eq. 3.1) determining it. As a result, the quasi-one-dimensional equations of the
jet dynamics (Eqs. 3.1 and 3.2) reduce to the autonomous continuity and momentum
balance equations in the following form:
f f W
0
t
s
4:49
f V f WV 1 P
f g qtotal
t
s
s
4:50
In Eqs. (4.49) and (4.50), t is time, s is an arbitrary parameter (coordinate) reckoned along
the jet axis, f(s,t) = a2 is the cross-sectional area (the cross-section is assumed to stay
circular even in bending jets a valid approximation for sufciently viscous jets,
according to Eq. 3.37 in Section 3.4; a(s,t) denotes its radius), W is the liquid velocity
along the jet relative to a cross-section with a certain value of s, the stretching factor
jR=sj, where R(s,t) is the position vector of the jet axis,V(s,t) is the absolute
velocity in the jet, is the liquid density, P(s,t) is the magnitude of the longitudinal
internal force of viscoelastic origin in the jet cross-section, is the unit tangent vector of
the jet axis, g is gravity acceleration and qtotal is the overall aerodynamic force imposed
on a unit jet length by the surrounding gas. Boldfaced characters denote vectors.
Let s be a Lagrangian parameter of liquid elements in the jet (e.g. their initial Cartesian
coordinate along the blowing direction). Then, W = 0, since the particles keep their
Lagrangian coordinate unchanged, and Eq. (4.49) is integrated to yield
a2 0 a20
4:51
where the subscript 0 denotes the initial values.

Consider, for example, two-dimensional (planar) bending perturbations of a jet
(fully three-dimensional bending will be considered in Section 4.7). Then, accounting
118
118 [89--178] 10.12.2013 10:00PM
for Eq. (4.51), the two projections of Eq. (4.50) on the directions of the local tangent and
unit normal n to the jet axis read:

qtotal;
V
1 Vn
1 P
Vn
kV
4:52
g
t
f
s
f s

qtotal;n
Vn
1 Vn
Pk
V
kV
gn
t
f
s
f
4:53
where k is the local curvature of the jet axis, and subscripts and n denote vector
projections on the local tangent and normal to the jet axis, respectively.
In the case of planar bending, the position vector of the jet axis is described as
R is; t js; t
4:54
with i and j being the unit vectors of the directions of blowing and normal to it,
respectively, while the geometric parameters and k are given by

1=2
2;s 2;s
4:55
;ss ;s ;ss ;s
k
3=2
2;s 2;s
4:56
The total aerodynamic force is comprised of the distributed longitudinal lift force
(a nonlinear analog of the linearized expressions given in Eqs. 3.17 or 4.4), the distributed drag force associated with the gas ow across the jet (see the rst Eq. 3.18) and
the pulling drag force similar to that of Eq. (4.21). Therefore
qtotal nqtotal;n qtotal;
2
3

2

2
=
sign
=
;s ;s
;s ;s 7
;ss ;s
6 ;s ;ss ;s
g U2g n4f
5

2
5=2 a
2
2
1 ;s =;s
;s ;s
0:81

2
2a Ug V
ag Ug V 0:65
g
4:57
where Ug is the magnitude of the absolute local blowing velocity of the gas and
corresponds to its projection on the local direction of the jet axis.
In addition, in Eqs. (4.52) and (4.53), the projections of the gravity acceleration g and
gn are equal to g = g and gn = gn, with g being its magnitude, and n the local
projection of the unit normal to the jet axis onto the direction of blowing.
The longitudinal internal force of viscoelastic origin in the jet cross-section2 P =
f( nn), where and nn are the longitudinal and normal deviatoric stresses in the
jet cross-section (see Sections 2.3 and 2.4). As usual, in the case of strong stretching
(see, for example, Section 4.4), >> nn and the latter can be neglected. Then, P = f,
119 [89--178] 10.12.2013 10:00PM
119
where the constitutive equation for is provided by the viscoelastic upper-convected

Maxwell model (UCM, see Section 2.3) in the form
1
1
2
2

t
t
t
4:58
It should be emphasized that, meanwhile, we consider the isothermal cases with viscosity
and relaxation time to be constant. Below in this section, the non-isothermal cases
will be considered separately.
Equations (4.51)(4.53) and (4.55)(4.58), which describe jet dynamics, are supplemented by the following kinematic equations that describe the axis shape:
V n Vn
t
4:59
Vn V n
t
4:60
The projections of the unit vectors associated with the polymer jet axis and n on the
directions of the unit vectors i and j associated with the blowing direction and normal to it
( and ) are given by the following expressions:
h

2 i1=2
n 1 ;s =;s
4:61

h

2 i1=2
n ;s =;s 1 ;s =;s
4:62
According to the theory of axisymmetric turbulent gas jets (Abramovich 1963, Yarin
2007), the gas ow eld is given by the following expression:
Ug ; Ug0 ;
4:63
where Ug0 is the gas velocity of the nozzle exit and the dimensionless function (,) is
given by
;
4:8=
1

; ;
4:8= 1 2 =8 2
0:05 4:8=
4:64
In Eq. (4.64), and are rendered dimensionless by L, the distance between the initial
cross-section and deposition screen, and is given by the second Eq. (4.29).
Neglecting secondary terms, the governing equations of the problem (4.52), (4.53)
and (4.58)(4.60) in the isothermal case can be reduced to the following dimensionless
form:
qtotal;
2
2 2
1
2 2 J
2
t
Re s
Fr
f
4:65
120
120 [89--178] 10.12.2013 10:00PM

2

i 1 2 n
2 h
2 ; ;s =;s sign ;s =;s
2
J ; 2 2 2 J

2
Re
Fr
t2
a
s
1 ;s =;s
4:66
where 1=De2 satises the following equation:

t
De2
4:67
In Eq. (4.65)

2

0:81
qtotal; a ; V 0:65 Rea a ; V
4:68
The following scales are used: L/Ug0 for t; L for s, and ; L1 for k; Ug0/L for and
; Ug0 for Ug, V and Vn; a0 for a (and a20 for f); g U2g0 a0 for qtotal,; / for the kinematic
turbulent eddy viscosity t (used below; see Section 4.2). As a result, the following
dimensionless groups arise in addition to L=a0 :
!1=2
U2g0
g
LUg0
2a0 Ug0
Ug0
; J ; Fr
4:69
; Rea
; De
Re
gL
g
L
where Re and Rea are the corresponding Reynolds numbers, Fr is the Froude number and
De is the Deborah number.
Based on the results of Sections 4.2, 4.3 and 4.5, we assume that turbulent pulsations in
the gas jet impose lateral perturbations at the origin of the polymer melt jet at s = sorigin.
Therefore, similarly to Eq. (4.39), the boundary conditions for Eqs. (4.65) and (4.66) are
given by the following dimensionless expressions:
jssorigin 0; jssorigin 0 expit
4:70
L
Ug0
4:71
where
0 0:06Re1=2 =
1=2
1=4
=0
with 0 being the dimensionless longitudinal stress in the jet inherited from the nozzle.
On the other hand, at the free end, s = sfree end the jet is assumed to be fully unloaded, i.e.
;s jsf ree end 1; jsf ree end 0
4:72
The initial condition for Eq. (4.67) at the moment when a liquid element leaves the nozzle
and enters the jet tbirth reads
jttbirth 0 1=De=20
4:73
In the non-isothermal case the polymer melt is surrounded by hot gas jet, which blows
into the space lled with cold gas. As a result, the gas jet is cooled and the encased
polymer melt jet is also cooled and solidies. Following Yarin (1993), we assume that the
121 [89--178] 10.12.2013 10:00PM
121
viscosity and relaxation time of the polymer melt vary with melt temperature T according
to the following expressions:

U 1
1
T0
U 1
1
; 0 exp
4:74
0 exp

T
R T T0
R T T0
where T0 is the melt and gas-jet temperature at the origin, 0 and 0 are the corresponding
values of the viscosity and relaxation time, U is the activation energy of viscous ow and
R is the absolute gas constant (see Eqs. 2.55 and 2.56 in Chapter 2).
The quasi-one-dimensional thermal balance equation for a jet element reads

cTf h T Tg 2a
t
4:75
where c is the specic heat, h is the heat transfer coefcient and Tg the local gas temperature.
Using Eq. (4.51) and rendering temperatures T and Tg dimensionless by T0, rearrange
Eq. (4.75) to the following dimensionless form

T
JC
2Nu
T Tg
t
Rea Prg 0
4:76
where Prg is the molecular Prandtl number for gas, C = cpg/c is the ratio of the specic
heat at constant pressure for the gas to the specic heat for the polymer melt, and the
Nusselt number Nu = (h2a/kg), with kg being the molecular thermal conductivity of gas,
is given by the following expression (Yarin et al. 2001):
1=2
Nu 0:495Re1=3
a Pr g
4:77
Substituting Eqs. (4.74) into Eq. (4.58) and using Eq. (4.76), we obtain the following
dimensionless rheological constitutive equation, replacing Eq. (4.67):

T Tg

2NuJC
1
4:78

1
T exp UA
t
Rea Prg De0 0
T
De0 2
with T=De0 =2 and the following two additional dimensionless groups
involved:
De0
0 Ugo
U
; UA
L
RT0
4:79
Using the theory of the axisymmetric turbulent gas jets (Yarin 2007), the temperature
eld in the gas jet is given by the following expression:

Prt 1=2 1 Tg
1
1
p
4:80
Tg ; Tg

2Prt
4:8=
0:05 6
1 2 =8
where Tg is the surrounding cold gas temperature far from the polymer jet, rendered
dimensionless by T0, the turbulent Prandtl number Prt = 0.75 and (,) is given by the
second Eq. (4.64).
122
122 [89--178] 10.12.2013 10:00PM
It should be emphasized that the other governing equations of the problem, e.g. Eqs.
(4.64)(4.66) and (4.68) do not change in the nonisothermal case.
The governing equation (Eq. 4.66) is a hyperbolic wave equation for (s,t), which
corresponds to propagation of the elastic compressionexpansion waves along the jet
(the elastic sound). The second governing equation (Eq. 4.67) responsible for the
propagation of the bending perturbations (s,t) is a hyperbolic wave equation if the jet is
stretched signicantly, i.e. =Re J2 ; > 0. If the longitudinal jet stretching fades
due to the elastic relaxation and the distributed lift force Jf2 ; H becomes dominant,
Eq. (4.67) changes type and becomes elliptic, since the factor =Re J2 ;
becomes negative. This behavior has already been mentioned in the linearized version of
this problem in Section 4.5, which has drastic consequences on perturbation growth.
The following parameter values were used in the numerical simulations: a0 = 0.12 cm,
L = 200 cm, Ug0 = 230 m s1, = 1 g cm3, = 102 g cm1s1, and g = 1.22 103 g cm3.
The corresponding values of Re, J and were about Re = 46000, J = 103 and = 103. The
effect of gravity was excluded, which corresponds to Fr = . The initial longitudinal
elastic stress was taken as 0 = 10 and 0 = 1. The relaxation time was taken as about =
0.1 s, which corresponds to De = 10. In addition, it was taken that C = 0.25 and = 0.3
(the latter corresponds to = 35 Hz; see Section 4.2). Also, Tg = 0.5 and UA = 10. The
value of Rea for air (g = 0.15 cm2 s1) is about 46000. However, at such values of Rea the
pulling aerodynamic drag described by Eq. (4.68) is insufcient to initiate meltblowing.
This might be related to the fact that the empirical Eq. (4.68) was established by Ziabicki
and Kawai (1985) in experiments with threadlines pulled through stagnant air. However,
drag imposed by blowing air may be dramatically increased due to the turbulent eddy
viscosity (which was absent in the experiments described by Ziabicki and Kawai 1985).
To account for that fact, the factor 0.65 in Eq. (4.68) was replaced by 1265. Then, the
aerodynamic drag as per Eq. (4.68) became quite sufcient to initiate meltblowing.
An implicit numerical scheme of a generalized CrankNicolson type, with the central
difference special discretization at three levels of time from Mattheij et al. (2005; p. 444)
was implemented to solve Eqs. (4.65) and (4.66) numerically.
The mean axisymmetric velocity and temperature elds in the central domain close to
the jet origin calculated using Eqs. (4.63), (4.64) and (4.80) are presented in Figure 4.19.
Since the turbulent Prandtl number Prt is less than 1 (Prt = 0.75), the temperature prole in
the gas jet is wider than the velocity prole.
The predicted congurations of the jet axis in the isothermal planar blowing process
are depicted in Figure 4.20. It can be seen that the polymer jet is pulled and strongly
stretched by the co-owing gas jet. The polymer jet also experiences lateral perturbations
due to turbulent eddies. These bending perturbations are signicantly enhanced by the
distributed aerodynamic lift force acting on the curved polymer jet. They also propagate
along the polymer jet as elastic waves, and are additionally swept by the local aerodynamic drag force acting on the jet elements. The congurations of the polymer jet can
become rather complicated and self-intersecting after a while (Figure 4.20a), which is
possible for a phantom jet, but forbidden for a real material one. Still, evolution similar
to that in Figure 4.20b points at possible self-intersection in meltblowing, even in the case
of a single jet considered here.
123 [89--178] 10.12.2013 10:00PM
123
0.8
0.8
U
0.7
0.5
Y
0.4
0.3
0.2
0.6
0.5
0.4
0.3
0.2
0.1
0.1
0.1
Figure 4.19
0.5
0.25
0.75
0.54566
0.542626
0.539592
0.536559
0.533525
0.530491
0.527457
0.524424
0.52139
0.518356
0.515322
0.512288
0.509255
0.509221
0.503187
0.6
TEM
0.7
0.0429403
0.040079
0.0372177
0.0343564
0.0314952
0.0286339
0.0257726
0.0229113
0.0200501
0.0171888
0.0143275
0.0114663
0.00860498
0.00574371
0.00288243
0.1
0.25
0.5
0.75
Axisymmetric velocity (left) and temperature (right) elds in the gas jet. Reprinted with permission
from Yarin et al. (2010). Copyright 2010, AIP Publishing LLC.
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5
1
1
0.8
0.75
0.6
0.5
0.4
H()
0.5
H()
0.2
0.25
0.4
0.5
at t = 50
at t = 35
at t = 15
0.6
0.8
1
0.5
0.25
0.2
(a)
Figure 4.20
0.5
X
Isothermal conditions
0.5
0.5
0.75
1.5
1
0
(b)
X
Isothermal planar blowing. (a) Three snapshots of the axis conguration of a polymer jet at the
dimensionless time t = 15, 35 and 50 (the corresponding dimensional times are 0.13 s, 0.3 s and
0.43 s, respectively). (b) A later snapshot of the polymer jet axis corresponding to the
dimensionless time t = 75 (the corresponding dimensional time is 0.65 s). Reprinted with
If one follows the individual material Lagrangian elements of the polymer jet, as in the
present simulations, their distribution along the jet allows visualization of nonuniform jet
stretching. Indeed, these material elements are visualized by symbols on the jet axis in
Figure 4.21. The larger the distance along the jet between two neighboring symbols, the
larger the local stretching of viscoelastic polymer melt in the corresponding jet section.
Figure 4.21 shows that, initially, jet stretching is quite signicant and grows with time
124
124 [89--178] 10.12.2013 10:00PM
0.8
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1 1.2 1.3
0.4
0.5
0.3
0.3
0.4
Y
0
0.2
0.4
at t = 50
at t = 35
at t = 15
0.6
0.8
0
(a)
Figure 4.21
0.5
1
X
1.5
H()
0.25
0.2
0.2
0.2
0.1
0.1
0.25
0.1
0.5
0.2
0.75
0.3
0
at t = 50
at t = 35
at t = 15
0.1
0.2
0.3
0.4
0.4
0
(b)
H()
0.6
0.4
0.75
0 0.10.20.30.40.50.60.70.80.9 1 1.11.21.31.41.5
1
X
Nonisothermal conditions
0.5
1.5
(a) Three snapshots of a polymer jet axis with isothermal planar blowing at dimensionless
times t = 15, 35 and 50 (the corresponding dimensional times are 0.13 s, 0.3 s and 0.43 s,
respectively) with the symbols denoting the material elements of the jet. (b) Same as (a), but with
nonisothermal planar blowing. Reprinted with permission from Yarin et al. (2010). Copyright
2010, AIP Publishing LLC.
(from snapshot to snapshot in Figure 4.21), however, it can deteriorate further on due to
the elastic recoil characteristic of viscoelastic polymer melts, as well as decreasing
stretching by the gas jet, which weakens down the ow.
Comparison of Figures 4.21a and b allows one to visualize the effect of cooling and
solidication of the polymer jet. In particular, it can be seen that the growth of the
bending perturbations of the jet is arrested due to cooling. However, the jet does not
become straight, but continues to sustain traveling bending perturbations in a similar way
to the exible solid threadlines described in Sections 4.2 and 4.3.
The linear and nonlinear theory of meltblowing developed in Sections 4.5 and 4.6 explains
the physical mechanisms responsible for jet congurations, and in particular, the role of
turbulent pulsations in the gas jet, the aerodynamic lift and drag forces and the longitudinal
viscoelastic stress in the polymer jet. The theory produces a plausible pattern of the evolution
of the jet over time. Self-intersection of polymer jets visualized by the two-dimensional
simulations can lead to jet rupture in the case of violent blowing, which results in so-called y
(small jet segments contaminating the surrounding atmosphere). The last, but not least, goal
of numerical modeling is prediction of the internal structure of the bers comprising nonwovens resulting from meltblowing. This information will be elucidated using the predicted
longitudinal stresses frozen in the jet at the deposition screen (see Section 8.3).
4.7
Fully three-dimensional blowing of single and multiple polymer jets

In this section a comprehensive model of three-dimensional congurations of polymer
jets in meltblowing from die exit to deposition screen is developed, following
125 [89--178] 10.12.2013 10:00PM
4.7 Fully three-dimensional blowing
125
Yarin et al. (2011), as a generalization of the model described in Section 4.6. A single
nonisothermal polymer jet with three-dimensional perturbations, or multiple jets
modeled simultaneously are considered, as well as deposition on a screen moving
normally to the blowing direction. All important properties of polymer melts are used
in the simulations, namely density, zero-shear viscosity and the viscoelastic relaxation
time. In addition, the material parameters responsible for temperature-dependent
variation of the material properties are used. All the other parameters are the operational parameters characterizing gas ow. The results include prediction of the ber
deposition patterns and ber size distributions in the resulting nonwovens. The
angular distributions in lay-down nonwovens are also predicted.
In meltblowing, the polymer melt is issued through a die nosepiece into a high-speed
subsonic coaxial air jet. The air jet pulls the polymer jet, dramatically accelerating and
stretching it, vigorous bending and apping take place, which additionally stretch the
polymer jet (see Sections 4.14.6). The gas jet is submerged in the surrounding atmosphere
at room temperature. As a result, it cools down, cooling down the polymer jet inside. The
latter solidies at some distance from the die exit and is deposited as a nonwoven onto a
grid-like screen moving normally to the blowing direction. Each individual jet forms a ber
in the nonwoven matrix, with ber diameters typically being in the range 210 m, albeit
submicron or several hundred micron bers have also been reported. The screen is typically
made of metal and moves with a speed of the order of 120 m s1. It moves over a chamber
connected to a vacuum pump to prevent air motion along the screen.
This section rst modies the model results of Section 4.4 to make them appropriate
for the description of the initial part of real jets in meltblowing, which are to be used as a
subroutine in the general numerical code for simulations of multiple three-dimensional
nonisothermal polymer jets depositing onto a moving screen. In real meltblowing
processes, polymer jets are issued from dies in a nosepiece at a relatively low speed of
about 10 cm s1. They are issued into approximately concentric gas jets with velocities of
the order of 100250 m s1. As a result, polymer jets experience a tremendous pulling
force at their initial part (of the order of 1 mm from the die) and rapidly accelerate. The
concentric gas jets are formed by oblique impingement of several gas jets surrounding a
polymer jet, which produces a signicant force directed toward the collecting screen.
This force sustains straight conguration of the initial part of the polymer jets. In
addition, polymer jets at the initial part are still sufciently thick and possess a signicant
bending stiffness, which precludes any bending, similarly to the situation encountered in
electrospinning (Reneker et al. 2000; see Chapter 5). It should be emphasized that over
such a short distance the velocity of the concentric submerged gas jet practically does not
fade due to viscous interaction with the surrounding air and will be assumed to be
constant in the present section.
The stretching aerodynamic drag force per unit jet length q is described similarly to
Eq. (4.21) as
0:81

2 Ug V a
q x Kag Ug V 2
4:81
g
126
126 [89--178] 10.12.2013 10:00PM
In Eq. (4.81), as before, x is the axial coordinate reckoned from the die exit, f (x) = a2
and a(x) are the cross-sectional area and radius, respectively, V(x) is the absolute axial
velocity of the polymer jet, Ug(x) is the absolute velocity of the concentric gas jet, and g
and g are the gas density and kinematic viscosity, respectively.
Equation (4.81) is, in fact, empirically based and stems from experiments in which a
wire or lament was pulled parallel to itself through stagnant air, as described in the book
edited by Ziabicki and Kawai (1985 and references therein). It was used in Sections 4.4
with the pre-factor established in the above-mentioned work, K = 0.65. While the general
structure of Eq. (4.81) is physically sound, the modeling in Section 4.6 demonstrated that
this value is probably underestimated and should be increased, since otherwise it is
impossible to achieve a proper pulling force to stretch viscoelastic polymer jets. The
underestimation of the value of K in the experiments described in the book edited by
Ziabicki and Kawai (1985 and references therein), most probably has deep physical
roots. Indeed, the situation where a wire is pulled through stagnant air with a certain
speed is not equivalent to the situation where a nonmoving wire (or a jet, as in
meltblowing) is subjected to parallel gas ow with the same speed, since in the latter
case turbulent eddy viscosity is expected to be higher, and thus the drag force, and
therefore K, should be higher. An additional circumstance that points at an additional
increase in the value of K is related to the fact that in the experiments where it was
established, velocities never approached levels characteristic of meltblowing and sufciently high turbulence levels were not reached. Due to all these circumstances, while
keeping the general structure of Eq. (4.81), the value of K was increased to K = 100,
which was validated a posteriori by comparison with the experimental data for meltblowing, as described below.
The quasi-one-dimensional continuity and momentum balance equations (Eqs. 4.19
and 4.20) for the straight part of meltblown viscoelastic polymer jets were solved
numerically using Eq. (4.81) for the pulling force. Similarly to Section 4.4, the upperconvected Maxwell model (UCM) is used as an appropriate rheological constitutive
model for viscoelastic polymeric liquids in such strong elongational ows as meltblown jets. As a result, distributions of the cross-sectional radius a(x), longitudinal
velocity V(x), the longitudinal and radial deviatoric stresses xx and yy, and the stress
xx = xx yy in the jet cross-section are found. As in Section 4.4, the equations were
rendered dimensionless by the following scales: V0 for V and Ug, the distance of the
initial straight part Lstraight for x, a0 for a, 0/0 (with 0 and 0 being the viscosity and
relaxation time of polymeric liquid at the die temperature T0, respectively) for deviatoric stresses, where the subscript 0 corresponds to the die exit where the radius (a0) and
velocity (V0) values are assumed to be given. The dimensionless groups involved
in the simulations are given similarly to Eqs. (4.29) by
R
g
Lstraight
2V0 a0
0 V0
;
; Re
; De
a0
g
Lstraight
4:82
with Re and De being the Reynolds and Deborah numbers, respectively; the secondary
dimensionless groups E and M are given following Eqs. (4.30) by
127 [89--178] 10.12.2013 10:00PM
g
2R
; M
0
De ReM
127
4:83
In Eq. (4.83) g denotes gas viscosity.

In the simulations of the initial straight part of the jet the following parameter values
corresponding to real practical meltblowing conditions were used. In a typical meltblowing process, polymer throughput through every die in a nosepiece of radius of the order of
100 m, is of the order of 0.31.0 g min1, which means that V0 0.17 0.5 m s1. The
typical value of gas velocity is of the order of 150 m s1. Using this data, the value of
the dimensionless gas velocity Ug/V0 is taken as 390, whereas the value of Re used is
Re = 7.5. The initial length scale of the straight part Lstraight is of the order of 1 mm, which
makes = 10. The relaxation time 0 is of the order of 102 s, which might be expected
for strong elongational processes, where the leading physical mechanism of relaxation
is associated with the recoil of stretched macromolecular coils (Doi 1980, Yarin 1993;
see Section 2.3). Then, a reasonable value of the Deborah number De is taken as
De = 1.59. Following Section 4.4, the values of M and R are both taken to be 0.001.
A sufciently long die channel allows longitudinal elastic stresses acquired in the ow
inside to relax. Therefore, the value of the longitudinal deviatoric stress xx0 is taken to be
as low as xx0 = 0.0001, which corresponds to a dimensional value of 0.00025 N m2.
It was found that any further decrease in the value of xx0 does not affect the results.
The predicted distributions of the jet radius and velocity, as well as of the longitudinal
deviatoric stress are plotted in Figures 4.22ac. The values of the radial deviatoric stress
are practically immaterial compared to xx, the result being similar to that in Figure 4.16
in Section 4.4, and thus not shown here.
Figure 4.22a shows that the value of the dimensionless radius at the end becomes 8.75
m. The gure demonstrates that an abrupt decrease in the cross-sectional radius happens
close to the exit from a die. Figure 4.22b demonstrates that at the end of the straight part
of the polymer jet its velocity is already about 30% of the velocity of the surrounding gas
jet. In comparison, the experimental data of Breese and Ko (2003) show that at a distance
of about 5 mm from the die, the velocity of the meltblown polymer jet had already been
increased up to 43% of the velocity of the surrounding gas jet, which is comparable with
our numerical prediction. A more detailed comparison with the experimental data of
Breese and Ko (2003) is, unfortunately, impossible due to the lack of a number of crucial
input parameters characterizing the experimental situation, as well as an insufcient
number of experimental points available. In addition, it should be mentioned that the
experimental data of Shambaugh (1988) show that at a die-to-collector distance (DCD)
of 6 mm the ber diameter was 49 m at a blowing speed of 123.2 m s1, whereas at a
DCD of 7 mm the ber diameter was 117 m. These values are about 510 times higher
than our prediction of 8.75 m. However, the difference can be attributed to the following
factors: (i) the lack of detailed rheological data for the polymer melt blown by
Shambaugh (1988); (ii) unspecied differences in the operating conditions; (iii) the asformed bers were deposited on a screen by Shambaugh (1988), which allows the bers
to relax and increase their cross-sectional size, which was not considered by the theory in
128
128 [89--178] 10.12.2013 10:00PM
100
50
80
40
v (m/s)
60
a (m)
120
30
60
40
20
20
10
0
0
0
(a)
0.2
0.4
0.6
Distance (mm)
0.8
0.2
0.4
(b)
0.6
0.8
Distance (mm)
45000
40000
xx(N/m2)
35000
30000
25000
20000
15000
10000
5000
0
0
(c)
Figure 4.22
0.2
0.4
0.6
0.8
Distance (mm)
1.2
Distributions of (a) dimensionless cross-sectional radius, (b) axial velocity and (c) longitudinal
deviatoric stress along the polymer jet. Reprinted from Yarin et al. (2011), with permission from
Elsevier.
this section. Nevertheless, the trend observed in the experiments of Shambaugh (1988)
and the numerical predictions here show that there was already a signicant decrease in
ber diameter at a distance of about 1 mm from the die.
Note also that Figure 4.22c shows that at the end of the straight part of the polymer jet,
quite signicant longitudinal stress has already been reached, which can be attributed to
strong stretching in the initial part of the jet.
The conditions at the end of the initial straight part of the jet should be taken as the
initial conditions for the bending part, which begins from there on and is described below.
After the straight part of the jet becomes sufciently thin, its bending stiffness becomes
very small (since it is proportional to a4) and small bending perturbations that are
imposed on the jet begin to grow. The description of the bending part of the jet is
based on the general momentless quasi-one-dimensional equations of the dynamics of
free liquid jets (Eqs. 4.49 and 4.50 in Section 4.6). However, unlike Section 4.6, in the
present section we deal not with planar bending of the jet, but rather with fully threedimensional bending.
Let s be a Lagrangian parameter of liquid elements in the jet (e.g. their initial Cartesian
coordinate along the blowing direction). Then, W = 0 in Eq. (4.49), since the particles keep
their Lagrangian coordinates unchanged, and after integration that equation reduces to
129 [89--178] 10.12.2013 10:00PM
a2 0;straight a20;straight
129
4:84
where subscript 0, straight denotes the values corresponding to the end of the straight part
of the jet predicted above in this section. Note that Eq. (4.84) is similar to Eq. (4.51) in
Section 4.6.
The local projections of the momentum balance equation (Eq. 4.50) on the tangent,
normal and binormal to the jet axis, after several minor terms are omitted for simplicity, read
qtotal;
V
1 f
g
t
f
f s
4:85
qtotal;n
Vn k
gn
t
4:86
qtotal;b
Vb
gb
t
f
4:87
where k is the local curvature of the jet axis and subscripts , n and b denote the
projections on the directions of the local unit tangent, normal and binormal, , n and b,
respectively.
The position vector R(s,t) and the corresponding stretching ratio (s,t) are expressed as
Rs; t is; t jHs; t kZs; t
4:88

1=2
R
2;s H2;s Z2;s
s
4:89
where , and denote projections of the position vector on the laboratory Cartesian
frame with the unit vector i corresponding to the direction of blowing and j and k the
other two unit vectors; see Eqs. (3.16).
The local curvature of the jet axis corresponding to the position vector of Eq. (4.88) is
q

2
2
2
Z;ss H;s H;ss Z;s ;ss Z;s Z;ss ;s H;ss ;s ;ss H;s
4:90
k
3
Each s in the subscript corresponds to differentiation in the s-direction.
The velocity components are related to the rate of the jet axis evolution according to
the following kinematic relations:
V n n
H
Z
n
n
t
t
t
4:91
V b b
H
Z
b
b
t
t
t
4:92
H
Z

t
t
t
4:93
130
130 [89--178] 10.12.2013 10:00PM
The aerodynamic force qtotal in the momentum balance equation (Eq. 4.50), which consists
of the normal (bending) component and the drag force, as well as the longitudinal pulling
component, is given by the following expression equivalent to Eq. (4.57):
qtotal g U2g f kn g U2g an2 signn n

2 2a Ug V 0:81
Cag Ug V
g
4:94
where Ug ; is the absolute local velocity of the surrounding gas jet given by Eqs.
(4.63) and (4.64). In the latter equation in the present case, is given by L/a0,straight .
For the constitutive equation for the longitudinal deviatoric stress the upperconvected Maxwell model (UCM) of Eq. (4.58) is used. In the long bending part of
the jet, variation of the rheological constitutive parameters with decrease in temperature T is felt and described using the expressions for the viscosity and relaxation
time given by Eqs. (4.74). The thermal balance equation for a jet element used to
predict T is given by Eq. (4.75).
The governing equations are rendered dimensionless, as in Section 4.6, using the
following scales: Ug0 for all the velocities, a0,straight for a, a20;straight for f, L/Ug0 for t,
g U2g0 a0;straight for qtotal, 0Ug0/L for all the stresses and L for all other lengths except
radius. As a result, Eq. (4.75) takes a form identical to Eq. (4.76), and the rheological
constitutive equation (Eq. 4.58) becomes Eq. (4.78). The initial condition for Eq. (4.78)
at the moment when a liquid element enters the bending part of the jet tbirth reads
jttbirth 0 T0 =De0 =20
4:95
As in Section 4.6, the temperature eld in the gas jet is given by Eq. (4.80).
The projections of the momentum balance equation (Eqs. 4.854.87) can be transformed
to the following dimensionless form (where some of the minor terms are omitted):
qtotal;
2
2 2
J
2
f
t
Re s2 Fr2
4:96
i 1 2 H
2 H h
2 ; H; Z 2
J2 ; H; Z b 2 2 2 b J
n signn 4:97
2
Re
Fr
t
a
s
i 1 2 Z b b
2 Z h
2 ; H; Z 2

2
b

J
;
H;
Z

b
J
n signn
Re
b Fr2
t2
a
2 s2
4:98
where

2

0:81
qtotal; Ca ; H; Z V Rea a ; H; Z V

;s
H;s
Z;s
;
;
4:99
4:100
131 [89--178] 10.12.2013 10:00PM
131

;ss H2;s Z2;s ;s H;ss H;s Z;ss Z;s
n q

2
2
2
4:101

H;ss 2;s Z2;s H;s ;ss ;s Z;ss Z;s
n q

2
2
2
4:102

Z;ss 2;s H2;s Z;s ;ss ;s H;ss H;s
n q

2
2
2
4:103
b n n ; b n n ; b n n
g
LUg0
; J ; Fr
Re
0
U2g0
!1=2
; Rea
gL
2a0;straight Ug0
g
4:104
4:105
Here Re and Rea denote the corresponding Reynolds numbers and Fr is the Froude
number.
Similarly to Eqs. (4.70) in Section 4.6, the boundary conditions for Eqs. (4.96)(4.98)
at the end of the straight part correspond to the overall effect of turbulent eddies and are
expressed as
origin 0; Horigin j H0 sinj t Zorigin H0 cosj t
4:106
in the cases of a single (j = 1) or multiple (j > 1) jets. In Eqs. (4.106) j are constants
monotonously increasing from jet to jet (if j > 1), which corresponds to a nosepiece with
the die exits located in the -direction.
In Eqs. (4.106), similarly to Eqs. (4.71)
0 0:06Re1=2 =
1=2
1=4
=0 j
j L
Ug0
4:107
where the dimensionless (and the corresponding dimensional) frequencies j (and j)

were chosen randomly from jet to jet, as described below. In Eq. (4.107), 0 is related to
the predicted dimensionless longitudinal stress at the end of the straight part of the jet, as
detailed below.
The free end of the jet is practically unloaded, which similarly to Eq. (4.72) corresponds to
;s jsend 1 H;s jsend 0 and Z;s jsend 0
4:108
On the other hand, when a material point of a jet touches a screen moving in the
Z-direction, its coordinate stays frozen at = 1, its H coordinate stays frozen at
the value it had at the moment of touching the screen, while its Z coordinate increases as
132
132 [89--178] 10.12.2013 10:00PM
0
0.2
Zeta
0.4
0.8
1
0.2
Et
Figure 4.23
0.6
i
Ks
1.2
0
0.2
1.4
Three snapshots of the polymer jet axis in isothermal three-dimensional blowing at the dimensional
time moments t = 15, 30 and 45 (the corresponding dimensional times are 0.13 s, 0.26 and 0.39 s,
respectively; all the parameters are the same as in the isothermal two-dimensional case in Figure 4.20).
touch Vscreen t
4:109
The predictions for isothermal three-dimensional blowing of a single jet are shown in
Figure 4.23. In this case the conguration of the jet axis is described using three
projections, = (s), H = H(s) and Z = Z(s), instead of the two, = (s) and H = H(s),
used in the planar case in Section 4.6. The tendency for self-intersection of the polymer
jet is clearly seen in the three-dimensional case similar to the two-dimensional case of
Section 4.6.
In a typical meltblowing process considered below in this section, the die-to-collector
distance (DCD) L is of the order of 1015 cm, Ug0 is of the order of 150 cm s1, and, at the
operating temperature T0 = 250350 C, the shear viscosity of the polymer melt is within
the range 520 g cm1 s1. Using these values, the Reynolds number Re is of the order of
40 000. In the simulations described below, the value of Re was taken as 46 000.
At the end of the straight part of the jet, the cross-sectional radius was found to be 8.75
m. This makes the value of Rea = 174.
The value of the factor K in the expression for the pulling force in Eq. (4.81) was taken
as K = 14 instead of K = 100 used in the calculations for the straight part of the jet above.
Such reduction in the value of K can be interpreted as the accounting for the fact that
pulling of the jet by air becomes additionally less effective when the jet is inclined
relative to the gas ow due to bending.
The relaxation time 0 = 0.1 s was used in the calculations for the bending part instead
of 0 = 0.01 s, used in the calculations for the straight part of the jet above. This
133 [89--178] 10.12.2013 10:00PM
133
corresponds to the fact that the strongest stretching happens in the straight part of the jet,
and thus the fastest relaxation processes are expected (Doi 1980, Yarin 1993). In the case
of weaker stretching, corresponding to the bending part, even the reptational relaxation
mode (Doi 1980, Yarin 1993) can play a role. The corresponding value of De0 was taken
as De0 = 400. The effect of gravity was neglected, which corresponds to the Froude
number Fr = . In addition, it was taken that C = 0.25, where C = cpg/c is the ratio of the
specic heat at constant pressure for the gas to the specic heat for the polymer melt.
The simulations described below in this section were conducted with 62 jets. Each jet
was excited by the cumulative effect of turbulent eddies described using Eqs. (4.106) and
(4.107). In these equations, the frequencies j were chosen according to the following
formula: j = 0pj, where pj are random numbers from the interval [0,1] generated anew
at each time step and initiated by the current time of the computer clock, i.e. nonrepeatable. The basic variants were simulated at a value of 0 = 0.3 (corresponding to
the cases considered in Section 4.6 and in Figure 4.23). For comparison, simulations
were also done with 0 = 0.1 and 0.5.
The value of the stress 0 for Eq. (4.107) is found from the value of xx,end,straight
calculated at the end of the straight part of the jet. Due to the different scales
used for nondimensionalization for the bending and straight parts, they are related
as 0 = xx,end straight/De0, which makes 0 = 41.84. In addition, as in Section 4.6,
Tg = 0.5, J = 103, = 103 and UA = 10.
The deposition screen was located at = 1, normal to the direction of blowing, and
could be moved in the direction with a given velocity Vscreen from 0.0150.15
(rendered dimensionless by Ug0), which corresponds to the dimensional range 2.25 to
22.5 m s1, which is of practical interest. The choice of the collector screen velocity range
also aims at elucidating its effect on the radius distribution, stretching of the nonwoven
lay-down and pattern formation. As described above, the jet deposition on the screen was
implemented in the following way: at the moment the free end of the jet had reached the
screen, its and coordinates were frozen, whereas its coordinate could vary in
accordance with the translational motion of the screen in the -direction (see Eq. (4.109).
The cross-sectional radius of each jet at the moment it touched the screen was also
frozen and used to produce jet size distributions, as discussed below. The jet size
distributions (see Figures 4.25 and 4.27) were recorded at ve different dimensionless
time moments t = 45, 60, 65, 70 and 75, which correspond to 0.03 s, 0.04 s, 0.043 s,
0.047 s and 0.05 s. The jet congurations at t = 75 corresponding to 0.05 s are shown in
Figures 4.24 and 4.26 for different values of screen velocity. For the sake of brevity, the
radius distribution and jet evolution are illustrated only for 0 = 0.3. In addition, a
detailed comparison of the maximum ber diameter, mean ber diameter, standard
deviation and number of jets reaching the collecting screen for different values of its
velocity and 0 is given in Table 4.1 for two dimensionless time moments t = 45 and 75.
The experimental data on meltblowing from Breese and Qureshi (2006) reveal that for
a DCD of 22 cm the maximum and mean ber diameters are 35 m and 18.3 m,
respectively, for a research line. For a commercial line, Breese and Qureshi (2006) report
maximum and mean ber diameters of 19 m and 7.5 m, respectively. In comparison,
Table 4.1 shows that our numerically predicted values of the maximum ber diameter are
134
134 [89--178] 10.12.2013 10:00PM
H
0.9
0.6
0.3
0
1.6
1.5
1.2
0.8
0.4
0.4
H
Z
1
2
0.5
0
0.2
0
H
0.5
0.8
11
0.6
1.6
0.4
1.2
Z
0.8
0.4
1.6
Figure 4.24
1.2
0.8
0.4
0.4
Jet congurations in the case of the dimensionless screen velocity Vscreen = 0.015 (which
corresponds to 2.25 m s1) at t = 75 (which corresponds to 0.05 s) for = 0.3. Reprinted from Yarin
et al. (2011), with permission from Elsevier.
within the range 17.9440 m, in reasonable agreement with the experimental data. The
experimental data in Shambaugh (1988) for polypropylene (Mw = 175 kDa and Mn = 32
kDa) show that the mean value of the ber diameter was 6 m with a standard deviation
(SD) of 3 m, and for polypropylene with a different molecular weight (Mw = 120 kDa
and Mn = 30 kDa) it was 5 2 m. In addition, the numerical results listed in Table 4.1
show that the predicted mean ber diameter is in the range 8.9814.52 m with a range of
135 [89--178] 10.12.2013 10:00PM
135
Vscreen = 0.015
Vscreen = 0.015
Mean value 6.53 micron
SD 1.73 micron
Maximum 11.99 micron
Minimum 3.68 micron
No. of Jets 60
Frequency (%)
20.0
16.7
13.3
10.0
17.2

SD 3.07 micron
Minimum 2.59 micron
No. of Jets 58
13.8
Frequency (%)
23.3
10.3
6.9
6.7
3.4
3.3
0.0
0.0
0
(a)
6
8
10
Radius (micron)
12
14
6
8 10 12 14
Radius (micron)
(b)
Vscreen = 0.015
8.8
Frequency (%)
Frequency (%)
17.5
(c)
20

SD 2.47 micron
Minimum 3.51 micron
No. of Jets 56
17.9
8.9
0.0
0
18
Vscreen = 0.015

SD 2.96 micron
Minimum 4.01 micron
No. of Jets 57
26.3
16
0.0
10 12 14 16 18 20 22
(d)
Radius (micron)
6
8
10 12
Radius (micron)
14
16
18
Vscreen = 0.015
Frequency (%)
26.8

SD 4.14 micron
Minimum 3.78 micron
No. of Jets 56
17.9
8.9
0.0
0
(e)
Figure 4.25
10
15
20
25
30
35
Radius (micron)
Predicted radius distributions for the screen velocity Vscreen = 0.015 (2.25 m s1) at the time
moments (a) t = 45 (0.03 s), (b) 60 (0.04 s), (c) 65 (0.043 s), (d) 70 (0.047 s) and (e) 75 (0.05 s) for
= 0.3. Reprinted from Yarin et al. (2011), with permission from Elsevier.
SD from 1.98 to 10.86 m. These predictions are in a rather good agreement with the
experimental data. It should be emphasized that Shambaugh (1988) and Breese and
Qureshi (2006) lack many details needed for detailed characterization of the polymer
melts and the governing parameters used (e.g. the rheological behavior, the initial
temperature, polymer ow rate, collector speed, collecting technique, etc.), therefore
the comparison with these data is inevitably conditional.
136
136 [89--178] 10.12.2013 10:00PM
H
1
0.75
0.5
0.25
10
0
9
1.2
0.6
0.3
0.3
0.6
1.25 1 0.75 0.5 0.25 0 0.25 0.5
10.5
9.75
0.9
9
8.25
4
3
7.5
6.75
0.5
0.5
5.25
4.5
0
0.5
3.75
3
2.25
1.5
0.75
0
H
Figure 4.26
Jet congurations in the case of the dimensionless screen velocity Vscreen = 0.15 (which
corresponds to 22.5 m s1) at t = 75 (which corresponds to 0.05 s) for = 0.3. Reprinted from Yarin
137 [89--178] 10.12.2013 10:00PM
Vscreen = 0.15
Vscreen = 0.15
SD 0.99 micron
Maximum 8.15 micron
Minimum 2.65 micron
No. of Jets 62
Frequency (%)
40.3
32.3
24.2
16.1

SD 1.34 micron
Maximum 9.29 micron
Minimum 2.63 micron
No. of Jets 62
40.3
32.3
Frequency (%)
48.4
137
24.2
16.1
8.1
8.1
0.0
0.0
0
(a)
10 11 12
(b)

SD 2.03 micron
Minimum 2.42 micron
No. of Jets 62
10 11 12

SD 2.85 micron
Minimum 2.36 micron
No. of Jets 62
32.3
Frequency (%)
Frequency (%)
24.2
Vscreen = 0.15
Vscreen = 0.15
32.3
Radius (micron)
Radius (micron)
16.1
24.2
16.1
8.1
8.1
0.0
0.0
0
(c)
10
12
14
16
18
(d)
Radius (micron)
10 12 14 16 18 20 22 24
Radius (micron)
Vscreen = 0.15
SD 1.8 micron
Maximum 9.92 micron
Minimum 2.55 micron
No. of Jets 62
Frequency (%)
32.3
24.2
16.1
8.1
0.0
(e)
10
12
Radius (micron)
Figure 4.27
Predicted radius distributions for the screen velocity Vscreen = 0.15 (22.5 m s1) at the time
moments (a) t = 45 (0.03 s), (b) 60 (0.04 s), (c) 65 (0.043 s), (d) 70 (0.047 s) and (e) 75 (0.05 s)
for = 0.3. Reprinted from Yarin et al. (2011), with permission from Elsevier.
138
138 [89--178] 10.12.2013 10:00PM
Table 4.1 The numerically predicted maximum ber diameter, mean ber diameter, standard deviation (SD)
and the number of jets reaching the collecting screen for different values of 0. The results for several values of
the screen velocity are shown at two dimensionless time moments t = 45 and 75.
Vscreen
Time moment
Max ber
diameter in m
Mean ber
diameter in m
SD in m
Number of jets
reaching the collecting screen
0.1
0.015
45
75
45
75
45
75
45
75
45
75
45
75
45
75
45
75
45
75
45
75
45
75
45
75
45
75
45
75
45
75
27.8
26.9
85.44
29.04
37.8
24.8
33.10
27.50
53.14
27.56
23.98
61.08
21.80
24.68
22.14
57.18
25.06
24.24
16.30
19.84
29.84
35.2
53.14
28.50
27.16
25.14
31.04
34.56
17.94
25
14.14
13.76
15.02
12.18
13.04
10.98
12.34
10.50
11.34
9.2
13.06
14.52
11.58
12.66
10.42
11.46
10.28
10.8
8.98
10.16
12.34
13.8
11.34
13.14
10.50
11.96
10.24
10.92
9.48
11.02
4.96
4.9
10.86
4.54
6.84
3.74
5.86
3.30
6.94
3.00
3.46
8.28
3.06
3.38
3.12
6.64
3.78
4.08
1.98
3.6
4.26
5.38
6.94
6.64
3.92
3.82
3.88
4.70
2.68
4.36
58
55
60
56
62
61
62
62
62
62
60
56
60
58
62
62
62
62
62
62
62
62
62
62
62
62
62
62
62
62
0.03
0.06
0.10
0.15
0.3
0.015
0.03
0.06
0.10
0.15
0.5
0.015
0.03
0.06
0.10
0.15
In spite of the above-mentioned uncertainty, the experimental data of Shambaugh

(1988) and Breese and Qureshi (2006) were used for comparison with our numerical
predictions in Figure 4.28. It can be seen that irrespective of the above-mentioned
uncertainties in the exact experimental conditions, the numerical results are in qualitative
agreement with the experimental data.
Figures 4.25 and 4.26 and Table 4.1 show that the values of the basic frequency 0 and
screen velocity Vscreen do not affect signicantly the values of the mean ber diameter.
With regard to the effects of these two parameters, no visible trends were found in the
numerical results. As is evident from Figures 4.25 and 4.27, at 0 = 0.3 the radius
distributions appear to be approaching the normal distribution as the screen velocity
139 [89--178] 10.12.2013 10:00PM
139
30
45
40
Ref 11
20
35
Frequency (%)
Frequency (%)
25
Numerical
present work
15
10
30
Ref 11
25
Numerical present
work
Ref 13
20
15
10
5
5
0
0
0
(a)
Figure 4.28
20
40
60
Diameter (micron)
80
(b)
10
15
20
Diameter (micron)
25
Comparison of experimental and numerically predicted radius distributions. The numerical

results correspond to = 0.3. In panel (a) Vscreen = 0.015, in (b) Vscreen = 0.06. Reprinted from
Yarin et al. (2011), with permission from Elsevier.
Vscreen increases. A similar approach to the normal distribution at higher values of Vscreen
was observed at the other values of 0 (not shown here).
It should be emphasized that the numerical results listed in Table 4.1 show that a few
jets from the initial 62 did not arrive at the collection screen in some cases. At a certain
moment in time the cross-sectional radius of those few jets became zero at a particular
cross-section. This manifests as cohesive jet breakup and probably corresponds to
so-called y formation. The observed cohesive jet breakup is more pronounced at
0 = 0.1 than at higher values of 0. Still, the maximum number of broken jets that
did not reach the screen was not more than 7 at 0 = 0.1.
Using the same approach, Sinha-Ray et al. (2013a) investigated the dependence of the
mass distribution pattern in nonwovens deposited on a moving screen, as well as the
angular orientation of the bers in them. The simulations were done at different speeds of
screen motion (normal to the primary blowing direction). Four different screen speeds
were chosen to cover a wide range of possible operational conditions: 2.25, 4.5, 9 and
22.5 m s1. Being rendered dimensionless with the velocity of gas blowing at the nozzle
exit Ug0, these values correspond to the dimensionless screen velocities 0.015, 0.03, 0.06
and 0.15. The deposits were probed at different cross-sections of deposited nonwovens at
the following four time moments from the beginning of the process: 0.03, 0.04, 047 and
0.05 s, which correspond to the following dimensionless times, respectively, 45, 60, 70
and 75 (rendered dimensionless by L/ Ug0, where L is of the order of the liquid ber
length of about 10 cm).
Note that when a jet segment lands on a moving screen, the following four (ad)
angular orientations shown in Figure 4.29 are possible. In this gure it is assumed that
node i precedes node i+1 on that jet, which means that node numbering is chosen such
that the leading node of the jet is labeled as node 1. For the four possible angular
orientations of the jet segment shown in Figure 4.29 the angle (not to be confused
with the relaxation time in the governing equations!) is dened as
140
140 [89--178] 10.12.2013 10:00PM
(Direction of the screen motion)
Location of the cross-section

i
i+1
180
180 +
i+1
i+1
(a)
(b)
i+1
360
i
(c)
(d)
H
(Spanwise direction)
Figure 4.29
Four different possible segment orientations relative to a screen cross-section. The cross-section
corresponds to = const. Reprinted from Sinha-Ray et al. (2013a), with permission from Elsevier.
tan
1

i1 i
abs
Hi1 Hi
4:110
This value of the angle directly corresponds to the angular orientation of segment (a) in
Figure 4.29. For segments (b)(d) the angular positions are determined as shown in the
gure.
The numerical predictions of the angular distributions of bers in several cross-sections
of the deposited nonwovens at different time moments and for different speeds of screen
motion are depicted in Figures 4.304.33. These gures show the angular distributions
found in the cross-sections taken in the middle of the deposited nonwoven along the
screen (between the leading and trailing edges of the deposit on the screen) corresponding
to the dimensionless time moments from the beginning of the process t = 45, 60, 70 and
75, respectively. In each of these gures, panels (a), (b) and (c) correspond to the following
screen velocities: 0.015 (2.25 m s1), 0.06 (9 m s1) and 0.15 (22.5 m s1), respectively.
Comparing the panels corresponding to the same time moments from the beginning of the
process, but to different screen velocities (say, Figures 4.30ac), one can see that as the
screen velocity increases, the angular distribution changes from bimodal to normal. Since
some jets make loops and cross a chosen cross-section several times, the number of jet
segments used in the calculation of the angular distributions might be larger than the total
number of jets. When the screen velocity increases, the number of the intersections
diminishes, which shows a reduced tendency to looping.
The predicted angular distributions are contrasted with the corresponding ber congurations in the deposited nonwovens in Figures 4.344.37. As in Figures 4.304.33,
the angular distributions were calculated for the cross-sections in the middle (between the
141 [89--178] 10.12.2013 10:00PM
141
Vscreen = 0.015, time = 45, position = 0.608
Total no. of loops - 217

Maximum - 321.5
Minimum - 11.7
Mean - 161.3
40

Maximum - 353
Minimum - 2.5
Mean - 126.5
30
40
No.of loops
No.of loops
60
20
20
10
0
0
(a)
50
100
150
200
250
300
350
(b)
Angle
100
200
300
400
Angle

50

Maximum - 295.1
Minimum - 80.9
Mean - 96.5
No.of loops
40
30
20
10
0
0
(c)
Figure 4.30
50
100
150
Angle
200
250
300
Predicted angular distributions of bers in the lay-downs corresponding to three different

screen velocities at the same dimensionless time moment t = 45 (the corresponding dimensional
time = 0.03 s) from the beginning of the process, which correspond to different locations along
the screens. The dimensionless screen velocities are denoted in the panels. The cross-section
locations (rendered dimensionless by L) corresponding to a given screen velocity and t = 45
are shown in the panels. (a) Vscreen = 0.015 (2.25 m s1), (b) Vscreen = 0.06 (9 m s1),
(c) Vscreen = 0.15 (22.5 m s1). Reprinted from Sinha-Ray et al. (2013a), with permission from
Elsevier.
leading and trailing edges) of the nonwoven mats deposited onto the moving screen. The
detailed ber congurations near the cross-sections considered, shown on the right in
Figures 4.344.37, demonstrate how individual bers loop at lower screen velocities and
become more straight at higher ones. Such ber stretching by faster-moving screens also
leads to the transformation of the bimodal angular distribution into a normal (a Gaussianlike) one. This can be attributed to the following reasons. At the lower screen velocities
the oncoming polymer jets, with their natural propensity to loop, as follows from our
numerical predictions in Sections 4.64.7, have enough time to follow their trajectories
over almost the same location on the screen. On the other hand, at higher screen speeds,
the attached jet segment is rapidly transferred forward and the oncoming jet segments are
inevitably straightened before deposition. This is visualized in detail by the ber congurations on the right in Figures 4.344.37. The nonwoven structures revealed in these
142
142 [89--178] 10.12.2013 10:00PM

Maximum - 354.9
Minimum - 19.1
Mean - 165.3
80
60
40
20

Maximum - 344.1
Minimum - 4
Mean - 123.53
50
No. of loops
No. of loops
40
30
20
10
0
0
(a)
50
100 150 200 250 300 350

Angle
50
(b)
100 150 200 250 300 350

Angle
60

Maximum - 306.7
Minimum - 79.4
Mean - 108.7
50
No. of loops
40
30
20
10
0
0
(c)
Figure 4.31
50
100
150
200
250
300
Angle
Predicted angular distributions of bers in the lay-downs corresponding to three different screen
velocities at the same dimensionless time moment t = 60 (the corresponding dimensional time =
0.04 s) from the beginning of the process, which correspond to different locations along the screens.
The dimensionless screen velocities are denoted in the panels. The cross-section locations
(rendered dimensionless by L) corresponding to a given screen velocity and t = 60 are shown in the
panels. (a) Vscreen = 0.015 (2.25 m s1), (b) Vscreen = 0.06 (9 m s1), (c) Vscreen = 0.15 (22.5 m s1).
Reprinted from Sinha-Ray et al. (2013a), with permission from Elsevier.
gures show that at the lowest screen velocity almost the whole screen is lled with loops
and looks more dense than the lay-downs obtained at higher screen velocities. The
variation in the deposit density associated with multiple loops can have an impact on
nonwoven strength, as discussed in Chapter 8.
The predicted deposited nonwoven patterns on the moving screens were also analyzed
by Sinha-Ray et al. (2013a) to calculate their mass distribution and inhomogeneity. For
mapping the mass distribution the following algorithm was used. The length between two
successive nodes i and i+1 (i) was subdivided into ve equal subsections. The ber
section between two successive nodes has a known polymer mass determined by the
operational conditions (the polymer throughput, and, to some extent, the blowing speed
and viscoelasticity). In a typical meltblowing process, the die-to-collector distance
(DCD) L is of the order of 1015 cm and the blowing speed (Ug0) is of the order of
150 m s1. The shear viscosity of polymer melt is within the range 520 g cm1 s1). In the
143 [89--178] 10.12.2013 10:00PM
143
80
40

Maximum - 312.3
Minimum - 7.4
Mean - 112.1
30
No. of loops
No. of loops
60
20
20
10
0
0
(a)
40
Maximum - 348.3
Minimum - 13.6
Mean - 164.5
50
100
150 200
Angle
250
300
350
50
100
(b)
150 200
Angle
250
300
350

Maximum - 289.8
Minimum - 53.8
Mean - 103
40
No. of loops
30
20
10
0
0
(c)
Figure 4.32
50
100
150
200
250
300
Angle
velocities at the same dimensionless time moment t = 70 (the corresponding dimensional time
= 0.047 s) from the beginning of the process, which correspond to different locations along the
screens. The dimensionless screen velocities are denoted in the panels. The cross-section locations
(rendered dimensionless by L) corresponding to a given screen velocity and t = 70 are shown in
the panels. (a) Vscreen = 0.015 (2.25 m s1), (b) Vscreen = 0.06 (9 m s1), (c) Vscreen = 0.15 (22.5 m s1).
Reprinted from Sinha-Ray et al. (2013a), with permission from Elsevier.
simulations the polymer density was taken as 1 g cm3 and L was taken to be equal to
10 cm. As was established above in the present section, the relaxation time in the bending
part is equal to 0.1 s. The nozzle radius was taken as 100 m (the diameter of 200 m). As
was shown above in the present section, at the end of the straight part of the jet (which is
the beginning of the bending part of the jet) the cross-sectional radius of the polymer jet is
equal to 8.75 m. The subdivision of i into ve subsections attributes to each of them the
same one-fth of the total mass, which is approximately correct if the jet section between
nodes i and i+1 is not too stretched and the accuracy is not lost. Each of the subsection
masses was attributed to its middle point. Coordinates and H of such massive points
on the screen were known. The whole screen was subdivided spanwise into 10 strips.
Each of these strips was dissected into 10 boxes in the direction of the screen motion. As a
result, the whole screen was subdivided into 100 boxes. Masses of all subsections of all
144
144 [89--178] 10.12.2013 10:00PM

80
40

Maximum - 356.9
Minimum - 13.6
Mean - 164.9
40
20
20
10
0
0
(a)

Maximum - 344.12
Minimum - 2.22
Mean - 125.8
30
No. of loops
60
No. of loops
50
100 150 200 250 300 350 400
(b)
Angle
50
100
150
200
250
300
350
Angle

60

Maximum - 321.5
Minimum - 11.7
Mean - 161.3
No. of loops
50
40
30
20
10
0
0
(c)
Figure 4.33
50
100
150
200
250
300
Angle
velocities at the same dimensionless time moment t = 75 (the corresponding dimensional time 0.05 s)
from the beginning of the process, which correspond to different locations along the screens. The
dimensionless screen velocities are denoted in the panels. The cross-section locations (rendered
dimensionless by L) corresponding to a given screen velocity and t = 75 are shown in the panels. (a)
Vscreen = 0.015 (2.25 m s1), (b) Vscreen = 0.06 (9 m s1), (c) Vscreen = 0.15 (22.5 m s1). Reprinted
from Sinha-Ray et al. (2013a), with permission from Elsevier.
jets located inside a certain box were added, divided by four and equally attributed to the
box corners, while the and H coordinates of such points were known. This procedure
yields mass mapping with respect to and H. The procedure is described schematically in
Figure 4.38. The mass distributions found using the above procedure are plotted in
Figures 4.394.42. These gures correspond to different time moments from the beginning of the meltblowing process and each of them incorporates four screen velocities.
Figures 4.394.42 show that as the screen velocity increases, the nonwoven stretch in the
-direction also increases at any xed time moment. On the other hand, they also show
that the nonwoven stretch in the spanwise H direction does not change that much. This
can be explained by the fact that the spanwise width of the nonwoven forms very close to
the landing location of all the jets involved, which happens practically simultaneously.
145 [89--178] 10.12.2013 10:00PM
145
(a)
No. of loops
60
0.75

Maximum - 321.5
Minimum - 11.7
Mean - 161.3
0.7
0.65
0.6
40
0.55
0.5
20
0.8
0.7
0
0.6
0
50
100
150
200
Angle
250
300
350
0.5
(b)
50
0.7
40
Maximum - 287.1
No. of loops
Minimum - 6.38
0.6
Mean - 143.7
30
0.5
20
0.8
10
0.7
0.6
0
0
Figure 4.34
50
100
150
Angle
200
250
300
0.5
Comparison of the predicted angular distributions with the corresponding ber congurations.
Panels (a), (b), (c) and (d) display the results for dimensionless screen velocities of 0.015 (2.25 m s1),
0.03 (4.5 m s1), 0.06 (9 m s1) and 0.15 (22.5 m s1), respectively. The time moment corresponding to
all the cross-sections considered is the same (t = 45; the corresponding dimensional time 0.03 s), while
their -positions are different, since they increase proportionally to Vscreen. All the cross-sections
chosen correspond to the middle part of the deposited nonwoven (between the leading and trailing
edges) where the lay-downs are already fully developed. On the right-hand side of each panel,
the deposited congurations of all jets and three representative jets are shown. Reprinted from
Sinha-Ray et al. (2013a), with permission from Elsevier.
146
146 [89--178] 10.12.2013 10:00PM
(c)
0.7
40

Maximum - 352.2
Minimum - 40.4
No. of loops
30
0.6
Mean - 132.6
0.5
20
10
0
0
50
100
150
200
Angle
250
300
350
0.9
0.8
0.7
0.6
0.5
400 0.4
(d)
50
Maximum - 273
No. of loops
0.7
Minimum - 81.9
40
Mean - 99.4
0.6
30
0.5
20
0.9
0.8
0.7
0.6
0.5
0.4
10
0
0
Figure 4.34
50
100
150
Angle
200
250
(cont.)
However, the results show that the deposited nonwoven mass can be distributed in a
nonuniform manner over the strip formed on the moving screen. In particular, Figure 4.39
shows that at t = 45, at lower screen velocities of 0.015 (2.25 m s1) and 0.03 (4.5 m s1)
(see Figures 4.39a and b), the collected nonwoven mat is denser around the screen center.
147 [89--178] 10.12.2013 10:00PM
147
(a)
0.9
0.8
Maximum - 354.9
Minimum - 19.1
Mean - 165.3
No. of loops
80
60
0.7
0.6
0.5
40
0.7
20
0.6
0
0
50
100
150
200
Angle
250
300
350
0.5
(b)
0.8
50
0.7
40
Total no. of loops -180
0.6
No. of loops
Maximum - 351.15
Minimum - 7
30
Mean - 153.7
0.5
20
1
10
0.8
0
0
Figure 4.35
50
100
150
200 250
Angle
300
350
400
0.6
Panels (a), (b), (c) and (d) display the results for the dimensionless screen velocities of 0.015
(2.25 m s1), 0.03 (4.5 m s1), 0.06 (9 m s1) and 0.15 (22.5 m s1), respectively. The time moment
corresponding to all the cross-sections considered is the same (t = 60; the corresponding
dimensional time = 0.04 s), while their -positions are different, since they increase proportionally
to Vscreen. All the cross-sections chosen correspond to the middle part of the deposited nonwoven
(between the leading and trailing edges) where the lay-downs are already fully developed. On the
right-hand side of each panel, the deposited congurations of all jets and three representative jets
are shown. Reprinted from Sinha-Ray et al. (2013a), with permission from Elsevier.
148
148 [89--178] 10.12.2013 10:00PM
(c)
0.7

40
0.65
Maximum - 329
0.6
Minimum - 25.6
30
Mean - 124.1
No. of loops
0.55
0.5
20
0.7
10
0.6
0.5
0
0
50
100
150
200
250
300
350
X Axis Title
(d)
70
60
50
No. of loops
0.75

Maximum - 295.5
Minimum - 52.5
Mean - 98.1
0.7
0.65
0.6
40
0.55
30
0.5
0.3
20
0.6
0.8
0.8
10
0.6
0.4
0
50
100
150
200
250
300
Angle
Figure 4.35
(cont.)
On the other hand, at higher screen velocities of 0.06 (9 m s1) and 0.15 (22.5 m s1)
(see Figures 4.40c and d), the deposited nonwoven is more uniformly distributed in the
spanwise direction. This is probably related to the fact that at higher screen velocities,
tensile stresses imposed by the screen on the jets through their deposited sections are
higher, and they do not have enough time to relax before the oncoming jet sections are
149 [89--178] 10.12.2013 10:00PM
149
(a)
80
0.9
70
0.8
Maximum - 348.3
Minimum - 13.6
60
0.7
No. of loops
Mean - 164.5
0.6
50
0.5
40
30
0.8
20
0.6
10
0
0
50
100
150
200
Angle
250
300
350
0.4
0.8
(b)
50
0.7

Maximum - 286.8
40
No. of loops
Minimum - 17.4
0.6
Mean - 132
30
0.5
20
1
0.9
0.8
0.7
0.6
0.5
10
0
0
50
100
150
200
250
300
Angle
Figure 4.36
dimensional time is 0.047 s), while their -positions are different, since they increase proportionally
150
150 [89--178] 10.12.2013 10:00PM
(c)
0.7
40
30
No. of loops
0.6

Maximum - 317.2
Minimum - 5.4
Mean - 124.2
0.5
20
0.9
0.8
10
0.7
0.6
0
0
50
100
150
200
Angle
250
300
350
0.5
(d)
0.7
V screen = 0.15 time = 75, position = 0.7392
40
Maximum - 314.2
Minimum - 77.2
Mean - 97.8
No. of loops
30
0.65
0.6
0.55
0.5
20
0.8
0.7
10
0.6
0.5
0.4
0
0
Figure 4.36
50
100
150
200
Angle
250
300
0.3
(cont.)
deposited. Therefore, these tensile stresses affect jet motion toward the screen, rigidly
directing them to the preceding landing spots. At lower screen speeds this does not
happen, since small tensile stresses relax and cannot prevent jets from forming loops near
the center of the blowing gas jet.
At later time moments, the deposited nonwovens obviously stretch further in the
direction of the screen motion (Figures 4.404.42). It should be emphasized that
the mass distribution pattern stays the same, with the nonwoven part deposited during
151 [89--178] 10.12.2013 10:00PM
151
(a)

0.8
No. of loops
80

Maximum - 356.9
Minimum - 13.6
Mean - 164.9
60
0.7
0.6
0.5
40
0.9
0.8
20
0.7
0.6
0
0
50
100
150
200
250
300
350
400
Angle
0.5
(b)
No. of loops
30
Maximum - 341.5
Minimum - 6.9
Mean - 133.3
20
0.8
0.7
0.6
0.5
10
0
0
Figure 4.37
50
100
150
200
Angle
250
300
350
1
0.9
0.8
0.7
0.6
0.5
dimensional time = 0.05 s), while their - positions are different, since they increase proportionally
the preceding period being shifted in the -direction with the screen, while a newly
formed deposit is emerging at a new landing location. Note that the mass mapping at
different screen velocities is done by the plotter with the same colors assigned to
different density ranges, which makes judgments based on just color impossible when
152
152 [89--178] 10.12.2013 10:00PM
(c)
Vscreen = 0.06 time = 75, position = 0.7392
Maximum - 278.6
Minimum - 11.8
Mean - 114.2
30
No. of loops
0.8
0.7
0.6
20
0.5
10
1
0.9
0.8
0.7
0.6
0.5
0.4
0
0
(d)
50
100
200
250
300

40
0.75

Maximum - 280.2
Minimum - 55.4
Mean - 109.6
30
No. of loops
150
Angle
0.7
0.65
0.6
20
0.55
10
0
0
Figure 4.37
50
100
150
Angle
200
250
300
1.2
1.1
1
0.9
0.8
0.7
0.6
0.5
(cont.)
comparing panels (a), (b), (c) and (d) in Figures 4.394.42. Rather, the actual values of
the deposited nonwoven density should be compared to each other.
Sinha-Ray et al. (2013a) also used the numerical code to evaluate the effect of polymer
polydispersity on the resulting nonwoven lay-down. The rhelogical upper-convected
Maxwell model (Eq. 4.58) used is based on the lumped viscosity and relaxation time of
the polymer melt, which practically precludes modeling of the polydispersity effect on
the scale of a single jet. However, in the present case of multiple jets, each jet can be
attributed its own polymer relaxation time, which qualitatively mimics polydispersity
of the polymer in nonwovens. That means that each polymer jet is assigned its own value
of the Deborah number (the dimensionless relaxation time). In addition, the effect of
153 [89--178] 10.12.2013 10:00PM
153
A
C
D
E
C
D
E
F
B
(a)
(b)
(c)
Schematic of the algorithm used for mass mapping to calculate mass distribution in nonwovens,
predicted numerically. (a) The overall structure of several jets deposited on the screen. (b) One of
the jet segments between two consecutive nodes (A and B). The segment AB is subdivided into ve
equal sub-segments, AC, CD, DE, EF and FB, all of which have a mass equal to one fth of the
mass of AB. The mass of each sub-segment is assigned to its middle point, i.e. the ve masses are
assigned to points A, C, D, E and F. (c) The masses of all the sub-segments in a box on the
screen are added, divided by four and equally assigned to its four corners. The grid of such
massive corner points is used to map the corresponding mass distribution. Reprinted from SinhaRay et al. (2013a), with permission from Elsevier.
Figure 4.38
16
10
3.5
14
12
2.5
10
8
6
1.5
0.5
(a)
(cm)
x105
4
(cm)
12
5
6
H (cm)
10
x105
2.5
2
1.5
(b)
1
0.5
5
6
H (cm)
10
x106
x10
25
50
12
20
30
20
10
(cm)
(cm)
10
(c)
Figure 4.39
4
H (cm)
10
(b)
cm2
40
10
15
10
H (cm)
Area density (in g

) of the deposited nonwoven mats at t = 45 (0.03 s) for four different
screen velocities: (a) 0.015 (2.25 m s1), (b) 0.03 (4.5 m s1), (c) 0.06 (9 m s1) and (d) 0.15 (22,5 m s1).
For mass mapping, the density of the polymer is taken to be 1 g cm3. The range of the scale columns on
the right is 105. Reprinted from Sinha-Ray et al. (2013a), with permission from Elsevier.
154
154 [89--178] 10.12.2013 10:00PM
14
12
8
6
4
2
0
0
6
H (cm)
10
12
20
15
10
5
0
10
(c)
0
H (cm)
1
0.5
0
0
6
H (cm)
10
12
12
10
8
6
4
2
0
2
6
H (cm)
10
40
35
30
25
20
15
10
5
0
10
Figure 4.41
70
60
50
40
30
20
10
0
1
(d)
14
x106
10
16
(cm)
10
10
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
H (cm)
Area density (in g cm2) of the deposited nonwoven mats at t = 60 (0.04 s) for four different screen
velocities: (a) 0.015 (2.25 m s1), (b) 0.03 (4.5 m s1), (c) 0.06 (9 m s1) and (d) 0.15 (22.5 m s1).
For mass mapping the density of the polymer is taken to be 1 g cm3. The range of the scale columns
on the right is 105. Reprinted from Sinha-Ray et al. (2013a), with permission from Elsevier.
Figure 4.40
(cm)
1.5
(cm)
(cm)
25
(c)
15
(b)
11
10
9
8
7
6
5
4
3
2
1
30
2.5
x106
35
(a)
20
0
H (cm)
10
12
x105
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
35
30
25
1.5
20
15
10
0.5
5
0
5
(b)
x105
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
x10
40
(cm)
(a)
x10
25
6
H (cm)
10
12
x106
(cm)
(cm)
10
x10
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
(cm)
16
(d)
90
80
70
60
50
40
30
20
10
0
6
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
0
2
4
H (cm)
10
12
Mass mapping (the area density in g cm2) of the deposited nonwoven mats at t = 70 (0.047 s)
for four different screen velocities: (a) 0.015 (2.25 m s1), (b) 0.03 (4.5 m s1), (c) 0.06 (9 m s1),
(d) 0.15 (22.5 m s1). For mass mapping the density of the polymer is taken to be 1 g cm3.
The range of the scale columns on the right is 105. Reprinted from Sinha-Ray et al. (2013a),
155 [89--178] 10.12.2013 10:00PM
155
x10
16
4.5
14
12
3.5
10
(a)
1.6
1.4
30
1.2
20
1
0.8
10
0.6
0
4
1.8
1.5
40
x10
50
2.5
(cm)
(cm)
18
0.4
0
0.5
0.2
0
0
6
H (cm)
10
12
(b)
6
H (cm)
10
12
x10
45
0.8
0.7
25
0.6
20
0.5
15
0.4
10
0.2
0.1
(c)
0.3
10
Figure 4.42
0
H (cm)
10
(cm)
(cm)
30
4.5
90
0.9
35
x106
100
40
80
70
3.5
60
50
2.5
40
30
1.5
20
10
0.5
(d)
2
4
H (cm)
10
12
Mass mapping (the area density in g cm2) of deposited nonwoven mats at t = 75 (0.05 s) for
four different screen velocities: (a) 0.015 (2.25 m s1), (b) 0.03 (4.5 m s1), (c) 0.06 (9 m s1),
(d) 0.15 (22.5 m s1). For mass mapping, the density of the polymer is taken to be 1 g cm3. The
range of the scale columns on the right is 105. Reprinted from Sinha-Ray et al. (2013a), with
polydispersity on polymer melt viscosity can be modeled by assigning separate values of

the activation energy UA to each polymer jet. Effectively, that might be also interpreted as
different solidication rates for different polymer jets. It should be emphasized that the
solidication rate is related to the crystallization rate, which in turn can be different for
different chain lengths (polydispersity). Bansal and Shambaugh (1998) showed that ber
attenuation in meltblowing reaches a plateau at a DCD of about 4 cm, which was
attributed to polymer crystallization.
The effects of the jet-to-jet variation of the Deborah number and the activation energy
UA associated with polydispersity on nonwoven lay-down was studied systematically in
the numerical simulations of Sinha-Ray et al. (2013a) and compared to the experimental
data reported by Begenir (2008). As in Begenirs work (2008), the screen velocity was
taken as 30 m min1, which makes the dimensionless value Vscreen = 0.0033. Note that the
experiments do not distinguish the direction of a ber in the lay-down, i.e. they do not
distinguish the congurations (c) and (d) in Figure 4.29 from (a) and (b). Therefore, for
comparison with the experiments, one has to lump the former with the latter, as is shown
schematically in Figure 4.43. Then, the entire range of the ber orientations considered
shrinks from 360 to only 180.
156
156 [89--178] 10.12.2013 10:00PM
(Direction of screen motion)
Location of cross-section
i/i+1
i/i+1
180-
i+1/i
(a)
i+1/i
(b)
H
(Spanwise direction)
Figure 4.43
Two different distinguishable segment orientations relative to screen cross-section. The crosssection corresponds to = const. Reprinted from Sinha-Ray et al. (2013a), with permission from
Elsevier.
The effect of the Deborah number on the angular orientation in the nonwoven lay-down
modeled at a constant value of activation energy UA =10 (the corresponding dimensional
value ~ 33 kJ mol1) is elucidated in Figure 4.44. In particular, this gure depicts the results
for the Deborah number De = 100 for all jets (Figure 4.44a) and De = 400 for all jets
(Figure 4.44b). It also shows the results for different values of De between the jets, in the
range 100 De 400 in equal steps of De for all 62 jets (Figure 4.44c), as well as in the
range 100 De 400 in equal steps of De for all 62 jets (Figure 4.44d). The values of
the relaxation time 0 corresponding to the range 40 De 400 are 0.01 0 0.1s.
Figures 4.44ad depicts the predicted lay-down patterns for two different cross-sections
along the receiving screen = 0.2 and 0.5 (dimensional values of 2 and 5 cm, respectively).
In addition to the angular distributions, these gures also display the corresponding ber
congurations. Figures 4.44a and b show that at a lower Deborah number (shorter
relaxation time, faster relaxation of the elastic stresses) the angular distribution is wider
than at a higher Deborah number. In addition, it can be seen that although the screen
velocity was the same in Figures 4.44a and b, the lay-down spread on the screen is larger for
a lower Deborah number. These observations point to the fact that as the relaxation becomes
faster, the elastic stresses in the jets become lower and jet excursions easier. Comparing the
predicted results with a xed Deborah number for all the jets with the experimental data of
Bansal and Shambaugh (1998), one can observe that the experimentally measured angular
distributions are wider. However, the numerical results obtained accounting for polydispersity (i.e. with the Deborah number varying from jet to jet) qualitatively mimic the trend
reported by Bansal and Shambaugh (1998) and show widening of the angular distribution
(Figures 4.44c and d). It should be emphasized that there is no signicant difference
between the spread of the angular distributions predicted for 100 De 400 and
157 [89--178] 10.12.2013 10:00PM
157
(a)
0.80.4 0 0.4 0.8 1.2 1.6
1.5
De = 100 and UA = 10
De = 100 and UA = 10
Position of cross-section = 0.2
20.8
21.1
1.2
0.3
0
0.3
0.6
Frequency (%)
0.6
Frequency (%)
0.9
10.4
10.6
0.9
1.2
1.5
0.0
0.0
50
100
150
50
100
150
Angle (Degrees)
Angle (Degrees)
(b)
0.75
De=400 and UA=10
De = 400 and UA=10
Position of cross-sectin=0.5
Position of cross-section=0.2
0.5
37.7
0.25
0
0.25
0.25 0.5 0.75
Frequency (%)
Frequency (%)
33.9
22.6
11.3
28.3
18.9
9.4
0.0
0.0
20
40
60 80 100 120 140 160 180

Angle (Degrees)
20
40
60
80 100 120 140 160 180

Angle (Degrees)
(c)
27.3
0.5
0
0.4
H
Figure 4.44
0.8
1.2
16.9
21.9
Frequency (%)
Frequency (%)
2.5
1.5
De varied between 100 to 400 and UA=10
11.3
16.4
10.9
0.5
5.5
0.0
20 40 60 80 100 120 140 160 180

Angle (Degrees)
5.6
0.0
20
40
60 80 100 120 140 160 180

Angle (Degrees)
The angular distributions predicted for two different cross-sections of nonwoven lay-down on a
screen corresponding to = 0.2 and 0.5. (a) Fixed Deborah number De =100, and xed
dimensionless activation energy of solidication UA = 10; (b) xed De = 400 and xed UA = 10; (c)
De varied from jet to jet in the range 100400, xed UA = 10; (d) De varied from jet to jet in the
range 40400, xed UA = 10; (e) xed De = 400, UA varied from jet to jet in the range 325; (f) De
varied from jet to jet in the range 100400, UA varied from jet to jet in the range 325; (g) De varied
from jet to jet in the range 100400, UA varied from jet to jet in the range 253. Reprinted from
Sinha-Ray et al. (2013a), with permission from Elsevier.
158 [89--178] 10.12.2013 10:00PM
(d)
2

31.6
26.5
1.6
0.8
0.4
0
0.4
1.5 1 0.5 0
0.5
Frequency (%)
1.2
Frequency (%)
26.3
21.1
15.8
10.5
0.8
5.3
1.2
0.0
20
40
17.7
8.8
0.0
60 80 100 120 140 160 180

Angle (Degrees)
50
100
Angle (Degrees)
150
(e)
0
0.25
0.5
0.75
De = 400 and UA varied between 3 to 25

1
1
30.9
0.75
Frequency (%)
0.25
27.7
Frequency (%)
0.5
De = 400 and UA varied between 3 to 25

13.9
24.7
18.5
12.3
6.2
0.25
0.0
0.0
0 20 40 60 80 100120140160180
Angle (Degrees)
20
40
60 80 100 120 140 160 180

Angle (Degrees)
(f)
De varied from 100 to 400 and UA from 3 to 25
1.8
1.5

21.6
1.2
21.4
0.6
0.3
0
0.3
0.6
16.2
Frequency (%)
Frequency (%)
0.9
10.8
5.4
16.0
10.7
5.3
0.0
0.9
0.0
0.0
0
0 0.25 0.5 0.75 1
20
40
60 80 100 120 140 160 180

Angle (Degrees)
20
40
60 80 100 120 140 160 180

Angle (Degrees)
(g)
1.75

1.5

1.25
1
0.5
0.25
0
0.25
Frequency (%)
Frequency (%)
26.5
24.1
0.75
16.1
8.0
0.5
17.7
8.8
0.0
0.75
0.0
0.0
1
0.4
0
0
0.4
0.8
H
Figure 4.44
(cont.)
20
40
60
80 100 120 140 160 180

Angle (Degrees)
20 40 60 80 100 120 140 160 180

Angle (Degrees)
159 [89--178] 10.12.2013 10:00PM
159
40 De 400 (Figures 4.44c and d, respectively). This shows that the angular distribution
saturates below a certain value of the Deborah number.
To study the effect of polydispersity on the solidication rate, the dimensionless
activation energy was varied from jet to jet in the range 325 (a dimensional value of
10 corresponds to 33 kJ mol1; Cheremisinoff 1998), while the value of the Deborah
number was xed at De = 400 (Figure 4.44e). Comparing the results depicted in
Figures 4.44d and e, one may observe that the jet-to-jet variation of the activation
energy of viscosity at a xed value of De has a minor effect on the width of the angular
distribution in the nonwoven lay-down. When the activation energy and the Deborah
number both vary from jet to jet, as in Figures 4.44f and g, the angular distribution
widens (in Figure 4.44f the Deborah number was in the range 100 De 400, while
the dimensionless activation energy varied from 3 for the rst jet to 25 for the last; in
Figure 4.44g the Deborah number was in the range 100 De 400, while the
dimensionless activation energy varied from 25 for the rst jet to 3 for the last). It
can be seen that when varying these two parameters, the wider angular distributions
obtained mimic the trend reported by Bansal and Shambaugh (1998). However, based
on the results shown in Figures 4.44c and d it can be stated that the effect of
polydispersity can be attributed mostly to variation of the relaxation time rather than
to the activation energy of solidication.
The comprehensive quasi-one-dimensional model of the dynamics of threedimensional excursions of polymer jets in meltblowing outlined in the present section
and implemented with 62 jets was used to simulate a number of characteristic cases and
compare the results with the available experimental data. The results obtained revealed
the following. The model is capable of predicting plausible patterns of ber lay-down in
nonwovens formed on a moving screen, as well as the ber size distributions for different
operating conditions. The predicted values of the mean ber diameter and its standard
deviations were found to be in fairly good agreement with the experimental data from the
literature, even though the experimental reports do not specify the rheological behavior
of the polymer solutions, and the giving detailed pattern of the initial interaction of gas
and polymer jets. It was also found that the model reveals both bimodal and normal types
of ber size distributions in nonwovens, in qualitative agreement with the experimental
data. The numerical predictions also suggest that the effect of the turbulence spectrum in
the gas ow, as well as the screen speed, do not affect signicantly the values of the mean
diameter of deposited bers, since no visible trend was found. It was also predicted that
the ber size distribution approaches normal as the screen velocity increases. The
numerical results also demonstrated cohesive breakup of several jets before they reach
the screen, which might be akin to so-called y formation. The numerical results obtained
for the angular distribution and lay-down density in nonwovens deposited onto a moving
screen were also predicted for different operational parameters, including polymer
polydispersity.
The comprehensive model in this section reveals the theoretical/numerical capabilities
for meltblowing modeling comparable to that for electrospinning (Reneker et al. 2000,
Yarin et al. 2001, Thompson et al. 2007) discussed in detail in Chapter 5.
160 [89--178] 10.12.2013 10:00PM
160
4.8
Subsonic and supersonic solution blowing of monolithic

and coreshell fibers
This section introduces the solution-blowing process, which is similar to meltblowing,
but uses polymer solutions instead of melts. As a result, solvent evaporation rather than
melt cooling is responsible for jet solidication, and solution blowing is basically an
isothermal method. Unlike meltblowing, which produces microbers, solution blowing
results in nanobers. Isothermal subsonic blowing of polymer solutions as a method of
forming monolithic bers was introduced by Medeiros et al. (2009) and Sinha-Ray et al.
(2010b). In the latter work both monolithic and coreshell nanobers were formed using
solution blowing. Subsonic blowing employs converging nozzles that do not allow
supersonic regimes, with a gas velocity at the nozzle exit above the chocking velocity,
equal to the local speed of sound of about 310 m s1.
Sinha-Ray et al. (2010b) formed monolithic polyacrylonitrile (PAN) nanobers, and
polymethyl methacrylate (PMMA)polyacrylonitrile (PAN) coreshell nanobers by solution blowing. The experimental setups used for solution blowing and co-blowing are
depicted in Figure 4.45. In particular, the setup for solution blowing used to form monolithic PAN nanobers is shown in Figure 4.45a. The polymer solution was delivered from a
syringe pump to a reservoir attached to a core nozzle. The core nozzle was surrounded by a
concentric annular nozzle. Polymer solution from the reservoir was issued through the core
nozzle into the surrounding high-speed gas jet ow (at 230250 m s1) from the annular
nozzle, similarly to meltblowing in Sections 4.14.7. The gas jet accelerated and stretched
the core jet of PAN solution. The latter was undergoing not only stretching, but substantial
bending instability (see Sections 3.4 and 4.34.7). Stretching and bending instability
resulted in signicant thinning of the polymer solution jet, while solvent was gradually
Syringe pump
core solution
Syringe pump
Shell solution
Polymer solution
Seservoir
Reservoir
Concentric
nozzle
Concentric
nozzle
D.C. source
High pressure
cylinder
High pressure
cylinder
(a)
Figure 4.45
Blown nanofiber
mat
Blown nanofiber
mat
(b)
Schematic drawing of the setup for solution blowing of monolithic nanobers (a), and for
the electrically assisted co-blowing of coreshell nanobers (b). Reprinted from
Sinha-Ray et al. (2010b), with permission from Elsevier.
161 [89--178] 10.12.2013 10:00PM
4.8 Sub- and supersonic solution blowing
161
(a)
(b)
Figure 4.46
Solution blown monolithic polymer (PAN) jet. (a) The multiexposure envelope and (b) a snapshot.
Reprinted from Sinha-Ray et al. (2010b), with permission from Elsevier.
evaporating. At a certain stage, the polymer precipitated, and nanobers solidied and
were collected at a solid collector below the nozzle exit.
The setup used for co-blowing of coreshell PMMAPAN nanobers is sketched in
Figure 4.45b. In this case two different polymer solutions were supplied by two different
syringe pumps. The core solution (PMMA) was supplied into a core nozzle. The core
nozzle was surrounded by a reservoir, to which the shell (PAN) solution was supplied.
PMMA solution was issued through the core nozzle surrounded by an annular orice
through which the shell solution could ow, i.e. at the reservoir exit, a coreshell polymer
jet was formed. The coreshell polymer jet was issued inside a concentric nozzle, which
issued a gas jet surrounding the coreshell polymer jet. Working with PMMAPAN
solutions, it was found that both polymers should be electried, in addition, through the
coreshell nozzle, so that a potential difference of 3 kV was applied between the nozzle
and a grounded nanober collector. The application of the electric eld in blowing
PMMAPAN solutions makes the process akin to co-electrospinning (see Chapter 5).
It should be emphasized that for co-blowing of coreshell bers of some polymers
(see Section 4.9) there was no need in any electrication at all, and the process can be
purely aerodynamic.
A long-exposure photograph of a solution blown jet in ight taken by a DSLR Nikon
D-70s camera (Figure 4.46a) shows large amplitude bending similar to, albeit more
vigorous than, that of the blown solid exible threadline observed in Sections 4.2 and 4.3.
Figure 4.46b shows a snapshot of the jet in ight taken by a CCD camera.
SEM (scanning electron microscopy) images of solution-blown monolithic PAN
nanobers are shown in Figure 4.47. It can be seen that the nanober diameter range is
within 150250 nm (Figure 4.47b) and the individual nanobers are quite uniform in the
longitudinal direction.
An example of the optical images of co-blown PMMAPAN coreshell bers
obtained using an optical microscope is shown in Figure 4.48a. The image suggests a
162
162 [89--178] 10.12.2013 10:00PM
(a)
(b)
10 m
Figure 4.47
1 m
(a) The overall view of a PAN nanober mat produced using solution blowing. (b) A magnied
area of the PAN nanober mat. Reprinted from Sinha-Ray et al. (2010b), with permission
from Elsevier.
0.54 m
Shell (PAN) 2 m
Core (PMMA)
10 m
(a)
0.2 m
(b)
0.2 m
(c)
Figure 4.48
(a) An optical image of coreshell (PMMAPAN) bers. (b) and (c) TEM images of
mesoscopic carbon tubes formed from carbonized PMMAPAN coreshell bers. Reprinted
from Sinha-Ray et al. (2010b), with permission from Elsevier.
core 0.54 m in diameter (larger than the wavelength of blue light, and thus visible
optically) surrounded by a shell of 2 m.
Formation of coreshell instead of monolithic bers in co-blowing can be substantiated by their transformation into hollow mesoscopic carbon tubes after heat treatment.
163 [89--178] 10.12.2013 10:00PM
4.8 Sub- and supersonic solution blowing
163
Core
Sheath
M<1
M<1
Flow
(a)
Figure 4.49
(b)
(a) The core prole in a convergingdiverging Laval nozzle, which allows supersonic gas ow,
for a certain stagnation gas pressure at the exit. (b) Silvent 209L Laval nozzle. The supersonic gas
jet is issued from the core and the surrounding sheath ow is subsonic. This design relieves shock
waves in the core ow, thereby eliminating noise. In addition, the sheath subsonic ow helps to
sustain the supersonic core jet for a longer distance. Sinha-Ray et al. (2013b). Reproduced with
permission from The Royal Society of Chemistry.
Heat treatment can completely eliminate the PMMA core and carbonize the PAN shell,
thus converting the coreshell bers into carbon nanotubes. TEM (transmission electron
microscopy) images of such carbon nanotubes are shown in Figures 4.48b and c. They
demonstrate that carbonized bers in co-blown coreshell mats are indeed hollow, with a
through bore of about 50150 nm diameter. It should be emphasized that the optical
images of the as-blown coreshell bers in Figure 4.48a show outer diameters of about 2
m, whereas the TEM images of the carbonized bers reveal shrinkage, which resulted in
outer diameters in the range 400600 nm.
Sinha-Ray et al. (2013b) introduced supersonic electrically assisted solution blowing.
The nozzle they used to issue the gas jet was a convergingdiverging Laval nozzle. After
the sonic ow speed is reached in the nozzle throat at the end of the converging part, the
velocity continues to grow in the supersonic diverging part of the nozzle (Loitsyanskii
1966). At a certain stagnation gas pressure, supersonic ow at the nozzle exit can be
issued. Figure 4.49a shows the conguration of a Laval nozzle, where the subsonic
diverging part corresponds to Mach number M < 1, and the supersonic diverging part to
Mach number M > 1 (the Mach number is the ratio of the local gas velocity to the local
speed of sound).
The compressed air was supplied from the house pressure line at a supply (stagnation)
pressure of 80 Psi (552 kPa), and the compressed air line was connected to a Silvent 209L
Laval nozzle. The air jet was supersonic up to a distance of 15 mm from the nozzle exit,
and its velocity at the nozzle exit was about 560 m s1.
Polymer (nylon 6) was issued from a needle normal to the direction of the supersonic
gas jet at a rate of Q = 0.1 ml h1. High DC voltage was applied to the needle, while the
Laval nozzle was grounded. Therefore, the polymer jet was stretched and attracted by
electric forces toward the Laval nozzle, as shown in Figure 4.50. Close to the nozzle exit,
the polymer solution jet makes an abrupt kink (Figure 4.50), being entrained by the
supersonic gas jet. The polymer jet experiences stretching due to the pulling aerodynamic
force and due to the aerodynamically driven bending instability (see Section 4.3). As a
result the polymer jet thins, solvent evaporates, the nylon 6 precipitates and the solidied
nanobers are deposited onto a ber collector, which could be fabricated from
164
164 [89--178] 10.12.2013 10:00PM
(a)
(b)
Needle
Nozzle exit
5mm
(c)
(d)
Electric force
Gas flow
Figure 4.50
A nylon 6 (15 wt% solution in formic acid) jet issued from the needle is attracted to the exit
of the Laval nozzle where it makes a kink, since it is entrained and stretched by a supersonic gas jet.
A vigorous stretching and bending instability has set in, so that the jet looks blurred, even though it
is recorded by a CCD camera. Sinha-Ray et al. (2013b). Reproduced with permission from The
Al foil coated
with nanofiber
Figure 4.51
Bare
Al foil
A macroscopically thick nanober sample collected on heavy-duty Al foil with holes to

suspend bers, which became milky compared to bare Al foil. The ruler above the samples is in cm.
Sinha-Ray et al. (2013b). Reproduced with permission from The Royal Society of Chemistry.
conductive materials such as heavy-duty aluminum foil, stainless steel and copper, as
well as nonconducting materials, such as glass and plastics. A plain impermeable sheet of
collector material was located on a rigid substrate normal to the gas-ow direction and no
backward ow resulting from the impinging gas jet was observed. The process could be
continued as long as required and thick nanober mats collected (see Figure 4.51).
Seemingly defectless nanobers were collected in an electrically assisted supersonic
solution-blowing setup (Figure 4.52). The ber size distributions showed that the
majority of the bers were in the range 2050 nm, with only a few outliers in the range
165 [89--178] 10.12.2013 10:00PM
4.9 Blowing of natural biopolymer fibers
165
28.0
Frequency (%)
(a)
21.0
Mean = 53.3 nm
S.D. = 10.32 nm
Maximum = 78.2 nm
Minimum = 22.3 nm
14.0
7.0
0.0
10 20 30 40 50 60 70 80 90
Diameter (nm)
24.0
Frequency (%)
(b)
Mean = 42 nm
S.D. = 10.06 nm
Maximum = 60 nm
Minimum = 28 nm
18.0
12.0
6.0
0.0
10
Figure 4.52
20
30 40 50 60
Diameter (nm)
70
(a) Nanobers formed when a 25G needle was used. (b) Nanobers formed when a 32G needle
was used. In both cases, SEM images of the nanober mats are shown on the left, individual
nanobers in the middle and the ber size distributions on the right. Sinha-Ray et al. (2013b).
Reproduced with permission from The Royal Society of Chemistry.
6080 nm. Therefore, electrically assisted supersonic solution blowing allows ondemand formation of nanobers in a range unattainable by subsonic solution blowing,
discussed above, or electrospinning (Chapter 5).
When a jet of 15 wt% nylon 6 solution in formic acid is stretched from a needle of exit
diameter dn 0.25 mm (as for a 25G needle) and becomes a nanober of diameter df 50
nm, a material element in the polymer jet stretches 0.15(dn/df)2 107 times, irrespective
of whether it started stretching and shrinking from the needle exit or inside the needle.
Therefore, the corresponding stretching ratio is of the order of 107. For dn 0.1 mm (as
for a 32G needle) and df 20 nm the stretching ratio is even higher, of the order of 1010.
The entire stretching process takes about 1 ms (a conservative estimate), and thus the rate
:
of strain is enormous, 1010 s1 (even for the lowest stretching ratio value of 107). For
comparison, note that in the electrospinning process discussed in detail in Chapter 5, the
:
rate of strain is much lower, 103 s1 .
4.9
Blowing of natural biopolymer fibers

Modern technology is experiencing a signicant shift toward biodegradable, or renewable,
or just green agricultural biomolecular materials (Kaplan 1998, Rhim and Ng 2007). The
166
166 [89--178] 10.12.2013 10:00PM
shift is driven by consumer awareness, stringent regulations regarding synthetic waste and
a necessity to reduce dependence on synthetic materials. As a result, signicant efforts are
directed toward production of biodegradable materials from cotton, cellulose, corn, potato
(Kim et al. 2005, 2006, Greene 2007, Yu and Fan 2008) and soy protein (Sinha-Ray 2011)
and development of biodegradable composites (Chabba and Netravali 2005, Huang and
Netravali 2007, 2008). In particular, there is a growing emphasis on making soy-based
green materials. The reason behind the drive towards soy-based materials can be
attributed to the fact that soy protein (sp) is one of the cheapest and most widely grown
vegetable products with a wide range of uses. Booming SoyDiesel production
(Ahmed et al. 1994, Klass 1998) facilitates increasing production of soy, while using
only soy oil and leaving behind abundant (about 80%!) residual soy protein. In addition to
the use of soy protein as a nutrient, it has great industrial value too. Soy protein contains
sialic acid, which is able to inhibit and/or capture the inuenza virus as receptor. Soy
protein can be used as plastics, adhesives and re retardants (Johnson et al. 1984,
Hettiarachchy et al. 1995, Kalapathy et al. 1995). As a result of all these advantages,
there is a growing need and market for nanobers containing soy protein isolates, utilizing
their high surface-area-to-volume ratio, which facilitates their functionality. Several
attempts have been undertaken to produce such bers through electrospinning (Alemdar
and Sain 2008, Phiriyawirut et al. 2008, Vega-Lugo and Lim 2009, Cho et al. 2010, Fung
et al. 2010) or wet spinning (Zhang et al. 2003). In these works electrospinning mostly
relied on blends of a ber-forming polymer and soy protein, since pure soy protein bers
are not exible and it is very hard to achieve submicron bers. Moreover, in such solutionbased methods as electrospinning and wet spinning, dissolution of soy protein can be
facilitated by denaturation by means of thermal treatment (Rangavajhyala et al. 1997,
Zhang et al. 2003), alkali, for example NaOH, sodium sulte and urea (Rangavajhyala
et al. 1997, Lee et al. 2003), as well as pH and ionic strength variation (Lee et al. 2003).
Note also that to improve the tensile strength of protein/polymer bers, cross-linking was
applied (Zhang et al. 2003). Soy-protein-based monolithic and coreshell nanobers were
rst produced by Sinha-Ray et al. (2011) using the solution-blowing method described in
Section 4.8. Soy functionality would add value to the growing nanober market, which is
expected to reach 2.2 billion US$ by 2020 (BCC Research Reports, 2013).
In the present section solution blowing of soy-protein-containing nanobers is discussed following Sinha-Ray et al. (2011). We denote here as monolithic bers made of
sp/polymer blends and distinguish them from coreshell bers where sp is located in the
core, and the polymer is predominantly in the shell. It should be emphasized that
currently it is impossible to form pure sp bers. Indeed, pure soy protein is very brittle
and does not lend itself to ne ber formation. In work by Phiriyawirut et al. (2008)
electrospinning with pure sp was attempted and resulted in spraying only.
For solution blowing of monolithic and coreshell nanobers containing soy protein,
Sinha-Ray et al. (2011) used blends with polyamide-6 (nylon-6). Nylon-6, a common
industrial ber-forming polymer, was chosen since it is soluble in formic acid, which is
also a solvent for soy protein isolate, enabling formation of homogeneous solution
blends. Several soy protein isolates from ADM Specialty Food Ingredients were used:
PRO-FAM 955 (sp 955) and PRO-FAM 974, which have the same composition, but
167 [89--178] 10.12.2013 10:00PM
167
differ in the way they were processed, even though it might be expected that their
denaturation in solution would make them practically indistinguishable.
For solution blowing of monolithic nanobers, blends of nylon-6 and sp 955 in formic
acid were used with a sp/nylon-6 ratio of 40/60 (w/w). Soy protein was completely
dissolved in the acid. It should be mentioned that the pH of formic acid is 2, which is
signicantly lower than pH 4.5 corresponding to the isoelectric point of soy protein. The
color of the soy protein solution changed to transparent brownish instead of opaque,
which corresponded to a complete denaturation of soy protein, as explained by
Gennadios et al. (1993). For the solution blowing of coreshell bers the shell was
formed from a 20 wt% nylon-6 solution in formic acid. The core was prepared from a
blend of sp 955 and nylon 6 in the ratio 1.3/1 (w/w). Nylon-6 was added to the core to
increase viscoelasticity, which facilitates solution blowing.
The aim of Sinha-Ray et al. (2011) was to achieve the highest possible content of soy
protein in solution-blown bers. It was found that for monolithic bers the maximum
possible soy protein content in formic acid was 1/8.5, i.e. 0.1176 g g1. Any further
increase in the soy protein content disrupted the solution-blowing process. On the other
hand, in solution blowing of coreshell bers, the jet was additionally stabilized by the
polymer shell, which allowed a higher soy protein content in formic acid of 1.3/8.7.
Solutions used for blowing experiments were characterized rheologically in a simple
shear experiment. Figure 4.53 depicts the data corresponding to two solutions: (i) a
solution of sp 955 in formic acid, and (ii) a solution of sp 955 and nylon-6 in formic acid.
In both cases the ow curves show pronounced shear-thinning behavior, albeit the shear
viscosity of the sp solution is two to three orders of magnitude lower than that of the sp/
nylon-6 solution.
Solution-blown monolithic nanobers obtained from a blend of soy protein and nylon6 are shown in Figure 4.54. They have no visible grains of protein, as it was completely
dissolved in formic acid. Note that the bers in the SEM images in Figure 4.54 look rather
14000
36
34
12000
Viscosity (cp)
Viscosity (cp)
32
30
28
26
10000
8000
6000
24
4000
22
2000
20
0
(a)
Figure 4.53
20
40
Shear rate (1/s)
60
80
(b)
3
4
5
Shear rate (1/s)
Shear viscosity versus shear rate of two solutions: (a) soy protein (sp 955) solution in formic acid,
(b) sp 955 and nylon-6 solution in formic acid. Triangles correspond to the data acquired when
ramping up the shear rate, while squares correspond to ramping down. Reprinted with permission
from Sinha-Ray et al. (2011). Copyright 2011, American Chemical Society.
168
168 [89--178] 10.12.2013 10:00PM
(b)
(a)
EMS
SEl
4.0kv
X3,300
1m
WD 13.0mm
(c)
EMS
SEl
4.0kv
X13,000
1m WD 14.0mm
(d)
500nm
Figure 4.54
Panels (a) and (b) show SEM images of the monolithic bers formed by blowing a blend of nylon-6
and soy protein PRO-FAM 955 in formic acid. The magnied view of a section of (a) is shown
in panel (b). The bers are rather curly. Panel (c) shows, for comparison, bers formed by blowing a
blend of soy protein PRO-FAM 974 and nylon-6 in a 17/83 ratio in formic acid. The macroscale
web shown in panel (d) corresponds to the nanobers of panel (c). Reprinted with permission
curly, and some knobs are visible in Figure 4.54c, where another type of soy protein
(PRO-FAM 974) was used for comparison. A macroscale web image corresponding to
Figure 4.54c is shown in Figure 4.54d. In dry monolithic nanobers formed using
solution blowing of soy protein (PRO-FAM 955) blends with nylon-6, a protein/polymer
ratio of sp/nylon = 40/60 was achieved, as determined by the initial solution composition
after the solvent had evaporated.
PRO-FAM 974 product data shows nominal amino acid and other mineral contents,
0.62 wt% sulfur from cystine and methionine, and the elements Na (1.21.6 wt%), K
(0.024 0.3 wt%), Ca (0.05 0.15 wt%), P (0.71.0 wt%), Fe (0.008 0.013 wt%) and
Mg (0.025 0.10 wt%). Energy-dispersive X-ray spectroscopy (EDS) was employed to
detect the unique atomic markers from soy protein isolate to differentiate from nylon-6 in
monolithic nanobers shown in Figure 4.54c. Figure 4.55a shows the EDS spectra for a
formic acid solution cast lm of soy protein isolate PRO-FAM 974. The unique markers,
P and S are clearly seen. An EDS spectra of 40/60 sp/nylon-6 monolithic-blend nanobers containing some beads is shown in Figure 4.55b. It should be emphasized that the
blend of sp/nylon-6 blown to form the bers used in Figure 4.55 was absolutely clear,
which is an indication that it was homogeneous and no undissolved protein grains were
present. When such a blend is drawn and rapidly precipitated during the solution-blowing
169 [89--178] 10.12.2013 10:00PM
169
14000 C
12000
Counts
Counts
10000
8000
Na
6000
4000
P
S
2000
0
0
0.5
1.5
(a)
2 2.5 3 3.5
Energy (KeV)
4.5
5000
O
4500 C
4000
3500
3000
2500
2000
1500
1000
Na Al
500
0
0 0.5 1 1.5
(b)
P S
2 2.5 3
Energy (KeV)
3.5
4.5
800
C
Counts
600
400
S3
200
Na
Al
S
50 m
(d)
0
0
(c)
2
3
Energy (KeV)
2500
Na Al
Counts
2000
1500
1000
500
0
0
(e)
Figure 4.55
0.5
1.5
2 2.5 3 3.5
Energy (Kev)
4.5
EDS spectra of (a) cast PRO-FAM 974 soy protein lm; (b) 40/60 sp/nylon-6 nanobers with
beads; (c) a single bead from 40/60 sp/nylon-6 blend; (d) SEM image of ber-only region (box S3)
used for EDS collection in (e); (e) nanober-only region. Reprinted with permission from SinhaRay et al. (2011). Copyright 2011, American Chemical Society.
process, solid nanobers are formed. When some droplets are disconnected and emitted
by a high-speed gas stream, beads appear. The unique markers (P, S) are clearly seen in
the nanober, as was seen in the cast lm. The quantities of S from the soy protein amino
acids are comparable to theoretical (0.6 wt% in pure sp and 0.2% in the 40/60 sp/nylon-6
blend), as shown in Table 4.2. Note that Al was omitted in the analysis since it was due to
the aluminum foil nanober-collection substrate EDS spectra were collected with an
SEM (scanning electron microscope) zoomed in to a single bead, as shown in
170
170 [89--178] 10.12.2013 10:00PM
Table 4.2 Elemental composition from EDS spectrum in

Figure 4.55b (40/60 sp/nylon-6 bers plus beads).
Element
Weight %
Std. Dev
Carbon
Oxygen
Sodium
Phosphorus
Sulfur
Calcium
Total
84.82
14.56
0.14
0.25
0.20
0.03
100
1.84
0.73
0.05
0.09
0.07
0.01
Table 4.3 Elemental composition from EDS spectrum in

Figure 4.55 e (40/60 sp/nylon-6 bers only).
Element
Weight %
Std. Dev.
Carbon
Oxygen
Sodium
Phosphorus
Sulfur
Calcium
Total
87.94
11.29
0.15
0.40
0.16
0.06
100
1.95
0.63
0.06
0.14
0.06
0.03
Figure 4.55c. The P and S peaks are clearly visible, indicating the presence of soy protein.
Finally, EDS spectra were collected with an SEM zoomed in to a nanober-only region,
as shown in the box labeled S3 of Figures 4.55d (image) and e (spectrum). P and S are
again clearly visible at levels consistent with a monolithic blend (Table 4.3). These EDS
spectral data suggest compositional uniformity of a monolithic sp/nylon-6 solution,
resulting in nanobers and some beads with the same composition. The EDS analysis of
the beads showed both S and P, which proves that soy protein precipitated in the beads. The
EDS analysis of a visually ber-only region (Figure 4.55e box labeled S3) clearly revealed
P and S. In particular, such a region in box S3 in Figure 4.55d shows bundles of physically
separated nanobers with no matrix. This proves that the soy protein and nylon-6 coprecipitated to form monolithic bers. In other words, the soy protein was mixed with
nylon-6 in the nanobers and not lying in between the bers.
The SEM images in Figure 4.56 show the cross-sectional view (Figure 4.56a) and the
side view (Figure 4.56b) of nylon-6 based coreshell nanobers. Denaturation of soy
protein by formic acid results in complete solubilization of the former, while the
evaporation of the latter after bers are formed leaves the porous nylon-6 shell seen in
Figure 4.56b. The mechanisms of pore formation in nanobers were discussed in the
context of electrospinning by Dror et al. (2007) and Srikar et al. (2008). They can be
attributed to such factors as direct permeation of the shell by vapor escaping from the
core, as well as to breath gures, i.e. solvent evaporation cooling, which triggers water
171 [89--178] 10.12.2013 10:00PM
(a)
EMS
Figure 4.56
171
(b)
SEl
4.0kv X33,000
100 nm WD 16.0 mm
EMS
SEl
4.0kv X43,000
100 nm WD 16.0 mm
Panels (a) and (b) show SEM images of coreshell bers made with fully denaturated soy protein
and nylon-6. The core is clearly visible in the cross-sectional view in panel (a) because of a
signicant electron contrast between soy protein and nylon-6. The core boundary in panel (a) is also
traced by a white contour, to make it more visible in printed reproduction, and is marked by an
arrow. The side view (b) shows that the nylon-6 shell of this coreshell ber is porous. Reprinted
with permission from Sinha-Ray et al. (2011). Copyright 2011, American Chemical Society.
droplet condensation onto nanobers from the surrounding humid air. The image in
Figure 4.56a shows a signicant grayness gradient between the core and the shell,
which is attributed to the signicant electron contrast between soy protein in the core
and nylon-6 in the shell.
In dry coreshell nanobers the sp/nylon-6 ratio was 32/68, as determined by the
initial solution compositions and the corresponding ow rates, after the solvent had
evaporated.
Fluorescence imaging was used by Sinha-Ray et al. (2011) to map soy protein inside
bers formed by solution blowing of blends of nylon-6 with sp in formic acid, and core
shell bers with sp in the core and nylon-6 in the shell. The ber mats were washed twice
with phosphate-buffered saline (PBS) and then blocked with PBS containing 3% bovine
serum albumin (BSA; 500 l per well) at room temperature for 30 min. The ber mat was
then washed twice with PBS. After that, a rabbit anti-soy antibody at a dilution of 1:2000
in PBS was added overnight at 4 C. Thereafter, the ber mats were washed twice more
with PBS, and then a goat anti-rabbit IgG FITC antibody (Cat. No.10004302, Cayman,
USA) was added at a dilution of 1:100 in PBS for one hour at room temperature. Finally,
the ber mats were washed twice with PBS. Fluorescent images were recorded using a
Nikon microscope (Eclipse E800). Figure 4.57 demonstrates that soy protein is uniformly distributed in the bers with the stained soy protein visible everywhere as a green
color. It should be emphasized that soy protein can show some uorescence under the
illuminating light. However, typically staining is required for more reliable observation.
On the other hand, nylon-6 does not possess any uorescence of its own, and cannot be
stained using the aforementioned procedure.
The coreshell bers are depicted in Figure 4.58. The images reveal that the diameter
of the green uorescent bers in panels (b) and (d) is smaller than the corresponding
diameter of the same bers in the black-and-white optical images in panels (a) and (c). In
172
172 [89--178] 10.12.2013 10:00PM
(b)
(a)
10 m
Figure 4.57
10 m
Optical image (panel (a): black and white) versus uorescent images (panel (b): in color) of the
bers formed by solution blowing a blend of nylon-6 and soy protein in formic acid. Images (a) and
(b) were taken at exactly the same place in a ber mat. Both images were taken with a 100
microscope objective. The green color indicates the presence of soy protein. Reprinted with
permission from Sinha-Ray et al. (2011). Copyright 2011, American Chemical Society.
particular, in panel (a) the optical image of a ber has a diameter of 1 m, whereas the
uorescent image of the same ber is only 0.4 m in diameter. This proves that soy
protein is, indeed, located in the core of the coreshell bers, as intended. Note that the
uorescent markers could easily reach the soy protein core via the nylon-6 shell through
multiple pores, clearly visible in Figure 4.56, and also reported previously for other types
of coreshell bers (Dror et al. 2007, Srikar et al. 2008).
Nonwovens produced by solution blowing are comprised of entangled individual
bers. They practically never end until the process is stopped. In this sense, they are
similar to electrospun nanober mats, discussed in Chapter 5. Therefore, measurements
of ber lengths and aspect ratios are meaningless. On the other hand, cross-sectional
diameter can vary, even along the individual bers due to the effect of turbulent
pulsations in the gas and elastic wave propagation (see Sections 4.24.7). Therefore,
measurements of ber diameter distribution can be important for the structural characterization of solution-blown nonwovens and their further applications. Figure 4.59 depicts
the diameter distributions measured from SEM images of sp/nylon-6 bers blown from
blends, while Figure 4.60 contains the corresponding data for sp/nylon-6 coreshell
bers. It should be emphasized that the blend-blown (monolithic) bers all belong to the
submicron range, whereas the coreshell bers can involve infrequent outliers as large as
2 m. Still, the coreshell bers are submicron on average.
The dry monolithic nanobers produced from soy protein and nylon-6 had a protein/
polymer ratio of 40/60. The dry coreshell bers produced from protein/nylon-6 in the core
and nylon-6 in the shell had a protein/polymer ratio of 32/68. The monolithic nanobers
blown from sp/nylon-6 blends were basically nonporous on the scale 10100 nm, since
images at this scale (e.g. Figure 4.54c) do not resolve any visible porosity, which still might
be present on a smaller scale. The coreshell sp/nylon-6 bers had pores in the shell on the
scale of 10 nm (Figure 4.56b). Porous shells probably allow low-molecular-weight uids to
173 [89--178] 10.12.2013 10:00PM
173
(a)
(b)
1 m
0.4 m
10 m
(c)
10 m
Figure 4.58
10 m
(d)
10 m
Optical images (panels (a) and (c): black and white) and uorescent images (panels (b) and (d): in
color) of coreshell bers with soy protein and a small amount of nylon-6 in the core and 20 wt%
solution of nylon-6 in formic acid used to form the shell. The images in panels (a) and (b),
corresponding to exactly the same place in the ber mat were taken with a 40 microscope
objective. The images in panels (c) and (d) corresponding to exactly the same place in the ber mat
were taken with a 100 microscope objective. Green uorescence indicates the presence of soy
protein. Reprinted with permission from Sinha-Ray et al. (2011). Copyright 2011, American
Chemical Society.
communicate with the core when coreshell bers are submerged in them, which make
them attractive for many applications. The presence of pores in the shell allows staining of
the core with primary rabbit anti-soy antibody and secondary FITC antibody to observe soy
protein in uorescence images. It was shown that in the monolithic nanobers, soy protein
is distributed everywhere in the bers, in agreement with EDS data from SEM. In the core
shell bers, soy protein is located only in the ber cores.
Monolithic and coreshell nano- and microbers containing signicant amounts of
soy protein hold great potential as an effective method of utilizing of the one of the most
abundant natural materials. In particular, it is foreseen as an effective method of utilization and valorization of residual material left by Soy Diesel production, which can be
useful in various applications of nonwovens.
174
174 [89--178] 10.12.2013 10:00PM
Mean value 330 nm

Standard deviation 117 nm
Maximum 786 nm
Minimum 166 nm
23.8
Frequency (%)
19.0
14.3
9.5
4.8
0.0
0
Figure 4.59
100
200
300 400
500
Diameter (nm)
600
700
800
Diameter distribution of sp/nylon-6 blend solution-blown nanobers. Reprinted with permission

23.8
Mean value 910 nm

Standard Deviation 313 nm
Maximum 2051 nm
Minimum 582 nm
Frequency (%)
17.9
11.9
6.0
0.0
0
200
400
600
800 1000 1200 1400 1600 1800 2000 2200

Diameter (nm)
Figure 4.60
Diameter distribution ofsp/nylon-6 coreshell solution-blown bers. Reprinted with permission

4.10
References
Abramovich, G. N., 1963. The Theory of Turbulent Jets. The MIT Press, Boston.
Alemdar, A., Sain, M., 2008. Isolation and characterization of nanobers from agricultural
residues Wheat straw and soy hulls. Bioresource Technology 99, 16641671.
175 [89--178] 10.12.2013 10:00PM
4.10 References
175
Antonia, R. A., Satyaprakash, B. R., Hussain, A. K. M. F., 1980. Measurements of dissipation rate
and some other characteristics of turbulent plane and circular jets. Phys. Fluids 23, 695700.
Ahmed, I., Decker, J., Morris, D., 1994. How Much Does it Take to Make a Gallon of Soydiesel?
Report for the National SoyDiesel Development Board, Jefferson, Missouri.
Appel, D. W., Drost, A. D., Lau, J. C., 1988. Slotted melt-blown die head. US Patent No. 4720252.
Argentina, M., Mahadevan, L., 2005. Fluid-ow-induced utter of a ag. Proc. Natl. Acad. Sci.
USA 102, 18291834.
Bansal, V., Shambaugh, R., 1998. On-line determination of diameter and temperature during
meltblowing of polypropylene. Ind. Eng. Chem. Res. 37, 17991806.
Cambridge.
BCC Research Reports, 2013. Available at http://bccresearch.wordpress.com/2010/06/11/globalmarket-for-nanober-products-to-reach-22-billion-by-2020. Accessed July 31, 2013.
Begenir, A., 2008. Structure-Process-Property Relationships in Elastic Nonwovens Made from
Multi-Block Elastomers. PhD Thesis, Fiber and Polymer Science, North Carolina State
University.
Berger, R. M., 1996. Bicomponent bers and tobacco smoke lters formed therefrom. US Patent
No. 5509430.
Breese, R. R., Ko, W. C., 2003. Fiber formation during melt blowing. Int. Nonwovens J. Summer,
2128.
Breese, R. R., Qureshi, U. A., 2006. Inuence of process conditions on melt blown web structure.
Part IV Fiber Diameter. J. Eng. Fibers Fabrics 1, 3246.
Buehning, P. G., 1991. Melt blowing die. US Patent No. 4986743.
Buntin, R. R., 1976. Battery separators made from polymeric bers. US Patent No. 3972759.
Cheremisinoff, N. P., 1998. Advanced Polymer Processing Operations. Noyes Publications, Saddle
River, New Jersey.
Chabba, S., Netravali, A. N., 2005. Green composites Part 2: Characterization of ax yarn and
glutaraldehyde/poly(vinyl alcohol) modied soy protein concentrate composites. J. Mater. Sci.
40, 62756282.
Cho, D., Nnadi, O., Netravali, A., Joo, Y. L., 2010. Electrospun hybrid soy protein/PVA bers.
Macromol. Mater. Eng. 295, 763773.
Courant, R., Friedrichs, K. O., 1948. Supersonic Flow and Shock Waves. Interscience Publications,
New York
Debye, P., Daen, J., 1959. Stability considerations of nonviscous jets exhibiting surface or body
tension. Phys. Fluids 2, 416421.
Doi M., 1980. Molecular rheology of concentrated polymer systems. J. Polymer Sci., Polym. Phys.
Ed. 18, 10051020.
Dror, Y., Salalha, W., Avrahami R., Zussman E., Yarin A. L., Dersch, R., Greiner, A.,
Wendorff, J. H., 2007. One-step production of polymeric micro-tubes via co-electrospinning.
Small 3, 10641073.
Entov, V. M., Kestenboim, Kh. S., 1987. Mechanics of ber formation. Fluid Dynamics 22, 677686.
Fabbricante, A., Ward, G., Fabbricante, T., 2000. Micro-denier nonwovens materials made using
modular die units. US Patent No. 6114017.
Frost, W., Moulden, T. H. (Editors), 1977. Handbook of Turbulence, Volume 1. Plenum Press,
New York.
Fung, W., Yuen, K., Liong, M., 2010. Characterization of brous residues from agrowastes and the
production of nanobers. J. Agri. Food Chem. 58, 80778084.
176
176 [89--178] 10.12.2013 10:00PM
Gennadios, A., Brandenburg, A. H., Weller, C. L., Testin, R. F., 1993. Effect of pH on properties of
wheat gluten and soy protein isolate lms. J. Agric. Food Chem. 41, 18351839.
Greene, J., 2007. Biodegradation of compostable plastics in green yard-waste compost environment. J. Polym. Environ. 15, 269273.
Hettiarachchy, N. S., Kalapathy, U., Myers, D. J., 1995. Alkali-modied soy protein with improved
adhesive and hydrophobic properties. J. Am. Oil Chem. Soc. 72, 14611464.
Howard, R. E., Young, J., 1993. Nonwoven webs of microporous bers and laments. US Patent
No. 5230949.
Huang, X., Netravali, A. A. N., 2007. Characterization of ax ber reinforced soy protein
resin based green composites modied with nano-clay particles. J. Compos. Sci. Technol. 67,
20052014.
Huang, X., Netravali, A. A. N., 2008. Environmentally friendly green materials from plant-based
resources: Modication of soy protein using gellan and micro/nano-brillated cellulose.
J. Macromol. Sci., Part A: Pure Appl. Chem. 45, 899906.
Johnson, L. A., Myers, D. J., Burden, D., 1984. Early uses of soy protein in Far East. US Inform 3,
282284.
Kalapathy, U., Hettiarachchy, N. S., Myers, D., Hanna, M. A., 1995. Modication of soy proteins
and their adhesive properties on wood. J. Am. Oil Chem. Soc. 72, 507510.
Kanno, T., Matsushima, Y., Suzuki, M., 1992. High-strength non-woven fabric, method of
producing same and battery separator constituted thereby. US Patent No. 5089360.
Kaplan, D. L. (Editor), 1998. Biopolymers from Renewable Resources. Springer, Berlin.
Kim, C. W., Frey, M. W., Marquez, M., Joo, Y. L., 2005. Preparation of submicron-scale,
electrospun cellulose bers via direct dissolution. J. Polym. Sci., Part B: Polym. Phys. 43,
16731683.
Kim, C. W., Frey, M. W., Marquez, M., Joo, Y. L., 2006. Structural studies of electrospun cellulose
nanobers. Polymer 47, 50975107.
Klass, D. L., 1998. Biomass for Renewable Energy, Fuels, and Chemicals, Academic Press,
New York.
Komatsu, M., Narukawa, K., Yamamoto, N., 1988. Process for producing hydrophilic polyolen
nonwoven fabric. US Patent No. 4743494.
Kwok, K. C., Bolyard, Jr., E. W., Riggan, Jr., L. E., 1999. Meltblowing method and system. US
Patent No. 5904298.
Lau, J. C., Haynes, B. D., 1998. Apparatus for the production of bers and materials having
enhanced characteristics. US Patent No. 5711970.
Lee, K. H., Ryu, H. S., Rhee, K. C., 2003. Protein solubility characteristics of commercial soy
protein products. J. Am. Oil Chem. Soc. 80, 8590.
Moscow, 1970.
Marheineke, N., Wegener, R., 2007. Fiber dynamics in turbulent ows: Specic Taylor drag.
SIAM J. Appl. Math. 68, 123.
Mattheij, R. M. M., Rienstra, S. W., ten Thije Boonkkamp, J. H. M., 2005. Partial Differential
Equations. SIAM, Philadelphia.
177 [89--178] 10.12.2013 10:00PM
4.10 References
177
McCulloch, J. G., 1999. The history of the development of melt blowing technology. Int.
Nonwovens J. 8, 6672.
Medeiros, E. S., Glenn, G. M., Klamczynski, A. P., Orts, W. J., Mattoso, L. H. C., 2009. Solution
blow spinning: A new method to produce micro- and nanobers from polymer solutions. J. Appl.
Polym. Sci. 113, 23222330.
Mende, T., Sakai, T., 1991. Melt-blowing die. US Patent No. 5017112.
Mende, T., Sakai, T., 1992. Melt-blowing method having notches on the capillary tips. US Patent
No. 5171512.
Midkiff, D. G., 2001. Filtration media and articles incorporating the same. US Patent No. 6322604.
Mozelack, B., Schmitt, R. J., Barboza, S. D., Jana, P., Nguyen, S. N., Gschwandtner, R. R.,
Connor, R. D., Yingling, T. W., 2003. Apparatus for making melt-blown lter cartridges. US
Patent No. 6662842.
Nishioi, H., Ogata, S., Tsujiyama, Y., 1994. Microbers-generating bers and a woven or
non-woven fabric of microbers. US Patent No. 5290626.
Pall, D. B., 1996. Melt-blown brous web. US Patent No. 5582907.
Phiriyawirut, M., Rodchanacheewa, N., Nensiri, N., Supaphol, P. 2008. Adv. Mat. Res. 5557, 733736.
Pope, S. B., 2000. Turbulent Flows. Cambridge University Press, Cambridge.
Prentice, J. S., 1978. Laminated non-woven sheet. US Patent No. 4078124.
Rangavajhyala, N., Ghorpade, V., Hanna, M., 1997. Solubility and molecular properties of heatcured soy protein lms. J. Agric. Food Chem. 45, 42044208.
Rhim, J. W., Ng, P. K. W., 2007. Natyral biopolymer-based nanocomposite lms for packaging
applications. Crit. Rev. Food Sci. Nutr. 47, 411433.
Rubhausen, A., Roock, D., 1993. Apparatus for blow-extruding laments for making a eece. US
Patent No. 5248247.
Schwartz, E., 1983. Apparatus and process for melt-blowing a berforming thermoplastic polymer
and product produced thereby. US Patent No. 4380570.
Schwartz, E., 1995. Apparatus and process for uniformly melt-blowing a berforming thermoplastic polymer in a spinnerette assembly of multiple rows of spinning orices. US Patent No.
5476616.
Shambaugh R. L., 1988. A macroscopic view of the melt-blowing process for producing microbers. Ind. Eng. Chem. Res. 27, 23632372.
Shipp, Jr., P. W., Vogt, C. M., 1987. Melt-blown material with depth ber size gradient. US Patent
No. 4714647.
Sinha-Ray, S., Lee, M. W., Sinha-Ray, S., An, S., Pourdeyhimi, B., Yoon, S. S., Yarin, A. L., 2013b.
Chem. C 1, 34913498.
Sinha-Ray, S., Yarin, A. L., Pourdeyhimi, B., 2010a. Meltblowing: I-Basic physical mechanisms
and threadline model. J. Appl. Phys. 108, 034912.
Sinha-Ray, S., Yarin, A. L., Pourdeyhimi, B., 2010b. The production of 100/400 nm inner/outer
diameter carbon tubes by solution blowing and carbonization of coreshell nanobers. Carbon
48, 35753578.
Sinha-Ray, S., Yarin, A. L., Pourdeyhimi, B., 2013a. Prediction of angular and mass distribution in
meltblown polymer laydown. Polymer 54, 860872.
178
178 [89--178] 10.12.2013 10:00PM
Sinha-Ray, S., Zhang, Y., Yarin, A. L., Davis, S. C., Pourdeyhimi, B., 2011. Solution blowing of
soy protein bers. Biomacromolecules 12, 23572363.
Srikar, R., Yarin, A. L., Megaridis, C. M., Bazilevsky, A. V., Kelley, E., 2008. Desorption-limited
mechanism of release from polymer nanobers. Langmuir 24, 965974.
Terakawa, T., Nakajima, S., 1996. Spinneret device for conjugate melt-blow spinning. US Patent
No. 5511960.
Thompson, C. J., Chase, G. G., Yarin, A. L., Reneker, D. H., 2007. Effect of parameters on nanober diameter determined from electrospinning model. Polymer 48, 69136922.
Tikhonov, A. N., Arsenin, V. Y., 1977. Solutions of Ill-Posed Problems. Winston, New York.
Vega-Lugo, A., Lim, L., 2009. Controlled release of allyl isothiocyanate using soy protein and
poly(lactic acid) electrospun bers. Food Res. Int. 42, 933940.
Wente, V. A., 1954. Manufacture of Superne Organic Fibers. US Department of Commerce,
Ofce of Technical Services Report No. PBI 11437, Naval Research Laboratory, Report 4364.
Wente, V. A., 1956. Superne thermoplastic bers. Industrial and Engineering Chemistry 48,
13421346.
Yarin, A. L., 2007. Self-similarity. Section 2.3 in Springer Handbook of Experimental Fluid
Mechanics (Ed. Tropea, C., Yarin, A. L., Foss, J.). Springer, Berlin, pp. 5782.
Yarin, A. L., Sinha-Ray, S., Pourdeyhimi, B., 2010. Meltblowing: II-Linear and nonlinear waves on
viscoelastic polymer jets. J. Appl. Phys. 108, 034913.
Yu, G., Fan, Y., 2008. Preparation of poly(D, L-lactic acid) scaffolds using alginate particles.
J. Biomater. Sci. Polym. 19, 8798.
Zhang, X., Min, B. G., Kumar, S., 2003. Solution spinning and characterization of poly(vinyl
alcohol)/soybean protein blend bers. J. Appl. Polym. Sci. 90, 716721.
Ziabicki, A., Kawai, H. (Editors), 1985. High-Speed Fiber Spinning. John Wiley & Sons,
New York.
179 [179--261] 10.12.2013 8:33PM
Electrospinning of
micro- and nanofibers
This chapter deals with the mechanisms and electrohydrodynamic modeling of the
physical processes resulting in electrospinning of nanobers with cross-sectional
diameters approximately in the range 100 nm to 1 m. These involve the physical
nature of uids used in electrospinning, leaky dielectrics, discussed in Section 5.2, and
the formation of the precursor of electrospun jets, the Taylor cone, described in
Section 5.3. Polymer jets in electrospinning possess an initial straight section,
which is discussed in Section 5.4. Experimental observations of the key element of
the electrospinning process, the electrically driven bending instability, which is
similar to the aerodynamically driven jet bending of Chapters 3 and 4, are covered in
Section 5.5. Section 5.6 describes the theory of the bending instability in electrospinning. Multiple jet interaction in electrospinning and needleless electrospinning
are discussed in Section 5.7. Co-electrospinning and emulsion electrospinning of
coreshell bers (Section 5.8) are based on similar physical principles to electrospinning of monolithic nanobers. The electrostatic eld-assisted assembly techniques
developed with the aim of positioning and aligning individual nanobers in arrays
and ropes are discussed in Section 5.9. Melt electrospinning of polymer bers is briey
outlined in Section 5.10.
5.1
Electrospinning of polymer solutions

Electrospinning of polymer solutions, liquid crystals, suspensions of solid particles
and emulsions employs an electric eld of the strength about 1 kV cm1. The rst
US patent on electrospinning was issued to Formhals (1934), but interest to this
process was dormant until electried jets of polymer solutions and melts were
investigated as routes to the manufacture of polymer nanobers (Baumgarten 1971,
Larrondo and Manley 1981ac, Doshi and Reneker 1995, Reneker and Chun 1996).
In electrospinning, the electric force results in an electrically charged jet owing
out from a pendant or sessile droplet (see Figure 5.1). After the jet ows away
from the droplet in a nearly straight line, it bends into a complex path and other
changes in shape occur, during which electrical forces stretch and thin it by very
large ratios, quite similar to the effects of the aerodynamic forces in melt- and
solution blowing discussed in Chapter 4. After the solvent evaporates, solidied
nanobers are left.
180
180 [179--261] 10.12.2013 8:33PM
5 Electrospinning of micro- and nanofibers
Insulating tube
Syringe
Syringe pump
ON
OFF
. . .
Pendant drop
Power supply
H
Envelope
cone
Flat copper
collector
PC
.
LPF
Electronic filter
Figure 5.1
Schematic drawing of the electrospinning process, showing the jet path, and some parts of the
experimental setup. Reprinted from Theron et al. (2004), with permission from Elsevier.
5.2
Leaky dielectrics
The theory of electricity traditionally dealt with either perfectly conducting solids termed
conductors, or perfectly nonconducting, polarizable solids termed dielectrics ( Landau
and Lifshitz 1984, Smythe 1989, Feynman et al. 2006), or solid semiconductors (Yu and
Cardona 2010). Electrolytes and certain aspects of their ows were studied in the
framework of electrochemistry (Levich 1962), while such liquids as transformer oils,
refrigerants and various organic solvents were considered to be insulating liquids (i.e.
liquid dielectrics) and used in this capacity. Relatively recently, the concept of leaky
dielectrics was introduced (Melcher and Taylor 1969, Russel et al. 1989, Saville, 1997,
Castellanos and Perez 2007, Chang and Yeo 2010). This concept attributes a certain
degree of electric conductivity to the so-called insulating liquids. It resulted from a
number of hydrodynamic phenomena arising after the electric eld is applied to such
liquids. These phenomena were attributed to charge transport, in other words, revealed
the so-called insulating liquids as conductors, or more accurately, poor conductors
(Melcher and Taylor 1969, Ostroumov 1979, Saville, 1997, Castellanos and Perez
2007, Chang and Yeo 2010). It was demonstrated that such liquids are akin to weak
electrolytes, represent ionic conductors and can possess either positive or negative
net charge. The origin of the ions in such liquids is still debated. It can be attributed to
self-dissociation of the liquid in strong electric elds, or to dissociated impurities
accompanied or not by irreversible Faradaic reactions at the electrodes (Theron et al.
2004, Castellanos and Perez 2007, Yazdani and Seyed-Yagoobi 2009), or direct charge
injection (Castellanos and Perez 2007, Shrimpton 2009). In particular, low-conductivity
181 [179--261] 10.12.2013 8:33PM
5.2 Leaky dielectrics
181
hydrocarbons do not ionize and the ions responsible for their electrical conductivity
originate from spontaneous dissociation of tiny amounts of admixtures within
them (Castellanos and Perez 2007). When the electrical conductivity is associated with
dissociated impurities, irreversible Faradaic reactions are practically eliminated on such
electrodes as gold covered by a thin titanium layer, and so-called perfect polarization near
the electrodes associated with only pre-existing admixture ions can be assumed (Green
et al. 2002). On the other hand, a direct charge injection is possible using electron beams
or corona discharges from needles or blades over the free surface of a carrier liquid
(Castellanos and Perez 2007). Ion-exchange membranes can also be used to create pools
of ions suitable for injection as the electric eld is applied. Direct charge injection from
electrodes submerged into a carrier liquid without any dissociating additives is also
implied sometimes.
A possible mechanism for net charge formation in electrospinning is illustrated
in Figure 5.2. Direct measurements show that conductivity of solutions used in
electrospinning is typically associated with their solvents (Theron et al. 2004). Assume
that the ions present are provided by dissociated pre-existing admixtures, and the original
solution is electro-neutral. When such polymer solution enters a plastic syringe with an
inserted positive electrode, as shown in Figure 5.2, negative ions drift toward the
electrode and discharge there. The drift velocity of ions vd is given as (Levich 1962,
Bockris and Reddy 1970)
vd
ezi
E
6a
5:1
where e is the absolute value of the electron charge, zi is the valence of the ions, is the
solvent viscosity, a is the mean radius of the ions and E is the effective electric
eld strength inside the syringe. Taking as an estimate |E| = 0.3 kV cm1, zi = +1,
= 1031 Pa s, e = 1.6 1019 C and a = 108cm, |vd| is estimated to be of the order
of 1 cm s1 to 103 cm s1. On the other hand, the uid velocity in the syringe
during electrospinning is of the order of 104 cm s1, which is much smaller than the
drift velocity of the ions. The residence time of liquid elements in the syringe near the
submerged electrode is on the scale of at least several seconds. Therefore, negative ions
in the case depicted in Figure 5.2 have enough time to reach the positive electrode and
discharge there. On the other hand, positive ions are repelled from this electrode. As a
result, the balance between the negative and positive ions is broken, and a net bulk
charge builds up in the owing polymeric uid. Therefore, at the syringe exit, the
polymer jet will be positively charged in the situation depicted in Figure 5.2. The rate of
replenishment of the ion content is dictated by the volumetric ow rate and does not
vary that much in electrospinning. On the other hand, the rate of withdrawal and
discharge of the opposite charges by the electrode is dependent on the residence
time of ions in contact with the needle. At higher ow rates, fewer (negative) ions
are withdrawn and discharged at the electrode. Therefore, the uid carries a lower
(positive) net charge at the time it arrives at the syringe exit and forms the polymer
jet. This conclusion holds even though some ions end up at the inner wall of the
syringe. As a result, the net bulk charge in the jet is expected to decrease at higher
182
182 [179--261] 10.12.2013 8:33PM
Copper wire connected

to a positive high voltage
Copper
brushes
Insulation
material
Copper piston
Rubber
band
e- e-
Copper needle
Negative ions
E
Positive ions
Vd
Polymer
solution
Applied electric field

inside the syringe
Plastic syringe
Orifice (0.3 mm)
Figure 5.2
A possible mechanism for the appearance of net charge in electrospinning. Reprinted from Theron
volumetric ow rates. It should be emphasized that net charges in electrospun polymer

jets rapidly migrate to the free surface, as the direct calculations show (Reznik et al.
2006). Therefore, electrospun jets represent a typical leaky dielectric case, where
all uncompensated charges are located at the free surface, while the bulk is fully
electro-neutral. The interactions of the net surface charges with the imposed electric
eld and with each other provide the driving forces for all phenomena characteristic of
electrospinning.
Magnetic properties of electrolytes (of whatever origin) are such that their
magnetization is negligibly small. Since the overall (ohmic plus convective) electric
currents in electrospinning are in the nA range, the force resulting from any interaction of
electrospun jets with any available magnetic eld is typically negligibly small, and one
can disregard any magnetic effects.
Table 5.1 lists values of the dielectric constant (relative permittivity) and electric
conductivity e of several typical polymer solutions in water and organic solvents, as
well as for the pure liquids and solvents relevant in electrospinning. All these liquids
can be characterized as leaky dielectrics, since their conductivity is very low. For
comparison, the electric conductivities of copper, mercury and molten sodium chloride
are 5.8 107 S m1 (at room temperature), 1.04 104 S m1 (at room temperature) and
390 S m1 (at 1173.15 C), respectively.
183 [179--261] 10.12.2013 8:33PM
5.3 Taylor cone and jet initiation
183
Table 5.1 Representative values of and for several leaky dielectric liquids (polymer solutions in different solvents, pure
solvents and several pure liquids): polyethylene oxide (PEO), polyacrylamide (PAA), polyvinyl alcohol (PVA), polyurethane
(PU), polycaprolactone (PCL), methylene chloride (dichloromethane, MC), dimethylformamide (DMF) and tetrahydrofuran
(THF). For convenience note that conversion of the SI unit for the electric conductivity (S m1) into the corresponding
Gaussian unit is done as 1 S m1 = 9 109 s1.
Liquid
e (mS/m)
Distilled water
Ethanol (95%)
Acetone
Transformer oil
Glycerin
MC/DMF (40/60)
MC/DMF (75/25)
THF/ethanol (50/50)
2% PEO Mw = 6 105 in ethanol/water (40/60)
2% PEO Mw = 106 in ethanol/water (40/60)
1% PEO Mw = 4 106 in ethanol/water (40/60)
2% PEO Mw = 106 in water
6% PAA Mw = 2.5 105 in ethanol/water (40/60)
6% PVA Mw = 104 in ethanol/water (50/50)
6% PU Tecoex in THF/ethanol (50/50)
8% PCL Mw = 8 104 in acetone
10 % PCL Mw = 8 104 in MC/DMF (75/25)
88.75
24.55
20.7
2.24
39.1
69.47
29.82
21.3
15.79
67.09
66.71
66.12
81.96
79.5
65.99
16.75
25.2
18.55
0.447
0.0624
0.0202
108106
1.0
0.150
0.505
0.273
0.037
0.85
0.81
1.102
9.43
24.47
3.73
0.093
0.142
0.191
5.3
Taylor cone and jet initiation

When a charged sessile or pendant drop of polymer solution in contact with an inserted
electrode experiences the electric eld existing between this electrode and a counter
electrode at some distance from it, it acquires a stable shape if the potential difference is
not too high. The electric eld imposes electric Maxwell stresses pulling the drop
toward the counter electrode and stretching the drop. On the other hand, surface tension
always tends to minimize the drop surface (which is equivalent to minimization of
surface energy) making it the shape of a spherical segment. A possible drop shape
arising as a result of the interplay between Maxwell stresses and restoring
surface tension is shown in Figure 5.3a. If the electric Maxwell stresses prevail in a
supercritical electric eld, the equilibrium cannot be sustained and jetting begins from
the drop tip (Figure 5.3b).
Steady-state drop shapes result from the equilibrium of electric forces and surface
tension, irrespective of the viscoelastic forces, which all vanish in statics. Liquid in a
static drop attached to an electrode located at some distance from a counter electrode, can
be considered as a perfect ionic conductor, even though this is a leaky dielectric liquid
(Taylor 1964). The reason that this assumption is valid in the present case is related to the
184
(a)
Figure 5.3
184 [179--261] 10.12.2013 8:33PM
(b)
(a) An electried drop stretched by electric Maxwell stresses resembling a conguration similar to
the Taylor cone. (b) A jet is issued from the tip 2 ms after, since the electric eld is supercritical.
Reprinted from Reneker and Yarin (2008), with permission from Elsevier.
characteristic charge relaxation time C = /(4e). Note that here and hereinafter all the
equations that contain terms that depend on the electric eld are expressed in Gaussian
(CGS) units, and the values of all the corresponding parameters are given in CGS units
(Landau and Lifshitz 1984). This is especially convenient and customary in cases where
both electrical and mechanical parameters are involved, in particular, in the discussion of
electrospinning. The values of the electric potential, the electric eld strength, the electric
current and conductivity are also converted to SI units for convenience. Plausible
values for the parameters of polymer solutions used in electrospinning are 40
and e= 9 102 to 9 106 s1, which is 107 to 103 S m1 (see Table 5.1). Therefore,
c = 0.00035 to 3.5 ms. If a characteristic hydrodynamic time H >> C, then the electric
behavior of a leaky dielectric reduces to that of a perfect conductor, even though it is
actually a poor conductor compared to such good conductors as metals. In the present
section H ~ 1 s is of the order of the residence time of uid particles in drops sustained at
the needle exit. Therefore H >> C, and the approximation of a perfect conductor is fully
justied. Then, the excess anions or cations have enough time to escape to the drop
surface, where they are distributed nonuniformly. The drop surface is equipotential, since
liquid behaves as a perfect conductor, with the excess charges distributed in a way that
maintains a zero electric eld inside the liquid.
Considering the critical drop shape that can be sustained as an ultimate equilibrium
conguration resulting from the competition of electric and surface forces, we follow
Yarin et al. (2001b), which allows us to uncover some intrinsic features of the Taylor
cone. We consider an axisymmetric liquid body kept at a potential (0 + const) with its tip
at a distance a0 from an equipotential plane, a counter electrode (Figure 5.4). The
distribution of the electric potential = + const is described by the spherical
coordinates R and , and by the cylindrical coordinates and z (see Figure 5.4). The
free surface is assumed to be in equilibrium, i.e. the electrical forces acting on the drop in
Figure 5.4 are balanced by the surface tension forces. The potential 0 can always be
185 [179--261] 10.12.2013 8:33PM
185
Fluid body at potential

(0 + constant
R
a0
= Constant
Figure 5.4
Axisymmetric innite uid body kept at potential 0 = 0 + const. at a distance a0 from an

equipotential plane kept at = const. Reprinted with permission from Yarin et al. (2001b).
expressed in terms of the surface tension coefcient and a0, specically as 0 = C

(a0), which follows from the dimensional analysis, with C being a dimensionless
factor. Due to the dimensional arguments, the general representation of is, in the present
case, = 0F1(R / a0, ), where F1 is a dimensionless function (Yarin 2012). The value of
the potential throughout the space that surrounds the liquid drop is given by

R
1=2
a0 F
; const:
5:2
a0
where F = C F1 is a dimensionless function.
At distances R >> a0, one can assume that the inuence of the gap a0 is small. Then, the
function F should inevitably approach the following power-law scaling:

1=2
R
R
F
;
5:3
a0
a0
with () being a dimensionless function, whereupon Eq. (5.2) takes the asymptotic
self-similar form, independent of a0
R1=2 const:
5:4
Note that power-law scalings resulting in self-similar solutions are common in boundarylayer theory (see, for example, Schlichting 1979, Zeldovich 1992, Yarin 2007,
and references therein). In particular, such self-similar solutions for jets and plumes,
considered as issuing from a point source, in reality correspond to the non-self-similar
solutions of the boundary-layer (Prandtl) equations for jets and plumes being issued from
186
186 [179--261] 10.12.2013 8:33PM
nite-size nozzles, at distances much larger than the nozzle size, i.e. they represent
remote asymptotics (Dzhaugashtin and Yarin 1977). The self-similar solution for capillary waves generated by the weak impact of a droplet of diameter D onto a thin liquid
layer emerges at distances much greater than D from the center of impact (Yarin and
Weiss 1995). The self-similar solution for the electric eld in Eq. (5.4) is motivated by
precisely the same idea, and is expected to correspond to the limiting behavior of all nonself-similar solutions at distances R >> a0.
The solution (Eq. 5.4) should also satisfy the Laplace equation, which governs the
electric potential (Landau and Lifshitz 1984, Smythe 1989, Feynman et al. 2006). Thus,
one nds the function (Taylor 1964) as
P1=2 cos
5:5
where P (cos ) is a Legendre function of order .

The free surface becomes equipotential only when corresponds to the single zero of the
function P (cos ) in the range 0 , which is 0 = 130.7 (Taylor 1964).
Then, according to the self-similar equilibrium solution, a uid body similar to the one
shown in Figure 5.3a should be enveloped by a cone with the half-angle at its tip equal to
= T = 0 = 49.3, which is the Taylor cone (the cone-like drop shown in Figure 5.3a,
albeit with a half-angle different from that predicted). The shape of the liquid body in
Figure 5.4 would then approach the Taylor cone asymptotically as R ! . (Note also that
Pantano et al. (1994) considered a nite drop attached to a tube). The self-similarity
assumption implicit in Taylors theory and leading to Eqs. (5.3) and (5.4) also implies
that ! as R ! , which is quite peculiar. Below in this section we show that relevant
non-self-similar solutions do not follow this trend as R ! , which means that they are
fundamentally different from the self-similar solution corresponding to the Taylor cone.
The experimental data of Taylor (1964) and numerous subsequent works show that
drops acquire a static shape that does not depend on the initial shape. This static shape is
stable if the strength of the electric eld does not exceed a critical level at which
jetting begins from the drop tip. As the electric eld strength approaches the critical
value, the drop shape approaches that of a cone with a rounded tip. The radius of
curvature of the tip can become too small to be seen in an ordinary photograph, as
discussed below. Nevertheless, the tip should be rounded, since otherwise the electric
D
x = x0 < 1
x = Constant
C
=1
0
Figure 5.5
= Constant
a0(~100 nm)
x=0
Prolate spheroidal coordinate system about a hyperboloidal liquid body BCD. Reprinted with
permission from Yarin et al. (2001b). Copyright 2001, AIP Publishing LLC.
187 [179--261] 10.12.2013 8:33PM
187
eld strength would become innite at the tip. Detailed calculation of the exact droplet
shape near the tip is an involved nonlinear integro-differential problem, since the electric
eld depends on the drop shape, and vice versa (see below in this section). To simplify
such calculations, approximate methods were proposed (e.g. Taylor 1964). In these
approximate methods, a plausible shape for a drop is chosen that would satisfy the stress
balance between the electric eld and the surface tension in an approximate way. In the
present problem any likely droplet shape must be very close to a hyperboloid of
revolution (BCD in Figure 5.4), as was assumed by Yarin et al. (2001b). The tip of the
hyperboloid BCD is situated at a distance a0 from the equipotential surface z = 0, and the
range in which a solution for the electric potential is sought corresponds to 0 0 < 1
and 1 . The surface of a hyperboloid BCD corresponds to = 0 (see Figure 5.4).
Coordinate isolines are also shown in Figure 5.4, with the lines where = const.
representing ellipsoids, and the lines where = const. representing hyperboloids.
The solution of the Laplace equation for the electric potential about a hyperboloid of
revolution reads
0
n1 =1
const:
n1 0 =1 0
5:6
and the equilibrium between the electric Maxwell stresses and surface tension yields

1=2
1 0
0 a0 1=2 41=2 n
1 20
5:7
1 0
Relating the prolate spheroidal coordinates and to the spherical ones R and , one can
nd that as ! cos , the radial coordinate R ! , and Eqs. (5.6) and (5.7) yield

1 cos
1=2
jR! a0 1=2 41=2 1 20 n
const:;
5:8
1 cos
2
The latter equation shows that the asymptotic value of is nite as R ! . The electric
potential does not tend to innity as the self-similar solution (Eqs. 5.4 and 5.5) implies.
Also, in spite of the fact that R >> a0, the dependence on a0 does not disappear from Eq.
(5.8), in contrast to the self-similar behavior of Taylors solution, given by Eqs. (5.4) and
(5.5). We thus have an example of a non-self-similar solution with a nonfading inuence
of the value of a0, even when R >> a0. The details of the shape of the tip at small distances
of the order of a0 affect the solution for at any R >> a0. In other words, the solution for
the electric eld about a hyperboloid depends on the value of a0 everywhere, while the
eld surrounding a Taylor cone does not depend on a0 at R >> a0. The electric eld
surrounding the hyperboloidal bodies is always affected by the value of a0, even when R
approaches innity. This behavior is quite distinct from that of the boundary-layer theory
cases of jets from a nite orice and of plumes originating at a nite source, where the
inuence of the size of the orice or source rapidly fades away.
The following observation should be mentioned. In the case of the parabolic governing
equations (the boundary layer theory; Schlichting 1979, Zeldovich 1992, Dzhaugashtin
and Yarin 1977), or an equation with a squared parabolic operator (the beam equation
188
188 [179--261] 10.12.2013 8:33PM
describing self-similar capillary waves derived by Yarin and Weiss 1995), self-similar
solutions attract the non-self-similar ones and thus are realizable. On the other hand,
in the present case, the Laplace equation governing the electric eld is elliptic, and its
self-similar solution does not attract the non-self-similar one and therefore could hardly
be expected to be realizable. Moreover, a similar phenomenon was found in the problem
described by the biharmonic (the elliptic operator squared) equation, namely in the
problem of a wedge under a concentrated couple. The latter is known in elasticity theory
as the SternbergKoiter paradox (Sternberg and Koiter 1958).
The calculated cone, which is tangential to the critical equilibrium hyperboloid (just
before a jet is ejected), differs from a Taylor cone. Indeed, in Figure 5.6 the dependence of
0 /(a0) on 0, according to Eq. (5.7) is shown. The maximum potential at which a
stationary shape can exist corresponds to 0 0:834 and 0 4:699a0 1=2 :The
value 0 corresponds to the critical hyperboloid. The half-angle at the tip of the cone
to which the critical hyperboloid leans is 33:5o , which is signicantly smaller than
the angle for the Taylor cone T = 49.3. The critical hyperboloid is much sharper than
the one corresponding to the Taylor cone.
It should be emphasized that the left-hand branch of the curve, with a positive
slope, in Figure 5.6 can be realized pointwise, since higher electric potentials correspond
to sharper hyperboloids. By contrast, the right-hand branch represents still sharper
hyperboloids for lower electric potentials, which cannot be realized in the usual
experiments with stable uid drops. The latter means that the right-hand branch corresponds to unstable solutions.
The theoretical predictions for critical drop shapes, according to the self-similar Taylor
cone and non-self-similar solutions, were compared to the experimental data from Yarin
et al. (2001b). Two experiments, using sessile and pendant droplets, were performed. In
the sessile drop experiment (Figure 5.7a) a droplet was created at the tip of an inverted
pipette by forcing the liquid through the pipette slowly using a syringe pump. The liquid
5.0
(a0)1/2
4.0
3.0
2.0
1.0
0
0.2
0.4
0.6
0.8
1.0
0
Figure 5.6
Relation between the electric potential 0 and the hyperboloid shape parameter 0. Reprinted with
permission from Yarin et al. (2001b). Copyright 2001, AIP Publishing LLC.
189 [179--261] 10.12.2013 8:33PM
189
Lamp
High voltage
Lamp
Camera
High voltage
Camera
Pump
(a)
Figure 5.7
(b)
(a) Sessile drop experiment. (b) Pendant drop experiment. Reprinted with permission from Yarin
et al. (2001b). Copyright 2001, AIP Publishing LLC.
used was an aqueous solution of polyethylene oxide, with a molecular weight of 400 kDa
and a weight concentration of 6%.
For drop sizes of the order of 0.1 cm, the evaporation process lasts not less than 600 s.
This is much more than the time required to reach a steady state and to complete
measurements (of the order of 1 s). Therefore, evaporation effects when the photographs
were taken were negligible. All the experiments were done at room temperature. Drop
congurations were quite reproducible for a given capillary size, which was not varied in
the experiments. The electric potential was applied between the drop and a at metal
collector plate held above the droplet. The drop was kept at ground potential for
convenience. The potential difference was increased in steps of about 200 V, each step
a few seconds long, until a jet formed at the tip of the droplet. Images of the drop were
made with a video camera. The shape of the drop during the step that preceded the
formation of the jet corresponded to the critical shape. Two linear lamps were mounted
vertically, behind and on either side of the drop. The shape and diameter of the drop were
demarcated by reection of light, seen as a white line on the image recorded by the video
camera. Diffuse back lighting was used for the pendant drop (Figure 5.7b). The drops
were photographed at a rate of 30 frames per second. The observed shape of the drops
was compared to the calculated shapes in Figure 5.8a. In the pendant drop experiment
(Figure 5.7b) the polymer solution was placed in a spoon with a 1 mm hole in its bowl.
The electric potential was applied between the drop and a at plate. The experimental
result is shown in Figure 5.8b.
The predicted non-self-similar solutions for the critical hyperboloids approach the conical asymptotes with a half-angle of 33:5, shown by the solid lines in Figure 5.8.
Cones with a half-angle of T = 49.3, which represent the self-similar solution for the
Taylor cone, are shown in Figure 5.8 by dashed lines. The half-angle at the tip shown in
the photographs of Figures 5.8a and 5.8b in region C, where the inuence of the pipette is
small, is 30.5. Even closer to the tip, in region B, the observed half-angle is 37.5. Both
190
190 [179--261] 10.12.2013 8:33PM
100 m
1 mm
C
C
(a)
(b)
B
B
A
A
1 mm
2 mm
(c)
Figure 5.8
(d)
(a) Videograph of the critical drop shape observed for a sessile droplet. The bottom of the drop was
constrained to the inner diameter of the pipette on which it sat. The drop is symmetrical about the
white line. The symmetry axis is not exactly vertical due to camera tilt, the tilt of the pipette and the
tilt of the electric eld direction. The half-angles predicted in this section in the framework of nonself-similar theory are indicated by the solid lines. The half-angle associated with the self-similar
solution for the Taylor cone is indicated by the dashed lines. This image was not enhanced or
cropped. The outlines of the pipette can be seen at the bottom. (b) Part of the image in (a), processed
with Scion Image Find Edges. No useful data about the location of the edge were found in region
A. Lines tangential to the boundary segments in region B indicate a half-angle of 37.5. Lines
tangential to the boundary segments in region C indicate a half-angle of 30.5. The lower parts of
the boundary were not used because they were constrained by the pipette. (c) Critical drop shape
observed for a pendant drop. (d) Part of the image in (c) processed with Scion Image Find Edges.
Lines tangential to the boundary segments in region A indicate a half-angle of 31. Lines tangential
to the boundary segments in region B indicate a half-angle of 26. Reprinted with permission from
Yarin et al. (2001b). Copyright 2001, AIP Publishing LLC.
these angles are closer to the non-self-similar hyperboloidal solution (33.5) than to the selfsimilar solution for the Taylor cone (49.3). Calculation predicts that the hyperboloid
approaches within 5 m of the intersection of the asymptotes, but there is not enough
resolution in the images for this to be seen. Half-angles were measured, as shown in
Figure 5.8. For the sessile drop the measured half-angle near the tip, in region B, was
37.5 and in region C it was 30.5. For the pendant drop the measured half-angle near the
tip, in region A, was 31 and in region B it was 26. All these angles are closer to the
hyperboloidal solution than to the self-similar solution for the Taylor cone.
Note also that the electrode used in the experiments was submerged in the liquid inside
the pipette so the inuence of the actual electrode on the shape of the droplet was
191 [179--261] 10.12.2013 8:33PM
191
minimal. The lower part of the drop shown in Figure 5.8a is also affected mechanically
by the pipette wall, which restricts the diameter of the base of the drop, which is why the
free surface deviates from the predicted solid line in Figure 5.8a near the bottom.
Michelson (1990) stated that, according to experimental data, a stable cone can be
obtained for a range of angles, but typically the half-angle was close to 45. Both Taylor
(1964) and Michelson (1990) worked with low-molecular-weight liquids, which are
prone to perturbations and atomization. These perturbations might lead to premature
jetting before a true critical shape could be achieved. This could explain the larger (and
varying) values of recorded in their experiments. Harris and Basaran (1993)
calculated critical congurations of liquid drops affected by the electric eld in a parallel
capacitor numerically using the boundary-element method. One of the arrangements
considered, an initially hemispherical droplet supported by an electrode, was close to the
experimental situation discussed here (Yarin et al. 2001b). The numerical predictions
for this case (Figure 42 in Harris and Basaran 1993) showed that the apparent cone halfangle is about 40 or less, which is closer to the critical half-angle 33:5o predicted by
the non-self-similar theory than to T = 49.3 corresponding to the self-similar theory.
Wohlhuter and Basaran (1992) using nite-element analysis calculated steady-state
shapes of pendant/sessile drops in an electric eld. Cheng and Miksis (1989)
considered steady-state shapes of drops on a conducting plane. Their drops, however,
were considered to be polarizable dielectrics (nonconductors) with no free charges
embedded at the free surface. In a situation characteristic of electrospinning, the uid
behavior corresponds to that of ionic conductors. Therefore neither the electric context in
electrospinning, nor the drop shapes can be related to those predicted in Cheng and
Miksis (1989).
The numerically predicted value for the half-angle of the calculated drop shape, which
is signicantly less than 49.3, may be an indication of failure of the self-similarity
assumption, similar to what was discussed above. However, due to inaccuracies intrinsic
in numerical methods in cases in which a singularity is formed, a denite statement
cannot be made. According to Stone et al. (1999), who characterized both boundary- and
nite-element calculations related to the present problem, all the numerical studies
either assume a rounded end and/or cannot resolve the structure in the neighborhood
of a nearly pointed end. As usual, close to singularities, insight can be gained by
approximate models, for example, the slender body approximation (Sherwood 1991,
Li et al. 1994, Stone et al. 1999), or the hyperboloidal approximation considered here.
It should be emphasized that following Taylor (1964), most of the works assume
the liquid in the drop to be a perfect conductor. In a number of studies, however,
cases where liquid in the drop is an insulator, were also considered (Li et al. 1994,
Ramos and Castellanos 1994, Stone et al. 1999). Two self-similar conical solutions
with half-angles of 0 49:3 exist when the ratio of the dielectric constants is in the
range 17.59 d / s , where d corresponds to the drop, and s corresponds
to the surrounding uid (the ratio d / s = corresponds to a fully conductive drop).
For d / s < 17.59 equilibrium conical solutions do not exist. Deviation of the experimental half-angles toward values signicantly below 49.3 can, in principle, be attributed to one of the two solutions for the range of d / s where two solutions exist.
192
192 [179--261] 10.12.2013 8:33PM
The choice between these solutions based on stability arguments leads to a rather
puzzling outcome that the Taylor cone branch is unstable, and that very small halfangles should be taken, which is in contradiction with experiments (Li et al. 1994, Stone
et al. 1999). However, the assumption that liquids could be considered to be insulators
actually holds only on timescales shorter than the charge relaxation times, H < C. The
latter are of the order of 1010103 s according to the estimates of Ramos and
Castellanos (1994), and above in this section. Since in the experiments the residence
time of liquid elements in the cone H is of the order of 1 s and is much longer than the
charge relaxation time, conductivity effects should dominate dielectric effects (Ramos
and Castellanos 1994). In insulating dielectric liquids, due to nonzero electric shear stress
at the cone surface, ow is inevitable inside the drop (Ramos and Castellanos 1994). In
the experiments discussed above such a ow was not seen. The absence of this ow is
consistent with the fact that the behavior of polymer solutions can be closely approximated by that of a perfectly conductive liquid, as was assumed.
It is of interest to estimate the radius of curvature rc at the hyperboloid tip at the critical
potential that corresponds to the onset of jetting:
r c a0
1 20
220
5:9
With parameters corresponding to the critical hyperboloid, this expression yields

rc = 5.69 105 cm, which is near the wavelengths of visible light and is too small to be
seen in an ordinary photograph. Dimensions of polymer molecules, such as the radius of
gyration in the solution, are typically around 10 nm (106 cm), and therefore can be neglected
compared to rc.
In a group of works related to the development of pure liquid alloy ion sources (LAIS),
e.g. Driesel et al. (1996, and references therein) several additional physical processes
that may be relevant within the context of Taylor cone formation, were revealed. The
most important is eld-driven evaporation of metal ions from the tip of the cone, leading
to the emergence of ion emission currents and space charge. These phenomena are
totally irrelevant in an electrospinning context, due to the following reasons. According to
Driesel et al. (1996), eld-driven evaporation is impossible unless a jet-like protrusion is
formed on top of the Taylor cone. The characteristic radius of curvature of the protruding tip
should be of the order of 11.5 nm, and the corresponding electric eld strength of the order
of 1.5 105 kV cm1. These conditions could never be realized in electrospinning, where,
unlike LAIS, the huge electric eld strengths needed for eld-driven evaporation could
not even be approached. Moreover, the apex temperatures corresponding to eld-driven
evaporation and the accompanying effects are of the order of 6001000 C. Such temperatures would produce drastic chemical changes in any polymer solution. In addition, in
the experiments with Taylor cones by Yarin et al. (2001b), space charge and electrical
currents in air were occasionally measured. It was shown that the occurrence of these
phenomena was always a consequence of corona discharge, and could be reduced to a
very low level. Overall, one can conclude that eld-driven evaporation and ion current
effects on the half-angle of the observed cones can be totally disregarded.
193 [179--261] 10.12.2013 8:33PM
193
For low-viscosity liquids, tiny droplets can easily be emitted from the cone tip.
Sometimes droplet emission begins at close to 45 (Michelson 1990) or, sometimes,
close to 49 (Fernandez de la Mora 1992). It should be emphasized that single tiny
protrusions, jets and droplets of submicron size at the top of the Taylor cone are invisible
in ordinary photographs. It is difcult to judge when a jet emerges, since the cone tip may
oscillate as each droplet separates. At higher voltage, atomization of the cone tip can lead
to signicant space charge from the electrically charged droplets emitted. In Fernandez
de la Mora (1992) it was shown that the backward electric effect of the charged droplets
on the tip of the cone leads to reduction of its half-angle to the range 325 546. For
the highly viscoelastic liquids used in electrospinning, atomization is virtually impossible. Breakup of tiny polymer jets, threads and laments is always prevented by
viscoelastic effects and the corresponding huge elongational viscosity (Yarin 1993,
Reneker et al. 2000, Stelter et al. 2000, Yarin et al. 2001a). Therefore, it is highly
improbable that the reduced values of the half-angle found in the experiments with
polymeric liquids described above can be attributed to a space charge effect similar to that
of Fernandez de la Mora (1992).
Transient subcritical drop shapes in electric elds were also explored. The shape
evolution of small drops attached to a conducting surface and subjected to relatively
strong electric elds was studied, both experimentally and numerically in Reznik et al.
(2004) in relation to electrospinning of nanobers. Three different scenarios of drop
shape evolution were distinguished, based on numerical solution of the Stokes equations
for perfectly conducting drops: (i) in sufciently weak (subcritical) electric elds, drops
are stretched by the electric Maxwell stresses and acquire steady-state shapes where
equilibrium is achieved by means of surface tension, (ii) in stronger (supercritical)
electrical elds the Maxwell stresses overcome surface tension, and jetting is initiated
from the drop tip if the static (initial) contact angle of the drop with the conducting
electrode is s < 0.8; in this case the jet base acquires a quasi-steady, nearly conical
shape with vertical half-angle 30, which is signicantly smaller than that of the
Taylor cone (T = 49.3), (iii) in supercritical electric elds acting on drops with contact
angles in the range 0.8 < s < there is no jetting and almost the whole droplet jumps
off, similar to gravity or drop-on-demand dripping. Reznik et al. (2004) used the
boundary integral method to describe the ow eld corresponding to the axisymmetric
creeping ow inside conducting drops and the electric eld surrounding them. The
parameter representing the relative importance of electric and capillary stresses is the
electric Bond number, dened as BoE E2 =, where is the characteristic drop size
and E is the applied electric eld. Supercritical scenarios mentioned above correspond
to BoE values larger than a certain critical value BoE,cr depending on the value of the
contact angle s. In the supercritical cases, jetting from the droplet tip emerges. In
Figure 5.9 the predicted and measured shapes of a polycaprolactone (PCL) drop are
shown at different moments. In this case the numerical predictions slightly underestimate
the stretching rate, but the overall agreement is fairly good. The shift could be attributed
to the neglect of inertia in these calculations. However, that is not the case: the values of
the tip velocity uz measured in the experiments were: for curve 1 in Figure 5.9, 0 cm s1,
for curve 2, 0.058 cm s1, for curve 3, 0.110 cm s1, for curve 4, 0.142 cm s1, for curve 5,
194
194 [179--261] 10.12.2013 8:33PM
2.0
10
10
9
9
8
7 8
6
1.5
7
6
5
4 5
3 4
3
2 2
1
1.0
0.5
0
1.0
Figure 5.9
0.5
0
r
0.5
1.0
Measured and predicted shapes of a PCL drop at different time moments: (1) t = 0, (2) 101.5, (3)
201.5, (4) 351.5, (5) 501.5, (6) 601.5, (7) 651.5, (8) 701.5, (9) 731.5, (10) 756.5. Time is given in
ms. The calculation results are shown by solid lines for the right-hand side of the drop only. Their
numerals are located at their tip points (corresponding to r = 0). The experimental shapes are plotted
as dotted lines. On the left-hand side the values of r are articially made negative. Reznik et al.
(2004). Courtesy of Cambridge University Press.
0.167 cm s1, for curve 6, 0.221 cm s1, for curve 7, 0.353 cm s1, for curve 8, 0.485 cm
s1, for curve 9, 0.638 cm s1 and for curve 10, 0.941 cm s1; the corresponding values in
the calculations are quite similar. The viscosity of PCL = 212 P, the density = 1.32 g
cm3, and the droplet size = 0.1 cm. Therefore, the highest value of the Reynolds
number corresponding to Figure 5.9 is Re = us /= 5.86 104, which hardly produces
any inertial effects. Experiments on drop evolution in a high-voltage electric eld were
also conducted by Zhang and Basaran (1996). They used a low-viscosity uid (water).
The ow behavior of drops in this case was quite distinct from that of the highly viscous
uids used in electrospinning of nanobers.
The predicted drop shapes corresponding to the above-mentioned scenarios (i) and (ii)
are shown in Figure 5.10; in this case the critical value of the electric Bond number BoE,cr
is about 3.04.
It should be emphasized that the average half-angle of the cone below the jet base in
Figure 5.10b is approximately 25 to 30. Reznik et al. (2004) have not been able to nd
an approach to the Taylor cone from subcritical regimes in their dynamical numerical
simulations. The fact that the early supercritical regimes exhibit jets protruding from the
cones with = 25 to 30 favors the assumption that the critical drop congurations
(which are very difcult to achieve numerically) are close to those predicted by Yarin
et al. (2001b) with a half-vertical angle of 33.5, rather than to T = 49.3. This
195 [179--261] 10.12.2013 8:33PM
195
2.7
2.0
2.6
6
5
2.5
1.5
4
z
2.4
3
z 1.0
2
2.3
0.5
2.2
0
0
0.5
1.0
1.5
2.1
r
(a)
Figure 5.10
0.1
r
0.2
(b)
Drop evolution corresponding to the contact angle s = /2; (a) BoE = 3.03: the subcritical case,
curve (i) shows the initial drop shape at t = 0, the subsequent curves correspond to time intervals
t = 1; (b) BoE = 3.24: the jetting stage emerging in the supercritical case, (1) t = 12.001, (2) 12.012,
(3) 12.022, (4) 12.03, (5) 12.037, (6) 12.041. Time is rendered dimensionless by H = /. Reznik
et al. (2004). Courtesy of Cambridge University Press.
assumption, however, should be treated with caution, since all the examples considered
correspond to slightly supercritical dynamical cases, where the half-angles can be
smaller because of the presence of the protrusion. It should be added that Taylor (1964)
and Yarin et al. (2001b) considered innite liquid bodies: a cone or a hyperboloid of
revolution, respectively. Comparison of these two idealized models with experimental or
less-idealized numerical situations, where drops are nite and attached to a needle or a
plane surface, should be made with caution. The base parts of the drops are mechanically
affected by the needle wall, which restricts the diameter of the drop (Yarin et al. 2001b).
Such a restriction is, however, much less important for a drop attached to a plane surface,
as in Reznik et al. (2004). On the other hand, near the drop tip, any effect of mechanical
restrictions and electric stresses resulting from charge distribution in areas far from the tip
should be small, which is why both Taylor cones and hyperboloids could be compared
with experiments and numerical calculations for nite droplets.
196
196 [179--261] 10.12.2013 8:33PM
Notz and Basaran (1999) carried out a numerical analysis of drop formation from a
tube in an electric eld. The ow in the drops was treated as an inviscid potential ow. In
a subcritical electric eld when no jetting is initiated such a model predicts undamped
oscillations of the drop. Obviously, such behavior, as well as that in supercritical jetting,
is incompatible with the creeping ow case characteristic of electrospinning.
Experiments with levitated drops, also corresponding to the low-viscosity limit, revealed
thin jets issuing from drop poles and totally disintegrating over 5 s (Duft et al. 2003).
This case is also incompatible with the case of electrospinning, dominated by the very
high elongational viscosity characteristic of spinnable polymer solutions.
When the critical potential for static cone formation is exceeded and jetting begins, in
the case of polymer solutions the jets are stable to capillary perturbations, but are subject
to bending instability, which is usually observed in the electrospinning process (see
Section 5.5). On the other hand, in the case of low-viscosity liquids, or neutralization of
the electric charge (Fong et al. 1999), free liquid jets of polymer solutions are subject to
capillary instability (see Section 3.3), which sometimes leads to an almost immediate
breakup of the jet (Fernandez de la Mora 1992).
In electrospinning, charged polymeric jets possess an initial almost straight section,
where the growing bending perturbations are still very small. Therefore, it is of interest to
describe the jet prole corresponding to the almost straight section. As noted above, the
cone half-angle in the transient region, where the viscous inertialess ow transforms into
a jet, is 30. Then, for a description of the ow in the transient region and in the jet it is
natural to use the quasi-one-dimensional equations (see Sections 3.1 and 3.2), which has
been done in a number of works (Melcher and Warren 1971, Kirichenko et al. 1986, Li
et al. 1994, Ganan-Calvo 1997a,1997b, 1999, Cherney 1999a,1999b, Stone et al. 1999,
Hohman et al. 2001a, Feng 2002, 2003) with different degrees of elaboration. The
solution of these equations should also be matched to the ow in the drop region.
Cherney (1999a,b) used the method of matched asymptotic expansions to match the jet
ow with a conical semi-innite meniscus. As a basic approximation for the drop shape a
Taylor cone of T = 49.3 was chosen. This choice seems to be rather questionable in
light of nding that the Taylor cone represents a self-similar solution of the Laplace
equation to which non-self-similar solutions do not necessarily tend, even in the case of a
semi-innite meniscus, as discussed above in the present section. Moreover, even in the
situation considered, complete asymptotic matching has never been achieved.
Figures 2b, 3 and 4 in Cherney (1999a) depict discontinuities in the transition region
from the meniscus to the jet; namely, the solutions for the velocity, the potential and the
eld strength, and the free-surface conguration are all discontinuous. A similar discontinuity in the distribution of the free-surface charge density is depicted in Figure 2 in
Cherney (1999b). In that work it is mentioned that rigorous studies of the whole
transition region require signicant effort and must be a subject of separate work. The
rigorous asymptotic matching is not yet available in the literature, to the best of our
knowledge. Moreover, Higuera (2003) pointed out a formal inconsistency in Cherneys
(1999a,1999b) analysis. Approximate approaches were tested to tackle the difculty. In
particular, Ganan-Calvo (1997a,1997b, 1999), Hohman et al. (2001a) and Feng (2002,
2003) extended the quasi-one-dimensional jet equations through the whole drop up to its
197 [179--261] 10.12.2013 8:33PM
5.4 Straight part of the jet
197
attachment to the nozzle. Such an approach is quite reasonable, but only as a rst
approximation, since the equations are formally invalid in the drop region, where the
ow is fully two-dimensional. Also, in the electric part of the problem there is a need to
take into account the image effects at the solid wall, which is not always done. When it is
done, however (e.g. Hohman et al. 2001a), it does not necessarily improve the accuracy
of the results. Fortunately, Feng (2002) showed that all the electrical pre-history effects
are important only in a very thin boundary layer, adjacent to the cross-section where the
initial conditions are imposed (in his case at the nozzle exit). As a result, there is a
temptation to apply quasi-one-dimensional jet equations similar to those of Feng (2002),
but moving the jet origin to a cross-section z* > 0 in the drop (the value of z* is of the
order of the apparent height of the drop tip). Based on this idea, Reznik et al. (2004)
matched the ow in the jet region with that in the drop. By this means, they predicted the
currentvoltage characteristic I = I(U) and the volumetric ow rate Q in electrospun
viscous jets, given the potential difference applied, with I being the electric current and U
the applied voltage. The predicted dependence I = I(U) is nonlinear, due to the convective
mechanism of charge redistribution superimposed on the conductive (ohmic) one. For U
of the order of 10 kV and uid conductivity e = 104 S m1, realistic current values I of
the order of 102 nA were predicted.
Note also that two-dimensional calculations of the transition zone between a drop and
the electrically pulled jet at its tip were published by Hayati (1992), Higuera (2003) and
Yan et al. (2003).
5.4
Straight part of the jet

When jetting is initiated in a supercritical electric eld, three zones near the beginning of
the electrospun jet can be distinguished (Han et al. 2008). The modied Taylor cone
zone is nearest to the orice and is followed by the transition zone, which is in turn
followed by the straight thin segment region, the jet per se (see Figure 5.11). The shape
of the static Taylor cone is modied by the ow of the solution and the electrical charge
when the jet begins (Reznik et al. 2004; see Section 5.3). The static Taylor cone was
discussed in detail in Section 5.3. Only the ow-affected modied Taylor cone is
discussed in the present section.
In the electrically driven jets arising in electrospinning, signicant stretching of
polymer solutions begins in the transition zone, between the modied Taylor cone,
where the stress is small, and the beginning of the thin jet zone, where the liquid can
already be signicantly pre-stretched. As a result of this pre-stretching, the electrically
charged jet can possess a signicant initial stress in a cross-section of diameter 2ae
(Figure 5.11), which might affect its further evolution. The rate of strain in this strong and
extremely short pre-stretching process in the transition zone is of the order of 1001000 s1,
as revealed by the experimental data discussed below. Therefore, the viscoelastic jets can
possess signicant initial longitudinal viscoelastic stresses generated in the preceding ow
domain (the transition zone). Moreover, the high longitudinal stresses in the jet affect
its thinning, i.e. its shape. The shape changes affect the charge distribution at the jet
198
198 [179--261] 10.12.2013 8:33PM
Taylor Transition
cone
zone
2ab
Jet
2ae
1.0 mm
Figure 5.11
Modied Taylor cone zone, with the transition zone, and the beginning of the thin jet in an
electrically-driven jet of polymer solution arising in electrospinning. The cross-sectional radii at the
beginning and end of the transition zone are denoted ab and ae, respectively. The sketch of the
experimental setup is given in Figure 2.11 in Chapter 2. Reprinted from Han et al. (2008), with
surface, which has a dramatic effect on the further evolution of the jet. In electrospinning of
viscoelastic polymer solutions, the length of the initial straight part of the electried jets is
determined by the level of the longitudinal viscoelastic stresses and electric forces. In
particular, the longitudinal stresses are a strong stabilizing factor in relation to possible
electrically driven bending instability, similar to their stabilizing effect onthe aerodynamically driven instability (see Section 4.5 in Chapter 4). Moreover, until the jet becomes
sufciently thin, it possesses a signicant bending stiffness that scales as a4, with a being
the cross-sectional radius (see Section 4.5 in Chapter 4), which also promotes the existence
of a straight part in electrospun jets.
A detailed investigation of the transitional and straight zones of electrospun jets,
including their proles, thinning rates and the longitudinal viscoelastic stresses in
them, was conducted by Han et al. (2008). They measured the diameter of the jet as a
function of the position along the path, using a digital video camera attached to an optical
microscope. A polymer solution (polyethylene oxide, PEO, Mw = 400 kDa, used as a 6 wt%
aqueous solution) was held in a glass pipette that had a tip several centimeters long
with an 800-m inner diameter. A copper wire was immersed in the solution and
connected to a high-voltage power supply that generated DC voltage up to 13 kV. A
grounded horizontal plate was placed below the pipette tip to serve as a collector
electrode. The cross-sectional radius in the transition zone a = a(x) (ae a ab in
199 [179--261] 10.12.2013 8:33PM
199
Figure 5.11) was measured from the images taken (Figure 5.12). Then, the longitudinal
velocity V = V(x) was calculated from the quasi-one-dimensional continuity equation (see
Chapter 3) as V = Q/(a2), where Q is the volumetric ow rate sustained in the experiment,
neglecting solvent evaporation, at the short straight part of the jet. A laser Doppler
velocimeter was also used to measure the velocity of the thin jet following the transition
zone (where a ae) independently. The velocity proles in the transition zone and the jet
were combined and matched each other smoothly (Figure 5.12). Similar trends were
elucidated in the voltage range 35 kV.
The stretching rate in the transition zone is equal to dV/dx; values were found using the
data in Figure 5.12 and were in the range 1001000 s1. This is a tremendously high rate of
strain for 6% polymer solutions, which can hardly be achieved in other experiments.
The longitudinal stresses in the jet were measured using propagation of small bending
perturbations imposed by a lateral impactor, as described in Section 2.3. The measured
distributions of xx are depicted in Figure 5.13. The initial stresses at the beginning of the
thin jets appeared to be of the order of 10100 kPa (the rst points on the left in
Figures 5.13 and 5.14). The latter value is larger than those reported for the uncharged
jets (Gill and Gavis 1956, Goren and Gavis 1961, Bazilevskii et al. 1985) by two orders of
magnitude. This reveals the strong electrical forces affecting polymer solutions in the
transition zone prior to formation of a thin jet. The higher the applied voltage (at a xed
interelectrode distance), the higher the value of the longitudinal stress generated near the
beginning of the thin jets. Figures 5.13 and 5.14 show that relaxation effects in straight, thin
jets dominate the stretching effect of the electric stresses acting on them, which means that
the ow there is weak, following the terminology used in polymer rheology (see
Section 2.3 in Chapter 2). Therefore, the longitudinal stresses relax along the jet. This
result, in fact, invalidates many theoretical models of the straight part of viscoelastic
electrospinning jets published so far, because such models disregard the existence of
these high initial longitudinal stresses generated in the transition zone between the
Taylor cone and the beginning of the thin jet. Moreover, the results in Section 3.4
in Chapter 3 (in particular, curve 1 in Figure 3.9b) show that growth of bending perturbations is delayed by initial high longitudinal stresses in the jets. Even though this result was
rst found (Figure 3.9b) for aerodynamically driven bending instability, it can be
immediately recast for electrically driven bending instability, since the equivalence of the
two phenomena has been established (see Section 3.2). This equivalence allows the
following interpretation of the results presented in Figures 5.13 and 5.14: the electrically
driven bending instability is impossible from the tip of the modied Taylor cone up to
distances of about 2 to 4 cm along the jet, because the stabilizing longitudinal stresses are
sufciently high. This indeed happens, as was found by Reneker et al. (2000) and Yarin
et al. (2001a) and illustrated in Section 5.5. Moreover, the results depicted in Figures 5.13
and 5.14 suggest that the straight section of an electrospun jet should be longer if the jet is
formed at a higher applied voltage. Again, the predicted increase in the length of the
straight segment of the jet with increasing applied voltage is seen in the experiments by
Reneker et al. (2000) and Yarin et al. (2001a) and illustrated in Section 5.5.
Due to the mass conservation in the transition zone (where there is practically no
solvent evaporation yet), the elongation of a material element is = (ab/ae)2, where ab and
200
200 [179--261] 10.12.2013 8:33PM
(a)
2.0
2
1.5
1.0
V (m/s)
a (mm)
1
0.5
2
0.0
3
0.0
0.5
1.0
1.5
2.0
2.5
3.0 3.5
x (mm)
4.0
4.5
5.0
5.5
6.0
(b)
3
2.0
1.8
1.6
1.4
1.2
1.0
0
0.8
V (m/s)
a (mm)
0.6
0.4
0.2
0.0
3
0.2
0.0
Figure 5.12
0.5
1.0
1.5
2.0
2.5
3.0 3.5
x (mm)
4.0
4.5
5.0
5.5
6.0
Prole of the transition zone following the modied Taylor cone and the beginning of the jet. The
cross-sectional radius is shown as a = a(x) (the symmetric prole on the middle-left), where the
axial coordinate x is zero at the rst observable cross-section of the drop. At the end of the transition
zone and the beginning of the thin jet, x = xe and a = ae, respectively. The velocity distribution
measured is shown below the prole a = a(x). The change in the jet diameter after the transition
zone is practically unseen on this scale, which is the reason that the velocity distribution in panel (b)
extends beyond the prole of the transition zone. The applied voltage was 3 kV in (a) and 5 kV in
(b). Reprinted from Han et al. (2008), with permission from Elsevier.
(a)
201 [179--261] 10.12.2013 8:33PM
65000
3.0 kV
60000
4.0 kV
55000
5.0 kV
50000
45000
xx (Pa)
40000
35000
30000
25000
20000
15000
10000
5000
0
0
2
t (ms)
(b)
65000
3.0 kV
60000
4.0 kV
55000
5.0 kV
50000
45000
xx (Pa)
40000
35000
30000
25000
20000
15000
10000
5000
0
0
Figure 5.13
10
15
20
30
25
x (mm)
35
40
45
50
55
Longitudinal stress distribution over a straight unperturbed electried jet for different applied
voltages. The interelectrode distance was held at 5.5 cm. (a) Stress versus time t for an individual
material element moving along the jet, xx = xx(t). (b) The Eulerian presentation as xx = xx(x), i.e.
stress versus position x. Symbols show the experimental data for different applied voltages. The
rst points on the left correspond to the end of the transition zone and the beginning of the thin jet at
x = xe 25 mm, where x is measured from the rst observable cross-section of the drop as in
Figures 5.11 and 5.12. Reprinted from Han et al. (2008), with permission from Elsevier.
202
202 [179--261] 10.12.2013 8:33PM
ae are the cross-sectional radii at the beginning and end of the transition zone, respectively. Polymer solutions react to strong and almost instantaneous stretching almost as an
elastic neo-Hookean (rubber-like) body with practically no relaxation involved (Yarin
1993; see Section 2.4 in Chapter 2). The stress xxe at the end of the transition zone (the
initial stress in the thin jet) is then given by
4
ab
2
xxe 2G 2G
5:10
ae
where G is the modulus of elasticity, and ab and ae are measured from photographs of the
transition zone (Figure 5.11). The viscoelastic relaxation is neglected in Eq. (5.10), since
it is expected (and proved a posteriori below in this section) that the residence times of
uid elements in the transition zone tr are shorter than the relaxation time .
The initial longitudinal stress xxe should be attributed to tm tme = 0 (i.e. to x = xe) in
each experiment depicted in Figure 5.14, where tm is time reckoned for a material uid
element in the jets, which passes the end of the transition zone x = xe at the moment tm = tme.
The values of tme for these experiments are listed in the caption to Figure 5.14. It should
be emphasized that the values of tme incorporate all the pre-history of the uid elements
100000
3.0 kV
4.0 kV
xx (Pa)
5.0 kV
10000
1000
0
10
20
30
40
50
60
70
80
tm tme(ms)
Figure 5.14
The Lagrangian presentation of the stress as a function of time, xx = xx(tm tme). The time
moments at which the material element exits the transition zone and reaches the beginning of the
thin jet are denoted by tme. The values of tme were 15145, 1880 and 1183 ms for voltages of 3 kV, 4
kV and 5 kV, respectively. The distributions of stress xx versus tm tme are plotted on a semilogarithmic scale. Symbols show the experimental data for different applied voltages; lines are
plotted using Eqs. (5.11) and (5.12). Reprinted from Han et al. (2008), with permission from
Elsevier.
203 [179--261] 10.12.2013 8:33PM
203
Table 5.2 Characteristics of the 6wt% PEO solution (Mw = 400 kDa) calculated from
Eqs. (5.10), (5.11) and (5.12).
U (kV)
xxe(Pa)
G (Pa)
(ms)
(Pa s)
3.0
4.0
5.0
14431
26455
64050
53
163
395
8.3
3.0
3.0
119
80
190
(at 0 x xe) as the elements moved from the very beginning of the modied Taylor cone,
where the ow is extremely slow and the residence time is large. That is why the values of
tme (and tm) are of the order of 103104 ms. Then, tting Eq. (5.10), with ab and ae being
measured from images similar to that in Figure 5.11, reveals the corresponding values of
the modulus of elasticity G. It was found that G = 53, 163 and 395 Pa for the applied
voltages of 3.0, 4.0 and 5.0 kV, respectively. Stress relaxation in a material liquid element
moving along the jet seen in Figure 5.14 can be interpreted using an expression derived by
integration of the equations of the upper-convected Maxwell model (UCM) of Section 2.3
in Chapter 2, which yields
xx xxe exptm tme =ef f
ef f
1 2dV=dxmax ef f
5:11
5:12
where eff is the effective relaxation time, is the Rouse relaxation time corresponding to
contraction of macromolecules to their equilibrium length (Yarin 1990, 1993) and (dV /
dx)max is taken as the largest positive value of the stretching rate. Some details of the
integration resulting in Eqs. (5.11) and (5.12) are shown below, using Eqs. (5.13) and (5.14).
Fitting Eqs. (5.11) and (5.12) to the experimental data in Figure 5.14, one can nd xxe
and (shown in Table 5.2). Therefore, both characteristic rheological parameters of
polymer solutions, G and , can be established in one experiment using Eqs. (5.10)
(5.12) (see Table 5.2). The initial elongational viscosity in the jet is then estimated as
= xxe (see Table 5.2).
It should be emphasized that Eqs. (5.11) and (5.12) follow from integration of the
following rheological constitutive equations of the UCM model:

dxx
dV 1
dV
2
xx 2
5:13
dtm
dx
dx

dyy
dV 1
dV

yy
5:14
dtm
dx
dx
valid for uniaxial elongation in a jet, where xx and yy are the longitudinal and lateral
deviatoric stresses in the jet, respectively, while the longitudinal stress xx = xx yy. The
initial conditions are: t = tme, xx = xxe, yy = yye (and thus xx = xxe = xxe yye).
The integration leading to Eqs. (5.11) and (5.12) is done for dV/dx = (dV / dx)max= const.
The result shows, as expected in jets, that xx >> yy, and thus xx xx xxeexp{ [1/2
204
204 [179--261] 10.12.2013 8:33PM
(dV / dx)max](tmtme)}. The fact that the experimental data show that the stress xx
decreases when a material element moves along the jet, means that in the present case
the viscoelastic relaxation reduces the tensile stress faster than the elongation due to the
electric eld increases the stress, and thus 1/ > 2 (dV / dx)max. In this sense, the ow in the
straight thin part of the jet is weak, whereas the ow in the preceding transition zone is
strong, since it results in a build-up of longitudinal stress. It should also be noted that Eqs.
(5.11) and (5.12) account for jet stretching by the electric eld, which determines the value
of (dV / dx)max. In a more detailed analysis, the integration of Eqs. (5.13) and (5.14) could
be modied to include the fact that dV/dx could vary along the jet. This modication could
be made in the framework of a conjugate problem, incorporating the quasi-onedimensional equations of the mass and momentum balance for jets, the rheological
constitutive equations (Eqs. 5.13 and 5.14) and the electric eld description accounting
for the interaction of the surface charges on the jet with the electric eld. Examples of
numerical solutions of similar conjugate problems can be found in Feng (2003) and Carroll
and Joo (2006), although these examples do not account for the initial stresses. In the
present section we avoid dealing with the conjugate problem and detailed calculation of the
electric eld about the jet, because we use the experimental velocity data for V(x) to
evaluate dV/dx. Since the experimental data obviously reect the effects of the electric eld
via V(x), the parameter values obtained and presented in Table 5.2 are affected by the
electric eld properly, with no need for further calculation.
The relaxation times found for the 6 wt% PEO solution (Mw = 400 kDa) are in the
range = 38 ms (Table 5.2). The residence time in the transition zone is of the order of tr
(0.5 101cm) / (0.5 102 cm/s) 1 ms (see Figures 5.11 and 5.12). Therefore, the
assumption tr < does indeed hold, and the relaxation effects in the transition zone can be
neglected, as was done in Eq. (5.10).
Note that electrospun jets consist of the three distinct zones discussed above, in
addition to the zone of electrically driven bending instability discussed in Sections 5.5
and 5.6. In order along the jet these are: (i) a modied Taylor cone (a very weak, almost
Newtonian ow), (ii) a transition zone (strong elongational ow at rates of 1001000
s1), (iii) a straight thin jet (weak elongational ow at a strain rate averaged over the entire
length of the straight segment that is of the order of 20 s1, Reneker et al. 2000 and
Sections 5.5 and 5.6) and (iv) bending loops (with initially strong elongational ow and
strain rates of the order of 1000 s1, which decrease rapidly with length (Reneker et al.
2000 and Sections 5.5 and 5.6).
Xu et al. (2003) and Reneker et al. (2007) developed an optical method for measuring
cross-sectional diameters of the straight part of an electrospun jet based on diffraction of a
laser beam. The tapered shape of a typical jet is shown in the plot of jet diameter versus
distance in Figure 5.15. One can easily monitor the changes in diameter of an electrospun
jet by observing the position along the path of a distinctive color band. The length along
this path provides a visual indication of the taper rate of the jet. A higher taper rate
produces a narrower color band.
Xu et al. (2003) and Reneker et al. (2007) also applied a particle-tracing technique to
characterize the velocity in the straight part of eletrospun jets. Tracer particles were
incorporated into the polymer solution and their speeds were measured by observing
205 [179--261] 10.12.2013 8:33PM
205
Distance across the jet

8
10
12
14
16
18
20
Figure 5.15
Diameter of the straight segment of a jet as a function of position along the jet. The graph shows the
results of a series of measurements of the diffraction of a laser beam. The photograph shows the
corresponding range of interference colors. Reprinted from Reneker et al. (2007), with permission
from Elsevier.
particle movement during electrospinning, using high-speed photography at 2000 frames

per second. The largest beads were 106 m in diameter. The video images showed that the
size of the beads was close to the diameter of the electrospinning jet at its origin. Figure 5.16
shows the successive bead positions in 12 video frames during electrospinning. The
position of the particle in each frame was marked by circle. The particle speed was
calculated by dividing the distance between circles by the time elapsed between neighboring frames. Fluid velocity as a function of position along the jet axis was measured. The jet
velocities were assumed to be equal to the velocities of the beads. The jet velocities were in
the range 15 m s1. The acceleration was 590 m s2 for a voltage of 42 V mm1, 499 m s2
for 52 V mm1, 497 m s2 for 67 V mm1 and 130 ms2 for 75 V mm2. These high
acceleration values show that gravity acceleration of 9.8 m s2 can be neglected during
electrospinning (which agrees with the theoretical estimates in Reneker et al. 2000). These
measurements also showed that uid jets spun at lower voltage have higher acceleration. In
addition, it was found that when spun at lower voltages, an electrospun jet tends to have a
higher velocity and a lower diameter at the same position along the jet axis (Figure 5.17).
The corresponding rates of elongation dV/dx are shown in Figure 5.18. It can be seen that at
206
206 [179--261] 10.12.2013 8:34PM
Un
it:
m
1.
38
0.
56
0.
81
0.
71
0.
84
0.
67
0.
2000 fps
82
1.
03
1.
1 mm
22
1.
63
2.
Figure 5.16
50
Successive positions of a glass particle in a jet. Reprinted from Reneker et al. (2007), with
18
16
Jet diameter D (micron)
14
12
10
42 V mm1
y = 0.5182x2 10.417x + 56.502
2
R = 0.9865
67 V mm1
y = 0.0722x2 2.8125x + 30.372
2
R = 0.9987
6
4
52 V mm1
y = 0.1776x2 5.1364x + 39.502
2
R = 0.9967
2
0
0
10
15
20
Distance from the spinneret L (mm)

Figure 5.17
Jet diameter measured under different voltages. Reprinted from Reneker et al. (2007), with
the same positions (e.g. x < 6 mm), the strain rate has higher values when the spinning
voltage is lower. It is instructive to see that the strain rate does not increase monotonically
along the straight section of the jet. Instead, it reaches a maximum and then decreases.
However, the uid velocity does increase monotonically along the jet axis (Figure 5.12).
207 [179--261] 10.12.2013 8:34PM
207
1200
42V/mm
42 V mm1
52 V mm1
52V/mm
Strain rate (1/s1)
1000
800
67V/mm
67
V mm1
600
400
200
0
0
10
15
20

Figure 5.18
Calculated strain rate along the jet axis at different voltages. Reprinted from Reneker et al. (2007),
Since the extensional component of the velocity gradient dominates the shear component in the elongational ow in electrospun jets, they must be very effective in stretching
the polymer macromolecular chains. Polymer chains coil up in both theta and good
solvents due to the entropy contribution to the free energy. A strong elongational ow is
able to stretch polymer chains, resulting in a coilstretch transition (de Gennes 1974, 1979).
In this case, viscous forces exerted by the ow overcome the entropic elasticity of the
polymer chains. The strength of the ow in electrospinning is characterized by the elongational strain rate dV/dx. The response of the molecular chains to the ow eld depends on
the relaxation time, which characterizes entropic elasticity of macromolecular chains. The
product of the strain rate and the relaxation time determines whether stretching or relaxation will dominate in the elongation process (de Gennes 1974, see Section 2.4 in
Chapter 2). If the ow is weak and molecular chains are able to recover their original
conformation in a short time, the product of the strain rate and relaxation time will be small,
and relaxation will dominate. Thus, the coiled conformations prevail. If the ow is strong
and polymer chains need a long time to regain their original conformation, the product of
strain rate and relaxation time will be large. Molecular chains have no time to rearrange in
the ow eld. Deformation is stored and added up. In this case, a coilstretch transition
occurs and the coil quickly reaches a stretched state. Figure 5.19 shows the product of the
relaxation time of the polymer solution and the strain rates of electrospinning jets spun at
different voltages. PEO/water solutions for this experiment had an initial concentration of
6%. In Figure 5.19 the jet spun at the lowest voltage shows the highest value for the product
:
:
of strain rate dV=dx and relaxation time . Larger values of the product signify
better molecular chain stretching and alignment by the ow. It should be emphasized that a
:
coilstretch transition is expected for values of > 0.5 (de Gennes 1974). The latter shows
208
208 [179--261] 10.12.2013 8:34PM
60
Strain rate x relaxation time (s)
42 V mm1
52 V mm1
50
40
67 V mm1
30
20
10
0
0
10
15
20
25

Figure 5.19
The product of the strain rate and the relaxation time at positions along the straight section of the jet.
Reprinted from Reneker et al. (2007), with permission from Elsevier.
that the three jets of Figure 5.19 are already strong at their straight sections and should
result in stretched macromolecules.
Birefringence experiments allow one to investigate the molecular chain alignment
under different spinning voltages, as done by Xu et al. (2003) and Reneker et al. (2007).
Figure 5.20a shows birefringence on a jet spun at 44 V mm1, which is the lowest eld at
which an electrospinning jet could be maintained in this experiment. Drying jets usually
showed birefringence before the spinning stopped. In these experiments, ow rates were
very low. The thinner jets were stretched more easily by the electric eld. Figure 5.20b
shows that when the electrospinning voltage was increased, the jet birefringence became
almost undetectable. The lower spinning voltage facilitates chain alignment in the
:
electrospinning process. This result complies with the data on the rate of strain and
:
the product shown in Figures 5.18 and 5.19, respectively.
Birefringent jets observed during electrospinning always have an outer layer
brighter than the core. The following points are the possible reason. First, drying
happens at the outer edges due to evaporation and a shell structure with a higher
polymer concentration and a higher relaxation time could be expected to appear. This
shell can support most of the stress and thus result in better stretching and orientation
in the outer layer. Second, molecules on the outer surface of the jet have fewer degrees
of freedom than those in the bulk. Chains with less freedom are easier for the ow to
align. Third, like charges always stay away from each other due to repulsion forces.
Since the electrical relaxation time across the jet is rather short, essentially all of the
charge is concentrated in the surface layer and there is a higher charge density at the
outer surface than in the bulk. Thus, stronger electrical stretching forces are exerted on
the surface layer of the jet.
209 [179--261] 10.12.2013 8:34PM
5.5 Bending instability: experimental observations
44 V mm1
Electrospinning jet
6% PEO/Water
209
59 V mm1
Electrospinning jet
6% PEO/Water
26 micron
Width of the view: 360 micron

(a)
Figure 5.20
(b)
Jet birefringence observed at different electrospinning voltages. Reprinted from Reneker et al.
(2007), with permission from Elsevier.
Figure 5.21 shows the birefringence of the as-spun ber obtained under the same
conditions and from the same jet as in Figure 5.20. Although the ber diameter is much
smaller than that of the straight section of the jet, the birefringence intensity is a lot
stronger. This implies that further chain alignment took place in the bending loops
following the straight part (see Sections 5.5 and 5.6).
5.5
Electrically driven bending instability: experimental observations

Figure 5.1 depicts a sketch of the apparatus typically used in experiments with electrospinning (Reneker and Chun 1996, Reneker et al. 2000, 2007). The jet ows downwards
from the tip of a pendant drop of uid towards a collector at a distance h below the
droplet. An electrical potential difference, which is typically of the order of several
kilovolts, is established between the surface of the liquid drop and the collector. The
distance h is of the order of 10 cm, i.e. the electric eld strength is of the order of 1 kV cm1.
The collector is usually a good electric conductor. The charged nanobers may be collected
on an insulator, although a way to neutralize the charge carried by nanobers must be
provided in order to collect many layers of nanobers. Airborne ions from a corona
discharge provide an effective way to neutralize the charge on the jets and on the nanobers. Nanobers may also be collected on the surface of a liquid.
As discussed in Sections 5.3 and 5.4, at a supercritical voltage, a jet with an initial
straight section is issued from the drop tip. The jet carries away excess ions that migrated
210
210 [179--261] 10.12.2013 8:34PM
PEO
Electrospun fiber
5.7 micron
Figure 5.21
Birefringence of an electrospun ber. Reprinted from Reneker et al. (2007), with permission from
Elsevier.
10 mm
Figure 5.22
Stereographic images of an electrically driven bending instability. The exposure time was 0.25 ms.
The arrow marks the maximum lateral excursion of a loop. Reprinted with permission from
Reneker et al. (2000). Copyright 2000, AIP Publishing LLC.
to the surface when the potential was applied. Reneker et al. (2000) recorded the jet using
an electronic camera at about 2000 frames per second, with exposure times as short as
0.01250.25 ms. The light source was a 50 W halogen lamp with a faceted parabolic
reector. A Fresnel condenser lens was used to project an image of the halogen lamp and
its reector onto the region occupied by the jet. The Fresnel lens had a focal length of
19 cm and a diameter of 30 cm. The central 15 cm diameter part of the Fresnel lens was
covered so that the camera received the light scattered from the jet superimposed upon
the dark background produced by the covered part of the Fresnel lens.
Images for stereographic viewing were obtained with reduced magnication so that a
region about 1-cm wide is shown in each image in Figure 5.22. These paired images were
211 [179--261] 10.12.2013 8:34PM
211
viewed stereoscopically during playback to produce a slowed down, three-dimensional

image of the moving jet. The jet can be imagined surrounded by an envelope cone with
the vertex at the point where bending instability sets in (see Figure 5.22). The expanding
spiral in Figure 5.22 is a simple example of the kinds of jet paths that were observed.
After a short sequence of unstable bending back and forth, with growing amplitude, the
jet follows a bending, winding, spiraling and looping path in three dimensions. The jet in
each loop grows longer and thinner as the loop diameter and circumference increase.
Some jets, which are shown in Figures 5.235.25, drift downwards at a velocity much
slower than the downward velocity of the smaller loops close to the vertex of the
envelope cone. After some time, segments of a loop suddenly develop a new bending
instability, similar to, but at a smaller scale than, the rst. Each cycle of bending
instability can be described in three steps. Step 1: a smooth segment that was straight
or slightly curved suddenly develops an array of bends; Step 2: the segment of the jet in
each bend elongates and the array of bends becomes a series of spiraling loops with
growing diameters; Step 3: as the perimeter of the loops increases, the cross-sectional
diameter of the jet forming the loop grows smaller; the conditions for Step 1 are reestablished on a smaller scale, and the next cycle of bending instability begins. This cycle
was observed to repeat at an even smaller scale by Reneker et al. (2000). It was inferred
that more cycles occur, reducing the jet diameter even more and creating nanobers.
After the second cycle, the axis of a particular segment may point in any direction. The
polymer solution jet solidies as it dries and electrospun nanobers are collected some
distance below the envelope cone. The vector sum of forces from the externally applied
electric eld and the charge momentarily held in space by the jet causes the charged
segments of the jet to drift towards the collector. Except for the creation of the pendant
droplet, the electrospinning process discussed in this section depends only slightly on the
gravity force and air drag.
Figure 5.23 shows the jet entering the upper left corner, near the end of the straight
segment of a jet, and the vertex of the envelope cone, where the rst bending instability
grew. Several segments of the jet are shown, including segments from slow moving loops
that formed earlier. All these segments are connected by segments that are not seen. Two
smooth segments cross each other in this image as they run nearly horizontally across the
2 mm
Figure 5.23
Image of the end of the straight segment of the jet. The exposure time was 0.25 ms. Reprinted with
permission from Reneker et al. (2000). Copyright 2000, AIP Publishing LLC.
212
212 [179--261] 10.12.2013 8:34PM
bottom of the image. These two segments are noticeably thinner than the jet entering the
image because the jet elongated as time evolved. These slow-moving segments were a part
of large loops and were affected by the disturbance of the applied electrical eld caused by
the presence of both charged segments of the jet and charged nanobers below the region
being observed. Such slow moving segments remained in view for many frames.
Two thinner segments that formed even earlier are also seen in Figure 5.23. One runs
across the top half of the image, and the other runs across the bottom half. In the lower of
these segments, the successive bends (Step 1 of the second cycle) were apparent. In the
upper segments, the bends had already developed into spiraling loops (Step 2 of the
second cycle). The pattern of dots visible in the lower left corners of Figures 5.235.25
was caused by the pattern of facets of the reector of the halogen lamp used to illuminate
this experiment. These dots are not evidence of the familiar capillary instability that may
cause a liquid jet to become a series of droplets (see Section 3.3 in Chapter 3). No
capillary instability was observed in the experiment of Reneker et al. (2000).
Using a set of images created by an electronic camera, it was often possible to follow the
evolution of the shape of spiraling segments, such as those shown in Figure 5.23, back to
the straight segment that entered the upper left hand corner of the image. In Figures 5.24
and 5.25, the light ellipse in the rst image marks a segment that evolved in an interesting
way. The selected segment of the jet was followed forward in time, from the moment it
entered the region contained in the images until it elongated, looped, became unstable,
bent, entered the next cycle, and ultimately became too thin to form an image.
Figure 5.24 starts with a bend near the end of the straight segment of a jet entering the
image at the upper left. The onset of electrically driven bending instability occurred just
before the jet entered the image. The straight segment of the jet extended upward, and is
not shown. The segment of the jet that is highlighted by the white ellipse was followed for
27.5 ms in a series of images that were recorded at 0.5 ms intervals. The thinner segments
of the jet were emphasized by using Photopaint 6 software to reproduce them. Places
where the faint image of the jet was ambiguous are indicated by dots, seen, for example,
in the image at 22.5 ms.
Eleven images were selected from this series of 55 to show the evolution of the
highlighted segment. The time intervals between the images that are shown vary.
Many images that show only a gradual evolution of the path were omitted to simplify
Figure 5.24. The time at which the rst image was captured is taken as time zero. The
elapsed time at which each of the following images was recorded is given in Figure 5.24.
The looping segment being observed at zero time elongated for 10 ms in Figure 5.24.
Its further elongation was not followed, because the loop had extended entirely across the
image. The rate of increase in length of the highlighted segment was around 120 mm s1.
After 22 ms the visible part of the highlighted segment still appeared in Figure 5.24 as a
smooth, slightly curved line. In the short time interval between 22.0 and 22.5 ms, this
long segment suddenly became instable. A linear array of bends appeared, marking the
beginning of the second cycle. The lateral amplitude of the bends grew to about 1 mm,
and the spatial period of the bends along the segment was also about 1 mm.
These smaller bent segments of the jet continued to elongate, but the images of the
trajectories grew fainter and soon were ambiguous. The elongation and the associated
213 [179--261] 10.12.2013 8:34PM
Figure 5.24
0 ms
16 ms
2 ms
22 ms
3 ms
22.5 ms
5 ms
23 ms
8 ms
25 ms
213
27.5 ms
Evolution of an electrical bending instability. The exposure times were 0.25 ms and the width of
each image is 5 mm. Reprinted with permission from Reneker et al. (2000). Copyright 2000, AIP
Publishing LLC.
thinning presumably continued as long as the charge on the jet supplied enough repulsive
electric force. Meanwhile, the elongational viscosity increased as the jet stretched and
dried. Eventually the jet solidied and elongation stopped.
The rst image in Figure 5.25 shows a selected segment that was tracked back to
the highlighted area near the bottom of the straight segment. This loop grew in
diameter as the jet elongated and became thinner. After 18 ms, an array of bends
that had a relatively long wavelength developed. These bends evolved gradually
into the path shown at 30.5 ms. Then a tertiary array of bends developed on the
highlighted segment during the next 0.5 ms, and quickly evolved to the path shown
at 31.5 ms. The growth of the tertiary excursions was followed until 38.5 ms after
the rst image, at which point the jet was so thin that its image could no longer be
followed.
An example of an electrospun nanober mat deposited on a collector is shown in
Figure 5.26.
214
Figure 5.25
214 [179--261] 10.12.2013 8:34PM
0 ms
30.5 ms
16.5 ms
31.5 ms
18 ms
32 ms
22 ms
37.5 ms
24.5 ms
38.5 ms
Images of secondary and tertiary cycles of bending instabilities. The exposure time was 0.25 ms
and the width of each image is 5 mm. Reprinted with permission from Reneker et al. (2000).
The circled region in Figure 5.27 shows a jet that split into two that splayed apart, with
the axis of the thinner branch generally perpendicular to the axis of the primary jet (Reneker
et al. 2000). The thinner jet disappeared within a few milliseconds, in some cases because it
rapidly became even thinner, and in other cases because its path left the eld of view. No
bending instability was observed in the thinner segment, probably because it was not
observed long enough for instability to develop. Only a few such events were observed in
the thousands of images of jets of poly(ethylene oxide) (PEO) solution examined.
Before the high frame-rate, short exposure-time images in Figures 5.24 and 5.25 were
available, visual observations and video images of electrically driven PEO solution jets
were interpreted as evidence of a process that splayed the primary jet into many smaller
jets. The smaller jets were supposed to emerge from the region just below the apex of the
envelope cone. Figure 5.28a shows an image from a video frame with an exposure of 16.7
ms. The envelope cone was illuminated with a single bright halogen lamp that projected a
narrow beam through the envelope cone, toward, but not directly into, the lens, so that
215 [179--261] 10.12.2013 8:34PM
Acc.V Spot Magn Det WD

10.0 kV 3.0 10000x SE 11.8 DIM 049290
Figure 5.26
215
5 m
Electrospun nanober mat collected on a counter electrode. Reprinted with permission from
Lembach et al. (2010). Copyright 2010, American Chemical Society.
5 mm
Figure 5.27
A jet splits off the primary jet and splays in a different direction. Reprinted with permission from
Reneker et al. (2000). Copyright 2000, AIP Publishing LLC.
most of the light that entered the video camera was scattered from the jets. Figure 5.28b
shows a jet similar to that shown in Figure 5.28a that was illuminated with light from two
halogen lamps and photographed with a video camera. The two lamps were above and
behind the jet. One was to the left and the other to the right. This provided a broader
source of illumination than that used for Figure 5.28a, but not as uniform as with the
Fresnel lens. An exposure time of 1 ms was used. The part of the straight jet with a small
bending amplitude is visible, as are the loops containing segments, which had turned so
that the axis of the segment formed a high angle with the axis of the straight segment. The
parts of the jet nearer the vertex of the envelope cone appeared only as short, unconnected
lines. Spectacular reections of the beam of light, called glints, from one or the other of
the two halogen lamps off nearly horizontal segments of downward-moving loops were
shown to be the cause of these bright spots. Similar bright spots moved downwards
216
216 [179--261] 10.12.2013 8:34PM
3.0cm
cm
3.0
(a)
Figure 5.28
1.5 cm
cm
1.5
(b)
Images of an electrospinning jet with longer camera exposure times: (a) 16.7 and (b) 1.0 ms.
Reprinted with permission from Reneker et al. (2000). Copyright 2000, AIP Publishing LLC.
during the longer exposure of Figure 5.28a, and created the lines that are prominent in
this gure.
Note that the video frame rate of 30 frames per second was not fast enough to follow
the smooth development of the jet path. At this frame rate, for any particular frame, the
preceding and the following frames showed loops and spirals in completely different
positions. Only after the illumination was improved, as described in detail by Reneker
et al. (2000), and the high frame-rate electronic camera used, was it obvious that the
envelope cone was occupied by one long, owing, continuous and ever thinner PEO
solution jet. The repeated cycles of ever smaller electrically driven bending instability
created a complex path in which the directions of the axes of the connected segments
were often different and changing, sometimes by large angles.
Cross-sections of the as-spun nanobers are typically roughly circular. However,
noncircular cross-sectional shapes (presumably due to collapse of the polymer matrix
during solvent evaporation) were also revealed by the morphological analysis of
Koombhongse et al. (2001). The collapse may be so strong that electrospun ribbons
appear (cf. Yarin et al. 2007).
5.6
Electrically driven bending instability: theory

Bending instability of polymer jets in electrospinning was discovered by Reneker et al.
(2000). The term bending instability in electrospinning is reminiscent of a similar term
applied to the related aerodynamically driven bending instability discussed in Chapters 3
217 [179--261] 10.12.2013 8:34PM
5.6 Bending instability: theory
217
and 4. This term stems from the name introduced in the seminal work of Weber (1931),
who rightly recognized the basic similarity between bending jets and elastic bar bending
in classical EulerBernoulli theory (Landau and Lifshitz 1970). In some following works
on electrospinning, e.g. Hohman et al. (2001b), bending instability is called whipping
instability.
Dealing with the bending instability of electrospun jets, we consider polymer solutions
to be perfect dielectrics with frozen charges. This is justied by the fact that the bending
instability we are going to tackle is characterized by a characteristic hydrodynamic time, H
1 ms, and thus the charge relaxation time C > H (C = 3.5 ms for e = 107 S m1). Under
such conditions the same liquid that behaved as a perfect conductor in the Taylor cone (see
Section 5.3), behaves as a perfect dielectric in the bending jet issued from the Taylor cone
(these two limits encompass the dual nature of leaky dielectrics described in Section 5.2).
Then, the conductive electric current along the jet can be neglected, and charge transport
can be attributed entirely to the jet ow (the charge is frozen in the liquid).
The reason for the bending instability observed in the experiments in Section 5.5 may
be understood in the following way. In a frame of reference moving with a rectilinear
electried jet, the electrical charges can be regarded as a static system of charges
interacting according to Coulombs law (without the external eld). Such systems are
known to be unstable, according to Earnshaws theorem (Jeans 1958). To illustrate the
instability mechanism that is relevant in the context of electrospinning, we consider three
point-like charges, each with a value e and originally in a straight line at A, B and C, as
shown in Figure 5.29. Two Coulomb forces having magnitudes F = e2/r2 push against
charge B from opposite directions. If a perturbation causes point B to move off the line,
by a distance , to B, a net force F1 = 2Fcos = (2e2/r3) acts on charge B in the direction
perpendicular to the line, and tends to cause B to move further in the direction of the
perturbation, away from the line between xed charges, A and C. The growth of the small
Jet axis
F
r
l1
fl
B
F1
B
r
C
Figure 5.29
fl
F
Illustration of the Earnshaw instability, leading to bending of an electried jet. Reprinted with
permission from Reneker et al. (2000). Copyright 2000, AIP Publishing LLC.
218
218 [179--261] 10.12.2013 8:34PM
bending perturbation that is characterized by is governed in the linear approximation by

the equation
m
d2 2e2
3
dt2
1
5:15
where m is the mass.

1=2
The growing solution of this equation, 0 exp2e2 =m 31 t, shows that small
perturbations increase exponentially. The increase is sustained because the electrostatic
potential energy of the system depicted in Figure 5.29 decreases as e2/r when the
perturbations, characterized by and r, grow. This mechanism is responsible for the
observed bending instability of jets in electrospinning described in Section 5.5.
If charges, A, B and C are attached to a liquid jet, forces associated with the liquid tend
to counteract the instability caused by the Coulomb forces. For very thin liquid jets, the
inuence of the shearing force related to the bending stiffness can be neglected in
comparison with the stabilizing effect of the longitudinal forces since the shearing forces
are of the order of a4, which is much smaller than the longitudinal forces, which are of the
order of a2 (see Sections 3.1 and 3.2 in Chapter 3). The longitudinal force f at the crosssection where the bending instability sets in is determined by the end of the straight
section of the jet, as described in Section 5.4. The forces f are directed along BC or BA in
Figure 5.29, and are opposite to the local Coulomb force F. If F is larger than the
viscoelastic resistance f, the bending perturbation continues to grow, but at a rate
diminished by f.
Surface tension always counteracts the bending instability because bending always
leads to an increase in the area of the jet surface (Yarin 1993). Surface tension resists the
development of too large a curvature by the perturbation ABC in Figure 5.29, and
therefore limits the smallest possible perturbation wavelengths, albeit surface tension
effects are typically negligibly small compared to the electric and viscoelastic forces in
electrospinning.
In the dynamics of thin vortices in uids the localized-induction approximation is
widely used to describe velocity induced at a given vortex element by the rest of
the vortex line (Arms and Hama 1965, Aref and Flinchem 1984, Pozrikidis 1997, Yarin
1997, Batchelor 2002). A similar approach may be used to calculate the electric force
imposed on a given element of an electried jet by the rest of it. Consider an
enlarged element of a curved jet, as shown in Figure 5.30. We assume that the
arc length is reckoned along the jet axis from the central cross-section of the
element, where = 0. We denote the coordinates reckoned along the normal and
binormal by y and z, so that the position vector of point A on the surface of the element
ROA = yn + zb. The position vector of point B on the jet axis, close to the element
considered is thus given by
1
ROB j k0 j 2 n
2
5:16
219 [179--261] 10.12.2013 8:34PM
219
n (y)
ROB
th
xi
of
a
et
Pa
A
ROA
b (z)
2L
Figure 5.30
Sketch of an enlarged element of a curved jet and the associated normal, binormal and tangent
vectors, n, b and , respectively. Reprinted with permission from Yarin et al. (2001a). Copyright
2001, AIP Publishing LLC.
where k0 is the curvature of the jet axis at point O and is a unit tangent vector. Therefore

1
2
5:17
RBA ROA ROB y j k0 j n zb
2
Denote the cross-sectional radius of the jet element by a, assume that charge is uniformly
distributed over the jet surface with a surface density e, and denote the charge per unit
jet length by e = 2ae. Then, the Coulomb force acting at a surface element near point A
from the jet element situated near point B is given by
dFBA
ed De add
j RBA j3
RBA
5:18
where is the polar angle in the jet cross-section.

Substituting Eq. (5.17) into Eq. (5.18) and accounting for the fact that y = a cos, and
z = a sin, we obtain from Eq. (5.18)

a cos jk0 j 2 =2n a sin b

dFBA edDe add 2
5:19
a a cos j k0 j2 j k0 j 2 4 =4 2 3=2
For a thin jet, as a0 all the terms containing a in the numerator of Eq. (5.19) can be
safely neglected; also, in the denominator the term a cos jk0 j2 is negligibly small
compared to 2. Then, using Eq. (5.19) we calculate the electric force acting on a
220
220 [179--261] 10.12.2013 8:34PM
particular element of the jet, assuming that the length of the element is 2L, with L being a
cut-off for the integral, to be determined later on:
2 L
L
j k0 j2 n=2
i
Fe
d
dFAB e2 d
d h
5:20
0
L
L
a2 2 j k0 j2 4 =4 3=2
The latter yields
Fe e d
"
L=a
dx
L=a
x
a1 x2 3=2
jk0 j x2 n=2
1 x2 3=2
#
5:21
The force in the axial direction obviously cancels, whereas the total force becomes

L
2
Fe e n
j k j nd
5:22
a
This shows that the net electric force acting on a jet element is related to its curvature k =
k0, and acts in the direction of the normal to the jet axis (see Eq. 3.19 in Chapter 3).
The magnitude of the net force acting on a jet element due to the action of the surface
tension forces is equal to
F ajd aj a j k j nd
5:23
where is the surface tension coefcient.

Therefore, the net normal (lateral) force acting on a jet element is given by the sum of
the electric and surface tension forces, Eqs. (5.22) and (5.23), as

L
2
dF j k j nd a e n
5:24
a
The cut-off length L is still to be found, which will be covered below.
The electrical bending force given by Eq. (5.22) closely resembles the aerodynamic bending force per jet length d in the case of small bending perturbations of
jets rapidly moving in gas discussed in detail in Section 3.4 in Chapter 3.
According to Eq. (3.17), the aerodynamic bending force per jet length d is
given by the following expression:
Faer g V20 a20 j k j nd
5:25
where g is the gas density, V0 is the jet velocity, and a0 the jet cross-sectional radius,
which does not change for small perturbations.
This force comprises the only difference between aerodynamically driven and electrically driven bending. Comparing Eq. (5.22) (with e = e0) with Eq. (5.25), one can see that
all the results obtained in Section 3.4 in Chapter 3 for aerodynamically driven bending
may also be used here in the case of electrically driven bending, if one replaces the factor
gV20 by e20 nL=a0 =a20 :
221 [179--261] 10.12.2013 8:34PM
221
For example, in the case of an electried jet of a Newtonian liquid of viscosity , the
characteristic equation for the growth rate of small bending perturbations (Eq. 3.33) is
recast in the following form, appropriate for electrically driven bending:

3 4
e20 nL=a0 2
2

0

5:26
4 a20
a40
a30
where is the perturbation growth rate and = 2a0 / is the dimensionless wavenumber.
This equation accounts for the shearing force and moment of forces in the jet crosssection (thus, accounting for the bending stiffness). Equation (5.26) shows that the
destabilizing electric force overcomes the stabilizing effect of the surface tension if

L
e20 n
> a0
5:27
a0
This result closely resembles the threshold of the aerodynamically driven bending
instability, Eq. (3.34) in Chapter 3.
Solving Eq. (5.26), one can nd the wavenumber and the growth rate of the
fastest-growing bending perturbation:

1=6
8 a20 e20 nL=a0

5:28
9 2
a0
a20

2
2=3
e0 nL=a0 =a20
3a40
1=3
5:29
Here 2a0 = , where is the wavelength of the fastest-growing perturbation.

Equations (5.28) and (5.29) for an electrically driven bending instability closely
resemble Eqs. (3.35) and (3.36) of Chapter 3 for an aerodynamically driven bending
instability.
Compare Eq. (5.26) with the characteristic equation for electrically driven bending
perturbations of an inviscid liquid column ( = 0), derived by Taylor (1969) his Eq.
(12). Expanding this equation in the long-wave limit as the dimensionless wavenumber
! 0; we nd that it reduces to Eq. (5.26) with the term n1= instead of n(L /a0).
This fact denes the cut-off length L, since the result of Taylor (1969) is exact. Thus
taking n(L /a0) = n1= and neglecting the minor surface tension effect in Eq. (5.28),
we reduce the latter to the form

8 e20
1

1=6
5:30
n
9 2

which denes , and thus the cut-off length L. The latter appears to be very short, of the
order of a0 (Yarin et al. 2001a).
The electrically driven bending perturbations of highly viscous liquids grow much
faster than the capillary ones (driven by the surface tension), if the condition
222
222 [179--261] 10.12.2013 8:34PM
2
>> 1
e20 nL=a0
5:31
is fullled. This condition represents the recast condition (Eq. 3.37) of Chapter 3 for
aerodynamically driven bending perturbations.
For very thin jets one can neglect, in the rst approximation, the effect of the shearing
force in the jet cross-section, as well as the bending stiffness in the quasi-onedimensional equations of the dynamics of free liquid jets (Eqs. 3.13.3) in Chapter 3.
This was already done when studying bending perturbations of polymer jets in meltblowing (see Sections 4.54.7 in Chapter 4), as well as above in the present section. The
simplication stems from the fact that the shearing force and the entire moment-ofmomentum balance equation (Eq. 3.3) are of the order of a4, whereas the other terms in
the momentum balance equation (Eq. 3.2) are of the order of a2, where a is the crosssectional radius, assumed to be small. In the momentless approximation, i.e. neglecting
the bending stiffness, and using a Lagrangian parameter s frozen into the jet elements,
we reduce the quasi-one-dimensional (Eqs. 3.1 and 3.2) to the following form:
0 f 0
f 0 f 0
5:32
V
P
L
U0
j k j Pn g0 f 0 k j k j a e2 n n e k
h
t
s
a
5:33
Equation (5.32) is the continuity equation with being the geometrical stretching ratio,
so that ds = d, and f = a2 the cross-sectional area. The subscript 0 denotes the
parameter values at time t = 0. Equation (5.33) is the momentum balance equation with
being the liquid density, V its velocity, P the longitudinal force in the jet cross-section
(of viscoelastic origin in the case of electrospinning of polymer jets), gk gravity acceleration and U0/h the outer eld strength (the outer eld is assumed to be parallel to the
unit vector k, with U0 being the value of the electrical potential at the jet origin, and h the
distance between the origin and a grounded collector). It should be emphasized that on
the right-hand side of the momentum equation (Eq. 5.33) we account for the longitudinal
internal force of rheological origin acting on the jet (the rst two terms), the gravity force
(the third term), the bending electrical force and the stabilizing effect of the surface
tension (the fourth term) (see Eqs. 5.22 and 5.23), and for the electric force imposed on
the jet by the eld created by the potential difference between the jet origin and the
collector (the fth term). Note that Lagrangian parameterization of the jet axis was also
used in the theory of meltblowing in Section 4.6 in Chapter 4.
Equations (5.32) and (5.33) are supplemented by a kinematic relation similar to Eq.
(3.15) of Chapter 3. In the present case, where the Lagrangian parametrization of the jet
axis is used, it takes the form
R
V
t
where R is the position vector of a point on the jet axis.
5:34
223 [179--261] 10.12.2013 8:34PM
223
Introducing the Cartesian coordinate system associated with the exit of the needle
issuing the jet or a grounded collector, with unit vectors i, j and k, and accounting for the
expansions
R iX jY kZ
5:35
V iu jv kw
5:36
we obtain, from the projections of Eqs. (5.33) and (5.34), the following system of scalar
equations:

u
P
L
0 f 0 X j k j PnX j k j a e2 n
5:37
nX
t
s
a
0 f 0

v
P
L
Y j kj PnY j k j a e2 n
nY
t
s
a
0 f 0
w
P
Z j kj PnZ j k j
t s

L
U0
nZ g0 f 0 e
a e2 n
a
h
5:39
X
Y
Z
u;
v;
w
t
t
t
5:40
The following geometric relations should be added:

1=2
X2;s Y2;s Z2;s
2
jkj 4
5:41
1 X
1 Y
1 Z
; Y
; Z
s
s
s
5:42
1 X
1 Y
1 Z
; nY
; nZ
j k j s
j k j s
j k j s
5:43
nx
5:38
X2;s Y2;s Z2;s

2 31=2
X2;ss Y2;ss Z2;ss X;s X;ss Y;s Y;ss Z;s Z;ss
5
3
X2;s Y2;s Z2;s
5:44
Employ the simplest linear version of the upper-convected Maxwell (UCM) model of
viscoelasticity (see Section 2.3 in Chapter 2). With the present Lagrangian parametrization,
it reduces to the following equation, reminiscent of the nonlinear Eq. (4.58) in Chapter 4:
1 G
G

t
t
5:45
224
224 [179--261] 10.12.2013 8:34PM
where is the normal (longitudinal) stress in the jet cross-section. In Eq. (5.45) we
neglect purely viscous stresses compared to the elastic ones, assuming a very strong
stretching, and denote by G the modulus of elasticity and by the viscosity. Also, the rate
of stretching is found as
1 X;s u;s Y;s v;s Z;s w;s
t
2
5:46
It should be emphasized that any other reliable rheological constitutive equation could
replace Eq. (5.45) in the framework of the present model. In Section 5.7, for example, the
nonlinear upper-convected Maxwell model identical to Eq. (4.58) in Chapter 4 is used as
well, following the work of Theron et al. (2005) on multineedle electrospinning, who
showed that the difference between the predictions based on the linear and nonlinear
UCM models is not large.
The longitudinal force P is given by
P
0 f 0
5:47
Note that Eqs. (5.45) and (5.47) imply that the total stress and the corresponding
deviatoric stress are practically equal to each other, whereas the radial deviatoric stress
in the jet cross-section nn is, as usual, negligibly small (see Section 4.6 in Chapter 4 and
Section 5.4). A detailed proof of this fact can be also found in Stelter et al. (2000).
Therefore, in Eq. (5.47) P is determined by instead of nn = nn.
In addition, the equation of the charge conservation in a jet element holds:
e e0 0
5:48
The system of Eqs. (5.37)(5.48) allows one to nd the jet conguration in space at any
moment of time. To study the nonlinear evolution of the electried polymer solution jets in
electrospinning, these equations were solved numerically in Reneker et al. (2000). It should
be emphasized that direct estimates show that the aerodynamic drag force, gravity and
surface tension have negligibly small effects on electrospinning (Reneker et al. 2000). A
more general version of these equations, accounting for solvent evaporation and jet solidication was proposed by Yarin et al. (2001a). Note that Frikrikh et al. (2003) claimed that
surface tension can play a signicant role at the last stage of bending in electrospinning.
However, their model does not account for viscoelasticity, solvent evaporation and polymer
solidication (all playing dominant and increasing roles at the last stage of jet thinning due to
bending). As a result, in the model of Frikrikh et al. (2003) a disproportionately large role is
attributed to surface tension, as was noted in Thompson et al. (2007).
The system of Eqs. (5.37)(5.48) or its generalized versions, accounting in particular
for solvent evaporation, were solved numerically by Reneker et al. (2000), Yarin et al.
(2001a), Kowalewski et al. (2005) and Thompson et al. (2007). Several examples of the
numerical solutions found in these works are discussed below.
Figures 5.31ae illustrate the development of a typical jet path predicted numerically.
The jet ows continuously from a pendant drop in response to the electric eld
225 [179--261] 10.12.2013 8:34PM
225
15.2
20
15.1
Z (cm)
Z (cm)
18
15.0
16
10
2
3
(a)
X (cm
m)
10
X (c
14
14.9
2
14.8
10
(b)
Y (cm)
10
15
Y (cm)
15.2
15.2
15.0
Z (cm)
Z (cm)
14.8
14.8
20
X (cm
20
10
0
10
20
20
14.4
15
10
(c)
10
15
m)
X (c
14.4
14.6
20
14.0
(d)
Y (cm)
20
10
10
20
30
Y (cm)
16
Z (cm)
15
14
X (cm
40
20
0
20
40
40
13
(e)
Figure 5.31
30
20
10
10
20
30
Y (cm)
Numerically predicted development of bending instability. Solvent evaporation and solidication

are not accounted for. (a) t 0:19, (b) 0.39, (c) 0.59, (d) 0.79 and (e) 0.99; time is rendered
dimensionless by the viscoelastic relaxation time. Reprinted with permission from Reneker et al.
(2000). Copyright 2000, AIP Publishing LLC.
226
226 [179--261] 10.12.2013 8:34PM
established by the externally applied potential between the drop and the collector. This
electric eld causes the jet to be charged as it leaves the pendant drop. Figures 5.31ae
show how the bending instability develops along the jet. At t 0:99 in Figure 5.31e the
instantaneous path of the jet is similar to the pattern recorded in experiments using a highspeed video camera, such as those shown in Figure 5.22.
In Figures 5.31ae, a long segment near the vertex of the envelope cone is plotted at
various times and scales to show details of the jet path. The entire length of both the
straight segment and the spiral part is shown at the same scale in the inset at the upper
right of each part (ae). An ellipse in each inset encloses the part of the jet path
shown in the corresponding coordinate box. The pendant drop was always at X = 0,
Y = 0 and Z = h.
The experimental evidence shows a self-similar, fractal-like process of development of
electrically driven bending instabilities. The diameter of the rst generation of bending
loops becomes larger and the jet becomes thinner. Then much smaller bending perturbations set in on those loops and also begin to grow. This self-similar process continues at
smaller and smaller scales until viscoelastic force or solidication of the jet arrest further
bending. The numerical results in Figures 5.31ae describe only the emergence and
growth of the rst cycle of loops. This is a consequence of the fact that the Lagrangian
nodes of the jet segments increase enormously in the simulation of the development of
the rst cycle. No new nodes were added, except at the top of the rectilinear segment.
Therefore, the capability of the computer code to elucidate smaller details in the path
decreases as the jet elongates enormously.
Figure 5.31 shows that the qualitative pattern of jet behavior in electrospinning can be
predicted without accounting for evaporation and solidication. A quantitative comparison can be made, only accounting for evaporation and solidication, which was done by
Yarin et al. (2001a). Figure 5.32a shows the predicted path of a jet accounting for
evaporation and solidication, whereas Figure 5.32b was calculated without accounting
for these effects. Due to evaporation and solidication each loop of the jet becomes more
viscous with time, and its elastic modulus increases. As a result, the resistance to bending
increases and the radius of the bending loops in Figure 5.32a (with evaporation and
solidication) is smaller than that in Figure 5.32b (without evaporation and solidication). The radius of the bending perturbations of the jet calculated accounting for the
evaporation and solidication effects compares well with that found experimentally (cf.
Figure 5.22). The shape of the envelope cone can be easily seen by the naked eye, or
using a camera with a long exposure time (see Figure 5.33). The two bright lines
bifurcating in Figure 5.33 from a point highlighted by the arrow resulted from a specular
reection of light from segments near the maximum lateral excursion of each loop. Each
loop moved downward during the long exposure time of the camera and created the
bright lines seen in Figure 5.33, which dene the envelope cone of the bending jet during
the electrospinning process. For comparison with the results of the calculations, the
generatrix of the envelope cone in Figure 5.33 is also represented in Figure 5.34.
The calculations showed that evaporation and solidication have a strong effect on the
predicted shape of the envelope cone. Two theoretical curves: without evaporation and
solidication, and with these effects accounted for are presented in Figure 5.34. It can be
227 [179--261] 10.12.2013 8:34PM
227
20
18
16
14
Z (cm)
12
10
8
6
4
2
10
X (cm)
10
0
10
10
15
Y (cm)
(a)
16.0
Z (cm)
15.5
15.0
14.5
40
X (cm)
20
0
20
40
14.0
30
20
10
10
20
30
40
Y (cm)
(b)
Figure 5.32
(a) Jet path calculated accounting for evaporation and solidication. (b) Jet path calculated without
accounting for evaporation and solidication. Reprinted with permission from Yarin et al. (2001a).
clearly seen that the result accounting for evaporation and solidication agrees fairly well
with the experimental data.
The envelope visible in the experiment does not extend beyond a radius of about
3 cm, whereas the theory allowed for further growth of the radius to 10 cm. The reason
may be that after the jet had solidied in the experiment, it became much more rigid, i.e.
228
228 [179--261] 10.12.2013 8:34PM
4 cm
Figure 5.33
Shape of the envelope cone created by an electrically driven bending instability. The complicated
image in the lower part of the gure is a consequence of the long exposure time (~16 ms) used to
observe the envelope cone, and the time-varying path of the jet in that region. Reprinted with
permission from Yarin et al. (2001a). Copyright 2001, AIP Publishing LLC.
Radius of the envelope cone (cm)
10
Theory, without
evaporation
Theory, with
evaporation
Experiment
0
10
12
14
16
18
Distance from spinneret (cm)

Figure 5.34
Shape of the envelope cone: experiment versus theory. Points show the calculated radii of
successive loops. Experimental points were measured from a photograph. Reprinted with
permission from Yarin et al. (2001a). Copyright 2001, AIP Publishing LLC.
unstretchable, whereas in the theoretical calculations the solidied jet is still described
as a liquid (albeit highly viscous, with a high elastic modulus), which still allows for
some additional stretching. Actually, the comparison in Figure 5.34 shows that the
calculations should be stopped once the radius of the envelope cone has achieved a
value of 34 cm.
The theoretical results suggest that stretching of material elements along the jet makes
it possible to achieve very high draw-ratio values in the electrospinning process.
Assuming that the initial polymer concentration in the jet was 6%, the cross-sectional
229 [179--261] 10.12.2013 8:34PM
229
radius of a dry ber (af), after elongation and solvent evaporation have been completed, is
related to the initial radius of the jet (a0), by the material balance equation
a2f 13:92 a20 3:99 104 0:06
5:49
For a0 = 150 m, this yields af = 196.7 nm. The corresponding draw ratio due to
elongation is equal to (a0/af)2 0.06 = 34815. It should be emphasized that if the jet is
straight and stationary, like in the ordinary ber spinning processes (see Sections 1.2
1.4), the ratio of the ber velocity at the winding bobbin Vf to the initial one in the
spinline V0 becomes
Vf V0
a20 0:06
34815 V0
a2f
5:50
For the experimentally measured value of V0 0.1 m s1, the velocity Vf would be
Vf 3481:5 m s1 10 speed of sound!
5:51
This obviously is not true. The paradoxical value of Vf in Eq. (5.51) results from the fact
that an enormous elongation of the ber cannot be achieved at a distance of about 10 cm
along a straight line. The electrically driven bending instability supplies the mechanism
of strong elongation via fractal-like looping, which allows reduction of the nal radius af
to the range of nanobers, even though Vf 1 m s 1.
The predicted draw ratio in electrospinning versus the vertical distance of the segment
from the drop tip is shown in Figure 5.35. It is instructive to see the envelope cone too
(the dashed line in Figure 5.35), since it shows where the draw ratio grows. Along the
Draw ratio,
without evaporation
10
Draw ratio,
with evaporation
30000
Draw ratio
20000
Envelope cone
10000
2
12
40000
0
0
10
12
14
16
18

Figure 5.35
Predicted draw ratio of a segment of the jet along its length. The dotted line was generated using
results not accounting for evaporation and solidication. Reprinted with permission from Yarin
et al. (2001a). Copyright 2001, AIP Publishing LLC.
1 x 106
12
800 x 103
10
8
600 x 103
Envelope cone
6
400 x
103
4
Viscosity
200 x 103
0
0
Figure 5.36
Viscosity, (g/(cm.s))
230
230 [179--261] 10.12.2013 8:34PM
4
6
8
10
12
14
16
18
Calculated viscosity along the jet at t = 6 ms. The calculated radius of the envelope cone continued
to grow after the viscosity reached the plateau. Reprinted with permission from Yarin et al. (2001a).
straight part of the jet, which is about 6 cm long, the draw ratio achieves a value of about
1000. In the bending loops inside the envelope cone the draw ratio increases by another
factor of 25, to a value of 25 000. Without evaporation and solidication being accounted
for in the model, the draw ratio extracted from the calculation increased very rapidly, as
shown in Figure 5.35.
The predicted distribution of the viscosity along the jet at t = 6 ms is shown by the solid
line in Figure 5.36. Viscosity slowly increases along the straight part of the jet. When
bending perturbations begin to grow rapidly, the velocity of motion increases and the
evaporation process strongly intensies, which can be clearly seen when comparing the
viscosity prole with that of the envelope cone shown in Figure 5.36 by the dashed line.
Fast evaporation strongly increases the polymer fraction in the jet, which leads to
solidication manifested by the appearance of a high viscosity plateau at a distance of
about 2 cm from the beginning of the envelope cone. The calculations showed that at the
beginning of the plateau, nanobers have already been formed, since the cross-sectional
radius of the ber was already about 640 nm. A wide parametric study of the effect of
different parameters on jet tapering in electrospinning was done in the numerical
simulations of Thompson et al. (2007).
The high value of the area reduction ratio and the associated high longitudinal strain
rate imply that the macromolecules in the electrospun jets should be stretched and axially
oriented. It should be emphasized that most electrospun nanobers, even those made
from a styrenebutadienestyrene triblock co-polymer, are birefringent (Fong and
Reneker 1999).The results of the numerical simulations in Reneker et al. (2000) showed
that the longitudinal strain rate was different at different places along the jet. The
longitudinal strain rate for three different parts of the jet was determined. In particular,
in the bending loops the longitudinal strain rate is of the order of 103 s1. Polymer physics
231 [179--261] 10.12.2013 8:34PM
5.7 Multineedle and needleless electrospinning
231
suggests that strong stretching of macromolecular coils occurs when the strain rate
multiplied by the relaxation time is greater than 1/2 (Chang and Lodge 1972, de
Gennes 1974). Taking = 102 s as an estimate of the relaxation time, we nd that the
condition for the coilstretch transition is fullled in electrospinning. This does not
necessarily mean that stretched and axially aligned macromolecules are still be predominant in electrospun nanobers which typically are collected semi-wet, and thus can
relax, albeit that the experimental data in Zussman et al. (2003a) show that a portion of
the aligned macromolecules undergoes ow-induced crystallization and forms axially
aligned brillar crystals.
5.7
Branching, garlands, multineedle and needleless electrospinning

Being a dominating phenomenon, bending instability, under certain conditions, is
accompanied by a sequence of secondary jet branches emanating from the primary jet.
Yarin et al. (2005) described an experiment in which many closely spaced branches along
the jet were observed during electrospinning of a polycaprolactone (PCL) solution. The
apparatus was similar to that in Figure 5.1. Polycaprolactone, with a molecular weight of
120 kDa, was dissolved in acetone at concentrations near 15%. Polymer solutions were
electrospun from a drop hanging from a glass pipette with a tip opening in the range 300
400 m. Branching jets were produced when the electrical potential difference between
the tip and collector was in the range 315 kVand the distance between the tip and ground
was in the range 1570 mm (see Figures 5.37 and 5.38). No stable jets were produced at 2
kV, even after a jet was started by touching the drop with an insulating rod and pulling out
a charged uid segment. The electric eld strengths for these experiments ranged from 57
to 500 V mm1. Adjacent branches can lower their electrostatic interaction energy by
extending in different azimuthal directions and interactions between branches and the
charges on nearby loops of the primary jet may also affect the direction of a branch.
The jet and the branches are tapered; bending and branching may occur together. The
stereographic image of the azimuthal directions of the branches provided reliable information about the location and direction of the branches in three-dimensional space. For
several typical jets, the distance between two adjacent branches was measured as a
function of time, starting at the frame in which the two branches were rst observed
and continuing until one of the branches passed out of the eld of view. The increase in
the distance between adjacent branches was rapid at rst and became much slower after
the distance had doubled.
The branches did not occur continuously. As time progressed, the ow shown in
Figure 5.37 showed the following sequence of three types of events: (i) a straight
segment, (ii) the onset of bending instability which usually generated a garland, discussed below in more detail, and (iii) a nearly straight and relatively long segment, from
which the branches appeared and grew rapidly, that extended more than half way across
the eld of view of the segment where the branches grew. Branches grew rapidly after
small branches appeared. This sequence of three events repeated about 10 times per
second. Branches were also observed when the dried nanobers were examined
232 [179--261] 10.12.2013 8:34PM
232
Figure 5.37
Lateral jets from PCL solution. These stopped motion images of a jet were taken by a high-speed
camera at a frame rate of 2000 frames per second. Every 24th frame is shown here, so the time
separation between the frames shown is about 12 ms. The 15% PCL solution was electrospun at
5 kV, and the gap distance from pipette tip to copper plate collector was 70 mm. The width of each
frame is about 14 mm. The exposure time of each frame was 0.1 ms. Branches are usually initiated
in the straight segment and continue to elongate while the primary jet undergoes electrically driven
bending instability. The vertical gray line is due to light from the drop scattered by the camera. This
line is not part of the jet. Stereographic images show that every segment of the primary jet,
including those segments where a branch is present, moved radially outward and downward as the
segment elongated. Reprinted with permission from Yarin et al. (2005). Copyright 2005, AIP
Publishing LLC.
microscopically, but the tangled paths of the collected nanobers made it impractical to
measure the relatively long distance between the branches. Branching can be profuse,
with many long, closely spaced and rapidly growing branches. Jets with larger diameters
corresponding to higher voltage values tend to have more branches. The bending
instability and the occurrence of branching co-exist, with only minor interactions, even
when both instabilities are fully developed, as in Figure 5.38.
An innitely long static jet of incompressible liquid with a uniform circular crosssection, in a radial electric eld was the theoretical model used by Yarin et al. (2005) as a
233 [179--261] 10.12.2013 8:34PM
Figure 5.38
233
A thick jet with many closely spaced branches and a high taper rate is shown. The still images of
PCL solution were taken by a high-speed camera at a frame rate of 2000 frames per second.
Bending and branching began only a short distance from the tip. The 15% PCL solution was
electrospun with 10 kV, and the gap distance from pipette tip to the copper plate collector was
70 mm. The width of this frame is about 12 mm. The exposure time was 0.1 ms. Reprinted with
starting point to describe the mechanism leading to quasi-periodic branching of jets in

electrospinning. This approximation is related to the fact that branching takes place on a
background of bending, albeit quite independently (see Figures 5.37 and 5.38). The
electrical conductivity is supposed to be large enough to assume that excess charge is
always at the surface. The surface of the jet can respond to the presence of electrical
Maxwell forces in the following way. If any element of the charged surface moves
outward in response to the electrical forces, the motion of that element will extract energy
from the electric eld, in order to form a hill. The lateral surface area associated with
the growing hill must increase because volume is conserved, and no ow from/to the
ends can occur in an innitely long jet of incompressible liquid. The energy required to
form the undulating surface of hills and valleys is provided by the electric eld. In the
stability analysis by Yarin et al. (2005), which is related to the earlier results of Saville
(1971) and Yarin (1979), it is argued that the longest allowed static wavelength along the
jet axis leads to the observed quasi-regular spacing of the branches.
A smooth jet with a circular cross-section is the only stable shape at small electricalpotential differences. Not every undulating shape can occur in equilibrium as the
potential is increased, but some static undulations of the jet surface inevitably occur, as
shown in the study by Yarin et al. (2005). Near the highest peaks of the static undulations,
shape perturbations, which increase the radius or the curvature (see Figure 5.39), grow
rapidly and give rise to branching. The predictions of the distances between branches at
higher voltage values (7.5 and 10 kV) were commensurate with the measurements (Yarin
et al. 2005).
234
234 [179--261] 10.12.2013 8:34PM
2164
Maximal curvature (cm1)
2163.30
2163.40
2163
2161.61
2162
2160.68
2161
2160.42
2160.23
2160.42
2160
2159.05
2159.83
2159.55
2159
5
4
ns)
3
a
i
ad
(r
2
1
2161.61
2160.42
0
2163.30
2159.05
2159.83
2160.68 2160.42
2160.23
2163.40
2159.55
0.0476
0
0.0357
0.0238
Dis
tan
Figure 5.39
ce a
lon
0.005
0.0119
g je
t, z
(cm
0
0
0.0
05
y (cm)
0.005
0.
005
cm
x(
The lower panel shows a shaded perspective drawing of ve cycles of the longest wavelength mode
on the calculated surface of the jet. The maximum curvatures in the cross-sections along the jet
were also calculated and values of the highest curvatures were plotted on the unrolled surface in
the upper part of the gure. The locations of the highest curvatures of the surface are identied by
lines in the shaded drawing. Reprinted with permission from Yarin et al. (2005). Copyright 2005,
AIP Publishing LLC.
Overall, the electrohydrodynamical theory of branching proposed by Yarin et al.

(2005) showed that the surface of a conducting liquid jet can acquire complicated static
equilibrium undulations under the combined effects of electric Maxwell stresses and
surface tension, as the electrical stresses increase. A perfectly smooth cylindrical jet
develops static undulations with complex shapes in a cylindrical electric eld. It was
235 [179--261] 10.12.2013 8:34PM
235
shown that such undulating surfaces could become unstable at the sites of the highest
local surface curvature on the tops of the longest wavelength undulations. This instability
leads to the emanation of lateral branches from the primary jet at these unstable sites.
Reneker et al. (2002) reported that electrospinning of a PCL solution in acetone caused
the dramatic appearance of a uffy, columnar network of bers that moved slowly in
large loops and long curves. The name garland was given to the columnar network.
Open loops of the single jet came into contact just after the onset of bending instability
and then merged into a cross-linked network that created and maintained the garland.
Side branches can also contribute to garland formation. Contacts between loops occurred
when the plane of some of the leading loops of the jet rotated around a radius of the loop.
Then, a small following loop, expanding in a different plane, intersected a leading loop
that was as many as several turns ahead. Mechanical forces overcame the repulsive forces
from the charge carried by the jet, the open loops in ight made contact and merged at the
contact point, to form closed loops. The merged contacts were established when the
momentum of the segments and the tension in the jet forced a uid segment into contact
with another segment, in spite of repulsive Coulomb forces. Upon contact, surface
tension immediately tended to hold the jets together, while the charge tended to ow
away from the point of contact. Since the electric charge moved almost with the motion
of the liquid, the large elongation of the jet segments led to a dramatic decrease in the
charge per unit length of the jet. Segments separated by relatively large distances along
the path of the jet contacted each other during the complicated motions associated with
bending and branching. The two liquid sections in contact merged, due to the effect of
surface tension, if the reduction of the surface energy due to merging was greater than the
local increase in the energy of the electric eld.
Jets carrying higher charge (PEO) did not create garlands, while PCL jets carrying
lower charge (Theron et al. 2003, 2004) resulted in garlands under certain conditions.
The closed loops constrained the motion to form a uffy network that stretched and
became a long, roughly cylindrical column a few millimeters in diameter. This garland,
which was electrically charged, developed a path of large open loops that are characteristic of a large-scale electrically driven bending instability (Figure 5.40). Over a long
period of time, the uffy garland never traveled outside a conical envelope similar to, but
larger than, the conical envelope associated with the bending instability of a single jet.
Electrospinning of mutually interacting jets issued from multiple needles was investigated by Theron et al. (2005). Several experimental settings were used. In the rst, setup
A in Figure 5.41a, nine identical syringes, containing identical solutions, were arranged
in a 3 3 matrix. In the second, setup B in Figure 5.41b, the nine syringes were arranged
in a row. The electric repulsions between the jets led to outward deection of the jets.
Theron et al. (2005) also modeled multineedle electrospinning using the generalized
electrohydrodynamic theory outlined in Section 5.6. A number of additional new
elements introduced in the model in Theron et al. (2005) are as follows. Firstly, the
uniform capacitor electric eld was replaced with an electric eld between a sharp
electried needle and a large at ground collector. In the simplest form this eld can be
represented by a eld between a pointwise charge opposite a conducting plate (or a mirror
image of a charge of the opposite sign). Also, mutual Coulombic interactions between the
236
236 [179--261] 10.12.2013 8:34PM
56 mm
(a)
10 mm
(b)
Figure 5.40
(a) Motion of a curly garland from a 15% PCL solution is shown. The white lines demarcate
segments of the garland that advanced downward. The slope of the top line corresponds to 0.58 m s1,
and the slope of the lower lines, successively, to 0.7, 1.05, 1.36 and 2.04 m s1. Since only every fth
frame is shown, the time separation between the frames was about 20 ms. The experiment was
conducted with a voltage of 7.5 kV and a 140-mm interelectrode gap; the recording speed was 250
frames per second, with 2 ms exposure time. (b) A zoomed-in view. Reprinted from Reneker et al.
(2002) and Reneker and Yarin (2008), with permission from Elsevier.
charged jet elements were accounted for, not only for a given jet, but for all the jets in the
array. To describe the rheological behavior of polymer solutions, the nonlinear upperconvected Maxwell (UCM) model (see Section 2.3 in Chapter 2) was used in addition to
the linear Maxwell model employed in Section 5.6 (Eq. 5.45). Similarly to the experimental observations, the numerical simulations of Theron et al. (2005) demonstrated
that mutual Coulombic interactions inuence the paths of individual electried jets in
237 [179--261] 10.12.2013 8:34PM
Jet nr.
(a)
Jet nr. 1
4 cm
(b)
1 cm
(c)
Figure 5.41
(a) A photograph of a nine-jet electrospinning process where the jets were arranged in a 33 matrix
(setup A). The photograph was taken with a slow shutter speed (200 ms). An arrow indicates the
three-dimensional direction of each of the jet axes. The distance, ds, between the nozzles in the
image was 5 cm. (b) Photographs were taken at long exposure times (200 ms) of a nine-jet
electrospinning process using setup B. The distance between the individual syringes ds = 4 cm.
Front view of jets 1 to 9 with arrows to indicate the directions of the main axes of the
electrospinning envelopes. (c) A side view of all the jets in setup B. Jet 1 is the frontmost jet in the
image. Reprinted from Theron et al. (2005), with permission from Elsevier.
238
238 [179--261] 10.12.2013 8:34PM
electrospinning. Furthermore, the semi-vertical angle of the electrospinning envelope

cones of the two jets on the edges was larger than for the inner jets (2 to 8 in
Figure 5.41b). The values of the latter angles were obtained by measuring the semiangle between the two bright lines bifurcating for a specic jet envelope in Figure 5.41b,
where the measured angle for jet 5 is indicated by a double-headed arrow. The semivertical angle of the electrospinning envelope for the inner jets lies between 25 and 30,
whereas for jet 1 it is about 40. When rotating Figure 5.41b by 90, the semi-vertical
angle of the envelope cone in the direction perpendicular to the line on which the needles
are located is revealed. The latter angle, estimated from Figure 5.41c, is between 50 and
75. This conrms that the inner envelope cones are, in fact, squeezed along the line on
which the needles are located. A similar result is revealed by the simulations of Theron
et al. (2005). Although the electrospinning envelopes are squeezed, the electrically
driven bending instability of all the jets is similar to that familiar for single jets. The
results of the modeling suggest that both the nonlinear UCM model and the linear
Maxwell model provide a reasonable and quite close description of the viscoelastic
behavior of jets in electrospinning.
Reasonable process stability and uniformity of the as-spun nanober mats can be
achieved with an interneedle distance of about 1 cm and 9 nozzles on a square of about
4 cm2. This results in a jet distribution density of 2.25 jets cm2, and in a production rate
of the order of 4 ml cm2 min1.
When a single jet is issued from a single needle in a multijet setup, one has to use many
needles to achieve a high production rate (Dosunmu et al. 2006, Filatov et al. 2007). This
is technologically inconvenient due to the complexity of the system involved and the
high probability of clogging. While multiple jetting from a polymer solution droplet has
been observed (Xu and Reneker 2006, Reneker and Yarin 2008), as Figure 5.42 shows, it
has not been implemented in stable operation in either single or multineedle setups. On
Figure 5.42
Four jets issued from a single drop. Each jet developed electrically driven bending instability,
which resulted in visible loops. Reprinted from Reneker and Yarin (2008), with permission from
Elsevier.
239 [179--261] 10.12.2013 8:34PM
Figure 5.43
239
Development of jetting from a spherical layer of a viscous liquid subjected to a high electric eld.
The experiment was originally conducted by Yu. K. Stishkov. Reprinted with permission from
Miloh et al. (2009). Copyright 2009, AIP Publishing LLC.
c
e
H
a
f
b
d
Figure 5.44
Schematic drawing of the experimental setup. (a) Layer of magnetic liquid, (b) layer of polymer
solution, (c) counter electrode located at a distance H from the free surface of the polymer, (d)
electrode submerged in magnetic uid, (e) high-voltage source and (f) strong permanent magnet or
electromagnet. Reprinted from Yarin and Zussman (2004), with permission from Elsevier.
the other hand, multiple self-organized, needleless, electrically driven jetting from
planar, cylindrical or spherical sufaces subjected to super-critical electric elds is in
principle possible, and has been predicted and demonstrated (see Figure 5.43) for both
Newtonian and viscoelastic polymeric liquids (Taylor and McEwan 1965, Ostroumov
1979, Yarin and Zussman 2004, Yarin et al. 2005, Lukas et al. 2008, Miloh et al. 2009).
Such jetting typically requires a rather high electric eld strength, which can easily result
in a corona discharge in air. To overcome this difculty, Yarin and Zussman (2004)
introduced needleless electrospinning based on a two-layer system, with the lower layer
240
240 [179--261] 10.12.2013 8:34PM
being a ferromagnetic suspension and the upper layer a polymer solution, subjected to a
normal magnetic eld provided by a permanent magnet or a coil (Figure 5.44). As a
result, steady vertical spikes of magnetic suspension perturbed the interlayer interface, as
well as the free surface of the uppermost polymer layer. When a normal electric eld was
applied in addition, the perturbations of the free surface became sites of jetting directed
upward. Multiple electried jets underwent strong stretching by the electric eld and
bending instability, the solvent evaporated and solidied nanobers were deposited on
the upper counter electrode, as in an ordinary electrospinning process. The jet density in
such a needleless process was estimated as 26 jets cm2 instead of the above-mentioned
value of 2.25 jets cm2 in multineedle electrospinning. As a result, a 12-fold increase in
the production rate is expected when this method is used instead of separate needles. In
addition, the design problems related to multiple needles, as well as clogging, can be
eliminated.
Needleless electrospinning is realized in the NanospiderTM electrospinning apparatus
commercialized by Elmarco, which is shown in Figure 5.45.
5.8
Co-electrospinning and emulsion spinning of coreshell fibers

Coaxial electrospinning or co-electrospinning of coreshell micro- and nanobers
was born 2003 as it bifurcated from electrospinning. Through electrospinning,
co-electrospinning inherited roots in polymer science and electrohydrodynamics,
while some additional genes from textile science and optical ber technology
were spliced in addition. Co-electrospinning also engulfed emulsion electrospinning.
Co-electrospinning rapidly became widely popular and its applications proliferated
into such elds as biotechnology, drug delivery and nanouidics. It also triggered
signicant theoretical and experimental efforts directed at a better understanding and
control of the process. Similarly to electrospinning, the situation with co-electrospinning
is reminiscent of a babushka nested doll, with multiple interdisciplinary interactions
exposed kaleidoscopically one by one and novel opportunities emerging down the road.
Coaxial electrospinning (co-electrospinning) was rst introduced by Sun et al.
(2003). Co-electrospinning formed a novel class of nanobers with a coreshell
structure. Similarly to electrospinning, co-electrospinning employs electric forces
acting on polymer solutions in DC electric elds, resulting in signicant
stretching of polymer jets due to direct pulling and growth of electrically driven
bending perturbations. However, co-electrospinning was also inspired by such
traditional elds as melt spinning of coreshell polymer bers (Ziabicki 1976)
and formation of polarization-maintaining coreshell optical glass bers (Yarin
1995; see Section 6.8 in Chapter 6). It is worth mentioning that electrically driven
coreshell jets were also studied in relation to electrospraying (Loscertales et al.
2002), albeit the hydrodynamic issues and the products are completely different. In
the case of electrospraying, the jets are rapidly atomized into tiny coreshell
droplets, with no viscoelasticity or jet bending involved, whereas in the case of
241 [179--261] 10.12.2013 8:34PM
5.8 Co-electrospinning and emulsion spinning
241
(a)
2
1
End face
1
2
2 1
1
End face
1
Shaft axis
1
1
Shaft
1
1 Radical groove
2 Cord
(b)
(c)
Figure 5.45
NanospiderTM electrospinning apparatus. (a) Sketch of the rotary element, (b) design of the
electrodes, (c) the photograph shows multiple jets issued upwards. Petras et al. (2009).
co-electrospinning, the polymer jets stay intact to form nanobers, and viscoelasticity and jet bending are the dominant phenomena.
In co-electrospinning, a plastic syringe with two compartments containing different
polymer solutions or a polymer solution (shell) and a nonpolymeric Newtonian liquid
or even a powder (core) is used to initiate a coreshell jet (Figure 5.46). A coreshell
242
242 [179--261] 10.12.2013 8:34PM
Electrode
Rubber
band
Air pressure
inlet of the
inner chamber
Air pressure
inlet of the
outer chamber
Inner and outer
polymer solutions
Outer
chamber outlet
Figure 5.46
A double-compartment plastic syringe for co-electrospinning features separate supplies of core and
shell materials. Sun et al. (2003). Copyright Wiley-VCH Verlag GmbH & Co. KGaA. Reproduced
with permission.
Shell capillary
Core capillary
Figure 5.47
Inner chamber
outlet
PAN
solution
PMMA
solution
Flow pattern in a compound droplet attached to a coreshell needle with polyacrylonitrile, PAN,
solution (shell) and poly(methyl methacrylate), PMMA, solution dyed with malachite green (core).
The protruding core needle is visible. Zussman et al. (2006). Copyright Wiley-VCH Verlag
GmbH & Co. KGaA. Reproduced with permission.
droplet appears at the exit of the coreshell needle attached to the syringe and acquires a
shape similar to the Taylor cone due to the pulling action of electric Maxwell stresses
acting on the liquid (Figure 5.47). Liquid in the cone, subjected to sufciently strong
(supercritical) electric eld, issues a compound jet, which undergoes electrically driven
bending instability characteristic of the ordinary electrospinning process (Sun et al.
2003, Yarin et al. 2007). Strong jet stretching resulting from the bending instability
is accompanied by enormous jet thinning and fast solvent evaporation. As a result,
243 [179--261] 10.12.2013 8:34PM
243
66 nm
2.11 um
44 nm
7.32 um
50 nm
(a)
Figure 5.48
(b)
(a) TEM micrograph of a coreshell nanober. The core and shell solutions are PSU and PEO,
respectively; scale bar 50 nm. Sun et al. (2003). (b) Optical image of coreshell bers, with an outer
ber diameter of around 7 m and a core diameter of about 23 m; scale bar 50 m. Zussman et al.
(2006). Copyright Wiley-VCH Verlag GmbH & Co. KGaA. Reproduced with permission.
the jet solidies and coreshell bers are depositing on a counter electrode.
Some polymer pairs can result in coreshell nanobers (Figure 5.48a, polysulfone/
polyoxyethylene, PSU/PEO), others in microbers (Figure 5.48b, poly(methyl
methacrylate)/polyacrylonitrile, PMMA/PAN].
Co-electrospinning rapidly gained popularity and was implemented by a number of
groups (Li and Xia 2004a, 2004b), Loscertales et al. 2004, Yu et al. 2004, Zhang et al.
2004, Li et al. 2005). Two comprehensive literature reviews of co-electrospinning are
available (Yarin et al. 2007, Yarin 2011).
Conventional co-electrospinning requires a coreshell nozzle attached to a doublecompartment syringe as shown in Figure 5.46 or the supply of two polymer solutions by
means of two separate syringe pumps and pipelines leading to a coreshell needle.
Therefore, co-electrospinning setups are more complicated than for ordinary electrospinning, which involve a single needle and pump (Figure 5.1). However, coelectrospinning of coreshell polymer nanobers is also possible using an ordinary
single-needle electrospinning setup if an emulsion of two polymer solutions is used as
a working liquid (Bazilevsky et al. 2007). Solutions of PMMA in DMF (dimethylformamide) and PAN in DMF were blended and left for one day. During that time the blend
decomposed into PMMA/DMF droplets of about 100 m in diameter dispersed in a PAN/
DMF matrix (see Figure 5.49). The resulting emulsion was electrospun using an ordinary
electrospinning setup (Figure 5.50a, left). In this case the coreshell Taylor cone at the
needle exit is only transient: it appears periodically when a PMMA/DMF droplet is
entrained into the tip of a single-liquid Taylor cone created by the PAN/DMF matrix
(Figure 5.50a, right). A coreshell jet is issued from a transient compound coreshell
244
244 [179--261] 10.12.2013 8:34PM
100 m
Figure 5.49
PMMA/PAN emulsion about one day after mixing equal amounts of each polymer in DMF to
create a homogeneous blend containing 6 % wt. PMMA and 6% wt. PAN. The PMMA/DMF
droplets are dispersed in the surrounding PAN/DMF matrix. Reprinted with permission from
Bazilevsky et al. (2007). Copyright 2007, American Chemical Society.
From syringe pump
High voltage
2 m
10 m
(a)
Figure 5.50
(b)
(c)
(a) Emulsion co-electrospinning of a PMMA/PAN blend in DMF using a single needle. The inset
(right) shows magnied detail of the needle orice, which is shown in black on top. The PMMA/
DMF droplets are shown in white, whereas the PAN/DMF matrix is gray. In this case the coreshell
bers were collected on the edge of a rotating wheel serving as a grounded electrode. A plain
grounded electrode is equally possible. (b) Optical image of as-spun coreshell microber collected
on a glass slide located between the electrodes. The ber has the largest diameter produced by this
method, since ber stretching at the location of the collecting glass slide had not yet terminated. The
PMMA core is highlighted in yellow. (c) A carbonized PMMA/PAN ber is converted into a
carbon tube, namely, the PMMA core is thermally eliminated, while the PAN shell is carbonized
(see Section 7.1 in Chapter 7). Reprinted with permission from Bazilevsky et al. (2007). Copyright
2007, American Chemical Society.
Taylor cone only during the time of its existence before the PMMA/DMF droplet trapped
in its tip expires. Therefore, in principle, the as-spun bers should not possess an intact
PMMA core. However, as found by inspecting the as-spun bers (see Figure 5.50b), the
number of core disruptions is very small. The reason for that is in the very strong
stretching of material elements in the co-electrospun jet. For example, the length of a
ber section of diameter d ~ 1 m co-electrospun from one trapped PMMA/DMF droplet
of diameter D ~ 100 m should be about D3/d2 ~1 m.
245 [179--261] 10.12.2013 8:34PM
245
For the employed ow rates, it is estimated that a PMMA/DMF droplet is consumed

within a few milliseconds. This is signicantly longer than the charge relaxation time C
of such polymer solutions. Therefore, all the electric charges have enough time to escape
to the free (outer) surface of the compound Taylor cone. As a result, stretching of the tip
of the PMMA/DMF droplet currently trapped in the Taylor cone (Figure 5.50a, right) is
due to strong suction (negative pressure, traction), which is related to viscous forces
alone, as discussed below.
The as-spun bers obtained by Bazilevsky et al. (2007) had outer diameters in the
range 0.55 m and possessed a coreshell structure (Figure 5.50b), similar to that
attained via coreshell needles. Co-electrospinning of coreshell nano- and microbers
from a single needle was also implemented in several other works (Kim et al. 2005, Li
et al. 2007, Hong et al. 2008, Zhang et al. 2009). Sometimes the process is termed
emulsion electrospinning (Xu et al. 2006, Angeles et al. 2008) and viewed as an
agglomeration of droplets in the core of an emulsion jet (Xu et al. 2006). This scenario
can hardly be realized, and a realistic physical mechanism is depicted in Figure 5.50a.
Moreover, it should be emphasized that emulsion electrospinning in many cases does not
result in coreshell polymer bers but rather in bers with a dispersed phase embedded as
separate blobs. These cases are not co-electrospinning according to our denition. Only
those cases that result in coreshell bers with a very long continuous core, as recognized
by Bazilevsky et al. (2007), represent the part of emulsion electrospinning encompassed
by co-electrospinning.
A theoretical analysis of the mechanism responsible for the formation of coreshell
bers from a single needle was given by Bazilevsky et al. (2007). It resembles ows
characteristic of spherical journal bearings or hydraulic suspenders (in the latter case
the ow direction is opposite to that in the Taylor cone and the jet; Loitsyanskii 1966).
The lower part of the outer PAN/DMF volume can be approximated by a sphere. In
addition, the trapped PMMA/DMF (inner) volume is also assumed to maintain initially
a spherical shape. Thus, we are dealing with a ow inside the outer sphere (radius R)
and around a xed spherical droplet of radius R, whose center is displaced by e in the zdirection (Figure 5.51). This analysis considers the inner droplet to be xed in space.
This assumption could be relaxed without major implications for the overall trends
predicted. The gap h (between the inner sphere and the outer liquid/gas interface),
where PAN solution ows vary approximately as h() = e cos , where = R R <<
R and is the angular deection from the z-axis (see Figure 5.51). The ow of PAN
solution in this gap is axisymmetric and inertialess. The corresponding Stokes equations (Loitsyanskii 1966, Happel and Brenner 1991), simplied for a narrow gap in the
lubrication approximation, read
p
2 v r
2 0
5:52
1 p
2 v
2 0
R
5:53

246
246 [179--261] 10.12.2013 8:34PM
e
0
R
R
=0
z
Figure 5.51
h()
Schematic of the PAN/DMF ow around a spherical PMMA/DMF droplet trapped over the Taylor
cone issuing the co-annular jet, which upon drying, forms a continuous coreshell ber. The Taylor
cone and the issued jet are aligned with the z-axis. The needle orice (not shown) is just above the
top of the area shown. Reprinted with permission from Bazilevsky et al. (2007). Copyright 2007,
American Chemical Society.
vr
1 v sin
0
R sin
5:54
where = r R (r is the radial spherical coordinate centered at the drop center O), p
denotes pressure, vr and v the radial and angular velocity components, respectively, and
the viscosity of the PAN solution in this ow.
From the continuity equation (Eq. 5.54), it follows that v = O(Rvr/), i.e., v >> vr. From
the dynamic equations (Eq. 5.52 and 5.53) it follows that p==p=R~vr =v ~=R,
and thus p = p() in the gap. Therefore, integrating Eq. (5.53), we obtain
v h=Rdp=d 2 =2Rdp=d, where the following boundary conditions
were used: = 0 (the inner drop surface), v = 0 (neglecting the entrainment of PMMA/
DMF by the shear stress from the outer PAN/DMF ow; no ow inside the inner droplet);
and = h (the outer drop surface), v = 0 (stress free). Then, the average angular
velocity in the gap is
h
V 1=h v d h2 =3Rdp=d
5:55
0
Integrating the continuity equation (Eq. 5.54) over the gap, and using the boundary
conditions at = 0 and h, where vr = 0, and the expression for V, we obtain a particular
form of the Reynolds lubrication equation governing the pressure distribution in the gap
in the present case

d
dp
h3 sin
0
5:56
d
d
The integral of Eq. (5.56) yields h3 sin dp/d = C, where C is a constant. Under the
present circumstances, the ow emanating from the Taylor cone tip consists only of PAN/
DMF. Consequently, the volumetric ow rate Q supplied by the syringe pump is
channeled in its entirety through the gap h. The overall mass balance in the gap thus
247 [179--261] 10.12.2013 8:34PM
247
yields 2RhVsin = Q. Note that Q is negative, as is V, due to the coordinate system

chosen. With the above expression for V, it is Q = 2Rh sin [h2/3R] dp/d that infers
C = 3Q/2. Using the expression for C, as well as h() = e cos , we can deduce
dp
3Q
1

d
2 e cos 3 sin
5:57
We are interested only in the proximity of the Taylor cone, i.e., << 1, where cos 1
and sin . In this small angle limit, the dominant term in the pressure distribution given
by Eq. (5.57) becomes
p
3Q
23 1
3
ln
5:58
where = e/ and p < 0, which means that the surface of the inner droplet is pulled
downward (Figure 5.51). It should be emphasized that the shear stress r in the gap is
much lower than the pressure p, since r/p = O(/R) << 1. This means that the shear stress
can be safely neglected and the trapped droplet will be affected only by the pulling effect
of the negative pressure (traction).
According to Eq. (5.58), pulling on the surface of the inner droplet becomes
very strong in the vicinity of the Taylor cone, where << 1. The average pressure over
the section 0 < < * << 1 is P = 3Q/[23(1 )3](ln * 1), where, for example, for
* = 1, (ln * 1) = 5.6.
The trapped PMMA/DMF droplet may deform in response to the pulling effect of
pressure. However, the viscoelastic response of the PMMA/DMF solution opposes any
such deformation. The interfacial tension is another force that would oppose droplet
deformation. However, the interfacial tension is too weak (the same solvent on both
sides) to have an effect. In the expected strong elongational ow near the tip, the
viscoelastic response of the inner liquid should be predominantly elastic and
governed by the momentum balance 2G(L/)2 = P, where G is the elasticity modulus of
the PMMA/DMF solution and L/ is the stretching ratio at the droplet tip. The latter yields
L/ ~ [(Q)/3G]1/2. Therefore, PMMA solutions with lower G, and PAN solutions with
higher , as well as ows with higher Q, would result in easier stretching and thinning of
the PMMA core and, in turn, thinner coreshell bers (G and can be varied by
manipulating molecular weight and concentration). Introducing the viscoelastic relaxation
time as ~ /G, the stretching ratio can be written as L/ ~ [(Q)/3]1/2. Taking the
estimate Q ~1 mL h1, ~ 0.1 s and ~ 0.5 102 cm, we can deduce L/ ~ 10, which
indicates a high stretch ratio, suggesting that the PMMA/DMF solution can be pulled into
the PAN/DMF jet, thus leading to formation of a coreshell jet and subsequent formation of
coreshell bers.
It is worth mentioning that the overall force pulling the PMMA/DMF droplet along
the z-axis is Fz 3(Q)R2*2/3(1)3. This force tends to push the droplet
downward, and in the inertialess approximation is balanced by the viscous drag
(Loitsyanskii 1966) Fd 8R4U/3, where U = de/dt is the center-of-mass velocity of
the PMMA/DMF droplet relative to the PAN/DMF solution. This yields U (Q) *2/R2,
248
248 [179--261] 10.12.2013 8:34PM
and, in turn, U ~ 106105 cm s1. This low velocity suggests that the inner droplet
would barely move within the few milliseconds available for it to be fully stretched and
sucked into the Taylor cone. This provides an a posteriori verication of the initial
assumption of a xed inner droplet within the outer PAN/DMF solution ow.
Almost from the very beginning, co-electrospinning was used to form nanotubes.
Precursors appropriate for a further sol-gel transformation of nanobers at the
post-processing calcination stage (see Section 8.1 in Chapter 8) were blended in shell
solutions, which allowed their conversion into ceramic tubes, while the core was
removed chemically (Li and Xia 2004a,2004b, Li et al. (2005). These ceramic tubes,
however, were never used in nanouidic devices, to our knowledge. Co-electrospinning
of coreshell nanobers with PMMA solution in the core and PAN solution in the
shell opened the way to production of robust, long carbon nanotubes appropriate
for nanouidic applications (Zussman et al. 2006, Bazilevsky et al. 2007). At the
post-processing heat-treatment stage (see Section 7.1 in Chapter 7), the PMMA in the
core was completely eliminated, whereas the PAN in the shell was carbonized,
resulting in turbostratic amorphous carbon nanotubes (Figures 5.50c and 5.52), which
are stiff enough to sustain pressures of the order of 10 bar in the bore. Formation of
nanotubes en masse by such a relatively cheap method as co-electrospinning promises
benets compared to the traditional methods of microfabrication.
The rst efforts to demonstrate ows inside hollow polymer microtubes formed via coelectrospinning were undertaken by Dror et al. (2007), where the tubes were lled with
water, and wettability- and evaporation-driven motion of the meniscus in the bore
was observed. Pressure-driven ows inside co-electrospun and carbonized micro- and
nanotubes were also demonstrated (Bazilevsky et al. 2008, Srikar et al. 2009, Yarin 2011;
see Section 9.4 in Chapter 9).
Figure 5.52
SEM image of fractured surfaces of a single turbostratic carbon nanotube. Zussman et al. (2006).
Copyright Wiley-VCH Verlag GmbH & Co. KGaA. Reproduced with permission.
249 [179--261] 10.12.2013 8:34PM
5.9 Alignment of electrospun nanofiber mats
249
(b)
(a)
Shell (PAN)
0.95 m
Core (DCPD)
1.82 m
0.94 m
5 m
Figure 5.53
Core (DCPD)
2.13 m
Shell (PAN)
5 m
Optical images of co-electrospun coreshell PAN/DCPD nanobers. Panels (a) and (b) illustrate
different ber morphologies. The scale bar in both images is 5 m. Sinha-Ray et al. (2012).
Sinha-Ray et al. (2012) employed co-electrospinning and emulsion electrospinning

to encapsulate self-healing materials (liquid monomers) inside polymer bers (polyacrylonitrile, PAN). Note that they also demonstrated that solution co-blowing
described in Section 4.8 of Chapter 4 was equally good for this purpose. Two types
of healing materials were used (dicyclopentadiene, DCPD, and isophorone diisocyanate). The presence of self-healing materials inside polymer bers was corroborated by
using optical microscopy (see Figure 5.53), energy-dispersive X-ray spectroscopy
and preferential imprinting of uorescent dye. They also fabricated proof-of-concept
carbon-ber/epoxy composites reinforced with ultrathin nanobrous interlayers of
self-healing coreshell nanobers. When damaged, these nanobers release selfhealing agents (monomers) into the surrounding polymeric matrix, which contains
dispersed catalysts. As a result, self-healing agents polymerize and heal the damaged
composite.
5.9
Alignment of electrospun nanofiber mats

In a number of applications, in situ alignment of as-spun nanobers/nanotubes is
required. This is achieved by means of an electrostatic lens, which delivers the electried
jet to the required location, and electrostatic repulsion forces between different sections
of the jet, which stay charged for a while after being deposited onto a grounded electrode.
The rst example of this type was introduced by Theron et al. (2001), and some additional geometric arrangements based on the same principle were proposed by Deitzel
et al. (2001), Zussman et al. (2003), Sundaray et al. (2004) and Li et al. (2004).
A sketch of the experimental apparatus used by Theron et al. (2001), Zussman et al.
(2003, 2006), Bazilevsky et al. (2008) and Srikar et al. (2009) is shown in Figure 5.54. The
electried polymer jet owed downward from a needle toward a rotating vertical grounded
disk. The disk (with a diameter of 200 mm) was made of aluminum and had a tapered edge
250
250 [179--261] 10.12.2013 8:34PM
Plastic syringe
Electrospinning
envelope
Rotatable
table
Rotating
disk
collector
Axis
Figure 5.54
Schematic drawing of the electrospinning process with a sharpened disk, showing the double-cone
envelope of the jet. The disk can be equipped with a table that assists in collection of the nanobers.
The table can be rotated about the Z-axis when the disk rotation is temporarily stopped to enable
layer-by-layer collection at a desired angle between the nanober array layers. Reprinted with
permission from Zussman et al. (2003). Copyright 2003, AIP Publishing LLC.
with a half-angle of 26.6 in order to create a strongly converging electrostatic eld. The jet
owed away from the needle in a nearly straight line and then bent into a complex path that
was contained within a nearly conical region (the envelope cone). The growing loops of
bending instability represent the part of the jet where nanobers are formed due to strong
stretching. Then, due to the sharpened grounded disk (an electrostatic lens), at a certain
point above the wheel the envelope cone started to shrink, resulting in an inverted envelope
cone with its apex at the wheels edge. During the electrospinning process, the disk was
rotated at a constant speed to collect nanobers at its sharp edge. The linear speed at the tip
of the disk collector was v = 11 m s1. As the as-spun jet reached the wheels edge, it was
wound around the wheel.
A small table (5 4 mm) made of aluminum can be attached to the disk edge to
facilitate collection of nanobers and to detach them further on. The table can be rotated
about its Z-axis when the disk rotation is temporarily stopped, hence the direction of the
collected nanobers can be controlled. Nanobers were collected over a 10 s period. The
resulting two-dimensional nanober arrays are shown in Figure 5.55. Typical threedimensional nanober arrays (crossbars) collected on the table on the disk are depicted in
Figure 5.56. The collected nanobers show a high order of alignment. The diameter of
251 [179--261] 10.12.2013 8:34PM
5.9 Alignment of electrospun nanofiber mats
251
(a)
(b)
Figure 5.55
High-resolution SEM images of aligned nanobers that were collected on a carbon tape attached to
a table on the edge of a grounded rotating disk. In (a), the diameter of the nanobers varies from 100
to 300 nm. The pitch (center-to-center) varies from 1 to 2 m. In (b), the diameter of the bers
varies from 200 to 400 nm. The pitch varies from 1 to 1.5 m. Theron et al. (2001). IOP
Publishing. Reproduced with permission from IOP Publishing. All rights reserved.
the nanobers in this case was also nonuniform and varied in the range 1080 nm. When
nanobers were electrospun onto the wheels sharp edge without the table, nanoropes of
nanobers were obtained. Two HR-SEM images of a rope of nanobers are shown in
Figure 5.57. The nanobers (which are actually loops of the same jet) are in contact, and
nearly parallel for long distances. Not only monolithic, but also coreshell bers can be
collected using a sharpened rotating disk (Zussman et al. 2006). A collected nanorope
can be carbonized and the bers made hollow (see Figure 5.58). Such aligned arrays of
turbostratic carbon nanotubes were used in nanouidic experiments by Bazilevsky et al.
(2008; see Section 9.4 in Chapter 9).
252
252 [179--261] 10.12.2013 8:34PM
(a)
5 m
(b)
2 m
Figure 5.56
Typical SEM images of crossed arrays of nanobers collected on an aluminum table on top of the
disk. The structures were obtained in a sequential assembly process with orthogonal placement
directions. (a) A two-step assembly, (b) a four-step assembly. Reprinted with permission from
Zussman et al. (2003). Copyright 2003, AIP Publishing LLC.
Figure 5.57
Typical high-resolution SEM images of two ropes of aligned nanobers. The density is about 100
nanobers per m2. Reprinted from Reneker et al. (2007), with permission from Elsevier.
5.10
Electrospinning of polymer melts

Larrondo and Manley (1981a1981c) were the rst to report electrospinning of molten
polypropylene. The ber diameters were close to 50 m. Molten polymers generate
much higher viscoelastic stresses and resist much stronger stretching than polymer
solutions. Therefore, much higher electric elds would be required to electrospin them.
Also, melt electrospinning would signicantly differ from solution electrospinning, since
in the former process solidication is related to cooling, rather than solvent evaporation
and polymer precipitation, as in the latter. Too high electric elds cannot be achieved in
air, since at eld strengths above 30 kV cm1 sparks and corona discharge set in.
Therefore, Rangkupan and Reneker (2003) tried to electrospin polyethylene terephthalate in a vacuum chamber. The jets were issued from Taylor-cone-like drops and
253 [179--261] 10.12.2013 8:34PM
5.10 Electrospinning of polymer melts
(b)
(a)
5 m
100 m
15-Nov-06
WD11.9mm 10.0kv x250
200um
(c)
15-Nov-06
WD15.5mm 10.0kv
x3.5k
10um
(d)
2 m
10 m
15-Nov-06
Figure 5.58
253
WD 9.9mm 10.0kv x2.0k
WD10.3mm 10.0kv
20um
x6.0k
5um
SEM images of a bundle of carbon tubes made using co-electrospinning onto a rotating sharpened
disk, followed by carbonization. Longitudinal views at low (a) and high (b) magnication. Low (c)
and high (d) magnication cross-sectional views of the end of the bundle. The large gap shown in
the top middle of (c) is a result of handling. Some nonhollow bers seen in (d) could have been
created by intermittent co-electrospinning, or could indicate tube ends crimped during cutting of
the bundle. From Bazilevsky et al. (2008). Reproduced with permission from The Royal Society of
Chemistry.
10 m
Figure 5.59
Microber mat electrospun from molten polyethylene terephthalate in a vacuum. The ber
diameter was in the range 26 m. Reprinted from Reneker and Yarin (2008), with permission from
Elsevier.
254
254 [179--261] 10.12.2013 8:34PM
t = 0 ms.
t = 4 ms.
t = 16 ms.
t = 20 ms.
t = 32 ms.
t = 36 ms.
t = 8 ms.
t = 12 ms.
1 cm.
Figure 5.60
t = 24 ms.
t = 40 ms.
t = 28 ms.
t = 60 ms.
Electrospinning molten polycaprolactone in air. A patch of bending loops developing and moving
down the jet is framed by a square. The next patch is developing in the areas surrounded by an
ellipse. The process is intermittent, with this scenario repeating itself. Reprinted from Reneker and
Yarin (2008), with permission from Elsevier.
possessed long (more than 10 cm) straight sections; bending instability set in only at the
very end and was not well developed. As a result, only microbers were formed and no
nanobers were observed (Figure 5.59). Buckling of the jets (see Section 3.5 in
Chapter 3) as they impinged onto the chamber wall and branching were also observed.
On the other hand, it was possible to electrospin molten polycaprolactone (PCL) in
air. Long initial straight segments of the jets were observed, as well as intermittent
electrically driven bending instability (see Figure 5.60). Such jets also buckled when
impacting onto a counter electrode.
Melt electrospinning of microbers was also described by several other groups, for
example, by Lyons et al. (2004),
5.11
References
Angeles, M., Cheng, H. L., Velankar, S. S., 2008. Emulsion electrospinning: composite bers from
drop breakup during electrospinning, Polym. Adv. Technol. 19, 728733.
Aref, H., Flinchem, E. P., 1984. Dynamics of a vortex lament in a shear ow. J. Fluid Mech. 148,
477497.
255 [179--261] 10.12.2013 8:34PM
5.11 References
255
Arms, R. J., Hama, F. R., 1965. Localized-induction concept on a curved vortex and motion of an
elliptic vortex ring. Phys. Fluids 8, 533559.
Cambridge.
Baumgarten, P. K., 1971. Electrostatic spinning of acrylic microbers. J. Colloid. Interface Sci.
36, 7179.
Bazilevskii, A. V., Entov, V. M., Rozhkov, A. N., 1985. Elastic stresses in capillary jets of dilute
polymer solutions. Fluid Dynam. 20, 169175.
Bazilevsky, A. V., Yarin, A. L., Megaridis, C. M., 2007. Co-electrospinning of coreshell nano/
microbers using a single nozzle technique, Langmuir 23, 23112314.
Bazilevsky, A. V., Yarin, A. L., Megaridis, C. M., 2008. Pressure-driven delivery through carbon
tube bundles, Lab on a Chip 8, 152160.
Bockris, J. O. M., Reddy, A. K. N., 1970. Modern Electrochemistry, Volumes 1 and 2, Plenum
Press, New York.
Carroll, C. P., Joo, Y. L., 2006. Electrospinning of viscoelastic Bodger uids: modeling and
experiments. Phys. Fluids 18, 053102.
Castellanos, A., Perez, A. T., 2007. Electrohydrodynamic systems. Chapter C21 in Springer
Handook of Experimental Fluid Mechanics (Eds. C. Tropea, A. L. Yarin and J. Foss),
Springer, Berlin.
Chang, H. -C., Yeo, L. Y., 2010. Electrokinetically Driven Microuidics and Nanouidics.
Cambridge University Press, Cambridge.
Cheng, K. J., Miksis, M. J., 1989. Shape and stability of a drop on a conducting plane in an electric
eld. Physico-Chem. Hydrodyn., 11, 920.
Cherney, L. T., 1999a. Structure of Taylor cone-jets: limit of low ow rates. J. Fluid Mech. 378,
167196.
Cherney, L. T., 1999b. Electrohydrodynamics of electried liquid menisci and emitted jets.
J. Aerosol Sci. 30, 851862.
Chang, H., Lodge, A. S., 1972. Comparison of rubberlike-liquid theory with stress-growth data for
elongation of a low-density branched polyethylene melt. Rheol. Acta 11, 127129.
de Gennes, P. G., 1974. Coil-stretch transition of dilute exible polymers under ultrahigh velocity
gradients. J. Chem. Phys. 60, 50305042.
de Gennes, P. G., 1979. Scaling Concepts in Polymer Physics. Cornell University Press, Ithaca,
New York.
Deitzel, J. M., Kleinmeyer, J. D., Hirvonen, J. K., Tan, N. C. B., 2001. Controlled deposition of
electrospun poly(ethylene oxide) bers. Polymer 42, 81638170.
Doshi, J., Reneker, D. H., 1995. Electrospinning process and applications of electrospun bers.
J. Electrostatics 35, 151160.
Dosunmu, O. O., Chase, G. G., Kataphinan, W., Reneker, D. H., 2006. Electrospinning of polymer
nanobres from multiple jets on a porous tubular surface. Nanotechnology 17, 11231127.
Driesel, W., Dietzsch, Ch., Muhle, R., 1996. In situ observation of the tip shape of AuGe liquid
alloy ion sources using a high voltage transmission electrone microscope. J. Vac. Sci. Technol.
B14, 33673380.
Dror, Y., Salalha, W., Avrahami R., Zussman E., Yarin A. L., Dersch, R., Greiner, A.,
Wendorff, J. H., 2007. One-step production of polymeric micro-tubes via co-electrospinning.
Small 3, 10641073.
Duft, D., Achtzehn, T., Muller, R., Huber, B. A., Leisner, T., 2003. Rayleigh jets from levitated
microdroplets. Nature 421, 128128.
256
256 [179--261] 10.12.2013 8:34PM
Dzhaugashtin, K. E., Yarin, A. L., 1977. Numerical simulation of nonself-similar wall jet. J. Eng.
Phys. 32, 420426.
Feng, J. J., 2002. The stretching of an electried non-Newtonian jet: A model for electrospinning.
Phys. Fluids 14, 39123926.
Feng, J. J., 2003. Stretching of a straight electrically charged viscoelastic jet. J. Non-Newton. Fluid
Mech. 116, 5570.
Fernandez de la Mora, J., 1992. The effect of charge emission from electried liquid cones. J. Fluid
Mech. 243, 561574.
Feynman, R. P., Leighton, R. B., Sands, M., 2006. The Feynman Lectures on Physics. Vol. 2,
Pearson/Addison-Wesley, San Francisco.
Fundamentals and Applications in Separation and Filtration Processes. Begell House, New
York.
Fong, H., Chung, I., Reneker, D. H., 1999. Beaded nanobers formed during electrospinning.
Polymer 40, 45854592.
Fong, H., Reneker, D. H., 1999. Elastomeric nanobers of styrene-butadiene-styrene triblock
copolymer. J. Polym. Sci., Polym. Phys. Ed. 37, 34883493.
Formhals, A., 1934. Process and apparatus for preparing articial threads. US Patent No.
1975504.
Frikrikh, S. V., Yu, J. H., Brenner, M. P., Rutledge, G. C., 2003. Controlling the ber diameter
during electrospinning. Phys. Rev. Lett. 90, 144502.
Ganan-Calvo, A. M., 1997a. On the theory of electrohydrodynamically driven capillary jets. J.
Fluid Mech. 335, 165188.
Ganan-Calvo, A. M., 1997b. Cone-jet analytical extension of Taylors electrostatic solution and the
asymptotic universal scaling laws in electrospraying. Phys. Rev. Lett. 79, 217220.
Ganan-Calvo, A. M., 1999. The surface charge in electrospraying: its nature and its universal
scaling laws. J. Aerosol Sci. 30, 863872.
Gill, S. J., Gavis, J., 1956. Tensile stress in jets of viscoelastic uids. II. J. Polym. Sci. 21, 353362.
Goren, S., Gavis, J., 1961. Transverse wave motion on a thin capillary jet of a viscoelastic liquid.
Green N. G., Ramos A., Gonzales A., Morgan H., Castellanos A., 2002. Fluid ow induced by
nonuniform ac electric elds in electrolytes on microelectrodes. III. Observation of streamlines
and numerical simulation. Phys. Rev. E 66, 026305.
Harris, M. T., Basaran, O. A., 1993. Capillary electrohydrostatics of conducting drops hanging
from a nozzle in an electric eld. J. Colloid Interface Sci. 161, 389413.
Happel, J., Brenner, H., 1991. Low Reynolds Number Hydrodynamics. Kluwer, Dordrecht.
Hayati, I., 1992. Eddies inside a liquid cone stressed by interfacial electrical shear. Colloids and
Surfaces 65, 7784.
Higuera, F. J., 2003. Flow rate and electric current emitted by a Taylor cone. J. Fluid Mech. 484,
303327.
Hohman, M. M., Shin, M., Rutledge, G., Brenner, M. P., 2001a. Electrospinning and electrically
forced jets: I. Stability theory. Phys. Fluids 13, 22012220.
Hohman, M. M., Shin, M., Rutledge, G., Brenner M. P., 2001b. Electrospinning and electrically
forced jets: II. Applications. Phys. Fluids 13, 22212236.
257 [179--261] 10.12.2013 8:34PM
5.11 References
257
Hong, C. K., Yang, K. S., Oh, S. H., Ahn, J. H., Cho, B. H., Nah, C., 2008. Effect of blend
composition on the morphology development of electrospun bers based on PAN/PMMA
blends, Polymer Int. 57, 13571362.
Jeans, J., 1958. The Mathematical Theory of Electricity and Magnetism. Cambridge University
Press, Cambridge.
Kim, C. W., Frey, M. W., Marquez, M., Joo, Y. L., 2005. Preparation of submicron-scale, electrospun cellulose bers via direct dissolution. J. Polym. Sci., Part B: Polym. Phys. 43, 16731683.
Kirichenko, V. M., Petrianov-Sokolov, I. V., Suprun, N. N., Shutov, A. A., 1986. Asymptotic radius
of a slightly conducting liquid jet in an electric eld. Sov. Phys. Doklady 31, 611614.
Koombhongse, S., Liu, W., Reneker, D. H., 2001. Flat polymer ribbons and other shapes by
electrospinning. J. Polym. Sci., Polym. Phys. Ed. 39, 25982606.
Kowalewski, T. A., Blonski, S., Barral, S., 2005. Experiments and modeling of electrospinning
process. Bull. Polish Acad. Sci. 53, 385394.
Landau, L. D., Lifshitz, E. M., 1984. Electrodynamics of Continuous Media. Pergamon, Oxford.
Larrondo, L., Manley, R. S. J., 1981a. Electrostatic ber spinning from polymer melts.
I. Experimental observations on ber formation and properties. J. Polym. Sci., Polym. Phys.
Ed. 19, 909920.
Larrondo, L., Manley, R. S. J., 1981b. Electrostatic ber spinning from polymer melts. II.
Examination of the ow eld in an electrically driven jet. J. Polym. Sci., Polym. Phys. Ed. 19,
921932.
Larrondo, L., Manley, R. S. J., 1981c. Electrostatic ber spinning from polymer melts. III.
Electrostatic deformation of a pendant drop of polymer melt. J. Polym. Sci., Polym. Phys. Ed.
19, 933940.
Lembach, A., Tan, H. B., Roisman, I. V., Gambaryan-Roisman, T., Zhang, Y., Tropea, C.,
Yarin, A. L., 2010. Drop impact, spreading, splashing and penetration in electrospun nanober
mats. Langmuir 26, 95169523.
Levich, V. G., 1962. Physicochemical Hydrodynamics. Prentice Hall, Englewood Cliffs, New
Jersey.
Li, D., McCann, J. T., Xia, Y., 2005. Use of electrospinning to directly fabricate hollow nanobers
with functionalized inner and outer surfaces. Small 1, 8386.
Li, D., Wang, Y., Xia, Y., 2004. Electrospinning nanobers as uniaxially aligned arrays and layerby-layer stacked lms. Adv. Mater. 16, 361365.
Li., D., Xia, Y., 2004a. Direct fabrication of composite and ceramic hollow nanobers by electrospinning, Nano Lett. 4, 933938.
Li, D., Xia, Y., 2004b. Electrospinning of nanobers: reinventing the wheel? Adv. Mater. 16, 1151
1170.
Li, H., Halsey, T. C., Lobkovsky, A., 1994. Singular shape of a uid drop in an electric or magnetic
eld. Europhys. Lett. 27, 575580.
Li, X. H., Shao, C. L., Liu, Y. C., 2007. A simple method for controllable preparation of polymer
nanotubes via a single capillary electrospinning, Langmuir 23, 1092010923.
Moscow, 1970.
Loscertales, I. G., Barrero, A., Guerrero, I., Cortijo, R., Marquez, M., Ganan-Calvo, A. M., 2002.
Micro/nano encapsulation via electried coaxial liquid jets. Science 295, 16951698.
258
258 [179--261] 10.12.2013 8:34PM
Loscertales, I. G., Barrero, A., Marquez, M., Spretz, R., Velarde-Ortiz, R., Larsen, G. 2004.
Electrically forced coaxial nanojets for one-step hollow nanober design, J. Am. Chem. Soc.
126, 53765377.
Lukas, D., Sarkar, A., Pokorny, P., 2008. Self-organization of jets in electrospinning from free
liquid surface: A generalized approach. J. Appl. Phys. 103, 084309.
Lyons, J. Li, C. Ko, F., 2004. Melt-electrospinning part I: processing parameters and geometric
properties. Polymer 45, 75977603.
Melcher, J. R., Taylor, G. I., 1969. Electrohydrodynamics: a review of the role of interfacial shear
stresses. Annu. Rev. Fluid Mech. 1, 111146.
Melcher, J. R., Warren, E. P., 1971. Electrohydrodynamics of a current-carrying semi-insulating jet.
J. Fluid Mech. 47, 127143.
Michelson, D., 1990. Electrostatic Atomization. Adam Higler, Bristol and New York.
Miloh, T., Spivak, B., Yarin, A. L., 2009. Needleless electrospinning: electrically-driven
instability and multiple jetting from the free surface of a spherical liquid layer. J. Appl.
Phys. 106, 114910.
Notz, P. K., Basaran, O. A., 1999. Dynamics of drop formation in an electric eld. J. Colloid
Interface Sci. 213, 218237.
Ostroumov, G. A. 1979. Interaction of the Electric and Hydrodynamic Fields. Nauka, Moscow (in
Russian).
Pantano, C., Ganan-Calvo, A. M., Barrero, A., 1994. Zeroth-order, electrohydrostatic solution for
electrospraying in cone-jet mode. J. Aerosol. Sci. 25, 10651077.
Petras, D., Mares, L., Cmelik, J., Fiala, K., 2009. Device for production of nanobers through
electrostatic spinning of polymer solutions. US Patent Application 20090148547.
Pozrikidis, C., 1997. Introduction to Theoretical and Computational Fluid Dynamics. Oxford
University Press, New York.
Ramos, A., Castellanos, A., 1994. Conical points in liquid-liquid interfaces subjected to electric
elds. Phys. Lett. A184, 268272.
Rangkupan, R., Reneker, D. H., 2003. Electrospinning process of molten polypropylene in
vacuum. J. Metals, Mater. Miner. 12, 8187.
Reneker, D. H., Chun, I., 1996. Nanometer diameter bers of polymer, produced by electrospinning. Nanotechnology 7, 216223.
Reneker, D. H., Kataphinan, W., Theron, A., Zussman, E., Yarin, A. L., 2002. Nanober garlands
of polycaprolactone by electrospinning. Polymer 43, 67856794.
23872425.
Reneker, D. H., Yarin, A. L., Fong, H., Koombhongse, S., 2000. Bending instability of
electrically charged liquid jets of polymer solutions in electospinning. J. Appl. Phys. 87,
45314547.
Reznik, S. N., Yarin, A. L., Theron, A., Zussman, E., 2004. Transient and steady shapes of droplets
attached to a surface in a strong electric eld. J. Fluid Mech. 516, 349377.
Reznik, S. N., Yarin, A. L., Zussman, E., Bercovici, L., 2006. Evolution of a compound droplet
attached to a coreshell nozzle under the action of a strong electric eld. Phys. Fluids
18, 062101.
Russel, W. B., Saville, D. A., Schowalter, W. R., 1989. Colloidal Dispersions. Cambridge
University Press, Cambridge.
259 [179--261] 10.12.2013 8:34PM
5.11 References
259
Saville, D. A., 1971. Stability of electrically charged viscous cylinders. Phys. Fluids 14, 1095
1099.
Saville, D. A., 1997. Electrohydrodynamics: The Taylor-Melcher leaky dielectric model. Annu.
Rev. Fluid Mech. 29, 2764.
Schlichting, H., 1979. Boundary Layer Theory, McGraw-Hill, New York.
Sherwood, J. D., 1991. The deformation of a uid drop in an electric eld: a slender-body analysis.
J. Phys. A24, 40474053.
Shrimpton, J., 2009. Charge Injection Systems: Physical Principles, Experimental and Theoretical
Work. Springer, Berlin.
Sinha-Ray, S., Pelot, D. D., Zhou, Z. P., Rahman, A., Wu, X. -F., Yarin, A. L., 2012. Encapsulation
of self-healing materials by coelectrospinning, emulsion electrospinning and solution blowing
and intercalation. J. Mater. Chem. 22, 91389146.
Smythe, W. R., 1989. Static and Dynamic Electricity. McGraw-Hill, New York.
Srikar, R., Yarin, A. L., Megaridis, C. M., 2009. Fluidic delivery of homogeneous solutions through
carbon tube bundles, Nanotechnology 20, 275706.
Stelter, M., Brenn, G., Yarin, A. L., Singh, R. P., Durst, F., 2000. Validation and application of a
novel elongational device for polymer solutions. J. Rheol. 44, 595616.
Sternberg, E., Koiter, W. T., 1958. The wedge under a concentrated couple: a paradox in the twodimensional theory of elasticity. J. Appl. Mech. 25, 575581.
Stone, H. A., Lister, J. R., Brenner, M. P., 1999. Drops with conical ends in electric and magnetic
elds. Proc. R. Soc. Lond. A455, 329347.
Sun, Z., Zussman, E., Yarin, A. L., Wendorff, J. H., Greiner, A., 2003. Compound core/shell
polymer nanobers by co-electrospinning. Adv. Mater. 15, 19291932.
Sundaray, B., Subramanian, V., Natarajan, T. S., Xiang, R. Z., Chang, C. C., Fann, W. S.
2004. Electrospinning of continuous aligned polymer bers. Appl. Phys. Lett. 84,
12221224.
Taylor, G. I., 1964. Disintegration of water drops in an electric eld. Proc. Roy. Soc. London
A280, 383397.
Taylor, G. I., 1969. Electrically driven jets. Proc. Roy. Soc. London A 313, 453475.
Taylor, G. I., McEwan, A. D., 1965. The stability of a horizontal uid interface in a vertical electric
eld. J. Fluid Mech. 22, 115.
Theron, S. A., Yarin, A. L., Zussman, E., Kroll, E., 2005. Multiple jets in electrospinning:
experiment and modeling. Polymer 46, 28892899.
Theron, A., Zussman, E., Yarin, A. L., 2001. Electrostatic eld-assisted alignment of electrospun
nanobers. Nanotechnology 12, 384390.
Theron, A., Zussman, E., Yarin, A. L., 2003. Measurements of the governing parameters in the
electrospinning of polymer solutions. Polymer Preprints 44, 6162.
Theron, S. A., Zussman, E., Yarin, A. L., 2004. Experimental investigation of the governing
parameters in the electrospinning of polymer solutions. Polymer 45, 20172030.
Thompson, C. J., Chase, G. G., Yarin, A. L., Reneker, D. H., 2007. Effect of parameters on
nanober diameter determined from electrospinning model. Polymer 48, 69136922.
Wohlhuter, F. K., Basaran, O. A., 1992. Shapes and stability of pendant and sessile dielectric drops
in an electric eld. J. Fluid Mech. 235, 481510.
Xu, H., Reneker, D. H., 2006. Chapter 3 in Polymer Nanobers, ACS Symposium Series 918 (Ed.
D. H. Reneker and H. Fong), American Chemical Society, Washington, DC.
260
260 [179--261] 10.12.2013 8:34PM
Xu, H., Yarin, A. L., Reneker, D. H., 2003. Characterization of uid ow in jets during electrospinning. Polymer Preprints 44, 5152.
Xu, X. L., Zhuang, X. L., Chen, X. S., Wang, X., Yang, L., Jing, X., 2006. Preparation of core-sheath
composite nanobers by emulsion electrospinning. Macromol. Rapid Commun. 27, 16371642.
Yan, F., Farouk, B., Ko, F., 2003. Numerical modeling of an electrically driven liquid meniscus in
the cone-jet mode. Aerosol Sci. 34, 99116.
Yarin, A. L., 1979. Stability of a jet of viscoelastic liquid in the presence of a mass ux at its surface.
J. Eng. Phys. 37, 904910.
Yarin, A. L., 1990. Strong ows of polymeric liquids: 1. Rheological behavior. J. Non-Newton.
Fluid Mech. 37, 113138.
Yarin, A. L., 1995. Surface-tension-driven low Reynolds number ows arising in optoelectronic
technology. J. Fluid Mech. 286, 173200.
Yarin, A. L., 1997. On the mechanism of turbulent drag reduction in dilute polymer
solutions: dynamics of vortex laments. J. Non-Newton. Fluid Mech. 69, 137153.
Yarin, A. L., 2007. Self-similarity. Section 2.3 in Springer Handbook of Experimental Fluid
Mechanics (Eds. Tropea, C., Yarin, A. L., Foss, J.), Springer, Berlin, pp. 5782.
Yarin, A. L., 2011 Coaxial electrospinning and emulsion electrospinning of coreshell bers.
Polym. Adv. Technol. 22, 310317.
Yarin, A. L., Kataphinan, W., Reneker, D. H., 2005. Branching in electrospinning of nanobers.
J. Appl. Phys. 98, 064501.
Yarin, A. L., Koombhongse, S., Reneker, D. H., 2001a. Bending instability in electrospinning of
Yarin, A. L., Koombhongse, S., Reneker, D. H., 2001b. Taylor cone and jetting from liquid droplets
in electrospinning of nanobers. J. Appl. Phys. 90, 48364846.
Yarin, A. L., Weiss, D. A., 1995. Impact of drops on solid surfaces: self-similar capillary waves, and
splashing as a new type of kinematic discontinuity. J. Fluid Mech. 283, 141173.
Yarin, A. L., Zussman, E., 2004. Upward needleless electrospinning of multiple nanobers.
Polymer 45, 29772980.
Yarin, A. L., Zussman, E., Wendorff, J. H., Greiner, A., 2007. Material encapsulation in coreshell
micro/nanobers, polymer and carbon nanotubes and micro/nanochannels. J. Mater. Chem
17, 25852599.
Yarin, L. P., 2012. The Pi-Theorem. Applications to Fluid Mechanics and Heat and Mass Transfer.
Springer, Berlin.
Yazdani, M., Seyed-Yagoobi, J., 2009. Electrically induced dielectric liquid lm ow based on
electric conduction phenomenon. IEEE Trans. Dielectr. Electr. Insul. 16, 768777.
Yu, J. H., Fridrikh, S. V., Rutledge, G. C., 2004. Production of submicrometer diameter bers by
two-uid electrospinning. Adv. Mater. 16, 15621566.
Yu P. Y., Cardona, M., 2010. Fundamentals of Semiconductors. Physics and Material Properties.
Springer, Berlin.
Zeldovich, Ya. B., 1992. Limiting laws of freely rising convective currents. Selected Works of Ya.B.
Zeldovich, Vol. 1. Chemical Physics and Hydrodynamics. Princeton University Press, Princeton
(rst published in 1937).
Zhang, J. F., Yang, D. Z., Xu, F., Zhang, Z. P., Yin, R. X., Nie, J., 2009. Electrospun coreshell
structure nanobers from homogeneous solution of poly(ethylene oxide)/chitosan.
Macromolecules 42, 52785284.
261 [179--261] 10.12.2013 8:34PM
5.11 References
261
Zhang, X., Basaran, O. A., 1996. Dynamics of drop formation from a capillary in the presence of an
electric eld. J. Fluid Mech. 326, 239263.
Zhang, Y., Huang, Z. M., Xu, X., Lim, C. T., Ramakrishna, S., 2004. Preparation of coreshell
structured PCL-r-Gelatin bi-component nanobers by coaxial electrospinning, Chem. Mater.
16, 34063409.
Zussman, E., Rittel, D., Yarin, A. L., 2003a. Failure modes of electrospun nanobers. Appl.
Phys. Lett. 82, 39583960.
Zussman, E., Theron, A., Yarin, A. L., 2003. Formation of nanober crossbars in electrospinning.
Appl. Phys. Lett. 82, 973975.
Zussman, E., Yarin, A. L., Bazilevsky, A. V., Avrahami, R., Feldman, M., 2006. Electrospun
Polyacrylonitrile/Poly(methyl methacrylate)-derived carbon micro-/nanotubes. Adv. Mater.
18, 348353.
262 [262--296] 10.12.2013 9:35PM
Additional methods and materials

used to form micro- and nanofibers
This chapter covers several additional methods used to form micro- and nanobers.
Some of them have already achieved maturity, such as the island-in-the-sea method
discussed in Section 6.1, melt fracture in meltblowing processes (Section 6.2) and the
ash spinning process (Section 6.3). Some others are still relatively exotic or under
development, like the two methods based on Couette shear ow described in Section 6.4,
or the centrifugal spinning method in Section 6.5. Nontraditional materials used for
nanober formation, discussed in Section 6.6, include liquid crystals, conducting polymers, biopolymers and denatured proteins. Finally, Sections 6.7 and 6.8 discuss the
specics of drawing glass optical bers, and in particular, polarization-maintaining
optical bers with multilobal cladding (Section 6.8).
6.1
Island-in-the sea multicomponent fibers and nanofibers

Microscopic bi- and multicomponent bers can be formed using melt spinning
(Section 1.2 in Chapter 1), meltblowing (Section 4.1 in Chapter 4), or integrated
processes such as spunbonding (Section 1.5 in Chapter 1). The additional polymer
components are supplied to the main polymer through separate inner spinnerets inserted
into the main outer spinneret similarly to formation of coreshell bicomponent bers in
co-electrospinning (Section 5.8 in Chapter 5) and solution blowing (Section 4.8 in
Chapter 4). Cross-sections of such bi- and multicomponent bers are reminiscent of
islands in the sea, which explains the name of this technology (Nakajima 2000). In some
cases the islands can merge and form winged structures, as seen in Figure 6.1a.
The sea material can be either dissolved, or mechanically split or brillated by water
jets, resulting in nanobers (the former islands) with diameters of about 300 nm
(Nakajima 2000, Pourdeyhimi et al. 2008). Consolidated nonwoven webs produced by
this method possess high surface area (~ 5 m2 g1) with variable morphologies and
compositions (see Figure 6.1b and Pourdeyhimi and Chappas 2008) and form exceptionally strong assemblies with fully developed bers (Durany et al. 2009).
6.2
Fibers from melt fracture in meltblowing processes

Meltblowing, discussed in Chapter 4, has been the king of microber production for
many decades it is hard to imagine what the nonwovens industry would be today in the
263 [262--296] 10.12.2013 9:35PM
6.2 Fibers from melt fracture
(a)
Figure 6.1
263
(b)
(a) Typical winged ber cross-section showing both the core and shell materials of a bicomponent
ber. (b) The shell can be removed by dissolution after web formation, as with these winged bers
in a consolidated web. The scale bar is 10 m, the separate wing section is sub-micron.
Pourdeyhimi and Chappas (2012, 2013).
absence of meltblowing. This has been an enabling technology leading to incredible

innovations in medical, hygiene, ltration, adsorbents and many other market segments.
Typical meltblowing can produce bers in the range of 2 m at high throughputs
(>80 kg m1 h1 beam1). This is several orders of magnitude higher than electrospinning, discussed in Chapter 5; however, the bers are also signicantly larger.
Recently, it was demonstrated that it is possible to produce bers down to 300 nm or
less by using novel dies and reduced throughput (at about 10 kg m1 h1 beam1; Hassan
et al. 2013). One notable advantage of meltblowing is that it is done with molten polymer
(the use of solvents is frowned upon by many today due to environmental concerns) and
uses readily available, well-developed polymer stock. And meltblowing technology is
not staying static. There are signicant developments underway that are and will be
resulting in major advances in the ability to produce large quantities of micro- and
nanobers. Consider, for example, developments enabled by systems that rely on
polymer melt fracture to yield sub-micron bers. Note that this is different from melt
spinning instability, also called melt fracture, discussed in Section 1.2 in Chapter 1. Here,
a polymer lm is formed and then brillated by high shear stresses caused by air streams
impinging onto the lm. The net result is bundles of micro- and nanobers. A critical
feature of melt fracture is the ability to use the technology to form ne bers. The origin
of the latest technologies being practiced is the patents of Torobin and Findlow (2001)
and Reneker et al. (2002). Both these patents use annular lms that are fractured by air
streams to form ne bers. In the patent by Chhabra and Isele (2007), the rst claim
states: A method of forming a nonwoven web comprising the steps of: (i) forming bers
from a melt lm brillation process in a rst uid stream, (ii) providing a coating
substance in the rst uid stream, (iii) applying the coating substance onto the surface
of the bers, and (iv) depositing the coated bers on a surface to form a web. The latest
264
264 [262--296] 10.12.2013 9:35PM
6 Additional methods and materials
Negative charge
electrostatic
V
inducer
110
2
105
101
103
113
104
108
102
107
106
111
30
109
Polymer
solution
22
2
Positive charge
ground plate
A. (USP3.081 . 519)
Figure 6.2
112
60 Cycle
alternating current
static dissipator
23
33
32
21
20
31
Inventor
Walter steuber
112
B.
C.
D.
Flash spinning process. Blades and White (1963).
commercial technology developed and introduced by PGI (Arium Technology) is also

based on similar concepts, but is extended to a curtain spinning (Textile World 2011).
6.3
Fibers from flash spinning processes

DuPonts web page states that the discovery of DuPont Tyvek was due to a DuPont
researcher, Jim White, who, in 1955, noticed white polyethylene uff coming out of a
pipe in a DuPont experimental lab. A program to develop the new material was set up,
and a year later DuPont submitted a patent proposal for strong yarn linear polyethylene
(DuPont 2013). In 1959, a pilot facility was established for trial applications. In 1965, the
new engineered sheet structure was registered under the trademark name DuPont
Tyvek, and was introduced commercially in April 1967. DuPont was the sole manufacturer of ash-spun nonwovens until the mid-1990s, when it was joined by Asahi Kasei
Kogyu Kabushiki Kaika of Japan.
Here, we focus on the DuPont process. Blades and White (1963) were granted a US
patent that describes the process as one forming bers/webs from a polymer in solution, at a
temperature higher than the normal boiling point of the solvent, under high pressure,
through a small orice into a zone of considerably lower pressure (Figure 6.2). Upon exit,
the solvent ashes off leaving behind an extrudate in the form of a three-dimensional
interconnected network of very ne brillated laments. A critical feature of the bers
formed is the irregular cross-sections, leading to an incredibly large surface area. Todays
Tyvek approaches a specic surface area of 30 to 40 m2 g1. Indeed, the patent described
that bers thus formed would have surface areas higher than 2 m2 area per gram of material.
6.4
Fibers from polymer solutions in Couette flow

Zwijnenburg and Pennings (1976) and Pennings et al. (Chapter III in Ciferri and Ward
1979) observed ow-induced crystallization of polyethylene from 0.1% p-xylene
265 [262--296] 10.12.2013 9:35PM
6.4 Fibers from polymer solutions
265
Variable speed
take-up roil
2
05
=0
Wire
Coloured
fluid
.05
.2
.4
.6
.8
Outer
cylinder
Inner
cylinder
solution
Polymer
= 1.0
Rotor
150
Solvent
112
Seed
135
=0
Teflon bearing
05
Couette flow
(a)
Figure 6.3
(b)
(a) Couette apparatus where polymer solution is contained in the gap between a rotating inner
cylinder (the rotor) and a cylindrical outer shell (at rest). A seed crystal is inserted using a thin
wire through a small Teon pipe on the right. As the ow-induced crystallization of polyethylene
onto the seed crystal begins, the growing linear crystals form a ber that is withdrawn through
the pipe by a pulley. Zwijnenburg and Pennings (1976), with kind permission of Springer
Science+Business Media. (b) Taylor vortices in TaylorCouette ow. Taylor (1923). Courtesy
of the Royal Society of London.
solutions in the shear ow in the Couette apparatus sketched in Figure 6.3a. The
apparatus presumably realizes simple shear ow. However, this ow may become
unstable and a system of Taylor vortices sets in. These toroidal vortices encircle the
rotor (Figure 6.3b). The ow between the two neighboring vortices is elongational and
can result in a coilstretch transition and formation of linear brillar crystals called shishkebab structures, which are discussed in more detail in Section 2.6 of Chapter 2. Note,
that other ows with signicant velocity gradients, e.g. the Poiseuille laminar pipe ow,
can also be used to form ow-induced linear brillar crystals as described by Pennings
et al. (Chapter III in Ciferri and Ward 1979).
Continuous bers of polyethylene with lengths of several hundred meters were pulled
from the Teon pipe of the apparatus shown in Figure 6.3a. The bers have diameters on
the scale of 100 m (Figure 6.4) and consist of bundles of parallel brillar shish-kebab
crystals.
Velev et al. (2011) used precipitation of polymer macromolecules in Couette ow in a
method that is akin to wet spinning (Section 1.4 in Chapter 1). A startup company,
XanoShearTM, is developing a commercial process based on this approach. The diameter
of the resulting polymer nanobers is on average 500 nm, the length is unspecied, but
266
266 [262--296] 10.12.2013 9:35PM
25 m
Figure 6.4
Scanning electron micrograph of polyethylene bers formed in the Couette apparatus.

Zwijnenburg and Pennings (1976), with kind permission of Springer Science+Business Media.
should be relatively short, and the minimum production rate of the batch process is more
than 3.6 kg h1 (Luo et al. 2012).
6.5
Centrifugal spinning, forcespinning

A recently introduced process for making nanobers called forcespinning (Sarkar
et al. 2010) is essentially akin to cotton candy machines and rotary disk atomizers,
which use centrifugal force to issue liquid jets from rotating drums or disks. In
forcespinning, polymer solutions or melts are issued from a heated rotating disk-like
spinneret through orices on its circumference (Figure 6.5). Nanobers from a
number of polymers (e.g. poly(ethylene oxide), polypropylene and polystyrene)
were formed from solutions and melts using this method. Forcespinning is characterized by a high production rate, which can be regulated by the spinneret rotation
speed in the range 30005000 rpm.
A similar idea was realized by Dosunmu et al. (2006), where polymer jets were issued
radially from a porous tube rotating about its axis. The technological challenge of such
methods as those of Dosunmu et al. (2006) and Sarkar et al. (2010) is in the difculties in
collecting large areas of continuous nanober mats. Regardless, this is the process that is
being commercialized by the company Fiberio (2013).
6.6
Electrospinning of liquid crystals, conducting polymers,

biopolymers and denatured proteins
Ordered liquid crystalline phases are mainly classied as nematic, cholesteric and
smectic (Donald et al. 2006). The structural basis for each of them is a state of matter
in which the degree of molecular order is intermediate between the perfect, threedimensional, long-range positional and orientational order found in solid crystals, and
267 [262--296] 10.12.2013 9:35PM
6.6 Electrospinning of liquid crystals
267
In
Environment chamber
Spinneret
Oven
Heater
Fiber
collector
Out
Figure 6.5
Motor
Bruke
Forcespinning apparatus. Top left: schematic. Top right: a prototype device. Bottom: spinneret
sketch. Reprinted from Sarkar et al. (2010), with permission from Elsevier.
the absence of long-range order found in isotropic liquids, gases and amorphous solids.
Aromatic polyamides contain conjugated rings along the main chain, and the most wellknown macroscopic bers produced from them are poly(p-phenylene terephthalamide)
(PPTA) bers, commercialized under the trade name Kevlar 49 (Dupont).
Electrospinning of a liquid crystalline solution of PPTA in sulfuric acid was demonstrated by Srinivasan and Reneker (1995). They collected nanobers on the free surface
of water in a grounded bath where PPTA solution precipitated, and solidied bers
were formed. The annealing treatment of these nanobers improved the local crystallographic packing so that the electron diffraction patterns from the electrospun nanobers
were similar to X-ray diffraction patterns from commercial Kevlar bers (Reneker and
Yarin 2008). Aromatic heterocyclic polymers represent lyotropic materials and are
characterized by wholly aromatic molecular structures with fused heterocyclic rings
along the main chain. Acetoxypropyl cellulose (APC), a lyotropic cholesteric liquid
crystalline polymer, was successfully electrospun by Canejo et al. (2008) as microbers (~2.75 m) using a standard electrospinning setup with a solid counter electrode
collector. Liquid crystal elastomers (LCE) were electrospun by Krause et al. (2007) as
microbers with an average diameter of 1.5 m from a solution of photo-cross-linkable
polymer using a conventional electrospinning apparatus. They obtained only beaded
bers, and cross-linking was conducted using UV light as a post-process. Nanobers of
a liquid crystalline block co-polymer [BB-5(3-Me)] were electrospun by Nakashima
et al. (2010).
Aromatic co-polyester polymers possess thermotropic behavior and are characterized
by a molecular structure with a high degree of linearity and rigidity that allows formation
268
268 [262--296] 10.12.2013 9:35PM
of ordered phases over a wide temperature range; three examples are Vectra, Xydar
and Ekonol (Pegoretti and Traina 2009). Medeiros Araujo et al. (2013) used electrospinning of a liquid crystalline polyester (Vectran) alone and in blends with
poly(ethylene oxide), PEO, to form nanobers, and characterized their crystalline
and mechanical properties prior to and after removal of the PEO. They showed that
pure Vectran solutions possess insufcient spinnability to produce continuous and
uniform nanobers using electrospinning. The use of a hostguest approach, with the
host being a high-molecular-weight PEO, was employed to improve the spinnability of
Vectran. Then, continuous and uniform nanobers with an average diameter of 195 nm
were formed. After that, PEO was eliminated by heat treatment at 250 C for 24 h in air.
Finally, annealing of the Vectran nanober mats increased thermal stability due to
enhancement of the structural order. Correspondingly, tensile tests showed a signicant
improvement in the mechanical properties. The mechanical behavior in tensile tests
changed from relatively soft to stiff, and the elongation at break has also increased after
PEO was fully eliminated.
MacDiarmid et al. (2001) and MacDiarmid (2002) formed nanobers by electrospinning a conducting polymer, polyaniline doped with camphorsulfonic acid (PAn.
HCSA). Electrospun nanobers from a conducting conjugated polymer, polyaniline
(Versicon) solutions in concentrated sulfuric acid (9598%) containing 1518% by
weight of the polymer, were reported by Srinivasan (1994), Reneker et al. (2007) and
Reneker and Yarin (2008). The electrospun bers were green in color in the conducting
state and were often curved and looped. The diameter of the polyaniline bers ranged
from about 40 nm to about a micron. Such bers can potentially be used in electromechanical actuators to convert electrical energy to mechanical energy using doping/
dedoping or charge-transfer reactions that cause large dimensional changes in the system.
MacDiarmid (2002) and Srinivasan (1994) proved that polyaniline nanobers were
electrically conducting.
Electrospinning is a very promising technique for the fabrication of luminescent
nanobers based on conducting, photosensitive polymers. The high value of the area
reduction ratio and the associated high stretching rate (~1000 s1) imply that macromolecules in the nanobers should be elongated and oriented in the direction of the jet
axis (see Chapter 5). Conjugated polymers are typically nonspinnable. Therefore, lightemitting conjugated polymers are blended and electrospun, together with a spinnable
host polymer, as guests. Utilizing the high degree of chain extension, chain alignment and structural order attainable, in principle, in electrospinning of the host
polymers, one can expect to induce a similar order in the guest conjugated macromolecules incorporated in nanobers. Then, nanobers which combine the mechanical
properties of the host polymer with the electric conductivity and the anisotropic linear
and nonlinear optical properties of the conjugated guest polymer are expected.
Electrospinning of a blend of conducting polyaniline (PAn.HCSA) and highly spinnable
poly(ethylene oxide), PEO, was demonstrated by MacDiarmid (2002), Norris et al.
(2000) and MacDiarmid et al. (2001). The electrospun bers were uniform and had
diameters ranging between 950 and 2100 nm. The bers formed a nonwoven mat with
high porosity and relatively low conductivity compared to cast lms.
269 [262--296] 10.12.2013 9:35PM
6.7 Nanofibers containing nanoparticles
269
Biopolymers/proteins typically have three-dimensional, rigidly xed, globular molecular structures. Therefore, their solutions do not possess sufcient viscoelasticity and are
not spinnable. However, they can be denatured by means of thermal treatment, alkali, for
example NaOH, sodium sulte and urea, as well as pH and ionic strength variation,
which convert protein macromolecules into a kind of a loose nylon. Such denatured
protein solutions can be electrospun into nanobers, as was done with bovine serum
albumin (BSA) (Kowalczyk et al. 2008, Dror et al. 2008). Denatured proteins can be also
solution blown, as was done with soy protein isolate (Sinha-Ray et al. 2011, Khansari
et al. 2012; for more detail see Section 4.9 in Chapter 4). Electrospun collagen nanobers
are expected to nd important biomedical applications, which are discussed in
Chapter 11.
6.7
Nanofibers containing nanoparticles and nanotubes

Zhang and Yarin (2011) introduced a novel method to functionalize nanober surfaces
with carbon or silica nanoparticles by dip coating. They applied dip coating of electrospun and carbonized nanober mats in poly(furfuryl alcohol) (abbreviated as FFA) to
increase surface roughness by means of FFA-derived carbon nanoparticles produced at
the ber surface. Also, they used dip coating in tetraethylorthosilicate (abbreviated as
TEOS) to decorate carbon nanobers with TEOS-derived silica nanoparticles at the
surface. Dip coating is an inexpensive technique that is easier to implement than some
other methods of nanober decoration with silica nanoparticles. It also results in a higher
loading capacity. Examples of electrospun and carbonized nanobers decorated with
nanoparticles using dip coating are shown in Figure 6.6.
EMS
Figure 6.6
SEI
5.0 kV x60,000 100 nm WD 9.0 mm
EMS
SEI
5.0 kV x19,000 1 m
WD 9.0 mm
SEM images of electrospun and carbonized nanobers. (Left) Grain-like structures at the ber
surface resulting from a carbonized FFA lm applied in the dip coating process. (Right)
Electrospun carbon nanobers decorated with silica nanoparticles. Reprinted with permission
from Zhang and Yarin (2011). Copyright 2011, American Chemical Society.
270
SEI 5.0 kV x10,000 1 m WD 9.0 mm
EMS
Figure 6.7
EMS
Figure 6.8
270 [262--296] 10.12.2013 9:35PM
EMS
SEI
5.0 kV x10,000 1 m WD 9.0 mm
(Left) PAN bers with copper nanoparticles inside. (Right) Copper nanoparticles replaced by
silver. Zhang and Yarin (unpublished, 2012).
SEI
5.0 kV x40,000 100 nm WD 9.0 mm
EMS
SEI 5.0 kV x40,000 100 nm WD 9.0 mm
Images of the individual nanobers. (Left) PAN ber with copper nanoparticles inside. (Right)
Copper nanoparticles replaced by silver. Zhang and Yarin (unpublished, 2012).
To decorate carbon nanobers with silver and palladium nanoparticles, Zhang and
Yarin (unpublished, 2012) blended copper nanoparticles with polyacrylonitrile, PAN,
solutions that were electrospun. The resulting nanober mats contained Cu nanoparticles
on the surfaces as well as inside the individual nanobers. The as-spun nanober mats
were submerged in a silver nitrate bath and an ion-exchange reaction took place. As a
result, all Cu nanoparticles on the nanober surface were converted into Ag nanoparticles. Probably most of the Cu nanoparticles embedded in the bulk were also reached
by silver nitrate through the individual ber pores and converted into embedded, but still
accessible silver particles. The images in Figures 6.7 and 6.8 illustrate the as-spun Cucontaining nanober mats and the same mats decorated by silver nanoparticles. In
addition, in Figure 6.9 it can be seen that copper nanoparticles can disappear completely,
271 [262--296] 10.12.2013 9:35PM
6.7 Nanofibers containing nanoparticles
EMS
SEI
5.0 kv x5,000
EMS
Figure 6.9
EMS
Figure 6.10
1 m WD 9.0 mm
SEI
5.0 kv
EMS
x40,000
5.0 kv
SEI
100 m
x30,000 100 m
271
WD 9.0 mm
WD 9.0 mm
Copper nanoparticles replaced by silver. Some holes are left at the previous locations of copper
nanoparticles. Zhang and Yarin (unpublished, 2012).
SEI
5.0 kv x4,000
1 m WD 9.0 mm
EMS
SEI
5.0 kv
x25,000 1 m
WD 9.0 mm
Pd-decorated nanobers. Zhang and Yarin (unpublished, 2012).
while tiny silver nanoparticles form inside the craters and down into the ber body. A
similar ion exchange can be done with PAN nanober mats containing Cu nanoparticles
after they are submerged in a bath containing Pd(NO3)2 or Pd(Cl)2. The resulting Pddecorated nanobers are shown in Figure 6.10.
272
272 [262--296] 10.12.2013 9:35PM
a0
a0
V0, T0
ap
Vp, Tp
V0, T0
a1
V1, T1
V1, T1
(a)
Figure 6.11
(b)
Drawing of optical bers. (a) Drawing from a heated preform. (b) Drawing from spinneret hole.
Yarin (1982, 1993), with kind permission of Springer Science+Business Media.
Overall, electrospinning and solution blowing with appropriate post-processing yield a

plethora of ways to develop nanotextured catalyst supports.
6.8
Drawing of optical microfibers

Optical bers typically are formed from glass and have cross-sectional diameters of the
order of 100 m. There are two main methods used to form optical bers: (i) ber
drawing from a cylindrical glass preform locally heated to a high enough temperature for
the glass to melt and ow (Figure 6.11a); (ii) spinning through a spinneret hole from a
tank containing melt, which is similar to melt spinning of polymers (Figure 6.11b, cf.
Section 1.3 in Chapter 1). In both cases the material is supplied at a constant velocity V0,
and as a ber is drawn, it is cooled down by the surrounding cold gas atmosphere, and
solidies.
To ensure glass purity, preforms are typically produced using the following method,
sometimes called modied chemical vapor deposition (MCVD). A homogeneous mixture of gases, for example, methane, oxygen and tetrachlorosilane (CH4, O2 and SiCl4,
respectively) is supplied to a Bunsen burner and burned as a torch. The corresponding
overall reactions are
CH4 2O2 ! CO2 2H2 O
6:1
SiCl4 gas O2 ! SiO2 condensed glass particles 2Cl2
6:2
Tetrachlorosilane oxidizes vigorously only at the high temperature provided by a methane ame, since the reaction in Eq. (6.2) is only slightly exothermic and has a high
activation energy. Roughly speaking, the thermal effect of the second reaction is
273 [262--296] 10.12.2013 9:35PM
6.8 Drawing of optical microfibers
273
y
3
1
Gas mixture
input
x
Figure 6.12
4
4
z
3
Sketch of the N-MCVD process for the production of preforms. 1: support tube, 2: round
burner, 3: linear cooler, 4: linear heater. Copyright 1991, IEEE. Reprinted, with permission, from
Doupovec and Yarin (1991).
negligibly small compared to the rst. At the end of the torch a hot jet of combustion
products carrying glass particles moves along a cold seed rod. Thermophoresis (Rosner
2000) moves glass (silica) particles across the gas jet until they deposit onto the rod. A
uffy layer of glass particles grows on the rod like an icicle. After a while, this protopreform with its supporting seed rod is removed and heated up to sinter the deposited
glass layer, producing a preform of the type shown in Figure 6.11a via an MCVD process.
Alternatively, a hollow glass support tube can be used as a seeding surface, with a gas
jet burning inside. Down the ow, hot combustion products carrying glass particles move
through the tube, as shown in Figure 6.12. Then, thermophoretic deposition happens
inside the tube. It can be manipulated to be nonaxisymmetric, as in Figure 6.12, where
particle deposition by thermophoresis happens onto the tube surfaces in contact with cold
strips (3) and is prevented on the heated surfaces (4). The nonsymmetric deposition is
benecial for drawing polarization-maintaining optical bers, as described in
Section 6.9. This nonsymmetric process is called N-MCVD.
It should be emphasized that preforms with a refractive index prole in cross-section
can be easily formed using other gases instead of tetrachlorosilane, for example germanium tetrachloride, GeCl4, so that GeO2 (germania) particles are deposited on the
growing preform.
Steady-state congurations of the glass jet arising in the drawing of optical bers from
solid glass preforms or molten glass, as in Figure 6.11, can be found by solving the general
quasi-one-dimensional equations of the dynamics of free liquid jets (Yarin 1982, 1993). In
this case, the continuity and momentum balance equations (Eqs. 3.1 and 3.2, with 3.8), as
well as their reduced versions (Eq. 1.3 and 1.5, with 1.10, take the following form:

d
dV
f V Q;
3f
0
6:3
dx
dx
Here the cross-sectional area f = a2, with a being the cross-sectional radius, and V is
longitudinal velocity. Also, Q a21 V1 is the volumetric ow rate, with a1 and V1 being
the ber cross-sectional radius and winding velocity at the receiving bobbin, respectively, is viscosity and x is the axial coordinate in the jet. As in Section 1.2 in Chapter 1,
surface tension, as well as gravity and air drag, are neglected as irrelevant in the optical
274
274 [262--296] 10.12.2013 9:35PM
ber-drawing process. Glass viscosity varies with temperature T according to Arrhenius

law (Doremus 1973)

U
0 exp
6:4
RT
where 0 and U are the pre-exponential factor and the activation energy of the viscous
ow, respectively, and R is the absolute gas constant (cf. similar Eq. 2.54 in Chapter 2 for
polymeric liquids). Note that in the framework of Eq. (6.4) there is no difference between
the solid glass in the preform and liquid molten glass. The former differs from the latter
only by a much higher value of viscosity corresponding to the lower temperature.
In the nonisothermal process under consideration, the following quasi-onedimensional thermal balance equation is added to Eqs. (6.3) and (6.4):
cQ
dT
2aqw
dx
6:5
where and c are density and specic heat, and qw is the heat ux at the jet surface. In the
case of ber drawing from a preform, the heat ux qw < 0 at the heating section, and at the
subsequent cooling section, qw > 0. A section of a preform can be heated by a surrounding ring-like furnace or by laser beam. In the case of ber drawing from a spinneret hole,
qw > 0 along the entire jet.
In the case of ber drawing from a preform, the latter is heated from an initial temperature
T0 at x = 0 to a temperature Tp at the end of the heating section at x = (see Figure 6.11a).
Then, the temperature of the molten glass decreases to a value T1 at a certain distance 1, i.e.
at x = L = + 1. In fact, the distances and 1 are considered to be unknown if temperatures
Tp and T1 are given. Alternatively, if the distances and 1 are given, Tp and T1 are to be
found. The solution found in Yarin (1982, 1993) provides the following relations between
these two pairs of quantities, which can be used in any of the above-mentioned cases:

p
p

Tp T0 V0
V0
V0
6:6
ln
Vp
A
2AU=RT2p

p A T1
1 A0 =A
1 EL 0
6:7
1 1 ln p
A Tp
E A0 =A
p
Here the draw ratio E = V1 / V0 is given, and correspondingly, a1 is found as a1 a0 = E.
Also
p
2 qw
Aw p
6:8
c Q
and in the heating zone, where qw < 0, Aw = A > 0, whereas in the cooling zone, where
qw > 0, Aw = A < 0. In addition, the following notations are used:
L
cTp Q
U
;
2qp a0
RTp
6:9
275 [262--296] 10.12.2013 9:35PM
275
with qp = qw(Tp) < 0, and thus L* > 0. The velocity Vp at the end of the heating zone in
Eq. (6.6) is determined from the following equation:
!
!
1
1
1
1
p p A p p A0
6:10
V0
V1
Vp
Vp
Correspondingly, one can nd the cross-sectional radius ap at the end of the heating
zone as
p
E A0 =A
6:11
ap
1 A0 =A
where ap is rendered dimensionless by a1.
The parametric dependence of the preform radius a on x in the heating zone, as well as
the corresponding temperature distribution, are given by the following dimensionless
expressions:

p p

a E
E ap exp T 1
6:12
n
h

p
io L
6:13
x T T0 1 ln 1 1 ap = E exp T 1
L
where overbars denote dimensionless parameters, the radius a is rendered dimensionless
by a1, the coordinate x by L = + 1, and temperature T by Tp.
In the cooling zone, the jet radius a and the corresponding temperature distribution are
given by the following dimensionless expressions:
p

1 E
a ap 1 p
T

1

1
6:14
exp
E A0 =A
p
p

1 A0 =A
1 E
EL A
1
p
x
T

1
ln
T

1

exp
L A0
L
E A0 =A
E A0 =A
6:15
The force P required for ber drawing is found as

2
3

120 a0 qp exp
ap
4
5
1 p
P
cTp
E
6:16
The stationary conguration of the optical ber drawn from a spinneret hole
(Figure 6.11b) is given by
p
p

a E 1 E exp T 1 1
6:17

T 1 1 ln E1=2 E1=2 1 exp T 1
x
6:18
T1 1 21 ln E
276
276 [262--296] 10.12.2013 9:35PM
p
Here radius is rendered dimensionless by a1 = a0/ E and temperature by the initial
temperature, and the longitudinal length scale L1 is given by

cT0 V1 a1
1
ln E
6:19
L1
T1 1
2q0 w
2
where qw = qw (T0) > 0. Note that L1 corresponds to the length at which temperature
reduces to the value T1. If L1 is given and equal to the distance to the winding bobbin,
then T1 is the ber temperature at the bobbin, which is found from Eq. (6.19).
The tensile force required for drawing optical bers from a spinneret hole is given by
p

120 exp E 1 a1 q0 w
6:20
P
cT0
It should be emphasized that in the limit of ! 0 the solution (Eq. 6.17) does not
approach the isothermal solution a E1x=2 given by the rst expression (Eq. 1.14) in
Chapter 1, since Eq. (6.17), as well as all the other results in the present section,
correspond to the limit >> 1, and do not allow to tend to zero.
To illustrate the results obtained, take the following parameter values. The initial crosssectional radius of the preform a0 = 0.19 102 m, the initial temperature T0 = 300 K, the
temperature to which the preform will be heated up Tp=1873 K, the winding velocity at
the receiving bobbin V1 = 0.3 m s1, which corresponds to the cross-sectional radius of
the optical ber a1 = 62.5 m, the ber temperature at the bobbin T1 = 300 K. In addition,
the glass density = 2.2 103 kg m3 and its specic heat c = 1.043 103 J kg1 K1. The
heat ux from the heater at the highest temperature is assumed to be qp = qw(Tp) = 92.7
104 J m2 s1, with being the emissivity of the glass.
The congurations of bers drawn from preforms calculated using Eqs. (6.12)(6.15)
are shown in Figures 6.136.15. The vertical bars on the ber proles mark the end of the
heating zone. In all cases the draw ratio E = 924 is much larger than the critical value of
E = 20.22 for the onset of the draw resonance instability (Section 1.2 in Chapter 1), since
cooling below the heating zone is a strong stabilizing factor (Yarin 1993). When the
dimensionless activation energy of the viscous ow is increased beyond a value of 10,
with the other parameters xed, the ber conguration does not change further in
comparison with curve 2 in Figure 6.13. The reduction of the heat ux at the heating
2.5
a.103, m
1
0
Figure 6.13
2
2.5
5.0 x.103, m
Optical ber drawn from a heated preform. Parameter values: = 1, A/A = 1. Curve 1 corresponds
to = 7 (the lower dimensionless activation energy of the viscous ow), curve 2 to = 10 (the
higher dimensionless activation energy of the viscous ow). The draw ratio E = 924. From Yarin
(1982, 1993), with kind permission of Springer Science+Business Media.
277 [262--296] 10.12.2013 9:35PM
277
2.5
a.103, m
1
0
Figure 6.14
2.5
5.0
x.103, m
Optical ber drawn from a heated preform. Parameter values: = 7, A/A = 1. Curve 1
corresponds to glass emissivity = 1, curve 2 to = 0.5. The draw ratio E = 924. Yarin (1982,
1993), with kind permission of Springer Science+Business Media.
2.5
a.103, m
1
2 3
0
Figure 6.15
2.5
5.0 x.103, m
Optical ber drawn from a heated preform. Parameter values: = 7, = 1. Curve 1 corresponds
to A/A = 0.2, curve 2 to A/A = 1, curve 3 to A/A = 3. The draw ratio E = 924. From Yarin
(1982, 1993), with kind permission from Springer Science+Business Media
2.5
a.103,
m
2
0
0
Figure 6.16
2.5
5.0
x.103, m
Optical ber drawn from a spinneret hole. The initial temperature of molten glass is T0 = 1873
K. Curve 1 corresponds to the heat removal rate qw = qw(T0) = 92.7 103 J m2 s1, curve 2 to
qw = qw(T0) = 92.7 104 J m2 s1. In both cases = 7 and E = 924. Yarin (1982, 1993), with
kind permission from Springer Science+Business Media.
and cooling stages due to reduced ber emissivity, with the other parameters xed, results
in a less abrupt ber thinning, as curve 2 in Figure 6.14 shows. An increase in the heat
removal rate at a xed heat ux in the heater (the other parameters xed) results in a
noticeable enhancement of ber thinning, as is seen in Figure 6.15. This is also accompanied by a signicant growth in the pulling force P required to sustain the ber drawing
process.
Note that a similar solution was also obtained for the drawing of glass capillaries
(hollow optical bers) from a heated preform (Yarin et al. 1989).
In the case of drawing optical bers from a spinneret hole, the results found using Eqs.
(6.17) and (6.18) are illustrated in Figure 6.16. It can seen that, similarly to ber drawing
from a preform (Figure 6.15), a higher cooling rate makes ber thinning more abrupt
(curve 2 in Figure 6.16).
278
278 [262--296] 10.12.2013 9:35PM
1
,
mm
0.1
0.01
0.001
0
Figure 6.17
10
20
30
40
50
60
x, mm
Optical ber spinning from a spinneret hole. The solid line shows the analytical result, Eqs. (6.17)
and (6.18), the dashed line corresponds to the numerical two-dimensional solution of Huynh and
Tanner (1983), the symbols to the experimental data of Glicksman (1968). From Yarin (1993),
courtesy of Pearson Education.
To compare the analytical results of the present section for optical ber drawing from a
spinneret hole with the experimental data of Glicksman (1968) and the numerical
solution of the corresponding problem by Huynh and Tanner (1983), take the following
values for the parameters. The glass density = 2.4 103 kg m3, its specic heat c =
1.0466 103 J kg1 K1, and the initial temperature of molten glass T0 = 1499.7 K. The
pre-exponential in Arrhenius law (Eq. 6.4) is taken as 0 = 0.948 109 Pa s, and the
dimensionless activation energy = 25.2. The radius of the spinneret hole a0 = 838 m
and the outow velocity at the exit from the spinneret hole V0 = 0.3124 102 m s1. Heat
is removed by convection and thermal radiation. According to Huynh and Tanner (1983),
qw,conv = qw,conv(T0) = 0.43 105 J m2 s1 and qw,rad = qw,rad(T0) = 0.554 105 J m2 s1.
Therefore, the overall heat removal rate is qw = qw,conv(T0) + qw,rad(T0) = 0.984 105 J
m2 s1. The draw ratio E = 5310, and the ber length L1 = 0.06 m.
The analytical results of the present section are compared with the experimental data
and the numerical results in Figure 6.17. The comparison shows that the analytical
asymptotic results are fully corroborated by the experimental data and direct numerical
simulations. The tensile force predicted by Eq. (6.20) is P = 173.6 105 N, which is in
good agreement with the value of P = 200 105 N corresponding to the experimental
data (Huynh and Tanner 1983).
6.9
Polarization-maintaining optical microfibers and multilobal fibers

Standard monomode optical bers drawn from preforms prepared by means of the
modied chemical vapor deposition method (MCVD), as described in Section 6.8
possess axially symmetric layers (if any), e.g. the core, cladding and outer matrix. Two
independent orthogonally polarized modes can be propagated in such axisymmetric
layered bers above the cut-off wavelength. However, this polarization state is nearly
279 [262--296] 10.12.2013 9:35PM
6.9 Polarization-maintaining optical microfibers
279
r
0
1
Figure 6.18
Initial conguration in the cross-section of a preform. The circular domain 0 is the core, where a
signal propagates, 1 is the stress cladding and 2 is the outer matrix. 0 is the boundary between the
core and cladding, 2 is the boundary between the cladding and outer matrix and 2 is the outer
boundary of the cross-section. Yarin (1990, 1995), with kind permission from Springer Science
+Business Media.
degenerate, the resulting birefringence is small and the polarization state of the propagating wave suffers from unstable uctuations due to changing ambient conditions
(Doupovec and Yarin 1991). Therefore, polarization-maintaining optical bers with
cross-sectional asymmetry are called for.
The polishing method used for creation of preforms for drawing of polarizationmaintaining optical bres is based on the following principles (Kaminow and
Ramaswamy 1979). The initial cross-section of a glass preform is shown in
Figure 6.18, where domain 0 corresponds to the core through which a signal propagates,
domain 1 to the cladding that serves to impose stresses on the core and domain 2 to the
outer matrix of the preform (as well as the ber that will be drawn from it).
Cylindrical preforms with three different coaxial glass layers are formed by the MCVD
method described in Section 6.8. The material (glass) of the core differs from the
cladding and outer matrix in composition and physical properties. In its turn, the material
(glass) of the cladding differs from that of the outer matrix.
To form nonsymmetric polarization-maintaining bers, a part of the outer matrix is
removed (polished) as shown by the dashed lines in Figure 6.18, for example and the
preform is placed in a furnace and heated until the cladding and outer matrix soften. The
core remains hard. Surface tension at the boundary 2 begins to round it off. The resulting
ow of molten glass deforms the boundary 1 subjected to interfacial tension, which is
lower than the surface tension at the boundary 2. Deformation of the boundary 1 causes
it to lose its circular form. Meanwhile, the boundary 0 remains unchanged since the core
continues to be hard. Note that the case of a negligibly small core (effectively, a two-layer
preform) is also of interest.
Cooling and solidifying of the preform at some intermediate moment of time yields a
hard preform with a noncircular cladding boundary 1, whereas the outer boundary 2 is
already practically circular (the boundary 0 is always circular). It should be emphasized
280
280 [262--296] 10.12.2013 9:35PM
that the duration of the heat treatment should not be very long, since in the end the
boundary 1 will also begin to approach a circle if the interfacial tension is nonzero.
Owing to the difference in the thermoelastic properties of the materials in the cladding
and the outer matrix, an anisotropic eld of elastic stresses is created in the hard preform
cross-section (as well as in the optical ber drawn from it), which results in desirable
birefringence. Accordingly, the core becomes capable of transmitting signals with a
certain polarization.
One of our goals in the present section is to predict what shape the cladding will have
for a given initial shape of the outer surface of the matrix and material parameters of the
two- and three-layer performs, as was done by Yarin (1995). We also address the inverse
problem prediction of the polished shape of the outer matrix needed in order to arrive at
a prescribed shape for the cladding.
The second method of preform production employs the modied chemical vapour
deposition process (MCVD; Kaminow and Ramaswamy 1979) or the nonsymmetric
variant (N-MCVD; Doupovec and Yarin 1991), described in Section 6.8. In the present
case, in both techniques glass particles are thermophoretically deposited from a gas ow
onto the inner surface of a glass substrate tube, creating a coating. Afterwards the tube is
heated, softens and begins to collapse. The latter means that creeping ow of highly
viscous liquid (glass) directed towards the center of the cross-section arises under the
action of surface tension, which tends to reduce the free surface area, lling up the cavity
with material. Thus, the slow viscous ow of the glass is driven by surface tension and
(perhaps) a pressure differential between the inner and outer tube surfaces (Geyling et al.
1983). In the present section we aim to describe analytically the collapse of a substrate
tube with radially nonsymmetric layers inside, following Yarin (1995) and Yarin et al.
(1993). The simplest model system with a single-layer coating, shown in Figure 6.19, is
considered. The viscous ow arising in this geometry differs signicantly from the
collapse of axisymmetric tubes (Lewis 1977, where the interfacial tension was taken to
1
0
1
2
Figure 6.19
Single-layer coating in the tube-collapse process. Layers 1 and 2 represent the deposited domain
and the substrate tube material domain, respectively. Boundaries 0, 1 and 2 denote the inner, the
median and the outer interfaces, respectively. Yarin (1995). Courtesy of Cambridge University
Press.
281 [262--296] 10.12.2013 9:35PM
281
be zero and pressure difference was accounted for; Das and Gandhi 1986, where the
interfacial tension was also taken to be zero and viscosity/temperature dependence was
accounted for).
With the system shown in Figure 6.19 collapsed, solidication results in a nonsymmetric two-layer preform (the case of a negligibly small core), which possesses birefringence and polarization-maintaining properties for the same reasons as for the polishing
method the difference in the thermoelastic properties of the materials.
In the polishing method, a preform is heated in a furnace by a convective medium at
distant temperature T. A reasonable value of the heat transfer coefcient h = 1.5 102 W
m2 K1 (Paek and Runk 1978). Taking the radius of the unpolished outer matrix R2 = 0.6
102 m and the thermal conductivity of glass k = 0.3 102 W m1 K1 (Paek and Runk
1978), we obtain the inverse Biot number Bi1 = k/(hR2) = 33.3. The relationship
between the temperature at any radius in a cylinder and the temperature on the centerline
can be found from the known solution for conductive heat transfer or its graphic
representation in the form of a Heisler chart (e.g. Figure 4.11 in Bejan 1993). As a result,
the temperature eld in the cylinder is nearly homogeneous for Bi1 = 33.3 during the
entire heating process, as is usually supposed for preforms and optical bres (e.g. see
Paek and Runk 1978).
However, the temperature of the cylinder changes with time. For Bi1 = 33.3 it takes an
approximate time t0 = 40R22 /T, (T is the thermal diffusivity) to heat the cylinder to
temperature T. Taking the glass density = 2.2 103 kg m3 and the specic heat
c = 1.05 103 J kg1 K1, as well as the above-mentioned value of k (Paek and Runk
1978), we obtain T = 0.13 104 m2 s1 and t0 = 111 s. All the above estimates also hold
for a polished preform.
The viscosity of glass ts the Arrhenius-type equation (Eq. 6.4). Molten glass is a
highly viscous Newtonian liquid. If we take the temperature T, corresponding to the
working point of a glass Tw, (at which the viscosity is 103 Pa s), then the isothermal ow
(with T = T) resulting from the action of surface tension begins when t0 = 111 s. The
ow is negligible until the viscosity sharply decreases when temperature reaches T Tw
(Doremus 1973). The value of the temperature T, corresponding to the working point
Tw (at which the viscosity equals 103 Pa s), typically ranges from 1800 to 2000 C (Paek
and Runk 1978, Oh 1979).
We can also take T higher than that corresponding to the working point Tw, in which
case the temperature will continue to increase after ow has begun. However, temperature remains homogeneous over a cross-section of the preform. The viscosities of the
glasses forming the cladding and outer matrix will change during such overheating. If,
however, the activation energies of viscous ow are identical in the cladding and outer
matrix (which is a realistic assumption for several pairs of glasses), U1 = U2, then
according to the Arrhenius law (Eq. 6.4), the viscosity ratio 1/2 is temperature
independent and thus, time independent an important fact that can be used below to
generalize the solution obtained for the case of the polishing method with overheating.
(The subscripts 1 and 2 denote the activation energies and viscosities of the cladding and
the outer matrix, respectively.)
282
282 [262--296] 10.12.2013 9:35PM
Note that even in the worst case of U1 U2 in an overheated preform, the situation may
be considered approximately as an isothermal one, since the characteristic time for
temperature eld saturation 1 R22 /T is small compared to the characteristic time of
ow development, 2 R2/2 (2 is the surface tension). Indeed, for R2 = 0.6 102 m,
T = 0.13 104 m2 s1, = 103 Pa s and 2 = 0.3 kg s2 (Paek and Runk 1978), we obtain
1/2 = R22/(T) = 0.138.
Let us now estimate the Reynolds number characteristic of the polishing method. To this
end we will prove that the ow is viscosity dominated. In the given surface-tensiondriven ow the characteristic velocity is of the order of 2/. Therefore, the Reynolds
number Re = 2R2/2 (which is also the inverse Ohnesorge number). To estimate the value
of Re we take, as above, = 2.2 103 kg m3, R2 = 0.6 102 m, = 103 Pa s, corresponding
to the working point, and 2 = 0.3 kg s2. As a result we obtain Re = 3.96 106. This
Reynolds number is much less than unity and will remain so even when the preform is
overheated (to prevent the onset of instabilities). The Reynolds number estimates the
ratio of inertial to viscous forces. In the given problem additionally the ratio of the inertial
term with time derivative in the NavierStokes equation to the viscous terms is also of the
order of Re, since the characteristic time of the ow is R2/2. Thus viscous forces
dominate inertial forces and the ow can be considered to be quasi-steady creeping ow
(with boundary conditions that obviously are functions of time).
The creeping ow under consideration is planar, and its solution does not depend on
the axial coordinate z. In polar coordinates r and (Figure 6.18), we obtain the inertialess
Stokes equations (Happel and Brenner 1991) in each of the domains 1 and 2 in the form
2

p
vr 1 vr 1 2 vr vr 2 v

0
6:21

r2
r
r r r2 2 r2 r2
2

v 1 v 1 2 v 2 vr v
1 p
0

2 2 2
r2
r
r r
r
r r2
6:22
rvr v
0; vz 0
6:23
where p is pressure, vr, v and vz are the components of the velocity vector, and is the
viscosity (different, in the general case, in domains 1 and 2 in Figure 6.18). At the
beginning we consider the case of a steady-state uniform temperature, which is high
enough for viscous ow of both glasses 1 and 2 in the cladding and outer matrix.
The solution of Eqs. (6.21)(6.23) must satisfy the no-slip conditions at the boundary
of the solid core 0, as well as the kinematic and dynamic conditions at the boundaries 1
and 2 presented in the form
ri ri Ri i Ri 1 i ; t; i 1; 2
6:24
where R1 and R2 are the initial and outer radii of the cladding and the unpolished outer
matrix, respectively, and i and i are the dimensional and dimensionless perturbations of
the boundaries compared to the corresponding circles, respectively.
283 [262--296] 10.12.2013 9:35PM
283
The above-mentioned no-slip, kinematic and dynamic conditions take the form
0 : vr1 0; v1 0 at r R0
6:25
1
v1 1
; vr1 vr2 ; v1 v2 ; nn1 nn2 q1 ; n1 n2

t 1 1
at r r1 R1 1
1: vr1
6:26
v2 2

; nn2 q2 ; n2 0
2 : vr2 1
t 1 2
at r r2 R2 2
6:27
Here and hereinafter the subscripts 1 and 2 denote quantities related to the cladding and
outer matrix, respectively. The capillary pressures q1 and q2 at the boundaries 1 and
2, in accordance with the Laplace equation, are equal to the products of the interfacial
(or surface) tension and the sum of the principal curvatures of the corresponding
boundary surface. The stresses in the liquid material are denoted by nn and n, the
subscript n referring to the normal to the boundary and to the tangent.

Introducing the stream function vr r1 =; v =r , we reduce Eqs.
(6.21) and (6.22) to the biharmonic equation for
2
4 2 3 1 2 1 4 2 2 3
2 4
1 4
0
r4 r r3 r2 r2 r3 r r4 2 r3 r2 r2 r2 2 r4 4
6:28
where is the two-dimensional Laplace operator.

The general analytical solution of the problem in the form of a Fourier series can be
found under the assumption that the perturbations of the boundaries compared to the
circles are small, i.e. i are assumed to be sufciently small compared to unity. The details
are available in the paper by Yarin (1995). Here we illustrate the solutions obtained. To
calculate particular cases, one has to employ data on the material properties of molten
glass, such as viscosity, surface and interfacial tension. Determination of surface and
interfacial tension of molten glass involves experimental difculties, owing to the
necessarily high temperature and high viscosity. The following facts are known from
the literature(e.g. Morey 1954, Scholze 1991): the surface tension of molten glass is
practically temperature independent (an increase of 100 K reduces surface tension by
about 13%); its variation due to the effect of added oxides (such as B2O3, which is
considered below) is typically small, of the order of several percent. To the best of our
knowledge, there is no direct measurement of interfacial tension between two molten
glasses. Accordingly, one can virtually neglect surface-tension gradients in nonisothermal problems, and expect that interfacial tension is small compared with surface tension.
The interfacial tension of molten glasses has been measured indirectly by Grigoryants
et al. (1989). The experiments were carried out with two-layer quartz optical bers with
borosilicate quartz glass in the cladding and pure quartz in the outer matrix. Fibers with
noncircular cladding formed by the polishing method were subjected to prolonged
284
284 [262--296] 10.12.2013 9:35PM
1
4
Figure 6.20
Two-layer preform production comparison with experiment. The dashed curves 1 and 2 show the
initial conguration of the boundaries of the cladding and outer matrix. The solid curve 3 shows the
predicted steady-state shape of the boundary of the cladding after rounding-off of the softened
preform under the action of surface tension. Curve 4 shows the nal shape of the boundary of the
cladding that was observed in the experiment and curve 5 shows the circumference of the outer
matrix of the preform after rounding-off, both in the theory and experiment. Experimental data of
Grigoryants et al. (1989): 1/ 2 = 0, 1/2 = 0.2 and = 2.27. Yarin (1995). Courtesy of
Cambridge University Press.
heating. No rounding-off of the boundary of the cladding was observed in this experiment, which shows that interfacial tension between borosilicate quartz glass and pure
quartz is, indeed, approximately zero. For this reason, the case 1/ 2 = 0 is considered
below as a basic one, with 1 being the interfacial tension and 2 the surface tension.
The solutions obtained by Yarin (1995) allow us, however, to treat the general cases, with
1/ 2 0, as well. The results of these calculations are also discussed below.
First, consider a two-layer preform, which corresponds to 0 = R0/R1 = 0. We compare
theoretical results with the experimental data of Grigoryants et al. (1989) corresponding
to the case 1/ 2 = 0, the viscosity ratio 1/2 = 0.2 and = R2/R1 = 2.27. The comparison
is presented in Figure 6.20 and shows that the predictions based on the analytical solution
of the linearized problem agree fairly well with the experimental data.
The satisfactory agreement of the predictions with the experimental data for large
initial perturbations of the circular outer boundary of the preform, as shown in
Figure 6.20, demonstrates that the analytical solution obtained is accurate far beyond a
linear approximation. Note also that the value 1/ 2 = 0 used in the calculation is in
agreement with the experiment on prolonged heating of the perform, discussed above.
The results shown in Figures 6.206.22 were obtained with 19 modes of the Fourier
series. In Figure 6.21 some additional representative predictions for two-layer performs
are shown. The corresponding values of the parameters are listed in Table 6.1.
In Figure 6.21 and Table 6.1 it can be seen that vefold and tenfold increases in the
viscosity ratio 1/2 at xed values of and 1/ 2 = 0 had only minor effects on the
results. In the case of = 4, the results (Figures 6.21d and 6.21e) show that the nal form
of the cladding becomes similar to that of a bow tie. In optoelectronic applications a bowtie form of the cladding boundary is preferable because it provides higher birefringence
(Yarin 1990, Bernat and Yarin 1992).
285 [262--296] 10.12.2013 9:35PM
285
Table 6.1 Values of the parameters corresponding to Figures 6.21a k.

Figure
1/2
Figure
1/2
6.21a
6.21b
6.21c
6.21d
6.21e
2
2
2
4
4
1
0.2
2
1
0.2
6.21f
6.21g
6.21h
6.21i
6.21j
6.21k
2
2
2
2
2
2
1
1
0.2
0.2
2
0.2
When the interfacial tension is negligible, 1/ 2 = 0, the cladding boundary asymptotically approaches a nal noncircular form (shown in Figures 6.20 and 6.21), whereas the
outer boundary of the outer matrix tends asymptotically to a circle under the action of
surface tension. Calculations were also performed with nonzero interfacial tension and
values of and 1/2 similar to those shown in Table 6.1. For 1/ 2 = 0.1, two
characteristic timescales of the process can be distinguished. The rst is based on the
surface tension, 2R2/2. On this timescale the ow development is almost identical to
that for 1/ 2 = 0. The most deformed shapes of the cladding boundaries are practically
indistinguishable from those shown in Figure 6.21, whereas the outer boundary of the
outer matrix acquires a circular shape. Such an intermediate asymptotic form persists for
rather a long time. However, on a timescale based on the interfacial tension, 2R2/1,
which is ten times longer than 2R2/2 in the given case, these intermediate asymptotic
forms of the cladding disappear, since the cladding should return to the trivial equilibrium
form with a circular boundary, minimizing the interfacial energy. This, indeed, takes
place, and the analytical solution obtained describes monotonic evolution of the cladding
forms of Figure 6.21 back to a circular shape, which the boundary of the cladding
eventually assumes as time increases. During this process, the outer boundary of the
outer matrix continues to be virtually circular.
Several calculations have been done at relatively large interfacial tension, 1/ 2 = 0.5,
and the values of and 1/2 shown in Table 6.1. At such a value of 1/ 2, the
intermediate quasi-steady asymptotic form of the cladding, described above, practically
disappears (since the timescales of 2R2/2 and 2R2/1 are close to each other), the
deformation of the inner (cladding) boundary relative to a circle being only very slight
throughout the rounding-off process of the outer boundary. Therefore, glasses with such
high interfacial tension are completely inappropriate for use with the polishing method.
All the results above show solutions of the direct problem when the initial conguration of the outer matrix obtained by polishing is given, in order to predict the nal
shape of the cladding boundary. The analytical solution of Yarin (1995), however, may
also be used to address the inverse problem: what should be the shape of the outer matrix
after polishing, to arrive at a given shape for the cladding boundary?
Bearing in mind that in optoelectronic applications a bow-tie shape of cladding is
preferable, we predict the initial shape of the outer matrix in a two-layer system, which
allows one to arrive at the nal conguration of the cladding, for example, in the form of
the ovals of Cassini
286 [262--296] 10.12.2013 9:35PM
2
1
2
1
(a)
(b)
2
2
1
(c)
(d)
2
2
2
(e)
(f)
2
(g)
(h)
2
2
1
1
2
2
1
1
(i)
2
2
1
1
(k)
(j)
Figure 6.21
Predicted shapes of two-layer preforms for various initial cross-sections of the outer matrix. The
dashed curves 1 and 2 show the congurations of the boundaries of the cladding and outer matrix at
the initial moment, and the solid curves 1 and 2 show these boundaries at the end of the process at
steady state. The parameters corresponding to panels (a)(k) are given in Table 6.1. Yarin (1995).
Courtesy of Cambridge University Press.
287 [262--296] 10.12.2013 9:35PM
x2 y2 2c2 x2 y2 a4 c4
287
6:29
Here the Cartesian coordinates

are those of Figure 6.18 and the constants a and c satisfy
p
the inequality c5a5c 2 (the only condition under which the ovals resemble a bow tie).
The area of oval A is equal to the area of an equivalent unperturbed circle, which yields
the radius of the latter:
2
2
1=2
c
c
c4
1 4 sin 2
d
6:30
A 2a2 E 2 R21 ; E 2
a
a
a
0
Knowing R1 from the rst Eq. (6.30), one can represent the oval (Eq. 6.29) as a
perturbation of a circle of radius R1 in the form of Eq. (6.24). As a result, the perturbation
of this circle is given by the expression
"
#1=2

1=2 i1=2
1h
2
1
1
6:31
cos 2 cos 2 2 4 1
2E
p
a
; 15 5 2
6:32
c
The results for the inverse problem are shown in Figure 6.22, corresponding to = 2,
1/2 = 0.2, 1/ 2 = 0 and = 1.1. The value of chosen corresponds to the required
cladding shape.
The boundary shape 2 in Figure 6.22 is a rather large perturbation of a circle. The
ellipticity of this curve is, however, close to 2, which ts the borderline of the range of
applicability of the linearized theory, as corroborated in Bernat and Yarin (1992) by
comparison of similar predictions with the results of direct numerical simulations.
2
2
1
Figure 6.22
Solution of the inverse problem. Predicted initial shape of the outer boundary (curve 2) leading to
the nal shape of the cladding boundary (curve 1) in the form of an oval of Cassini with = 1.1.
The dashed curves 1 and 2 show the initial congurations of the boundaries, and the solid curves 1
and 2 show the nal ones. Yarin (1995). Courtesy of Cambridge University Press.
288
288 [262--296] 10.12.2013 9:35PM
Therefore, it might be speculated that the result in Figure 6.22 represents a valid
approximation of a nonlinear solution.
Note that such inverse problems are characteristic of engineering (Shercliff 1981).
Computers are practically useless in solving such problems. Therefore, even an approximate analytical solution, like that of Figure 6.22, might be very instructive and helpful in
this case.
Consider now three-layer preforms similar to the sketch in Figure 6.18. An example of the
direct problem corresponding to 0 = R0/R1 = 0.2, = R2/R1 = 2, 1/2 = 0.2 and 1/ 2 = 0
is shown in Figure 6.23. The predicted nal shape of the cladding (curve 1), corresponding
to the initial polishing of the outer matrix, is close to that of an ellipse.
The effect of the solid core size on the nal shape of the cladding is illustrated in
Table 6.2. There we show the second Fourier coefcient of the cladding boundary b21f (=
b21 at t = ) as a function of 0 (this coefcient is related to the ellipticity). The coefcient
b21f (< 0) determines the birefringence of the ber (Yarin 1990, Bernat and Yarin 1992)
and its value (as well as the birefringence) decreases as the core radius increases. Indeed,
birefringence is given by
Table 6.2 The effect of the solid core size on the

nal shape of the cladding for three-layer performs.
0
jb2lf j
0.1
0.2
0.3
0.4
0.5
0.2823
0.2776
0.2683
0.2523
0.2260
2
1
Figure 6.23
2
1
Three-layer preform. The dashed curves 1 and 2 show the shapes of the boundaries of the cladding
and outer matrix at the initial moment. The solid curves 1 and 2 show these boundaries at the end
of the process, in the steady state. Circle 3 shows the core boundary. 0 = 0.2, = 2, 1/2 = 0.2 and
1/ 2 = 0. Yarin (1990, 1995), with kind permission from Springer Science+Business Media.
Courtesy of Cambridge University Press.
289 [262--296] 10.12.2013 9:35PM
B Cxx yy
CE
k2 k1 1 34 b21f
1
289
6:33
where C is the stress-optic coefcient, and E and denote the elastic parameters of solid
glass, Youngs modulus and Poissons ratio, respectively. The latter two parameters are
assumed to be the same in all three domains (core, cladding and outer matrix). The
thermoelastic parameters ki are given by
ki
1
i T T0i ; i 0; 1; 2
1
6:34
Here i denote the corresponding linear thermal expansion coefcients, different in

domains 0, 1 and 2 (core, cladding and outer matrix, respectively), and T is temperature,
which is supposed to be constant in the cross-section. Also, T0i are the temperature values,
which correspond to the absence of thermoelastic stresses. These temperatures differ in
different domains, which is the main reason for the thermoelastic stresses frozen in solid
bers.
As the greatest part of the energy in polarization-maintaining optical bers is transmitted through the core, the difference between the normal stresses xx and yy (see the
axes x and y in Figure 6.18) in the core characterizes the expected birefringence (see axes
x and y in Figure 6.18, and Eq. (6.33). The results for the mechanical anisotropy in the
core (xx yy) obtained using analytical theory and direct numerical simulations using
the nite element method from Yarin (1990) and Bernat and Yarin (1992) are shown in
Figure 6.24, where the following parameter values are used: E = 7830 kg mm2,
= 0.186, T T0i = 1000 C, 0 = 2 = 5 107 C1, 1 = 106 C1 and the core
radius R0 comprises 7% of the outer matrix radius R2. The gure shows that at higher
ellipticity (lower value of b/a), birefringence increases signicantly. The analytical
results are in excellent agreement with the numerical ones in the range 0.2 < b/a < 1.
Several ramications of the hydrodynamic solution for polishing method of two- and
three-layer preforms are worth mentioning. The rst of them is the surface-tension-driven
evolution of cross-sections of multilobal textile bers formed from a noncircular spinneret hole (Ziabicki 1976). This single-layer case corresponds to a degenerate three-layer
one when 0 = 0 (no core), 1 = 2 (equal viscosities of the cladding and outer matrix),
and 1 = 0 (zero interfacial tension). The perturbation of the outer surface relative to a
circle of radius R2 is given by the Fourier series
2
b02 t X
an2 t sin n bn2 t cos n
2
n1
6:35
with the Fourier coefcients given by
n 2; an2
b02 b020 ; a12 a120 ; b12 b120

2 n
2 n
an20 exp
t ; bn2 bn20 exp
t
22 R2
22 R2
6:36
6:37
290
290 [262--296] 10.12.2013 9:36PM
xx yy [kg mm2]
a = 0.3
a = 0.7
0
0.0
0.2
0.4
0.6
0.8
1.0
b/a
Figure 6.24
The mechanical anisotropy in the core (which is the birefringence divided by the stress-optic
coefcient) of a three-layer optical ber with elliptical cladding as a function of the aspect ratio b/a,
where a and b are the cladding semi-axes. The values of a in the panel are rendered dimensionless
by the radius of the outer matrix R2. The two sets of results correspond to two different values of a,
shown near the curves. The analytical results are shown by curves, the numerical ones by symbols.
Copyright 1991, IEEE. Reprinted, with permission, from Bernat and Yarin (1992).
This solution can be applied to any initial shape of cross-section, which is given by its
Fourier coefcients ani0 and bni0. Equations (6.36) and (6.37) show how the Fourier
coefcients will diminish in time under the action of surface tension. If the ber does not
solidify fast enough (not longer than the time 2R2/2), the effect of the noncircular
spinneret hole will be practically lost and the ber cross-section will be almost circular.
The second ramication is related to generalization to the case when rounding-off of
the polished outer matrix and the corresponding deformation of cladding are accompanied by simultaneous ber drawing in the direction z normal to the cross-section.
Consider, for example, a two-layer preform. We assume that along the bre axis z
there exists a velocity vz, such that vz =z Dzz t; vz =r vz = 0 and
vr =z v =z 0. Here Dzz(t) is a given stretching rate. Formally we consider the
case of a slow uniform stretching, for example by two clamps moving in opposite
directions when vr, v and the area of the preform cross-section are independent of z,
whereas vz is constant over the cross-section. In ber drawing by a winding bobbin from
a heated preform, all the above assumptions are applicable only when variation of the
parameters along a threadline is gradual, which is the case under certain conditions (Yarin
1993). Then, the evolution of the cross-sectional boundaries over time will be similar to
that predicted without stretching, except that the same congurations will correspond to
different times depending on Dzz(t), Yarin (1995). Therefore, the solutions shown in
Figures 6.206.22 also show the nal steady-state cross-sectional congurations of the
optical bers drawn from a polished preform. Polishing of a 1 cm preform is easy, while
291 [262--296] 10.12.2013 9:36PM
291
2
1
(a)
2
1
1
(d)
(b)
2
1
(e)
2
(c)
1
(f)
Figure 6.25
The time evolution of the collapse process at various dimensionless times for the surface tension
ratios 0 0 =2 1, 1 1 =2 0 and the viscosity ratio 1 1 =2 1 (curve 1:
deposited domain, curve 2: substrate tube). (a) t 0 (the initial shapes of the boundaries); (b)
t 0:3; (c) t 0:5; (d) t 0:7; (e) t 0:75; (f) t 0:827 (the nal state). Time t is rendered by
2R20/2. Yarin (1995), Courtesy of Cambridge University Press.
292 [262--296] 10.12.2013 9:36PM
292
Figure 6.26
SEM image of a preform cross-section after chemical etching. The image is rotated to have the
longer axis vertically for comparison with Figure 6.25f. The preform was formed using the
nonsymmetric modied vapor deposition method (N-MCVD). Copyright 1991, IEEE. Reprinted,
with permission, from Doupovec and Yarin (1991).
achieving a pre-determined noncircular cladding in a 100 m optical ber is nontrivial. A

nonuniform temperature eld along the threadline also results only in time re-calculation,
but does not affect the boundary congurations (Yarin 1995).
Consider now the results obtained for the nonsymmetric modied vapor deposition
method (N-MCVD) sketched in Figure 6.19. In this case the boundaries 1 and 2 in
Figure 6.19 may be taken approximately as circles, and the interface 0 as an ellipse with
a semi-axis ratio = a/b satisfying the inequality
1 1 = 2
6:38
Here * = R20/R10 and 1 = R20/R00, with R10 and R20 being the initial radii of the
boundaries 1 and 2, respectively, and R00 = (ab)1/2 is the radius of the circle, about
which the boundary 0 is perturbed.
The initial perturbations of the boundaries are thus given by
0 1 sin 2 1 cos 2
1=2
; 1 2 0
6:39
The rst expression (Eq. 6.39) describes an elliptical perturbation of a circle of radius
R00. In the calculations the following values of the geometrical parameters were used:
293 [262--296] 10.12.2013 9:36PM
293
2
2
(a)
Figure 6.27
(b)
The nal states of the collapse for various surface tension ratios 1 ; 0 1, 0 1, the initial
geometry is the same as in Figure 6.25a (curve 1: deposited domain, curve 2: substrate tube).
(a) 1 0:1; (b) 1 0:1. Yarin (1995). Courtesy of Cambridge University Press.
(a)
Figure 6.28
(b)
The nal states of the collapse for various surface tension ratios 1 ; 0 1, 1 0, the initial
geometry is the same as in Figure 6.25a (curve 1: deposited domain, curve 2: substrate tube).
(a) 1 0:1; (b) 1 10. Yarin (1995). Courtesy of Cambridge University Press.
1 = 1.457, * = 1.44, = (1/*)2 = 1.023, which means that the elliptical boundary 0 at
the major axis practically touches 1 (see Figure 6.25a).
The time evolution of the surface-tension-driven collapse process is shown in
Figure 6.25. The resulting bow-tie-like shape (Figure 6.25f) is in qualitative agreement
with the one found experimentally by Doupovec and Yarin (1991; Figure 6.26).
Unfortunately, in this experiment the exact parameters of the deposited layer are unknown,
which does not allow for a quantitative comparison of prediction with experiment.
294
294 [262--296] 10.12.2013 9:36PM
The nal shapes of the surface-tension-driven collapse for the same initial condition as
in Figure 6.25, but for different values of 1 and 1 are shown in Figures 6.27 and 6.28,
respectively. Variation in 1 , as well as in 1 does not change the fact that a bow-tie-like
shape of the cladding appears, as can be seen in these gures. Note, that the conguration
shown in Figure 6.27b is very similar to that observed experimentally, shown in
Figure 6.28.
6.10
References
Bejan, A., 1993. Heat Transfer. John Wiley & Sons, New York.
Bernat, V., Yarin, A. L., 1992. Analytical solution for stresses and material birefringence in optical
bers with noncircular cladding. J. Lightwave Technology 10, 413417.
Blades, H., White, J. R., 1963. Fibrillated strand. US Patent No. 3081519.
Canejo, J. P., Borges, J. P., Godinho, M. H., Brogueira, P., Teixeira, P. I. C., Terentjev, E. M., 2008.
Helical twisting of electrospun liquid crystalline cellulose micro- and nanobers. Advanced
Materials 20, 48214825.
Chhabra, R., Isele, O. E. A., 2007. Coated nanober webs. US Patent No. 7291300.
Ciferri, A., Ward, I. M., 1979. Ultra-high Modulus Polymers. Applied Science Publications, London.
Das, K., Gandhi, K. S., 1986. A model for thermal collapse of tubes: applications to optical glass
bers. Chem. Eng. Sci. 41, 7381.
Donald, A. M., Windle, A. H., Hanna, S. 2006. Liquid Crystalline Polymers. Cambridge University
Press, Cambridge.
Doremus, R. S., 1973. Glass Science. John Wiley & Sons, New York
Dosunmu, O. O., Chase, G. G., Kataphinan, W., Reneker, D. H., 2006. Electrospinning of polymer
nanobres from multiple jets on a porous tubular surface. Nanotechnology 17, 11231127.
Doupovec, J., Yarin, A. L., 1991. Nonsymmetrical modied chemical vapor deposition (NMCVD) process. J. Lightwave Technology 9, 695700.
Dror, Y, Ziv, T., Makarov, V., Wolf, H., Admon, A., Zussman E., 2008. Nanobers made of
globular proteins. Biomacromolecules 9, 27492754.
DuPont, 2013. Available at http://www2.dupont.com/Residential_Construction/en_US/science/
sci_const.html. Accessed August 4, 2013.
Durany, A., Anantharamaiah, N., Pourdeyhimi, B., 2009. High surface area nonwovens via
brillating spunbonded nonwovens comprising islands-in-the-sea bicomponent laments:
Structureprocessproperty relationships. J. Mater. Sci. 44, 59265934.
Fiberio, 2013. Available at http://beriotech.com/. Accessed August 4, 2013.
Geyling, F. T., Walker K. L., Csentits, R., 1983. The viscous collapse of thick-walled tubes.
Proceedings of the ASME Applied Mechanics, Bioengineering and Fluids Engineering
Conference, Houston, TX, Paper 83-APM-27.
Glicksman, L. R., 1968. The dynamics of a heated free jet of variable viscosity liquid at low
Reynolds numbers. Trans. ASME, J.Basic Eng., Series D 90, 343354.
Grigoryants, V. V., Entov, V. M., Ivanov, G. E., Chamorovskii, Y. K., Yarin, A. L., 1989.
Formation of two-layer preforms for optical bers with shaped cores. Soviet Physics Doklady
34, 368370.
Happel, J., Brenner, H., 1991. Low Reynolds Number Hydrodynamics. Kluwer, Dordrecht.
Hassan, M. A., Yeom, B. Y., Wilkie, A., Pourdeyhimi, B., Khan, S. A., 2013. Fabrication of nanober
meltblown membranes and their ltration properties. J. Membrane Science 427, 336344.
295 [262--296] 10.12.2013 9:36PM
6.10 References
295
Huynh, B. P., Tanner, R. I., 1983. Study of the non-isothermal glass bre drawing. Rheol. Acta 22,
482499.
Kaminow, I. P., Ramaswamy, V., 1979. Single polarization optical bers and methods of fabrication. US Patent No. 4179189.
Khansari, S., Sinha-Ray, S., Yarin, A. L., Pourdeyhimi, B., 2012. Stress-strain dependence for soyprotein nanober mats. J. Appl. Phys. 111, 044906.
Kowalczyk, T., Nowicka, A., Elbaum, D., Kowalewski, T. A., 2008. Electrospinning of bovine
serum albumin. Optimization and the use for production of biosensors. Biomacromolecules 9,
20872090.
Krause, S., Dersch, R., Wendorff, J. H., Finkelmann, H., 2007. Photocrosslinkable liquid crystal
main-chain polymers: thin lms and electrospinning. Macromol. Rapid Comm. 28, 20622068.
Lewis, J. A., 1977. The collapse of a viscous tube. J. Fluid Mech. 81, 129135.
Luo, C. J., Stoyanov, S. D., Stride, E., Pelan, E., Edirisinghe, M., 2012. Electrospinning versus
bre production methods: from specics to technological convergence. Chem. Soc. Rev. 41,
47084735.
MacDiarmid, A. G., 2002. Synthetic metals: A novel role for organic polymers (Nobel lecture).
Angew. Chem. Int. Ed. 40, 25812590. Also in Synth. Metals 125, 1122.
MacDiarmid, A. G, Jones, Jr., W. E., Norris, I. D., Gao, J., Johnson, Jr., A. T., Pinto, N. J., Hone, J.,
Han, B., Ko, F. K., Okuzaki, H., Llaguno, M., 2001. Electrostatically- generated nanobers of
electronic polymers. Synth. Metals 119, 2730.
Medeiros Araujo, T., Sinha-Ray, S., Pegoretti, A., Yarin, A. L., 2013. Electrospinning of blend of
liquid crystalline polymer with poly(ethylene oxide): vectran nanober mats and their mechanical properties. J. Mater. Chem. C 1, 351358.
Morey, G. W., 1954. The Properties of Glass. Reinhold Publishers, New York.
Nakajima, T. (Editor), 2000. Advanced Fiber Spinning Technology. Woodhead Publishing Ltd.,
Cambridge.
Nakashima, K., Tsuboi, K., Matsumoto, H., Ishige, R., Tokita, M., Watanabe, J., Tanioka, A., 2010.
Control over internal structure of liquid crystal polymer nanobers by electrospinning.
Macromol. Rapid Comm. 31, 16411645.
Norris, I. D., Shaker, M. M., Ko, F. K., MacDiarmid, A. G., 2000. Electrostatic fabrication of
ultrane conducting bers: polyaniline/polyethylene oxide blends. Synth. Metals 114, 109114.
Oh, S. M., 1979. Cooling rates of optical bers during drawing. Ceramic Bull. 58, 11081110.
Paek, U. C., Runk, R. B., 1978. Physical behavior of the neck-down region during furnace drawing
of silica bers. J. Appl. Phys. 49, 44174422.
Pegoretti, A., Traina, M., 2009. Liquid crystalline organic bres and their mechanical behavior. In
Handbook of Tensile Properties of Textile and Technical Fibres. (Ed. A. R. Bunsell), Woodhead
Publishing Ltd, Cambridge, UK, 354436.
Pourdeyhimi, B, Chappas, W. J., 2008. High surface area ber and textiles made from the same. US
Patent Publication 2008/0108265.
Pourdeyhimi, B, Chappas, W. J., 2012. High surface area ber and textiles made from the same. US
Patent No. 8129019.
Pourdeyhimi, B, Chappas, W. J., 2013. Composite lter media with high surface area bers. US
Patent No. 8410006.
Pourdeyhimi, B., Fedorova, N., Sharp, S., 2008. High strength, durable micro and nano-ber fabrics
by brillating bicomponent islands in the sea bers. US Patent Publication 2008/0108265.
Reneker, D. H., Chun, I., Ertley, D., 2002. Process and apparatus for the production of nanobers.
US Patent No. 6382526.
296
296 [262--296] 10.12.2013 9:36PM
23872425.
Rosner, D. E., 2000. Transport Processes in Chemically Reacting Flow Systems. Dover
Publications, New York.
Sarkar, K., Gomez, C., Zambrano, S., Ramirez, M., de Hoyos, E., Vasquez, H., Lozano, K., 2010.
Electrospinning to ForcespinningTM. Mater. Today 13, 1214.
Scholze, H., 1991. Glass: Nature, Structure and Properties. Springer, New York.
Shercliff, J. A., 1981. Reections of a new editor. J. Fluid Mech. 106, 349356.
Sinha-Ray, S., Zhang, Y., Yarin, A. L., Davis, S. C., Pourdeyhimi, B., 2011. Solution blowing of
Srinivasan, G., 1994. Structure and Morphology of Electrospun Polymer Fibers. PhD thesis.
Department of Polymer Science, The University of Akron.
Srinivasan, G., Reneker, D. H., 1995. Sructure and morphology of small diameter electrospun
Aramid bers. Polym. Int. 36, 195201.
Taylor, G. I., 1923. Stability of a viscous liquid contained between two rotating cylinders. Phil.
Trans. R. Soc. A 223, 289343.
Textile World, 2011. Available at http://www.textileworld.com/Articles/2011/November/
PGI_Debuts_Arium_Technology.html. Accessed August 4, 2013.
Torobin, L., Findlow, R. C., 2001. Method and apparatus for producing high efciency brous
media incorporating discontinuous sub-micron diameter bers, and web media formed thereby.
US Patent No. 6183670.
Velev, O. D., Smoukov, S., Geisen, P., Wright, M., Gangwal, S., 2011. A continuous process for
nanober fabrication based on shear and antisolvent-based polymer precipitation. Invention
Disclosure, North Carolina State University.
Yarin, A. L., 1982. Stationary congurations of bres formed under nonisothermal conditions.
Applied Mechanics and Technical Physics 23, 865870.
Yarin, A. L., 1990. Hydrodynamic analysis of the process of making three-layer optical bers and
calculation of the eld of elastic stresses and birefringence. J. Applied Mechanics and Technical
Physics 31, 361367.
Yarin, A. L., 1995. Surface-tension-driven low Reynolds number ows arising in optoelectronic
technology. J. Fluid Mech. 286, 173200.
Yarin, A. L., Bernat, V., Doupovec, J., Miklos, P., 1993. The viscous collapse of radial nonsymmetric composite tubes. J. Lightwave Technology 11, 198204.
Yarin, A. L., Rusinov, V., Gospodinov, P. Radev, S., 1989. Quasi one-dimensional model of
drawing of glass microcapillaries and approximate solutions. Theor. Appl. Mech. (Bulg. Acad.
Sci.) 20, 5562.
Zhang, Y., Yarin, A. L., 2011. Carbon nanobers decorated with poly(furfuryl alcohol)- derived
carbon nanoparticles and tetraethylorthosilicate-derived silica nanoparticles. Langmuir 27,
1462714631.
Zwijnenburg, A., Pennings, A. J., 1976. Longitudinal growth of polymer crystals from owing
solutions. Colloid Polymer Sci. 254, 868881.
297 [297--318] 10.12.2013 9:39PM
Tensile properties of nanofibers
This chapter is devoted to the tensile properties of individual nanobers and nanober
mats. The experimental results for the former are discussed in Section 7.1, and for the
latter in Section 7.2. Phenomenological theory describing the stressstrain dependence of
nanober mats is outlined in Section 7.3, and the corresponding micromechanical theory
is described in Section 7.4.
7.1
Tensile tests on individual nanofibers

To perform tensile tests on Nylon 6,6 (PA66) (Figure 7.1), Zussman et al. (2006) collected
individual nanobers using the electrostatic lens (the sharpened wheel collector) described
in Section 5.9 of Chapter 5 (see Figure 5.54). Then, one end of an individual ber was
mounted on the tip of an AFM cantilever (that served as a force-sensing element) and the
other end on the etched tip of a stainless steel wire (that served as a pulling element). The
nanober ends were secured at these positions with epoxy glue.
The cantilevers were considered as EulerBernoulli beams of rectangular crosssection, with b and h being the width and thickness, and the length L being more than
5b. Since in the present case the cantilever deection angles are always smaller than 10,
the general solution for a clamped cantilever transversally loaded at the free end (tip)
reduces to (Landau and Lifshitz 1970)
P Ke
7:1
where P is the applied force acting perpendicular to the cantilever at the tip, is the
resulting tip displacement and the elastic constant Ke = 3EI/L3. In this expression, I is the
cross-sectional moment of inertia and E is Youngs modulus. The rst fundamental
eigenfrequency of the cantilever is given by (Timoshenko 1961)
1:8752 EI=S1=2 =L2
7:2
where is the cantilever densit and S = bh the cross-sectional area.

Using Eq. (7.2), the expression for Ke becomes
Ke 0:2472bhL2
where 3/1.7754 = 0.2427.
7:3
298 [297--318] 10.12.2013 9:39PM
298
7 Tensile properties of micro- and nanofibers
Figure 7.1
SEM micrograph of electrospun PA66 nanobers. Zussman et al. (2006). Courtesy of John
Wiley & Sons.
Each cantilevers rst eigenfrequency, , was measured using a vibrometer, and then
Ke was calculated as per Eq. (7.3). Note that the value of could also have been
calculated using the expression for the moment of inertia I = bh3/12 and a value of E,
as well as Eq. (7.2). However, as E is not known with a high degree of accuracy, better
results can be achieved using direct measurement of , as was done by Zussman et al.
(2006).
For the tensile test, the cantilever end was loaded normally to its neutral axis at the tip
via the attached nanober pulled by a wire at a linear velocity of 4 m s1. The other end
of the cantilever was clamped. The deection, , of the AFM cantilever was observed
using an Olympus BX51 microscope with 500 and 1000 magnications (longdistance focal length objective), equipped with a CCD camera. A series of images
taken during a tensile test is shown in Figure 7.2.
Figure 7.3 shows a typical stressstrain curve (until failure) of the individual electrospun nanobers collected at 20 m s1 and at 5 m s1, along with a commercial microber
(Nilit, P55; diameter 38 m) that was prepared by melt spinning followed by cold
drawing. The average values of Youngs modulus, maximal stress and elongation at
break of the electrospun nanobers are presented in Table 7.1. The results represent 10
experiments at a take-up wheel velocity of 5 m s1 and 22 experiments at 20 m s1.
At maximal tensile stress, the nanobers were broken far from the clamped ends;
hence, the stress concentration at the nanober ends was excluded. The measured
Youngs moduli and maximal stresses of the electrospun nanobers have broad distributions, even though the bers were processed under similar conditions. The distributions
may be partially attributed to experimental inaccuracies; however, the main contribution
is probably due to an increase in the heterogeneities of the nanostructures. The nanobers
299 [297--318] 10.12.2013 9:39PM
299
7.1 Tensile tests on individual nanofibers
Table 7.1 Youngs modulus, maximal stress and elongation at break of electrospun nanobers.
Fiber take-up
velocity [m s1]
Ave. diameter
[nm] (SD)
Ave. E
[MPa] (SD)
Ave. xx,max
[MPa] (SD)
Ave. f
[%] (SD)
5
20
570 (40)
550 (120)
453 (150)
950 (390)
110 (14)
150 (49)
66 (15)
61 (19)
Nanofiber
Stainless-steel
wire
AFM
cantilever
50 m
Figure 7.2
50 m
50 m
A series of images taken during a tensile test up to break of an electrospun PA66 nanober. The
ber was attached to the tip of an AFM cantilever and pulled transversally by a stainless steel wire.
The deection of the cantilever, , is proportional to the applied force and was measured throughout
the tensile test. Zussman et al. (2006). Courtesy of John Wiley & Sons.
collected at 5 m s1 revealed an initial plateau at ~20 MPa (see Figure 7.3), which is
consistent with alignment of the macromolecules in the amorphous regions during the
initial stage of stretching. Above 20 MPa, the load is also supported by the crystalline
phase and the stress increases signicantly. Similarly shaped stressstrain curves were
observed in most of the experiments when nanobers were collected at 5 m s1. However,
nanobers collected at 20 m s1 did not reveal an initial plateau, apparently due to prestretching by the wheel during collection. When observing nanobers using SEM after
the tensile tests, local failures were not observed aside from a ductile fracture (see
Figure 7.4). The average elongation at break was found to be f = 61% and 66% for
nanobers collected at 20 m s1 and 5 m s1, respectively, and reects the high ductility of
these bers.
For comparison, Zussman et al. (2006) analyzed microbers formed from PA66
(which is identical to the material used in the electrospinning process) with a diameter
of 38 m (SD = 0.5 m), prepared by melt spinning followed by cold drawing. The
average maximum tensile stress of these bers was xx,max = 590 MPa (SD = 55 MPa),
with an average elongation at break of about 50%, and average Youngs modulus
E = 1.2 GPa (SD = 0.13 GPa); see Figure 7.3. It should be emphasized that the
mechanical properties of the electrospun nanobers in Zussman et al. (2006) were
300
300 [297--318] 10.12.2013 9:39PM
Electropsun (take-up velocity 20 m s1)

Electropsun (take-up velocity 5 m s1)
Melt spun + cold drawn
600
550
500
450
Stress [MPa]
400
350
300
250
200
150
100
50
0
0
10
20
30
40
50
60
70
Strain %
Figure 7.3
Typical stressstrain curves of individual PA66 nanobers collected at take-up velocities of 5 m s1

and 20 m s1, and a commercial microber that was prepared by melt spinning followed by cold
drawing. Zussman et al. (2006). Courtesy of John Wiley & Sons.
Figure 7.4
Fracture surface of a PA66 electrospun nanober. Zussman et al. (2006). Courtesy of John
Wiley & Sons.
301 [297--318] 10.12.2013 9:39PM
301
7.1 Tensile tests on individual nanofibers
nm
L1
150
100
50
0
50
100
1.6
Trench (width 1m)
0.8
1.2
L2
Nylon 6 fiber
2
0.4
1.6
1.2
0.8
Korea
WD15.0mm
0.4
15.0kv x70k 500nm
SE
(a)
(b)
(nm) 100
DH = 66.1 nm
L1
75
DW = 195 nm
50
25
0
25
0.0
0.5
1.0
1.5
2.0 (m)
0.5
1.0
1.5
2.0 (m)
(nm) 0
10
20
30
L2
40
0.0
(c)
Figure 7.5
(a) An individual nanober deposited across a trench for measurement of Youngs modulus. (b)
AFM landscape of the ber and trench. (c) Fiber and trench proles. Sinha-Ray et al. (2013).
measured right after the electrospinning process, without any post-processing. Hence, the
comparison of the mechanical properties of post-processed microbers with those of asspun nanobers is rather problematic. From a utilitarian perspective, as-spun polyamide
6,6 nanobers can be employed for various applications in tissue engineering, where the
nanometer-range size is an advantage, while the mechanical properties are comparable to
those of live tissues.
Another, albeit related, method of measuring tensile properties of individual nanobers was employed by Sinha-Ray et al. (2013) to measure Youngs modulus of supersonically solution-blown 2050 nm nylon 6 bers (cf. Section 2.6 in Chapter 2 and
Section 4.8 in Chapter 4). Individual solution-blown 50 nm nylon 6 nanobers were
deposited across a trench, as shown in Figure 7.5. The tip of an atomic force microscope,
AFM (XE-100, Park Systems, Rtip=10 nm) was pressed onto an individual ber with
force P and the corresponding deection was measured. The results of the nanoindentation experiment were processed using formulae following from EulerBernoulli
bar bending theory (Landau and Lifshitz 1970) in the limit of small deections
302
302 [297--318] 10.12.2013 9:40PM
PL3
PL3
; E
192I
48I
7:4
where I is the cross-sectional moment of inertia and L is the length of the suspended ber.
The rst Eq. (7.4) implies clamp conditions on both sides, the second, zero-moment
conditions. The stronger O. . .H bonding in solution-blown 50 nm nylon 6 manifests
itself in unusually strong nanobers. The values of Youngs modulus E measured using
nanoindentation were in the range 11.5146.04 GPa. These values of E are signicantly
higher than E = 23.5 GPa, found in ordinary macroscopic post-treated nylon 6 bers.
Some other measurements of mechanical properties of individual electrospun nanobers were reported by Tan and Lim (2004), Gu et al. (2005), Inai et al. (2005), Kim et al.
(2005b), Tan et al. (2005) and Pai et al. (2011a). Note that Tan et al. (2005) and Pai et al.
(2011a) respectively reported increasing tensile strength of microbers and nanobers
with decreasing diameter. In addition, Arinstein et al. (2007) attributed an abrupt increase
in Youngs modulus of electrospun nanobers as their diameter decreases to the connement effect on the supramolecular structure, albeit the Grifth mechanism relating
stronger small bers to smaller surface defects also could not be discarded (Grifth
1920, Wendorff et al. 2012).
In addition, note the report of Papkov et al. (2013) that stiffness of electrospun
nanobers increases when their cross-sectional diameter decreases below 150 nm.
These authors studied polyacrylonitrile nanobers and attributed their stiffness to
low crystallinity, which results from rapid solvent evaporation and solidication. In
the case of even smaller supersonically solution-blown nylon 6 nanobers (SinhaRay et al. 2013) discussed above, this factor is also present, albeit the bers are not
amorphous and contain a new phase (see Section 2.6 in Chapter 2), which could be
responsible for the enhanced stiffness. Also, formation of the extended-chain
crystals in electrospun poly(ethylene oxide) nanobers was reported in Zussman
et al. (2003), which shows that strong uniaxial elongational ows of polymer
solutions accompanying nanober formation are capable of producing new crystalline phases.
For electrospun polyacrylonitrile (PAN) nanobers with embedded single-wall carbon
nanotubes, SWCNTs, the forcedisplacement curves obtained by indentation of the AFM
tip revealed values of Youngs modulus in the range 60140 GPa at 0 to 4 wt% of
SWCNTs in PAN nanobers with diameters in the range 50 to 200 nm (Ko et al. 2003).
The high values of Youngs modulus measured in this work can be attributed to SWCNTs
and their strong binding to the PAN matrix. It should be emphasized that such nanobers
represent nanocomposites reinforced with SWCNTs. On the other hand, nanobers can
be used as reinforcing elements in macroscopic composites. An interfacial toughening
technique based on entangling small bers at interfaces (interlayers) is specically
suitable for toughening ber-reinforced polymer composites (Dzenis 2004, 2008).
These toughening small bers can be compliant continuous electrospun or solutionblown coreshell nanobers with healing agents in the core, which polymerize when
such material is damaged (Sinha-Ray et al. 2012).
303 [297--318] 10.12.2013 9:40PM
7.2 Tensile tests on nanofiber mats
7.2
303
Tensile tests on nanofiber mats

Tensile tests with nanober mats electrospun from blends of poly(vinyl chloride) and
polyurethane, PVC/PU, yield values of Youngs modulus up to 11.8 MPa, yield stress of
1.03 MPa, ultimate tensile stress of 3.73 MPa and elongation at break of 456% for 50/50
blends (Lee et al. 2003). For blends with a higher content of PU, a fully elastic response
was recorded up to an ultimate strength of 7.04 MPa, and elongation at break of 1210%
for pure PU mats. Youngs modulus of the PU mats was, however, much lower, namely
0.62 MPa (Lee et al. 2003). For electrospun PU mats an ultimate tensile stress of the order
of 40 MPa at an elongation of about 700% was reported (Pedicini and Farris 2003a). For
electrospun mats of poly(trimethylene terephthalate), PTT, with nanober diameters in
the range 200 to 600 nm, the ultimate tensile stress was about 4 MPa at an elongation of
about 300% (Khil et al. 2004). For mats of electrospun gelatin nanobers with diameters
in the range 100 to 340 nm, the tensile modulus was of the order of 117 to 134 MPa and
the ultimate tensile strength was in the range 2.93 to 3.40 MPa (Huang et al. 2004). Pai
et al. (2011b) measured a tensile strength of about 4 MPa for electrospun poly(trimethyl
hexamethylene terephthalamide) bers with diameters in the range 288 nm1.4 m.
Tensile tests of solution-blown biopolymer (soy protein) containing nanober mats
were conducted by Khansari et al. (2012). A typical stressstrain dependence measured
for soy protein/nylon 6 nanober mats is shown in Figure 7.6. It can be seen that at
relatively small strains , the tensile stress xx depends on practically demonstrating a
0.8
Rupture point
Plastic zone
0.2
Failure zone
0.4
Elastic zone
xx (MPa)
0.6
0.0
0
(%)
Figure 7.6
Tensile stress versus strain for a sample of soy protein/nylon 6 solution-blown nanober mat.
Sample rupture occurs at xx,rupture = 0.7 MPa and rupture = 4.5%. Reprinted with permission from
Khansari et al. (2012). Copyright 2012, AIP Publishing LLC.
304
304 [297--318] 10.12.2013 9:40PM
Table 7.2 Average mechanical properties of soy protein/nylon 6 nanober mats.
Figure 7.7
Ave. width
(mm)
Ave. thickness
(mm)
Ave. E
(MPa)
Ave. Y
(MPa)
Ave. u
(MPa)
Ave.
rupture(%)
Ave.
xx,rupture
12.07
0.20
19.56 6.48
0.56 0.15
2.26 0.71
4.52 0.92
0.67 0.10
Typical sample rupture pattern for soy protein/nylon 6 nanober mat. Reprinted with permission
from Khansari et al. (2012). Copyright 2012, AIP Publishing LLC.
linear elastic Hookean response. At higher strains, > 3%, the response becomes nonlinear, which can be attributed to the onset of plasticity. At 4.5% the sample failed.
The morphology of sample failure corresponding to Figure 7.6, is illustrated in
Figure 7.7. In most cases, samples failed in the middle (see Figure 7.7). Typically, the
failure stress and strain were in the range xx,rupture = 0.40.9 MPa and rupture = 410%,
respectively.
The average values of Youngs modulus E and the yield stress Y found for several
samples of soy protein/nylon 6 nanober mats are listed in Table 7.2. These parameter
values are established by tting the phenomenological theory discussed in some detail in
Section 7.3 to the experimental data. Table 7.2 also contains the specic strain energy,
dened as u 0 xx d.
7.3
Phenomenological model of stressstrain dependence of nanofiber mats

Phenomenological models that span the elastic and plastic ranges of stress response of
solids to deformation date back to the seminal works of Prager (1939, 1942), Truesdell
305 [297--318] 10.12.2013 9:40PM
7.3 Phenomenological model
305
(1952, 1953) and Green (1956). They bridge a Hookean elastic body to an ideally plastic
body that ows with a constant yield stress as soon as the von Mises condition is fullled.
Different terms were applied to such materials, e.g. alternatively, hypo-elasticity or plasticity. More recently, materials of this type, with deviations from Hookean linear behavior
and the associated irreversibility of deformation were understood as elastic-viscoplastic and
close counterparts of viscoelastic polymeric liquids (Rubin and Yarin 1993, 1995).
Following the latter work, the phenomenological rheological constitutive equation prone
to behavior reminiscent of that of meltblown nonwovens (see below) is taken in the form

d
2
2 2 1
rv rvT : DI 2D
I
7:5
dt
3
3
where d/dt denotes the material time differentiation, denotes the deviatoric stress tensor,
rv is the velocity gradient tensor and D its symmetric part (the rate-of-strain tensor), I is
the unit tensor, is the Lame coefcient responsible for the elastic behavior (in the case
of an incompressible body assumed here, = E/3, with E being Youngs modulus) and
: D denotes the scalar product of two tensors. The dimensionless factor 2 is included in
Eq. (7.5) for correspondence with Greens version of the plastic rheological constitutive
equation (Green 1956).
In uniaxial stretching of an axisymmetric specimen or a strip in the x-direction, the
quantity in Eq. (7.5) is determined as with = const. being the rate of strain.
Then, integrating Eq. (7.5), for the longitudinal deviatoric stress we obtain
xx
2
tanh
7:6
with t being the strain.

Since : I = 0, for an axisymmetric specimen, one nds that the lateral deviatoric
stresses yy = zz = xx/2. Then, the longitudinal stress xx = xx yy is equal to
xx
3
tanh
7:7
As tends to innity, xx = Y with Y being the yield stress. Then, accounting for the fact
that 3 = E, one nds that = E/Y, and Eq. (7.7) reduces to the following expression
established by Green (1956):

E
xx Ytanh
7:8
Y
which obviously recovers Hookes law xx= E, as tends to zero.
For uniaxial stretching of a planar strip similar to that depicted in Figure 7.7, integrating Eq. (7.5) and accounting for the fact that >> 1, we obtain
r
r !
8
2E
xx
Ytanh
7:9
3
3Y
306
306 [297--318] 10.12.2013 9:40PM
0.8
0.8
0.7
0.7
1
0.6
0.5
xx (MPa)
xx (MPa)
0.6
0.4
0.3
0.2
0.5
0.4
0.3
0.2
0.1
0.1
0.0
(a)
1
2
0.0
0
Figure 7.8
(%)
(b)
3
4
(%)
Comparison of phenomenological elastic-plastic model (Eq. 7.9), with experimental stress-strain

data for a soy protein/nylon 6 nanober mat. In panel (a), Eq. (7.9) is tted to the experimental
data up to the rupture point. Panel (b) shows the overall stressstrain data corresponding to panel
(a). Black symbols (1) depict the experimental data, lines (2) the phenomenological model,
Eq. (7.9). Reprinted with permission from Khansari et al. (2012). Copyright 2012, AIP
Publishing LLC.
which obviously recovers Hookes law for this case, xx = (4/3)E as tends to zero.
Equation (7.9) was tted to the experimental data for stressstrain curves in uniaxial
stretching of nonwoven strips blown from a soy protein/nylon 6 blend, and the resulting
values of Youngs modulus E and yield stress Y are listed in Table 7.2. Note that Eq. (7.9)
is capable of reproducing only the elastic and plastic parts of stressstrain curves similar
to that shown in Figure 7.6, rather than the descending part corresponding to the failure
zone. This is illustrated in Figure 7.8. In addition, comparisons of the phenomenological
model with the results of tensile tests of meltblown mats are discussed in Section 7.4.
7.4
Micromechanical model of stressstrain dependence of nanofiber mats

A micromechanical model of the tensile strength of nanober mats was proposed in
Khansari et al. (2012). In this section it is discussed in detail and compared to the
experimental data. Consider the orientational probability density function for(,t) in
nanober mats subjected to uniaxial stretching, as in the experiments of Section 7.2,
with being the angle relative to the direction perpendicular to the stretching direction,
and t being time. It assumes that ber segments cross any cross-section normal to the
stretching direction with certain inclinations (cf. Figure 4.29 in Section 4.7 of
Chapter 4), and the corresponding probability density function for(,t) varies with time
as stretching goes on. At the moment when stretching has begun, t = 0 and f0() is the
probability density function of the meltblown lay-down, as predicted from the numerical
model described in Section 4.7 in Chapter 4. The model fully determines angles in the
307 [297--318] 10.12.2013 9:40PM
7.4 Micromechanical model
307
original lay-down for all ber segments that intersect a given cross-section normal to the
stretching direction. This fully determines f0() at t = 0 when stretching of the mat begins.
The probability density function at any moment during sample stretching for(,t) can
be found from the following FokkerPlanck equation
f or
f or sin 2
t
7:10
where the stretching rate is assumed to be constant.

The solution of Eq. (7.10) satisfying the initial condition reads
f or
exp2
exp4sin2
2 cos 2
7:11
where t is the strain (Khansari et al. 2012).

It is easy to see that Eq. (7.11) automatically satises the normalization condition
2
f or d 1
7:12
0
The tensile strength of different bulk materials *, including individual nanobers, is

affected by many factors that are not under control, and therefore can be treated as a
mathematical expectation of many scattered values that might be measured in repeated
experiments (Yarin 2008, Khansari et al. 2012). Consider a material with n potential
defects per unit volume that might be responsible for a local rupture. These potential
defects can be opened due to stretching in any direction if an appropriate local stress
arises. We can treat these defects as potential rupture surfaces, with their banks being
initially conglutinated (see Figure 7.9). A defect results in rupture when its banks are
pulled apart by an appropriate stress, normal to the surface. Each defect, in fact,
represents multiple conglutinated surfaces radiating spherically and symmetrically.
X2
s22
s22
O
X1
X3
Figure 7.9
Sketch of a defect and its rupture. A closed defect with conglutinated banks is depicted on the left.
A defect ruptured by stresses in the x2 direction is depicted on the right. Reprinted with permission
from Khansari et al. (2012). Copyright 2012, AIP Publishing LLC.
308
308 [297--318] 10.12.2013 9:40PM
Each of these surfaces could be opened by an appropriate normal stress. As explained

before, the defect-opening process can be considered as random.
The probability density function of a defect to be opened by a normal stress 11 (related
to stretching along an axis Ox1 (see Figure 7.9), whereas the conglutinated surface is
normal to this axis) is denoted by F(11), and the probability of the defect to be ruptured
by a stress from the interval [11, 11+d11] is p1= F(11)d11. Rupture in different
directions is considered to be mutually independent. Therefore, the number of ruptured
defects in a unit volume subjected to stretching, for example, along three normal axes
Ox1, Ox2 and Ox3 is
dN nF11 d11 F22 d22 F33 d33
7:13
This number is related to the joint probability density function f(11, 22, 33), so that
dN nF11 F22 F33 d11 d22 d33 f 11 ; 22 ; 33 d11 d22 d33
7:14
Therefore, the number of defects in a unit volume that will not be ruptured at all in such
three-axial stretching by stresses 1, 2 and 3 is given by
N0 n F11 d11 F22 d22 F33 d33 n1 2 3
2
3
7:15
1
f 11 ; 22 ; 33 d11 d22 d33 1 ; 2 ; 3
3 2 1
Accordingly, the number of defects ruptured in a unit volume in this case is N = n N0.
The choice of a coordinate system is arbitrary, and the same rupture process can be
described using an arbitrarily oriented Cartesian coordinate frame Ox, Oy and Oz. Then,
the number of intact defects is equal to N0 = (xx, xy, xz, yy, yz, zz), where xx, etc.
denote the corresponding components of the stress tensor . Consider now the previously
used Cartesian axes Ox1, Ox2, and Ox3 as the principal axes of the stress tensor , with
1, 2, and 3 becoming, correspondingly, the principal stresses. The number of intact or
ruptured defects should not depend on the directions of the arbitrarily chosen axes Ox,
Oy and Oz, which means that the function should depend only on the three invariants
of the stress tensor
I1 xx yy zz 1 2 3
7:16
I2 xx yy yy zz xx zz 2xy 2yz 2xz 1 2 2 3 1 3
7:17
I3 xx yy zz 2xy yz xz xx 2yz yy 2xz zz 2xy 1 2 3
7:18
Equations (7.15)(7.18) result in the following functional equation:

n1 2 3 1 2 3 ; 1 2 2 3 1 3 ; 1 2 3
Its solution reads
7:19
309 [297--318] 10.12.2013 9:40PM
i A Bi expCi ; i 1; 2; 3
309
7:20
with A, B and C being constants.

When no stress is applied, the number of intact defects in a unit volume N0 = n. Then,
Eqs. (7.19) and (7.20) yield N0 = nA3, and therefore A = 1.
From Eq. (7.20), one nds that
i Fii dii 1 Bi expCi

7:21
i
which results in
Fii BCii C B expCii
7:22
Since materials have a certain strength, F(0) = 0, and thus C = B, which yields
Fii B2 ii expBii
7:23
It is easy to see that Eq. (7.23) satises the normalization condition.

The macroscopic measurable mathematical expectation of the defect-opening stress,
*, is identical to the tensile strength of the material, which is considered to be a known
parameter. Therefore, using Eq. (7.23), we obtain

7:24
ii Fii dii B2 2ii expBii dii
0
which yields B = 2/*. Therefore, the probability density function of rupture under
stretching in the i-th direction, ii, is given by
Fii
4
ii exp2ii =
2
7:25
Assume that all individual nanobers behave as Hookean elastic solids until they rupture.
We aim to show below that the plasticity of a macroscopic nanober mat can result from
rupture of individual nanobers in the mat under uniaxial stretching.
Fibers in the mat experience different stretching from the overall macroscopic
axial stretching imposed on the sample . Indeed, for an inclined ber the strain i is
given by
i sin 2
7:26
According to Eq. (7.25), if an initially unloaded ber was stretched to a certain stress ,
its probability of staying intact, Pintact, is

2
2
exp
7:27
Pintact 1

Then, the longitudinal stress in the mat is given by the following expression:
310
310 [297--318] 10.12.2013 9:40PM
xx Em
2
0
i sin 2 Pintact f or ; d
7:28
where Em is proportional to Youngs modulus.

Substituting Eqs. (7.11), (7.26) and (7.27) into Eq. (7.28), and accounting for the fact
that for an individual Hookean ber, according to Eq. (7.26), = Efsin2 with Ef being
the Youngs modulus of individual nanobers, we arrive at the following micromechanical stressstrain relation for nanober mats under uniaxial elongation:

2
2Ef sin 2
2Ef sin 2
4
exp
sin 1
xx Em exp2

0
7:29
f 0 0 d

sin 2 exp4 cos 2
The dimensionless tensile strength */Ef in Eq. (7.29) fully determines the character of
deviation of the dependence of xx on from the linear Hookes law, and thus
effectively controls mat plasticity. In the limit of small strains when 0, Eq. (7.29)
reduces to xx 3=8Em . The latter corresponds to Hookean behavior. The Hookean
limit should correspond to that of the phenomenological model of Eq. (7.9), which
means that
Em
32
E
9
7:30
The micromechanical model (Eq. 7.29) was tted to the experimental data for solutionblown soy protein/nylon 6 nanober mats. One case of such tting is shown in
Figure 7.10b (for comparison Figure 7.10a depicts predictions of the phenomenological
model, Eq. 7.9, for this case). It can be seen that the micromechanical model (Eq. 7.29)
0.7
1
2
1
0.3
xx (MPa)
0.2
0.3
2
1
2
0.0
0
Figure 7.10
0.4
0.1
0.0
(a)
0.2
0.1
0.5
0.4
0.6
2
2
0.5
xx (MPa)
0.7
0.6
3
(%)
7
(b)
(%)
Soy protein/nylon 6 stressstrain curve tted with (a) phenomenological and (b) micromechanical
models. Sample No. 1 from Table 7.3. Black symbols (1) depict the experimental data, lines
(2) the corresponding theoretical results. Reprinted with permission from Khansari et al. (2012).
311 [297--318] 10.12.2013 9:40PM
311
Table 7.3 Youngs modulus, yield stress and the relative bond-rupture stress */Ef for soy protein/nylon 6
nanober mats.
Sample
Width
(mm)
Thickness
(mm)
E (phenom.
model) (MPa)
E (micromech.
model) (MPa)
Y (MPa)
*/Ef
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
11.91
11.47
12.43
11.47
11.21
11.38
11.53
11.65
11.99
12.01
11.59
11.68
10.87
11.6
12.72
13.38
14.14
13.44
12.83
12.29
0.22
0.22
0.22
0.22
0.22
0.24
0.2
0.22
0.22
0.16
0.16
0.22
0.22
0.22
0.22
0.16
0.16
0.17
0.15
0.18
12.8
17.58
14.26
20.88
19.69
24.01
38.02
24.25
14.87
21.79
21.58
17.62
21.93
14.55
18.74
14.58
20.22
40.28
16.50
21.46
12.8
17.58
14.26
20.88
19.69
24.01
38.02
24.25
14.87
21.79
21.58
17.62
21.93
14.55
18.74
14.58
20.22
40.28
16.50
21.46
0.53
0.46
0.46
0.53
0.6
0.6
0.78
0.53
0.49
0.79
0.69
0.59
0.49
0.56
0.67
0.69
0.39
0.77
0.21
0.57
0.071
0.047
0.058
0.047
0.055
0.047
0.038
0.041
0.058
0.060
0.062
0.057
0.042
0.062
0.062
0.076
0.041
0.035
0.030
0.041
ts the data in the elastic and plastic part of the stressstrain dependence just as well as the
phenomenological model (Eq. 7.9), albeit the latter is incapable of describing the last part
corresponding to the catastrophic rupture of the sample. Similar comparisons were done
for nine different samples and the results are presented in Table 7.3. The tted values of
Em of the micromechanical model were recalculated to the values of Youngs modulus E
of the phenomenological model, using the relation Eq. (7.30), and found to be in full
agreement with the values of E found directly by tting the phenomenological model
(Table 7.3). Note that the micromechanical model (Eq. 7.29) does not involve the yield
stress Y. Instead, it involves the relative characteristic bond-breaking stress */Ef, which
is responsible for plastic effects. Its values for the 20 samples are also presented in
Table 7.3.
A more detailed comparison of the predictions of the micromechanical model (Eq.
7.29) was conducted using the results of the tensile tests for a meltblown mat supplied by
Metallocene Polypropylene, Exxon Achieve 6936 G, 1800 MFR. The mat was formed at
an average polymer mass ow rate of 0.3 ghm (g hole1 min1), with a die-to-collector
distance of 12.5 cm. The surrounding air velocity was 117.64 m s1, while the collector
speed was kept at 9 m min1.
The angular distributions of bers in the lay-down were numerically modeled at a
screen velocity of 9 m min1, as described in Section 4.7 of Chapter 4. The predicted laydown pattern and the evolution of the oncoming jets at a dimensionless time step of
312
312 [297--318] 10.12.2013 9:40PM
0.9
0.6
0.3
Z
0
0.3
0.9
0.6
0.6
0.9
0.3
Z
0.0
0.5
0.5
1.2
0.3
0.6
0.9
1.2
0.6
0.3
0
0.3
H 0.6
0
0.3
0.6
0.9
0.9
0.9
0.6
0.3
0
0.3
0.6
0.9
0
0.9
0.6
0.3
1.2
Figure 7.11
Jet congurations in the case of screen speed Vscreen = 9 m min at t = 45 (which corresponds
to 0.03 s). Predictions are based on the method of Yarin et al. (2011).
45 (0.03 s) are shown in Figure 7.11. It can be seen that because of the low screen speed in
the case under consideration, the oncoming polymer jets make numerous loops, allowing
for the excursion of the deposition points over the screen. It should be emphasized that
the experiments do not distinguish the angular directions of bers in the lay-down in
congurations (c) and (d) in Figure 4.29 in Section 4.7 in Chapter 4, i.e. from an
experimental point of view they do not differ from (a) and (b). Therefore, for comparison
with experiments, one has to lump the former pair with the latter, as shown schematically
in Figure 4.43 in Section 4.7 in Chapter 4. Then, the whole range of ber orientations
considered shrinks from 360 to only 180. Comparison of the numerical predictions of
the angular distribution of bers in the lay-down at a screen cross-section of = 0.2 with
the experimental data is shown Figure 7.12. It can be seen that the numerical predictions
313 [297--318] 10.12.2013 9:40PM
313
0.6
Frequency
0.5
0.4
Experimental value
0.3
Numerical prediction
0.2
0.1
0
0
Figure 7.12
10 20 30 40 50 60 70 80 90 100 110 120 130 140 150 160 170

Angle (Degree)
Comparison of the experimental and predicted angular distributions. It can be seen that the
numerical results slightly over-predict the orientation in the machine direction 90. Predictions are
based on the method of Yarin et al. (2011).
reproduce the experimental data fairly well. Note also that the numerical results discussed
in Section 4.7 in Chapter 4 showed that with an increase in screen speed, a wide-angular
distribution of bers in the lay-down similar to that in Figure 7.12 evolves toward a
normal distribution, with the peak shifting towards 90 (corresponding to the machine
direction). In the present case the screen speed is much lower than in Section 4.7. As a
result, the angular distribution predicted and measured in Figure 7.12 is wide and nonGaussian.
The numerically predicted angular distribution f0() was used as a starting point to nd
the current probability density function of the ber orientation for(,t) when a nonwoven
is subjected to a tensile test. The predicted evolution of the probability density function
for with strain in a tensile test is based on Eq. (7.11) and is depicted in Figure 7.13a. It can
be seen from this gure that as the strain increases (i.e. the stretching continues), the
bers are more and more predominantly oriented in the direction of stretching, i.e., 90
and 270 (they are not distinguished experimentally, as discussed above).
Figure 7.13b shows a partially ruptured meltblown sample in a tensile test. The banks
of the failed sample part on the left-hand side are still spanned by many bers, which are
perfectly oriented in the direction of stretching, in agreement with the theoretical
predictions in Figure 7.13a.
The micromechanical model in Eqs. (7.28) and (7.29) involves the distribution of the
probability density function of ber orientation in a stretched nonwoven sample for,
which is already predicted above. Then, a section of the stressstrain curve was described
using Eq. (7.29), which was tted to the experimental data obtained in the tensile tests to
establish the values of the material parameters Em and Ef/* (see Figures 7.14 and 7.15).
These gures correspond to two stretching rates, 1 and 20 mm min1, respectively. All the
material parameter values found for these two rates are listed in Table 7.4. Equation (7.9)
of the phenomenological model was also tted to such experimental data, illustrated in
314
314 [297--318] 10.12.2013 9:40PM
Probability densityfunction (W)
3
2.5
Before stretching
At a strain of 30%
At a strain of 60%
2
1.5
1
0.5
0
0
100
200
300
Fibers oriented
towards 90
Angle (degree)
(b)
(a)
Figure 7.13
1 cm
(a) Theoretical predictions of the evolution of the probability density function of the angular
ber distribution for with stretching. The distributions of the probability density function
correspond to the following strain values in the tensile test: 0, before stretching of the nonwoven
has started, 30% and 60%. The predictions show that the probability density function evolves
into delta-function-like peaks in the stretching direction corresponding to 90 and 270. (b) A
meltblown sample partially ruptured in the tensile test. The two arrows point toward the grips and
indicate the stretching direction. It can be seen that the bers spanning the banks of the ruptured
sample part on the left are oriented in the direction of stretching, as predicted in panel (a).
Predictions are based on the method of Yarin et al. (2011).
1.8
1.1
1.6
1.4
0.7
0.5
Experimental
Theoretical prediction
(micromechanical
model)
0.3
0.1
0
Figure 7.14
1.2
1
0.8
0.6
Experimental
0.4
(micromechanical
model)
0.2
0
0.01
(a)
Stress (MPa)
Stress (MPa)
0.9
0.02
Strain
0.03
0.04
(b)
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Strain
Comparison of the theoretical prediction of the micromechanical model with experimental

stressstrain curves for a stretching rate of 1 mm min1. Panel (a) shows that the micromechanical
model is in fairly good agreement with the experimental data up to 4% strain, which incorporates
the inception of plastic behavior after the elastic linear part. Panel (b) shows the entire stressstrain
curve. Predictions are based on the method of Yarin et al. (2011).
Figures 7.16 and 7.17 for the stretching rates of 1 and 20 mm min1, respectively. This
allowed us to establish the corresponding Youngs modulus E and yield stress Y, which
are listed in Table 7.5. It should be emphasized that the meltblown samples in the
experiments underwent signicant stretching of the order of ~6070%. This made it
imperative to distinguish between the engineering stress and the true stress, which are
related as xx 0xx 1 , with xx being the true stress and xx the engineering stress.
315 [297--318] 10.12.2013 9:40PM
315
Table 7.4 The material parameters of the micromechanical model, Em and Ef/ * corresponding
to the two stretching rates.
Stretching rate (mm min1)
Sample
Em (MPa)
Ef/*
1
2
3
4
5
6
7
8
9
10
105.173
78.22
78.22
40.53
166.76
74.92
80.92
59.95
110.89
115.27
30
25
22
18
30
21.3
20
19.5
25
21
20
1.2
2
1.8
1.6
0.8
Stress (MPa)
Stress (MPa)
0.6
0.4
Experimental
0.2
(micromechanical model)
1.4
1.2
1
Experimental
0.8
0.6
(micromechanical model)
0.4
0.2
0
0
(a)
Figure 7.15
0.01
0.02
Strain
0.03
0.04
(b)
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Strain
Comparison of the theoretical prediction of the micromechanical model with experimental stress
strain curves for a stretching rate of 20 mm min1. Panel (a) shows that the micromechanical model
is in fairly good agreement with the experimental data up to 4% strain, which incorporates the
inception of plastic behavior after the elastic linear part. Panel (b) shows the entire stressstrain
In Figures 7.14a7.17a a zoomed-in view of the comparison of theoretical and experimental data up to 45% strain is shown, while Figures 7.14b7.17b depict the entire
stressstrain curve obtained experimentally in comparison with the theoretical
predictions.
Table 7.4 shows that the micromechanical material parameter Em varies widely from
sample to sample for the lower stretching rate: from 40.53 MPa to 166.76 MPa. On the
other hand, for the higher stretching rate, variation in the values of Em from sample to
sample is less pronounced: Em varies from 59.95 MPa to 115.27 MPa. The micromechanical parameter Em is related to the macroscopic Youngs modulus E via
Eq. (7.30). Therefore, the values of E following from the micromechanical model are,
correspondingly, from 11.39 MPa to 46.9 MPa for the lower stretching rate, and from
316
316 [297--318] 10.12.2013 9:40PM
1.2
1.8
1.6
1.2
Stress (MPa)
Stress (MPa)
1.4
0.8
0.6
0.4
1
0.8
0.6
Experimental
(phenomenological model)
0.2
Experimental
0.4
0.2
0
0
0
0.01
0.02
0.03
0.04
Strain
(a)
Figure 7.16
0.05
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Strain
(b)
Comparison of the theoretical predictions of the phenomenological model with experimental

2
1.2
1.8
1.6
0.8
Stress (MPa)
Stress (MPa)
0.6
Experimental
0.4
0.2
1.4
1.2
Experimental
1
0.8
0.6
0.4
0.2
0
0
0
(a)
Figure 7.17
0.01
0.02
0.03
Strain
0.04
0.05
(b)
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Strain
Comparison of the theoretical predictions of the phenomenological model with the experimental
16.86 MPa to 32.42 MPa for the higher stretching rate. On the other hand, Table 7.5
shows that the phenomenological model yields E from 18.26 MPa to 49.68 MPa for the
lower stretching rate, and from 32.63 MPa to 60.86 MPa for the higher stretching rate.
Tables 7.4 and 7.5 also show that the ratio Ef/*, which corresponds to plasticity in the
framework of the micromechanical model and does not vary much between samples. Its
mean values are in the range 21.36 to 25 with the standard deviation being 5.19 (20.78%
of the mean value) and 2.16 (10.17% of the mean value) for the lower stretching rate and
higher stretching rate, respectively. On the other hand, the phenomenological yield stress
Y varies signicantly from sample to sample for the higher stretching rate. Indeed,
Table 7.5 shows the mean value of Y to be in the range 0.586 MPa to 0.74 MPa and
the standard deviation to be 0.13 (21.77% of the mean value) and 0.14 MPa (19.15% of
the mean value) for the lower and higher stretching rates, respectively.
317 [297--318] 10.12.2013 9:40PM
317
Table 7.5 The material parameters of the phenomenological model, E and Y, corresponding to the two
stretching rates.
Stretching rate (mm min1)
Sample
E (MPa)
Y (MPa)
1
2
3
4
5
6
7
8
9
10
45.64
31.43
28.08
18.26
49.68
32.63
35.55
25.37
48.71
60.86
0.74
0.62
0.56
0.39
0.62
0.65
0.73
0.58
0.79
0.95
20
It should be emphasized that Figures 7.147.17 show that neither micromechanical

nor phenomenological models were able to describe the entire experimental stressstrain
curves for meltblown samples, not to mention the rupture section at the end. This fact
indicates that the assumption of a constant yield criterion is probably inaccurate for
meltblown mats and a better material hypothesis is called for with meltblown samples
that can be stretched up to ~6070 %. On the other hand, for solution-blown soy protein/
nylon 6 samples that cannot be stretched above 45 %, both models can accurately
describe stressstrain curves up to the rupture zone.
7.5
References
Arinstein, A., Burman, M., Gendelman, O., Zussman, E., 2007. Effect of supramolecular structure
on polymer nanobre elasticity. Nature Nanotechnology 2, 5962.
Dzenis, Y., 2004. Spinning continuous bers for nanotechnology. Science 304, 19171919.
Dzenis, Y., 2008. Structural nanocomposites. Science 319, 419420.
Green, A. E., 1956. Hypo-elasticity and plasticity. Proc. R. Soc. London A 234, 4659.
Grifth, A. A., 1920. The phenomena of rupture and ow in solids. Philos. Trans. R. Soc. London,
Ser A, 221, 163198.
Gu, S. Y., Wu, Q. L., Ren, J., Vansco, G. J., 2005. Mechanical properties of a single electrospun
ber and its structures. Macromol. Rapid Commun. 26, 716720.
Huang, Z. M., Zhang, Y. Z., Ramakrishna, S., Lim, C. T., 2004. Electrospinning and mechanical
characterization of gelatin nanobers. Polymer 45, 53615368.
Inai, R., Kotaki, M., Ramakrishna, S., 2005. Structure and properties of electrospun PLLA single
Khansari, S., Sinha-Ray, S., Yarin, A. L., Pourdeyhimi, B., 2012. Stress-strain dependence for
soy-protein nanober mats. J. Appl. Phys. 111, 044906.
Khil, M. S., Kim, H. Y., Kim, M. S., Park, S. Y., Lee, D. R., 2004. Nanobrous mats of poly
(trimethylene terephthalate) via electrospinning. Polymer 45, 295301.
318
318 [297--318] 10.12.2013 9:40PM
Kim, G. M., Lach, R., Michler, G. H., Chang, Y. W., 2005. The mechanical deformation process of
electrospun polymer nanocomposite bers. Macromol. Rapid Commun. 26, 728733.
Ko, F., Gogotsi, Y., Ali, A., Naguib, N., Ye, H. H., Yang, G. L., Li, C., Willis, P., 2003.
Electrospinning of continuous carbon nanotube-lled nanober yarns. Adv. Mater. 15, 11611165.
Lee, K. H., Kim, H. Y., Ryu, Y. J., Kim, K. W., Choi, S. W., 2003. Mechanical behavior of electrospun ber mats of poly(vinyl chloride)/polyurethane polyblends. J. Polym. Sci., Polym. Phys.
Ed. B 41, 12561262.
Pai, C. L., Boyce, M. C., Rutledge, G. C., 2011a. Mechanical properties of individual electrospun
PA 6(3)T bers and their variation with ber diameter. Polymer 52, 22952301.
Pai, C. L., Boyce, M. C., Rutledge, G. C., 2011b. On the importance of ber curvature to the elastic
moduli of electrospun nonwoven ber meshes. Polymer 52, 61266133.
Papkov, D., Zou, Y., Andalib, M. N., Goponenko, A., Cheng, S. Z. D., Dzenis, Y. A., 2013.
Simultaneously strong and tough ultrane continuous nanobers. ACS Nano, 7, 33243331.
Pedicini, A., Farris, R. J., 2003. Mechanical behavior of electrospun polyurethane. Polymer 44,
68576862.
Prager, W., 1939. On isotropic materials with continuous transition from elastic to plastic state.
Proceedings of the 5th International Congress for Applied Mechanics, Cambridge Mass. Sept
1226 1938. John Wiley & Sons, New York, 234237.
Prager, W., 1942. Fundamental theorems of a new mathematical theory of plasticity. Duke Math. J.
9, 228233.
Rubin, M. B., Yarin, A. L., 1993. On the relationship between phenomenological models for
elastic-viscoplastic metals and polymeric liquids. J. Non-Newton. Fluid Mech. 50, 7988.
Rubin, M. B., Yarin, A. L., 1995. Corrigendum: J. Non-Newton. Fluid Mech. 57, 321.
Sinha-Ray, S., Lee, M. W., Sinha-Ray, S., An, S., Pourdeyhimi, B., Yoon, S. S., Yarin, A. L., 2013.
Chem. C 1, 34913498.
Sinha-Ray, S., Pelot, D. D., Zhou, Z. P., Rahman, A., Wu, X. -F., Yarin, A. L., 2012. Encapsulation
of self-healing materials by coelectrospinning, emulsion electrospinning and solution blowing
and intercalation. J. Mater. Chem. 22, 91389146.
Tan, E. P. S., Lim, C. T., 2004. Physical properties of a single polymeric nanober. Appl. Phys. Lett.
84, 16031605.
Tan, E. P. S., Ng, S. Y., Lim, C. T., 2005. Tensile testing of a single ultrane polymer ber.
Biomaterials 26, 14531456.
Timoshenko, S. P., 1961. Theory of Elastic Stability, McGraw-Hill, New York.
Truesdell, C., 1952. The mechanical foundations of elasticity and uids. J. Rat. Mech. Anal. 1, 125300.
Truesdell, C., 1953. Corrections and additions to The mechanical foundations of elasticity and
uid dynamics. J. Rat. Mech. Anal. 2, 593616.
Wendorff, J. H., Agarwal, S., Greiner, A., 2012. Electrospinning. Wiley-VCH, Weinheim.
Yarin, A. L., 2008. Stimuli-responsive polymers in nanotechnology: Deposition and possible effect
on drug release. Mathematical Modelling of Natural Phenomena 3, No. 5, 115.
Zussman, E., Burman, M., Yarin, A. L., Khaln, R., Cohen, Y., 2006. Tensile deformation of
electrospun Nylon 6,6 nanobers. J. Polym. Sci., Part B- Polymer Physics 44, 14821489.
Zussman, E., Rittel, D., Yarin, A. L., 2003. Failure modes of electrospun nanobers. Appl. Phys.
Lett. 82, 39583960.
319 [319--336] 10.12.2013 9:49PM
Post-processing
This chapter discusses several post-processing approaches applied to as-spun nanobers

to change their structure and/or enhance certain properties. Section 8.1 describes carbonization, sol-gel transformation and calcination, as well as metal-plating, used to make
stiff, hollow or thermally and electrically conducting bers. Sections 8.2 and 8.3 are
devoted to cross-linking of solution-blown soy protein/nylon 6 nanobers. The collected
ber mats can be bonded both chemically (using aldehydes and ionic cross-linkers, as in
Section 8.2), and physically (by means of wet and thermal treatment, as in Section 8.3) to
increase the tensile strength and therefore widen the range of applications of these green
nonwovens. Chemical cross-linkers bond different amino groups, primary amides and
sulfhydryl groups in the protein structure, which is benecial for the enhancement of
tensile strength. It is shown that treatment with ionic cross-linkers results in nanober
mats with a higher Youngs modulus. Covalent bonds formed by aldehyde groups have a
smaller effect on the mat strength. As cross-linked nanobers are exposed to heat, the
bonds formed between amino groups in the bers are broken and they became less
aggregated. In addition, in Section 8.3 it is shown that wet conglutination of soy protein/
nylon 6 nanober mats leads to partial physical cross-linking of nanobers and, consequently, to an increase in Youngs modulus. An enhancement of the tensile strength of
soy protein nanober mats, as well as a slight plasticizing effect, can also result from
exposure to water.
8.1
Carbonization, sol-gel transformation, calcination and metallization

To eliminate the poly(methyl methacrylate) (PMMA) core and carbonize the polyacrylonitrile (PAN) shell in electrospun core/shell nanobers, Zussman et al. (2006) and
Bazilevsky et al. (2007) used the following route. The as-spun bers were placed on
alumina substrates in a tube furnace and stabilized in air for 30 min at 250 C. The bers
were then carbonized by heating in nitrogen, rst at 750 C for 1 h, and then at 1100 C
for another hour. The ramp rate was 5 C min1 between 250, 750 and 1100 C plateaus.
Similar routes to carbonization were also implemented by Zhang and Yarin (2011) to
produce extremely rough bers from electrospun PAN nanobers containing carbon
black (CB) dip-coated in poly(furfuryl alcohol) (FFA), shown on the left in Figure 6.3, or
carbon bers electrospun from phenolic resin (both for electrode usage). The electrical
conductivity of the carbonized PAN + CB nanober mat was 126 S m1; for PAN + CB
320
320 [319--336] 10.12.2013 9:49PM
8 Post-processing
with a single FFA treatment, the conductivity was 290 S m1; for PAN + CB with a
double FFA treatment, the conductivity was 286 S m1. The conductivity of carbon
nanober mats derived from phenolic resin was 243 S m1. These are the bulk carbon
conductivity levels expected for pyrolysis temperatures in the range 10001050 C
(Wang et al. 2003). Carbonization of electrospun nanobers was also conducted by
Kim et al. (2007) to achieve porous carbon nanobers with hollow cores.
In the co-electrospinning experiments of Li and Xia (2004), the Ti(OiPr)4 sol-gel
precursor was originally added into the poly(vinyl pyrrolidone) (PVP)/alcohol solution
used to form the ber shell. Mineral oil was co-electrospun as the ber core. The
following post-processing was used to form TiO2 nanotubes from the coreshell bers.
The as-spun bers were left in air for about 1 h to completely hydrolyze the Ti(OiPr)4
precursor. Then, the mineral oil was extracted from the core by immersing the bers in
octane overnight. After that, hollow nanotubes made of pure titania were formed using
ber calcinations in air at 500 C for 1 h.
Metallization of nanobers by means of electroplating was conducted by Sinha-Ray
et al. (2011a). The resulting copper- and silver-plated thorny devil and cactus-like
nanobers are discussed in Section 9.5 in Chapter 9 in relation to their application for
drop/spray cooling of high-heat-ux surfaces.
8.2
Chemical cross-linking
Biodegradable polymers attract attention in relation to such applications as food packaging, construction materials, composite llers, wood adhesives, particle boards, etc. For
example, forming micro- and nanobers using soy protein isolate (SPI) derived from
such raw materials as soybean is considered to be highly attractive (Sinha-Ray et al.
2011b, Khansari et al. 2012; see Sections 4.9 in Chapter 4 and 7.2 in Chapter 7).
However, many products made of biodegradable polymers possess low strength and
high hydrophilicity.
The basic building blocks of proteins are amino acids, which are linked by different
covalent and ionic bridges (e.g. amide, disulde, etc.). The reactivity of proteins depends
on the side chains of their free amino acids. The labile groups in the side chains are
attacked by cross-linking agents and the resulting dints serve as the sites for efcient
inter- and intramolecular cross-linking. The chemically reactive blocks in amino acids
include carboxylic, primary and secondary amine groups, cystine, lysine, arginine,
guanidyl, and sulfhydryl groups. These reactive groups participate in cross-linking
triggered by chemical cross-linkers or thermal treatment.
Solubility of soy protein in a solvent is determined by the competition between
proteinprotein interactions and proteinsolvent interactions, which is related to the
isoelectric point of soy protein. Therefore, soy protein solubility can be effectively
inuenced by pH, ionic strength, temperature and concentration. The most widely used
cross-linkers for soy proteins include aldehyde groups, with formaldehyde being the
oldest and most common agent (Sinha-Ray et al. 2012). Formaldehyde cross-links
protein polyamide chains by reacting with -NH, -OH and -SH groups. This reaction
321 [319--336] 10.12.2013 9:49PM
8.2 Chemical cross-linking
321
produces methylene bridges between polymer molecules. Glyoxal is a small molecule

compared to most aldehyde compounds and mostly bonds amino acid side chains in one
molecule. The cross-linking effect of glyoxalization is restricted to intermolecular
structure. Also, the cross-linking effect of zinc ions is determined by the way they
bond to protein chains. In addition to these, sodium borohydride, known to be a strong
reducing agent, can also be used as a cross-linker.
Chemical cross-linking of solution-blown (cf. Section 4.9 in Chapter 4) nanober mats
(50/50 SPI/nylon 6) was studied by Sinha-Ray et al. (2012). Nanober mats were cut into
several pieces and each piece weighed carefully. For chemical cross-linking, the weighed
samples were immersed in a solution with a specied weight percentage of cross-linker
relative to the nanober mat. The weight ratio of cross-linking agents to nanober
samples was 5, 10 and 20 wt/wt%. This procedure was followed for four different
types of cross-linking agents: formaldehyde, glyoxal, zinc sulfate and sodium borohydride. After adding a cross-linking solution to the samples, open vials were left at room
temperature for 24 h to dry out completely. Drying in a hood was used instead of vacuum
drying, since the latter would be too fast compared with the rate of the cross-linking
reactions, and the samples would be either non-cross-linked or partially cross-linked. It
should be emphasized that Sinha-Ray et al. (2012) used only samples prepared from the
same batch for cross-linking at different concentrations to avoid variability between
samples. All the resulting cross-linked nanober mats were subjected to uniaxial tensile
tests to measure the mechanical properties, as described in Section 7.2 in Chapter 7.
It was found that chemical cross-linking of ber mats above 50 wt/wt% cross-linker to
nanober mat ratio, led to visible macroscopic cracks in the samples. Therefore, crosslinking experiments were conducted only with 5, 10 and 20 wt/wt% ratios, except for the
case of heat treatment discussed below, where a 50 wt/wt % ratio was also used.
Cross-linking of soy protein with aldehydes is an example of a carbonylamine
reaction. In soy protein isolates the absence of aspargine and glutamine implies that
guanidyl groups are a potential source of cross-linking via methylene bridging. SPI 955
contains a very reactive lysine amino acid [~6.3% of the entire protein content], which
acts as the preferential site for cross-linking due to its conformational freedom and
external surface availability because of steric effects. In addition to methylene bridging,
sulfhydryl groups also participate in sulde linkage. The reaction kinetics reveal that
complete cross-linking occurs in a time frame of 24 h, which is why samples were left in
the cross-linker solution for 24 h to facilitate complete inter- and intraber cross-linking.
The covalent bonds thus formed restrict protein macromolecule mobility and rotation,
which facilitates an increase in Youngs modulus and ber strength. The reduced
exibility of cross-linked protein chains makes nanober mats more brittle and results
in reduction of strain at rupture point, rupture.
Cross-linking of soy protein nanober mats with ZnSO4 relies on metal chelation and
ionic bonding. Soy protein isolates contain many polar amino acids, with which ZnSO4
forms stable ionic bonds. In addition, Zn2+ forms chelating complexes with soy protein,
which increases the strength of the nanober mats.
Sodium borohydride (NaBH4) is a very strong reducing agent. Whenever protein
molecules come in contact with it, the labile disulde group of the polar amino acid
322
322 [319--336] 10.12.2013 9:49PM
8 Post-processing
(a)
(b)
10 m
Figure 8.1
10 m
(c)
(d)
(e)
(f)
SEM images of pristine nanober mat are shown in panels (a) and (b). In panel (b) merged
nanobers are indicated by arrows. SEM images of nanober mats treated with 20 wt% of: (c)
formaldehyde, (d) glyoxal, (e) zinc sulfate, and (f) NaBH4. Panel (e) reveals that there is excess zinc
sulfate deposited on the mat, as shown by arrows. Panel (f) shows that there are sharp features
formed on the nanober mat (shown by arrows). Reprinted with permission from Sinha-Ray et al.
(2012). Copyright 2012, American Chemical Society.
cystine is attacked, which results in sulfhydryldisulde exchange. This, in turn, results

in opening up of the inter- and intramolecular disulde bonds, which are readily oxidized
by air and cross-linked across the chain.
SEM images of a pristine solution-blown nanober mat are shown in Figures 8.1a
and b, and cross-linked samples with 20 wt% of the cross-linkers are shown in
Figures 8.1cf. In particular, Figures 8.1cf show SEM images of samples treated
with formaldehyde, glyoxal, zinc sulfate and sodium borohydride, respectively. The
SEM images of the non-cross-linked nanobers (Figure 8.1b) show that the bers
merge together at several places shown by arrows. This stems from the fact that when
nanobers are collected on a rotating drum, they still remain semi-wet and can merge.
323 [319--336] 10.12.2013 9:49PM
(a)
Figure 8.2
(b)
323
(c)
SEM images of nanobers cross-linked with: (a) 10 wt% zinc sulfate, and (b) 10 wt% and
(c) 5 wt% of NaBH4. In panel (b) arrows point to the sharp features. Reprinted with permission
The degree of merging is higher when the relative humidity is higher. It can be seen that
for the organic cross-linkers, formaldehyde (Figure 8.1c) and glyoxal (Figure 8.1d), the
nanober morphology does not change and there is no deposit or lm formed. These
observations show that the reaction between the nanober mats and the organic crosslinkers was completed. However, for the ionic cross-linker, zinc sulfate (Figure 8.1e),
there are visible deposits of zinc sulfate on the nanobers (shown by arrows) in
comparison to non-cross-linked bers (Figures 8.1a and 8.1b), which shows that an
excess of cross-linker was left. It can also be seen that at some places the nanobers
became fused. This can stem from two different processes: (i) as shown in Figure 8.1b,
the semi-wet pristine nanobers could be merged, or (ii) additional merging could
occur since soy protein is partially soluble in water. However, it should be emphasized
that most of the nanobers were not fused or merged, as shown in Figure 8.2a, where
although the zinc sulfate concentration is low (10 wt/wt%) and thus, the water content is
higher, no such merger is visible. This means that in most of the cases there was no ber
dissolution. For NaBH4 (Figure 8.1f), it can be seen that there are some sharp crystalline features visible on the nanobers (shown by arrows), which means that anything
higher than a 20 wt% mass ratio of the ionic cross-linker would be denitely excessive.
It was found that when the mass ratio of zinc sulfate was decreased to 10 wt%, there
were no more deposits on nanobers (Figure 8.2a). However, for NaBH4 it was found
that even at 10 wt% mass ratio, some nanobers with fewer sharp features were still
visible (Figure 8.2b). Only when the mass ratio of NaBH4 was reduced to 5 wt% did
these features practically disappear (Figure 8.2c). To resolve the chemical nature of
these sharp structures, an elemental analysis was done on a nanober mat treated with
10 wt% NaBH4. Two different places were used: smooth and rough patches (shown by
two arrows in Figure 8.2b). It was found that at both places the amount of sodium was
comparable (~68% of the total signal in both cases). If the sharp features were only
comprised of excess NaBH4, the elemental analysis would have shown higher amounts
of sodium compared with the smooth ber part. Comparable amounts of sodium at both
locations clearly show that the reaction is complete. Therefore, these sharp features are
most probably remnants of broken pieces caused by handling. This conclusion is
supported below by the results of tensile tests that show that cross-linking with
NaBH4 made nanober mats the most brittle.
324
8 Post-processing
2.0
1.5
1.5
3
3
1.0
2
3
2
1
4
0.5
xx (MPa)
XX (MPa)
324 [319--336] 10.12.2013 9:49PM
1.0
4
3
1
10
(a)
15
20
25
30
10
(b)
(%)
15
20
25
(%)
1.0
1.0
0.8
0.8
3
2
0.6
xx (MPa)
XX (MPa)
0.0
0
1
2
3
2
3
2
0.2
4
0.6
0.4
3
4
0.2
2
4
0.3
0.0
0
Figure 8.3
0.5
3
0.0
(c)
4 1
0.4
10
(%)
15
20
(d)
10
12
(%)
Stressstrain curves of cross-linked soy protein nanober mats for different cross-linkers at various
concentrations. In all panels, curve 1 shows the stressstrain dependence for untreated soy protein
nanobers used for control; curve 2 corresponds to 5 wt/wt% cross-linker/nanober mat ratio,
curve 3 to 10 wt/wt% cross-linker/nanober mat ratio and curve 4 to 20 wt/wt% cross-linker/
nanober mat ratio. Panel (a) shows stressstrain curves for soy protein/nylon 6 (50/50 wt/wt%)
when formaldehyde was used as a bonding agent; in panel (b) glyoxal was used as a cross-linking
agent; anel (c) corresponds to zinc-sulfate-treated samples and panel (d) shows sodium
borohydride-treated nanobers. Reprinted with permission from Sinha-Ray et al. (2012).
Copyright 2012, American Chemical Society.
Figures 8.3ad illustrate the typical stressstrain dependences of soy protein/nylon 6

(50/50 wt/wt %) nanober mats after cross-linking in the presence of 5, 10 and 20 wt/wt%
of different cross-linkers. It can be seen that sodium borohydride and zinc sulfate mostly
affected the strength of the nanober mats, whereas samples treated with formaldehyde
and glyoxal show more plastic behavior than those treated with NaBH4 and ZnSO4. This
clearly demonstrates that the ionic agents were more effective in cross-linking in comparison to the aldehydes, which will be discussed in more detail below.
It should be emphasized that non-cross-linked samples from different batches were
used in panels (a)(d) in Figure 8.3. Even though they result in almost the same Youngs
325 [319--336] 10.12.2013 9:49PM
325
modulus values, their maximum strain at rupture varies signicantly (see Figure 8.4), due
to variability of the process, namely, interber bonding, relative humidity, etc.
While using formaldehyde as a cross-linker, Sinha-Ray et al. (2012) found that an
addition of formaldehyde resulted in an increase in the Youngs modulus of the soy
protein nanober mat and a reduction in the strain at rupture (rupture) compared to the
corresponding non-cross-linked samples (see Figure 8.4a). The maximum increase in the
Youngs modulus corresponded to a ratio of 20 wt/wt% of cross-linker to nanober mat
in the cross-linking process. The average Youngs modulus of non-cross-linked soy
protein/nylon 6 nanober mats was measured as 16.512.39 MPa, whereas a value of
66.8116.05 MPa was achieved for a formaldehyde to nanober mat ratio of 20 wt/wt%.
In addition, the maximum strain at rupture for the non-cross-linked samples was
9.632.88%. This was reduced to 3.472.00% for formaldehyde-cross-linked samples
with a 20 wt/wt% cross-linker/nanober mat ratio. An increase in formaldehyde content
in the cross-linking process resulted in lower strain at rupture, which implies reduced
plasticity of the cross-linked nanober mats. Consequently, lower plasticity observed in
samples cross-linked with higher than 20 wt/wt% formaldehyde mass ratios resulted in
noticeable ruptures while drying at room temperature. Therefore, tensile tests could not
be conducted with these samples. The average Youngs modulus of comparable pure
nylon 6 solution-blown nanober mats is 8.590.88 MPa. Therefore, chemically modied soy protein nanober mats with formaldehyde used as a cross-linker reveal a higher
Youngs modulus than for the corresponding pure nylon 6 nanober mats.
The effect of glyoxal as a cross-linker is specied in Figure 8.4b. The glyoxal/mat
weight ratio in the cross-linking process varied in the range 020 wt/wt%. The Youngs
modulus and yield stress increased with glyoxal percentage up to 10 wt/wt%. It can be
seen that using glyoxal as a cross-linker led to values of Youngs modulus for soy protein
nanober mats almost ve times higher than that for non-cross-linked samples, which is
reported as E = 13.876.36 MPa. The value of Youngs modulus of glyoxal-cross-linked
soy protein nanober mats reached E = 66.9517.48 MPa for 10 wt/wt% cross-linker/
nanober mat ratio. The average maximum strain at rupture was 7.653.26% for noncross-linked samples, and reduced to 3.241.23% for a 10 wt/wt% glyoxal/nanober mat
ratio in the cross-linking process. This result reveals increased brittleness of nanobers
due to cross-linking with glyoxal. Increasing glyoxal content above 20 wt/wt% in the
cross-linking process did not further improve the mechanical properties of the samples,
due to high brittleness. It led to observable cracks in the nanober mat structure and
resulted in fragile nanober samples.
Comparison of the data for glyoxal cross-linking with those for formaldehyde crosslinking shows that the former results in higher values of Youngs modulus E than the
latter, up to a cross-linker mass ratio of 20 wt/wt%. Note that in the case of formaldehyde,
the E value increased monotonically with increase in formaldehyde content. Conversely,
in the case of glyoxal, the E value increased up to 10 wt/wt% of glyoxal and then
decreased when the mass ratio increased to 20 wt/wt%. Also, the strain at rupture was
lower in the case of glyoxal compared to that for formaldehyde.
The higher stiffness achieved using glyoxal compared to that with formaldehyde up to
10 wt/wt% can be explained as follows. Both glyoxal (OCHCHO) and formaldehyde
80
66.81
57.55
60
50
40
36.15
30
20
16.51
10
0
10
0
5
Formaldehyde content/mat (wt/wt%)
90
Average Youngs modulus (MPa)
(b)
80
66.95
70
59.22
60
54.31
50
40
30
20
13.87
10
0
(c)
5
10
Glyoxal content/mat (wt/wt%)
100
93.6
80
60
48.82
20
37.25
13.56
0
0
5
10
Zinc sulfate content/mat (wt/wt/%)
20
140
(d)
121.74
120
100
84.9
90.88
80
60
40
20
16.43
0
5
10
0
Sodium borohydride/mat (wt/wt/%)
Figure 8.4
12
10
20
9.63
8
6
4.79
4.05
3.47
2
0
0
5
10
20
12
10
8
7.65
6
4.39
4
3.24
3.74
2
0
20
120
40
14
20
Average maximum strain at rupture point (%)
70
0
90
(a)
326 [319--336] 10.12.2013 9:49PM
5
10
20
6
4.32
4
1.96
1.31
0.71
0
0
5
10
Zinc sulfate solution/mat (wt/wt/%)
20
3.63
2
0.82
0.53
0.43
0
0
5
10
20
Sodium borohydride/mat (wt/wt/%)
Average Youngs modulus and maximum strain at rupture for the same batch of cross-linked
soy protein samples at different concentrations of: (a) formaldehyde, (b) glyoxal, (c) zinc sulfate
and (d) sodium borohydride, used as cross-linkers. Fiber mats revealed lower plasticity as the
cross-linker content increased. Reprinted with permission from Sinha-Ray et al. (2012).
327 [319--336] 10.12.2013 9:49PM
327
(HCHO) have aldehyde groups. However, glyoxal has more available aldehyde groups
facilitating more cross-linking sites than formaldehyde, which results in higher strength
glyoxal-cross-linked nanober mats compared to formaldehyde-treated ones. Note also
that nanober mats subjected to cross-linking are porous uffy materials. The open
porosity of nanober mats resulted in easier cross-linker access, and high E values
comparable to those of solid sheets in papers by de Carvalho and Grosso (2006) and
Rhim et al. (2000).
The results of Sinha-Ray et al. (2012) revealed that Youngs modulus E reached
maximum values for cross-linker/nanober mat ratios in the cross-linking process in
the range 1020 wt/wt%. In the case of formaldehyde, an increase in its content in the
range 1020 wt/wt% did not result in any appreciable change in the value of E, as it did in
the range 510 wt/wt%. In the case of glyoxal the mean value of E corresponding to
20 wt/wt% of glyoxal decreased compared to that for 10 wt/wt%. One can speculate, as
de Carvalho and Grosso (2006), that an increase in the cross-linker content might have
plasticized the samples. However, the results of Sinha-Ray et al. (2012) show that as the
cross-linker/nanober mat ratio increased, the strain at rupture decreased. This implies
that the cross-linkers did not plasticize the samples at higher concentrations. The
reduction in the value of E at higher cross-linker mass ratios can be attributed to the
fact that as the aldehyde content increased, the number of possible inter- and intraber
linkages between protein chains also increased. Therefore, the material became overstretched and microcracks appeared. This resulted in earlier rupture and lowered strength,
as revealed experimentally.
Adding zinc sulfate solution to soy protein nanober mats resulted in about a sevenfold
increase in the average Youngs modulus of samples that were cross-linked at 20 wt/wt%
zinc sulfate/nanober mat ratio, compared to untreated samples. As nanober mats became
stronger due to the effect of the ionic bonding agent, the brittleness of the mats became
considerable, as the reduction in the maximum strain at rupture from 4.321.54% for noncross-linked samples to 0.710.32% for those cross-linked at a 20 wt/wt% zinc sulfate/
nanober mat ratio clearly shows. Sinha-Ray et al. (2012) demonstrated that cross-linking
with ZnSO4 has a stronger effect compared to aldehyde compounds. Figure 8.4c shows the
overall trends for soy protein/nylon 6 (50/50 wt/wt %) mats, in particular, in Youngs
modulus and maximum strain at rupture at different zinc sulfate contents.
Tensile tests were also conducted with soy protein/nylon 6 samples that were crosslinked using sodium borohydride. The results revealed that the maximum strength for
such nanober samples was achieved at a 20 wt/wt% cross-linker/nanober mat ratio. As
with the other types of cross-linkers, stronger nanober mats were less plastic. Using
sodium borohydride resulted in almost seven times stronger nanober mats compared to
untreated ones (see Figure 8.4d).
Zinc sulfate and sodium borohydride had the strongest effects on soy protein nanober
strength compared to the same formaldehyde- or glyoxal-to-nanober mat ratio.
Therefore, stronger proteinprotein interactions were achieved in nanober mats that
were chemically treated with sodium borohydride and zinc sulfate solutions. This implies
that ionic bonds formed between polymeric chains in the protein structure are stronger
than the inter-and intramolecular bonds formed by covalent cross-linking agents.
328
328 [319--336] 10.12.2013 9:49PM
8 Post-processing
Cross-linked soy protein/nylon 6 nanober mats were heated by Sinha-Ray et al.

(2012) up to 80 C for 20 min. This was done to reveal the effect of heat treatment on
cross-linked nanober mats for different cross-linkers used at different contents.
Figure 8.5 shows the average Youngs modulus for soy protein nanober samples as a
function of cross-linker content for three different agents used for nanober treatment:
formaldehyde, glyoxal and zinc sulfate solution. In each case, the average Youngs
moduli of thermally treated and untreated cross-linked samples are compared for each
of the cross-linker contents. Figure 8.5 shows that the average Youngs modulus of heattreated cross-linked samples is lower than the Youngs modulus of comparable untreated
samples.
In a different set of experiments, soy protein/nylon 6 (40/60 wt/wt %) monolithic
nanobers were cross-linked with the following three of the above-mentioned crosslinking agents: formaldehyde, glyoxal and zinc sulfate. For each bonding agent, the
cross-linking procedure was conducted with four different cross-linker weight to nanober mat ratios: 1, 5, 10 and 50 wt/wt %. For each cross-linker concentration, one half of
nanober samples from a batch were heat treated after being chemically cross-linked.
The second half from the same batch that were not heat treated, were used for control.
Thermal treatment was conducted as follows. After being exposed to a specied concentration of chemical cross-linker for 24 h, the samples were left at 80 C for 20 min on a
glass slide on a hotplate. Then, the mechanical properties of the heat-treated cross-linked
samples were compared with control (non-heat-treated) samples from the same batch. As
a result, the effect of heat treatment on different covalent and ionic bonds in cross-linked
samples was elucidated. Heat treatment of non-cross-linked proteins results in stronger
inter- and intramolecular cross-linking, mostly between cystine and lysine amino acids
owing to the presence of labile disulde bonds, which results in a higher Youngs
modulus and lower strain at rupture. This effect is, in part, due to the fact that heated
samples contain less moisture. Therefore, the inevitable plasticizing effect of water is
reduced due to heat treatment, and heat-treated samples reveal a higher Youngs modulus
and appear to be more brittle. Note also that if non-cross-linked nanober mats were
subjected to heat treatment, nylon 6 present in the samples would soften (the glass
transition temperature of nylon is in the range 4757 C) and conglutinate nanobers at
certain locations. Such conglutination would result in an increase in strength of the heattreated non-cross-linked nanober mats, as shown below. Note that heating cleaves the
methylene bridges of aldehyde-xated proteins. Also, when a chelated complex of chitin
with zinc is heated, the release of zinc into water increases, which clearly shows that heat
treatment results in cleavage of the chelated complex (Qin et al. 2007). Therefore, it is
expected that for chemically cross-linked samples subjected to heat treatment, the interand intraprotein linkages formed by covalent or chelated and ionic bonds will break. It is
also expected that when the heat source is removed and the samples cooled to room
temperature, the broken bonds cannot restore themselves completely. This is because
after heat treatment, when the cross-linkers are in their frozen state, they lose mobility
and cannot cross-link the protein with the same efcacy as they do in solution. As a result,
for thermally broken bonds it is energetically favorable to form bonds with the nearest
possible amino acids instead of the exotic linkages formed by the cross-linkers before
329 [319--336] 10.12.2013 9:49PM
329
60
52.42
50
40
36.16 36.95
30
24.92
21.84
20.59
20
17.58
13.48
10
0
(a)
10
50
Average Youngss modulus (MPa)
70
9
8
7
5.87
6
4.66
4.24
4.17
4
3
3.35
2.6
2.43
2.08
2
1
0
1
10
50
200
173.5
180
160
140
123.74
118.43
120
100
89.1
80
60
40
20
26.89
16.08
29.86
19.12
0
1
(b)
10
50
140
116.28
120
100
87.92
80
63.27
60
40
20
55.56
27.93
21.49 18.28
14.75
(c)
Figure 8.5
10
Zinc sulfate content/mat (wt/wt%)
50
Average maximum strain at rupture point (%) Average maximum strain at rupture point (%)
220
6
5
4.03
4
3
3.88
2.95
1.81
0.87
1.16
0.3
0.35
0
1
10
50

5
4
3.52
3
2.61
2.5
2.03
1.96
1.32
1.15
0.57
0
1
10
50
Zinc sulfate content/mat (wt/wt%)
Youngs modulus and the average maximum strain at rupture for both thermally treated and
untreated soy protein/nylon 6 nanober mats (40/60 wt/wt%) that were cross-linked using: (a)
formaldehyde, (b) glyoxal and (c) zinc sulfate solutions. Right columns correspond to cross-linked
nanober mats that were heat treated for 20 min at 80 C on a glass slide left on a hotplate. Left
columns illustrate the data for cross-linked nanober samples that were not exposed to any heat
treatment. A decrease in the average Youngs modulus of heat-treated samples (right columns)
results from the fact that covalent or ionic bonds are destroyed during heated. In panel (a), the
maximum strain at rupture for both thermally treated samples and those which did not undergo heat
330
330 [319--336] 10.12.2013 9:49PM
8 Post-processing
the heat treatment. This results in increased exibility at the expense of lowered strength,
in spite of conglutination of nylon-6 in soy protein nanober mats, as mentioned before.
As the amount of cross-linking agent in the cross-linking process increases, nanober
mats, in most cases, become more brittle. The heat treatment of cross-linked samples
tends to diminish this effect, as seen in Figure 8.5, and expected according to the
discussion in the previous paragraph. As a result of heat treatment, some cross-linked
sites are broken and protein chains recover their mobility, which makes the nanober
mats more plastic.
Sinha-Ray et al. (2012) also studied the water solubility of soy protein/nylon 6 (50/50
wt/wt%) nanober samples that were modied with different cross-linking agents. In
these experiments, nanober samples (soy protein/nylon 6 50/50 wt/wt%) were put
inside a metal grid enclosure and immersed in de-ionized water for 24 h at room
temperature; the water was constantly stirred. After immersion, the samples were withdrawn and left at room temperature for 24 h to dry out completely. Water-solubility test
data for monolithic non-cross-linked samples and monolithic samples chemically bonded
using various agents are reported in Table 8.1. The table also contains results for core
shell soy protein/nylon 6 nanober mats (without any cross-linking). It can be seen that
coreshell soy protein/nylon 6 nanober samples revealed signicantly lower weight
loss in water compared to either cross-linked or non-cross-linked samples. Since the soy
protein is in the core and protected by the nylon 6 in the shell, such coreshell nanobers
possess enhanced water longevity compared to all monolithic bers (cross-linked or not).
Overall, among the monolithic nanobers, cross-linked samples did not show much
difference in weight loss compared to non-cross-linked samples. For comparison,
Yamashita (2007) reported that cast soy protein isolate lms, after being left in a 50 ml
beaker for 24 h at 25 C, lost 28.691.1 wt%.
In summary, the experiments of Sinha-Ray et al. (2012) showed that in chemical crosslinking, two covalent cross-linkers (formaldehyde and glyoxal) and two ionic crosslinkers (zinc sulfate and sodium borohydride) can increase nanober mat stiffness almost
34, 5, 7 and 7 times, respectively. Note that only samples that were prepared from the
same batch collected on a rotating drum were used in tensile tests of non-cross-linked and
cross-linked nanober mats, to minimize the effect of the process variability. For all
cross-linked and non-cross-linked samples, tensile tests were conducted several times to
establish the reported values of Youngs modulus and strength. When an increase of 37
Figure 8.5
(cont.)
treatment was found at 10 wt/wt% formaldehyde/nanober mat ratio. In panel (b), as glyoxal
percentage in the cross-linking process increased, the sample brittleness increased, which
corresponds to diminished values of rupture. However, at these glyoxal concentrations, the
plasticizing effect of heat treatment is small. Panel (c) illustrates the results for ionically bonded
nanobers when zinc sulfate solution was used. As the zinc sulfate content in the chelation process
increases beyond 5%, the sample brittleness increases as well, which results in lower values of
rupture. Note that the plasticizing effect of heat treatment is lower for higher contents of zinc sulfate.
Reprinted with permission from Sinha-Ray et al. (2012). Copyright 2012 American Chemical
Society.
331 [319--336] 10.12.2013 9:49PM
8.3 Physical cross-linking
331
Table 8.1 Weight loss data for 50/50 wt/wt% soy protein/nylon 6 monolithic nanober mats (cross-linked and non-crosslinked) and coreshell nanober mats. Samples were left in water for 24 h at room temperature. Each sample was
weighed before immersion and in two days, after complete drying, to determine the percentage of lost weight as
L = (1W2/W1)100%, where W1 is the sample weight before immersion, and W2 is the weight after immersion and drying.
Soy protein/nylon 6
nanober samples
Ave. weight
loss (%)
Soy protein/nylon 6
nanober samples
Ave. weight
loss (%)
Monolithic non-cross-linked
Monolithic 5 wt/wt%
glyoxal cross-linked
Monolithic 10 wt/wt%
Monolithic 5 wt/wt%
formaldehyde cross-linked
21.85
19.97
Coreshell non-cross-linked
Monolithic 5 wt/wt%
zinc sulfate cross-linked
Monolithic 5 wt/wt%
sodium borohydride cross-linked
Monolithic 20 wt/wt %
5.28
15.48
21.80
17.07
17.05
16.65
23.65
24.60
28.29
18.32
17.94
15.51
times is observed between cross-linked and non-cross-linked samples, it cannot be

attributed to process variability, i.e. to ber diameter, porosity, interconnections, etc.
from batch to batch. The results show that ionic bonding in soy protein structure results in
a higher Youngs modulus compared to aldehyde-treated bers. Heat treatment mostly
plasticizes cross-linked nanober mats. In experiments on mass loss in water, it was
shown that the best longevity is achieved with coreshell nanober mats, where soy
protein is located in the core.
8.3
Physical cross-linking
For thermal bonding of soy protein/nylon 6 nanober mats, Sinha-Ray et al. (2012) used
rectangular samples, which were ironed for 1 min at 55 C. This resulted in partial
conglutination and cross-linking of nanobers at the intersection points. After such
treatment, nanober mats were left at room temperature for 15 min to cool. The
rectangular samples were used in tensile tests in order to reveal their stressstrain
characteristics. Figure 8.6a compares stressstrain curves for samples of soy protein/
nylon 6 (40/60 wt/wt %) nanober mats that underwent the thermal bonding process with
those that did not (taken from the same batch of samples). It can be clearly seen that
thermal bonding increases the Youngs modulus, the yield stress and the tensile stress of
soy-protein-containing nanober mats. The corresponding average data for Youngs
modulus E, yield stress Y, and specic strain energy U 0 xx d are combined in

Figures 8.6bd.
332
332 [319--336] 10.12.2013 9:49PM
8 Post-processing
1.2
1.0
0.8
xx (MPa)
25
0.6
2
2
0.4
1
0.2
20
15
5
0
10
15
20
25
(%)
(b)
Average yield stress (MPa)
0.8
0.61
0.38
0.4
0.2
0.0
Figure 8.6
Thermally bonded
nanofiber samples
30
Non-bonded
nanofiber samples
Average specific strain energy (MPa)
(a)
(c)
12.89
10
0.0
0.6
18.76
Thermally bonded Non-bonded nanofiber

nanofiber samples samples
12
10
(d)
8.09
6.34
6
4
2
0
Thermally bonded
nanofiber samples
Non-bonded nanofiber
samples
Comparison of thermally bonded and non-bonded samples. Panel (a) shows the stressstrain curves
of soy protein/nylon 6 (40/60 wt/wt %) nanober mats: curve 1, thermally bonded (at 55 C under
compression) nanober mat; curve 2, non-bonded nanober mat. The normal stress in the
stretching direction is denoted xx, the tensile strain (%). Panel (b) shows the average Youngs
modulus, panel (c) the average yield stress and panel (d) the average specic strain energy. The
average Youngs modulus, the yield stress and specic strain energy increase when nanober
samples are exposed to heat treatment. Reprinted with permission from Sinha-Ray et al. (2012).
The average Youngs modulus of non-treated nanober mats was found to be 12.895.34
MPa. The Youngs modulus of the nanober mat increased to 18.765.17 MPa when
exposed to post-heat-treatment under compression for only 1 min. When exposed to heat
treatment, nylon 6 present in the samples softens and forms conglutination points, which
result in physical cross-linking of nanobers and an increase in the average Youngs
modulus. Also it was observed that samples became more brittle after heat treatment.
The original average maximum strain at rupture rupture for soy protein nanober mats was
found to be 8.191.71%, whereas ironing of the samples at 55 C for 1 min resulted in an
rupture value of 6.861.17 %.
Another type of post-treatment studied by Sinha-Ray et al. (2012) was wet bonding.
Prior to wet bonding, nanober mat samples were submerged in water and then
333 [319--336] 10.12.2013 9:49PM
8.3 Physical cross-linking
333
immediately taken out. After that, wet samples with dimensions of about 1 cm in width
and 2.5 cm in length were compressed under a mass load of 150 g (i.e. under a pressure
of 6 kPa) for 24 h at room temperature until partially dried. After removing the load,
these samples were left at room temperature for another day to dry out completely.
The dried samples underwent a uniaxial tensile test using an Instron machine
(model 5294, with 100 N capacity on pneumatic grips) and their stressstrain curves
were measured. As a result, the effect of wet conglutination under a load was evaluated.
This effect stems from interber conglutination in the wet state at intersection points.
Indeed, the soy protein isolate used by Sinha-Ray et al. (2012) was partially soluble in
water. In the wet state the soy protein of different nanobers merged at the intersection
points and formed bonds on drying.
It was found that due to wet conglutination and the resulting cross-linking effect,
the overall mechanical properties of soy-protein-containing nanober mats were
enhanced. The Youngs modulus showed an increase of about 65%, which can be
attributed to bonds formed at the wet intersection points. Correspondingly, the
specic strain energy U increased by approximately 33%. The average yield stress
stayed practically unchanged. However, after wet bonding, nanober mats were also
plasticized, as both soy protein isolate and nylon 6 absorb water, which is the reason
that the strain at rupture does not decrease, although the strength increases. Figure 8.7
compares the average mechanical properties of pre-wetted and wet-conglutinated
nanobers with those of corresponding untreated samples. Figure 8.8 shows SEM
images of soy protein/nylon 6 (40/60 wt/wt %) nanobers after they were pre-wetted
and wet-conglutinated under a load of 150 g. The images in Figure 8.8 demonstrate
that after wet-conglutination under load, the individual nanobers keep their
individuality.
The humid aging of soy protein nanober mats was explored by Sinha-Ray et al.
(2012) as follows. Nanober samples were left in water at 80 C for 1 h. After that, the
samples were extracted from water and left at room temperature for 24 h to dry out
without applying any pressure and then used in tensile tests. This experiment revealed
the mechanical properties of soy protein/nylon 6 nanober mats after the exposure
to severe humidity conditions and elevated temperature. A typical stressstrain curve
for humid-aged nanobers after their immersion in hot water for 1 h is depicted in
Figure 8.9a, where it is compared with the stressstrain curve for a corresponding
non-treated sample.
The results obtained for the humid-aged samples demonstrate the effect of wet aging in
hot water on soy protein nanober mats. An increase of about 16% in Youngs modulus
and doubled specic strain energy were recorded. On the other hand, the yield stress
practically did not. The mechanical properties of the humid-aged nanober samples are
compared with those for the corresponding untreated samples. In particular, Figure 8.9d
illustrates an enhanced plasticity range for humid-aged nanober mats. The maximum
strain at rupture (rupture) is reported as 5.951.04 for untreated samples, whereas this
parameter increased to 11.493.44 for nanober mats after the humid-aging experiment.
Therefore, it is shown that while soy protein-containing nanober mats retained their
334
334 [319--336] 10.12.2013 9:49PM
8 Post-processing
1.2
1.0
1
2
xx (MPa)
0.8
0.6
2
0.4
2
0.2
35
1
25
25.33
20
15.33
15
10
5
0
0.0
2
(%)
(a)
(b)
0.8
0.65
0.6
0.61
0.5
0.4
0.3
0.2
0.1
0.0
Figure 8.7
12
10
9.34
7.08
6
4
2
0
Wet-bonded
soy protein/nylon 6
nanofiber samples
(c)
Non-treated
soy protein/nylon 6
nanofiber samples
14
0.9
0.7
Wet-bonded
soy protein/nylon 6
nanofiber samples
10 12 14 16 18 20 22 24
Average yield stress (MPa)
30
Wet-bonded
soy protein/nylon 6
nanofiber samples
Non-treated
soy protein/nylon 6
nanofiber samples
Non-treated
soy protein/nylon 6
nanofiber samples
(d)
Comparison of pre-wetted, wet-conglutinated and non-treated samples. Panel (a): curve 1 shows
the stressstrain curve of a pre-wetted, wet-conglutinated nanober sample under 150 g load.
The stressstrain curve of a corresponding non-treated nanober sample from the same batch is
shown as curve 2. It can be seen that the wet-conglutinated sample reveals higher Youngs
modulus and specic strain energy compared to the untreated one. Panel (b) shows the average
Youngs modulus, panel (c) the average yield stress and panel (d) the average specic strain
energy. Reprinted with permission from Sinha-Ray et al. (2012). Copyright 2012, American
Chemical Society.
strength under conditions of extreme humidity and temperature, they were also signicantly plasticized compared to the original samples.
In summary, the physical cross-linking (thermal bonding) of nanobers under compression led to an almost 50% increase in Youngs modulus, as well as slightly enhanced
brittleness of the samples. Pre-wetting and wet conglutination under a 6 kPa load resulted
in samples with a Youngs modulus of almost 65% higher than for the corresponding
non-treated samples. The higher maximum strain at rupture of humid-aged nanober
mats is indicative of the plasticizing effect of water.
335 [319--336] 10.12.2013 9:49PM
(b)
(a)
Figure 8.8
SEM images of soy protein/nylon 6 nanobers after pre-wetting and wet conglutination under a
load of 150 g. Nanobers kept their individuality under 6 kPa pressure (as shown by arrows).
Reprinted with permission from Sinha-Ray et al. (2012). Copyright 2012, American Chemical
Society.
1.2
1.0
xx (MPa)
1.4
0.8
0.6
2
0.4
0.2
25
20
21.47
18.64
15
10
5
0
0.0
2
(a)
10
12
14
16
(b)
(%)
15
12.15
10
5.57
5
(c)
Figure 8.9
Aged
soy protein/nylon
6 nanofibers
18
Non-treated soy protein/

Aged
soy protein/nylon nylon 6 nanofibers
6 nanofibers
Average maximum strain at rupture (%)
30

nylon 6 nanofibers
15
11.49
10
5.95
5
0
Aged
soy protein/nylon
6 nanofibers

nylon 6 nanofibers
(d)
Comparison of humid-aged and non-treated samples. (a) Stressstrain curve for a humid-aged (in
hot water) soy protein nanober sample is shown as curve 1 and for the corresponding untreated
sample as 2. Signicantly higher values of Youngs modulus and specic strain energy were found
for the humid-aged nanober samples. (b) The average Youngs modulus, (c) specic strain energy
and (d) maximum strain at rupture of soy protein/nylon 6 (40/60 wt/wt %) nanober mats aged in
hot water for 1 h compared to those of the corresponding non-treated samples. Reprinted with
permission from Sinha-Ray et al. (2012). Copyright 2012, American Chemical Society.
336 [319--336] 10.12.2013 9:49PM
336
8 Post-processing
8.4
References
Bazilevsky A. V., Yarin, A. L., Megaridis, C. M., 2007. Co-electrospinning of coreshell nano/
microbers using a single nozzle technique, Langmuir 23, 23112314.
de Carvalho, R. A., Grosso, C. R. F., 2006. Properties of chemically modied gelatin lms.
Brazilian J. Chem. Eng. 23, 4553.
Khansari, S., Sinha-Ray, S., Yarin, A. L., Pourdeyhimi, B., 2012. Stress-strain dependence for soyprotein nanober mats. J. Appl. Phys. 111, 044906.
Kim, C., Jeong, Y. I., Ngoc, B. T. N., Yang, K. S., Kojima, M., Kim, Y. A., Endo, M., Lee, J. W.,
2007. Synthesis and characterization of porous carbon nanobers with hollow cores through the
thermal treatment of electrospun copolymeric nanober webs. Small 3, 9195.
Li., D., Xia, Y., 2004. Direct fabrication of composite and ceramic hollow nanobers by electrospinning, Nano Letters 4, 933938.
Qin, Y., Zhu, C., Chen, J., Liang, D., Wo, G., 2007. Absorption and release of zinc and copper ions
by chitosan bers. J. Appl. Polym. Sci. 105, 527532.
Rhim, J. W., Gennadios, A., Handa, A., Weller, C. L., Hanna, M. A., 2000. Solubility, tensile, and
color properties of modied soy protein isolate lms. J. Agric. Food Chem. 48, 49374941.
Sinha-Ray, S., Khansari, S., Yarin, A. L., Pourdeyhimi, B., 2012. Effect of chemical and physical
cross-linking on tensile characteristics of solution-blown soy protein nanober mats. Ind. Eng.
Chem. Res. 51, 1510915121.
Sinha-Ray, S., Zhang, Y., Yarin, A. L., 2011a. Thorny devil nano-textured bers: The way to
cooling rates of the order of 1 kW/cm2. Langmuir 27, 215226 (2011).
Sinha-Ray, S., Zhang, Y., Yarin, A. L., Davis, S. C., Pourdeyhimi, B., 2011b. Solution blowing of
Wang, Y., Santiago-Aviles, J. J., Furlan, R., Ramos, I., 2003. Pyrolysis temperature and time
dependence of electrical conductivity evolution for electrostatically generated carbon nanobers. IEEE Trans. Nanotechnol. 2, 3943.
Yamashita, S., 2007. Heat-induced antigen retrieval: Mechanisms and application to histochemistry. Prog. Histochem. Cytochem. 41, 141200.
Zhang, Y., Yarin, A. L., 2011. Carbon nanobers decorated with Poly(furfuryl alcohol)- derived
carbon nanoparticles and Tetraethylorthosilicate-derived silica nanoparticles. Langmuir 27,
1462714631.
Zussman, E., Yarin, A. L., Bazilevsky, A. V., Avrahami, R., Feldman, M., 2006. Electrospun
Polyacrylonitrile/Poly(methyl methacrylate)-derived carbon micro-/nanotubes. Adv. Mater. 18,
348353.
337 [337--358] 10.12.2013 7:51PM
Applications of microand nanofibers
This chapter outlines several applications of electrospun and solution-blown nanobers

and their mats. In the case of lters and membranes (Section 9.1), industrial application
has already begun, but a number of research questions are still open. Applications of
nanober mats as uffy electrodes benecial for fuel cells and Li-ion batteries have
recently attracted signicant attention and are the focus of Section 9.2. Two recent
approaches based on nanobers were proposed in the eld of cooling of high-heat ux
microelectronics (Section 9.3) and nanouidics (Section 9.4).
9.1
Filters and membranes

Filter materials are used for air, water and blood ltration, while membranes are used in
separation processes, in particular, for bioseparation and pathogen removal for direct
blood transfusion. Filters can remove particles, droplets, bacteria, viruses or even
individual molecules from a carrier uid owing through them, or, in principle, possess
advanced detection and response features that are practically absent in todays products
(see Chapter 10).
An interesting area of focus for current research involves electrostatic, brous lter
particle-capture mechanisms. There are currently no reliable and reproducible methods
available for directly measuring the electrostatic particle capture force for bers and
webs. Charged nanoparticles embedded in lter media are called electrets. Electret lter
media for use in air lter products should be designed and optimized to minimize
electrostatic decay and losses in ltration efciency. While an electrostatic force can be
determined by electrostatic force microscopy (EFM) for single bers (Kim et al. 2006),
the role of ber geometry, surface area and web structure on the total electrostatic force
generated in lter structures, and the manner in which the lters perform with particle
loading, is not yet uncovered (see Filatov et al. 2007).
The monograph by Filatov et al. (2007) covers multiple applications of nanobers in
lters. Filter test methods were applied to conventional lters with a sandwiched electrospun nanober layer (Wang et al. 2008). The results showed that nanober lters are
more benecial than conventional berglass lters for particles larger than approximately
100 nm. On the other hand, for smaller particles, the nanober lters do not perform
better. Today, it would be impractical to create lters wholly composed of nanobers
because of the rapid pressure drop, which is a direct consequence of Darcys law of
338 [337--358] 10.12.2013 7:51PM
338
9 Applications of micro- and nanofibers
Figure 9.1
Particles adsorbed on a nanober medium (df = 200 nm) at early (left) and late (right) stages of
particle loading; numerically calculated pattern. Reprinted from Maze et al. (2007), with
ltration (Barenblatt et al. 1990). In a hybrid structure, ne bers contribute to ltration/

sensing efciency, while coarser bers contribute to rigidity. Useful hierarchical structures of both ne and coarse coiled bers can be created easily by electrospinning.
Models of the performance of ber structures were developed to predict dynamic drop/
ber interaction and evaluate the effect of ber wettability, roughness, exibility and
electric charge for a range of applications. There are several cell model studies describing
adsorption on a single ber or an ordered array of bers (e.g. Yarin et al. 2006). However,
current cell models are exclusively developed for clean media and fail to predict
performance of lters with adsorbed particles. Also, cell models are only developed for
media made up of a single ber component, while it is often desirable to create hybrid
structures and composites by combining micro- and nanobers, resulting in media with
multimodal ber diameter distributions. While microstructural numerical models for
predicting the capture efciency and pressure drop of brous media can be readily
produced (Filatov et al. 2007), more detailed models are needed to predict the position
of the adsorbed particles. This is important in order to understand fouling, caking and
biofouling, as well as to develop signaling and sensing. The modeling results shown in
Figure 9.1 represent a step in that direction.
9.2
Electrodes for fuel cells, batteries, supercapacitors

and electrochemical reactions
Such nanotextured materials as electrospun or solution-blown nanober mats hold great
promise as high-surface-area electrodes for dye-sensitized solar cells, fuel and microbial
cells, lithium ion (Li-ion) batteries and supercapacitors (Cavaliere et al. 2011). In the present
section several types of such electrodes are considered in brief. We begin with the electrodes
for microbial biocells formed and implemented by Chen et al. (2011). They used
339 [337--358] 10.12.2013 7:51PM
9.2 Electrodes for fuel cells
Figure 9.2
339
Bioelectrochemical cell (microbial fuel cell) for the simultaneous study of several different
electrode materials used by Chen et al. (2011). Reproduced with permission from The Royal
Society of Chemistry.
electrospinning or solution blowing of polyacrylonitrile (PAN)-based nanobers, followed

by carbonization (see Section 8.1 in Chapter 8). The resulting carbon nanotextured mats
were used to signicantly enhance the current density of bioelectrocatalytic anodes in a
microbial fuel cell, shown in Figure 9.2. Three different ber mat materials were studied in
comparison to conventional polycrystalline graphite and carbon felt: (i) three-dimensional
porous carbon bers, produced by gas-assisted electrospinning (hereafter denoted as
GES-CFM), (ii) electrospun carbon bers (ES-CFM) and (iii) solution-blown carbon bers
(SB-CFM). The latter two materials were additionally modied to incorporate 15% carbon
black (CB) to increase the porosity and conductivity of the resulting electrodes. These
modied materials are denoted as ES-CFM15%CB and SB-CFM15%CB.
The enhanced bioelectrocatalytic performance of the microbial fuel cell shown in
Figure 9.2 is attributed to a structure that provides a habitat for the growth of biolms of
340
340 [337--358] 10.12.2013 7:51PM
(a)
(b)
(c)
(d)
(e)
(f)
1.5 mm
Junctions
Figure 9.3
Scanning electron microscopic images of: (a) carbon felt, (b) an electroactive biolm grown on
carbon felt, (c) GES-CFM; (d) an electroactive biolm grown on GES-CFM, (e) high-resolution
image of GES-CFM illustrating the occurrence of interbre junctions, (f) cross- sectional view
of a GES-CFM electrode. Chen et al. (2011). Reproduced with permission from The Royal Society
of Chemistry.
electroactive bacteria Geobacter sulfurreducens up to a maximum density supplemented

by efcient substrate supply. Figure 9.3 shows the electrode architectures studied. The
intact three-dimensional biolms formed on low-specic-weight, high-area GES-CFM
(Figure 9.3d) benet from the fact that the individual bers of the nonwoven are
interconnected (Figure 9.3e). These interconnections, which allow optimum electron
transfer and conduction, are formed during the electrospinning process by conglutination
of the individual nanobers due to a high solvent content.
Table 9.1 summarizes the projected (geometric) current density data and physical
parameters of the tested electrode materials. The current density data correspond to the
maxima of the respective semi-batch experiments (Figure 9.4), averaged over at least
three independent experimental sets. Table 9.1 illustrates that conventional polycrystalline carbon graphite shows the lowest bioelectrocatalytic current density of all tested
341 [337--358] 10.12.2013 7:51PM
341
Table 9.1 Cumulative data on electrocatalytic current densities obtained by Chen et al. (2011) at different
electrode materials. The substrate was 10 mM acetate.
Electrode material
Current densitya
(A m2)
Specic weight
(g m2)
Specic current density

(mA g1)
Polycrystalline graphite
Carbon felt
GES-CFM
ES-CFM
ES-CFM15%CB
SB-CFM
SB-CFM15%CB
13
16
30
21
15
17
21
NA
333
42
126
88
437
183
NA
48
714
243
172
56
128
Substrate supply
(20 mM acetate)
30
Substrate supply
(10 mM acetate)
Projected (geometric) current density

NA not applicable.
20
jgeom./
A m2
10
0
0
Time / d
Figure 9.4
Biocatalytic current generation at a GES-CFM modied carbon electrode in a model semi-batch

experiment. The GES-CFM electrode was modied by a wastewater-derived secondary
biolm grown in a half cell experiment under potentiostatic control. The electrode potential was
0.2 V. Chen et al. (2011). Reproduced with permission from The Royal Society of Chemistry.
samples. The corresponding current value of 13 A m2 is still quite high. This is

explained by the fact that Chen et al. (2011) preselected (secondary) biolms with an
optimized performance, in combination with a reaction temperature of 35 C. Carbon felt
revealed a higher (by 23%) performance than polycrystalline graphite. The highest
current density was obtained with GES-CFM. As illustrated in Figure 9.4, biolms
grown on a GES-CFM electrode delivered maximum current densities of 30 A m2.
342
342 [337--358] 10.12.2013 7:51PM
These are the highest current densities achieved so far with an electroactive microbial
biolm.
The other material types, ES-CFM and SB-CFM, also yielded current densities higher
than commercial carbon felt, up to 21 A m2 (Table 9.1). Interestingly, the addition of
carbon black did not deliver unambiguous results. In the case of solution-blown material,
SB-CFM, addition of 15% CB increased the current density from 17 to 21 A m2. On the
other hand, the electrospun ES-CFM did not benet from the addition of CB. In the latter
case the current density decreased from 21 to 15 A m2. The reason for this behavior is
not known.
The origin of the performance differences between the different electrospun and
solution-blown materials is so far not fully understood. All materials possess an extremely
high porosity of up to 99%. Porosity represents the fraction of void space in the material and
is of utmost importance. Indeed, high porosity not only lowers the amount of material to a
minimum, but also maximizes penetration of microorganisms and the diffusional substrate
supply. However, additional parameters, like the occurrence of junction points (see
Figure 9.3e), may represent a key factor that leads to an additional performance gain.
Similar nanotextured carbonized nanober electrodes used as a substrate for cathode
material in a Li-ion battery were prepared by electrospinning PAN with 1115 wt% CB,
followed by carbonization, by Dimesso et al. (2012, 2013; see Figure 9.5a). The nanobers are fairly uniform in length and diameter; some of them can appear brighter due to
electron-induced charging. Such carbon nanober mats form a fully developed threedimensional network characterized by high surface area, as in the case of electrodes for
microbial biocells discussed above. Due to their high surface area they can be used as a
suitable substrate for the deposition of LiFePO4 or LiCoPO4 by using a simple sol-gel
process.
The LiFePO4 or LiCoPO4 cathode materials were prepared by a Pechini-assisted
reversed polyol process. The carbon nanober substrates were cut into small, highly
porous disks with diameters of about 57 mm and thicknesses of the order of 12 mm.
Acc.V Spot Magn WD

15.0 kv 3.0 5000x 7.7 Elektrospun
a
Figure 9.5
10 m
Acc.V Spot Magn WD

15.0 kv 3.0 5000x 6.8 Elektrospun
10 m
SEM micrographs (a) of carbon nanobers prepared by electrospinning of PAN with 15 wt%CB,
and (b) of an LiFePO4/C nanober composite prepared by the assisted Pechini sol-gel method.
Reprinted with permission from Dimesso et al. (2012). Copyright 2012, AIP Publishing LLC.
343 [337--358] 10.12.2013 7:51PM
343
(a)
4.0
Voltage (V)
3.6
3.2
2.8
2.4
50
100
150
200
t (h)
(b)
4.0
Voltage (V)
C/25
C/10
3.6
3.2
2.8
20
40
60
80
100
120
140
160
1)
Specific capacity (mAhg

Figure 9.6
(a) Charge and discharge curves (in the 2.54.0 V range) versus time, (b) voltagecapacity
curves for the second discharge, measured at rates of C/25 and C/10 at room temperature, for
LiFePO4/C nanober composites. Reprinted with permission from Dimesso et al. (2012).
Coating was successfully performed on the disks by soaking in an aqueous solution

containing lithium, iron salts and phosphates at 70 C for 24 h. After drying, the
composites were annealed at 600 C for 5 h under nitrogen (Figure 9.5b).
The cycling curves as a function of time, shown in Figure 9.6a, indicate a very good
reversible intercalation/deintercalation process into the composite, conrmed by a high
specic capacity of 156 mA h g1. The discharge proles, in Figure 9.6b, differ slightly. The
discharge prole measured at C/10 is noticeably curved, the voltage of 3.4 V drops as the
cell discharges due to polarization. This is due to diffusion resistance within the composite
electrode. On the other hand, a less curved prole can be observed by measuring at a
discharging rate of C/25, which conrms that the sample is of good quality. One can also
observe that the specic capacities of the composites correspond to 92% of the theoretical
value (at C/25) and 89.5% (at C/10).
344
344 [337--358] 10.12.2013 7:51PM
Following the works of Chen et al. (2011) and Dimesso et al. (2012, 2013), Kumar et al.
(2013a) used carbonized electrospun polyacrylonitrile (PAN) nanobers as a metal-free
catalyst system for the reduction of CO2 into a potential fuel (CO). It was shown that such a
catalyst possesses a negligible overpotential (0.17 V) for CO2 reduction, ~13 times higher
CO2-reduction current density compared to a silver metal catalyst and long-term stability.
The enhanced performance of such carbonized electrospun electrodes was attributed
to their nanobrillar structure, since they signicantly outperformed carbonized cast
PAN lms.
9.3
Thorny devil nanofibers: enhancement of spray cooling

and pool boiling
The Leidenfrost effect is familiar to anyone who once sprinkled drops of water on a very
hot skillet or pan. At temperatures of about 150 C, instead of an instantaneous ash
evaporation, the Leidenfrost effect surprisingly allows water droplets to survive for several
minutes and skid and roll over the hot surface. Due to the initial intense evaporation at the
bottom of the drops, a vapor layer between the drops and the skillet is generated, with
pressure sufcient to levitate them (Figure 9.7). The Leidenfrost effect is not only an
amazing demonstration of kitchen physics, but also a technically and industrially important
phenomenon that severely restricts heat removal from high-heat-ux surfaces, since conduction of heat through the vapor layer is negligibly small compared to the latent heat of
water evaporation, which might otherwise be exploited to remove heat. This is very bad for
cooling high-heat-ux surfaces using generally effective methods such as drop or spray
cooling. Indeed, these methods rely on the heat ux associated with the latent heat of
evaporation per unit time, while the Leidenfrost effect prolongs the evaporation process
enormously and diminishes the heat ux correspondingly. Even at lower temperatures,
before the Leidenfrost effect sets in, there are a number of factors detrimental to spray
cooling of hot surfaces: the receding motion of the contact line after droplet spreading and
possible complete bouncing from cold or hot surfaces due to surface tension, as well as
signicant atomization during boiling at the hot surface (Yarin 2006) (Figures 9.8 and 9.9).
Leidenfrost drop
~ 0.1 mm
Vapor pillow
~ 0.2 mm
Extremely hot surface

Figure 9.7
Leidenfrost effect: water droplet levitated on a vapor pillow above a very hot surface.
345 [337--358] 10.12.2013 7:51PM
9.3 Thorny devil nanofibers
345
Deposition
Prompt splash
Corona splash
Receding break-up
Partial rebound
Complete rebound
Figure 9.8
Morphology of drop impact on a dry surface at room temperature. Rioboo et al. (2001). Reprinted
from Rioboo et al. (2001), Copyright 2001, with permission from Begell House, Inc.
30
Volume loss (%)
25
20
15
10
5
0
0
Figure 9.9
50
100
150
Time (ms)
200
250
300
Mass losses due to atomization during cooling of a bare copper substrate at 102.7 C. Reprinted
with permission from Sinha-Ray et al. (2011). Copyright 2011, American Chemical Society.
Progressive miniaturization of microelectronic, optoelectronic and radiological devices inevitably concentrates resistive Joule heating into smaller and smaller volumes. For
example, robotic devices such as unmanned aerial vehicles (UAVs), which carry electrooptical, infrared and other sensors, video equipment, targeting systems and running
signal intelligence systems or real-time image processing in a small space are already
346
346 [337--358] 10.12.2013 7:51PM
approaching the limits of self-burning (Child 2009, Kinney 2009). Drop and spray
cooling in such situations could offer an attractive alternative (Mudawar 2001, Child
2009, Kinney 2009) if it were not for drop bouncing and the Leidenfrost effect.
A simple remedy to the Leidenfrost effect is provided by polymer or metal-plated
nonwoven nanober mats of about 100 m thickness, created and deposited by electrospinning (Srikar et al. 2009b, Lembach et al. 2010, Sinha-Ray et al. 2011, 2013a,
Weickgenannt et al. 2011a, 2011b, Sinha-Ray and Yarin 2013,a). It was shown that
such nanotextured surfaces signicantly modify the outcomes of drop impact, practically
eliminating the receding motion of the contact line and bouncing on cold surfaces, as well
as water atomization due to boiling on hot surfaces. They also diminish the Leidenfrost
effect or even completely eliminate it. Drop pinning on electrospun nanotextured surfaces has an intuitive physical reason. Pore sizes in the mats are of the order of d 1 m,
whereas drop sizes are of the order of D 1001000 m. The motion of a massive drop
impacting onto a nanober mat with velocity V0 (of the order of 1 m s1) is abruptly
stopped by the surface. A part of its kinetic energy is redirected along the top surface as
the drop spreads out and is gradually converted into surface energy or dissipated due to
viscosity as the drop spreads. The rest of the drop kinetic energy is channeled into a few
pores of the nanotextured surface. Because the drop and pore sizes are incommensurate,
such channeling of the kinetic energy of a big drop into a few tiny pores results in
the initial velocity U of pore lling being much higher than V0, namely U (D/d)V0
(a detailed theory can be found in Lembach et al. 2010 and Weickgenannt et al. 2011a).
High values of U are akin to the high speeds of Munroe jets studied, in particular, by
Birkhoff et al. (1948) and widely used in ballistic penetration. They are much larger than
the wettability-related LucasWashburn velocity VLW = dcos/(8H) where and are
the surface tension coefcient and viscosity of water, respectively, and is the contact
angle (Washburn 1921, Levich 1962). Therefore, water can penetrate into the nanober
mat pores, irrespective of their wettability.
To compare wettability-related phenomena with dynamic penetration into nanober
mats, consider Figures 9.10 and 9.11. The advancing contact angle on a polyacrylonitrile
(PAN) nanober mat, a = 103, is signicantly larger than that on a cast sample
(3040), which is explained by the fact that air entrapped in the pores (about 9095%
porosity) facilitates hydrophobicity, even though the nanobers are made of a relatively
wettable material. This corresponds to a so-called CassieBaxter state. The receding
contact angle approaches zero, indicating very large hysteresis. A rolling-off angle for
water drops on a nanober mat practically does not exist. Drops of about 0.5 mm softly
deposited on nanober mats look spherical (Figure 9.10). They almost do not change
their shape when ipped upside down, and never detach from the surface. It should be
emphasized that Figure 9.10 shows that PAN nanober mats behave statically as hydrophobic materials (due to the presence of air in the pores and several submicron scales of
texture), even though cast PAN is partially wettable.
A totally different outcome was observed when a water drop of 2 mm diameter impacted
onto the same PAN nanober mat at a speed of 2 m s1. As Figure 9.11a shows, the drop
rst spreads on the nanober mat surface as on a dry, rigid, completely wettable substrate
and then remains pinned in the spread-out conguration and does not recede, which means
347 [337--358] 10.12.2013 7:51PM
Figure 9.10
347
Droplet deposited softly on partially wettable PAN nanober mat. Reprinted with permission from
Lembach et al. (2010). Copyright 2010, American Chemical Society.
that static hydrophobicity does not characterize dynamic behavior under the conditions of
drop impact, and cannot prevent water penetration into these nanober mats. The latter
means that dynamic transition from the CassieBaxter to the Wenzel state is possible for
such systems, even though the static transition is not. As a result of the dynamic penetration
of water into pores, a spread-out drop becomes pinned on nanotextured mats: it becomes a
circular millipede and never recedes or bounces back (see Figure 9.11).
The thermal consequences of such millipede-like behavior are illustrated in
Figure 9.12, where the optical and infrared images of two identical drops released from
the same height onto a bare, hot substrate (50-m thick stainless-steel foil) and onto the
same substrate covered by a PAN nanober mat are compared. At an initial temperature
of 60 C the drop on the nanober mat spreads much wider than on bare steel foil, cools a
much wider area to a lower temperature and evaporates much faster. The latter corresponds to a much higher heat ux being removed from the hot surface (Srikar et al.
2009b, Weickgenannt et al. 2011a).
But is pinning of a spread-out drop on a nanober mat sufcient for an anti-Leidenfrost
effect where drop levitation is eliminated, even on surfaces heated well above the boiling
temperature? The images in Figures 9.13 and 9.14 illustrate what happens. The panels in
Figure 9.13a show that, following impact on a bare steel substrate at an initial temperature of 220 C, the drop spreads, experiences perturbations due to capillary waves and
incipience boiling, but stays intact, and by t = 0.03 s fully levitates over the surface. At a
higher temperature of 300 C (Figure 9.13b), after impacting the bare stainless steel foil,
the drop shatters into a main secondary droplet and several tiny satellites. At t = 0.03 s the
main secondary droplet and satellites all levitate. Therefore, at both temperatures, drop
impacts onto the bare hot substrates depicted in Figures 9.13a and b follow the familiar
scenario of the Leidenfrost effect.
Nothing like that happens following drop impact onto the same substrate covered by
PAN nanobers containing carbon black (CB) nanoparticles (Figure 9.14). The water
348
348 [337--358] 10.12.2013 7:51PM
(a)
(b)
(c)
(d)
Figure 9.11
Different modes of drop impact onto a PAN nanober mat: (a) deposition, (b) ngering without
splash, (c) receding splash and (d) advancing splash. The time span is 1.5 ms, the drop diameter is
2 mm and the impact speed is 1.7 m s1 (a), 2 m s1 (b), 2.3 m s1 (c) and 2.7 m s1 (d). Reprinted
with permission from Lembach et al. (2010). Copyright 2010, American Chemical Society.
drop spreads out as before and experiences surface perturbations due to capillarity and
boiling. However, it does not recede or shatter and stays in full contact with the mat
surface and the substrate underneath like a millipede. This manifests the anti-Leidenfrost
effect of electrospun nanotextured surfaces, which allows the entire latent heat of
evaporation to be utilized for cooling.
However, this high-heat-ux cooling can be further facilitated by employing metalplated nanobers, which introduces the much higher thermal diffusivities of metals
compared to those of polymers. And this is still not all. Copper- and silver-plated
electrospun nanobers acquire very unique rough surfaces, which can be designed on
demand (Sinha-Ray et al. 2011).. Such surfaces can be rightfully called thorny devil
(Figure 9.15a) or cactus-like bers (Figure 9.15b) after a small Australian thorny devil
349 [337--358] 10.12.2013 7:51PM
349
60[C]
55
50
10 mm
(a)
45
t = 200 s
40
60 [C]
55
50
45
10 mm
(b)
Figure 9.12
t = 30 s
40
Drop evaporation on (a) a bare steel foil, and (b) on a steel foil covered with an electrospun PAN
nanober mat of thickness h = 1.05 mm. The optical images in (a) and (b) are on the left, the
corresponding infrared on the right. Reprinted with permission from Weickgenannt et al. (2011a).
Copyright 2011, by the American Physical Society.
5 mm
(a)
t = 0.2 ms
1.5 ms
5 ms
30 ms
4000 ms
1.5 ms
5 ms
30 ms
4000 ms
5 mm
(b)
Figure 9.13
t = 0.2 ms
The Leidenfrost effect emerging after a water drop impacts bare steel foil at: (a) 220 C and
(b) 300 C. Reprinted with permission from Weickgenannt et al. (2011a). Copyright 2011,
American Physical Society.
lizard or the familiar plants, both nding their habitats in deserts. The anti-Leidenfrost
effect of these metal-plated bers is fully visible in Figure 9.16 where, after impact on a
bare copper substrate, the water drop fully bounces back (and practically does not
provide any cooling), whereas it spreads, pins itself, penetrates the pores and reaches
the substrate on nanobers. In spray cooling, heat transfer is predominantly a surface
effect, which is facilitated by any increase in the contact surface. Nanober mats alone
serve this purpose reasonably well, whereas such fractal-like surfaces as those of thorny
devil or cactus-like bers additionally amplify the surface area enormously. As a result,
drop evaporation time tev on such metal-plated mats is reduced by more than two orders
of magnitude compared to that on polymer mats (tev is less than 50 ms in Figure 9.16 for
copper-plated bers versus 510 s for polymer bers). This then explains the tremendous
350
350 [337--358] 10.12.2013 7:51PM
5 mm
(a)
t = 0.2 ms
1.4 ms
5 ms
30 ms
500 ms
1.4 ms
5 ms
30 ms
60 ms
5 mm
t = 0.2 ms
(b)
Figure 9.14
The anti-Leidenfrost effect of a PAN+CB nanober mat of thickness h = 0.5 mm deposited on a

stainless steel foil. (a) 220 C and (b) 300 C. Reprinted with permission from Weickgenannt et al.
(2011a). Copyright 2011, American Physical Society.
(a)
(b)
Acc.V Spot Magn

5.00 kV 4.0 5000x
Figure 9.15
Det WD
SE 8.7
10 m
EMS
SEI
5.0kV
X1,600
10 m
WD 15.0 mm
Scanning electron microscope images of metal-plate bers: (a) one of the biggest individual
copper-plated thorny devil bers; (b) a layer of silver-plated nanobers. Reprinted with permission
cooling rate of about 0.6 kW cm2 recorded after water drop impacts onto 23-mm thick
copper substrates with a bonded 30-m thick copper thorny devil nanober mat at its
surface (Sinha-Ray et al. 2011).
The results show that electrospun nanotextured bers strongly adhere to hot surfaces, even at temperatures as high as 300 C. The presence of a nanober mat dramatically changes drop impact on a coated surface. After the initial impact and spreading of
the drop over the mat surface up to 35 times its initial diameter, the bottom of the
whole puddle, including its contact line, is pinned, while water practically instantaneously lls the pores. No receding motion, bouncing, or Leidenfrost levitation and
skittering over the surface are possible anymore. Even drop atomization, characteristic
of boiling on high-temperature surfaces is eliminated. As a result of the antiLeidenfrost effect of nanotextured surfaces, cooling rates of 0.6 kW cm2 can be
351 [337--358] 10.12.2013 7:51PM
Copper nanofibers
Bare copper
t = 0 ms
351
1 mm
Boiling water
18 ms
34 ms
50 ms
Figure 9.16
Water drop impact onto a bare 23-mm thick copper substrate (left) and the same substrate with a
bonded 30-m thick mat of copper thorny devil nanobers. In both cases the target temperature is
172.2 C. Reprinted with permission from Sinha-Ray et al. (2011). Copyright 2011, American
Chemical Society.
achieved, which holds great potential for further miniaturization of micro- and optoelectronic and radiological devices.
Spray cooling can result in surface ooding and the formation of an intact liquid layer
on the surface. This cooling regime is similar to pool boiling and it denitely diminishes
heat removal. Still, thorny devil nanobers at the surface reveal signicant useful
phenomena. The experimental and theoretical results of Jun et al. (2013) showed that
the heat ux and the heat transfer coefcient of boiling on nanotextured surfaces with
copper-plated nanobers were about 38 times higher than those on bare copper surfaces.
This stems from the fact that nanotextured surfaces promote bubble growth by increasing
the average temperature of uid surrounding growing bubbles. Hence, nanotextured
surfaces composed of copper-plated nanobers facilitate bubble growth rate and, thus,
increase the bubble detachment frequency. On the other hand, the critical heat ux (CHF)
for the nanotextured surfaces was found to be very close to its counterpart for the bare
copper surfaces. The critical superheating temperature of nanotextured surfaces was
about 10 C lower than that of corresponding smooth surfaces (Figure 9.17). This result
shows that nanotextured surfaces have signicant benets for surface cooling compared
to smooth surfaces.
Beyond the CHF, as usual, the heat ux decreases as the wall superheating temperature
increases, which corresponds to transitional boiling where the formation of a vapor lm
reduces the heat ux. In the transition region the heat ux for nanober mats diminishes
much more slowly than on the corresponding bare copper surface, as shown in
Figure 9.17. This is because the bubble departure frequency and the density of nucleation
sites on a bare copper surface are much lower than those on a nanober mat (Figure 9.18).
This implies that the nanober mat tends to prevent formation of the vapor lm.
352
352 [337--358] 10.12.2013 7:51PM
70
Bare surface
Nanofiber-covered surface
60
q(W/m2)
50
40
30
20
10
0
0
10
20
30
40
50
60
Tsat (C)
Figure 9.17
Nucleate boiling and transition boiling data for ethanol on a smooth bare copper surface and a
nanober-covered surface. Heat ux versus surface superheating temperature. Reprinted from Jun
(a)
(b)
Bare copper
(c)
(d)
Bare copper
Figure 9.18
Nanofiber mats
Nanofiber mats
Images of pool boiling of water on bare surfaces and surfaces coated with nanober mats at a
hotplate temperature of 200 C in panels (a,b) and 400 C in panels (c,d). Reprinted from Jun et al.
(2013), with permission from Elsevier.
353 [337--358] 10.12.2013 7:51PM
353
9.4 Nanofluidics
9.4
Nanofluidics
Parallel arrays of carbon nanotubes produced by co-electrospinning and shown in
Figure 5.58 in Chapter 5 were installed in nanoudic devices by Bazilevsky et al. (2008)
and Srikar et al. (2009a). The experimental setup used is shown in Figure 9.19. In addition,
Sihna-Ray et al. (2009) and Sinha-Ray and Yarin (2010) used parallel arrays of electrospun
nanobers as sacricial templates to form parallel nanochannels in carbon blocks.
A typical carbon tube bundle was about 1 cm long and 200 m in diameter. It was
inserted into the open tip of a glass capillary (of 1 mm outer diameter and 0.58 mm inner
diameter) and secured there by epoxy cement (Figure 9.20). The epoxy was placed in a
manner that did not interfere with the inlet/outlet of any tube in the bundle, but penetrated
into the gaps between the tubes at the application location. Thus, the capillary orice was
effectively sealed, and uid could exit only through the tube bundle. The uid supply
system consisted of two standard plastic syringes of volumes 10 and 1 mL, a digital
pressure meter, a 7-mL plastic air chamber and two stopcocks, shown in Figure 9.19. All
parts of this apparatus were connected by thin silicon tubing. The chamber was used to
adjust air pressure smoothly and maintain it at a nearly constant level during a test. After
pressurizing the air chamber by means of the 10-mL syringe (on the left in Figure 9.19),
the one-way stopcock was closed. To reach higher pressure, this pressurizing step was
repeated several times. The pressurized chamber sustained nearly constant pressure
during long ow tests, even though air was owing through the open three-way stopcock
into the glass capillary and then into the carbon tube bundle. The second syringe
(vertically placed in Figure 9.19), which was connected directly to the three-way stopcock, was used only in the liquid-ow experiments to ll (with a liquid) the silicon pipe
leading to the glass capillary. This simple system allowed adjustable operational pressure
(in excess with respect to atmospheric) up to 4 bar at the carbon tube bundle entrance.
To visualize discharge of gas (air) or liquid (n-decane) from the carbon tube bundle, the
tip of the glass capillary was immersed in a shallow pool of water and observed under a
Pressure gauge
PC
Three-way
stopcock
Stopcock
CCD
Glass
capillary
Air chamber
Figure 9.19
Sketch of the experimental setup. Bazilevsky et al. (2008). Reproduced with permission from The
354
354 [337--358] 10.12.2013 7:51PM
microscope (Figure 9.19). As air was being discharged from the bundle tip, macroscopic
merger bubbles were visible in the water (Figure 9.20a). When n-decane, which is immiscible with water, was being discharged, a growing merger droplet was clearly visible in the
water pool (Figure 9.20b). In the n-decane-in-water case, a situation was also encountered
Epoxy
Water
CNT bundle
Water
1 mm
1 mm
Bubble
(a)
N-decane
droplet
(b)
N-decane
Water
Bubble
1 mm
(c)
Gas bubble
Epoxy plug
1 mm
N-decane
droplet
(d)
1 mm
(e)
Figure 9.20
(a) Growing merger air bubble at the tip of the carbon tube bundle immersed in a shallow water pool
(Figure 9.19). The epoxy secures the conduit bundle at the end of the glass capillary. (b) Growing
merger n-decane droplet in a water pool. (c) Growing merger n-decane droplet with a gas bubble
growing inside it. (d) Gas bubble trapped inside the glass capillary upstream of the solid epoxy
plug. (e) This image was captured after the gas bubble shown in d was completely depleted.
Bazilevsky et al. (2008). Reproduced with permission from The Royal Society of Chemistry.
355 [337--358] 10.12.2013 7:51PM
9.4 Nanofluidics
355
where the merger n-decane droplet contained a bubble, presumably consisting of air
(Figure 9.20c). In this case, gas was entrained from a bubble entrapped upstream in the
glass capillary in the region inside the capillary where the epoxy plug ended (Figure 9.20d);
this gas bubble was eventually depleted (Figure 9.20e). The discharge process was observed
and recorded at room temperature (2021oC) using a simple stereo-zoom microscope
(1040, Edmund Scientic) with a CCD camera (Figure 9.19).
Sinha-Ray et al. (2009) showed experimentally and theoretically that bi-layer ows of
liquid and gas in an individual nanochannel (Figure 9.20d) can result in an over-limiting
regime. Then, a higher ow rate of liquid can appear, compared to the case when the same
liquid ows through the same nanotube under the same pressure drop, but occupies
the whole bore. As a consequence, it is possible to release more liquid than predicted by
the Poiseuille law, in spite of the fact that in the bi-layer ow liquid does not occupy the
whole bore. This paradoxical result stems from the fact that the low-viscosity gas layer
ows much faster than the liquid layer in contact with it. The gas layer applies signicant
shear stress to the liquid layer and entrains it. This is effectively a giant quasi-slip
phenomenon that happens in relatively large nanotubes (of the order of 500 nm) where
the no-slip condition denitely holds.
A high-throughput nanouidic method of producing stimuli-responsive drug-containing
nanoparticles with a narrow diameter distribution in the therapeutically important range
400600 nm was proposed in Sinha-Ray and Yarin (2010). They used the nanouidic setup
shown in Figures 9.19 and 9.21. A 6 wt% solution of surfactant (Brij-30) in n-decane with
2 vol% of the poly(N-isopropylacrylamide) (PNIPAM) polymerization reaction accelerator
(N,N,N,N-tetramethylethylenediamine, 99%) was released through the nanochannels
into a petri dish lled with an aqueous solution of NIPAM monomer, crosslinker BIS
(N,N-methylenebisacrylamide), [2-(methacryloyloxy) ethyl]trimethyl ammonium chloride
(METAC), 75% (to facilitate NIPAM polymerization at room temperature) and the initiator
APS (ammonium persulfate). The presence of surfactant prevented formation of large
merger droplets (shown in Figure 9.20b). It was expected that as soon as n-decane droplet
Water + monomer +
crosslinker + initiator
+ polymerization
temp. reducer
Oil + dye + surfactant +

accelerator
Carbon strip
with thousands
of parallel
nanopores
Magnetic
stirrer
Figure 9.21
Polymerization of PNIPAM nanoparticles using carbon nanochannels of diameter 300500 nm.

Reprinted with permission from Sinha-Ray and Yarin (2010). Copyright 2010, AIP Publishing LLC.
356
356 [337--358] 10.12.2013 7:51PM
Dye particles
Brij-30
Droplet of n-decane
and accelerator
Polymerized nanoparticles
with dye in it
Aqueous solution of
NIPAM + BIS + METAC + Initiator
(a)
(b)
(c)
Figure 9.22
Sketch of formation of PNIPAM nanoparticles: (a) n-decane nanodroplet containing dye,

surfactant and accelerator delivered into an aqueous solution of NIPAM, BIS, METAC and the
initiator; (b) formation of PNIPAM nanoparticles; (c) collected PNIPAM nanoparticles after ltration.
Reprinted with permission from Sinha-Ray and Yarin (2010). Copyright 2010, AIP Publishing LLC.
Figure 9.23
TEM images of dye-containing PNIPAM nanoparticles. Reprinted with permission from

Sinha-Ray and Yarin (2010). Copyright 2010, AIP Publishing LLC.
clusters containing the accelerator supplied through carbon nanochannels came in contact
with the aqueous solution of NIPAM monomers and the initiator in the petri dish, polymerization of PNIPAM would proceed at the droplet/cluster surface, encapsulating n-decane in
a PNIPAM shell and forming coreshell nanoparticles (Figure 9.22). A uorescent dye
(rhodamine 6G), used as a model drug for release purposes, was supplied in n-decane
through carbon nanochannels. It was expected that the dye would also be encapsulated in
the nanoparticle core. Most probably, it would also diffuse into the shell (Figure 9.22).
After 5 min of release, the resulting solution of nanoparticles from the petri dish was
ltered through a 200- nm lter paper and dried. TEM images of the PNIPAM nanoparticles with the embedded uorescent dye formed by this method are shown in
Figure 9.23. They are all submicron in size and some possess a well-dened coreshell
357 [337--358] 10.12.2013 7:51PM
9.5 References
357
structure, while others contain multiple small cores. The production rate was evaluated to
be of the order of 107 nanoparticles per second. The diameters of the PNIPAM nanoparticles are in the range 100600 nm, with the average values close to 300400 nm, as
measured by dynamic light scattering (DLS).
9.5
References
Barenblatt, G. I., Entov, V. M., Ryzhik, V. M., 1990. Theory of uid ows through natural rocks.
Kluwer, Dordrecht.
Bazilevsky, A. V., Yarin, A. L., Megaridis, C. M., 2008. Pressure-driven delivery through carbon
tube bundles, Lab on a Chip 8, 152160.
Birkhoff, G., Macdougall, D. P., Pugh, E. M., Taylor, G. I., 1948. Explosives in lined cavities.
J. Appl. Phys. 19, 563582.
Cavaliere, S., Subianto, S., Savych, I., Jones, D. J., Rozi, J., 2011. Electrospinning: designed
architectures for energy conversion and storage devices. Energy Environ. Sci. 4, 47614785.
Chen, S., Hou, H., Harnisch, F., Patil, S., Carmona-Martinez, A. A., Agarwal, S., Zhang, Y., SinhaRay, S., Yarin, A. L., Schroder, U., Greiner, A., 2011. Electrospun and solution blown carbon
nanober nonwovens for application as electrodes in microbial fuel cells. Energy Environ. Sci. 4,
14171421.
Child, J., 2009. FPGA boards and systems boost UAV payload compute density. COTS J. 2, 14.
Dimesso, L., Spanheimer, C., Jaegermann, W., Zhang, Y., Yarin, A. L., 2012. LiFePO43 D carbon
nanober composites as cathode materials for Li-ions batteries. J. Appl. Phys. 111, 064307.
Dimesso, L., Spanheimer, C., Jaegermann, W., Zhang, Y., Yarin, A. L., 2013. LiCoPO43 D carbon
nanober composites as possible cathode materials for high voltage applications. Electrochim.
Acta 95, 3842.
Fundamentals and Applications in Separation and Filtration Processes. Begell House, New York.
Jun, S., Sinha-Ray, S., Yarin, A. L., 2013. Pool boiling on nano-textured surfaces. Int. J. Heat Mass
Transf. 62, 99111.
Kim, J., Jasper, W. J., Hinestroza, J. P., 2006. Charge characterization of an electrically charged
ber via electrostatic force microscopy. J. Eng Fibers Fabrics 1, 3046.
Kinney, D., 2009. UAVs embrace the benets of direct spray cooling. COTS J. 2, 710.
Kumar, B., Asadi, M., Pisasale, D., Sinha-Ray, S., Rosen, B., Haasch, R., Abiade, J., Yarin, A. L.,
Salehi-Khojin, A. 2013. Renewable, metal-free and non-precious carbon nanober catalysts for
CO2 reduction. Nat. Commun. (in press).
Lembach, A., Tan, H. B., Roisman, I. V., Gambaryan-Roisman, T., Zhang, Y., Tropea, C., Yarin, A. L.,
2010. Drop impact, spreading, splashing and penetration in electrospun nanober mats. Langmuir
26, 95169523.
Levich, V. G., 1962. Physicochemical Hydrodynamics. Prentice Hall, Englewood Cliffs, New Jersey.
Maze, B., Wang, Q., Tafreshi, H. V., Pourdeyhimi, B., 2007. A simulation of unsteady state
ltration via nanober media at reduced pressures. J. Aerosol Sci. 38, 550571.
Mudawar, I., 2001. Assessment of high-heat-ux thermal management schemes. IEEE Trans.
Comp. Pack. Tech. 24, 122141.
Rioboo, R., Tropea, C., Marengo, M., 2001. Outcomes from a drop impact on solid surfaces.
Atomiz. Sprays 11, 155165.
358
358 [337--358] 10.12.2013 7:51PM
Sinha-Ray, S., Chando, P., Yarin, A. L., 2009. Enhanced release of liquid from carbon nanotubes
due to entrainment by air layer. Nanotechnology 20, 095711.
Sinha-Ray, S., Sinha-Ray, S., Yarin, A. L., Weickgenannt, C. M., Emmert, J., Tropea, C. 2013.
Drop impact cooling enhancement on nano-textured surfaces. Part II: Results of the parabolic
ight experiments [zero gravity (0 g) and supergravity (1.8 g)]. Int. J. Heat Mass Transf.
(in press).
Sinha-Ray, S., Yarin, A. L., 2010. Flow from macroscopically long straight carbon nanopores for
generation of thermo-responsive nanoparticles. J. Appl. Phys. 107, 0294903.
Sinha-Ray, S., Yarin, A. L., 2013. Drop impact cooling enhancement on nano-textured surfaces.
Part I: Theory and results of the ground (1g) experiments. Int. J. Heat Mass Transf. (in press).
Sinha-Ray, S., Zhang, Y., Yarin, A. L., 2011. Thorny devil nano-textured bers: The way to cooling
rates of the order of 1 kW/cm2. Langmuir 27, 215226.
Srikar, R., Gambaryan-Roisman, T., Steffes, C., Stephan, P., Tropea, C., Yarin, A. L., 2009b.
Nanober coating of surfaces for intensication of spray or drop impact cooling. Int. J. Heat
Mass Transf. 52, 58145826.
Srikar, R., Yarin, A. L., Megaridis, C. M., 2009a. Fluidic delivery of homogeneous solutions
through carbon tube bundles, Nanotechnology 20, 275706.
Wang, J., Kim, S. C., Pui, D. Y. H., 2008. Investigation of the gure of merit for lters with a single
nanober layer on a substrate. J. Aerosol Sci. 39, 323334.
Washburn, E. W., 1921. The dynamics of capillary ow. Phys. Rev. 17, 273283.
Weickgenannt, C. M., Zhang, Y., Lembach, A. N., Roisman, I. V., Gambaryan-Roisman, T.,
Yarin, A. L., Tropea, C., 2011a. Non-isothermal drop impact and evaporation on polymer
nanober mats. Phys. Rev. E 83, 036305.
Weickgenannt, C. M., Zhang, Y., Sinha-Ray, S., Roisman, I. V., Gambaryan- Roisman, T.,
Tropea, C., Yarin, A. L., 2011b. The inverse-Leidenfrost phenomenon on nanober mats on
hot surfaces. Phys. Rev. E 84, 036310.
Yarin, A. L., 2006. Drop impact dynamics: splashing, spreading, receding, bouncing. . . . Ann. Rev.
Fluid Mech. 38, 159192.
Yarin, A. L., Chase, G. G., Liu, W., Doiphode, S. V., Reneker, D. H., 2006. Liquid drop growth on a
ber. AIChE J. 52, 217227.
359 [359--379] 10.12.2013 7:57PM
10 Military applications
of micro- and nanofibers
Chemical warfare differs from conventional warfare, such as explosives, in the sense
that the toxicity of the chemicals used is very dangerous to the combatants in the
battleeld, or to the civil population being affected during a war or as a result of a
terrorist attack. The threat is amplied by concealment and low cost and relatively
easy production of these agents, which renders them weapons of mass destruction. It
becomes essential in such a situation that soldiers wear protective clothing, respirators,
face masks, gloves, etc. Initially German troops unleashed chemical warfare agents,
such as mustard agent, HD, during World War I in 1915, and this resulted in much more
concern during World War II. Thus research on decontamination of chemical and
biological warfare agents (CWAs) has been the subject of serious interest in many
research laboratories since World War II. Mustard gas or bis(2-chloroethyl) sulde is
one of the highly cytotoxic, readily methylating CWAs that are regulated under the
1993 Chemical Weapons Convention (CWC). It covalently binds to DNA bases or
forms disulde bonds with the thiol groups. This readily leads to programmed cell
death or the mutated DNA leads to cancer. The simultaneous danger associated with
this toxic agent is its high skin penetration owing to its high lipophilicity, which results
in immediate blister formation and huge uptake through skin within a short duration of
exposure (Ivarsson et al. 1992).
Conventional modes of decontamination involve activated charcoal or other such
heavy physical adsorbants of the contaminant. When compared with technologies
employing conventional catalysts and reactive sorbents, which are specic and sometimes time-consuming, nanotechnology-based nanomaterials are highly reactive, nonselective and multifunctional in character. Current development of nanoparticles such as
MgO, Al2O3, Fe2O3, ZnO and TiO2 and their incorporation into nanobers impart
enhanced catalytic, disinfection and sensing capabilities, photo-protection capability,
and stain-resistance and self-cleaning properties. The testing of these materials in textiles
and protective clothing against CWAs showed that they have potential as replacement
technology in such applications.
Protective clothing can be divided broadly into two types: a nonpenetrable polymer
design and a partially permeable charcoal-lined design. The nonpenetrable polymer
garment is made up of a self-contained breathing apparatus, which is able to protect
the wearer for 23 h of continuous use. In the case of the activated carbon-based partially
permeable design, such as the US Joint Service Lightweight Integrated Suit Technology
(JSLIST), a layer of charcoal, capable of adsorbing the toxic vapors of chemical warfare
360
360 [359--379] 10.12.2013 7:57PM
10 Military applications of micro- and nanofibers
agents, is sandwiched between two layers of fabric. The charcoal adsorbant used has a
surface area of more than 1000 m2 g1, and is impregnated with Cu, Ag, Zn, Mo and
triethylenediamine (Morrison 2002). Although the JSLIST suit is relatively lightweight,
contains activated carbon spheres and has a protection duration of 24 h, it is still too
heavy (about 2.6 kg) and creates heat stress. Also, most of the toxic chemical vapors are
physisorbed and not dismantled (only a few gases are decomposed), thereby creating
problems in disposal after use. The other signicant disadvantage to highlight here is that
the pores present in activated charcoal are blocked when used in humid or moist environments, and thereby the performance of the material is reduced.
Nanober technology has demonstrated its usefulness in day-to-day life and this
chapter summarizes its applications as a chemical- and bio-decontamination tool. The
chemical and biological warfare decontamination applications of these nanobers are
discussed in Sections 10.1 and 10.2, respectively, based on the end-user view point.
Such applications as protective clothing materials, decontamination wipes, sensors and
respirator masks are covered in Sections 10.310.6, respectively.
10.1
Nanofibers and chemical decontamination

The protective garment should possess the following benecial properties: (i) it should
react with chemical and biological contaminants and transform them into nontoxic
products; (ii) it should be lightweight; (iii) it should work well in all kinds of climates
and (iv) it should maximize the wearers comfort. Polymer nanobers are a good
candidate as a niche material for such applications due to their attractive features:
(i) they can act as a barrier material to lter microorganisms and ne particles; (ii) they
have good strength; (iii) they have a high surface area for enhanced sensing capability.
The uses of polymer nanobers with active chemicals/catalysts not only help in weight
reduction, but also help to increase the effective or available surface area for functionalization. This has two advantages: (i) greater adsorption (and enhanced decomposition)
of chemical and biological warfare agents and (ii) extended life of the media.
Nanobers are produced by a simple and versatile electrospinning technique. A typical
electrospinning set up used to fabricate nanobers is shown in Figure 5.1 in Chapter 5.
There are three main groups of parameters that affect the formation of bers from the
polymer solutions in electrospinning: (i) polymer properties (molecular weight, chemical
structure), (ii) solution properties (solvent, surface tension, viscosity, dielectric constant,
electrical conductivity, polymer concentration), (iii) processing conditions (applied voltage, needle-collector distance, polymer ow rate, temperature). The nanobers produced
from the electrospinning process have unique properties, such as high surface-area- tovolume ratio, tailorable pore sizes and exibility in the surface chemistry of the nanobers. Hence, they have been tested as textiles in protective clothing, as membranes in
liquid ltration (microltration, ultraltration and nanoltration) and as scaffolds in
tissue engineering applications. As such, the nanobers have the capability of ltering
out biological warfare agents. Nanobers, after incorporation of nanoparticles, chemicals
and catalysts, have been tested for decontamination of CWAs. It has been observed by
361 [359--379] 10.12.2013 7:57PM
10.1 Chemical decontamination
361
1010
Darcy flow resistance (m1)
PBI Electrospun
PTFE Microporous membrane (wind barrier)
109
PAN Electrospun
108
Tightly woven cotton fabric,

which shows humidity dependent permeability
Knit polyester fabric
107
Nonwoven batting (thermal insulation layer)

106
Figure 10.1
.2
.4
.6
Relative humidity
.8
1.0
High convective gas-ow resistance of microporous membranes and electrospun nonwovens.

Gibson et al. (1999). Flow resistance is dened as the ratio of the membrane thickness to its
hydraulic permeability, which is proportional to the pressure drop divided by the volumetric ow
rate, according to Darcys law. Courtesy of John Wiley & Sons. Copyright 1999, American Institute
of Chemical Engineers (AIChE).
Gibson et al. (1999) that the gas ow resistance of nanobers is higher than that of woven
fabrics, as shown in Figure 10.1. In the case of nanobrous materials, gas ow also varies
with the nature of the materials.
A hydrophobic polymer such as polybenzimidazole shows a higher magnitude of gasow resistance than a hydrophilic ber, such as polyacrylonitrile (Figure 10.1). On the
other hand, electrospun membranes show lower resistance to water vapor diffusion when
compared with commercial laminates (Figure 10.2). This is because of the difference in
the nature of the pores present in these materials.
10.1.1
Metal oxides embedded in nanofibers for toxic chemical decontamination

Toxic materials are adsorbed and decomposed on the surfaces of metal oxide nanoparticles such as MgO, ZnO, TiO2, CaO, Al2O3 and Fe2O3. The decomposition ability
of metal oxides is characterized in Figure 10.3. The catalytic performances of various
adsorbents such as aerogel-prepared magnesium oxide (AP-MgO) and conventionally
prepared magnesium oxide (CP-MgO) are compared with a bulk sample (commercially available magnesium oxide, CM-MgO). Adsorbent (100 or 200 mg) and pentane
(100 or 200 mL) was placed in a round bottom ask. To this, paraoxon (4.5 or 9 L)
was added and its disappearance was monitored by UV/Vis spectroscopy. The paraoxon from the solution was completely adsorbed on the AP-MgO sample and hence
lower absorbance was noticed. The high reactivity of AP-MgO mainly arises due to
the large surface area, the presence of high concentrations of reactive sites and
stabilized lattice planes (Rajagopalan et al. 2002). Figure 10.3 shows that the catalytic
362
362 [359--379] 10.12.2013 7:57PM
Water vapor diffusion resistance (s m1)
Commercial membrane laminate (Sympatex)
105
Commercial membrane laminate (Gore-Tex)

Polyacrylonitrile electrospun membrane
PTFE microporous membrane
104
Polybenzimidazole electrospun membrane
103
102
101
100
.2
Figure 10.2
.4
.6
Mean relative humidity
.8
Excellent water vapor transport properties of electrospun nonwovens. Gibson et al. (1999). Water
vapor diffusion resistance is dened as the ratio of the membrane thickness to the vapor diffusion
coefcient, which is proportional to the vapor concentration difference divided by the diffusion
mass ux, according to Ficks law. Courtesy of John Wiley & Sons. Copyright 1999, American
Institute of Chemical Engineers (AIChE).
Blank
2.0
CM-Mgo
Absorbance
1.5
1.0
AP-MgO (0.05 g)
CP-MgO
0.5
AP-MgO
0.0
0
Figure 10.3
20
40
60
Time/min
80
100
120
Adsorption of paraoxon (4.5 L) on magnesium oxide (0.1 g) dispersed in pentane, monitored by

UV/Vis spectroscopy (267 nm band). AP-MgO is aero-gel prepared MgO, CP-MgO is
conventionally prepared MgO, CM-MgO is commercial MgO, and Blank represents samples with
no addition. Rajagopalan et al. (2002). Copyright Wiley-VCH Verlag GmbH & Co. KGaA.
Reproduced with permission.
363 [359--379] 10.12.2013 7:57PM
10.1 Chemical decontamination
363
3.5
3.0
Act. Carbon S.A. = 900 m2 g1

AP-MgO S.A. = 400 m2 g1
Absorbance
2.5
2.0
Act. Carbon, 10 L
1.5
AP-MgO, 20 L
1.0
AP-MgO, 10 L
0.5
0.0
Act. Carbon, 4.5 L

AP-MgO, 4.5 L
0
Figure 10.4
20
40
60
Time/min
80
100
120
Adsorption of paraoxon on AP-MgO (0.1 g) compared with activated carbon (0.1 g) dispersed in
pentane, monitored by UV/Vis spectroscopy (267 nm band). Rajagopalan et al. (2002). Copyright
Wiley-VCH Verlag GmbH & Co. KGaA. Reproduced with permission.
performance of metal oxide nanoparticles (AP-MgO) is higher than the bulk material
(CM-MgO). The mechanism of degradation for metal oxides is found to be destructive
adsorption, whereas mere adsorption is taking place on the surface of conventional
charcoal. This is reected in the rate and amount of toxic-material decontamination,
which is shown in Figure 10.4. Ramakrishnan and Ramakrishna (2007) have reported
the potential of electrospun ceramic nanobers as a good destructive adsorbent for the
detoxication of a nerve agent simulant (dimethyl methylphosphonate) and a mustard
agent simulant (chloroethyl ethyl sulphide). Metal oxide nanobers were fabricated by
the combination of two conventional techniques, such as electrospinning and sol-gel.
Metal oxide precursors were mixed with the polymer solution and then subjected to
electrospinning, which upon further calcination at higher temperatures removed the
polymer and then produced metal oxide nanobers. Zinc titanate nanobers proved to
possess excellent decontamination ability over other mixed metal oxides such as
AlTiO3, MgTiO3, CuMo4, etc. (Ramakrishnan and Ramakrishna 2007). This study
shows that ceramic materials are potential destructive adsorbents of both simulant
types when compared to organic catalysts, which are mostly effective with only one
type of simulant.
MgO and Al2O3 nanoparticles incorporated into activated carbon nanobers were
shown to decontaminate nerve gas, 2-chloroethyl ethyl sulde (2-CEES). A linear
increase in decontamination of nerve gas was observed with increase in the amounts of
embedded metal oxide nanoparticles, with the surface area of the incorporated nanoparticles also playing a crucial role. Metal oxide nanoparticles with their surface reactive
oxide/hydroxyl groups form covalent/alkoxide bonds between 2-CEES and thus destroy
364
364 [359--379] 10.12.2013 7:57PM
Disappearance of 2-CEES (%)
100
0.10 mass fraction MgO Plus in ACNFs

0.10 mass fraction Al2O3 Plus in ACNFs
90
ACNFs
80
0.10 mass fraction MgO Plus in PAN-NFs
70
0.10 mass fraction Al2O3 Plus in PAN-NFs
60
50
40
30
20
10
0
0
30
60
90
120
150
180
210
240
Time (min)
Figure 10.5
2-chloroethyl ethyl sulde (2-CEES) decontamination abilities of MgO and Al2O3 nanoparticles
embedded in activated carbon nanobers. Reprinted with permission from Dadvar et al. (2012).
the adsorbed toxicants (Figure 10.5). The formation of covalent bonds and decontamination was studied with by FTIR (Dadvar et al. 2012).
10.2
Nanofibers for biowarfare decontamination

Silver ions are well known for their potent inhibitory and bactericidal action, as well as a
broad spectrum of antibacterial activities. Hence, there are many commercial applications of silver as a wound-protecting material against bacterial or microbial infections.
The exact mode of action of silver ions with bacteria or microbes is yet to be made clear,
but the general antimicrobial mechanism of action (Figure 10.6) of nanosilver includes:
(i) the release of silver ions and further generation of reactive oxygen species (ROS); (ii)
the interaction of ROS with cell membrane proteinous material, thus affecting their
routine function; (iii) the accumulation of nanosilver in the cell membrane, affecting
cell membrane permeability and (iv) proliferation into the bacterial cell where ROS are
generated, and silver ions are released and affect DNA function by means of disruption of
ATP production and DNA replication. By these multiple mechanisms the silver kills or
deactivates the bacteria (Marambio-Jones and Hoek 2010). The broad-spectrum antibacterial activity of poly(vinyl alcohol)/ polyethyleneimine (PVA/PEI) blend and Ag
nanoparticulate composite membranes has been demonstrated, with their antibacterial
performance against Gram-negative Escherichia coli (E. coli) and Pseudomonas aeruginosa (P. aeruginosa), and Gram-positive Staphylococcus aureus (S. aureus) and
Bacillus subtilis (B. subtilis). PVA/PEI/Ag nanohybrids were prepared by an in situ
reduction method using silver nitrate, PVA and PEI as precursor, linker and reductant,
365 [359--379] 10.12.2013 7:57PM
10.2 Biowarfare decontamination
365
H+ +
H H+
H+ H+
ROS
PO43
ROS
Nanoscale silver
Silver ion
Membrane protein
Figure 10.6
Diagram summarizing nanoscale silver interaction with bacterial cells. (Marambio-Jones and
Hoek 2010), with kind permission from Springer Science+Business Media.
respectively. All forms of silver antibacterials are sources of silver ions (Ag+) and slowly
release Ag+ ions over time, inhibiting bacteria. Neutral silver in its zero-valent form is
inert (Guggenbichler et al. 1999), but releases silver ions upon treatment with moisture
(Radheshkumar and Munstedt 2006). Silver nanoparticle-impregnated PVA/PEI membranes exhibited moisture vapor permeability, as indicated in Figure 10.7 and evident
from reduction in water content with Ag nanoparticle incorporation. Further the moisture
permeability resulted in strong antibacterial activity (Figure 10.8) against Gram-positive
and Gram-negative S. aureus, B. subtilis and E. coli bacteria (Wu et al. 2009). The zone
of bacterial inhibition clearly showed the high percentage of microbial inhibition with all
four bacterial strains tested. Thus the PVA/PEIAg nanohybrid is ideal to retain moisture
vapor permeability together with enhanced action against biological pathogens.
Zhang et al. (2013) showed that solution-blown soy protein nanober mats decorated
with silver nanoparticles possess signicant antibacterial activity against E. coli colonies
without exposure to UV light (Figure 10.9). Soy protein/nylon 6 solution-blown nanobers were decorated with silver nanoparticles as follows. Two different AgNO3 solutions in water were prepared: 4 wt% (solution A) and 1 wt% (solution B). Collected soyprotein-containing nanobers were immersed in solutions A or B and then left at room
temperature for 24 h to dry. Then, two different post-treatments of the solution-blown
samples were used. In the rst, nanobers soaked in solutions A or B were subjected to a
reducing agent solution (1 wt% NaBH4 solution in water). This was done to reduce
AgNO3 to silver nanoparticles embedded in nanobers. After the addition of NaBH4
solution to the nanobers, the samples were left at room temperature to dry out completely. The post-treated nanobers thus obtained are denoted as Ar and Br in
Figure 10.9. As the second post-treatment, solution-blown nanobers soaked in solution
366
366 [359--379] 10.12.2013 7:57PM
(a)
(b)
(c)
Blank
Figure 10.7
Polymer
Polymer Polymer
/Ag-1
/Ag-3
Polymer
/Ag-5
Polymer
/Ag-8
(a) Images characterizing the moisture vapor permeability of a polymer/Ag hybrid lm and a
control impermeable membrane (blank), (b) the moisture vapor transmission surface, (c) the results
of moisture vapor permeability under 32 C, 48 h. Reprinted from Wu et al. (2009), with permission
from Elsevier.
A were heat-treated at a temperature of 120 C to thermally decompose AgNO3 to

embedded silver nanoparticles. These post-treated samples are denoted as Ah in
Figure 10.9. The results show that antibacterial activity against E. coli signicantly
depended on the pre-treatment type.
Apart from nanosilver antibacterials, guanidine-based polymers were also reported
to exhibit antibacterial activity (Oule et al. 2008). Polyhexamethylene guanidine
hydrochloride (PHGH), an antimicrobial biocide polymer belonging to the guanidine
family, has been shown to be efcacious against S. aureus, P. aeruginosa, Salmonella
choleraesuis (S. choleraesuis), methicillin-resistant S. aureus (MRSA) and E. coli.
An electrospun polyacrylonitrile (PAN) nanober surface was modied with polyhexamethyleneguanidine hydrochloride (PHGH) so as to provide antibacterial properties
(Mei et al. 2012). This creates a surface capable of killing adherent bacteria and further
repelling the dead bacteria off the surface. Chemically the nitrile groups were reduced
to amino groups on the PAN nanober surface and then modied with PHGH. The
surface modications from nitrile functionality and the morphology were ascertained
by scanning electron microscopy (SEM), attenuated total-reectance Fouriertransform infrared spectroscopy (ATR/FTIR) and energy-dispersive spectroscopy
(EDS). It was observed that PHGH-immobilized nanobrous membranes possessed
367 [359--379] 10.12.2013 7:57PM
(a)
(e)
(a)
(b)
(c)
(d)
(e)
(b)
(a)
(a)
(b)
E. Coli
Figure 10.8
(b)
(e)
(d)
(c)
(d)
B. subtilis
S. aureus
(e)
367
(c)
(d)
(c)
P. aeruginosa
Images of the antibacterial test results of the microbes listed in systems of: (a) PVA/PEI, (b) PVA/
PEI/Ag-1, (c) PVA/PEI/Ag-3, (d) PVA/PEI/Ag-5 and (e) PVA/PEI/Ag-8. Quantitative AgNO3
(1, 3, 5 and 8%) was added to 100 ml of an aqueous dispersion solution containing 5 wt% PEI.
Reprinted from Wu et al. (2009), with permission from Elsevier.
antibacterial activity against both Gram-positive S. aureus and Gram-negative E. coli.

Further the PHGH-immobilized nanobrous membranes possessed highly effective
and prolonged antibacterial activities, even after three cycles of antibacterial interaction. Because of the hydrophilic glycerol diglycidyl ether spacer-functionalized membrane surface it also exhibited signicant easy-cleaning performance (Mei et al. 2012).
Electrospun CuO/TiO2 nanobers were fabricated by Yousef et al. (2012) and showed
prominent visible-light-mediated photocatalytic activity degrading reactive black 5
and reactive orange 16 dyes in aqueous solutions. The composite nanober membrane
effectively catalyzed a Klebsiella pneumoniae inactivation process. After exposure of
K. pneumoniae to the nanobers, the destruction of the cell wall and the cell membrane
was visualized through TEM analysis. The recyclability of the introduced photocatalyst with the same original photocatalytic activity was ascertained, together with
synergistic enhancement of photocatalytic activity. The same synergy accounts for
increased antimicrobial potential that impeded the growth of K. pneumoniae and
ultimately caused bacterial death (Yousef et al. 2012). An et al. (2009) successfully
prepared chitosan (CS)/ poly(ethylene oxide) (PEO) ultrane bers incorporating
silver nanoparticles (AgNPs) by electrospinning, to show the antibacterial activity of
the nanober membrane. In situ chemical reduction of Ag ions was achieved by
368
368 [359--379] 10.12.2013 7:57PM
1.32 cm2
1.08 cm2
Ar
Ah
Br
0.96 cm2
Figure 10.9
Inhibition of E. coli by silver-decorated soy-protein nanober mats. The areas of the clear domains
around nanober mat samples 1, 2 and 3 were 1.32, 1.08 and 0.96 cm2, respectively. No clear
domain and thus no inhibition is seen around the control nanober mat 4 (without silver). Reprinted
from Zhang et al. (2013), with permission from Elsevier.
embedding silver ions in CS/PEO solutions, and Ag/CS nano-colloids were obtained.
The electrospun ultrane bers containing the inorganic nanoparticles were characterized with XRD to show the presence of AgNPs, and Ag nanoparticles of size less
than 5 nm were observed under a transmission electron microscope. The complexation
interaction between Ag and CS was studied with X-ray photoelectron spectroscopy,
which suggested that there exist AgO bonds in the composite ultrane bers that
result in a tight combination of Ag and CS. Antimicrobial activity evaluation of the
electrospun Ag/CS/PEO brous nanocomposite membranes against E. coli showed
that the AgNPs in the ultrane bers signicantly increased the antibacterial activity
(An et al. 2009).
A dendrimer-based nanohybrid incorporating silver nanoparticles in a third-generation
dendritic poly(amidoamine) (PAMAM) grafted onto multiwalled carbon nanotubes
(MWNTs) was explored as an antimicrobial agent in solution (Yuan et al. 2008). The
reducing agent for reduction of silver ions to nano-silver was highly abundant amine
groups on the dendrimer-modied MWNTs (d-MWNTs). Silver nanoparticles were
precipitated from silver acetate aqueous solution upon treatment with the functionalized
dendrimers, resulting in carbon nanotubes/Ag nanohybrids (d-MWNTs/Ag). Facecentered cubic silver nanocrystals were obtained by functionalization with the resultant
45% d-MWNT-grafted PAMAM dendrimers. The comparison of antimicrobial properties of acid-treated MWNTs (MWNTs-COOH), d-MWNTs and d-MWNTs/Ag showed
that, against E. coli and P. aeruginosa, which are Gram-negative, both d-MWNTs and
369 [359--379] 10.12.2013 7:57PM
369
d-MWNTs/Ag were equally effective and were found to inhibit bacteria more strongly
than MWNTs-COOH. Against S. aureus (a Gram-positive bacteria), the d-MWNTs/Ag
nanohybrid showed 92.3% antimicrobial killing effect, while d-MWNTs exhibited
71.6% bacteria killing effect and MWNTs-COOH showed only 15.4%. Apart from the
silver ions from the nanoparticles, the ions released from the MWNT/Ag nanohybrids
might have contributed to bacterial inhibition, accounting for the higher inhibition
percentage. Thus nanohybrids can function as an antimicrobial disinfection coating in
purication processes (Yuan et al. 2008).
Porous antimicrobial silver nanoparticles/poly (methyl methacrylate)/poly (vinyl
alcohol) nanocomposites have been synthesized using radical-mediated dispersion
polymerization by Kong and Jang (2008a). Cost-effective and highly transparent,
poly(methyl methacrylate) (PMMA) was utilized for the synthesis of the nanocomposite. The incorporation of silver nanoparticles into the polymer could result in semitransparent antimicrobial coatings. Silver ions were dissolved in aqueous PVA solution
leading to coordination of the hydroxyl groups of PVA with silver ions leading the
synthesis to be targeted at specic reaction sites. After addition of azobisisobutyronitrile (AIBN), which acts a reductant as well as radical initiator, to the Ag+/PVA aqueous
solution, Ag/PVA complexes were linearly assembled in the shear ow and more
tightly assembled due to dipoledipole interactions between silver nanoparticles on
the surface. Then, the carboxyl groups of methyl methacrylate (MMA) form hydrogen
bonds with the hydroxyl groups of PVA coated with silver nanoparticles. Silver nanoparticle/PMMA nanobers were synthesized using a radical-mediated dispersion polymerization technique by injecting MMA monomer into the reaction medium and
polymerizing for 24 h. TEM images of PMMA composite nanobers revealed that a
high percentage of silver nanoparticles was incorporated into PMMA polymer nanobers with their surfaces exposed and this resulted in high antimicrobial activity as
evidenced from the minimum inhibitory concentration values. Thus semi-transparent
and low-cost porous PMMA/Ag spheres showed good antimicrobial activity, even
though the silver nanoparticles are mostly encapsulated within the PMMA chains
(Kong and Jang 2008a, Lee et al. 2008).
Tecophilic is a member of the hydrophilic polyether-based thermoplastic aliphatic
polyurethane family (Melaiye et al. 2005). This is a medical-grade polymer that is
capable of absorbing high amounts of water. Tecophilic electrospun nanobers encapsulating an imidazole silver complex were made, so as to exhibit remarkable hydrophilic properties. The gradual formation of the nanoparticles inside the ber was
observed using TEM. The silver complexes embedded in the as-spun bers were
decomposed by water treatment and released silver ions, which aggregated into silver
particles and exhibited bactericidal and fungicidal effects, as studied with a modied
KirbyBauer technique. The silver composite and the control pure electrospun
Tecophilic ber were tested on a lawn of Gram-positive, Gram-negative bacteria
(E. coli, P. aeruginosa, S. aureus) and fungi (Candida albicans, Aspergillus niger
and Saccharomyces cerevisiae). A zone of inhibition was observed as evidence of
bactericidal activity within and around the membrane after overnight incubation of the
agar plate at 35 C. Similar to the polymerization of pyrrole, chemical oxidation
370
370 [359--379] 10.12.2013 7:57PM
polymerization-mediated silver nanoparticle-embedded polyrhodanine nanober

synthesis was reported by Kong and Jang (2008b) and Kong et al. (2009), starting
from rhodamine as the polymer host and silver cations as initiators. As in the case of
pyrrole, rhodanine monomer reduced silver ions to silver nanoparticles mediated by
self-oxidation, which further complexed with rhodanine due to formation of coordination complexes, resulting in the formation of silver nanoparticle-embedded polyrhodanine nanobers. The synthesized silver nanoparticle/polyrhodanine nanober
nanocomposite exhibited broad-spectrum antibacterial activity (against Gram-negative
E. coli, Gram-positive S. aureus and C. albicans).
10.3
Functionalization of nanofibers for protective clothing applications

Various chemicals such as -cylcodextrin (-CD), o-iodosobenzoic acid (IBA), polyoxometalates, peroxides, oximes and chloramines have been synthesized to decontaminate
chemical warfare agents such as nerve agents, blister agents and blood agents. A catalyst
such as (3-carboxy-4-iodosobenzyl) oxy--cyclodextrin has been synthesized from -CD
and o-iodosobenzoic acid (IBA). Functionalization of nanobers with this catalyst was
carried out and then tested for decontamination of paraoxon, a nerve agent simulant. The
reactivity of the functionalized membranes was found to be 11.5 times faster than
activated charcoal (Ramakrishnan et al. 2006). This enhanced reactive property in
combination with higher porosity and hydrophilicity suggest that they are potential
candidates to use as lter media in protective clothing applications. The incorporation
of silver nanoparticles into nanobers has shown that they are capable of decontaminating Gram-negative bacteria such as E. coli and P. aeruginosa, which can be used as
bacterial lters for protection against biological warfare agents (Lala et al. 2007).
However, it should be noted here that the above nerve agent catalysts are reactive towards
other chemical warfare agents such as mustard agent, and silver does not have the
capability to kill Gram-positive bacteria. Polyelectrolyte polymer membrane, with its
ionic groups, can block nerve gas agents and hence has a potential application as a
protective clothing material. Linear and cross-linked poly(2-acrylamido-2-methyl-1propanesulfonic acid) (PAMPS)-lled nonwoven polypropylene were studied as transporters of water molecules mediated by ionic interactions and simultaneous blockers of
dimethyl methyl phosphonate, a san gas simulant. The membrane provides hydrophilic
pathways for water transportation, while simultaneously blocking the organic nerve gases
from crossing (Rivin et al. 2004). Sulfobetaine polymer membranes are mechanically less
stable and lose their integrity because of high water afnity. Linear and cross-linked
PAMPS-lled nonwoven polypropylene membranes were made so as to attain stability.
The tensile properties of PAMPS-lled nonwoven membranes are drastically improved
compared to PAMPS membranes without a nonwoven host. Further, the polyelectrolytelled nonwoven membranes exhibited higher selectivity for the nerve gas simulant and
lower vapor permeability of the same than water. The highest selectivity was observed with
a 9% cross-linked polyelectrolyte-lled membrane. Thus, polyelectrolyte-lled nonwoven
membranes with improved tensile properties, selective, high water vapor permeability and
371 [359--379] 10.12.2013 7:57PM
10.3 Protective applications
371
low nerve gas permeability paved the way for yet another novel nonwoven material for use
as a protective clothing material (Jung et al. 2010).
DNA is known to complex with toxic acridine orange and the effect was studied in the
presence of magnetic nanoparticles. Nanobers of super-paramagnetic nanoparticles and
DNA hybrids were made as promising materials for organic and inorganic detoxication,
especially to chelate with the toxic acridine orange dye. Iron oxide nanoparticles (Fe3O4;
10%) coated with oleic acid was incorporated into DNA and cetyltrimethylammonium
bromide complex nanobers so as to remove inorganic and organic chemical toxicants.
Enhanced and fast detoxicication was observed due to the super-paramagnetic behavior
of the embedded magnetic nanoparticles under the inuence of an external magnetic eld
(Navarathne et al. 2011).
As a lightweight, breathable protective fabric for chemical and biological decontamination, poly(vinyl alcohol-co-ethylene-g-diallylmelamine) (PVA-co-PE-g-DAM) nanobers layered on poly(propylene-g-diallylmelamine) (PP-g-DAM) meltblown nonwoven
fabric were developed. By controlling the level of coated nanobers, three nanobrous
membranes with similar surface morphology, hydrophilicity and transport properties were
prepared (Zhu et al. 2012). The N-halamine precursor moieties present in the functional
polymer membrane matrices were converted to active N-halamine structures by means of
treatment with a diluted sodium hypochlorite solution. Thus the active-chlorine-enriched
membrane was robust and also rechargeable. Active chlorine nanobrous membranes were
shown to be very potent and exhibited rapid biocidal effects against both E. coli and S.
aureus on contact. The excellent disinfectant effect was also observed during wet bacterial
penetration through the nanober membrane. The halamine nanobrous membrane was
also observed to be a versatile decontaminant for aldicarb, a carbamate pesticide, by
efcient detoxication within 30 min. Hence, the functional halamine nanober composite
membrane can serve as a dual purpose, ultra-light ltering medium for chemical and
biological protective clothing materials (Zhu et al. 2012).
Diisopropyluorophosphatases (DFPases) are produced by squid and are known
for their broad-spectrum neutralizating activity against organophosphates. DFPase
enzyme-containing nanobers, with their potential to neutralize harmful organophosphates used in warfare were fabricated by a co-electrospinning technique (Han et al.
2011). The large surface-to-volume ratio, enzyme-coated coreshell nanobers were
fabricated with a very high enzyme loading (>80%) in the shell without noticeable
loss of DFPase enzymatic activity. The enzyme/PEO nanobers were covered with
superhydrophobic ber membranes and water-repellent properties were observed
(Figure 10.10). Further, the fabric was found to be useful in toxic chemical decontamination and allowed only water vapor to pass through. The concentration of uoride
released from the nanober was normalized with the amount of protein embedded in the
nanober membrane and the resultant enzyme activity versus time was derived. The
evaluation of various coreshell combinations revealed that a coaxial combination of
PEO (in the core) and DFPase (in the shell) produced the highest enzyme activity
(~ 7.3 mM mg1). Further, the enzyme activity was sustained for up to 12 hours of
water treatment by means of a multilayer model with a superhydrophobic protection
layer over the enzyme/PEO ber mat.
372
372 [359--379] 10.12.2013 7:57PM
Syringe for
core polymer
Syringe for
sheath enzymes
Water droplet
Superhydrophobic fibers
Enzyme-coated fibers
Enzyme
Superhydrophobic fibers
DFPase sheath
Polymer core
(directly react with toxins)
eSpinnable polymer
(a)
Polymer core
Conformal DFPase sheath
DFPase/PEO blend
(directly reacts with chemicals)
(b)
Water vapors
High voltage Collector
(c)
Figure 10.10
Electrospinning of enzyme-containing bers: (a) coaxial dual solution electrospinning;

(b) coresheath coaxial ber with polymer core and conformal DFPase-containing shell;
(c) uniform DFPase/PEO blended ber using single solution electrospinning. Reprinted with
permission from Han et al. (2011). Copyright 2011, American Chemical Society.
10.3.1
Polymer nanocomposite fibers in protective clothing and textile applications

Nanocomposites consist of a matrix material such as a polymer, with a dispersion of
nanoparticles. Various nanoparticles, such as TiO2, MgO and Al2O3 were dispersed in a
polymer solution, followed by electrospinning to produce novel nanocomposites, which
have excellent electrical conductivity and mechanical strength, and good catalytic
properties to apply as self-decontaminating fabrics. Metallic and metal oxide nanoparticles have been reported as multifunctional catalysts capable of decontaminating
both chemical and biological warfare agents. The integration of these nanoparticles with
nanobers combines the excellent barrier properties of nanobers and the good catalytic
properties of nanoparticles. In defense applications, nanocomposites were developed to
offer protection against bullets, chemical and biological warfare agents, and to be used as
chemical and biological sensors. The combination of nanobers with nanoparticles
allows the absorbtion, deactivation and decontamination of harmful substances. The
gas-barrier property of nylon fabric was improved by coating with polyurethane/clay
nanocomposites by Joshi et al. (2006). They tested the hydrogen gas permeability of
nanocomposites in which the incorporation of clay (2 wt%) reduced permeability by
about 36% compared to a neat polyurethane coating at the same thickness. This showed
that nanocomposites hold great potential to improve the performance of the resulting
materials. Nanocomposite fabrics were fabricated by the combination of electrostatic
processes such as electrospinning and electrospraying techniques (Figure 10.11) by
Sundarrajan et al. (2009). In this method, electrospinning of polysulfone nanobers
and electrospraying of nanoparticles were carried out either simultaneously (in which
case the membranes are referred to as SIM membranes) or using layer by layer deposition
373 [359--379] 10.12.2013 7:57PM
10.3 Protective applications
Taylor
cone
373
High voltage
power supply
Electrospraying
Electrospinning
Syringe pump with
polymer solution
Rotating mandrel
collector
High voltage
power supply
Figure 10.11
Laboratory setup for coupling electrospinning and electrospraying techniques.
(referred to as LBL membranes). The activities of SIM and LBL membranes were found
to be 52% and 48%, respectively, in comparison to nanoparticles.
10.3.2
Radioactive wastes and their decontamination

Depleted uranium as a by-product of uranium enrichment, with altered isotopic
ratios (typically 235U/238U = 0.003) is used in military armaments owing to its high
density and pyrophoric characteristics to damage heavily the artillery targets
(Wilkinson 1962). In the Gulf Wars and in the Iraq invasion, depleted uranium was
employed, which has raised controversies around contamination of the environment
and humans with toxic and radioactive material (Bleise et al. 2003). Natural calamities
are another reason for contamination of the environment by means of radioactive
materials, and hospitals also contribute to such environmental pollution. Chitosan is
known for metal ion complexation with a variety of transition and transition earth
metals, and has the capacity to absorb metal ions from solution. Hence, it has potential
for radioactive metal complexation, decontamination and environmental remediation.
Cuiffo and Halada (2012) reported the methods they have developed to use physically
374
374 [359--379] 10.12.2013 7:57PM
and chemically modied chitosan for secondary containment of nuclear waste

barrels, and spent nuclear fuel and/or depleted uranium. It was also applied as a
decontamination wipe for corroded steel exposed to uranium and transuranic analogs,
using either wet application (for subsequent removal through heat lamp exposure) or
application as a solid sponge. Further, a potential application of chitosanzerovalent-iron slurry in permeable reactive barriers or barrier liners is reported by Zhu
(2012), together with the design of titanate nanobers and Ag2O composites for
radioactive material remediation. These nanobers can capture the radioactive cations
from water by exchanging sodium ions in the titanate interstices. Due to the adsorption
of heavier cations or higher charges, deformation of the titanate structure occurs
and the radioactive ion becomes trapped inside permanently. The adsorbed or entrapped radioactive contaminant can be disposed accordingly later. The Ag2O treatment
of titanate nanobers results in a radioactive iodine adsorption capacity of nanobers
by precipitating silver iodide Felix-Rivera et al. (2011).
10.4
Sensors
The development of sensors for microorganisms and related contamination is
demanding research techniques that will be useful for military end-users in the
case of probable biological warfare. Fingerprinting information and chemical composition of microorganisms can be sensed with Raman spectroscopy, which can be
used in near-eld or at long distances from the sample. Surface-enhanced Raman
scattering (SERS) spectroscopy was employed to identify biochemical components
of the cell walls and endospores of Bacillus thuringiensis (Bt). Silver ions were
reduced to silver nanoparticles (NPs) with hydroxylamine and sodium borohydride
and then the sensor was capped with sodium citrate. Activation of these hot spots,
and aggregation and surface charge modication of the silver NPs, was optimized
to obtain signal enhancements from Bt by SERS. An improved interaction, as
inferred from higher-intensity SERS signals, was obtained for silver NPs with
even a slightly positive surface charge in contact with Gram-positive bacteria (Bt)
containing negatively charged phosphate groups and carboxylic groups at the
bacterial cell surface. Hydroxylamine-silver NPs required slight aggregation and
no pH modications in order to get high-spectral-quality results in bringing out
SERS-mediated signatures of Bt (Felix-Rivera et al. 2011). Silver nanoparticle
deposition mediated by a simple electrochemical method has been proposed for
the detection of cyanide ions. The loss of redox activity is used for sensitive
detection of cyanide ions. The electrochemical behavior and morphology of the
silver nanoparticles were found to change and were characterized in the presence of
potassium cyanide (KCN) or diethyl cyanophosphonate (DECP), as well as in
alkaline media. These were investigated by cyclic voltammetry and scanning electron microscopy (SEM). DECP was used as a simulant of the chemical warfare
agent tabun and detection of cyanide ions was inferred from nanoparticle formation
(Fathi et al. 2011).
375 [359--379] 10.12.2013 7:57PM
10.6 Respirator masks
10.5
375
Nanofiber decontamination wipes

Nanober technology-based decontamination wipes are available on the market.
The 3M company has introduced 3M FinalWipe Surface Cleaning System in
which the nanober absorbs and traps the solvent within the wipe. Further, nanoberbased technology allows the wipe to be used 10 times. Recently nanober technology
has been employed to develop a product that is very useful for wartime military
applications. This wipe exceeds conventional loose carbon particle wipes for the
adsorption of toxic vapors. These composite nanobers work based on both adsorption
and absorption of the toxic contaminants. It can be used as a dry wipe to decontaminate
the wounded skin of a soldier or equipment. Moreover, the nanobers can be functionalized using biocidal materials so as to make them active for decontamination in
chemical and biological warfare applications.
10.6
Respirator masks
A respiratory facemask is essential for protection of people working in many domains,
especially soldiers, healthcare workers and laboratory researchers, against antimicrobial
or viral infections. During pandemic disease spreading among humans, there is a need for
functional respiratory masks so as to avoid contamination through inhalation.
Electrospun micro- and nanobers have been used in respirator facemasks since the
1950s (Filatov et al. 2007). The mechanism of virus replication mediated by host cell
interaction and the mode of action of antiviral drugs is depicted in Figure 10.12a.
Nanoparticle-mediated virus replication inhibition is depicted in Figure 10.12b and is
through the interaction of nanoparticles on the surface of viral glycol proteins (Galdiero
et al. 2011). There are factors such as the t that may compromise the protective effect of
a facemask by means of exposure of the nostril passage to the invading microbes. The
antimicrobial protective effect of masks such as N95 and three-layer surgical masks are
only maintained when the surface layer of the mask is hydrophobic and dry. It was found
that when tested N95 facemasks were not hydrophobic, the antimicrobial protection
effect was signicantly reduced when the masks were accidently moistened. When a
moistened mask surface becomes contaminated with infectious agents, microorganisms
may be able to penetrate into the protective layers along with the moisture droplets. Since
frequent replacement with clean masks or gowns cannot be done, this results in repeated
contamination. So as to minimize the risk of transmission of infectious agents through
contaminated protective respiratory masks, an antimicrobial silver/TiO2 nanoparticle
coating was developed. Silver ions have been recognized for their broad-spectrum
antimicrobial activities, but the usefulness of silver-containing antimicrobial agents is
restricted because of the development of bacterial resistance due to uncontrolled and
increasing usage of these agents. As a second microbicidal compound, titanium dioxide
has been incorporated into the nanoparticles, which exhibit high microbicidal activity in
the presence of ultraviolet light, water and oxygen, by means of generation of highly
reactive OH radicals. The antimicrobial potential of nanoparticle-coated masks showed
376
376 [359--379] 10.12.2013 7:57PM
(a)
(b)
Important steps in
the virus
replication cycle
Major classes of
antiviral drugs
Virus adsorbtion
Antiviral mechanisms
of metal nanoparticles
Viral infections
Virion binding compound

Virus particles
Virus penetration
Host cell
Binding and
penetration
compounds
of virus genome
translation/transcription
Transcription
blocking
compounds
Nucleus
Antiviral activity
Progeny
virions
Genome
Viral factors
Non-nucleoside
replication
rt inhibitors
inhibitors
Helicase-primase
Protein
synthesis
Protease
Virion
Assembly
Release
of virions
Figure 10.12
inhibitors
Genome
Cellular factors
Viral factors
Nucleoside/nuceotide
Polymerase inhibitors
Genome
Interaction with viral

surface glycoproteins
Cytoplasm
Channel-blocking
Uncoating
Expression
Metal
nanoparticles
Fusion inhibitors
Cellular factors
Viral replication
Block viral replication

Cell nucleus
assembly inhibitors
Neuraminidase
inhibitors
(a) Mechanisms of virushost cell interaction and antiviral drug action.

(b) Nanoparticle-mediated virus inhibition. (Galdiero et al. 2011; an open access journal).
Courtesy of MDPI Publishers.
100% reduction in viable E. coli and S. aureus when observed in the coated mask
materials after 48 h of incubation. Nanoparticles show promise, even provoking no
skin irritation when applied as a coating to the surface of protective clothing or masks,
together with reduction in the risk of transmission of infectious agents (Li et al. 2006).
Nanober-based multiple-use respirator masks with Agion silver/copper zeolite compounds permanently embedded in the ber have been developed, complying with
National Institute of Occupational Safety and Health N95 and N99 standards. The
respirator mask prevents microbial and other particle penetration. The active nanober
layer of the mask kills Streptococcus pyogenes, methicillin-resistant Staphylococcus
aureus and other bacteria, together with inactivation of H1N1, H5N1, the common
cold and other viruses.
Nanober technology has shown its usefulness in day-to-day life and this chapter
summarized its applications as a decontamination tool from a warfare perspective. The
chemical and biological warfare decontamination applications of these nanobers were
discussed based on the end-user viewpoint, such as clothing material, wipes, sensors and
respirator masks.
The material in the present chapter shows that research in the area has delivered
few products available for real-life applications and further potential products using
the technology are keenly awaited by end-users. A better understanding of the safety
aspects (Schinwald et al. 2012) can quicken the research on these awaited nanober
377 [359--379] 10.12.2013 7:57PM
10.7 References
377
products. Usage of natural material nanober composites can overcome nanotoxicity

to a considerable extent. Further avoidance of leachable incorporations into nanobers
and improving complexations, cross-linking or interactions between the materials used
will reduce the toxicology issues to a great extent.
10.7
References
An, J., Zhang, H., Zhang, J., Zhao, Y., Yuan, X., 2009. Preparation and antibacterial activity of
electrospun chitosan/poly(ethylene oxide) membranes containing silver nanoparticles. Colloid
Polym. Sci. 287, 14251434.
Bleise, A., Danesi, P. R., Burkart, W., 2003. Properties, use and health effects of depleted uranium:
A general overview. J. Environ. Radioact. 64, 93112.
Cuiffo, M., Halada, G., 2012. Cross-linked chitosan-based materials for decontamination, containment and barriers. Materials Issues in Nuclear Waste Management in the 21st Century. Meeting:
Materials Science and Technology 2012, Pittsburgh, PA.
Dadvar, S., Tavanai, H., Morshed, M., Ghiaci, M., 2012. The removal of 2-chloroethyl ethyl sulde
using activated carbon nanobers embedded with MgO and Al2O3 nanoparticles. J. Chem. Eng.
Data 57, 14561462.
Fathi, F., Schlitt, M., Pedersen, D. B., Kraatz, H. B., 2011. Chemical behavior of electrochemically
generated nanostructured silver surfaces. Langmuir 27, 1209812105.
Felix-Rivera, H., Gonzalez, R., Rodrguez, G. DelMar, Primera-Pedrozo, O. M., Rios-Velazquez,
C., Hernandez-Rivera, S. P., 2011. Improving SERS detection of bacillus thuringiensis using
silver nanoparticles reduced with hydroxylamine and with citrate capped borohydride. Int.
J. Spectrosc. 2011, 989504.
Fundamentals and Applications in Separation and Filtration Processes. Begell House, New York.
Galdiero, S., Falanga, A., Vitiello, M., Cantisani, M., Marra, V., Galdiero, M., 2011. Silver nanoparticles as potential antiviral agents. Molecules 16, 88948918.
Gibson, P. W., Schreuder-Gibson, H. L., Rivin, D., 1999. Electrospun ber mats: transport properties. AIChE J. 45, 190195.
Guggenbichler, J. P., Boswald, M., Lugauer, S., Krall, T., 1999. A new technology of microdispersed
silver in polyurethane induces antimicrobial activity in central venous catheters. Infection 27, 1623.
Han, D., Filocamo, S., Kirby, R., Steckl, A. J., 2011. Deactivating chemical agents using enzymecoated nanobers formed by electrospinning. ACS Appl. Mater. Interf. 3, 46334639.
Ivarsson, U., Nilsson, H., Santesson, J. (Eds.), 1992. A FOA Brieng Book on Chemical Weapons:
Threat, Effects, and Protection. National Defence Research Establishment, Ume.
Joshi, M., Banerjee, K., Prasanth, R., 2006. Polyurethane/clay nanocomposite based coating for
enhanced gas barrier property. Ind J. Fibre Text. Res. 31, 202214.
Jung, K. H., Ji, L., Pourdeyhimi, B., Zhang, X., 2010. Structureproperty relationships of polymerlled nonwoven membranes for chemical protection applications. J. Membr. Sci. 361, 6370.
Kong, H., Jang, J., 2008a. Antibacterial properties of novel poly(methyl methacrylate) nanober
containing silver nanoparticles. Langmuir 24, 20512056.
Kong, H., Jang, J., 2008b Synthesis and antimicrobial properties of novel silver/polyrhodanine
nanobers. Biomacromolecules 9, 26772681.
Kong, H., Song, J., Jang, J., 2009. One-step preparation of antimicrobial polyrhodanine nanotubes
with silver nanoparticles. Macromol. Rapid Commun. 30, 13501355.
378
378 [359--379] 10.12.2013 7:57PM
Lala, N. L., Ramaseshan, R., Bojun, L., Sundarrajan, S., Barhate, R. S., Ying-jun L., Ramakrishna, S.,
2007. Fabrication of nanobers with antimicrobial functionality used as lters: protection against
bacterial contaminants. Biotechnol. Bioeng. 97, 13571365.
Lee, E. M., Lee, H. W., Park, J. H., Han, Y., Ji, B. C., Oh, W., Deng, Y. L, Yeum, J. H., 2008.
Multihollow structured poly(methyl methacrylate)/silver nanocomposite microspheres prepared
by suspension polymerization in the presence of dual dispersion agents. Colloid Polym. Sci. 286,
13791385.
Li, Y., Leung, P., Yao, L., Song, Q. W., Newton, E., 2006. Antimicrobial effect of surgical masks
coated with nanoparticles. J. Hosp. Infect. 62, 5863.
Marambio-Jones, C., Hoek, E. M. V., 2010. A review of the antibacterial effects of silver nanomaterials and potential implications for human health and the environment. J Nanopart. Res. 12,
15311551.
Mei, Y., Yao, C., Fan, K., Li, X., 2012. Surface modication of polyacrylonitrile nanobrous
membranes with superior antibacterial and easy-cleaning properties through hydrophilic exible
spacers. J. Membr. Sci. 417418, 2027.
Melaiye, A., Sun, Z., Hindi, K., Milsted, A., Ely, D., Reneker, D. H., Tessier, C. A.,
Youngs, W. J., 2005. Silver(I)-imidazole cyclophane gem-diol complexes encapsulated by electrospun tecophilic nanobers: formation of nanosilver particles and antimicrobial activity. J. Am.
Chem. Soc. 127, 22852291.
Morrison, W., 2002. Overview of current collective protection ltration technology. Presented at
2002 NBC Defense Collective Protection Conference, USA, 2002. Available at www.natick.
army.mil, and www.natick.army.mil/soldier/jocotas/ColPro_Papers/Morrison.pdf. Accessed
August 4, 2013.
Navarathne, D., Ner, Y., Jain, M., Grote, J. G., Sotzing, G. A., 2011. Fabrication of DNAmagnetite
hybrid nanobers for water detoxication. Mater. Lett. 65, 219221.
Oule, M. K., Azinwi, R., Bernier, A. M., Kablan, T., Maupertuis, A. M., Mauler, S.,
Nevry, R. K., Dembele, K., Forbes, L., Diop, L., 2008. Polyhexamethylene guanidine
hydrochloride-based disinfectant: a novel tool to ght meticillin-resistant Staphylococcus
aureus and nosocomial infections. J. Med. Microbiol. 57, 15231528.
Radheshkumar, C., Munstedt, H., 2006. Antimicrobial polymers from polypropylene/silver
composites Ag+ release measured by anode stripping voltammetry. React. Funct. Polym. 66,
780788.
Rajagopalan, S., Koper, O., Decker, S., Klabunde, K. J., 2002. Nanocrystalline metal oxides
as destructive adsorbents for organophosphorus compounds at ambient temperatures. Chem.
Eur. J. 8, 26022607.
Ramakrishnan, R., Ramakrishna, S., 2007. Zinc titanate nanobers for the detoxication of
chemical warfare simulants. J. Am. Ceram. Soc. 90, 18361842.
Ramakrishnan, R., Sundarrajan, S., Yingjun, L., Barhate, R. S., Lala, N. L., Ramakrishna, S., 2006.
Functionalized polymer nanobre membranes for protection from chemical warfare stimulants.
Nanotechnology 17, 29472953.
Rivin, D., Meermeier, G., Schneider, N. S., Vishnyakov, A., Neimark, A. V., 2004 Simultaneous
transport of water and organic molecules through polyelectrolyte membranes. J. Phys. Chem. B
108, 89008909.
Schinwald, A., Murphy, F. A., Prina-Mello, A., Poland, C. A., Byrne, F., Movia, D., Glass J. R.,
Dickerson, J. C., Schultz, D. A., Jeffree, C. E., Macnee, W., Donaldson, K. 2012. The threshold
length for ber-induced acute pleural inammation: shedding light on the early events in
asbestos-induced mesothelioma. Toxicol. Sci. 128, 461470.
379 [359--379] 10.12.2013 7:57PM
10.7 References
379
Sundarrajan, S., Pliszka, D., Ramakrishna, S., Jaworek, A., Krupa, A., Lackowski, M., 2009.
A novel process for the fabrication of nanocomposites membranes. J. Nanosci. Nanotechnol.
9, 44424447.
Wilkinson, W. D., 1962. Uranium Metallurgy. Interscience Publishers, New York.
Wu, K. H., Yu, P. Y., Hsieh, Y. J., Yang, C. C., Wang, G. P., 2009. Preparation and characterization
of silver-modied poly(vinyl alcohol)/ polyethyleneimine hybrids as a chemical and biological
protective material. Polym. Degrad. Stabil. 94, 21702177.
Yousef, A., Barakat, N. A. M., Amna, T., Al-Deyab, S. S., Hassan, M. S., Abdel-hay A., Kim, H. Y.,
2012. Inactivation of pathogenic Klebsiella pneumoniae by CuO/TiO2 nanobers: A multifunctional nanomaterial via one-step electrospinning. Ceram. Int. 38, 45254532.
Yuan, W., Jiang, G., Che, J., Qi, X., Xu, R., Chang, M. W., Chen, Y., Lim, S. Y., Dai, J., Chan-Park,
M. B., 2008. Deposition of silver nanoparticles on multiwalled carbon nanotubes grafted
with hyperbranched poly(amidoamine) and their antimicrobial effects. J. Phys. Chem. C 112,
1875418759.
Zhang, Y., Lee, M. W., An, S., Sinha-Ray, S., Khansari, S., Joshi, B., Hong, S., Hong, J. H.,
Kim, J. J., Pourdeyhimi, B., Yoon, S. S., Yarin, A. L., 2013. Antibacterial activity of photocatalytic electrospun titania nanober mats and solution-blown soy protein nanober mats
decorated with silver nanoparticles. Catal. Commun. 34, 3540.
Zhu, H. Y., 2012. Adsorbents of titanate nanobers and nanotubes for the removal of radioactive
ions from contaminated water. Materials Issues in Nuclear Waste Management in the 21st
Century. Meeting: Materials Science and Technology 2012, Pittsburgh, PA.
Zhu, J., Bahramian, Q., Gibson, P., Schreuder-Gibson, H., Sun, G., 2012. Chemical and biological
decontamination functions of nanobrous membranes. J. Mater. Chem. 22, 85328540.
380 [380--431] 10.12.2013 8:15PM

and micro- and nanoparticles:
healthcare, nutrition, drug delivery
and personal care
Nanotechnology has profound applications in healthcare and has improved healthcare
research to a large extent. The therapeutic benets of nanotechnology in the eld of
medicine have resulted in new areas, such as nanomedicine, nanobiotechnology, etc.
Researchers in the eld are attempting to nd an effective nanoformulation to deliver
growth factors, supplements or drugs safely and in a sustained manner at the required site.
Their task is to attempt a different drug nanoformulation of existing blockbuster drugs
that brings improved efcacy and a therapeutic breakthrough. Thus the ultimate objective
of these nanotechnological drug-delivery systems is to ne tune the normal prole of
potent drug molecules in the body following their administration to signicantly improve
their efcacy and also minimize potential intrinsic severe adverse effects. For treatment
of breast cancer and non-small-cell lung cancer, Abraxane (paclitaxel) is employed as a
nanoparticular formulation, which increases drug delivery up to 70% in comparison
with solvent-based paclitaxel delivery. In this novel nanoformulation, Abraxis Bio
Sciences have used Bristol-Meyers Squibbs blockbuster drug paclitaxel (Taxol) and a
very common globular protein bovine serum albumin (BSA). There are numerous
nanotechnology-based drug-delivery systems such as nanocrystals, nanoemulsions,
lipid or polymeric nanoparticles, liposomes and nanobers. While nanoemulsions and
liposomal formulations did not make signicant advances, despite huge research spending, the polymeric nanoparticulate systems show more promise. Nanoparticles of a
polymeric nature nd application as drug-delivery systems and are advantageous due
their scalability, cost, controlled and targeted delivery, compatibility, degradability,
etc. Natural biopolymers are even better than the synthetic polymers in terms of biocompatibility and biodegradability. Nanoparticulate drug formulations alter the pharmokinetic prole of the therapeutic entity and program the release of the drug in sustained
or controlled manner. Thus, nanoparticle or nanoformulated drugs outperform conventional systemic delivery in terms of delivery of an encapsulated drug and its sustained
release. Slowly and surely nanoformulated drugs are coming onto the market, surpassing
systemic delivery, which is believed to be the only mode of administration for a wide
range of active pharmaceutical ingredients. Nanobrous drug-delivery systems are
being developed as potential scaffolds in tissue regeneration, wound healing and
cancer drug-delivery applications. In this chapter we are going to discuss two promising
nanotechnology-mediated drug-delivery tools, namely electrospun micro- and nanobers
381 [380--431] 10.12.2013 8:15PM
11.1 Scaffolds for tissue regeneration
381
and electrosprayed micro- and nanoparticles, which have a common synthetic procedure
mediated by an electrical potential difference. In particular, electrospun nanobrous scaffolds for tissue regeneration are discussed in Section 11.1. Empirical observations on drug
delivery from electrospun nanober mats are reviewed in Section 11.2, while the physical
mechanisms involved in drug delivery from nanobers are discussed in Section 11.3. The
rate of drug delivery from elecrospun nanober mats can be modulated using thermo- and
pH-responsive polymers, as discussed in Section 11.4. The applications of electrospun
nanober mats as health supplements for vitamin delivery and in cosmetic facial masks are
covered in Sections 11.5 and 11.6, respectively. Electrosprayed nanoparticulate systems for
drug delivery are discussed in Section 11.7.
11.1
Nanofibrous scaffolds for tissue regeneration

Tissue engineering or regenerative medicine is the application of therapeutic products
that utilize a combination of matrix scaffolds from natural or synthetic polymer sources
viable in human cell systems, or cell-responsive biomolecules derived from such cells,
for the repair, restoration or regeneration of cells or tissues damaged by injury, disease or
congenital defects. Electrospun nanobers are used in such tissue engineering as scaffolds or matrices to provide support for cells in order to express the regenerated
extracellular matrix (ECM). The nanobers used in drug-delivery systems, serve as an
ideal matrix to release the drug in a sustained manner through the slow biodegradation of
the natural (Table 11.1) or synthetic polymer substrates. Drug delivery is also possible
from electrospun nondegradable nanobers due to the desorption mechanism discussed
in detail in Section 11.3. In addition, many materials, such as ceramic nanocomposites,
metallic nanoparticles and metal complexes with various functional capabilities have
been incorporated into electrospun nanobers and have found application as drugdelivery systems and tissue-engineering scaffolds.
Research in the area of drug delivery and tissue engineering has witnessed a surge in
recent decades due to the potential to mitigate concerns related to human health. Drug
delivery and tissue engineering are closely related elds, where both drug-delivery
vehicles and tissue-engineered scaffolds need to be biocompatible and in some cases,
biodegradable. Stem-cell nanober scaffolds nd application as patches for cardiac
restoration, and wound healing patches are in development. Drug-loaded nanober
membranes are helpful for the sustained release of encapsulated drugs in applications
such as drug-eluting stents and patches (Figure 11.1).
11.1.1
Extracellular matrix (ECM) proteins

Collagen
Collagen (Col) is a brous protein found in animals, especially in the esh and
connective tissues of mammals. It is the most abundant protein in mammals, constituting up to 35% of the whole-body protein content, and is commonly created by
broblast cells. Collagen is mostly found as elongated brils in brous tissues such as
Biopolymer
Electrospinning
Mechanical
Chemical
Clinical use
Collagen
Hexauoro 2-propanol
(HFP)
Glutaraldehyde, EDC
in ethanol
Used in wounddressing

materials
Fibrinogen
Serum albumin (bovine)
HFP
Co-electrospinning
(shell polymer in organic
solvent and core in water)
Blend with synthetic

polymers PLA, PCL,
PLACL, PDO
Blend with PDO
PLACL PU /LiCl
Collagen and elastin
HFP
Blend with PEO, PLGA
Gelatin
Triuoro ethanol (TFE)
Blend with PANI,

PLGA, PCL
Hemoglobin
Chitosan
TFE
Water, TFA/ dichoromethane
(DCM)
PEO, PEO/ hydroxyapatite
Genipin, EDC
Glutaraldehyde
Drug delivery in general.

Nanoparticulate
albuminpaclitaxel
complex used for breast
cancer treatment
EDC/N-hydroxy
succinimide (NHS)
Glutaraldehyde
Drug carrier, capsules
Glutaraldehyde
Glutaraldehyde
Drug delivery
382 [380--431] 10.12.2013 8:15PM
Cross-linking method
Table 11.1 Natural polymeric nanobrous scaffolds/crosslinking. The following works dealt with: collagen (Matthews et al. 2002, Shields et al. 2004); brinogen
(Kim et al. 2011, Carlisle et al. 2009, Wnek et al. 2003, McManus et al. 2006, 2007a, b); serum albumin (bovine) (Dror et al. 2008); collagen and elastin (Lee et al. 2007,
Barnes et al. 2007, Heydarkhan-Hagvall et al. 2008); gelatin (Zhang et al., 2009); Hemoglobin (Barnes et al. 2006); Chitosan (Zhang et al. 2008a,b).
383 [380--431] 10.12.2013 8:15PM
383
Prospective applications of nanofibers in drug delivery and tissue regeneration

Stem-cell
loading
Differentiation
Nanofiber
cardiac mesh
Nanofibers for ocular

drug delivery
Cross-linked protein
nanofibers
Drug-loaded stent
(nanofibrous)
Figure 11.1
Stem-cell-loaded
nanofiber mesh
Baking layer
Nanofibrous drug reservoir
Diffusing drug
Adhesive layer
Nanofibrous transdermal
drug-delivery system
Cardiomyocyte
differentiation
Wound healing
Nanober scaffolds in development as tissue engineering and drug-delivery scaffolds.
tendon, ligament and skin, and is also abundant in cornea, cartilage, bone, blood
vessels, the gut and intervertebral disks. Collagen is a major ECM component. It
possesses a brous structure with brils of varying diameters (50500 nm). These
brils inuence cell behavior by allowing cell attachment to the nanotextured collagen
matrix. Cells seeded on this nanobrous matrix tend to maintain their normal phenotype and growth is guided along the ber orientation. The motivation behind the
biomimetic nanostrategies is to dictate, control and fabricate the morphology and
composition of the developed biomaterials, in which nanocrystallites of inorganic
hydroxyl apatite (HAp) are dispersed with preferential orientation so as to enhance
bone tissue regeneration. Because of potential biomedical applications, many studies
report fabrication of bone-mimicking biocomposites of HAp and bioactive organic
components such as collagen, gelatin, chondroitin sulfate, chitosan and amphiphilic
peptides by direct precipitation methods (Kikuchi et al. 2001, Chen et al. 2002),
poly(lactic acid) through a solvent-cast technique (Liao et al. 2004) and polyamide
by a solution method (Wei et al. 2003). Collagen and HAp biocomposite is a native
ECM mimic and has the potential to replace diseased skeletal bones. HAp has an
inorganic crystalline nature like that of natural bone and is biocompatible, bioactive
and osteoconductive in nature. High levels of type I collagen and several noncollagenous proteins (such as osteopontin, bone sialoprotein and osteocalcin) constitute
bone tissue. The dimensions of bone minerals are around 50 nm in length, 25 nm in
width and 25 nm in thickness (Sachlos et al. 2006). These biominerals have their long
crystallographic c-axis oriented parallel to each other and also aligned with collagen
tropocollagen molecules (Weiner and Traub 1989). Collagen is easily biodegraded and
resorbed by the body and facilitates excellent attachment to cells. However, its
mechanical strength is relatively low (Youngs modulus E ~ 100 MPa), especially
with respect to bone (E ~ 25 GPa; Clarke et al. 1993) and it is therefore highly crosslinked or found in composites, such as collagenglycosaminoglycans for skin regeneration (OBrien et al. 2004) or Col/HAp for bone remodeling (Venugopal et al.
2008c). The advantage of collagen and HAp devices in comparison with synthetic
384
384 [380--431] 10.12.2013 8:15PM
11 Applications of micro- and nanofibers and particles
PLGA devices (Carlson et al. 2004) is that the biocomposite signicantly inhibits the
growth of bacterial pathogens, which is often associated with prosthesis. Although
co-electrospinning of nano-HAp and collagen improved the mechanical properties of
the scaffold, much has to be done to exactly mimic the complex native nanostructured
architecture. Collagen supports cell adhesion and proliferation, and HAp acts as a seed
for biomineralization of osteoblasts in bone tissue regeneration (Landis et al. 1993).
The biocomposite of collagen and nano-HAp (Wahl and Czernuszka 2006, Wahl et al.
2007) is bioactive, osteoconductive, osteoinductive and is a natural choice for bone
grafting, as it mimics the bone components. Bone-like orientation of the c-axes of
HAp nanocrystals with regular alignment along collagen brils was fabricated and
demonstrated (Porter et al. 2005). The HAp/Col composite, designed to simulate
bone tissue, is produced using atelocollagen to reduce antigenicity by condensing a
Ca(OH)2/H3PO4 suspension (Venugopal et al. 2008c). It was noted that biomineralization on Col/HAp composite nanobrous scaffolds cultured with human fetal osteoblasts was signicantly higher than on collagen nanobrous scaffolds over a 10-day
culture period. It was found that mineral deposition was signicantly lower: up to 56%
in collagen nanobrous scaffolds without inorganic HAp. Thus it was demonstrated
that a Col/HAp composite nanobrous scaffold provided a space for cell adhesion and
growth, and also stimulated mineralization and the functional activity of osteoblasts
for bone tissue engineering.
Tissue-engineering scaffolds can successfully improve nerve regeneration by providing mechanical support and adhesion to growing neurites, and simultaneous physical or
chemical binding of ECM proteins on biodegradable electrospun scaffolds will further
enhance their outgrowth. Neurite outgrowth aligned along the nanobrous scaffolds
incorporating materials such as bronectin, collagen and micropatterned polymers will
further play a signicant function in guiding axons (Whitworth et al. 1995, Miller et al.
2001). The ECM, an environment desirable to mimic, is a complex mixture of polysaccharides and proteins, so that multicomponent systems, if engineered to promote
nerve regeneration, might be advantageous over single-component systems. ECM proteins are involved in specic interactions with neural tissues to produce biomimetic
scaffolds suitable for nerve regeneration (Mannello and Tonti 2007). Collagen, laminin,
hyaluronic acid and bronectin are the ECM components used for nerve regeneration
that are suitable for incorporation into synthetic scaffolds. Although collagen laments
were found to guide axonal regeneration, they have the disadvantage of high cost and low
mechanical strength.
Gelatin
As the main practical problem associated with collagen is its cost and the poor denition
of commercial sources, gelatin is used as an alternative for designing tissue-engineering
scaffolds. Gelatin (Gel) is a protein obtained from partial hydrolysis of collagen extracted
from skin, bone, cartilage, ligaments, etc. The composite of gelatin with other biodegradable synthetic polymers, known as bioarticial polymeric materials, has been well
adopted by researchers. This approach of blending gelatin with synthetic polymers
reduces cytotoxicity by means of cross-linking agents. Further, this new biomaterial,
385 [380--431] 10.12.2013 8:15PM
385
with excellent biocompatibility and improved mechanical and physical/chemical properties overcomes the obstacle associated with use of a single natural polymer (Venugopal
et al. 2008a). Interaction between cells and the scaffold material depends on various
physico-chemical properties of the material, particle size and surface properties,
including topography, roughness, surface energy and wettability. Chandrasekaran et al.
(2011) demonstrated better broblast cell density and morphology in a PLACL/gelatin
scaffold after plasma treatment, compared to nanobrous PLACL scaffolds (PLACL is
poly(lactic acid-co-caprolactone)). Moreover they observed the secretion of collagen on
broblasts grown on PLACL and PLACL/gelatin nanobrous scaffolds. Plasma-treated
PLACL/gelatin nanobrous scaffolds have a sufciently porous structure, hydrophilicity, mechanical stability and loose peripheral regions favourable for cell inltration. They
also provided enough space for broblast in-growth and thus resulted in the development
of a dermal substitute in skin tissue regeneration. Liu et al. (2009) fabricated threedimensional nanober-gelatin/apatite composite scaffolds that mimic both the nanoscale
native architecture and the chemical composition of natural bone ECM. With a new
thermally induced phase separation and porogen leaching technique three-dimensional
nanobrous gelatin scaffolds with well-dened macropores were designed. Inorganic
HAp deposited all along the three-dimensional porous structure is ideal for controlling
the surface topography and chemistry within complex nanostructures. In addition, it was
shown that these scaffolds have excellent biocompatibility and mechanical properties,
with enhanced osteoblast adhesion, proliferation and differentiation suitable for bone
tissue engineering.
Silk protein: silk fibroin

Silk broin is considered to be the most promising natural brous protein replacement for
collagen in bone tissue engineering due to its biocompatibility, slow biodegradation and
excellent mechanical properties. In the past few years, two natural silk sources (silkworm
silk, Bombyx mori and spider dragline silk, Nephila clavipes) have been processed for
making nanobers via electrospinning (Jin et al. 2002, Kim et al. 2003, Ohgo et al. 2003,
Zarkoob et al. 2004). In order to improve the electrospinnability of silk protein solutions
and to avoid potential inuences of hazardous polar organic solvents such as hexauoroisopropanol (Zarkoob et al. 2004), hexauoroacetone (Kim et al. 2003) and formic acid
(Ohgo et al. 2003) on the scaffolds biocompatibility, Jin et al. (2002) used all-aqueous
electrospinning by blending silk broin with ber-forming poly(ethylene oxide), PEO, at
a ratio of between 1:4 and 2:3. Methanol treatment of the electrospun scaffold renders
the scaffold water-insoluble because of a structural conformational change into a native
-sheet structure.
Silk-based biocomposite nanobers of HAp and bone morphogenetic protein-2
(BMP-2) as a growth factor were fabricated by Li et al. (2006a) to realize enhanced
bone-formation outcomes by culturing with human bone-marrow-derived mesenchymal
stem cells (hMSCs). It was observed that the inclusion of BMP-2 and HAp with electrospun silk broin nanobers resulted in the highest calcium deposition and upregulation
of BMP-2 transcription levels compared to other electrospun silk-based scaffolds.
386 [380--431] 10.12.2013 8:15PM
386
11.1.2
Carbohydrates as tissue-engineering scaffolds, amino polysaccharide chitosan

composites (calciumcarbohydrate composites)
Chitosan (CTS), an amino polysaccharide derived from the structural biopolymer chitin,
exists abundantly in crustacean shells (e.g., crabs) and plays a key role similar to that of
collagen in higher vertebrates. Chitosan retains a number of salient features, such as
structural similarity to glucoseaminoglycan found in bone, osteoconductivity, excellent
biocompatibility, tailorable biodegradability, low immunogenicity and better mechanical
properties, as an alternative to collagen in tissue engineering, but at a cheaper cost
(Yamaguchi et al. 2001, Muzzarelli and Muzzarelli 2002, Hu et al. 2004). However, it
is poorly electrospinnable, forms aggregates with nonelectrospinnable HAp nanoparticles and formulating a robust chitosan solution is a prerequisite to generating nanobrous HAp/chitosan biocomposite scaffolds. Because of these limitations in chitosan
electrospinning (Rusu et al. 2005, Yang et al. 2008), there are only a few reports on
nanobrous HAp/chitosan composites for bone tissue engineering.
With ultrahigh-molecular-weight poly(ethylene oxide) (UHMWPEO) as the support
polymer, Zhang et al. (2008a) demonstrated that chitosan nanobers could be prepared. After this, a modied two-step approach (Zhang et al. 2008b) via an in situ
co-precipitation synthesis route was designed to overcome the usual problem of nanoparticle agglomeration. This electrospinning process was adopted for preparation of
HAp/CTS nanocomposite nanobers with a higher (30 wt%) loading of HAp nanoparticles. It was conrmed by electron diffraction and X-ray diffraction analysis that
the acetic acid used for chitosan dissolution had minor or no inuence on the crystallinity of HAp nanoparticles incorporated within the nanocomposite nanobrous structure. The bone regeneration ability of the scaffold was assessed in these HAp/CTS
nanocomposite nanobrous scaffolds by conducting MTS and ARS assays of human
fetal osteoblasts cultured for up to 15 days. The results conrmed that the scaffolds had
signicantly enhanced bone formation, compared to an electrospun chitosan-only
scaffold.
11.1.3
Need to improve mechanical strength of natural protein and carbohydrate

nanofibrous scaffolds
Scaffolds composed of natural proteins and carbohydrate materials must be cross-linked
to maintain the structural integrity of the construct. To improve the stability of the protein
or carbohydrate nanobrous scaffolds and to reduce the rate of biodegradation, crosslinking of biopolymers becomes inevitable. This process of cross-linking also increases
the mechanical strength of the scaffold and improves aqueous stability.
11.1.4
The basics of cross-linking

The process of chemically joining two or more molecules by a covalent bond is termed
cross-linking (see Section 8.2 in Chapter 8). Cross-linking of proteins or carbohydrates
depends on the availability of particular chemical groups that are capable of reacting with
387 [380--431] 10.12.2013 8:15PM
387
the specic kinds of functional groups that exist in proteins and carbohydrates. In spite of
the complexity of protein or carbohydrate structure, just four functional groups are
applicable for the vast majority of cross-linking and chemical modication techniques:
1. Primary amine functionality (NH2): amine groups exist at the N-terminus of each
polypeptide chain and in the side chain of some amino acid residues.
2. Carboxyl groups (COOH): the carboxylic acid groups exist at the C-terminus of each
polypeptide chain and in the side chains of some amino acid residues.
3. Sulfhydryl functional group (SH): thiol groups often help in disulde bond formation and exist in the side chain of cysteine amino acid.
4. Carbonyl functional group (CHO): aldehyde groups often associated with carbohydrates and glycoproteins are formed by oxidation.
These residues, when interconnected via cross-linkers, become stable towards degradation
with improved mechanical strength for controlled drug-delivery and tissue-engineering
applications.
Conventional cross-linking agents

Glutaraldehyde in solution or as a vapor has been commonly used to cross-link proteinbased and amino-group-containing carbohydrate scaffolds. The glutaraldehyde crosslinking technique is inexpensive and efcient. It cross-links over a variety of distances
and reacts with many of the amino groups present in a protein or carbohydrate. The extent
of cross-linking in an electrospun scaffold is a function of the percentage of glutaraldehyde present in the cross-linking solution. The degree of scaffold cross-linking increases
with the percentage of glutaraldehyde present, but eventually reaches saturation, where
no more cross-linking can occur. However, there may be some cytotoxity and calcication issues associated with this xation agent.
Other methods of scaffold cross-linking have been developed, including a carbodiimide/ethanol technique. Carbodiimide is a zero-length cross-linker that has shown
nominal potential cytotoxity issues and can be used to modulate material properties
similar to (and in some instances, better than) glutaraldehyde xation. Genipin is a
natural cross-linker used as a substitute for glutaraldehyde, carbodiimide and isocyanate
cross-linkers because of the cytotoxicity associated with those materials. In spite of its
lower cytotoxicity compared to glutaraldehyde and other cross-linkers, its high cost
means that there are only a few reports available with genipin cross-linking. Thus there
exists a demand for a new effective, nontoxic and economic cross-linker.
11.1.5
Promising synthetic polyester polymer nanofibers for tissue engineering

The use of synthetic polymers as tissue engineering scaffolds facilitates higher mechanical stability compared to natural scaffolds. Further, it avoids the use of available crosslinking agents leading to slow degradation of the scaffold. These polymers also constitute
a new generation of biomaterials to mimic the structure and characteristics of native
ECM, such as the brillar structure, viscoelasticity, cell addition domains, growth-factor
binding and proteolytic sensitivity.
388
388 [380--431] 10.12.2013 8:15PM
Poly(lactic acid-co-caprolactone) (PLACL)

PLACL is a synthetic co-polymer of poly(lactic acid) and polycaprolactone with
a biodegradable nature. Electrospinning of PLACL into smooth muscle cell (SMC)
conduits has allowed researchers to create scaffolds simulating ECM with tailorable
mechanical properties. Topographically aligned submicron PLACL bers have similar
circumferential orientations to those of brils found in native artery. PLACL scaffolds
composed of aligned and random nanobers were designed to mimic native ECM, and
favorable interactions with smooth muscle cells (SMCs) and endothelial cells (ECs) were
studied by He et al. (2005) and Mo et al. (2004). To produce aligned PLACL nanobrous
scaffolds for articial blood-vessel engineering, an electrostatic lens, the sharp-edged
rotating wheel of Section 5.9 in Chapter 5, was successfully used (Theron et al. 2001, Xu
et al. 2004a). Cell culture of SMCs on thus-fabricated nanobrous scaffolds indicated
that the cells attach and migrate along the direction of aligned nanobers. Further, the
cells showed spindle-like morphology and expressed smooth cytoskeleton protein
(-actin) inside the cells. SMCs and ECs were found to interact and integrate well with
PLACL nanobrous scaffolds, forming a three-dimensional cellular network (Xu et al.
2004b). Thus the nanobrous scaffolds provide a new ECM-mimetic scaffold for the
formation of new blood vessels. Also, the synthetic polymer PLACL (He et al. 2006b)
was demonstrated to inuence cell adhesion, migration, proliferation and to support the
formation of three-dimensional blood vessels in vitro. HAp nanoparticles were electrosprayed on PLACL/gelatin nanobers and thus produced scaffolds with controlled
morphology, which were found to be suitable for bone tissue engineering. Gupta et al.
(2009a) used a simultaneous electrospraying and electrospinning concept to fabricate
PLACL/Gel/HAp nanobers and compared their mechanical and cellular properties
with blend electrospun PLACL/Gel/HAp scaffolds. Electrospun PLACL/Gel/HAp
(blend) nanobers had the drawback of trapping HAp inside the nanobers (diameter
198 107 nm), while the HAp nanoparticles were found uniformly sprayed, forming
a layer of HAp on the surface of the other PLACL/Gel/HAp scaffold (diameter
406 155 nm). The tensile stress for the HAp electrosprayed scaffold was higher than
for the PLACL/Gel/HAp (blend) scaffold and this was due to the fact that the electrospraying of HAp nanoparticles resulted in their supercial dispersion. A more signicant
increase in human fetal osteoblastic (hFOB) proliferation was observed on the HAp
electrosprayed scaffold than on the PLACL/Gel/HAp (blend) nanobers after 15 days of
cell seeding. Further, the electrosprayed scaffolds showed 50% higher biomineralization
than the PLACL/Gel/HAp (blend), thus proving the versatility of electrospraying compared to the blend technique with respect to scaffold design for bone tissue engineering.
Jin et al. (2011) demonstrated the differentiation of mesenchemyl stem cells (MSCs) into
the epidermal phenotype and undifferentiated MSC grown on Col/PLACL nanobrous
scaffolds. They observed that cells on Col/PLACL nanobers had a higher percentage of
differentiated cells (66.2%) than those on PLACL nanobers (33.3%), which is because
of the presence of collagen in the Col/PLACL scaffold, at a level equivalent to the
percentage of collagen present in the native ECM of human skin. Morphological and
immunocytochemical evaluations conrmed the differentiation of MSC into epidermal
389 [380--431] 10.12.2013 8:15PM
389
cells on electrospun Col/PLACL nanobers. Thus, rapid progress in the design of

biomaterial for skin substitutes, including stem cell differentiation, would result in the
development of an ideal dermal substitute.
Polycaprolactone (PCL)
Polycaprolactone (PCL) is a biodegradable synthetic polymeric material obtained from
ring-opening polymerization of -caprolactone using a catalyst such as stannous octoate.
It is an FDA-approved polymer with no toxicity, controllable degradation and is inexpensive. PCL is a semi-crystalline polymer that belongs to the family of -hydroxy
polyesters (e.g., poly(lactic acid), PLA, and poly(glycolic acid), PGA) and considerably
enhances the mechanical properties of nanober scaffolds due to its high tensile modulus
(400 MPa; Engelberg and Kohn 1991, Thomas et al. 2006). Since these mechanical
properties are in the range of the elastic modulus of collagen bers, PCL constitutes
promising cartilage tissue regeneration nanobrous scaffolds (Wise et al. 2008, 2009). In
addition it was well studied to show its biocompatibility with seeded hMSCs and support
of multilineage differentiation (Shin et al. 2004, Li et al. 2005a, 2005b, Wise et al. 2008,
2009). Ghasemi et al. (2008) electrospan random and aligned PCL/gelatin (70:30)
nanobers and found that the scaffold enhanced nerve cell proliferation and differentiation with improved neurite outgrowth along aligned PCL/gelatin nanobers, compared to control PCL nanobers. Similarly in comparison to pure PCL nanobers, PCL/
collagen nanobers were demonstrated to improve migration of Schwann cells (SCs) and
neurite orientation by Schnell et al. (2007). Chew et al. (2008) electrospan aligned PCL
nanobers and studied their potential application in the guidance of human SCs. Cell
cytoskeleton together with nuclei were found to align and elongate along the aligned PCL
nanober axes, emulating the structure of Bungner bands. Biocomposite PCL/chitosan
nanobers were electrospun with a ber diameter of 190 nm and seeded with rat SCs. In
vitro evaluation of these scaffolds demonstrated enhanced cell proliferation in PCL/
chitosan composite scaffold compared to electrospun PCL-only nanobers, thus providing an ideal scaffold for peripheral nerve regeneration (Prabhakaran et al. 2008a, 2008b,
Gupta et al. 2009b). Electrospun PCL nanobrous scaffolds were fabricated by
Yoshimoto et al. (2003), and MSCs from the bone marrow of neonatal rats were seeded
with osteogenic supplements. Cell penetration into the brous matrix was observed,
along with abundant ECM deposits in the cellpolymer constructs, thus favoring PCL as
a potential scaffold for bone tissue engineering.
PCL/HAp (Venugopal et al. 2008b), PCL/Col/HAp (Venugopal et al. 2007) and PCL/
Gel/HAp (Venugopal 2008a) nanobers were fabricated and tested for suitability as a
scaffold for bone tissue engineering. Electrospun PCL/HAp nanobrous scaffold after
plasma treatment were found to enhance wettability and thus accelerated the biodegradation rate of nanobrous scaffolds. It was observed that a PCL/HAp plasma-treated
scaffold showed mineral formation on the surface of osteoblast cell layers within six days
of culture. Scaffolds of gelatin/PCL (1:1) nanobers were electrospun (Zhang et al.
2005), with gelatin concentrations ranging from 2.5 to 12.5% w/v, and the membranes
showed improved mechanical properties and more favorable wettability than gelatin- or
PCL-only membrane. Bone-marrow stromal cells (BMSCs) seeded on the surface of
390
390 [380--431] 10.12.2013 8:15PM
Gel/PCL biocomposite scaffolds were seen to migrate up to 114 m inside the scaffold
within a week of culture, showing biocompatibility to be better than PCL-only nanobers. Thus, it was demonstrated that a weak and loosely interlaced nanobrous structure
provides the least obstruction to cellular movements. Also, appropriate molecular signals
on the scaffold surface can further guide the living cells into the three-dimensional
nanober matrix (Venugopal et al. 2008a). Porous (235 m) nanobers of electrospun
PCL/HAp/Col (Venugopal et al. 2007) biocomposite with ber diameters of 373 191
nm provided a large surface-area-to-volume ratio for cell attachment, with sufcient
space for bone-cell ingrowth and nutrient transportation. Further, the interconnected
porous scaffold provided sufcient mechanical support for bone tissue regeneration.
Poly(lactic-co-glycolic acid) (PLGA)

PLGA is a synthetic co-polymer that is used in FDA-approved therapeutic devices,
owing to its biodegradability and biocompatibility. PLGA is synthesized by means of
random ring-opening copolymerization of two different monomers, the cyclic dimers
(1,4-dioxane-2,5-diones) of glycolic acid and lactic acid taken in the desired ratio. As a
variety of polyester biomaterial, to determine its suitability in nerve tissue engineering,
in vivo nerve regeneration was attempted using PLGA nanobers by Bini et al. (2004).
They observed that the implanted nanober guided the sciatic nerve of rats and found no
inammatory response, together with successful nerve regeneration.
Poly(lactic acid) (PLLA)

PLLA is a thermoplastic aliphatic polyester derived from renewable resources, and can
be synthesized from cyclic d,l-lactide via ring-opening polymerization. Koh et al.
(2008) fabricated laminin-functionalized PLLA nanobers via three different methods
for nerve tissue regeneration. They found enhanced axonal extensions using electrospun PLLA-laminin nanobers and conrmed their suitability for nerve tissue engineering. Articial nerve grafts incorporating growth factors were fabricated to
inuence nerve regeneration, and the controlled release of NGF, EGF and bFGF
from polymer conduits was found to enhance the growth of axons (Aebischer et al.
1989, Cordeiro 1989, Powell et al. 1990). So as to understand the contact guidance of
NSCs by PLLA nanober scaffolds, Yang et al. (2005) fabricated random and aligned
PLLA polymeric nano- and microbers and found that the neural stem cell (NSC)
differentiation rate was higher on nanobers than on microbers, independent of
the ber alignment. Further, aligned nanobers highly supported NSC proliferation
and enhanced neurite outgrowth compared to microbers and random nanobers.
Upon seeding human fetal osteoblasts (hFOBs) on PLLA/HAp and PLLA/Col/HAp
(Cordeiro 1989, Liao et al. 2004), the biocomposite PLLA/Col/HAp nanobrous
scaffold was found to be superior to PLLA/HAp for effective bone regeneration and
mineralization. In addition, the tensile strength of the biocomposite scaffold was
higher than the collagen brous matrix (1.68 MPa) and even PCL/HAp scaffolds.
The ALP activity was 25% higher on biocomposite nanobrous scaffolds compared to
PLLA/HAp scaffolds after 20 days of cell culture. Bone nodule formation by hFOB
cultured on different electrospun nanobers was characterized by ARS staining, and
391 [380--431] 10.12.2013 8:15PM
11.2 Drug delivery
391
mineralization was found to be 57% higher on biocomposite scaffolds compared to

PLLA/HAp scaffolds. Elemental analysis by EDX measurement primarily showed
calcium and phosphorus deposition, indicating that osteoblasts seeded on PLLA/Col/
HAp nanobers formed biomineralized tissue. Scaffolds with HAp-containing composites enhanced the formation of new bone tissue with increased osteoblast adhesion,
osteointegration and calcium mineral deposition on its surface (Prabhakaran et al.
2009).
11.2
Drug delivery
Noh et al. (2006) found that chitin nanobers, with their high surface area available
for cell attachment and three-dimensional features, can enhance cell attachment,
growth and proliferation. Irradiated chitin solution was electrospun and nanobers
were collected on a drum. The activity of commercial chitin microbers was compared
with the electrospun nanobers. Cell attachment and spreading studies revealed that the
cellular response on chitin nanobrous matrices was signicantly higher than on chitin
microbrous matrices. The same trend was also observed with extracellular matrix
proteins (i.e. collagen type I)-coated nanobrous mats. Incorporation of collagen
type I into chitin nano- or microbers could further enhance normal human keratinocyte
and broblast regeneration. Nanocomposites of poly(N-isopropylacrylamide)-copolystyrene with gold nanoparticle functionalization facilitated the accumulation of
daunorubicin inside leukemia cells and thereby enhance drug uptake (Song et al.
2009). With these nanocomposites, synergistic enhancement of daunorubicin uptake
by drug-resistant leukemia K562 cells was demonstrated. Using electrochemical and
confocal uorescence microscopic studies it was realized that these nanocomposites
could be utilized to facilitate drug delivery and also for diagnosis of the cancer.
11.2.1
Nanofibers as anticancer drug-delivery systems (drug-eluting stents)

Many attempts have been made to develop a nanober drug-eluting stent (Figure 11.2) for
the treatment of cancer. As an implantable device, applying the emulsion electrospinning
technique, an amphipilic polyethylene glycolpoly(l-lactic acid) diblock co-polymer
nanober scaffold with paclitaxel and doxorubicin was fabricated (Xu et al. 2009).
By using the emulsion method, anticancer nanober templates having dual drugs were
Nanofibrous drug delivery system (drug eluting stent)

Electrospinning
Commercial stent
Figure 11.2
Drug coated polymer nanofiber stent
Nanobrous drug-eluting stents in development for cancer treatment.
392
392 [380--431] 10.12.2013 8:15PM
fabricated: hydrophobic, cell-division preventing paclitaxel, and water-soluble, DNAintercalating doxorubicin. Hydrophilic doxorubicin was released from the scaffold faster
than the hydrophobic paclitaxel. The dual-drug combination therapy for cancer drug
delivery was demonstrated with rat C6 glioma cells. The dual therapy showed higher
inhibition than a single-drug-loaded system. PEO/chitosan electrospun nanobrous mats
were fabricated and made porous by leaching PEO (Ma et al. 2011). Paclitaxel was
incorporated into these porous cationic chitosan scaffolds. The nanobrous mat loaded
with the prostate cancer chemotherapy drug was cross-linked by anionic hyaluronic acid
so as to sustain drug release. Tests with DU145 prostate cancer cell culture demonstrated
that cell numbers decrease with increasing drug loading in the mat and suggested that
the nanobrous mats can be used as a promising post-surgical, chemotherapy wounddressing material. Electrospinning-mediated bers of paclitaxel-loaded PLGA micro- and
nanobers were fabricated and the sustained drug release was determined. The cytotoxicity
IC50 value of the nanobrous scaffold with C6 glioma cell lines was compared with the
commercial Taxol (Xie and Wang 2006). It was demonstrated that paclitaxel-loaded PLGA
nanober discs reduce the tumor by up to 30 times, with a signicantly lower tumor
proliferation index after 41 days of treatment.
A nanobrous implantable delivery system for the drug cisplatin in lung cancer
treatment was fabricated by electrospinning with biodegradable poly (l-lactic acid) to an
average ber diameter of 700 nm and the sustained release of cisplatin was studied in PBS
with and without proteinase K (Chen et al. 2011). The researchers observed that the release
rate was stronger when enhanced with proteinase K than with just PBS. Further it was
demonstrated that the virgin cisplatin and the drug-loaded composite bers showed almost
equal anti-tumor activity against human lung tumor spc-a-1 cells after 24 h.
11.2.2
Antimicrobial nanofibrous drug-delivery systems (wound-healing patches)

Tetracycline-loaded PLLA and PEVA (poly(ethylene-co-vinyl acetate)) (Kenawy et al.
2002), itraconazole-loaded hydroxylpropylmethylcellulose (Verreck et al. 2003b), PLGA
incorporating cefazolin (Katti et al. 2004) and chlorotetracycline/tetracycline/amphotericin
B-loaded PLLA/PCL (Buschle-Diller et al. 2007) nanobers were electrospun for use
in the inhibition of surgical-site microbial infections. Mefoxin incorporated into a PLGA/
PLA co-polymer/PLA nanobrous system (Kim et al. 2004) and PCL nanobers with
a model antibiotic (Bolgen et al. 2007) were trialed to avoid post-surgical microbial
adhesions and infections. The enantiomeric naphthoquinones, alkannin and shikonin,
found as natural products and extracted from the roots of Alkanna tinctoria were electrospun (Kontogiannopoulos et al. 2011) using PLLA, PLGA and cellulose acetate polymers
for enhancement of wound healing. PVA/PVP/PEG hybrid nanobers incorporating up
to 3% aloe vera were designed as wound-dressing materials, with increased porosity and
reduced ber diameter. The increased porous nanostructure was fabricated so as to increase
oxygen and moisture supply to the wound repair. Electrospinning of the aloe vera blend
showed an increase in viscosity and conductivity with increasing aloe vera percentage.
It was also observed that the ber diameter decreases with increasing aloe vera percentage
(Uslu et al. 2010).
393 [380--431] 10.12.2013 8:15PM
11.3 Desorption for drug delivery
11.3
393
Desorption as a drug-delivery mechanism

Electrospun and solution-blown nanobers were sucessfully loaded with different drugs,
molecular markers and nanoparticles, for example specic proteins acting as growth
factors determining commitment of human mesenchymal stem cells (hMSCs) and their
further differentiation. The rate of hMSC commitment and differentiation towards a
specic lineage is determined by the release patterns of growth factors embedded in
the electrospun nanober structures. Release rates of different molecular compounds
from electrospun bers have been studied experimentally (Kenawy et al. 2002, Zong
et al. 2002, Verreck et al. 2003a, 2003b, Zeng et al. 2003, Kim et al. 2004, Chew et al.
2005, Jiang et al. 2005, He et al. 2006a, Huang et al. 2006, Liao et al. 2006, Luong-Van
et al. 2006, Moroni et al. 2006, Zhang et al. 2006). Several groups have electrospun
biocompatible polymer blends with solutions of proteins or growth factors, so that the
latter compounds are embedded in the resulting bers (Chew et al. 2005, Jiang et al. 2005,
Liao et al. 2006, Li et al. 2006a). The growth factors used included human -nerve
growth factor (Chew et al. 2005), recombinant human platelet-derived growth factor-bb
(PDGF-bb) (Liao et al. 2006) and BMP-2 (Li et al. 2006a), although the release mechanisms of these compounds were not fully elucidated in these works. Drug and dye release
from nanobers was experimentally studied in several publications (Kenawy et al. 2002,
Zong et al. 2002, He et al. 2006a, Moroni et al. 2006). In all these works, nanobers with
embedded proteins, growth factors, drugs or dyes were fabricated either via the standard
electrospinning process (as monolithic bers) (Kenawy et al. 2002, Zong et al. 2002,
Chew et al. 2005, Moroni et al. 2006, Liao et al. 2006), or via co-electrospinning (as
coreshell bers) (Jiang et al. 2005, He et al. 2006a, Huang et al. 2006, Liao et al. 2006).
Release from monolithic nanobers saturated at various levels, including 3040%
(Zong et al. 2002), 4060 wt% (Kenawy et al. 2002), 60% (Liao et al. 2006) and
8090% (Chew et al. 2005). The release range 8090 wt% was only achieved when the
bers degraded during the release process. It was shown that the release process does not
follow theoretical predictions based on the assumption of solid-state diffusion (Zong et al.
2002). Release from coreshell bers saturated at about 5060% (He et al. 2006a) or
4080% (Jiang et al. 2005). Only one set of data for release from coreshell bers in
the latter work demonstrated nearly complete release, i.e., about 90%. However, in that
particular case the drug-free shell totally degraded, and thus the drug in the core became
fully exposed to the surrounding water. There are some other reports where drug
release from coreshell nanobers was driven by degradation of the carrier rather than
by solid-state diffusion (He et al. 2006a, Huang et al. 2006). Moreover, in Liao et al.
(2006), an additional polymer, polyethylene glycol (PEG), was blended in the monolithic
bers or shells (in the coreshell bers) together with polycaprolactone (PCL). When such
monolithic or coreshell nanobers were subjected to the surrounding water, 80100%
release of the embedded proteins or growth factors was achieved when PEG was leached by
water, thus creating porosity and in turn enhancing release.
Solid-state diffusion of small- and medium-sized molecules inside polymers was
studied using molecular dynamics simulations (Li et al. 1997). The predicted values of
394
394 [380--431] 10.12.2013 8:15PM
the diffusion coefcient were of the order of 106104 cm2 s1, which is several orders
of magnitude higher than the values expected from solid-state diffusion alone. The
dramatic disagreement of such calculations with the experimental data demonstrated
by Li et al. (1997) most probably stems from insufcient information being available to
calculate the intermolecular forces. The inability of solid-state diffusion theories to
explain the data found in the release experiments was clearly recognized by Saltzman
and Langer (1989), where it was argued that retardation due to dissolution of the
admixture, or pore interconnectivity and tortuosity must be taken into account in order
to explain the low values of the effective diffusion coefcients and release of less than
100%, even at the end of very long experiments. The surface-area-to-volume ratio in
the experiments of Saltzman and Langer (1989) was relatively small and the pore sizes
of the order of 104105 nm, which led the authors to the assumption that surface-related
phenomena (e.g. desorption) could be neglected. This may not be the case when nanober mats are the carriers of a released admixture, since such mats have dramatically
higher surface-area-to-volume ratios. Moreover, individual nanobers can accommodate
nanoporosity, which results in a still higher surface-to-volume ratio than is normally
thought for nonporous materials.
The above-mentioned studies reporting release saturation well below 100% suggested
that solid-state diffusion may not be the leading mechanism of release in many cases.
This idea was further supported by combining experiment and theoretical analysis in a
paper by Srikar et al. (2008), who put forward desorption from the nanopore surfaces as
the release mechanism in such cases. The limiting desorption stage is accompanied by
fast diffusion in water lling the nanopores or being in contact with smooth areas of the
nanober surface. This revised understanding was used to elucidate how variations in
polymer molecular weight and concentration in the electrospun solutions control the
desorption process, and thus the release rate.
Srikar et al. (2008) used rhodamine 610 chloride uorescent dye embedded in electrospun nanobers as a model drug in their release experiments. The dye was chosen as a
model compound because the amount released from the bers over a given time period
can be quantied by measuring the uorescence of the water pool surrounding the
nanober mats. The dye release proles from three different samples of electrospun
PCL nanobers submerged in water are shown in Figure 11.3. It can be clearly seen that
release from these bers fully saturates at about 67% for 11 wt% PCL, 50% for 13 wt%
PCL and 32% for 15 wt% PCL in the electrospun solution.
A possible explanation for the reduction of the released amount of dye at higher initial
polymer concentrations is related to the trend that bers fabricated from more concentrated polymer solutions contain less nanoporosity (discussed below). Then, less dye
embedded at the nanopore surfaces of an individual nanober comes into contact with the
surrounding water in the well, thus reducing the amount of released dye. This explanation
is corroborated below by additional experimental data for PMMA nanobers, as well as
theoretically.
Release of rhodamine 610 chloride dye from electrospun PMMA nanobers with three
different molecular weights: (i) 120 kDa (24 wt% in solution), (ii) 350 kDa (15 wt% in
solution) and (iii) 996 kDa (15 wt% in solution) is depicted in Figure 11.4. The data
395 [380--431] 10.12.2013 8:15PM
395
70
(1)
Percentage dye release
60
50
(2)
40
(3)
30
20
10
0
0
50
150
100
200
250
Time (h)
Figure 11.3
Effect of initial polymer concentration on dye release from electrospun PCL nanober mats. The
initial amount of dye per unit polymer mass was practically identical (about 0.2%) in all three
samples, whereas the polymer concentration in solution was: (1) 11 wt%, (2) 13 wt% and (3) 15 wt
%. The solvent was 60/40 wt% DMF/MC. Reprinted with permission from Srikar et al. (2008).
clearly indicate a strong effect of molecular weight on the release rate. The lowest
molecular weight PMMA (120 kDa) could not be electrospun at 15% because of its
low viscoelasticity, which did not allow spinnability. Therefore, it was electrospun at
24 wt% and, hence, the release rate from this sample was affected in this case not only by
variation in the molecular weight, but also in the concentration, as compared to the other
two PMMA samples (350 kDa and 996 kDa, both at 15 wt%). The results for PCL
nanobers (Figure 11.3) show that an increase in polymer concentration should reduce
the dye release rate. However, in Figure 11.4, the 120 kDa PMMA sample at 24 wt%
revealed a much higher release rate compared to the other two PMMA samples, both
electrospun at 15 wt%. Therefore, the effect of the molecular weight on release rate is
dramatically stronger than the effect of initial polymer concentration. This could be
attributed to lower nanoporosity of nanobers produced from higher-molecular-weight
polymers, which decreases the overall desorption of dye molecules from nanopore
surfaces.
Nanoporosity of nanobers deserves additional discussion. Some groups (e.g. Casper
et al. 2004) studied surface porosity of polystyrene nanobers electrospun from solutions
in tetrahydrofuran (PS in THF) in humid air, and reported formation of surface pits at
humidity levels higher than 25%. This porosity was attributed to the interaction of water
vapor with the polymer solution and was related to the breath gures (solventevaporative cooling, which triggers water condensation) and spinodal decomposition
396
396 [380--431] 10.12.2013 8:15PM
1.6
(1)
1.4
1.2
1
0.8
(2)
0.6
(3)
0.4
0.2
0
0
500
1000
1500
Time (h)
Figure 11.4
Effect of polymer molecular weight on dye release from electrospun PMMA nanobers. The initial
amount of dye per unit polymer mass was practically identical (about 3%) in all three samples.
(1) PMMA with Mw = 120 kDa, (2) PMMA with Mw = 350 kDa, (3) PMMA with Mw = 996 kDa.
The solvent was pure DMF. Reprinted with permission from Srikar et al. (2008). Copyright 2008,
(polymer phase separation). The pit diameter increased with polymer molecular weight.
As a result of this surface porosity, the surface area of the bers increased by 3065%,
although no clear trend in the dependence of the area increase on polymer molecular
weight was found. In Srikar et al. (2008), however, all the bers were electrospun in an
arid environment with humidity of 1923%, where no surface porosity, similar to that of
Casper et al. (2004), would be expected. As reported in the latter reference, on the
background of relatively large surface pits of the order of several hundreds of nanometers
(micropores), numerous nanopores of the order of 10 nm (Figure 7 in Casper et al. 2004)
were visible, seemingly unrelated to the ambient humidity. These pores, as well as
smaller ones, are the nanopores relevant to the discussion of saturation of the release
process well below 100%. Indeed, one should expect that these pores penetrate deep
inside the nanobers, most probably forming an interconnected network with a tremendous surface area, much larger than the outer surface, and diminish in number as the
molecular weight increases. The reasons for such expectations are the following. The
experimental and theoretical results of Dror et al. (2007) showed that solvent evaporation
in electrospinning proceeds mostly via direct permeation of the polymeric matrix comprising the ber. The process is accompanied by nanopore formation and should be
initiated by nucleation and bubbling of solvent vapor in the bulk of the polymer solution
in the ying ber. Resistance to such bubbling is provided by the solution viscosity. In
397 [380--431] 10.12.2013 8:15PM
Figure 11.5
397
SEM image of broken as-spun coreshell nanobers coated with gold. Zussman et al. (2006b).
Copyright Wiley-VCH Verlag GmbH and Co. KGaA. Reproduced with permission.
concentrated polymeric systems the viscosity strongly increases with molecular weight
at constant concentration, namely ~ M 3:4
w (Doi and Edwards 1986). In addition, at
constant molecular weight, ~ C15/4, where C is the polymer concentration (Doi and
Edwards 1986). Therefore, nucleation and opening of a pore in polymeric systems
with higher molecular weights and initial polymer concentrations should be strongly
suppressed by increased viscous resistance, similarly to polymer devolatilization
(Yarin et al. 1999). As a result, nanoporosity at higher molecular weights and initial
polymer concentrations should be lower.
SEM images of porosity visible on the surface or in a broken cross-section of both
monolithic (single-phase) electrospun and coreshell co-electrospun bers were published by several groups (e.g. Zussman et al. 2006b and Casper et al. 2004). While these
images (Figure 11.5) denitely conrm the presence of nanoporosity in the bers, they do
not allow estimates of the depth of penetration of nanopores, their interconnectivity or
tortuosity. In addition, these images lack the necessary resolution to visualize nanopores
of the order of several nanometers in diameter (which are probably the most prevalent
nanopores). Therefore, the release rates from nanobers, which shed light on the
presence of nanopores and the physical processes taking place at the nanopore surfaces,
is of immense value for improved understanding and interpretation of the real role of
nanoporosity in many important applications.
In the case of electrospun nanobers, the nanopore diameters are much smaller than
the ber diameters (~10 nm versus several 100 nm, respectively). Therefore, on the scale
of a nanopore, the curvature of the ber surface is practically not felt, and the ber
diameter itself has no or little direct effect on the release rate.
398
398 [380--431] 10.12.2013 8:15PM
The release processes from the PMMA bers shown in Figure 11.4 last for about
1400 h, but only 0.5 to 1.4% of the dye has been released. In comparison, PCL bers in
Figure 11.3 released from 32 to 67% of the dye. The temptation to attribute the drastic
difference in the release rates between PMMA and PCL to the molecular weight difference should be met with caution. Indeed, the molecular weight of PCL is 80 kDa, which
is not much smaller than that of the rst of the PMMA samples (120 kDa), whereas the
difference in their release rates is drastic. This means that not only the nanopore surface
but also the chemical nature of the nanober material (in particular, its binding energy to
dye molecules in the presence of water) is important.
Rhodamine dye forms nonaggregated (isolated) molecular dispersions in such
polymer matrices as PMMA (which was conrmed by Srikar et al. 2008 using individual
ber images obtained with an optical uorescence microscope). Therefore, in the experiments corresponding to Figures 11.3 and 11.4, dissolution of dye crystals prior to the
desorption-driven release stage cannot delay or decelerate the release process. Also, the
dye concentrations in polymer solutions were well below the solubility limit (if any).
Therefore, solubility cannot limit the release process in the present case, unlike the
experiments of Saltzman and Langer (1989), where microscopic protein particles embedded in a polymer matrix dissolved before being released via diffusion through macroscopically large pores lled with water. On the other hand, in the work of Srikar et al.
(2008), dealing with release of a molecular dye dispersion from electrospun nanober
mats, surface phenomena (e.g. desorption from the pore surface) neglected by Saltzman
and Langer (1989) seem to be important and thus should be accounted for. Below we
assume that the overall nanopore surface area in a ber is signicantly larger than the
smooth outer surface area of the ber, which is a plausible approximation for the
morphology of electrospun nanobers (Zussman et al. 2006b).
Consider a nanopore in a nanober as sketched in Figure 11.6. The pore has length L
and cross-sectional radius b. The cross-section at x = 0 corresponds to the ber surface
and represents an opening to a water-lled bath where the ber is immersed. We assume
that in most nanobers the pores have openings to the outer surface of the ber and thus
are accessible to water from the outside, as the results of Dror et al. (2007) suggest.
A real nanopore can be tortuous rather than straight; however, this should have only a
minor effect on the release process outlined below. The time required for a nanopore to
ll with water is negligibly short compared to the duration of the dye release process.
Indeed, according to Washburns equation, the time needed to ll the nanopore is
f = 2L2/(b cos ), where and are water viscosity and surface tension, and
is the contact angle (Washburn 1921, Levich 1962). Taking as estimates L = 104 cm,
b = 10 nm, = 102 P, = 102 mN m1 and cos of the order of 1, we nd f = 2 s, which
is much shorter than the duration of the release processes (e.g. 100 h in Figure 11.3 or
1000 h in Figure 11.4). Therefore, nanopores are modeled as being lled with water
from the very beginning of the release process. Assume that the nanopore surface layer
contains dye embedded in the polymer matrix when the nanober was electrospun and
the solvent evaporated. Assume a uniform surface density of dye at the beginning of the
release process (t = 0), when the nanopore was just lled with water. Then, the initial dye
surface density is sd0 = p2aMd0/Mp, where p is the average polymer ber density, 2a is
399 [380--431] 10.12.2013 8:15PM
399
Nanopore
2b
x=0
c=0
c = cw
x
x=L
Figure 11.6
Schematic of a cylindrical pore inside a polymer matrix. The pore is open at x = 0 and closed
at x = L. Reprinted with permission from Srikar et al. (2008). Copyright 2008, American Chemical
Society.
the molecular size and Md0/Mp is the initial dye/polymer mass ratio in the nanober.
Surface density of the polymer is sp = p2a >> sd0. It should be emphasized that the
dimensions of both sd0 (and of the surface density of dye at t > 0, sd) and sp is kg m2.
Dye desorption from the nanopore surface results in a certain initial bulk concentration of
dye in water near the surface cw0, with dimensions kg m3 (subscript 0 denotes t = 0). This
quantity is proportional to the specic initial dye concentration sd0/sp at the nanopore
surface, i.e., cw0 = k(T)sd0/sp, where the factor k(T) is given by the Clapeyron-like (or
the Arrhenius-like) dependence k(T) = k0exp(E/RT) with the pre-exponential k0 having
dimensions kg m3, E being the desorption enthalpy (or the activation energy), R the
universal gas constant and T the temperature. Obviously at t > 0, a similar expression,
cw = k(T)sd/sp holds, with sd< sd0. Since the dye is assumed to be water soluble
(a good model of water-soluble growth factors or drugs), its molecules diffuse in water
with a diffusion coefcient D. The release rate from the nanopore is estimated by the mass
ux through the opening at x = 0 as J ~ Dcw0b2/L. The total mass of dye to be released is
estimated as G ~ sd0Lb, since, for narrow nanopores with L >> b, the dye mass at the pore
end at x = L is negligibly small compared to that of the lateral surface. In light of
the experimental results described above, we assume that dye can be released only
from the surface layer of the polymer, whereas dye buried in the polymer bulk will not
be released at all. Then, the characteristic time for the release process is r = G/J = L2/Deff,
where the effective diffusion coefcient is Deff = Dcw0b/sd0. Take as estimates the
following realistic values of the parameters: k0 = 1 kg m3, p = 103 kg m3, b = 10 nm,
2a = 1010 m, E = 52.7 kJ mol1, T = 300 K, D = 109 m2 s1. Then, Deff = 1019 m2 s1.
400
400 [380--431] 10.12.2013 8:15PM
The fact that the value of Deff is so small compared to the diffusion coefcient of dye
in water D is directly related to the fact that the release process has two stages. The rst
stage dye desorption from the nanopore wall is the limiting stage, whereas the second
stage dye diffusion in water is relatively fast. The rate of the entire process is dictated by
the rate of the limiting stage, and that is the reason that Deff/D << 1. For a typical nanopore
L = 1 m (a tortuous nanopore of that size could be easily accommodated in a typical
nanober of cross-sectional diameter 100 nm, several centimetres in length). Then, r ~ 107
s, which is more than 42 days. This value is in very good agreement with the typical
duration of release processes (including those in the experiments discussed above). Note,
that theories of release based on the assumption of solid-state diffusion (Ritger and Peppas
1987, Chew et al. 2005, Wada et al. 1995121, Miyajima et al. 1997, 1998, 1999), would
attribute Deff to solid-state diffusion, since to describe the duration of the experimentally
measured process one needs a diffusion coefcient much less than that in water (D = 10 m2
s1). These theories, however, cannot explain release saturation well below 100%, as found
in many cases. The present approach does provide a detailed physically plausible mechanism of controlled release. It relates Deff to the basic physical parameters. The present
desorption-limited model of release also explains saturation of the release process at levels
well below 100%.
Based on the pore model outlined above, consider the release process in detail. It
should be emphasized that the situation we describe resembles, to some extent, a
preparative chromatography (sorption) problem (Guiochon et al. 2006, Tikhonov and
Samarskii 1990), albeit totally different. Diffusion of dye in the water lling the nanopore
is described by the following dimensionless equation

Def f c 2 c 1
c
2
r
11:1
D t x
r r r
where c is the dye concentration in water, rendered dimensionless by p, x is the axial
coordinate in the nanopore, rendered dimensionless by its length L (x = 0 at the pore open
exit, x = 1 at the pore end, Figure 11.6), r is the radial coordinate in the pore, rendered
dimensionless by L, and t is the time, rendered dimensionless by the estimated characteristic time of the release process r = L2/ Deff. Since Deff/D << 1, the left-hand side in
Eq. (11.1) can be neglected and the dye concentration eld in the nanopore becomes
quasi-steady. The rst boundary condition to be enforced is c < at r = 0. The other
boundary and initial conditions will be discussed below. As a result, the quasi-steady
concentration eld of dye in the nanopore is given by
cx; r; t
X
A00 t
B00 tx
I0 n rA0n t cos n x B0n t sin n x
2
n1
11:2
where we expect a discrete eigenvalue spectrum n, since the problem is solved on

the nite interval 0 x 1and I0() is the modied Bessel function. The coefcients
An and Bn for n 0 will be found from other conditions, discussed below.
The surface concentration of dye at the nanopore surface is governed by the following
dimensionless mass balance
401 [380--431] 10.12.2013 8:15PM

sd
DT c

t
L2a r rb=L
401
11:3
where sd is rendered dimensionless by p2a. Also, due to the desorption law

cjrb=L f sd
11:4
where the factor f is given by f = (k0/p)exp(E/RT).

Equations (11.3) and (11.4) form the second boundary condition imposed on the dye
concentration eld (Eq. 11.2) in the nanopore. Combining Eqs. (11.3) and (11.4) and
using Eq. (11.2), we nd
0

X
1 dA00 dB00
dA n
dB0n
x
cos n x
sin n x
I0 n b=L
dt
dt
dt
2 dt
n1
Dr f X
n I1 n b=LA0n cos n x B0n sin n x
L2a n1
11:5
Equation (11.5) shows that the coefcients A0 and B0 do not depend on time, whereas
the other coefcients can be found from the following equations
dKn
Dr f n I1 n b=L

Kn
dt
L2a I0 n b=L
11:6
where Kn is either An or Bn (n 1).

It is clear that all modes with large values of n will vanish almost immediately,
according to Eq. (11.6). Also, b/L << 1 in the present case. Therefore, we have nb/L < <1
and thus, I1(nb/L) nb/2L, whereas I0(nb/L) 1. Then, Eq. (11.6) takes the form
dKn
Dbr f 2n
2
Kn
dt
L 2a 2
11:7
Note that r = L2/Deff = L2/[Dcw0b/sd0]. Also, from Eq. (11.4) cw0 = fsd0/2a (dimensional).
Therefore, Dbrf/(L22a) = 1, and the integration of Eq. (11.7) yields Kn = Kn0exp(2n t/2).
The result for the coefcients Kn clearly shows that the evolution with time follows the exp
(const. t) law (with the constant being of the order of one), where time t is rendered
dimensionless by r = L2/ Deff. The latter shows (in accordance with the simple estimate
given above) that the release proceeds on the timescale r = L2/ Deff (>> L2/ D), which is
determined by the rate-limiting stage of dye desorption. The initial values of the coefcients
Kn0 (either An0 or Bn0) are determined by the initial distribution of the dye surface
concentration over the nanopore. The initial surface density of dye at t = 0 is given by the
Fourier series
(
)
h
X
nx
nxi
sd0 1=f A00 =2
Bn0 sin
An0 cos
11:8
2
2
n1
where the coefcients, in fact, also appear in the distribution of cw0 due to Eq. (11.4).
402
402 [380--431] 10.12.2013 8:15PM
The distribution (Eq. 11.8) should not be uniform, even though dye might be uniformly distributed at the moment when the ber was formed. Indeed, immediately (as
compared to the release timescale r) after the nanopore is lled with water, the dye
distribution along the surface should accommodate a certain prole with sd0 = 0 at x = 0,
i.e. at the pore exit, where the initial mass ux is effectively innite (since in the bath c = 0
due to its very large size compared to that of the nanopore). This means that all the
coefcients in Eq. (11.8) should be zero, except Bn0 = Bn0 with n =1, 5, 9,. . . Also,
according to Eqs. (11.2), (11.4) and (11.8), A0 = A00 = 0, B0 = 0 and n = n/2.
Therefore, according to Eqs. (11.2) and (11.4), at t > 0 for 0 x 1:
sd 1=f
Bn0 exp2 n2 t=8 sin nx=2
11:9
n1;5;9;
Modes with n > 1 vanish very rapidly, and the surface density prole of dye will evolve
according to the following dimensional formula:
2
t
x
sd sd00 exp
sin
11:10
8 r
2L
where we take B10 = sd00 = p2aMd0/Mp.
It is interesting to note that in parallel, via Eq. (11.2), we have c=xjxL 0, which
means that we chose the eigenfunctions that do not accommodate dye release from the
nanopore bottom. This is immaterial anyway, since the release from the bottom is
negligibly small compared to the release from the lateral surface due to the condition
b/L << 1. In the general case, appropriate eigenfunctions in Eq. (11.8) could always be
chosen to accommodate dye release at the nanopore bottom x = L.
Equation (11.10) allows calculation of the mass of dye G released from the nanobrous material:

2
G
t
11:11
1 exp
8 r
Md0
where the nanoporosity factor = Msd0/(M sd0 + Mbd0) < 1, with M sd0 and Mbd0 being the
initial amount of dye at the nanober surfaces (in particular, at the nanopore surfaces) and
the initial amount of dye embedded in the ber bulk, respectively; the total initial amount
of dye in the ber Md0 = Msd0 + Mbd0. According to Eq. (11.11), the release process will
saturate at percentage dye release = 100%, which is less than 100%. The nanoporosity
factor appearing in Eq. (11.11) is affected by the nanopore surface area and thus should
depend on both polymer concentration in the electrospun solution and polymer molecular weight. Conversely, the characteristic time r depends only on the polymer density
p, and the desorption process parameters k0 and E, which characterize the intermolecular
forces binding dye molecules to the polymer surface in the presence of water. Therefore,
r, k0 and E should be essentially insensitive to the polymer concentration and molecular
weight. They should manifest only the chemical nature of the polymerdye interactions
responsible for sorptiondesorption processes in the presence of water.
403 [380--431] 10.12.2013 8:15PM
403
(1)
0.7
(2)
0.6
(3)
0.5
0.4
0.3
0.2
0.1
0.0
0
50
100
150
200
250
Time (h)
Figure 11.7
Theoretical curves and experimental data (symbols) for dye release from PCL nanobers. The
initial amount of dye per unit polymer mass was practically identical (about 0.2%) in all three
experimental samples, where the polymer concentration was 11 wt% (1), 13 wt% (2) and 15 wt% (3).
The experimental data is replotted from Figure 11.3 and shown by the corresponding symbols. The
curves corresponding to Eq. (11.11) closely t the data with an appropriate choice of r and .
Reprinted with permission from Srikar et al. (2008). Copyright 2008, American Chemical Society.
Comparison of Eq. (11.11) with the experimental data from Figures 11.3 and 11.4
allows one to establish the parameters of the dye desorption process for different
polymers in water, and in particular to estimate the desorption enthalpy (or the activation
energy) E. Two examples of such a comparison are shown in Figures 11.7 and 11.8.
When the value of r is established by comparing the experimental data to Eq. (11.11),
as in Figures 11.7 and 11.8, the value of Deff is found as Deff = L2/r . The value of L was
taken as 1 m. The values of r and Deff obtained are listed in Table 11.2, and appear
to be essentially insensitive to polymer concentration variation among the PCL samples
and to polymer molecular weight variations among the PMMA samples. Then, the
factor k(T) is found according to the theoretical relation k(T) = p2aDeff/(Db), with the
following values: 2a = 1010 m, b = 10 nm and D = 109 m2 s1. The values of the polymer
density p for the polymers used, and the corresponding values of k(T), are listed in
Table 11.2. From the Clapeyron-like desorption law, the desorption enthalpy E is found
as E RTnkT=k0 , with the value of the pre-exponential k0 taken as k0 = 1 kg m3
and T = 300 K. The values of the desorption enthalpy E found using this procedure for
PCL and PMMA nanobers are also listed in Table 11.2. Note that for all three PCL
nanober samples with different polymer concentrations the value of E is ~37 kJ mol1,
whereas for all three PMMA samples with different molecular weights it is ~45 kJ mol1.
This indicates that dye desorption from PCL is much easier than from PMMA. It also
shows that desorption enthalpy is a function of only intermolecular forces acting on a
404
404 [380--431] 10.12.2013 8:15PM
Table 11.2 Basic release parameters, including desorption enthalpy E and nanoporosity factor .
Nanober
11% PCL
13% PCL
15% PCL
PMMA (Mw = 120 kDa)
PMMA (Mw =350 kDa)
PMMA (Mw =996 kDa)
r
(h)
Deff
m2 s1
p
kg m3
k(T)
kg m3
E
kJ mol1
11.862
10.281
9.7526
308.425
262.489
246.74
2.341017
2.701017
2.851017
9.011017
1.061018
1.131018
1.14103
1.14103
1.14103
1.18103
1.18103
1.18103
2.67107
3.08107
3.25107
1.06108
1.25108
1.33108
37.7522
37.3958
37.2643
45.7931
45.3908
45.2365
0.666
0.496
0.320
0.0144
0.0066
0.0050
1.6
(1)
(2)
1.4
(3)
1.2
1.0
0.8
0.6
0.4
0.2
0.0
0
200
400
600
800
1000
1200
1400
Time (h)
Figure 11.8
Theoretical curves and experimental data (symbols) for dye release from PMMA nanobers. The
initial amount of dye per unit polymer mass was practically identical (about 3%) in all three
samples. (1) PMMA with Mw = 120 kDa, (2) PMMA with Mw = 350 kDa, (3) PMMA with
Mw = 996 kDa. The experimental data is replotted from Figure 11.4 and is shown by the
corresponding symbols. The curves corresponding to Eq. (11.11) closely t the data with an
appropriate choice of r and . Reprinted with permission from Srikar et al. (2008). Copyright 2008,
polymerdye system in the presence of water, and thus it is independent of polymer

concentration or molecular weight. It should be emphasized that if dye dissolution were
the limiting mechanism of the release process, as per Saltzman and Langer (1989), the
release parameters (e.g., E) should be characteristic of the dye alone and should not differ
between the polymers. Therefore, different values of E for PCL and PMMA releasing the
same rhodamine 610 chloride dye support the view that desorption of molecularly
405 [380--431] 10.12.2013 8:15PM
405
dispersed dye from the polymer matrices rather than dye dissolution is the ratecontrolling process.
The values of the nanoporosity factor listed in Table 11.2 appear to be sensitive to
polymer concentration and molecular weight variation. These parameters determine the
sample nanoporosity and thus the ultimate release value characterized by .
It was found that rigorous shaking on a clinical rotator and a higher overall sample
porosity (a larger interber spacing in the samples) also facilitated dye release, since they
enhance water penetration into nanober nanopores, as well as introduce a convective
component that makes desorption easier. The latter is reminiscent of the enhancement of
mechanical degradation of polymer macromolecules by stretching, which effectively
reduces the activation energy for polymer scission (Bueche 1955, 1960, Zhurkov and
Korsukov 1974, Termonia et al. 1985, Yarin 1991,1993). Therefore, the values of E in
Table 11.2, in fact, incorporate the effect of convection (resulting from shaking) on the
desorption process.
It is of interest to estimate whether the ber collection conditions could have had any
effect on the nanopore surface area and thus on the dye release rate in the given experiments. Indeed, the PCL nanober mats were collected on a planar grounded electrode,
whereas all other bers were collected on a grounded vertically rotating (1150 rpm)
wheel collector (see Figure 5.54 in Section 5.9 of Chapter 5) with a metal ribbon mounted
over the wheel blade. The wheel radius Rw = 7.5 cm and the linear velocity at its edge was
about 9 m s1. This velocity is close to the typical velocity values in the electrospun jets
(Reneker et al. 2007), and due to this, fast stretching of the jet by the wheel in this case
did not happen and signicant stress-hardening of individual nanobers was not
observed (Zussman et al. 2006a; see Figure 7.3 in Section 7.1 of Chapter 7). To estimate
the effect of ber stretching by the wheel on the surface area of nanopores, consider
uniaxial elongation of a ber with a stretching ratio . It is easy to show that the surface
area of a nanopore inclined by an angle to the stretching axis changes by the factor
of (sin2 + 2cos2)1/4(1sin2 + 2cos2)1/2. In particular, the axially oriented pores
( = 0) will experience an increase in their surface area by a factor of 1/2, whereas the
radially oriented pores ( = 90) will experience a decrease in their surface area by a
factor of 1/4. The largest value of in the additional stretching due to the rotating wheel
is given by max={[(Rw/H)+1]2 (Rw/H)2}1/2, where H is the distance from the needle to
the wheel top. In the present case Rw = 7.5 cm, H = 12 cm and max = 1.5. Therefore,
the surface area of the axially oriented pores will increase by a factor of 1.22, whereas
the surface area of the radially oriented pores will decrease by a factor of 0.9. For pores
with an intermediate orientation, the largest increase in the surface area is about 1.25
times. All these values are close to 1. Therefore, the variation of the pore surface area due
to the additional stretching by the wheel is insignicant, and the drastic difference
between the release rates from the PCL bers (electrospun on the grounded plate) and
the PMMA ones (electrospun on the wheel) is unaffected by wheel stretching.
Gandhi et al. (2009) studied the release of two model protein compounds from
electrospun polycaprolactone (PCL) nanober mats. They were bovine serum albumin
(BSA) and an anti-integrin antibody (AI). The latter was chosen as a model of antibody
therapy; in particular, anti-integrin antibodies are a promising class of therapeutic
406
406 [380--431] 10.12.2013 8:15PM
Table 11.3 Basic release parameters including the desorption enthalpy E and nanoporosity factor .
r
(h)
Nanober
11% PCL+BSA
13% PCL+BSA
15% PCL+BSA
p
kg m3
k(T)
kg m3
1.14103
1.14103
1.14103
2.24107
2.04107
1.42107
Deff
m2 s1
1.971017
1.791017
1.251017
14.13
15.53
22.24
E
kJ mol1
38.19
38.43
39.32
0.87
0.6
0.4
100
(1)
Percentage release
80
60
(2)
40
(3)
20
0
0
100
200
300
400
500
600
700
800
Time (h)
Figure 11.9
Effect of initial polymer concentration on BSA release from PCL nanober mats. The experimental
data are shown by symbols(n = 13; averagestandard deviation), the corresponding theoretical
results, Eq. (11.11), by curves. The nal release attained was 87% from 11% PCL (1,), 60%
from 13% PCL (2, ) and 40% from 15% PCL (3,), within the period indicated. The higher
release rates correspond to the higher released masses reached at the same time. Reprinted with
permission from Gandhi et al. (2009). Copyright (2009) American Chemical Society.
molecules for cancer and angiogenic diseases. The release kinetics were studied experimentally and interpreted in the framework of Eq. (11.11). The results shown, in
particular, in Figure 11.9 are consistent with a protein release mechanism dominated
by desorption from the polymer surface, while the polycaprolactone nanobers are not
dissolving at appreciable rates. The corresponding desorption parameters are listed in
Table 11.3. The value of the desorption enthalpy E for BSA release from electrospun
nanobers in Table 11.3, E 3839 kJ mol1, differs from the one for rhodamine
407 [380--431] 10.12.2013 8:15PM
11.4 Modulation of drug release rate
407
610 chloride release from such bers, E = 3738 kJ mol1 in Table 11.2. The difference is
attributed to the different intermolecular forces between PCL and BSA versus PCL and
rhodamine 610 chloride.
11.4
Modulation of drug release rate

Although electrospun nanobers are a promising material for drug-delivery applications,
practically no active control of the release process was available. To accomplish this
goal, stimuli-responsive properties were implemented in nanober mats. In particular, to
manipulate the release rate on demand, Zhang and Yarin (2009), Sinha-Ray et al. (2010)
and Zhang et al. (2011) electrospan nanobers containing hydrogels: poly(N-isopropylacrylamide) (PNIPAM) and the related co-polymers poly(methyl methacrylate/N-isopropyl
acrylamide), (P(MMA/NIPAM)), and poly(methyl methacrylate/N-isopropyl acrylamide/
acrylic acid), (P(MMA/NIPAM/AAc)). Such nanobers are easily tunable and their
smart behavior can be triggered by changes in temperature and pH. PNIPAM is thermoresponsive and shrinks above its lower critical solution temperature (LCST) of 32.8 C.
The LCST of P(MMA/NIPAM) co-polymers could be as high as 52 C, whereas P(MMA/
NIPAM/AAc) co-polymers are pH sensitive.
Sinha-Ray et al. (2010) had electrospun thermo-responsive nanobers containing
PNIPAM nano-raisins and actively controlled the release kinetics of a model compound (uorescent dye rhodamine 610 chloride) incorporated into these bers.
Electrospun nanobers containing poly(acrylic acid), PAA, poly(vinyl alcohol), PVA,
and PNIPAM were annealed (cross-linked) after electrospinning at 150 C for 2.5 h. To
observe the internal structure of the resulting PAA/PVA/PNIPAM nanobers, they were
stained with CuSO4 and observed using transmission electron microscopy (TEM). In
particular, staining was used to reveal the inhomogeneity of the individual PAA/PVA/
PNIPAM nanobers resulting from the aggregation of PNIPAM in the form of raisins,
as was suggested in the preceding theoretical work (Yarin 2008). In the nanobers, only
PNIPAM contains nitrogen (as distinct from PAA and PVA), which is more likely to
coordinate with copper ions Cu2+ than other atoms, such as O. As a result, the coordination effect of Cu2+ with nitrogen can be used to label PNIPAM for further investigation
using TEM. Copper stains were indeed observable on PNIPAM nano-raisins, since Cu
labels PNIPAM nano-raisins alone, which facilitated their observation in TEM.
A representative TEM image of PAA/PVA/PNIPAM nanobers is shown in
Figure 11.10. The overall view of a single nanober in Figure 11.10a reveals multiple
black dot-like objects. Zooming-in (Figure 11.10b) shows that these objects are
nano-raisins of about 2 nm in diameter. These nano-raisins are the PNIPAM islands
stained by Cu from CuSO4.
One of the most important features of hydrogels, including PNIPAM, is their swelling
behavior. The uptake of water at T < LCST leads to an expansion of the hydrogel, usually
increasing its volume by several hundred percent compared to the dry (shrunken) state at
T > LCST. It was reported that when increasing the temperature from 25 C to 70 C, the
electrospun PAA/PVA/PNIPAM nanober mats produced by Chen and Hsieh (2004)
408
408 [380--431] 10.12.2013 8:15PM
20 nm
(a)
Figure 11.10
nm
10 nm
(b)
PNIPAM nano-raisins in a PAA/PVA/PNIPAM nanober. (a) TEM image of a single nanober.

(b) A zoomed-in area of the nanober shown in (a). The black nano-raisins shown by arrows
are of the order of 2 nm in diameter. Reprinted with permission from Sinha-Ray et al. (2010).
shrank to about one-quarter of their original size. Experiments with the electrospun PAA/
PVA/PNIPAM nanober mats produced by Sinha-Ray et al. (2010), however, yielded a
very different result: they practically did not shrink when the LCST was crossed, as
opposed to normal hydrogels, (Chen and Hsieh 2004). The reason for this distinction is
most probably a signicant difference in the molecular weight of PNIPAM in the paper
by Sinha-Ray et al. (2010) (Mw = 2025 kDa) versus Mw = 79203 kDa for Chen and
Hsieh (2004). The short-chain PNIPAM molecules in the nanobers of Sinha-Ray et al.
(2010c) facilitated the aggregation of PNIPAM chains as nano-raisins in the PAA/PVA
matrix at the annealing stage, as conrmed by direct observations in Figure 11.10.
However, even though the nanobers of Sinha-Ray et al. (2010) did not show standard
hydrogel behavior, they showed positive thermo-responsive dye release, i.e. a signicantly faster dye release at T > LCST (of about 32 C), as compared to T < LCST, as
demonstrated in more detail below. This means that only isolated PNIPAM nanoraisins in the PAA/PVA/PNIPAM nanobers of Sinha-Ray et al. (2010) behaved as
nanogels, whereas the nanober mats as a whole did not behave as hydrogels (i.e. did not
swell or shrink when the LCST was crossed).
To investigate the release kinetics of the PAA/PVA/PNIPAM nanober mats with
nano-raisins, a model compound, namely the low-molecular-weight uorescent dye
rhodamine 610 chloride, was incorporated into the bers by dissolving known quantities
of the dye in the polymer solutions prior to electrospinning. All release studies were
carried out at 10 C, 25 C (room temperature), 40 C and 55 C: the rst two values are
below the lower critical solution temperature of PNIPAM, about 32 C, whereas the latter
two are above it. These temperatures were chosen to represent three temperature ranges
of interest. Comparing the release rates at 10 C and 25 C represents the release prole at
T < LCST. By comparing the release rates at 40 C and 55 C, it was possible to
409 [380--431] 10.12.2013 8:15PM
409
investigate the release at T > LCST, and most importantly, comparing the data acquired
from the experiments at 25 C and 40 C elucidated the effect of the PNIPAM nanogel
swelling/shrinkage transition occurring at about 32 C on the release kinetics.
The results shown in Figure 11.11 clearly indicate that the release of the dye increases
with temperature, in the case of nanobers with a low PNIPAM concentration, as well as
for bers with medium and high PNIPAM content. It is also clear that the release changes
dramatically when the temperature is increased from 25 C to 40 C in all three cases,
while the release percentages change only marginally when the temperature is increased
from 10 C to 25 C. In the case of nanober samples with low PNIPAM concentration
(as shown in Figure 11.11a), the overall release percentage is very similar at 10 C and 25 C,
being about 7% after six days. The result is similar for the two other PNIPAM concentrations, as shown in Figures 11.11b and 11.11c, where the corresponding release percentages were found to be 12% and 14% after six days in the case of bers with a medium
PNIPAM content, and 24 % and 28 % for bers with a high PNIPAM concentration.
Since no effect of PNIPAM swelling/shrinkage transition is expected to occur at T < LCST,
the dye release is controlled by desorption only, because the release approaches saturation
well below 100% in Figure 11.11. It looks like the temperature dependence of the desorption
process is such that the increase in the release rate at lower temperatures is almost indistinguishable within a temperature interval of 15 C.
(a)
70
60
10 C
25 C
40 C
55 C
Dye release (%)
50
40
30
20
10
0
0
20
40
60
80
100
120
140
160
Time (h)
Figure 11.11
Cumulative dye release versus time from PAA/PVA/PNIPAM nanober mats immersed in water:
(a) for nanober mats with a low PNIPAM concentration, (b) for nanober mats with a medium
PNIPAM concentration, (c) for nanober mats with a high PNIPAM concentration. The polymer
solutions used for electrospinning consisted of 3:3:2 (low PNIPAM concentration), 1:1:1
(medium PNIPAM concentration) or 3:3:4 (high PNIPAM concentration) blends of PAA, PVA and
PNIPAM, with an overall polymer content in DMF of 10 to 12 wt%. Reprinted with permission
410
410 [380--431] 10.12.2013 8:15PM
(b)
45
40
35
10 C
25 C
40 C
55 C
Dye release (%)
30
25
20
15
10
5
0
0
20
40
60
80
100
120
140
160
Time (h)
(c)
55
50
45
Dye release (%)
40
35
30
25
20
10 C
15
25 C
10
40 C
55 C
0
0
20
40
60
80
100
120
140
160
Time (h)
Figure 11.11
(cont.)
On the other hand, when the temperature is increased from 25 C to 40 C, the release
percentages increase signicantly for all PNIPAM concentrations, to 33% in the case of
nanobers with low PNIPAM content, to 39% for bers with a medium PNIPAM
concentration and to 47% for bers with a high PNIPAM content. This effect can be
attributed to the PNIPAM swelling/shrinkage transition causing a drastic change in the
release kinetics when the temperature exceeds 32 C (approximately equal to the LCST).
411 [380--431] 10.12.2013 8:15PM
411
According to Figures 11.11ac the nanober mats exhibit positive thermoresponsive release when the temperature crosses the LCST of PNIPAM, i.e. release
increases above LCST. This is notable since the majority of PNIPAM-based hydrogels
reported in the literature are negatively thermo-responsive, with the release rate
increasing at temperatures below, and decreasing at temperatures above, the LCST
(Qiu and Park 2001).
When the temperature is increased even further, from 40 C to 55 C, a signicant
difference in the overall release corresponding to nanobers with different PNIPAM
concentrations is observed. Experiments using nanobers with a low PNIPAM concentration (Figure 11.11a) show that the overall release increased signicantly from 33% to
63% in six days. On the other hand, for the two ber samples with medium and high
PNIPAM contents, an increase in temperature from 40 C to 55 C practically does
not increase the six-day release, which amounts to 39% and 40% (medium PNIPAM
concentration, Figure 11.11b) and 47% and 48% (high PNIPAM concentration,
Figure 11.11c), respectively. The release trends depicted in Figures 11.11b and c are
consistent with either of the two mechanisms proposed in the literature: further exposure
of the existing nanopores (Kim et al. 2005) or formation of new ones in the vicinity of
PNIPAM nanogel islands (Yarin 2008), since each of these mechanisms should dramatically occur at the temperature transition through LCST. The increase in the overall
release over the temperature range 40 C to 55 C in Figure 11.11a could be attributed to
the dependence of dye desorption on temperature. The gain in desorption in the temperature range 40 C to 55 C should be larger than in the range 10 C to 25 C, albeit the
results in Figures 11.11b and c show that it probably is smaller than the range of error for
the measurements and signicantly less than that in the range 25 C to 40 C, where the
swelling/shrinkage transition of PNIPAM nano-raisins occurs. Therefore, the signicant gain in release in the range 40 C to 55 C seen in Figure 11.11a for the low PNIPAM
concentration must be facilitated by some other, still unknown, mechanisms in addition
to the temperature dependence of desorption.
To further investigate the inuence of the swelling/shrinkage transition of PNIPAM
nano-raisins and isolate it from other possible effects, Sinha-Ray et al. (2010) conducted experiments where the temperature was modulated during dye release from the
electrospun nanober mats. Three temperature ranges were analyzed: 10 C to 25 C, 25
C to 40 C and 40 C to 55 C, where the temperature was switched between the low and
high temperature values every 1.55 hours. A signicant effect of PNIPAM nano-raisin
thermosensitivity on the release rate was expected in the interval from 25 C to 40 C
containing the LCST.
The experiments with temperature modulation for all PNIPAM concentrations
(Figures 11.12ac) indicated that when the temperature was increased from 25 C to
40 C, the slope of the percentage of dye released increased visibly and decreased
accordingly when the temperature was switched back from 40 C to 25 C. The same
effect was also apparent for the other two temperature ranges, even though it was
signicantly smaller. To quantify this observation, Sinha-Ray et al. (2010) calculated
the change of slope in the graphs in Figure 11.12. The corresponding results are shown
in Figure 11.13.
412
412 [380--431] 10.12.2013 8:15PM
(a)
45
10 C to 25 C
40
25 C to 40 C
Dye release (%)
35
40 C to 55 C
30
25
20
15
10
5
0
0
10
15
20
25
30
Time (h)
(b)
40
10 C to 25 C
25 C to 40 C
40 C to 55 C
Dye release (%)
35
30
25
20
15
10
5
0
0
10
15
20
25
30
35
Time (h)
Figure 11.12
Cumulative dye release versus time for nanober mats subjected to temperature modulation for:
(a) bers with a low PNIPAM concentration, (b) bers with a medium PNIPAM concentration,
(c) bers with a high PNIPAM content. (d) The corresponding temperature variation for each case
(a), (b) and (c). Reprinted with permission from Sinha-Ray et al. (2010). Copyright 2010,
Figure 11.13 shows a very signicant effect associated with the temperature response
facilitated by PNIPAM nano-raisins. The slope changes dramatically by as much as
300700 % when the temperature is switched from 25 C to 40 C. This was found for all
investigated PNIPAM concentrations. The change of slope observed in the other two
temperature ranges is only of the order of 10%. This clearly indicates how the temperature response of PNIPAM at the swelling/shrinkage transition outweighs all other effects
caused by a change in temperature in the intervals that do not contain the LCST.
413 [380--431] 10.12.2013 8:15PM
413
(c)
45
10 C to 25 C
40
25 C to 40 C
Dye release (%)
35
40 C to 55 C
30
25
20
15
10
5
0
0
10
15
20
25
30
Time (h)
(d)
60
Temperature (c)
50
10-25 C
40
25-40 C
40-55 C
30
20
10
0
0
10
15
20
25
30
35
Time (h)
Figure 11.12
(cont.)
In the case of nanobers with low PNIPAM concentration, the change of slope for the
temperature range from 40 C to 55 C is signicantly higher than when switching from
10 C to 25 C. This is consistent with the data shown in Figure 11.11a.
The results demonstrate positive thermo-sensitive dye release (an increase in the
release rate at temperatures above LCST) that accompanies shrinkage of 2-nm
nanogel raisins embedded in nanobers above the LCST. The latter leads either
to further exposure of the existing nanopores (Kim et al. 2005), or formation of new
ones in the vicinity of the shrinking PNIPAM nanogel raisins (Yarin 2008).
Indeed, in the latter case raisin shrinkage at elevated temperatures results in elastic
stresses and strains on the surrounding material, which can lead to nanocracking.
414
414 [380--431] 10.12.2013 8:15PM
Change of slope (%)
(a)
800
700
10 C to 25 C
600
25 C to 40 C
500
40 C to 55 C
400
300
200
100
0
100
200
10
15
Time (h)
(b)
600
10 C to 25 C
500
25 C to 40 C
Change of slope (%)
400
40 C to 55 C
300
200
100
0
100
200
10
15
20
25
30
35
Time (h)
Figure 11.13
Release rate modulation with temperature for: (a) nanobers with a low PNIPAM concentration,
(b) nanobers with a medium PNIPAM concentration, (c) nanobers with a high PNIPAM
concentration. Each data point represents the change in slope of the corresponding curves shown in
Figure 11.12. Reprinted with permission from Sinha-Ray et al. (2010). Copyright 2010, American
Chemical Society.
415 [380--431] 10.12.2013 8:15PM
415
(c)
400
10 C to 25 C
25 C to 40 C
Change of slope (%)
300
40 C to 55 C
200
100
100
200
10
15
20
25
30
Time (h)
Figure 11.13
(cont.)
This brings more dye into contact with water and facilitates dye release in those
areas.
Among the three different modications of P(MMA/NIPAM) co-polymers with different
NIPAM content published by Zhang and Yarin (2009), the LCST varied from 38 C to 52
C. The shrinkage percentage at crossing the LCST, when the corresponding electrospun
nanober mat samples were fully dipped in water, varied from 35.56 to 73.21 (which is
typical thermo-responsive hydrogel behavior), whereas the swelling percentage was 30.05
to 39.33% (which shows that the transition was sufciently reversible). The NIPAM mass
retained in four days was from 89.94 to 99.62 % (electrospun nanober mats), which means
that only between 0.4 to 10% of the NIPAM was dissolved from these co-polymers
submerged in water for four days. An example of the shrinkage of three different P
(MMA/NIPAM) co-polymer electrospun nanober mats when the temperature was
increased above their LCSTs is shown in Figure 11.14.
For the three different modications of P(MMA/NIPAM/AAc) co-polymers synthesized by Zhang and Yarin (2009), the pH response threshold varied from 4.4 to 6.4, and
the NIPAM mass retained in four days was 32.67 to 68.27% (electrospun nanober mats;
between 32 to 67% of the NIPAM was dissolved from these co-polymers submerged in
water for four days).
Release of the uorescent dye rhodamine 610 chloride from the P(MMA/NIPAM) and
P(MMA/NIPAM/AAc) co-polymers was used by Zhang and Yarin (2009) to probe their
thermo- and pH-driven mechanical responses. Pieces of co-polymer nanober mat with
an LCST of 52 C and encapsulated dye were located in a water bath, and the temperature
was periodically changed. Four series of experiments were carried out with a temperature
switch from 5 to 25 C for series 1, from 25 to 45 C for series 2, from 45 to 65 C for
series 3 and from 65 to 85 C for series 4 (see Figures 11.15 and 11.16). Unlike the release
416
(a)
Figure 11.14
416 [380--431] 10.12.2013 8:15PM
(b)
(a) Three samples of different P(MMA/NIPAM) co-polymers (in the form of electrospun
nanober mats), each one in its own compartment of a water-lled petri dish. The samples oat at
the free surface of the water layer; the water temperature visible on the thermometer in the image is
(a) 23 C and (b) 58 C. Zhang and Yarin (2009). Reproduced with permission from The Royal
experiments from nanobers with nano-raisins, which do not show a global swelling/
shrinkage transition at the LCST, as discussed above, these results describe dye release
from co-polymers that behave as ordinary hydrogels (possessing signicant global
swelling/shrinkage).
In Figure 11.15 series 1 and 2, with temperature variations below the LCST of 52 C,
show a relatively low cumulative release of the order of 1%. The temperature sensitivity
for these temperature intervals is still of the order of 10%, albeit that it increases to 200%
at 45 C (close to the LCTS); see Figure 11.16. In series 3 in Figure 11.15 the temperature
crosses the LCST. Over that temperature interval the cumulative release rapidly achieves
a level of about 10% and saturates at about 12%. Also, the release rate for this series
shows the largest thermal response (Figure 11.16). On the other hand, series 4, where
temperature varies above the LCST, demonstrated a cumulative release curve that
resembles that of series 3 (Figure 11.15). This means that the major release event happens
at the preparatory stage of heating to 65 C (the starting temperature of series 4), when
temperature inevitably crosses the LCST.
The response to pH modulation of P(MMA/NIPAM/AAc) co-polymer electrospun
nanober mats submerged in a water bath was probed by Zhang and Yarin (2009) at a
temperature of 37 C. There were three series of experiments with pH varying from 5
to 6 for series 1, from 6 to 7 for series 2 and from 7 to 8 for series 3 (see Figures 11.17
and 11.18).
The results of the experiments with dye release from electrospun nanober mats of P
(MMA/NIPAM/AAc) co-polymer with pH modulation presented in Figures 11.17 and
11.18 show that a drastic change in the cumulative release levels and the release rate
response happens between pH 6 and 7. This makes such co-polymer a potentially
important candidate for drug carriers capable of distinguishing between cancer tumors
417 [380--431] 10.12.2013 8:16PM
417
11.5 Health supplements
14
(4)
(3)
12
10
(2)
(1)
0
0
10
15
20
25
Time (h)
Figure 11.15
Cumulative dye release versus time for P(MMA/NIPAM) co-polymer electrospun nanober
mats subjected to temperature modulation in a water bath. See text for details of series 1 to 4. Zhang
and Yarin (2009). Reproduced with permission from The Royal Society of Chemistry.
and normal tissues. Indeed, cancer tumors and sites of inamation are typically more
acidic (pH 6.5) than normal tissues (pH 7.4) (Filippov et al. 2008).
11.5
Health supplements (vitamin-loaded nanofiber mats)

Cellulose acetate nanober mats incorporating -tocopherol (vitamin E) and transretinoic acid (vitamin A) as transdermal and dermal therapeutic agents (Figure 11.19),
respectively, were electrospun by Taepaiboon et al. (2007). Both the retin-A-loaded cast
lm and the retin-A-loaded as-spun nanober mat showed smooth release. The retin-Aloaded ber mat showed a gradual increase in cumulative release over the 6 h testing
period, with a maximum release of ~34%. Similar results were also observed for
tocopherol-loaded nanobers. However, vitamin E release was very rapid because of
the methanol-mediated partial dissolution of the bers from both of the vitamin-E-loaded
samples. The release of vitamin E from the loaded as-spun ber mat increased gradually
and monotonously to reach a maximum value of ~95% at 24 h (Taepaiboon et al. 2007).
Vitamin B12-loaded PNIPAM/PEO nanobers were fabricated as a thermo-responsive
drug-delivery system for the model drug and demonstrated to be a suitable carrier for
418
418 [380--431] 10.12.2013 8:16PM
500
400
(3)
Change of slope (%)
300
200
(4)
100
(2)
(1)
100
200
0
10
15
20
25
Time (h)
Figure 11.16
Change of the release rate due to temperature modulation corresponding to the experiments in
Figure 11.15 on electrospun nanober mats. Zhang and Yarin (2009). Reproduced with permission
from The Royal Society of Chemistry.
controlled release. The release kinetics of the vitamin was found to be dependent on the
ratio of PNIPAM/PEO and on the loading of the vitamin. Also, the release rate was high
at room temperature compared to body temperature release from the nanober matrix
(Song et al. 2011). Vitamin B12-loaded PCL nanobers as transdermal patches were also
reported recently (Madhaiyan et al. 2013).
11.6
Cosmetic facial masks

The nanotechnology revolution has proved to be valuable not only for drug delivery and
tissue engineering, but also in cosmetic applications. The tools of nanoparticulate drug
delivery, especially to the skin, namely lipidnanoparticle and lipiddrug conjugates,
have proven to be more safe and specic in the delivery of an active drug moiety as well
cosmetic ingredients into skin. Nanoparticles have the capacity to increase or decrease
the ux, tailor the drug depot location and size, and also selectively permeabilize the
stratum corneum. Still there is a need for knowledge of the interactions of nanoparticles
with skin furrows, hair follicles, eccrine ducts, etc. Present research is focusing on
biodegradable nanomaterials that aim to constrain toxicological and environmental
419 [380--431] 10.12.2013 8:16PM
419
11.7 Electrosprayed nanoparticulate drug delivery
70
(3)
60
(2)
50
40
30
(1)
20
10
0
0
10
15
20
25
Time (h)
Figure 11.17
Cumulative dye release versus time for electrospun nanober mats of P(MMA/NIPAM/AAc)
co-polymer subjected to pH modulation in a water bath. See text for details of series 1 to 3. Zhang
and Yarin (2009). Reproduced with permission from The Royal Society of Chemistry.
concerns with increased safety. Gold nanoparticles have potential applicability in tumor/
cancer cell imaging, therapeutics and in cosmetics. l-ascorbic-acid-based cosmetic and
dermatology products have been widely used because it acts as a free radical scavenger
and antioxidant. The chemical instability of ascorbic acid needs a suitable delivery
system for its cosmetic application. As a solution, an electrospun antiwrinkle nanober
face mask containing ascorbic acid, retinoic acid, gold nanoparticles and collagen was
produced by Fathi-Azarbayjani et al. (2010), for potential cosmetic applications. They
fabricated the material so as to enhance the product stability, so that it gradually dissolved
once moistened and released the active ingredients, ensuring maximum skin penetration.
Further, the high surface-area-to-volume ratio of the nanober mask helps to make
maximum contact with the skin surface and enhances a healthy skin appearance.
11.7
Electrosprayed nanoparticulate drug-delivery systems

Electrospraying is a technique akin to electrospinning. The same setup that is used in
electrospinning (Figure 5.1) is essentially used in electrospraying and it is based on the
same physical principles, as described in Sections 5.2 and 5.3 in Chapter 5. The only
difference is in the type of solutions used in electrospraying. Unlike electrospinning, they
420
420 [380--431] 10.12.2013 8:16PM
50
Percentage release rate change
40
30
20
10
(1)
0
10
(3)
20
30
40
(2)
50
0
10
15
20
25
Time (h)
Figure 11.18
Change of the release rate during the pH modulation experiments in Figure 11.17 on
electrospun nanober mats. Zhang and Yarin (2009). Reproduced with permission from The Royal
Nanofibrous drug-delivery system

(Transdermal and wound-healing patches)
Drug and polymer loaded
solution
Baking layer
Nanofibrous drug reservoir
Diffusing drug
Adhesive layer
Taylor cone
Grounded collector
High DC voltage
Figure 11.19
Drug-loaded
nanofibrous patch
Transdermal/wound-healing
nanofibrous patch
Nanobrous drug-delivery system for transdermal delivery and for wound healing. Reprinted from
Madhaiyan et al. (2013), with permission from Elsevier.
do not contain high-molecular-weight exible polymers and are not spinnable. As a

result, solution jets in electrospraying do not stay intact to transform into bers, but rather
rapidly atomize to produce droplets. After solvent evaporation, the solute in these
droplets precipitates and micro- and nanoparticles are formed. The advantage of electrospraying is that even protein-based nanoparticles can be synthesized by high-voltage
electrospraying. Such nanoparticles were found to be stable without any loss of bioactivity
421 [380--431] 10.12.2013 8:16PM
11.7 Electrosprayed nanoparticulate drug delivery
421
of either the drug or the encapsulating biomolecules. Electrostatic spraying was benecial
for bone tissue engineering, such as for coating titanium implants with calcium phosphate
nanoparticles (Siebers et al. 2004) and for developing bioceramic scaffolds of zirconia
(Chen et al. 2006). Electrospray nebulizers for producing microparticles of size range 25
m were designed for the purpose of inhaling medicines through the lungs. These
respirable size range particles are designed to deliver the medicine into the lower airways
without the loss of drug activity in the encapsulated medicine (Tang and Gomez 1994,
Clark 1995, Gomez et al. 1998, Ijsebaert et al. 1999, 2001, Kaerger and Price 2004,
Li et al. 2006b).
11.7.1
Nanoparticulate drug delivery for cancer treatment

Gulfam et al. (2012) electrosprayed nanoparticles of natural gliadin polymer for
controlled release of a cyclophosphamide anticancer drug (Figure 11.20). The particle
sizes of gliadin or gliadingelatin composite nanoparticles were found to increase with
Medium control
Particle control
12 hours
(a)
(b)
24 hours
Annexin V,
Propidium iodide
Drug-loaded particle
Drug control
(c)
Figure 11.20
(d)
Apoptosis of breast cancer cells cultured with 7% gliadin nanoparticles loaded with
cyclophosphamide. The breast cancer cells were cultured with (a) medium alone, (b) 7% gliadin
nanoparticles alone, (c) cyclophosphamide alone and (d) 7% gliadin nanoparticle loaded with
cyclophosphamide. Apoptotic and necrotic cells were stained by annexin V (green) and propidium
iodide (red). Scale bars are 300 m. Reprinted with permission from Gulfam et al. (2012).
422
422 [380--431] 10.12.2013 8:16PM
gelatin concentration, as observed by dynamic light scattering and X-ray diffractometry experiments. Also, the gliadingelatin composite nanoparticles released cyclophosphamide rapidly, whereas gradual release was noted with gliadin nanoparticles
over 48 h. By culturing breast cancer cells with cyclophosphamide-loaded 7% gliadin
nanoparticles for 24 h, down-regulation of Bcl-2 protein was observed with Western
blot analysis, which conrmed the apoptotic effect of the nanoparticle formulation
(Gulfam et al. 2012). Electrosprayed chitosan micro- and nanoparticles were attempted
for drug delivery, with chitosan as a natural biodegradable and biocompatible polymer
(Zhang and Kawakami 2010). Kim and Lee (2011) observed three-dimensional chitosan nanobrous networks by freeze-drying the electrosprayed chitosan nanoparticles.
Doxorubicin-loaded chitosan nanoparticles were electrosprayed with tripolyphosphate
stabilization (Songsurang et al. 2011). The controlled release of the drug from the
biopolymer nanoparticle was up to 70% over a period of 72 h. The governing parameters, such as applied voltage, working distance, needle gauze and ow rate were
manipulated so as to obtain nanoparticles of drug-loaded chitosan-tripolyphosphate.
Titanium dioxide (TiO2) nanoparticles loaded with paclitaxel (TiO2 electrosprayed
coreshell microspheres for paclitaxel delivery) were formed by electrospraying (Jing
et al. 2011). The particle size was observed to be in the 600 nm to 6 m range, with the
titania protecting the initial burst release of the drug. The drug release was triggered by
ultrasonic stimulation of the nanoformulation.
11.7.2
Cardiac applications of nanoparticulate formulations

Li et al. (2012) attempted to produce hydrogen-peroxide-releasing electrosprayed core
shell microparticles, together with cardiosphere-derived cells (CDCs) and thermo-sensitive
hydrogel, from hydroxyethyl methacrylate-oligo hydroxybutyrate (HEMA-oHB) for
cardiac regeneration. The oxygen-releasing scaffold improved survival and assisted in
cardiac differentiation of CDCs. Further the hydrogel system matched the stiffness of
cardiac tissue, supported cardiomyocyte formation and sustained the release of oxygen for
up to two weeks.
11.7.3
Electrosprayed versus electrospun drug-delivery systems

Nanobrous drug-delivery systems were realized as promising scaffolds for in vitro cell
culture, stem-cell regeneration and differentiation. In addition to the pharmaceutical
applications, nanobrous scaffolds still need to be tested in animal study experiments.
Apart from patch applications of nanobrous membrane scaffolds, their applicability in
biomedical implants such as drug-eluting stents is yet to be fully explored. It is very
interesting to note that research in the area of electrospun nanobers has resulted in a
larger number of publications than electrosprayed nanoparticles (Figure 11.21). Even
though both techniques are based on an electrostatic potential difference, the extent of
research work on nanobers is larger. This implies that electrosprayed biomedical nanoparticles, although they have huge potential applications in various domains, still remain
423 [380--431] 10.12.2013 8:16PM
423
11.8 References
2000
Electrospun Nanofiber
1800
Electrosprayed Nanoparticle
1600
Number of Publications
1400
1200
1000
800
600
400
200
0
200
1998
2000
2002
2004
2006
2008
2010
2012
2014
Publication year
Figure 11.21
Comparison of publications on electrosprayed nanoparticles and electrospun nanobers (based on

SciFinder search tag and numbers projected for year 2013).
unnoticed compared to electrospun nanobers. Hence researchers can utilize this relatively less explored nanotechnology tool for their application domain, and more signicantly in drug delivery or healthcare.
11.8
References
Aebischer, P., Salessiotis, A. N., Winn, S. R., 1989. Basic broblast growth factor released from
synthetic guidance channels facilitates peripheral nerve regeneration across long gaps.
J. Neurosci. Res. 23, 282289.
Barnes, C. P., Sell, S. A., Boland, E. D., Simpson, D. G., Bowlin, G. L., 2007. Nanober technology:
Designing the next generation of tissue engineering scaffolds, Adv. Drug. Deliv. Rev. 59,
14131433.
Barnes, C. P., Smith, M. J., Bowlin, G. L., Sell, S. A., Tang, T., Matthews, J. A., Simpson D. G.,
Nimtz, J. C., 2006. Feasibility of electrospinning the globular proteins hemoglobin and myoglobin. J. Eng. Fibers Fabrics 1, 1629.
Bini, T. B., Gao, S., Tan, T. C., 2004. Electrospun poly(L-lactide-co-glycolide) biodegradable
polymer nanober tubes for peripheral nerve regeneration. Nanotechnology 15, 14591464.
Bolgen, N., Vargel, I., Korkusuz, P., Menceloglu, Y. Z., Piskin, E., 2007. In vivo performance of
antibiotic embedded electrospun PCL membranes for prevention of abdominal adhesions.
J. Biomed. Mater. Res. B, Appl. Biomater. 81, 530543.
424
424 [380--431] 10.12.2013 8:16PM
Bueche, F., 1955. Tensile strength of plastics above the glass temperature. J. Appl.Phys. 26, 11331140.
Bueche, F., 1960. Mechanical degradation of high polymers. J. Appl. Polym. Sci. 4, 101106.
Buschle-Diller G., Cooper, J., Xie, Z., Wu, Y., Waldrup, J., Ren, X., 2007. Release of antibiotics
from electrospun bicomponent bers. Cellulose 14, 553562.
Carlisle, R. C., Coulais, C., Namboothiry, M., Carroll, D., Hantgan, R. R., Guthold, M., 2009. The
mechanical properties of individual, electrospun brinogen bers. Biomaterials 30, 12051213.
Carlson, G. A., Dragoo, J. L., Samimi, B., Bruckner, D. A., Benhaim, P., 2004. Bacteriostatic
properties of biomatrices against common orthopaedic pathogens. Biochem. Biophys. Res.
Commun. 321, 472478.
Casper, C. L., Stephens, J. S., Tassi, N. G., Chase, D. B., Rabolt, J. F., 2004. Controlling surface
morphology of electrospun polysterene bers: effect of humidity and molecular weight in the
electrospinning process. Macromolecules 37, 573578.
Chandrasekaran A. R., Venugopal J., Sundarrajan S., Ramakrishna S., 2011. Fabrication of a
nanobrous scaffold with improved bioactivity for culture of human dermal broblasts for
skin regeneration. Biomed. Mater. 6 015001.
Chen, F., Wang, Z. C., Lin, C. J., 2002. Preparation and characterization of nano-sized hydroxyapatite particles and hydroxyapatite/chitosan nanocomposite for use in biomedical materials.
Mater. Lett. 57, 658662.
Chen, H., Hsieh, Y. L., 2004. Ultrane hydrogel bers with dual temperature- and pH-responsive
swelling behaviors. J. Polym. Sci. A-Polym. Chem, 42, 63316339.
Chen, P., Shengwu, Q., Ding, Y. P., Zhu, Z. C., 2011. Preparation of cisplatin composite micro/
nanobers and antitumor activity in vitro against human tumor spc-a-1 cells. NANO: Brief Rep.
Rev. 6, 325332.
Chen, Q. Z., Boccaccini, A. R., Zhang, H. B., Wang, D. Z., Edirisinghe, M. J., 2006. Improved
mechanical reliability of bone tissue engineering (zirconia) scaffolds by electrospraying. J. Am.
Ceram. Soc. 89, 15341539.
Chew, S. Y., Mi, R., Leong, K. W., Hoke, A., 2008. The effect of the alignment of electrospun
brous scaffolds on Schwann cell maturation, Biomaterials 29, 653661.
Chew, S. Y., Wen, J., Yim, E., Leong, K., 2005. Sustained release of proteins from electrospun
biodegradable bers. Biomarcromolecules 6, 20172024.
Clark, A. R., 1995. Medical aerosol inhalers: past, present, and future. Aerosol Sci. Technol.
22 374391.
Clarke, K. I., Graves, S. E., Wong, A. T. C., Trift, J. T., Francis, M. J. O., Czernuszka, J. T., 1993.
Investigation into the formation and mechanical properties of a bioactive material based on
collagen and calcium phosphate. J. Mater. Sci. Mater. Med. 4, 107110.
Cordeiro, P. G., 1989. Acidic broblast growth factor enhances peripheral nerve regeneration
in vivo. Plast. Reconstr. Surg. 83, 10131019.
Doi, M., Edwards, S. F., 1986. The Theory of Polymer Dynamics. Clarendon Press, Oxford.
Dror, Y., Salalha, W., Avrahami R., Zussman E., Yarin A. L., Dersch, R., Greiner, A., Wendorff, J. H.,
2007. One-step production of polymeric micro-tubes via co-electrospinning. Small 3, 10641073.
Dror, Y, Ziv, T., Makarov, V., Wolf, H., Admon, A., Zussman E., 2008. Nanobers made of
globular proteins. Biomacromolecules 9, 27492754.
Engelberg, I., Kohn J., 1991. Physico-mechanical properties of degradable polymers used in
medical applications: A comparative study. Biomaterials 12, 292304.
Fathi-Azarbayjani, A., Qun, L., Chan, Y. W., Chan, S. Y., 2010. Novel vitamin and gold-loaded
nanober facial mask for topical delivery. AAPS PharmSciTech. 11, 11641170.
425 [380--431] 10.12.2013 8:16PM
11.8 References
425
Filippov, S., Hruby, M., Konak, C., Mackova, H., Spirkova, M., Stepanek, P., 2008. Novel
pH-responsive nanoparticles. Langmuir 24, 92959301.
Gandhi, M., Srikar, R., Yarin, A. L., Megaridis, C. M., Gemeinhart, R. A., 2009. Mechanistic
examination of protein release from polymer nanobers. Molec. Pharm. 6, 641647.
Ghasemi, L. M., Prabhakaran, M. P., Morshed, M., Nasr-Esfahani, M. H., Ramakrishna, S., 2008.
Electrospun poly(-caprolactone)/gelatin nanobrous scaffolds for nerve tissue engineering.
Gomez, A., Bingham, D., de Juan, L., Tang, K., 1998. Production of protein nanoparticles by
electrospray drying. J. Aerosol Sci. 29, 561574.
Guiochon, G., Felinger, A., Shirazi, D. G., Katti, A. M., 2006. Fundamentals of Preparative and
Nonlinear Chromatography. Elsevier, Amsterdam.
Gulfam, M., Kim, J., Lee, J. M., Ku, B., Chung, B. H., Chung, B. G., 2012. Anticancer drug-loaded
gliadin nanoparticles induce apoptosis in breast cancer cells. Langmuir 28, 82168223.
Gupta, D., Venugopal, J., Mitra, S., Giri Dev, V. R., Ramakrishna, S., 2009a. Nanostructured
biocomposite substrates by electrospinning and electrospraying for the mineralization of osteoblasts. Biomaterials 30, 20852094.
Gupta, D., Venugopal, J., Prabhakaran, M. P., Giri Dev V. R., Low, S., Choon, A. T.,
Ramakrishna, S., 2009b. Aligned and random nanobrous substrate for the in vitro culture of
Schwann cells for neural tissue engineering. Acta Biomater. 5, 25602569.
He, W., Yong, T., Ma, Z., Teo, W. E., Ramakrishna, S., 2005. Fabrication and endothelialization of
collagen-blended biodegradable polymer nanobers: potential vascular graft for blood vessel
tissue engineering. Tissue Eng. 11, 15741588.
He, C. L., Huang, Z. M., Han, X. J., Liu, L., Zhang, H. S., Chen, L. S., 2006a. Coaxial electrospun
poly(L-lactic acid) ultrane bers for sustained drug delivery. J. Macromol. Sci., Part B: Physics
45, 515524.
He W., Yong T., Teo, W. E., Ramakrishna, S., 2006b. Biodegradable polymer nanober mesh to
maintain functions of endothelial cells. Tissue Eng. 12, 24572466.
Heydarkhan-Hagvall, S., Schenke-Layland, K., Dhanasopon, A. P., Rofail, F., Smith, H.,
Wu, B. M., Shemin, R., Beygui, R. E., MacLellan, W. R., 2008. Three-dimensional electrospun
ECM-based hybrid scaffolds for cardiovascular tissue engineering. Biomaterials 29, 2907.
Hu, Q., Li, B., Wang, M., Shen, J., 2004. Preparation and characterization of biodegradable
chitosan/hydroxyapatite nanocomposite rods via in situ hybridization: a potential material as
internal xation of bone fracture. Biomaterials 25, 779785.
Huang, Z. M., He, C. L., Yang, A., Han, X. J., Yin, J., Wu, Q., 2006. Encapsulating drugs in
biodegradable ultrane bers through co-axial electrospinning. J. Biomed. Mater. Res. Part A
77A, 169179.
Ijsebaert, J. C., Geerse, K. B., Marijnissen, J. C. M., Lammers, J. W. J., Zanen, P., 2001. Electrohydrodynamic atomization of drug solutions for inhalation purposes. J. Appl. Physiol.
91 27352741.
Ijsebaert, J. C., Geerse, K. B., Marijnissen, J. C. M., Scarlett, B., 1999. Electrohydrodynamic
spraying of inhalation medicine. J. Aerosol Sci. 30 (Suppl. 1), 825826.
Jiang, H., Hu, Y., Li, Y., Zhao, P., Zhu, K., Chen, W., 2005. A facile technique to prepare
biodegradable coaxial electrospun nanobers for controlled release of bioactive agents.
J. Controlled Release 108, 237243.
Jin, G., Prabhakaran, M. P., Ramakrishna, S., 2011. Stem cell differentiation to epidermal
lineages on electrospun nanobrous substrates for skin tissue engineering. Acta Biomater.
7, 31133122.
426
426 [380--431] 10.12.2013 8:16PM
Jin, H. J., Fridrikh, S. V., Rutledge, G. C., Kaplan, D. L., 2002. Electrospinning Bombyx mori silk
with poly(ethylene oxide). Biomacromolecules 3, 12331239.
Jing, Y., Zhu, Y., Yang, X., Shen, J., Li, C., 2011. Ultrasound-triggered smart drug release from
multifunctional coreshell capsules one-step fabricated by coaxial electrospray method.
Langmuir 27, 11751180.
Kaerger, J. S., Price, R., 2004. Processing of spherical crystalline particles via a novel solution
atomization and crystallization by sonication (SAXS) technique. Pharmaceut. Res. 21 372381.
Katti, D. S., Robinson, K. W., Ko, F. K., Laurencin, C. T., 2004. Bioresorbable nanober-based
systems for wound healing and drug delivery: optimization of fabrication parameters. J. Biomed.
Mater. Res. B, Appl. Biomater. 70, 286296.
Kenawy, E., Bowlin, G., Manseld, K., Layman, J., Simpson, G., Sanders, E., Wnek, G., 2002.
Release of tetracycline hydrochloride from electrospun poly(ethylene-co-vinylacetate), poly
(lactic acid), and a blend. J. Controlled Release 81, 5764.
Kikuchi, M., Itoh, S., Ichinose, S., Shinomiya, K., Tanaka, J., 2001. Self-organization mechanism
in a bone-like hydroxyapatite/collagen composite synthesized in vitro and its biological reaction
in vivo. Biomaterials 22, 17051711.
Kim, J., Song, H., Park, I., Carlisle, C., Bonin, K., Guthold, M., 2011. Denaturing of single
electrospun brinogen bers studied by deep ultra-violet uorescence microscopy. Microsc.
Res. Techn. 74, 219224.
Kim, K., Luu, Y. K., Chang, C., Fang, D, Hsiao, B. S., Chu, B., Hadjiargyrou, M., 2004.
Incorporation and controlled release of a hydrophilic antibiotic using poly(lactide-co-glycolide)based electrospun nanobrous scaffolds. J. Controlled Release 98, 4756.
Kim, M. H., Kim, J. C., Lee, H. Y., Kim, J. D., Yang, J. H., 2005. Release property of temperaturesensitive alginate beads containing poly(N-isopropylacrylamide). Coll. Surfaces B 46, 5761.
Kim, M. Y., Lee, J., 2011. Chitosan brous three-dimensional networks prepared by freeze drying.
Carb. Polym. 84, 13291336.
Kim, S. H., Nam, Y. S., Lee, T. S., Park, W. H., 2003. Silk broin nanober. Electrospinning,
properties and structure. Polymer J. 35, 185190.
Koh, H. S., Thomas, Y., Chan, C. K., Ramakrishna, S., 2008. Enhancement of neurite outgrowth
using nano-structured scaffolds coupled with laminin. Biomaterials 29, 35743582.
Kontogiannopoulos, K. N., Assimopoulou, A. N., Tsivintzelis, I., Panayiotou, C., Papageorgiou, V. P.,
2011. Electrospun ber mats containing shikonin and derivatives with potential biomedical
applications. Int. J. Pharm. 409, 216228.
Landis, W. J., Song, M. J., Leith, A., McEwen, L., McEwen, B. F., 1993. Mineral and organic
matrix in normally calcifying tendon visualized in three dimensions by high voltage electron
microscopic tomography and graphic image reconstruction. J. Struct. Biol. 110, 3954.
Lee, S. J., Yoo, J. J., Lim, G. J., Atala, A., Stitzel, J., 2007. In vitro evaluation of electrospun
nanober scaffolds for vascular graft application. J. Biomed. Mater. Res. A, 83, 9991008.
Levich, V. G., 1962. Physicochemical Hydrodynamics. Prentice Hall, Englewood Cliffs.
Li, C., Vepari, C., Jin, H. J., Kim, H. J., Kaplan, D. L., 2006a. Electrospun silk-BMP-2 scaffolds for
bone tissue engineering. Biomaterials 27, 31153124.
Li, T., Kildsig, D. O., Park, K., 1997. Computer simulation of molecular diffusion in amorphous
polymers. J. Controlled Release 48, 5766.
Li, S. W., Jayasinghe, S. N., Edirisinghe, M. J., 2006b. Aspirin particle formation by electric-eldassisted release of droplets. Chem. Eng. Sci. 61 30913097.
Li, W.-J., Tuli, R., Huang, X., Laquerriere, P. Tuan, R. S., 2005a. Multilineage differentiation of human
mesenchymal stem cells in a three-dimensional nanobrous scaffold. Biomaterials 26, 51585166.
427 [380--431] 10.12.2013 8:16PM
11.8 References
427
Li, W.-J., Tuli, R., Okafor, C., Derfoul, A., Danielson, K. G., Hall, D. J., Tuan, R. S., 2005b. A
three-dimensional nanobrous scaffold for cartilage tissue engineering using human mesenchymal stem cells. Biomaterials 26, 599609.
Li, Z., Guo, X., Guan, J., 2012. An oxygen release system to augment cardiac progenitor cell
survival and differentiation under hypoxic condition. Biomaterials 33, 59145923.
Liao, I. C., Chew, S. Y., Leong, K. W., 2006. Alighned coreshell nanobers delivering bioactive
proteins. Nanomedicine 1, 465471.
Liao, S. S., Cui, F. Z., Feng, Q. L., 2004. Hierarchically biomimetic bone scaffold materials: NanoHA/Collagen/PLA composite. J. Biomed. Mater. Res. B 69, 158165.
Liu, X., Smith, L. A, Hu, J., Ma, P. X., 2009. Biomimetic nanobrous gelatin/apatite composite
scaffolds for bone tissue engineering. Biomaterials 30, 22522258.
Luong-Van, E., Grondahl, L., Ngiap Chua, K., Leong, K., Nurcombe, V., Cool, S., 2006. Controlled
release of heparin from poly(epsilon-caprolactone) electrospun bers. Biomaterials 27, 20422050.
Ma, G., Liu, Y., Peng, C., Fang, D., He, B., Nie, J., 2011. Paclitaxel loaded electrospun porous
nanobers as mat potential application for chemotherapy against prostate cancer. Carb. Polym.
86, 505512.
Madhaiyan, K., Sridhar, R., Sundarrajan, S., Venugopal, J. R., Ramakrishna, S., 2013. Vitamin B12
loaded polycaprolactone nanobers: a novel transdermal route for the water soluble energy
supplement delivery. Int. J. Pharm. 444, 7076.
Mannello, F., Tonti, G. A., 2007. Concise review: No breakthroughs for human mesenchymal and
embryonic stem cell culture: Conditioned medium, Feeder layer, or Feeder-free; Medium with
fetal calf serum, human serum, or Enriched plasma; Serum-free, Serum replacement nonconditioned medium, or Ad Hoc formula? All that glitters is not gold! Stem Cells 25, 16031609.
Matthews, J. A., Wnek, G. E., Simpson D. G., Bowlin, G. L., 2002. Electrospinning of collagen
nanobers. Biomacromolecules 3, 232238.
McManus, M., Boland, E. D., Koo, H. P., Barnes, C. P., Pawlowski, K. J., Wnek, G. E.,
Simpson, D. G., Bowlin, G. L., 2006. Mechanical properties of electrospun brinogen structures.
Acta Biomater. 2, 1928.
McManus, M., Boland, E., Sell, S., Bowen, W., Koo, H., Simpson, D., Bowlin, G. L., 2007a.
Electrospun nanobre brinogen for urinary tract tissue reconstruction. Biomed. Mater. 2, 257262.
McManus, M. C., Boland, E. D., Simpson, D. G., Barnes, C. P., Bowlin, G. L., 2007b. Electrospun
brinogen: feasibility as a tissue engineering scaffold in a rat cell culture model. J. Biomed.
Mater. Res. A 81, 299309.
Miller, C., Shanks, H., Witt, A., Rutkowski, G., Mallapragada, S., 2001. Oriented Schwann cell
growth on micropatterened biodegradable polymer substrates. Biomaterials, 22, 12631269.
Miyajima, M., Koshika, A., Okada, J., Ikeda, M., 1999. Mechanism of drug release from poly(Llactic acid) matrix containing acidic or neutral drugs. J. Controlled Release 60, 199209.
Miyajima, M., Koshika, A., Okada, J., Ikeda, M., Nishimura, K., 1997. Effect of polymer crystallinity
on papaverine release from poly(L-lactic acid) matrix. J. Controlled Release 49, 207215.
Miyajima, M., Koshika, A., Okada, J., Kusai, A., Ikeda, M., 1998. Factors inuencing the diffusioncontrolled release of papaverine from poly(L-lactic acid) matrix J. Controlled Release 56, 8594.
Mo, X. M., Xu, C. Y., Kotaki, M., Ramakrishna, S., 2004. Electrospun P(LLA-CL) nanober: a
biomimetic extracellular matrix for smooth muscle cell and endothelial cell proliferation.
Moroni, L., Licht, R., de Boer, J., de Wijn, J. R., van Blitterswijk, C. A., 2006. Fiber diameter and
texture of electrospun PEOT/PBT scaffolds inuence human mesenchymal stem cell proliferation and morphology and the release of incorporated compounds. Biomaterials 27, 49114922.
428
428 [380--431] 10.12.2013 8:16PM
Muzzarelli, C., Muzzarelli, R. A. A., 2002. Natural and articial chitosan-inorganic composites.
J. Inorg. Biochem. 92, 8994.
Noh, H. K., Lee, S. W., Kim, J. M., Oh, J. E., Kim, K. H., Chung, C. P., Choi, S. C., Park, W. H.,
Min, B. M., 2006. Electrospinning of chitin nanobers: degradation behavior and cellular
response to normal human keratinocytes and broblasts. Biomaterials 27, 39343944.
OBrien, F. J., Harley, B. A., Yannas, I. V., Gibson, L., 2004. Inuence of freezing rate on pore
structure in freeze dried collagen GAG scaffolds. Biomaterials 25, 10771086.
Ohgo K., Zhao C., Kobayashi M., Asakura T., 2003. Preparation of non-woven nanobers of
Bombyx mori silk, Samia cynthia ricini silk and recombinant hybrid silk with electrospinning
method. Polymer 44, 841846.
Porter, A., Patel, N., Brooks, R., Boneld, W., 2005. Effect of carbonate substitution on the ultrastructural characteristics of hydroxyapatite implants. J. Mater. Sci. Mater. Med. 16, 899907.
Powell, M. P., Sobarzo, M. R., Saltzman, W. M., 1990. Controlled release of nerve growth factor
from a polymeric implant. Brain Res. 515, 309311.
Prabhakaran, M. P., Venugopal, J., Casey C., Ramakrishna S., 2008a. Surface modied electrospun
nanobrous scaffolds for nerve tissue engineering. Nanotechnology 19, 455102.
Prabhakaran, M. P., Venugopal, J., Chyan, T. T., Hai, L. B., Chan, C. K., Tang, A. L.,
Ramakrishna, S., 2008b. Electrospun biocomposite nanobrous scaffolds for neural tissue
engineering. Tissue Eng. A 14, 17871797.
Prabhakaran, M. P., Venugopal, J., Ramakrishna, S., 2009. Electrospun nanostructured scaffolds
for bone tissue engineering. Acta Biomater. 5, 28842893.
Qiu, Y., Park, K., 2001. Environment-sensitive hydrogels for drug delivery. Adv. Drug Deliv. Rev.
53, 321339.
Ritger, P. L., Peppas, N. A., 1987. A simple equation for description of solute-release. I. Fickian and
non-Fickian release from non-swellable devices in the form of slabs, spheres, cylinders or discs.
J. Controlled Release 5, 2336.
Rusu, V. M., Ng, C. H., Wilke, M., Tiersch, B., Fratzl, P., Peter, M. G., 2005. Size-controlled hydroxyapatite nanoparticles as self-organized organic-inorganic composite materials. Biomaterials 26,
54145426.
Sachlos, E., Gotora D., Czernuszka, J. T., 2006. Collagen scaffolds reinforced with biomimetic
composite nano-sized carbonate-substituted hydroxyapatite crystals and shaped by rapid prototyping to contain internal microchannels. Tissue Eng. 12, 24792487.
Saltzman, W. M., Langer, R., 1989. Transport rates of proteins in porous materials with known
microgeometry. Biophys. J. 55, 163171.
Schnell, E., Kinkhammer, K., Balzer, S., Brook, G., Mey, J., 2007. Guidance of glial cell migration
and axonal growth on electrospun nanobers of poly--caprolactone and a collagen/
poly--caprolactone blend. Biomaterials 28, 30123025.
Shields, K. J., Beckman, M. J., Bowlin, G. L., Wayne, J. S., 2004. Mechanical properties and
cellular proliferation of electrospun collagen type II. Tissue Eng. 10, 15101517.
Shin, M., Yoshimoto, H., Vacanti, J. P., 2004. In vivo bone tissue engineering using mesenchymal
stem cells on a novel electrospun nanobrous scaffold. Tissue Eng. 10, 3341.
Siebers, M. C., Walboomers, X. F., Leeuwenburgh, S. C. G., Wolke, J. G. C., Jansen, J. A., 2004.
Electrostatic spray deposition (ESD) of calcium phosphate coatings, an in vitro study with
osteoblast-like cells. Biomaterials 25, 20192027.
429 [380--431] 10.12.2013 8:16PM
11.8 References
429
Sinha-Ray, S., Zhang, Y., Placke, D., Megaridis, C. M., Yarin, A. L., 2010. Resins with nanoraisins. Langmuir 26 1024310249.
Song, M., Wang, X., Wang, C., Pan, C., Fu, D., Gu, Z., 2009. Application of the blending of
PNIPAM-co-PS nanobers with functionalized Au nanoparticles for the high-sensitive diagnosis
of cancer cells. J. Nanosci. Nanotechnol. 9, 876879.
Song, F., Wang, X. L., Wang, Y. Z., 2011. Poly (N-isopropylacrylamide)/poly (ethylene oxide)
blend nanobrous scaffolds: Thermo-responsive carrier for controlled drug release. Coll. Surf.
B: Biointerfaces 88, 74954.
Songsurang, K., Praphairaksit, N., Siraleartmukul, K., Muangsin, N., 2011. Electrospray fabrication of Doxorubicin-Chitosan-Tripolyphosphate nanoparticles for delivery of Doxorubicin.
Arch. Pharm. Res. 34, 583592.
Srikar, R., Yarin, A. L., Megaridis, C. M., Bazilevsky, A. V., Kelley, E., 2008. Desorption-limited
mechanism of release from polymer nanobers. Langmuir 24, 965974.
Taepaiboon, P., Rungsardthong, U., Supaphol, P., 2007. Vitamin-loaded electrospun cellulose
acetate nanober mats as transdermal and dermal therapeutic agents of vitamin A acid and
vitamin E. Eur. J. Pharm. Biopharm. 67, 387397.
Tang, K., Gomez, A., 1994. Generation by electrospray of monodisperse water droplets for targeted
drug delivery by inhalation. J. Aerosol Sci. 25, 12371249.
Termonia, Y., Meakin, P., Smith, P., 1985. Theoretical study of the inuence of the molecular
weight on the maximum tensile strength of polymer bers. Macromolecules 18, 22462252.
Theron, A., Zussman, E., Yarin, A. L., 2001. Electrostatic eld-assisted alignment of electrospun
Thomas, V., Jose, M. V., Chowdhury, S., Sullivan, J. F., Dean, D. R., Vohra, Y. K., 2006. Mechanomorphological studies of aligned nanobrous scaffolds of polycaprolactone fabricated by
electrospinning. J Biomater. Sci. Polym. Ed. 17, 969984.
Tikhonov, A. N., Samarskii, A. A., 1990. Equations of Mathematical Physics. Dover, New York.
Uslu, I., Keskin, S., Gul, A., Karabulut, T. C., Aksu, M. L., 2010. Preparation and properties of
electrospun poly(vinyl alcohol) blended hybrid polymer with aloe vera and HPMC as wound
dressing. Hacettepe J. Biol. and Chem. 38, 1925.
Venugopal, J., Low, S., Choon, A. T., Bharath Kumar, A., Ramakrishna, S., 2008a.
Nanobioengineered electrospun composite nanobers and osteoblasts for bone regeneration.
Artif. Organs 32, 388397.
Venugopal, J., Low, S., Choon, A. T., Kumar, A. B., Ramakrishna, S., 2008b. Electrospunmodied nanobrous scaffolds for the mineralization of osteoblast cells. J. Biomed. Mater.
Res. 85A, 408417.
Venugopal, J., Low, S., Choon, A. T., Sampath Kumar, T. S., Ramakrishna. S., 2008c.
Mineralization of osteoblasts with electrospun collagen/hydroxyapatite nanobers. J. Mater.
Sci. Mater. Med. 19, 20392046.
Venugopal, J., Vadagama, P., Sampath Kumar, T. S., Ramakrishna, S., 2007. Biocomposite nanobers and osteoblasts for bone tissue engineering. Nanotechnology 18, 055101.
Verreck, G., Chun, I., Rosenblatt, J., Peeters, J., Dijck, A., Mensch, J., Noppe, M., Brewster, M.,
2003a. Incorporation of drugs in an amorphous state into electrospun nanober composed of
water-insoluble, nonbiodegradable polymer. J. Controlled Release 92, 349360.
Verreck, G., Sun, I., Peeters, J., Rosenblatt, J., Brewster, M., 2003b. Preparation and characterization of nanobers containing amorphous drug dispersions generated by electrostatic spinning.
Pharm. Res. 20, 810817.
430
430 [380--431] 10.12.2013 8:16PM
Wada, R., Hyon, S. H., Ikada, Y., 1995. Kinetics of diffusion-mediated drug release enhanced by
matrix degradation. J. Controlled Release 37, 151160.
Wahl, D., Czernuszka, J. T., 2006. Collagen-hydroxyapatite composites for hard tissue repair. Euro.
Cells Mater. 11, 4356.
Wahl D. A., Sachlos, E., Liu, C., Czernuszka, J. T., 2007. Controlling the processing of collagenhydroxyapatite scaffolds for bone tissue engineering. J. Mater. Sci. Mater. Med. 18, 201209.
Washburn, E. W., 1921. The dynamics of capillary ow. Phys. Rev. 17, 273283.
Wei, J., Li, Y. B., Chen, W. Q., Zuo, Y., 2003. A study on nanocomposite of hydroxyapatite and
polyamide. J. Mater Sci. 38, 33033306.
Weiner, S., Traub, W., 1989. Crystal size and organization in bone. Connect. Tissue Res. 21, 589595.
Whitworth, I. H., Brown, R. A., Dore, C., Green, C. J., Terenghi, G., 1995. Orientated mats of
bronectin as a conduit material for use in peripheral nerve repair. J. Hand Surg. J. British Soc.
Surg. Hand 20, 429436.
Wise, J. K., Cho, M., Zussman, E., Megaridis, C. M., Yarin, A. L., 2008. Electrospinning techniques to control deposition and structural alignment of nanobrous scaffolds for cellular orientation and cytosceletal reorganization in Nanotechnology and Tissue Engineering, pp. 243260.
(Eds. C. T. Laurencin and L. S. Nair), CRC Press, Taylor & Francis, Boca Raton, New York.
Wise, J. K., Yarin, A. L., Megaridis, C. M., Cho, M., 2009. Chondrogenic differentiation of human
mesenchymal stem cells on oriented nanobrous scaffolds: Engineering the supercial zone of
atricular cartilage. Tissue Eng. 15, 913921.
Wnek, G. E., Carr, M. E., Simpson, D. G., Bowlin, G. L., 2003. Electrospinning of nanober
brinogen structures. Nano Lett. 3, 213216.
Xie, J., Wang, C. H., 2006. Electrospun micro- and nanobers for sustained delivery of paclitaxel to
treat C6 glioma in vitro. Pharm. Res. 23, 18171826.
Xu, C. Y., Inai, R., Kotaki, M., Ramakrishna, S., 2004a. Aligned biodegradable nanobrous
structure: a potential scaffold for blood vessel engineering. Biomaterials 25, 877886.
Xu, C. Y., Inai, R., Kotaki, M., Ramakrishna, S., 2004b. Electrospun nanobers fabrication as
synthetic extracellular matrix and its potential for vascular tissue engineering. Tissue Eng.
10, 11601168.
Xu, X., Chen, X., Wang, Z., Jing, X., 2009. Ultrane PEGPLA bers loaded with both
paclitaxel and doxorubicin hydrochloride and their in vitro cytotoxicity. Eur. J. Pharm.
Biopharm. 72, 1825.
Yamaguchi, I., Tokuchi, K., Fukuzaki, H., Koyama, Y., Takakuda, K., Monma, H., 2001. Preparation
and microstructure analysis of chitosan/hydroxyapatite nanocomposites. J. Biomed. Mater. Res.
55, 2027.
Yang, D. Z., Jin, Y., Ma, G. P., Chen, X. M., Lu, F. M., Nie, J., 2008. Fabrication and characterization of chitosan/PVA with hydroxyapatite biocomposite nanoscaffolds. J. Appl. Polym. Sci.
110, 33283335.
Yang, F., Murugan, R., Wang, S., Ramakrishna, S., 2005. Electrospinning of nano/micro scale poly
(L-lactic acid) aligned bers and their potential in neural tissue engineering. Biomaterials
26, 26032610.
Yarin, A. L., 1991. Strong ows of polymeric liquids: 2. Mechanical degradation of macromolecules. J. Non-Newton. Fluid Mech. 38, 127136.
Yarin, A. L., 2008. Stimuli-responsive polymers in nanotechnology: Deposition and possible effect
on drug release. Math. Model. Nat. Phenom. 3, No. 5, 115.
431 [380--431] 10.12.2013 8:16PM
11.8 References
431
Yarin, A. L., Lastochkin, D., Talmon, Y., Tadmor, Z., 1999. Bubble nucleation during devolatilization of polymer melts, AIChE Journal 45, 25902605.
Yoshimoto, H., Shin, Y. M., Terai, H., Vacanti, J. P., 2003. A biodegradable nanober scaffold by
electrospinning and its potential for bone tissue engineering. Biomaterials 24, 20772082.
Zarkoob, S., Eby, R. K, Reneker, D. H., Hudson, S. D., Ertley, D., Adams, W. W., 2004. Structure
and morphology of electrospun silk nanobers. Polymer 45, 39733977.
Zhang, S., Huang, Y., Yang, X., Mei, F., Ma, Q., Chen, G., Ryu, S., Deng, X., 2009. Gelatin
nanobrous membrane fabricated by electrospinning of aqueous gelatin solution for guided
tissue regeneration. J. Biomed. Mater. Res. A. 90, 671679.
Zhang, S., Kawakami, K., 2010. One-step preparation of chitosan solid nanoparticles by electrospray deposition. Int. J. Pharm. 397, 211217.
Zhang, Y., Ouyang, H., Lim, C. T., Ramakrishna, S., Huang, Z. M., 2005. Electrospinning of
gelatin bers and gelatin/PCL composite brous scaffolds. J. Biomed. Mater. Res. B. Appl.
Biomater. 72, 156165.
Zhang, Y., Sinha-Ray, S., Yarin, A. L., 2011. Mechanoresponsive polymer nanoparticles, nanobers and coatings as drug carriers and components of microuidic devices. J. Mater. Chem.
21, 82698281.
Zhang, Y., Yarin, A. L., 2009. Stimuli-responsive copolymers of N-isopropyl acrylamide with
enhanced longevity in water for micro- and nanouidics, drug delivery and non-woven applications. J. Mater. Chem. 19, 47324739.
Zhang, Y. Z., Ouyang, H. W., Lim, C. T., Ramakrishna, S., Huang, Z. M., 2005. Electrospinning of
gelatin bers and gelatin/PCL composite brous scaffolds. J. Biomed. Mater. Res. B Appl.
Biomater. 72, 156165.
Zhang, Y. Z., Su, B., Ramakrishna, S., Lim, C. T., 2008a. Chitosan nanobers from an easily
electrospinnable UHMWPEO-doped chitosan solution system. Biomacromolecules 9, 136141.
Zhang, Y. Z., Venugopal, J. R., El-Turki, A., Ramakrishna, S., Su, B., Lim, C. T., 2008b.
Electrospun biomimetic nanocomposite nanobers of hydroxyapatite/ chitosan for bone tissue
engineering. Biomaterials 29, 43144322.
Zhang, Y. Z., Wang, X., Feng, Y., Li, J., Lim, C. T., Ramakrishna, S., 2006. Coaxial electrospinning
of (uorescein isothiocyanate-conjugated bovine serum albumin)-encapsulated poly(epsiloncaprolactone) nanobers for sustained release. Biomacromolecules 7, 10491057.
Zhurkov, S. N., Korsukov, V. E., 1974. Atomic mechanism of fracture of solid polymers. J. Polym.
Sci., Polym. Phys. Ed. 12, 385398.
Zeng, J., Xu, X., Chen, X., Leng, Q., Bian, X., Yang, L., Jing, X., 2003. Biodegradable electrospun
bers for drug delivery. J. Controlled Release 92, 227231.
Zong, X., Kim, K., Fang, D., Ran, S., Hsiao, B. S., Chu, B., 2002. Structure and process relationship of electrospun bioadsorbable nanober membranes. Polymer 43, 44034412.
Zussman, E., Burman, M., Yarin, A. L., Khaln, R., Cohen, Y., 2006a. Tensile deformation of
electrospun Nylon 6,6 nanobers. J. Polym. Sci., Part B- Polymer Physics 44, 14821489.
Zussman, E., Yarin, A. L., Bazilevsky, A. V., Avrahami, R., Feldman, M., 2006b. Electrospun
Polyacrylonitrile/Poly(methyl methacrylate)-derived carbon micro-/nanotubes. Adv. Mater.
18, 348353.

Fundamentos y Aplicaciones de Micro y Nanofibras

Caricato da

Informazioni sul documento

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

Fundamentos y Aplicaciones de Micro y Nanofibras

Caricato da

Copyright:

Formati disponibili

Uploaded by:

Ebooks Chemical Engineering

For More Books, softwares & tutorials Related to Chemical Engineering

i [1--2] 10.12.2013 5:29PM

Fundamentals and Applications of Micro- and Nanobers

ii [1--2] 10.12.2013 5:29PM

iii [3--3] 10.12.2013 5:33PM

Fundamentals and Applications

[4--4] 10.12.2013 5:36PM

University Printing House, Cambridge CB2 8BS, United Kingdom

v [5--8] 11.12.2013 3:13PM

Polymer physics and rheology

Polymer structure, macromolecular chains, Kuhn segment,

General quasi-one-dimensional equations of dynamics of free liquid jets,

History and outlook

Mass, momentum and moment-of-momentum balance equations

Melt- and solution blowing

Bending and apping of exible solid threadlines in a gas jet

Electrospinning of micro- and nanobers

[5--8] 11.12.2013 3:13PM

Electrospinning of polymer solutions

Additional methods and materials used to form micro- and nanobers

Tensile properties of micro- and nanobers

Tensile tests on individual nanobers

[5--8] 11.12.2013 3:13PM

Carbonization, sol-gel transformation, calcinations and metallization

Applications of micro- and nanobers

Micromechanical model of stressstrain dependence of nanober mats

Filters and membranes

Military applications of micro- and nanobers

Nanobers and chemical decontamination

Applications of micro- and nanobers, and micro- and nanoparticles:

Nanobrous scaffolds for tissue regeneration

[5--8] 11.12.2013 3:13PM

ix [9--12] 10.12.2013 9:55PM

x [9--12] 10.12.2013 9:55PM

xi [9--12] 10.12.2013 9:55PM

xii [9--12] 10.12.2013 9:55PM

1 [1--24] 10.12.2013 3:58PM

History and outlook

2 [1--24] 10.12.2013 3:58PM

3 [1--24] 10.12.2013 3:58PM

1.1 History and outlook

4 [1--24] 10.12.2013 3:58PM

5 [1--24] 10.12.2013 3:58PM

1.2 Melt spinning

Melt Spinning Polymer from Chip

6 [1--24] 10.12.2013 3:58PM

7 [1--24] 10.12.2013 3:58PM

1.2 Melt spinning

acting in cross-section x of the innitesimal threadline slice under consideration would

which means that

For Newtonian uids,

and thus using Eq. (1.8) we arrive at

8 [1--24] 10.12.2013 3:58PM

where the factor 3 is called the Trouton viscosity (Yarin 1993).

9 [1--24] 10.12.2013 3:58PM

1.2 Melt spinning

with the corresponding equation relating to .

Solutions of Eq. (1.23) are sought on the complex plane.

10 [1--24] 10.12.2013 3:58PM

where 0 and are the dimensionless perturbation amplitude and frequency,

11 [1--24] 10.12.2013 3:58PM

1.2 Melt spinning