Functionalization of Carbohydrates

on the Macro-, Nano-, and Molecular Scale

by

Wei Chung Chen

A thesis submitted to McGill University in partial
fulfillment of the requirements for the degree of

Doctor of Philosophy
Department of Chemistry
McGill University
Montreal, Quebec
Canada

Wei Chung Chen, 2015

September 2015

Abstract
The main objective of the present thesis is to modify and to study carbohydrates in order
to make use of their surface properties in bulk, on the nano and molecular scale. To
achieve this goal, several strategies were explored. In the case of bulk cellulose, the
silylation of wood pulp by gas-phase reaction with trichloromethylsilane (TCMS)
produced a material that we call Cellufoam. In this silylation reaction, the hydroxyl (OH) groups of cellulose are silylated, which suppresses celluloses affinity for water.
With a water drop contact angle of 150, the surface can be described as
superhydrophobic. By forcing water into Cellufoams fibre lumen and pressing the pulp
into a handsheet, the material was shown to expand upon drying. This is the result of the
interplay between capillary force and the force of elastic rebound. After each cycle of use
and rewetting, this dry-expansion property became less pronounced. The reduction in
performance is explained by the exposure of hydroxyl (-OH) groups caused by
mechanical damage after each cycle. On the nanoscale, cellulose nanocrystals (CNC)
were modified to stabilize heterogeneous metal-in-water mixtures. Silver nanoparticles
(AgNP) were both synthesized and stabilized by the use of cationic cellulose nanocrystals
(cCNC). It was shown that chloride and silver ions in solution can only yield silver
chloride in the presence of CNC. However, by using cCNC, where each particle is
smaller and exposes more surface area, AgNP was produced due to the larger presence of
OH groups. In this work, it was determined that reducing ends (C=O) and OH groups
are both capable of forming AgNP. Therefore, it can be proposed that reducing sugars are
not the only types of sugars with the ability to reduce metal ions. As a continuation of the
AgNP study, sucrose, a non-reducing sugar, was explored as a reducing agent and
stabilizer for the growth of AgNP. A stable colloidal suspension was formed without
heating on the order of hours. By high-resolution TEM, a core-shell structure was
observed in which the core is single crystal silver and the shell is composed of sucrose
based on elemental analysis. In comparison, arabinose and galactose also formed silver
colloidal suspensions, but these suspensions were less stable because their shells were
much thinner. The greater stability of the sucrose AgNP suspension also causes the
formation of a silver mirror on the surface of the glass vial. Under the same condition,
silver colloidal suspensions stabilized by reducing sugars succumbed to sedimentation. In
this thesis, it is shown that carbohydrates in their various forms possess very different
properties depending on their numerous functional groups. By silylation, cationization,
and the presence of aldehyde reducing ends, these materials can be made
superhydrophobic, stabilize metal nanoparticles, and form silver core-shell structures.

ii

Rsum
L'objectif principal de la prsente thse touche la modification et ltude des glucides en
vue dutiliser leurs proprits de surface, lchelle nanomtrique et molculaire. Afin
datteindre ce but, plusieurs pistes ont t explores. Dans le cas dune masse de
cellulose, le fait de rendre la pte de bois hydrophobe en la soumettant une raction en
phase gazeuse avec du trichloromthylsilane (TCMS) cre une matire appele
Cellufoam. Dans cette raction chimique, les groupes dhydroxyle (-OH) sont silyls, ce
qui inhibe laffinit entre la cellulose et leau. Quand langle de contact entre une goutte
deau et la surface est de 150, on parle de superhydrophobie. Lorsque de leau est insre
de force dans le lumen des fibres du Cellufoam et que la pte est presse en feuille
dessai, on remarque que le volume de la matire augmente en schant. Il sagit l de
linteraction entre la force capillaire et le rebond lastique. Aprs chaque cycle de
schage et de mouillage, la proprit dexpansion devient de moins en moins prononce.
Cette rduction de performance sexplique par lexposition des groupes dhydroxyle (OH), cause par des dommages mcaniques chaque cycle. lchelle nanomtrique, la
cellulose nanocristalline (CNC) est modifie pour stabiliser les mlanges htrognes de
mtaux en suspension dans leau. Les nanoparticules dargent (AgNP) sont la fois
synthtiss et stabiliss laide de la cellulose nanocristalline cationique (cCNC). On
remarque quune suspension forme dions chlorure, dions argent et de CNC ne peut
produire que du chlorure dargent. Cependant, si la CNC est remplac par la cCNC, pour
laquelle chaque particule est plus petite, mais expose une plus grande surface, la raction
produit de lAgNP, en raison de la grande prsence de groupes -OH. Dans cette thse, il a
t dtermin que les extrmits rductrices (C=O) et les groupes -OH sont tous les deux
capables de produire de lAgNP. Par consquent, on peut dduire que les sucres
rducteurs ne sont pas les seuls types de sucre capables de rduire les ions mtalliques.
Pour faire suite cette dcouverte, le saccharose, un sucre non rducteur, a t utilis
comme rducteur et stabilisateur dans la formation de lAgNP. En quelques heures, une
suspension collodale sest forme sans chaleur. laide dun MET haute dfinition, on
observe une structure de type cur-corce, dans laquelle le cur est compos dargent
monocristallin et lcorce, de saccharose, selon lanalyse lmentaire. En remplaant le
saccharose par larabinose ou le galactose, on obtient galement une suspension
collodale dargent, mais celle-ci est moins stable puisque lcorce est beaucoup plus
mince. En raison de la grande stabilit de la suspension dAgNP base de saccharose, un
miroir dargent se forme sur la surface de la fiole en verre. Sous des conditions
identiques, les suspensions collodales dargent stabilises par des sucres rducteurs se
sdimentent. Cette thse dmontre que les glucides, sous toutes leurs formes, possdent
des proprits qui varient selon leurs multiples groupements fonctionnels. Par silylation
et cationisation ainsi quen prsence dextrmits rductrices daldhyde, ces matires
peuvent devenir superhydrophobes, stabiliser des nanoparticules de mtal et former des
structures cur-corce dargent.

iii

Foreword
In addition to the introduction and conclusion chapters, this thesis includes five papers.
Chapters 2, 3, 4, and 5 each comprise one manuscript and appendix II is an abridged and
modified version of a manuscript. As of September, 2015, three papers have been
published and two are currently under review:
Chapter 2: Tejado A.; Chen W.C.; Alam M.N.; van de Ven, T.G.M. (2014)
Superhydrophobic foam-like cellulose made of hydrophobized cellulose fibres, Cellulose,
21 (3), 17351743.
Chapter 3: Chen, W.C.; Tejado, A.; Alam, M.N.; van de Ven, T.G.M.; (2015)
Hydrophobic Cellulose: A Material That Expands Upon Drying, 2015, Cellulose, 22 (4),
17351743.
Chapter 4: Chen, W.C.; Yang, H.; van de Ven, T.G.M. (2015) Cationic Cellulose
Nanocrystal Assisted Reduction of Silver(I) to Silver Nanoparticles, submitted to
Biomacromolecules, Manuscript ID: bm-2015-01045w.
Chapter 5: Chen, W.C.; van de Ven, T.G.M. (2015) Synthesis and Stabilization of Silver
Nanoparticles With Carbohydrate Shells, submitted to Carbohydr. Res., Manuscript ID:
CAR-D-15-00356
Appendix: Kaushik, M.; Chen, W.C.; van de Ven, T.G.M.; Moores, A. (2014) Imaging
Cellulose Nanocrystals by Transmission Electron Spectroscopy, Nordic Pulp and Paper
Res. J., 29 (1), 77-84. (Invited contribution to the special issue on Nanocellulose)
Contribution of Authors
All of the papers were co-authored by Dr. Theo van de Ven, the supervisor of this Ph.D.
project. All SEM and TEM microscopy found in this dissertation were performed with
the assistance of Dr. David Liu at the Facility for Electron Microscopy Research
(FEMR). The full manuscript of the work shown in the Appendix can be found in Madhu
Kaushik's dissertation. Other than the supervision and direction of Dr. van de Ven, all of
the work presented in this dissertation was performed by the author.

iv

Acknowledgements
Firstly, I would like to thank my supervisor, Dr. Theo van de Ven for his guidance in my
research over the past four and a half years. This Ph.D. project was great learning
experience thanks to the valuable advice and dedication that he has given me. I especially
feel a need to acknowledge Dr. van de Ven's patience in sharing his knowledge and
scientific expertise with me.

Although her name is not mentioned in the remainder of this dissertation, a major
contributor to my work is Fanny, who has been with me for the entirety of my postsecondary education. Without her encouragement or support, this work would not have
been possible. I need to thank her a second time for translating my abstract to French,
given her expertise as a translator.

I would also like to thank:

Alvaro Tejado, for his patience in working with me. His original work has helped us in
publishing two papers.
George Rizis, from whom I have learned all of my laboratory skills and from whom I
have learned about improving my relationships with people at the workplace.
All members of Dr. van de Ven's group, past and present: Amir A., Amir S., Dezhi,
Goeun, Han, Kevin, Leila, Nur, and Salman.
Colleen and Louis, for all of the help that they have provided me on this journey, which
are far too many to enumerate.
Chantal Marotte and other departmental staff. Their help is very much appreciated.

Table of content
Abstract............................................................................................................................................................ii
Rsum............................................................................................................................................................iii
Foreword.........................................................................................................................................................iv
Contribution of Authors...................................................................................................................................iv
Acknowledgements.........................................................................................................................................iv
Table of content...............................................................................................................................................vi
List of figures..................................................................................................................................................ix
Chapter 1..........................................................................................................................................................1
1.1. General Overview......................................................................................................................................1
1.1.1. Thesis Objective.....................................................................................................................................1
1.1.2. The Layout of the Thesis........................................................................................................................2
1.2. Introduction to Carbohydrates...................................................................................................................4
1.2.1. The Nanosctructure of Cellulose............................................................................................................7
1.3. The Prevention of Damage to Bulk Cellulose...........................................................................................7
1.3.1. Hornification and Hydrogen-Bonding....................................................................................................8
1.3.2. Hornification in this Thesis....................................................................................................................9
1.4. Cellulose Nanocrystals (CNC)................................................................................................................10
1.4.1. Sulphuric Acid Hydrolysis of Cellulose...............................................................................................10
1.4.2. Chemical Modification of Cellulose Nanocrystals...............................................................................12
1.4.3. Applications of Cellulose Nanocrystals................................................................................................13
1.5. Silver Nanoparticles................................................................................................................................14
1.5.1. Properties of Silver Nanoparticles........................................................................................................14
1.5.2. Synthesis and Stabilization of Silver Nanoparticles.............................................................................15
1.6. Contributions...........................................................................................................................................18
1.7. References...............................................................................................................................................19
Chapter 2:.......................................................................................................................................................22
2.1. Abstract....................................................................................................................................................22
2.2. Introduction.............................................................................................................................................23
2.3. Experimental Section...............................................................................................................................25
2.3.1. Materials...............................................................................................................................................26
2.3.2. Preparation of cellulose foam using low surface tension solvent.........................................................26
2.3.3. Chemical vapour deposition of trichloromethylsilane (TCMS)...........................................................27
2.3.4. Analyses................................................................................................................................................28
2.4. Results and Discussion............................................................................................................................28
2.5. Conclusion...............................................................................................................................................37
2.6. References...............................................................................................................................................37
Bridging section between Chapters 2 and 3...................................................................................................40
Chapter 3:.......................................................................................................................................................41
3.1. Abstract....................................................................................................................................................41
3.2. Introduction.............................................................................................................................................42
3.3. Experimental Section...............................................................................................................................43
3.3.1. Materials...............................................................................................................................................43
3.3.2. The preparation of cellulose foam........................................................................................................44
3.3.3. Sheet preparation and observation........................................................................................................44
3.3.4. Analyses................................................................................................................................................45
3.4. Results and discussion.............................................................................................................................45
3.5. Conclusion...............................................................................................................................................52
3.6. References...............................................................................................................................................53
Bridging section between chapters 3 and 4....................................................................................................55
Chapter 4:.......................................................................................................................................................56
4.1. Abstract....................................................................................................................................................56
4.2. Introduction.............................................................................................................................................57
4.3. Materials and methods.............................................................................................................................58
4.3.1. Materials...............................................................................................................................................58
4.3.2. Methods and characterization techniques.............................................................................................58

vi

4.3.3. Preparation of cationic cellulose nanocrystals (cCNC)........................................................................59

4.3.4. AgNP formation....................................................................................................................................59
4.3.5. Analysis................................................................................................................................................60
4.4. Results and discussion.............................................................................................................................61
4.5. Mechanism..............................................................................................................................................66
4.6. Conclusion...............................................................................................................................................69
4.7. References...............................................................................................................................................70
Bridging section between chapters 4 and 5....................................................................................................72
Chapter 5:.......................................................................................................................................................73
5.1. Abstract....................................................................................................................................................73
5.2. Introduction.............................................................................................................................................74
5.3. Materials and methods.............................................................................................................................75
5.3.1. Materials...............................................................................................................................................75
5.3.2. Methods and characterization techniques.............................................................................................75
5.3.3. AgNP formation....................................................................................................................................75
5.3.4. Analysis................................................................................................................................................75
5.4. Results and discussion.............................................................................................................................76
5.5. Conclusion...............................................................................................................................................83
5.6. References...............................................................................................................................................84
6. Conclusions, contributions, and suggestions for future work....................................................................85
6.1. Conclusions and contributions to original knowledge............................................................................85
6.1.1. Superhydrophobic cellulose handsheets...............................................................................................86
6.1.2. Superhydrophobic cellulose handsheet: A material that expands upon drying....................................88
6.1.3. Cationic cellulose nanocrystal assisted reduction of silver cations to silver nanoparticles.................89
6.1.4. Synthesis and stabilization of silver nanoparticles with carbohydrate shells.......................................91
6.1.5. Suggestions for future work.................................................................................................................92
Bridge to the appendix....................................................................................................................................94
Appendix I......................................................................................................................................................95
A-I-1. UV-VIS spectroscopy..........................................................................................................................95
A-I-2. X-ray diffraction..................................................................................................................................97
A-I-3. Electron diffraction............................................................................................................................100
A-I-4. X-ray photospectroscopy (XPS)........................................................................................................101
A-I-5. Energy-dispersive X-ray photospectroscopy (EDX).........................................................................102
A-II-6. Conductometric titration..................................................................................................................103
Appendix II:..................................................................................................................................................107
A-II-1. Abstract.............................................................................................................................................107
A-II-2. Introduction......................................................................................................................................108
A-II-3. Materials and methods......................................................................................................................110
A-II-3.1. Materials and equipment...............................................................................................................110
A-II-3.2. Preparation of CNC for never-dried samples................................................................................111
A-II-3.3. Samples preparation for TEM imaging.........................................................................................111
A-II-3.4. Synthesis of Pd nanoparticles onto CNCs (PdNPs@CNCs).........................................................112
A-II-3.5. Low dose and high resolution transmission electron microscopy.................................................112
A-II-4. Results..............................................................................................................................................113
A-II-4.1. Optimization of dispersion conditions for TEM imaging: pH and grid type................................113
A-II-4.2. The impact of CNC drying history on TEM imaging...................................................................116
A-II-4.3. Comparison of TEM images of CNCs under high resolution and low dose microscopy.............121
A-II-4.4. Morphology of modified/decorated CNCs with palladium nanoparticles....................................121
A-II-5. Discussion........................................................................................................................................122
A-II-6. Conclusions......................................................................................................................................125
A-II-7. References........................................................................................................................................126

vii

List of figures
Fig. 1.1. The equilibrium structures of glucose, an aldohexose..................................5
Fig. 1.2. Wood fibres split into various component parts: nanofibrils (4-10 nm),
protofibrils (10-20 m), microfibrils (50-100 m), and wood stem (2-3 m)....6
Fig. 1.3. Cross-section showing the deformation of a fibre caused by nanofibril
coalescence; shown with microfibrillation after refining....................................6
Fig. 1.4. The amorphous regions of cellulose chains can undergo preferential
hydrolysis, which results in the isolation of cellulose nanocrystals (CNC)......11
Fig. 1.5. The straight chain and the pyranose form of -D-glucose with carbon
positions marked................................................................................................12
Fig. 1.6. Schematic of a dead E. coli bacterium with its cell wall and DNA damaged
by silver granules formed from silver nanoparticles.........................................15
Fig. 2.1. Schematic illustration of two cellulose fibres crossing each other
perpendicularly with water and ethanol represented by blue areas: drying from
water (top) causes shrinkage of the fibres due primarily to collapse of the
lumen, while drying from ethanol (bottom) preserves the open fibre
structure.............................................................................................................23
Fig. 2.2. Experimental set-up for the reaction of cellulose with TCMS. The reaction
vessel and TCMS were preheated in a 65C oven. A mesh basket containing the
pulp sample was introduced and held firmly in place by a rubber stopper inside
the vessel. The sample was reintroduced into the oven and the gas-phase TCMS
was made to react with the sample....................................................................27
Fig. 2.3. (Left) Wilhelmy balance setup and equation used, where is the surface
tension; (right) a table showing the apparent foam density of wood pulp dried
from various solvents and their surface tension above the critical micelle
concentration (cmc). SKL (cmc = 0.8 wt%), Dowfax 2A1 (cmc = 0.007 wt%),
and Triton X45 (cmc = 0.0136 wt%).................................................................29
Fig. 2.4. 3D images of individual fibres obtained with Skyscan 1172 Micro CT XRay tomography: a) after being dried from water and b) from a low surface
tension solvent (ethanol). c) Fibre size distribution of bHKP fiber in water
before drying obtained by using HiRes fiber quality analyzer (FQA)..............30
Fig. 2.5. SEM images of bHKPcellulose fibers of a dried from water, b dried from
ethanol, c cross-sectional view dried from water and d cross-sectional view
dried from ethanol..............................................................................................30
Fig. 2.6. A plot of the surface tension with respect to surfactant concentration; in this
case, Triton X45 is used and has a known surface tension of 29 mN/m...........31
Fig. 2.7. SEM images of Q90 nonbeaten cellulose fibers a) dried from water, b)
dried from anionic surfactant, c) cross-sectional view dried from water and d)
cross-sectional view dried from anionic surfactant...........................................32
Fig. 2.8. TCMS adsorption as a function of reaction time; also shown is the coverage
assuming a surface area of 10 m2/g of kraft pulp (Alince & van de Ven, 1997)33
Fig. 2.9. Water droplet on the surface of Cellufoam captured by a CCD camera.....34
Fig. 2.10. Unmodified hydrophilic (left) and TCMS-hydrophobized (right) cellulose
fibres in contact with a drop of water. Hydrophilic fibers are drawn into the
water drop. Only a sufficiently hydrophobic fiber remains at the water air
interface and shows zero engulfment................................................................36

viii

Fig. 3.1. The dry-expansion of sheets made from 100 % softwood cellulose fibres. A
wet handsheet is pressed and placed onto a glass slide. Images of its cross
section were taken before drying (solid content of 50 %, left) and after drying
(solid content of 100 %, right)...........................................................................46
Fig. 3.2. Each subsequent cycle results in progressively thicker wet handsheets due
to flocculation that also causes a decrease in relative expansion when dry; the
contact angle also decreases implying that the material becomes less and less
hydrophobic (the Q90 pulp sample used in this figure is a different batch from
the Q90 sample of Fig. 3.1)...............................................................................47
Fig. 3.3. The dry-expansion of Cellufoam when drying during the first cycle (a) and
solid content of Cellufoam as a function of drying time (b)..............................48
Fig. 3.4. The proposed mechanism for the expansion of Cellufoam upon drying; the
illustration of the lumen and fibre wall is not drawn to scale and fibre crosssections are not necessarily tubular...................................................................48
Fig. 3.5. The relationship between the tensile index and solids content for handsheets
made from fibres of various lengths and curl indices: hardwood (HKP) is
mainly short fibres of low curl index and softwood (NDSKP and Q90) are long
fibres of high curl index.....................................................................................50
Fig. 4.1. The reaction scheme of CNC with QUAB 151 (a quaternary ammonium
epoxide) to produce cCNC and AgNP...............................................................58
Fig. 4.2. Conductometric titration curve of AgNO3 with cCNC; an equivalence point
at 7.1 mL (using a 0.01 M standard), cCNC (50 mL, 0.3% w/w suspension) was
calculated to contain 0.47 mmol/g of cationic charge groups...........................60
Fig. 4.3. AFM tapping mode image of unmodified CNC (left) and cCNC (right); the
height profile distribution of the two samples is plotted as two separate
histograms (bottom)...........................................................................................61
Fig. 4.4. The particle size distribution of AgNP from over 10 TEM images; (right) an
example of TEM imaging where AgNP is depicted as dark spheres.................63
Fig. 4.5. (Left) The XRD patterns of: 1) cationic wood pulp and AgCl 2) CNC and
AgCl 3) cCNC with AgNP; (Right) The 220 reflection of AgNP is shown for
sample 3.............................................................................................................64
Fig. 4.6. The EDX elemental analysis spectrum of silver nanoparticles viewed using
TEM at 200kV; the Ag peaks are prominent, but the Cl peaks are weak and
slightly overlapped by secondary Ag peaks; this technique is paired with XPS
for a complete analysis......................................................................................66
Fig. 4.7. The XPS spectra of AgNP (with 3-point moving average) from cationic
cellulose assisted reduction compared to a silver chloride standard (left and
middle); the C-C and C-O peaks are also compared before and after etching of
the surface by using Ar+ gas...............................................................................66
Fig. 5.1. The UV-VIS spectra of 5 different carbohydrates with their most intense
peaks shown at approximately 450 nm..............................................................77
Fig. 5.2. The reduction of silver cations to AgNP using galactose (reducing sugar)
proceeds on the order of minutes (A) and rapidly sediments after 120 minutes;
the reduction using sucrose (non-reducing sugar) proceeds on the order of
hours (B) and did not sediment..........................................................................78

ix

Fig. 5.3. A core-shell structure of AgNP and carbohydrates is seen by high resolution
TEM using galactose (A) and sucrose (C); a low-resolution image of the same
systems reveal no core-shell structure when using galactose (B), though a faint
shell can be seen in the case of sucrose (D) given the added thickness............80
Fig. 5.4. High-resolution TEM of Core-shell structures for higher molecular weight
carbohydrates: agarose (A) and chitosan (B).....................................................81
Fig. 5.5. The EDX spectra of AgNP stabilised by various carbohydrates with the
electron beam aimed at the silver core (A) and the sugar shell (B)...................81
Fig. 5.6. Electron diffraction pattern of the silver core (220) (A) and the amorphous
sugar shell (B); the diffraction patterns for all samples were closely similar;
these images were obtained using galactose stabilised AgNP...........................82
Fig. A-I-1. The contribution of absorption and scattering to the extinction (left); the
formation of a plasmonic dipole in spherical nanoparticles as determined by
computation (right); the colours of the dipole (right) indicate the electric field
enhancement where the middle slice exhibits maximal enhancement..............96
Fig. A-I-2. The crystal spacing (d) as well as the 2 angle are illustrated.................98
Fig. A-I-3. The top, middle, and lower rows represent the cubic (sc), the bodycentric cubic (bcc), and the face-centric cubic (fcc), respectively.....................99
Fig. A-I-4. A diagram illustrating the working principles of electron diffraction. . .101
Fig. A-I-5. The working principle of XPS...............................................................101
Fig. A-I-6. EDX works by ejecting core electrons and detecting the X-rays emitted
by the transition of electrons from high energy levels to fill these electron holes
.........................................................................................................................103
Fig. A-I-7 The plot on the left shows the pH versus titrant volume titration curve;
the plot on the right shows the conductivity versus titration volume titration
curve where three distinct regions are represented by different slopes...........104
Fig. A-I-8 The slope of the strong acid is always steeper than that of the weak acid;
the weak acid can have a positive or negative slope depending on its acid
constant; excess base is characterized by a significant rise in conductivity....105
Fig. A-II-1. Never-dried sample on carbon grid: (left) at pH 5-6; (right) at pH 3.5 115
Fig. A-II-2. Never-dried CNCs at pH 5-6: on formvar grid (left); silicon monoxide
grid (right)........................................................................................................115
Fig. A-II-3. Length and width distribution of CNCs bundles/rod of never-dried
CNCs at pH 3.5 on different type of grids, based on the counting of 200
particles on each type of grid...........................................................................116
Fig. A-II-4. Types of CNCs at pH 5-6, carbon grid: never-dried CNCs (left); freezedried CNCs (middle); spray-dried CNCs (right).............................................118
Fig. A-II-5. Spray-dried CNCs (at pH 5-6) onto a formvar grid (left), carbon grid
(middle) and silicon monoxide (right).............................................................119
Fig. A-II-6. Freeze-dried CNCs (pH 5-6 and carbon grid): high resolution TEM (top)
and low dose TEM (bottom)............................................................................119
Fig. A-II-7. TEM image of modified CNCs, PdNPs@CNCs..................................122
Fig. A-II-8. Histograms of CNCs single rods length (top) and width (bottom)
measured on 1000 particles for never-dried particles on carbon grids at 3.5 pH.
.........................................................................................................................124

Chapter 1
1.1. General Overview
In materials science, cellulose's abundance, renewability, and biodegradability have given
researchers many reasons to study this natural biopolymer to exploit and to improve upon
its properties. With the ever-increasing need for developing materials in a sustainable
manner, cellulose has often been considered for its properties in many industries. As a
highly abundant raw material produced by plants, bacteria, and sea-life, various forms of
carbohydrates and cellulose are readily available in many regions of the world. For this
reason, research groups have demonstrated interest in modifying and studying
carbohydrates of different molecular weights in order to implement them for a wide
variety of novel uses. As part of the Green Fibre Network, funding from Natural Sciences
and Engineering Council of Canada (NSERC) and the FPInnovations industrial research
chair has allowed for this research project to be conducted.
1.1.1. Thesis Objective
The present thesis describes the chemical modification of carbohydrates on the macro-,
the nano-, and the molecular scale to change its interaction either with water or in
heterogeneous systems. The objective of this work is to use saccharides to improve the
properties of various macro- and nanosized systems. On the macroscale, the prevention of
lumen collapse in cellulose handsheets during de-wetting was studied. The hornification
of wood fibres is characterized by dry-shrinkage resulting in structural changes to fibres.
This phenomenon was prevented by making cellulose hydrophobic with
trichloromethylsilane (TCMS) in the gas phase. In employing this chemical reaction, it
was found that the material properties of the cellulose handsheet can be greatly altered.

On the nanoscale, cellulose was used to both reduce and to stabilize metal nanoparticles
in a colloidal suspension. In an aqueous mixture of silver cations (Ag+), cationic cellulose
nanocrystals (cCNC) can both reduce Ag+ to silver nanoparticles (AgNP) and stabilize
their formation thus achieving good size distribution. On the molecular scale, silver
colloidal suspensions were again studied, but using various reducing and non-reducing
sugars of different molecular weights. It was observed that a core-shell structure selfassembles where the core is silver and the shell is made of carbohydrates.
1.1.2. The Layout of the Thesis
The introduction will be divided into five sections. The first section is a brief introduction
to the structure of cellulose. The following section deals with chemical modifications to
cellulose fibres to prevent hornification. The third section focuses on cellulose
nanocrystals (CNC) and means of studying these nanometer-sized organic nanoparticles.
The fourth section is a scientific literature review of the reduction and stabilization of
silver cations to silver nanoparticles using carbohydrates. The final section provides
detail on contributions for published work.
This thesis consists of eight chapters. Chapter 1, the introduction, is divided into four
sections, and provides fundamental information to help lay the foundation to the work
presented in this thesis. The stability of macroscopic and nanoscopic systems are
explained in this chapter.
Chapter 2 is the first of two projects focused on chemically modifying cellulose pulp to
possess low density and superhydrophobicity. This treatment solves the problem of
hornification and is published in journal titled Cellulose: A. Tejado, W.C. Chen, Md N.

Alam, T.G M. van de Ven, Superhydrophobic foam-like cellulose made of hydrophobized

cellulose fibres, Cellulose, 2014, 21 (3), 17351743.
Chapter 3 is a continuation of the project in chapter 2. In this case, the superhydrophobic
pulp is forced to uptake water in vacuo and subsequently pressed into a handsheet. This
handsheet, when dried, expands, which makes it the first reported material to possess dryexpansion properties. The work presented is published in Cellulose: W.C. Chen, A.
Tejado, Md N. Alam, T.G.M. van de Ven, Hydrophobic Cellulose: A Material That
Expands Upon Drying, Cellulose, 2015, 22 (4), 17351743.
Chapter 4 is a study on the cationic modification of CNC to form a new type of
nanoparticle that is capable of reducing silver (I) to silver (0) as size-controlled silver
nanoparticles (AgNP). This work is submitted to Biomacromolecules: W.C. Chen, H.
Yang, T.G.M. van de Ven, Cationic Cellulose Nanocrystal Assisted Reduction of Silver(I)
to Silver Nanoparticles, 2015, Manuscript ID: bm-2015-01045w
Chapter 5 is devoted to the use of reducing and non-reducing sugars to form and stabilize
AgNP in aqueous suspension. In this work, the use of non-reducing sugars to cause the
formation of core-shell AgNP and the silver mirror resulting from colloidal stability are
discussed. This work is submitted to Carbohydrate Research journal: W.C. Chen, T.G.M.
van de Ven, Synthesis and Stabilization of Silver Nanoparticles With Carbohydrate
Shells, 2015, Manuscript ID: CAR-D-15-00356.

Chapter 6 is the final chapter of the thesis and contains the conclusion of the main body
of work from all previous chapters, contributions to the known scientific literature, and a
broad view for future developments in the field of biopolymeric research.
There are two appendix chapters. The first one is a review of several instrumental and
analytical techniques used in previous chapters. The second appendix chapter is based on
work published in the Nordic Pulp and Paper Research Journal: M. Kaushik, W. C. Chen,
T. G. M. van de Ven, A. Moores, Imaging Cellulose Nanocrystals by Transmission
Electron Spectroscopy, Nordic Pulp and Paper Res J, 2014, 29 (1), 77-84. In this chapter
and special issue on nanocellulose, the nano-sized biopolymeric particles CNC are
imaged using transmission electron microscopy (TEM) under different conditions.
1.2. Introduction to Carbohydrates
Carbohydrates are naturally occurring compounds that comprise simple sugars called
monosaccharides; more complex sugars such as disaccharides and oligosaccharides, and
higher molecular weight macromolecules and polymers called polysaccharides. The
number of carbon atoms per sugar unit and the presence of aldehydes and ketones
determine the class of the carbohydrate. In this thesis, there is a heavy emphasis on
cellulose, which is a polymer of glucose units. The simple sugar glucose is an aldohexose
(presence of an aldehyde group and 6 carbons). A stereoisomer of cellulose is starch
because both biopolymers are composed of glucose units linked via the first and fourth
carbons. The defining feature between the two lies in the linkage between carbon 1 and 4.
In the case of cellulose, the 1-4 bond is an equatorial -bond while for starch, the bond is
axial (-bond).

Pentoses and hexoses (5 and 6-membered sugars) are capable of undergoing transitions
between straight chain, furanose (5-membered ring), and pyranose (6-membered ring)
form. Depending on the sugar, steric effects will cause the equilibrium to favour a
specific structure over the others. For the purpose of this thesis, sugars are drawn in the 6membered chair configuration because glucose is most sterically stable in this form.

Fig. 1.1. The equilibrium structures of glucose, an aldohexose

As mentioned previously, the basic building block of cellulose is D-glucose, but more
precisely, these individual sugar blocks are called D-anhydroglucopyranose units (AGU)
(Krssig, 1993). From the three hydroxyl (-OH) groups present in each AGU, the
functional groups at position C2 and C3 are more reactive in etherification reactions than
at position C6 because both behave as secondary alcohols. The -OH group at position C6
is least reactive in etherification reactions (Wang et al., 2014); however, in terms of
esterification reactivity, the order is reversed: C6 > C2 = C3 (Chen, 2014). Cellulose
chains are bundled together into microfibrillated cellulose particles that can vary greatly
in size due to the polydispersity of the bundles. Under normal conditions, protofibrils, the
elementary fibrils, are made up of 36 cellulose chains (Sommerville, 2006; Mutwil et al.,
2008). These protofibrils assemble into the tightly bound larger units called microfibrils,
which are then assembled into cellulose fibres (Fan et al., 2008).

Fig. 1.2. Wood fibres split into various component parts: nanofibrils (4-10 nm),
protofibrils (10-20 m), microfibrils (50-100 m), and wood stem (2-3 m)
The macroscopic structure of cellulose fibres is composed of nanofibrils arranged into
many layers of lamellar structures along the cell wall (Fahlen & Salmen, 2002). The
relevance of the macroscopic structure to the work presented in this thesis lies in the fact
that voids and pores form between nanofibrils, which allows water to permeate. Due to
weak-force interactions such as hydrogen bonding, these interstitial spaces can constrict
and irreversibly close as nanofibrils coalesce.

Fig. 1.3. Cross-section showing the deformation of a fibre caused by nanofibril

coalescence; shown with microfibrillation after refining

1.2.1. The Nanosctructure of Cellulose

Based on evidence from imaging (AFM, SEM, and electron tomography (Chundawat et
al., 2011)), microfibril aggregates are assumed to be the fundamental cohesive unit of
wood cell walls (Fernandes et al., 2011). These units comprise crystalline and amorphous
regions, which when subjected to strong acid treatment, result in the isolation of the only
the crystalline parts. These shorter rod-like particles are known in the scientific literature
as cellulose nanocrystals (CNC) (Ranby, 1951; Oke, 2010) and are also called cellulose
nanowhiskers or nanocrystalline cellulose (NCC). Although this material is an organic
material and is therefore less electron-dense, CNC can be imaged by TEM with relatively
good contrast. Another means of imaging CNC is by AFM, which also provides the
height profile of the nano-sized rods. The only drawback of AFM is the effect of shape
perturbation from the broadening phenomenon associated with the AFM tip. This causes
more box-shaped CNC from sources such as valonia to appear round (Habibi et al.,
2010). Not only are the dimensions of CNC different depending on its source and
treatment conditions, but other properties also vary. A detailed account on the analysis of
CNC is presented in section 1.4.
1.3. The Prevention of Damage to Bulk Cellulose
Chemical modification of bulk cellulose is a practice with an extensive history. A lot of
work has been conducted to chemically modify the functional groups within wood pulp
in order to provide the final product with enhanced properties where protection against
water damage is one type of enhancement under investigation within the scope of this
work. The large presence of hydroxyl (-OH) groups within the cellulose structure causes
this material to have an affinity for water; it also signals that chemical modifications can

be made at these sites. As far as paper-based products are concerned, irreversible damage
to the integrity of the fibre mat is often attributed to wetting. In the scientific body of
work on this phenomenon, the term 'hornification' is used to describe the irreversible
coalescence of microfibrils caused by the hydrogen bonding of polysaccharide chains
(Jayme, 1944). Subsequently, this coalescence also leads to lumen and pore collapse that
cellulose fibres undergo upon drying (Hubbe et al., 2007). It is clear that fibres from
certain trees are more likely to succumb to hornification than others due to differences in
fibre dimensions and composition. Nevertheless, there are numerous studies on reducing
the risk of hornification.
1.3.1. Hornification and Hydrogen-Bonding
Numerous methods have been devised to prevent hornification. The inactivation of
cellulose fibre surface is achieved by blocking the formation of hydrogen bonds. It has
been shown that solvent exchange, the use of extractives and fatty acids, and chemical
modification of the -OH groups can help prevent hornification (Brancato, 2008). In past
work, as little as 2.2% substitution by methylation of the -OH groups is sufficient to
disrupt internal bonding. Other ways of achieving this would be silylation. In the case of
substitution reactions, the goal is always to hydrophobize to some extent cellulose's
chemical structure. The issue with chemical modification is the higher expense associated
with this strategy. Instead of substitution, the treatment of wood pulp with a water-soluble
material can also prevent hornification. In this case, the material would bond to the -OH
groups thus obstructing contact between cellulose microfibrils during drying. It has been
found that the addition of sucrose and/or glycerol can perform this task at a loading
concentration of 20%. Furthermore, the effectiveness is limited to 1 cycle because these

blockers are washed out during repulping. Therefore, although the use of blockers is more
economically viable, they are ineffective in the case of paper recycling.
Studying and understanding the phenomenon of hornification is of high significance to
the forest industry. Despite centuries of research on the topic of wood pulp, no
economical industrial method can prevent irreversible damage to cellulose fibres due to
the effects of drying. The consequence of this irreversible change is a reduction in the
quality of the cellulose-based material. This is especially true in cases where wood pulp is
delignified because lignin and hemicellulose act as spacers for wood fibres, which
prevents lumen and pore collapse and are part of the integral make-up of wood.
Removing these constituents from wood pulp generally results in a loss of strength
properties. Significant issues associated with hornification are the reduced accessibility to
the buried cellulose chains, the loss of strength, and fibre mat integrity.
1.3.2. Hornification in this Thesis
Of the numerous techniques to help prevent hornification, solvent exchange and chemical
modification were two methods used in conjunction to solve the problem in the work
presented (Klemm et al., 1998). Nevertheless, other means such as the use of additives
(Nemati et al., 2011) and freeze-drying (Rder & Sixta, 2004) have also been shown to
prevent hornification. In all of the methods mentioned, the pore and lumen collapse is
assumed to be prevented based on the water retention volume (WRV). Imaging by means
of scanning electron microscopy and X-ray tomography can also be used to determine the
cross-sectional lumen structure.

In this thesis, two chapters of work are devoted to preventing lumen collapse when
drying cellulose fibre mats. This has been achieved by performing solvent exchange and
chemical modification. Analysis was performed by imaging using SEM and X-ray
tomography. Furthermore, contact angle goniometry was performed to assess the
hydrophobicity and the cross-sectional thickness of the fibre mat.
1.4. Cellulose Nanocrystals (CNC)
1.4.1. Sulphuric Acid Hydrolysis of Cellulose
It is known that sulphuric acid (H2SO4) hydrolysis of cellulose is responsible for the
breakdown of fibres into rodlike fragments (Roman & Winter, 2004). The reason for
which hydrochloric acid (HCl) is not used is explained by the colloidal stability of the
nanocrystalline particles in aqueous suspension. If HCl were used instead of H2SO4, then
aggregation occurs and results in the formation of microcrystalline cellulose (MCC)
(Araki et al., 1998). The sulfonate groups introduced onto the rodlike fragments esterify a
low amount of surface hydroxyl (-OH) groups. This chemical reaction is responsible for
the colloidal stability of cellulose nanocrystals (CNC) (Ranby, 1949). The amount of
sulfonate groups present can be determined by a potentiometric titration Gran plot. It was
found that higher reaction temperature, longer reaction times, and higher acid
concentration all lead to a higher amount of sulfonate groups esterified to the cellulose
nanorod surface. The problem with these conditions lies in the possibility of degradation
and decomposition of the sample "burning". It was found that, even lower temperature,
shorter reaction time, and lower sulphuric acid concentration, a sufficient presence of
sulfonate groups can be achieved. This, in turn, leads to a stable suspension.

10

Fig. 1.4. The amorphous regions of cellulose chains can undergo preferential hydrolysis,
which results in the isolation of cellulose nanocrystals (CNC)
The dimensions as well as the physical properties of CNC greatly depend upon the source
of the raw material and the treatment that resulted in the isolation of the CNC product. In
a published work from 1998 (Dong et al., 1998), the mean particle size of CNC was
determined for samples at a wide range of acid hydrolysis reaction time. It was found
that, using sulphuric acid to hydrolyse cellulose fibres, a shortening of the CNC length
occurred along with an increase in sulphur content. With a fixed acid concentration (64%
throughout the work), at a reaction temperature range of 26C to 45C, an ivory-coloured
suspension of CNC was observed depending on the reaction time. However, increasing
the reaction temperature resulted in a darkening of the sample to the point of charring at
65C in only 1 hour. In this same work, a detailed histogram and two clear TEM images
demonstrating the size distribution of the CNC nanorods were shown for samples that
underwent either 20 minutes or 4 hours of acid hydrolysis at 45C. In agreement with the
theory proposed by the author, longer exposure time to sulphuric acid resulting in shorter
CNC particles and more highly charged surfaces resulting in better dispersion.
The TEM imaging work shown in many published papers involves no negative staining
with heavy metals such as uranium salts. However, with increasing analysis on CNC,
such staining techniques have gained popularity in this field of study. It is important to
11

note, though, that TEM imaging can often produce artifacts, especially from drying. The
introduction of heavy metal salts can also obscure features in organic/inorganic systems
due to poor image contrast. The details of these issues associated with TEM imaging of
CNC is further discussed in the appendix of this thesis.
1.4.2. Chemical Modification of Cellulose Nanocrystals
The abundance of -OH groups for bulk cellulose is also observed in nanocellulose. After
all, the latter, when bundled up, is the structural building block for the former. Given that
these -OH groups are reactive, chemical modifications can be carried out to alter the
properties of CNC. Amongst the 3 hydroxyl groups per AGU, the reaction site at position
C6 acts as a primary alcohol and is susceptible to many more types of reactions than at
sites C2 and C3. In this chapter, a scientific review serves to summarize the numerous
reactions studied in the past on CNC.

Fig. 1.5. The straight chain and the pyranose form of -D-glucose with carbon positions
marked
For example, work published in 2007 (Dong & Roman, 2007) demonstrated the
fluorescent labelling of CNC using fluorescein-5'-isothiocyanate (FITC). This work
makes bioimaging possible when using CNC in fluorescence bioassay studies. Similar to

12

reactions with bulk cellulose, silylation is another important chemical modification to

CNC that renders the nanorods hydrophobic. The use of alkyldimethylchlorosilanes
(AKD) can make the surface of cellulose hydrophobic at a higher degree of substitution,
though this has been found to disrupt the structural integrity of the nanoparticles.
Amongst the countless reactions that can be discussed in this section, cationization will
be presented because this chemical modification has been extensively performed on
cellulose to stabilize silver nanoparticles in Chapter 4. Work performed in 2008 (Hasani
et al., 2008) showed that cationic ammonium groups can be tailored onto cellulose by
grafting epoxypropyltrimethylammonium chloride onto CNC. Unmodified CNC
possesses a slight negative charge due to the presence of sulfonate groups and, by
reversing the charge to positive, a stable aqueous suspension is still observed, though
with gelling properties at a higher degree of substitution. The degree of substitution is
measured by titration with silver nitrate (AgNO3) to measure the precipitation of chloride
anions with silver cations. In the present thesis, it was also assumed that silver chloride
(AgCl) nanoparticles would result from this titration. However, further investigation
suggested that silver nanoparticles can form instead.
1.4.3. Applications of Cellulose Nanocrystals
There are numerous possible applications of CNC. A popular use is the incorporation of
this material in nanocomposite polymers. Given its tensile strength, CNC is a good
candidate for reinforcing polymeric matrices. In the case of the aforementioned
length/diameter aspect ratio, a higher value is characteristic of a better reinforcing filler.
In a publication from 2004 (Samir et al., 2004), nanowhiskers of cellulose ranging from

13

an aspect ratio of 10 to 67 showed greater modulus increase for cellulose with higher
aspect ratios. This helps to demonstrate the tunability of CNC for various applications.
1.5. Silver Nanoparticles
1.5.1. Properties of Silver Nanoparticles
The study of metal nanoparticles is a very broad and active field in nanoscience and
technology linking to carbohydrates such as simple sugars and cellulose for synthesis and
stabilization (see section 1.5.2). Depending on size, shape, and surface morphology,
significant changes to the material's electrical, chemical, optical, electronic, and thermal
properties can be greatly altered (Raveendran et al., 2003). Noble metals such as silver
and gold nanoparticles are also effective in medicine and food packaging because of their
antimicrobial properties (Sharma et al., 2009). It has been demonstrated that silver
nanoparticles (AgNP) can hinder the growth of E. coli, S. aureus, and a list of other
bacteria. A noble metal nanoparticle such as gold can be used in biosensors for the
detection of pathogens and be used for controlled drug delivery (Ayala et al., 2012).
Although it has been long-established that silver has the ability to prevent bacterial
growth, convincing mechanistic evidence was only published in a paper 15 years ago
(Feng et al., 2000). In this work, TEM images of E. coli and S. aureus showed the uptake
of AgNP into the cell. The small dark nanoparticles of silver either adsorbed to the cell
wall of bacteria or permeated into the nucleus, thus condensing the DNA of the
bacterium. It has been proposed in that paper that the penetration of silver ions from the
nanoparticles into the cell wall can react with the thiol groups of proteins found in DNA,
which causes the DNA to condense. This mechanism leads to either damage or death of

14

the microorganism. In this same paper, it was also found that E. coli was more susceptible
to AgNP than S. aureus because the former is a Gram-negative bacterium while the latter
is Gram-positive. The peptidoglycan in the cell wall of Gram-positive bacteria protects
against silver nanoparticles and prevents the penetration and entry of silver ions into the
cytoplasm.

Fig. 1.6. Schematic of a dead E. coli bacterium with its cell wall and DNA damaged by
silver granules formed from silver nanoparticles
1.5.2. Synthesis and Stabilization of Silver Nanoparticles
In the past, numerous methodologies for the synthesis and stabilization of AgNP have
been published. They all produce similar results in which silver of a given shape no larger
than 100 nm is produced. This section will focus on the reduction of silver cations (Ag+)
to silver metal (Ag0) as explained in the scientific literature. Moreover, because many of
these publications describe a reduction and stabilization mechanism that often differ and
contradict one another, a combination of the knowledge gained from various work is
needed to try and understand what causes the formation of stable AgNP.

15

In Raveendran's 2003 publication (Raveendran et al., 2003), it was shown that starch can
reduce silver nitrate (AgNO3) to AgNP. The synthesis was carried out in argon gas at
40C for 20 hours. The nanoparticles were characterized by UV-VIS spectroscopy, which
shows a broad peak at 419 nm. This signal is in a wavelength range associated with
AgNP. In numerous published work, analysis by UV-VIS spectroscopy is considered
as a measure of the absorption of light with respect to wavelength, extinction due to
absorption and scatter must be considered. This is discussed in Appendix I. The size
distribution of the nanoparticles was measured by TEM and subsequent particle counting
and revealed a mean size of 5.3 nm. This work was amongst the earlier explorations of
using carbohydrates to synthesize AgNp discussed in this thesis, which is why the
reaction conditions appear fairly harsh and the analytical techniques fairly limited. The
nano-size metal particles that formed were a result of the high amount of hydroxyl (-OH)
groups, which facilitates the complexation of silver ions to the molecular matrix. In turn,
this prevents the aggregation of silver atoms, thus allowing for nanosized silver to be
produced.
Raveendran described the reducing sugar mechanism as a means of explaining how metal
ions can be reduced in the presence of carbohydrates; this is a popular view in the present
day. However, many other mechanisms have also been proposed to explain the reduction
mechanism. In Travan's 2009 publication (Travan et al., 2009), the reduction of silver
ions by using derivatives of chitosan was explored. It was proposed in this work that a
polyol reduction mechanism is responsible for the reduction of silver ions. More
specifically, a primary alcohol is oxidized to an aldehyde, followed by the formation of a
cyclic hemiacetal, which is oxidized to a lactone. Oxidation can provide electrons to

16

nearby metal ions, thereby forming metal nanoparticles. The use of 13C-NMR and IR
spectroscopy helped identify the presence of lactone moieties and carboxylic acid groups.
AgNP was, again, characterized by UV-VIS spectroscopy and TEM imaging. Raman
spectroscopy was also employed and showed an enhanced SERS effect when AgNP was
present; this suggests that there is a strong interaction between carbohydrate surfaces and
silver nanoparticle surfaces. This evidence corroborates the notion that the formation of
small AgNP is achieved by the presence of sugar units. However, the polyol reduction
mechanism differs from the reducing sugar mechanism proposed in the previous work by
Raveendran.
Ayala's 2012 publication in dealing with AgNP and gold nanoparticles (AuNP) showed
that both reducing and non-reducing sugars can cause reduction of noble metal ions to the
metal nanoparticle form. This work shows that metal reduction can occur even without
reducing chain ends. This work does not suggest that the reducing sugar mechanism is
wrong; in fact, no mechanistic discussion was included. It can be assumed, based on
numerous publications, that no true agreement on the reduction mechanism has been
made. The use of simple non-reducing sugars such as sucrose and fructose yielded more
prominent absorbance at 419 nm than reducing sugars such as maltose and and galactose.
In this same paper, starch, comprising non-reducing sugar units, yielded a less prominent
absorbance signal at 419 nm than glucose, a reducing sugar. However, starch is a high
molecular weight polydisperse polymer made up of sugar units while glucose is a small
molecule monosaccharide.

17

In the present thesis, the reduction of Ag+ ions to AgNP is studied under conditions that
differ from the papers referenced. In this case, an approach to modify cellulose making it
cationic was employed as a novel means of synthesizing AgNP (see chapter 4). On the
other hand, the use of simple sugars with or without reducing ends can also produce
varied results (see chapter 5). With the data obtained under specific conditions, a higher
level of understanding can be achieved to elucidate the mechanism of formation of AgNP.
1.6. Contributions
At the time of writing this thesis, three papers have been published and two more have
been submitted. In the chronological order of writing, the first publication was a
collaborative work with Dr. A. Moores from McGill University and graduate student M.
Kaushik. The paper titled "Imaging Cellulose Nanocrystals by Transmission Electron
Spectroscopy" was published in the Nordic Pulp and Paper Research Journal in 2013 (see
Appendix II). The author of the present thesis performed pH dependent TEM imaging on
TEM grids coated with hydrophilic, neutral, and hydrophobic layers as well as
contributing to the writing of the manuscript for the publication. This allowed for a more
thorough understanding of the dispersion of CNC particles on TEM grids. The second
publication with Dr. A. Tejado, Dr. M.N. Alam, and Dr. T.G.M. van de Ven titled
"Superhydrophobic foam-like cellulose made of hydrophobized cellulose fibres" was
published in Cellulose in 2014 (see chapter 2). The author of the present thesis performed
chemical modification of wood pulp and analytical work to demonstrate the low-density
and superhydrophobicity of the modified material as well as writing the manuscript for
the publication in full. A. Tejado is first author because he initiated the research and
provided the first samples for analysis. He also performed SEM and X-ray tomography

18

analysis on the pulp samples. In the third publication titled "Hydrophobic Cellulose: A
Material That Expands Upon Drying", experimental, analytical, and written work were
contributed by the author of this thesis (see chapter 3). Co-authors Dr. A. Tejado and Dr.
M.N. Alam provided experimental results and Dr. T.G.M. van de Ven was the principal
investigator. The two manuscripts currently under review involve the use of cationic
cellulose nanocrystals (cCNC) to reduce and stabilize silver nanoparticles (AgNP) (see
chapter 4) and the use of various carbohydrates with or without reducing ends to reduce
and stabilize AgNP (see chapter 5). In both papers, the primary author is the author of this
thesis and the principal investigator is Dr. T.G.M. van de Ven.
1.7. References
Araki, J., Wada, M., Kuga, S., Okano, T. (1998) Flow properties of microcrystalline
cellulose suspension prepared by acid treatment of native cellulose, Colloids Surf., A,
142, 75.
Ayala, G., de Oliveira Vercik, L.C., Villa Menezes, T.A., Vercik, A. (2012) Simple and
Green Method for Synthesis of Ag and Au Nanoparticles using Biopolymers and Sugars
as Reducing Agent. MRS Proceedings, 1386, mrsf11-1386-d11-03.
Brancato, A.A. (2008) Effect of Progressive Recycling on Cellulose Fiber Surface
Properties, Thesis, Georgia Institute of Technology, UMI Number: 3345944, section
2.4.2.5, 30-31.
Chen, H. Biotechnology of Lignocellulose: Theory and Practice. Springer, London,
England, 2014. ISBN 9400768982, 9789400768987.
Chundawat S.P.S., Donohoe B.S., da Costa S.L., Elder T., Agarwal U.P., Lu F. (2011)
Multi-scale visualization and characterization of lignocellulosic plant cell wall
deconstruction during thermochemical pretreatment, Energ Environ Sci., 4, 97384.
Dong, X.M., Revol, J-F., Gray D.G. (1998) Effect of microcrystallite preparation
conditions on the formation of colloid crystals of cellulose, 5 (1), 19-32.
Dong, S., Roman, M. (2007) Fluorescently Labeled Cellulose Nanocrystals for
Bioimaging Applications, J Am Chem Soc., 129 (45), 13810-13811.

19

Fahlen, J., Salmen, L. (2002) On the lamellar structure of the tracheid cell wall, Plant
Biol., 4 (3), 339-345.
Fan, J., Liu, J.F., He, J.H. (2008) Hierarchy of Wool Fibers and Fractal Dimensions, Int J
Nonlin Sci Num., 9 (3), 293-296.
Feng, Q.L., Wu, J., Chen, G.Q., Cui, F.Z., Kim, T.N., Kim, J.O. (2000) A mechanistic
study of the antibacterial effect of silver ions on Escherichia coli and Staphylococcus
aureus, J Biomed Mater Res., 52 (4), 662-8.
Fernandes, A.N., Thomas, L.H., Altaner, C.M., Callow, P., Forsyth, V.T., Apperley, D.C.,
Kennedy, C.J., Jarvish, M.C. (2011) Nanostructure of cellulose microfibrils in spruce
wood, Proceedings of the National Academy of Sciences of the United States of America,
108 (47), E1195-E1203.
Habibi, Y., Lucia, L.A., Rojas, O.J. (2010) Cellulose Nanocrystals: Chemistry, SelfAssembly, and Applications, Chem Rev, 110 (6), 3479-3500.
Hasani, M., Cranston, E.D., Westman, G., Gray, D.G. (2008) Cationic surface
functionalization of cellulose nanocrystals, Soft Matter, 4, 2238-2244.
Hubbe, M.A., Venditti, R.A., Rojas, O.J. (2007) How fibers change in use, recycling,
BioResources, 2 (4), 739-788.
Jayme, G. (1944) Mikro-Quellungsmessungen an Zellstoffen. Wochenbl Papierfabr, 6,
187-194.
Klemm, D., Philipp, B., Heinze T., Heinze, U., Wagenknecht, W., Comprehensive
Cellulose Chemistry, 1, Wiley-Verlag Weinheim, 1998.
Krssig, H.A. Cellulose - Structure, Accessibility and Reactivity, Yverdon, Gordon and
Breach Publishers, Switzerland, 1993.
Mutwil, M., Debolt, S., Persson, S. (2008) Cellulose synthesis: a complex complex, Curr
Opin Plant biol, 11(3), 252-257.
Nemati, M., Seyyedmohammadi, N., Samariha, A., Shakouri M.J. (2011) Studying the
Effect of Chemical Additives on Strength Properties of Recycled Paper, J Basic Appl Sci
Res, 1 (11), 2314-2318.
Oke, I. (2010) Nanoscience in nature: cellulose nanocrystals, Studies by Undergraduate
Researchers at Guelph, 3 (2), 77-80.
Ranby, B.G. (1949) Aqueous Colloidal Solutions of Cellulose Micelles, Acta Chemica
Scandinavica, 11, 649-650.

20

Ranby, B.G. (1951) The colloidal properties of cellulose micelles. Discussions Faraday
Soc, 11, 158-164.
Raveendran, P., Fu, J., Wallen, S.L. (2003) Completely "Green" Synthesis and
Stabilization of Metal Nanoparticles, J Am Chem Soc, 125, 13940-13941.
Roman, M., Winter, W.T. (2004) Nanocomposites of cellulose acetate butyrate reinforced
with cellulose nanocrystals, Biomacromolecules, 5 (5), 1671-1677.
Rder, T., Sixta, H., (2004) Thermal treatment of cellulose pulps and its influence to
cellulose reactivity, Lenzinger Berichte, 83, 79-83.
Samir, M., Alloin, F., Sanchez, J.Y., El Kissi, N., Dufresne, A. (2004) Preparation of
cellulose whiskers reinforced nanocomposites from an organic medium suspension,
Macromolecules, 37 (4), 1386-1393.
Sharma, V.K., Yngard, R.A., Lin, Y. (2009) Silver nanoparticles: Green synthesis and
their antimicrobial activities Adv Colloids Interface Sci,145, 83-96.
Somerville, C. (2006) Cellulose Synthesis in Higher Plants, Annu Rev Cell Dev Biol, 22,
53-78.
Travan, A., Pelillo, C., Donati, I., Marsich, E., Benincasa, M., Scarpa, T., Semeraro, S.,
Turco, G., Gennaro, R., Paoletti, S. (2009) Non-cytotoxic silver nanoparticlepolysaccharide nanocomposites with antimicrobial activity, Biomacromolecules, 10 (6),
1429-1435.
Wang, R., Fu, Y., Qin, M. (2014) Homogeneous Acylation and Regioselectivity of
Cellulose with 2-Chloro-2-Phenylacetyl Chloride in Ionic Liquid, BioResources, 9 (3),
5134-5146.

21

Chapter 2:
Superhydrophobic foam-like cellulose made of hydrophobized cellulose fibres
2.1. Abstract
Wood (kraft) pulp was first dried into a low-density foam-like material by solventexchange with anhydrous ethanol. X-Ray tomography showed that, while pulp fibres are
flat and resemble ribbons when dried from water, those dried from ethanol are quasitubular, inferring that capillary forces derived from a low surface tension solvent are not
strong enough to promote fibre lumen collapse, contrary to what happens in water. When
the resulting solid foam-like pulp was then subjected to a vapour phase reaction with
trichloromethylsilane (TCMS) a silicon based polymeric coating was created on the
surface of the fibres, and the totality of the hydroxyl groups (-OH) on the external surface
of cellulose fibres and the internal surface of macropores in the fibre wall became
silylated, whereas the surface of the mesopores was inaccessible to TCMS. The novelty
lies in the ability to modify both the external surface and the internal micropore structure
of cellulose fibres from 50 to 100 % silane coverage, which results in a novel
superhydrophobic material, with an apparent contact angle of approximately 150. This is
the first time cellulose is rendered hydrophobic both internally and externally. We refer to
the resulting foam as Cellufoam.

22

2.2. Introduction
For the past many decades, environmental concerns have led to a greater emphasis on
renewability, biodegradability, and eco-friendliness (Park et al., 2004). A biopolymer that
meets the criteria as a green material is cellulose. As one of the most abundant natural
resources in the world, cellulose can be used in its native state or processed to afford
countless applications (Chen et al., 2007).
From the point of view of advanced materials, celluloses hydrophilic character is unique
and can be modified for a number of different uses. In the case of foams and building
materials, rendering wood pulp hydrophobic has been greatly sought after and many
hydrophobization techniques have been investigated. Most of those attempts failed to
provide a durable hydrophobic character to cellulose products, and the reasons for this are
discussed in this work.

Fig. 2.1. Schematic illustration of two cellulose fibres crossing each other perpendicularly
with water and ethanol represented by blue areas: drying from water (top) causes
shrinkage of the fibres due primarily to collapse of the lumen, while drying from ethanol
(bottom) preserves the open fibre structure.

23

It is known from the literature that as cellulose fibres dry, strong attractive capillary
forces related to the surface tension of water cause a collapse of the internal opening
called the lumen (Tejado & van de Ven, 2010; Huang et al., 2011; Tejado & van de Ven,
2011), especially for delignified fibres, and closes the pores present in the cell wall. This
effect makes the interior portions of the fibres inaccessible to anything but water and to
some extent ions or molecules dissolved within it. This non-reversible fibre collapse is
typically referred to as hornification. In order to obtain dry pulp without fibre collapse,
the water content of never-dried pulp has to be substituted with a low surface tension
fluid (e.g. ethanol) by means of solvent-exchange (Li et al., 2013; Uetani & Yano, 2012).
It is known that solvent-exchanged kraft pulps maintain their internal porous structure,
which consists of mesopores with pore sizes of a few nanometers and surface areas in the
range 200300 m2 /g and macropores with pore sizes of around 75 nm and surface areas
of about 10 m2 /g (Alince, 1975; Alince & van de Ven, 1997). When solvent-exchanged
pulp is allowed to dry, the resulting product is a low-density cellulose foam-like material
(Fig. 2.1). In this work, foam-like pulp has been obtained by drying the original material
from, ethanol and several surfactant solutions in water just above critical micelle
concentration (cmc), all of them showing surface tensions in the order of one-third that of
water.
Aside from the foam-like material reported here, which relies on varying the surface
tension of the solvent before drying, aerogels have also been investigated for a long time
(Jin et al., 2011). Cellulose aerogels avoid the problems related to fibre collapse and
aggregation upon water-removal by either freeze-drying or supercritical drying.

24

Nevertheless, the resulting low-density and highly porous solid is brittle and cannot be
made as easily as the material investigated in this study (Wu et al., 2012).
Based on the combination of old knowledge and recent findings, a new approach for
cellulose hydrophobization (surface polymerization via a gas-phase reaction over dried
but non-collapsed pulp) has been developed which, for the first time, results in totally
hydrophobic and durable cellulose fibres. The macropores in the fibre wall, around 75 nm
in size, are sufficiently large for trichloromethylsilane (TCMS) gas molecules to freely
penetrate. From the data shown below, it appears that TCMS does not penetrate the
mesopores, at least not on the time scale of the experiments due to polymerization of
TCMS into siloxanes that obstruct the pores. As a consequence, vapour deposition
reaction with TCMS allows the production of a superhydrophobic material which we call
Cellufoam. In the case of an aprotic solvent-based reaction with TCMS, the results
showed poor yield and was not investigated further. Although the use of TCMS to render
cellulose hydrophobic is not a novel reaction (Li et al. 2007, 2008; Andresen et al., 2006;
Chinga-Carrasco et al., 2012), its application on non-collapsed fibres is a simple and
novel approach that leads to remarkable new cellulose products, as will be shown in this
and upcoming papers. The reaction between cellulose and TCMS also produces a byproduct not often discussed, but that is detrimental to the overall quality of the
Cellufoam: the production of hydrogen chloride gas, which readily adsorbs to pulp,
making its removal a challenge. This issue has been resolved by blowing air onto the
product immediately following vapour deposition.

25

2.3. Experimental Section

2.3.1. Materials
Beaten hardwood kraft pulp (bHKP) was provided by FPInnovations and softwood kraft
pulp (Q90) was provided by Domtar. The TCMS was purchased from Aldrich. Anhydrous
ethanol was purchased from Commercial Alcohols. Surfactants Triton XF5 and Dowfax
2A1 were obtained from Dow Chemical. Sulfonated Kraft Lignin (SKL) was obtained
from Westvaco. Papermaking machineries include the British Standard Disintegrator
(Noram XRCC1351), the British Automatic Press (Noram XRCC1435) and the British
handsheet maker (Noram 80063), all of which fulfil TAPPI standards.
2.3.2. Preparation of cellulose foam using low surface tension solvent
A sample of never-dried wood pulp (typically about 10 g) was placed in a plastic net of
500 m pore size, which allowed excess water to drain. The damp sample was resuspended in anhydrous ethanol and stirred for 24 h. The ethanol-soaked sample was
drained in the same net and re-suspended again in anhydrous ethanol for 24 h with
constant stirring. This process was repeated three more times until the amount of water
remaining is negligible (typically <0.5 wt%). The ethanol was drained and the
ethanolsoaked pulp was dried at 70C in an oven for 48 h. The resulting material is
moulded by the shape of the drying vessel, cylindrical in the case of a beaker, and has the
consistency of solid foam. A similar process was repeated where aqueous surfactant
solutions were used instead of ethanol. The mixture of wood pulp and this solution was
shaken to the point of foaming and, after draining and repeating a second time, the wet
pulp was put into the oven at 80C for another 24 h. The resulting material was, again,
moulded by the shape of the drying vessel just as before. This part of the experiment was

26

a proof of principle that demonstrated the feasibility of using surfactants instead of

ethanol. Chemical vapour deposition and all subsequent steps were performed using
ethanol-dried pulp.

Fig. 2.2. Experimental set-up for the reaction of cellulose with TCMS. The reaction
vessel and TCMS were preheated in a 65C oven. A mesh basket containing the pulp
sample was introduced and held firmly in place by a rubber stopper inside the vessel. The
sample was reintroduced into the oven and the gas-phase TCMS was made to react with
the sample
2.3.3. Chemical vapour deposition of trichloromethylsilane (TCMS)
The plastic net was sewn into a cylindrical basket with thread and needle. The cellulose
foam mass was broken into smaller fragments and placed into the cylindrical basket that
was subsequently hung from an Erlenmeyer flask top edge resulting in the sample being
suspended (Fig. 2.2). Liquid TCMS (12 mmol/ gcellulose) had been previously poured into
the flask, which was then sealed with a glass stopper and placed into the oven at 65C for
a period of 1-15 min. The resulting superhydrophobic pulp was first blown with air and
then rinsed with a generous amount of water to remove excess reagent and hydrogen

27

chloride produced during the reaction. The fragments of Cellufoam were re-dispersed in
ethanol and dried in the oven at 80C for 48 h to yield one single mass, which once more,
has been moulded by the shape of the drying vessel.
2.3.4. Analyses
The size distribution analysis of individual fibres was performed using the HiRes fiber
quality analyzer (FQA) (OpTest Equipment Inc.). The contact angle goniometry
measurements were acquired using the Contact Angle System OCA (Dataphysics). The
3D images of individual fibres were obtained by X-ray tomography using the 1172 Micro
CT (Skyscan). The surface tension measurements were carried out using the KSV Sigma
70 Tensiometer (KSV Instruments). The surface morphology of fibres was examined by
field emission high resolution scanning electron microscopy FE-SEM (S- 4700 Hitachi,
Tokyo, Japan). The fibres were pressed onto a double-sided tape adhered to a sample
holder surface and sputtered with gold and palladium for 2 min. Imaging was done with a
dispersive spectrometer. The applied accelerating voltage and current were 20 kV and 10
A, respectively.
2.4. Results and Discussion
A cellulose fibre is known to collapse (i.e. shrink by closing its lumen and pores) when it
is dried from a water suspension, acquiring a tough appearance and reducing its ability to
reabsorb water, in what has been traditionally called hornification(Jayme, 1944;
Minor, 1994). This behaviour has been avoided in the present work by using alternative
solvents of low surface tension and also decreasing the surface tension of water with
surfactants. As a result, foam-like materials made entirely of cellulose fibres have been

28

obtained, showing an apparent density in the order of 0.060.10 g/cm3 (Fig. 2.3) (~10-20
times lower than water-dried, depending on the procedure). The density was determined
by measuring the volume of the cylindrical foam pulp and by weighing the sample.

Fig. 2.3. (Left) Wilhelmy balance setup and equation used, where is the surface tension;
(right) a table showing the apparent foam density of wood pulp dried from various
solvents and their surface tension above the critical micelle concentration (cmc). SKL
(cmc = 0.8 wt%), Dowfax 2A1 (cmc = 0.007 wt%), and Triton X45 (cmc = 0.0136 wt%)
As shown by X-ray tomography (Fig. 2.4), when a sample of bHKP wood fibres dries
from water, its lumen is collapsed. This is demonstrated by the narrow aspect ratio of the
fibre cross-section. When dried from ethanol, the fibre cross-section suggests an open
lumen structure, roughly the same as the dimensions of wet fibres in water measured by
FQA (cf. Fig. 2.4c). It can be seen that, when dried from water, the fibres possess a
ribbon-like shape but, when dried from ethanol, they possess a quasi-tubular shape. This
can be explained by the fact that ethanol has a much lower surface tension value than
water (Azizian & Hemmati, 2003; Vazquez et al., 1995; Gunde et al., 1992) and is unable
to collapse the lumen by capillary forces. In an entangled network of fibres, lumen
collapse leads to flat ribbons, which, when in contact with other ribbons, result in a larger

29

binding area than noncollapsed fibres (van de Ven, 2008), leading to a more compact
structure. In contrast, in the absence of lumen collapse, the structure is more open.

Fig. 2.4. 3D images of individual fibres obtained with Skyscan 1172 Micro CT X-Ray
tomography: a) after being dried from water and b) from a low surface tension solvent
(ethanol). c) Fibre size distribution of bHKP fiber in water before drying obtained by
using HiRes fiber quality analyzer (FQA)

Fig. 2.5. SEM images of bHKPcellulose fibers of a dried from water, b dried from
ethanol, c cross-sectional view dried from water and d cross-sectional view dried from
ethanol

30

To further support the X-ray tomography data, SEM images (Fig. 2.5) also suggest flatter
fibres that pack more densely when dried in water compared to ethanol-dried hand sheets.
The lumen opening appears to be narrower in SEM images due to pressing and metal
coating when sputtered with gold and palladium. Nevertheless, the cross-sectional view
does point to an open lumen structure when dried from ethanol.
As shown in the table in Fig. 2.3, the surface tension of the surfactant solution is higher
than that of ethanol. However, for cost considerations, 0.01 % surfactant solutions were
prepared to replace ethanol. During drying, the surface tension reaches the minimum (i.e.
just above the cmc) (Fig. 2.6). Using these replacements, the apparent density was
lowered (0.10 g/cm3 ) to nearly the same value as ethanol (0.06 g/cm3 ) (Fig. 2.3).
However, drying from organic solvents still yield lower density cellufoam.

Fig. 2.6. A plot of the surface tension with respect to surfactant concentration; in this
case, Triton X45 is used and has a known surface tension of 29 mN/m

31

When wood pulp dries from a surfactant solution, again a foam-like pulp is obtained.
SEM imaging (Fig. 2.7) indicates that, indeed, the lumen is much narrower and, in some
cases, nearly closed. The reason for which the pulp displays such a low-density is its
overall packing. Although the individual fibres have partly collapsed lumens, the fibre
network is much looser than when the pulp is dried from water. It can be assumed that,
with surfactants, the surface tension is simply too high to prevent complete lumen
collapse.

Fig. 2.7. SEM images of Q90 nonbeaten cellulose fibers a) dried from water, b) dried
from anionic surfactant, c) cross-sectional view dried from water and d) cross-sectional
view dried from anionic surfactant
Trichloromethylsilane (TCMS) is a hygroscopic compound that almost instantly reacts
with air moisture and hydroxyl (-OH) groups to form siloxane and hydrogen chloride gas.
The hydrogen chloride gas hydrolyses in the presence of moisture to afford concentrated
hydrochloric acid (Tripp & Hair, 1993; Artus & Seeger, 2012). Removal of HCl has been

32

the challenging step in performing the proposed reaction. After several modifications to
the procedure, the best results were obtained from a time-dependent chemical vapour
deposition followed by blowing air toward the sample and lastly, rinsing the hydrophobic
pulp with a generous amount of deionised water. Without adequate removal of
hydrochloric acid, the white pulp burns in the presence of the strong acid and becomes
yellow in colour. A reaction time of 1 min was deemed sufficient for achieving
superhydrophobicity based solely on apparent contact angle measurements, because
increasing the exposure time does not increase the contact angle. Furthermore, with a
longer reaction time, polymerization of siloxane into a multilayer can congest the internal
structure of the fiber. An important feature of this process is that, although reaction times
of 1-15 min yield different degrees of silane coverage, a critical amount of coverage is
reached at just 1 min reaction time to achieve superhydrophobicity. Reaction times above
15 min result in yellowing and disintegration of pulp. Despite the fact that full coverage
cannot be reached, partial coverage of the external surface and the internal micropore
surface has been achieved.

Fig. 2.8. TCMS adsorption as a function of reaction time; also shown is the coverage
assuming a surface area of 10 m2/g of kraft pulp (Alince & van de Ven, 1997)

33

As shown in Fig. 2.8, the percent of TCMS coverage from 50 to 100 % of external
surface and the internal micropore surface of cellulose fibers. To determine percent
coverage, it is assumed that TCMS molecules cover an area of 10 2 (specific adsorption
of 2 mg/ m2) and the combined surface area of the external surface and the macropores is
about 10 m2/g for kraft pulp (Alince & van de Ven, 1997). By measuring the adsorption
of TCMS as a function of reaction time (assuming that TCMS is the only contributor to
the weight gained), the reaction capacity should be approximately 20 mg/m2. It is fair to
assume, in our work, that after 15 min of reaction, that full coverage of the macropores
has been achieved.

Fig. 2.9. Water droplet on the surface of Cellufoam captured by a CCD camera
A material is classified as hydrophilic when its contact angle with a drop of water is
inferior to 90 and as hydrophobic when the contact angle is 90 (Chen et al. 2013;
Drelich et al. 1996); in the latter case, superhydrophobicity is defined by a contact angle
>150 (Blossey, 2003; Lafuma & Qur, 2003; Wang et al., 2010). Cellulose, in its
natural form, has a contact angle of approximately 30 (Jonoobi et al., 2012; Erbil, 1997)
at the moment of initial contact. Due to the porosity and roughness of cellulose sheets,

34

water drops eventually get absorbed into the material. A replacement for commercial
plastics such as cellulose acetate, on the hand, has a contact angle of 55 (Erbil, 1997)
along with reduced porosity. In nature, certain plants and insects have evolved to produce
superhydrophobic coating at the surface of their anatomy. For example, the Indian cress
leaf (Otten & Herminghaus, 2004) has a contact angle of nearly 180, which is the
maximum degree of hydrophobicity. The Cellufoam described in this work has a contact
angle of roughly 150 at equilibrium, which marks it as a superhydrophobic material
(Fig. 2.9). Furthermore, the water drop evaporates before getting absorbed into the pores
of the material. The aforementioned contact angle values are summarized in Table 2.1. It
is important to note that a thorough study of surface energy in relation to contact angle
would require the analysis of advancing and receding contact angles for water droplets at
an incline or by using the pendant drop test method. Surface energy analysis was not
performed to ascertain the superhydrophobicity and stability of the material.
Table 2.1. Contact angles of various materials

The advantage of Cellufoam is the ease of production compared to other

superhydrophobic cellulosic materials in the literature. Furthermore, only Cellufoam is

35

rendered hydrophobic both inside and out with respect to the fibres. Ogawa et al. (Ogawa
et al., 2007) achieved 162 using 30 by-layers of poly(diallyldimethylammonium
chloride) (PDDA)/TiO2; Gonalves et al. (Gonalves et al., 2008) achieved 147 using
silica beads and 1H,1H,2H,2H-perfluorooctyl triethoxysilane (FOTS) (Table 2.1). These
are all fairly more complex techniques compared to chemical vapour deposition with
TCMS and, still, only the outer surfaces of these materials are rendered hydrophobic.

Fig. 2.10. Unmodified hydrophilic (left) and TCMS-hydrophobized (right) cellulose

fibres in contact with a drop of water. Hydrophilic fibers are drawn into the water drop.
Only a sufficiently hydrophobic fiber remains at the water air interface and shows zero
engulfment
Not only does Cellufoam possess unprecedented properties in bulk, its individual fibres
also possess qualities that have never been reported in the past. As shown by Fig. 2.10
(right), a fibre of the Cellufoam is shown to stick to the surface of a water droplet. This
would not have occurred with a natural fibre or when the hydrophobization reaction is
performed after conforming a structure i.e. paper; instead, partial water repellence and
partial engulfment would be expected in that case. As shown in Fig. 2.10 (left), a native
cellulose fibre will instantly enter the water droplet due to its affinity for water.

36

2.5. Conclusion
In this study, solvent-exchange has been used to dry wood pulp fibres without collapsing
the fibre lumens while minimizing the fibrefibre packing (preventing hornification). The
foam-like material obtained has been successfully rendered superhydrophobic through a
vapour phase reaction with TCMS which, combined, is a whole new approach for
hydrophobization of cellulose fibres on an individual scale and renders each cellulose
fibre hydrophobic inside (internal micropore) and outside (external surface) to reach a
maximum of 100 % silane coverage. The implication of such a material is significant:
low-density insulation and packaging materials can be made this way and research into
making Cellufoam flame-retardant (Gates & Hu, 2010) would allow its introduction into
industries such as electronics, automotive and construction.
2.6. References
Alince, B. (1975) Porosity of swollen solvent-exchanged cellulose and its collapse during
final liquid removal, Colloid Polym Sci, 253, 720-729.
Alince, B., van de Ven, TGM. (1997) Porosity of swollen pulp fibers evaluated by
polymer adsorption, In: Baker CF (ed) Transactions of the 11th fundamental research
symposium, Cambridge, UK. The Fundamentals of paper making materials, vol 2. Pira
International, Surrey, UK, 771-788.
Andresen, M., Johansson, L.S., Tanem, B.S., Stenius, P. (2006) Properties and
characterization of hydrophobized micro-fibrillated cellulose, Cellulose, 13, 665-677.
Artus, G.R.J., Seeger, S. (2012) Scale-up of a reaction chamber for superhydrophobic
coatings based on silicone nanofilaments, Ind Eng Chem Res, 51, 2631-2636.
Azizian, S., Hemmati, M. (2003) Surface tension of binary mixtures of ethanol + ethylene
glycol from 20 to 50C, J Chem Eng Data, 48, 662-663.
Blossey, R. (2003) Self-cleaning surfaces-virtual realities, Nat Mater, 2, 301-306.

37

Chen, X., Burger, C., Wan, F., Zhang, J., Rong, L., Hsiao, B.S., Chu, B., Cai, J., Zhang,
L. (2007) Structure study of cellulose fibers wet-spun from environmentally friendly
NaOH/urea aqueous solutions, Biomacromolecules, 8, 1918-1926.
Chen H., Amirfazli A., Tang T. (2013) Modeling liquid bridge between surfaces with
contact angle hysteresis, Langmuir, 29, 3310-3319.
Chinga-Carrasco, G., Kuznetsova, N., Garaeva, M., Leirset, I., Galiullina, G., Kostochko,
A., Syverud, K. (2012) Bleached and unbleached MFC nanobarriers: properties and
hydrophobisation with hexamethyldisilazane, J Nanoparticle Res, 14, 1280.
Drelich, J., Wilbur, J.L., Miller, J.D., Whitesides, G.M. (1996) Contact angles for liquid
drops at a model heterogeneous surface consisting of alternating and parallel
hydrophobic/hydrophilic strips, Langmuir, 12, 1913-1922.
Erbil, H.Y. (1997) Calculation of spreading pressure of water on cellulosic films from
contact angle data, Turk J Chem, 21, 332-345.
Gates, D., Hu, T. (2010) New class of non-leachable phosphorousbased flame retardants.
US patent: US7,728,091 B2.
Gonalves, G., Marques, P.A.A.P., Trindade, T., Neto, C.P., Gandini, A. (2008)
Superhydrophobic cellulose nanocomposites, J Colloid Interface Sci, 324, 42-46.
Gunde, R., Dawes, M., Hartland, S., Koch, M. (1992) Surface tension of wastewater
samples measured by the drop volume method, Environ Sci Technol, 26, 1036-1040.
Huang, F., Lanouette, R., Law, K.N. (2011) Characterization of jack pine early- and
latewood fibers in thermomechanical pulping, Ind Eng Chem Res, 50, 13396-13402.
Jayme, G. (1944) Mikro-Quellungsmessungen an Zellstoffen, Wochenbl Papierfabr, 6,
187-194.
Jin, H., Kettunen, M., Laiho, A., Pynnonen, H., Paltakari, J., Marmur, A., Ikkala, O., Ras,
R.H.A. (2011) Superhydrophobic and superoleophobic nanocellulose aerogel membranes
as bioinspired cargo carriers on water and oil, Langmuir, 27, 1930-1934.
Jonoobi, M., Mathew, A.P., Abdi, M.M., Makinejad, M.D., Oksman, K. (2012) A
comparison of modified and unmodified cellulose nanofiber reinforced polylactic acid
(PLA) prepared by twin screw extrusion, J Polym Environ, 20, 991-997.
Lafuma, A., Qur, D. (2003) Superhydrophobic states, Nat Mater, 2, 457-460.
Li, S., Xie, H., Zhang, S., Wang, X. (2007) Facile transformation of hydrophilic cellulose
into superhydrophobic cellulose, Chem Commun, 46, 4857-4859.

38

Li, S., Zhang, S., Wang, X. (2008) Fabrication of superhydrophobic cellulose-based

materials through a solution-immersion process, Langmuir, 24, 5585-5590.
Li, F.S., Lively, R.P., Lee, J.S., Koros, W.J. (2013) Aminosilanefunctionalized hollow
fiber sorbents for post-combustion CO2 capture, Ind Eng Chem Res, 52, 8928-8935.
Minor, J.L. (1994) Hornification-its origin and meaning, Prog Pap Recycl, 3, 93-95.
Ogawa, T., Ding, B., Sone, Y., Shiratori, S. (2007) Super-hydrophobic surfaces of layerby-layer structured film-coated electrospun nanofibrous membranes, Nanotechnology, 18,
165607.
Otten, A., Herminghaus, S. (2004) How plants keep dry: a physicists point of view,
Langmuir, 6, 2405-2408.
Park, H., Misra, M., Drzal, L.T., Mohanty, A.K. (2004) Green nanocomposites from
cellulose acetate bioplastic and clay: effect of eco-friendly triethyl citrate plasticizer,
Biomacromolecules, 5, 2281-2288.
Tejado, A., van de Ven, T.G.M. (2010) Why does paper get stronger as it dries, Mater
Today, 13 (9), 54-61.
Tejado, A., van de Ven, T.G.M. (2011) Effect of lumen collapse on the strength of drying
paper, Paper presented at the proceedings of 97th annual PAPTAC meeting, Montreal,
Canada, February 1-3, 9-11.
Tripp, C.P., Hair, M.L. (1993) Chemical attachment of chlorosilanes to silica: a two-step
amine-promoted reaction, J Phys Chem, 97, 5693-5698.
Uetani, K., Yano, H. (2012) Zeta potential time dependence reveals the swelling
dynamics of wood cellulose nanofibrils, Langmuir, 28, 818-827.
van de Ven, T.G.M. (2008) Capillary forces in wet paper, Ind Eng Chem Res, 47, 72507256.
Vazquez, G., Alvarez, E., Navaza, J.M. (1995) Surface tension of alcohol water + water
from 20 to 50C, J Chem Eng Data, 40, 611-614.
Wang, C.F., Chen, W.Y., Cheng, H.Z., Fu, S.L. (2010) Pressure-proof superhydrophobic
films from flexible carbon nanotube/ polymer coatings, J Phys Chem, 114, 15607-15611.
Wu, W., Wang, K., Zhan, M.S. (2012) Preparation and performance of polyimidereinforced clay aerogel composites, Ind Eng Chem Res, 51, 12821-12826.

39

Bridging section between Chapters 2 and 3

In chapter 2, superhydrophobization of cellulose wood pulp with trichloromethylsilane
(TCMS) resulted in a foam-like material of low density and an apparent water drop
contact angle of 150. The low density material, called Cellufoam, resulted from neverdried pulp that has been dried from ethanol by the solvent exchange method in order to
prevent lumen and pore collapse that is caused by the coalescence of cellulose micro- and
nanostructures. It has been found that many low surface tension solvents can also be used
to replicate this work. For industrial consideration, surfactants can be used to reduce
costs. Scanning electron microscopy and X-ray tomography indicate that fibres remained
open after drying from a low surface tension solvent, unlike for water, where fibres
appear to be completely collapsed. Further analysis was performed by varying the
reaction time in order to achieve greater contact angle and to modify the nanostructure of
cellulose. However, additional reaction time resulted in damage to the sample caused by
the formation of hydrochloric acid as a by-product of the reaction.
In chapter 3, the superhydrophobic material, Cellufoam, was studied for its drying
behaviour and its wet web strength (WWS) in order to understand the contribution that
entanglement provides to handsheets. In this work, a new drying property of
superhydrophobic cellulose was explored.

40

Chapter 3:
Hydrophobic cellulose: a material that expands upon drying
3.1. Abstract
A chemically modified hydrophobic cellulose material was wetted by force in vacuo and
allowed to dry under ambient conditions. Most known materials shrink upon drying and
swell upon wetting, a phenomenon known as dry-shrinkage; and thus are characterized
by a dry-shrinkage coefficient either equal or greater than zero. Different from
conventional materials, sheets of hydrophobic cellulose fibres expand upon drying, which
implies that they exhibit dry-expansion. This property is calculated as a negative dryshrinkage coefficient. We are unaware of any other material with this property. Such
sheets can expand to over 500 % in thickness upon drying in the first cycle of use. This
property degrades with each cycle because more hydrophilic areas come in contact with
water as a result of mechanical damage to the material, thus making the sheets less
hydrophobic. With increasing solid content, a decrease in tensile strength is observed,
which is opposite to the conventional trend in wet web strength. A mechanism for the
dry-expansion of this material is being proposed.

41

3.2. Introduction
When materials are placed in a wet environment, swelling occurs; the dimensions of the
material are altered so as to expand in at least one of its dimensions. As water is removed,
dry-shrinkage occurs and is a measure of the decrease in volume as a function of
decreasing moisture content. The dimensional change is quantified by a coefficient ()
that is defined as the dimensional change (%) divided by the change in moisture content
(%) (Marklund and Varna, 2009):
= dimensional change (volume %) / moisture content(%)

(1)

Equation 1 is used to describe two phenomena that are inherently linked. Hygroexpansion
and dry-shrinkage share similarities in that both are properties governed by the
coefficient b (Sampson and Yamamoto 2011). Equation 1 is limited in applicability by the
range of the moisture content, which generally is in a linear region of the thickness versus
moisture content curve (Uesaka and Qi 1994; Nanri and Uesaka 1993). In addition, the
moisture content of samples typically ranges from 0 to 20 % (Pulkkinen et al. 2009a),
which differs from the range observed in this experiment (050 %). In the literature, both
the hygroexpansion and the dry-shrinkage coefficient are always assumed to be positive,
or zero for solid materials that can take up water in their porous structure without a
change in volume. Cellulose is a material that has often been studied for its swelling
properties when exposed to water (Vainio et al., 2007; McMillin, 1973; Uesaka and
Moss, 1997). This natural product is hydrophilic and can take up water with ease into its
fibre structure. The swelling and dry-shrinkage of cellulose depends on many factors such
as fibre geometry (length, width, and wall thickness) and structural factors that result
from the various stages of pulp treatment. Despite thorough research in this field, it has

42

never been proposed that rendering wood pulp hydrophobic would result in dryexpansion. Referring to Eq. 1, this implies that there is an increase in volume with a
decrease in moisture content. In other words, as a wet sheet dries, it grows bigger in
overall size. It has been shown in the past that factors that influence changes most in
hydrophilic cellulose are the polydispersity of fibre size and wall thickness (Pulkkinen et
al., 2009b). In this work, it will be demonstrated that other factors account for this
property in hydrophobic cellulose. It has been reported in a prior publication (Tejado et
al., 2014) that hydrophobic cellulose foam (Cellufoam) of low density can be produced
by the hydrophobization of ethanol-dried cellulose in the vapour phase with
trichloromethylsilane (TCMS). Low-density cellulose foam has an open lumen structure
and the fibres are loosely packed so as to allow for maximum penetration of TCMS into
the fibre microstructure. Given Cellufoams hydrophobicity, when water is forced into its
porous structure, a sheet of this type of pulp can expand upon drying to many times its
original thickness with no apparent change in length or width.
3.3. Experimental Section
3.3.1. Materials
Beaten hardwood kraft pulp (bHKP) and northern softwood kraft pulp (NDSKP) were
provided by FPInnovations and softwood kraft pulp (Q90) was provided by Domtar.
Trichloromethylsilane was purchased from Aldrich. Anhydrous ethanol was purchased
from Commercial Alcohols. Papermaking machineries include the British standard
disintegrator (Noram XRCC1351), the British Automatic Press (Noram XRCC1435) and
the British handsheet maker (Noram 80063), all of which fulfil TAPPI standards. Wet
web strength (WWS) was measured using the TMI LabMaster tensile machine.

43

3.3.2. The preparation of cellulose foam

The preparation of cellulose foam using the chemical vapour deposition of
trichloromethylsilane (TCMS) has been discussed in a previous work (Tejado et al.,
2014). In brief, never-dried cellulose pulp was suspended in water and, through multiple
solvent exchange cycles, was suspended in anhydrous ethanol and allowed to dry as lowdensity foam. This material is rendered hydrophobic by vapour phase reaction with
TCMS. The reaction time and temperature of the hydrophobization have been set to 5
min at 60C for the entirety of the project.
3.3.3. Sheet preparation and observation
The hydrophobic cellulose foam, when dry, floats on water. Therefore, the foam was first
dispersed in anhydrous ethanol and, through multiple solvent exchange cycles with water,
was fully dispersed in and not on the surface of water. The ethanol content of the
suspension after this procedure is negligible. The suspension was then placed in a
Buchner flask sealed with a stopper and subjected to vacuum. This ensured the
penetration of water into the microporous structure. Finally, a 0.03 % w/w suspension of
treated pulp was used to prepare 100 g/m2 paper sheets using a standard British handsheet
machine. A metal template mold, having openings to create eight 25 mm wide and 55 mm
long paper strips, was placed over the metal screen of the handsheet maker. After sheet
formation, the template was removed and the pre-cut sheet was pressed between two
Teflon plates at 350 kPa for 5.5 min. The paper strips were allowed to dry in ambient
conditions (approximately 20C, 50 % humidity) and used to measure their wet web
strength (WWS) in a TMI LabMaster tensile machine at a crosshead rate of 25 mm/min.

44

It is important to note that the temperature and humidity within the room were not
controlled.
3.3.4. Analyses
The contact angle goniometry measurements were acquired using the Contact Angle
System (OCA) (Dataphysics). Using the same contact angle device, a profile of the sheet
was observed and used to measure thickness over time.
3.4. Results and discussion
The cellulose fibres exposed to the air were made hydrophobic by vapour deposition of
TCMS. Hydrophobic handsheets made from any of the three sources of wood pulp (Q90,
bHKP, NDSKP) always expand upon drying regardless of the conditions. This shows that
the sheets prepared from these pulps possess a negative dry-shrinkage coefficient. Based
on measurements of the largest increase in sheet thickness, it was determined that the
handsheets expanded to over 500 % the original thickness upon pressing in the first cycle.
As Fig. 3.1 shows, this was achieved for a sample of softwood kraft pulp (Q90).

45

Fig. 3.1. The dry-expansion of sheets made from 100 % softwood cellulose fibres. A wet
handsheet is pressed and placed onto a glass slide. Images of its cross section were taken
before drying (solid content of 50 %, left) and after drying (solid content of 100 %, right)
As illustrated in Fig. 3.2, the ability of the material to expand upon drying reduces after
each drying-rewetting cycle and one can relate the contact angle (CA) to the dryexpansion property measured for hydrophobic cellulose sheets. The contact angle of the
original material was 150 prior to mechanical treatment and degrades with each cycle
(Fig. 3.2a). The surface properties of the fibres are greatly affected by the wetting process
thus resulting in mild fibrillation or surface damage, which causes the fibres to become
less hydrophobic. As a result, the dry-shrinkage coefficient of the sheets is reduced in
each subsequent cycle.

46

Fig. 3.2. Each subsequent cycle results in progressively thicker wet handsheets due to
flocculation that also causes a decrease in relative expansion when dry; the contact angle
also decreases implying that the material becomes less and less hydrophobic (the Q90
pulp sample used in this figure is a different batch from the Q90 sample of Fig. 3.1)
It has been proposed in the past (Wallstrm et al., 1995) that the positive hygroscopic
expansion of cellulose pulp is a result of shrinking of the fibre walls. Upon drying, the
micro- and nano-structures collapse causing a deformation of the entire fibre that often
results in the closing of the fibre lumen.

Given that Cellufoam is a superhydrophobic material with a contact angle of about 150
when freshly reacted with TCMS and before being mechanically treated, the
microstructures of the fibres do not possess many free hydroxyl groups (Li et al., 2007).
This prevents the collapse of such structures by strong hydrogen bonding interactions. It
is important to note that, although rewetting the fibres restores lumens to their open
structure, hornification (Jayme, 1944; Suty et al., 2012) may occur when fibres are
irreversibly closed. From the results of weighing the sample during the drying process

47

and recording the sheet thickness, a mechanism by which Cellufoam expands upon
drying can be proposed.

Fig. 3.3. The dry-expansion of Cellufoam when drying during the first cycle (a) and solid
content of Cellufoam as a function of drying time (b)

Fig. 3.4. The proposed mechanism for the expansion of Cellufoam upon drying; the
illustration of the lumen and fibre wall is not drawn to scale and fibre cross-sections are
not necessarily tubular
The expansion of the handsheet follows a quasilinear growth in the range of 55-95 %
solid content as shown in Fig. 3.3a, corresponding to a dry-expansion coefficient = -12.
This increase in size accelerates when the sample is almost fully dried. It is postulated

48

that water bridges bind fibres together and hold the lumen nearly closed due to capillary
forces (Fig. 3.4). Lumen opening accounts for between 3 and 5 times the expansion as the
height of a closed lumen is 68 m (twice the wall thickness); when the lumen is fully
open, the fibre height is approximately 1530 m, which corresponds to the diameter of a
non-collapsed fibre (Tejado et al., 2014). This is consistent with the quasi-linear growth
accounting for the major part of the expansion. Furthermore, as the lumens open, more
water surface is exposed to the air thus accelerating the drying process (Figs. 3.3b, 3.4b).
At above 95 % solid content, the water droplets holding the fibres together by capillary
forces break, resulting in an elastic rebound of the fibres, thus causing the sheet to expand
further. At near dryness, a time-dependent fibre rearrangement has been observed to
contribute to the expansion. During the drying process, fibres near the top of the sheet
contribute more to the expansion than those near the bottom which are compressed and
cannot provide the same contribution. Thus, a density gradient is present whereby fibres
higher up provide a greater contribution to the expansion. Although the dry-expansion is
explained as a two-step mechanism, different regions of the sheet will exhibit elastic
rebound at a varied rate; the mechanism is not assumed to be uniform throughout the
sample. The reason why there is no observable contribution in the plane of the sheet (no
change in length or width) is that the sheet was pressed from the top-down. It is therefore
most dense in its thickness. The lumen opening mechanism is first to occur because
compressed lumens are the most elastically deformed structures in the fibre matrix. This
deformation exerts a large force on the bridging droplets within the once tubular
structure, causing these water bridges to break first. For droplets between fibres, there is
also a competition between capillary attraction and elastic fibre deformation. These

49

elastic forces are expected to be less than those in compressed lumens, which explains the
order of the sheet expansion mechanism proposed.

Fig. 3.5. The relationship between the tensile index and solids content for handsheets
made from fibres of various lengths and curl indices: hardwood (HKP) is mainly short
fibres of low curl index and softwood (NDSKP and Q90) are long fibres of high curl
index
The relationship between tensile index and solid content can be seen in Fig. 3.5., which
shows the wet web strength (WWS) of various sheets. The choice of fibres takes into
consideration the relationship between tensile strength and fibre entanglement, though
not explored in greater detail. In the case of wood pulp that has not been modified, a
noticeable increase in tensile strength occurs with drying. This relationship can be
somewhat altered by lowering the basis weight of the handsheet produced. In the case of
hydrophobic cellulose pulp, there is less adhesive interaction between fibres and the
tensile strength approaches near-zero at 100 % solid content. When a high basis weight
softwood pulp with longer fibres (NDSKP) is treated with TCMS and made into a

50

handsheet, the expansion is limited because the fibers are highly entangled, thus
preventing expansion, but maintaining WWS upon drying (Tejado and van de Ven, 2010).

The expanded handsheet made from Q90 softwood is a hydrophobic porous material
where 98 % of the total volume is air (Table 3.1). In this state, the fibres are loosely
bound to each other and individual fibres can easily separate from the dried handsheet.

As mentioned in a previous paper (Tejado et al., 2014), a range of reaction times could
have been chosen to perform the reaction between cellulose and TCMS. So long as the
reaction time is below 15 minutes, there is no risk of damage to the sample due to the
elevated acidity caused by the production of hydrogen chloride gas and air moisture. In
this work, it has been specified that a reaction time of 5 minutes was used. This is
because longer reaction times resulted in samples with reduced expansion or which failed
to expand upon drying. Although not studied in any great detail, it is hypothesized that
higher siloxane coverage would result in lumen congestion and the inability of the fibre
matrix to deform due to a decrease in fibre flexibility. It is important to note,
nevertheless, that no shrinkage of the sample has been observed in drying of Cellufoam
handsheets under any condition.

51

Table 3.1. Volumes occupied by cellulose fibres, water, and air in wet and dry samples

The dimensions of the sample mat were determined by measuring the width and length
with a ruler and the height with a caliper. The volume of cellulose fibres is determined by
dividing the weight of the dry sample by the known density of cellulose fibres d = 1.5
g/cm3 (Hermans et al., 1945).
3.5. Conclusion
The negative dry-shrinkage coefficient of cellulose fibres is due to the interplay between
the elastic deformation of fibres and the attractive capillary forces exerted by water
droplets, i.e. it is an example of an elastocapillary phenomenon. As such, it is likely that
other porous hydrophobic materials with the right elasticity can be engineered to
demonstrate properties similar to that of hydrophobic cellulose.

52

3.6. References
Hermans P.H.; Vermaas J.J.; Vermaas D. (1945) Density of cellulose fibers. II. Density
and refractivity of model filaments, J Polym Sci, 1 (3), 156-161.
Jayme, G. (1944) Micromeasurements of the swelling of pulps, PapierfabrikantWochenblatt Fuer Papierfabrikation, 6, 187-194.
Li, S.; Xie, H.; Zhang, S.; Wang, X. (2007) Facile transformation of hydrophilic cellulose
into superhydrophobic cellulose, Chem Commun, 46, 4857-4859.
Marklund, E.; Varna, J. (2009) Modeling the hygroexpansion of aligned wood fiber
composites, Compos Sci Technol, 69, 1108-1114.
McMillin C.W. (1973) Fibril angle of loblolly pine wood as related to specific gravity,
growth rate, and distance from pith, Wood Sci Technol, 7, 251-255.
Nanri, Y.; Uesaka, T., (1993) Dimensional stability of mechanical pulps: drying shrinkage
and hygroexpansivity, Tappi J, 76, 62-66.
Pulkkinen, I.; Fiskari, J.; Alopaeus, V. (2009a) The effect of hardwood fiber morphology
on the hygroexpansivity of paper, BioResources, 4, 126-141.
Pulkkinen, I.; Fiskari, J.; Alopaeus, V. (2009b) The effect of sample size and shape on the
hygroexpansion coefficient - a study made with advanced methods for hygroexpansion
measurement, Tappsa J, 26-33.
Sampson, W.W.; Yamamoto, J. (2011) The drying shrinkage of cellulosic fibres and
isotropic paper sheets, J Mater Sci, 46 (2), 541-547.
Suty, S.; Petrilakova, K.; Katuscak, S.; Kirschnerova, S.; Jablonsky, M.; Vizarova, K.;
Vrska, M. (2012) Change in the capability of cellulose fibres to retain water during
thermally accelerated ageing of paper, Cellul Chem Technol, 46, 631-635.
Tejado, A.; van de Ven, T.G.M. (2010) Why does paper get stronger as it dries? Mater
Today, 13 (9), 42-49.
Tejado, A; Chen, W.C.; Alam, M.N.; van de Ven, T.G.M. (2014) Superhydrophobic foamlike cellulose made of hydrophobized cellulose fibres, Cellulose, 21, 1735-1743.
Uesaka, T.; Moss, C. (1997) Effects of fiber morphology on hygroexpansivity of paper - a
micromechanics approach, In: Transactions of the fundamental research symposium, vol
1, Cambridge, 663679.
Uesaka, T.; Qi, D. (1994) Hygroexpansivity of paper - effects of fiber-to-fiber bonding, J
Pulp Paper Sci, 20, 175-179.

53

Vainio, A.; Sirvio, M.; Paulapuro, H. (2007) Observation on the microfibril angle of
finnish papermaking fibers. In: International paper physics conference, Gold Coast, 397403.
Wallstrm, L.; Lindberg, K.A.H.; Johansson, I. (1995) Wood surface stabilization, Holz
als Roh-und Werkstoff, 53, 87-92.

54

Bridging section between chapters 3 and 4

The previous two chapters involved the modification of bulk cellulose wood pulp.
Cellulose fibres are hundreds of microns long and tens of microns in diameter. These
fibres are composed of nanofibrils, as explained in the Introduction chapter. It was found
that all accessible hydroxyl (-OH) groups can be modified from the microstructure of
cellulose fibres. However, the weight increase of the modified cellulose pulp suggested
that the nanostructure of the material did not undergo modification. Furthermore,
although the superhydrophobization reaction resulted in an initial contact angle of 150,
with each cycle of use, the contact angle diminished. It was found that unreacted -OH
groups can be exposed by mechanical damage caused by washing, pressing, and drying.
In the next two chapters, the study shifts into the nano- and the molecular scale. With
reduction in size, inaccessible hydroxyl groups become scarce. In chemical reactions,
degree of substitution ranges between negligible and 12 mmol/g. Each increment of 4
mmol/g represents the full conversion of hydroxyl groups from either carbon 2, 3, or 6 in
the case of 1-4 beta-glucans. In chapter 4, we describe how CNC is modified to possess
cationic charge groups, which are shown to have a profound effect on the dimensions of
the nanocrystals. By exposing more hydroxyl groups in aqueous suspension, this form of
nanocellulose can reduce silver cations to silver nanoparticles in the presence of chloride.
This work demonstrates two competing reactions where both silver nanoparticles and
silver chloride nanoparticles can form.

55

Chapter 4:
Cationic Cellulose Nanocrystal Assisted Reduction of Silver (I) to Silver Nanoparticles
4.1. Abstract
Silver nanoparticles (AgNP) in the range of 5-22 nm in size were formed by the reduction
of silver nitrate using cationic cellulose nanocrystals (cCNC). Atomic force microscopy
suggests that the cationization of cellulose nanocrystals results in the formation of soluble
polysaccharide products as well as cCNC of reduced diameter. Elemental analysis with
electron dispersion x-ray spectrometry and x-ray photospectrometry suggest the
formation of AgNP of low chlorine content stabilized by polysaccharides. The metal
conversion yield is 70% based on analysis with inductively coupled plasma atomic
emission spectroscopy. Cationic wood pulp and unmodified CNC (in 0.01M NaCl) do not
provide the same reducing properties found in cCNC and afford only AgCl as the product
of precipitation in this competitive reaction between reduction and precipitation. Based
on our results, the titration of cationic cellulose nanocrystals with silver nitrate is a viable
method to produce silver nanoparticles.

56

4.2. Introduction
The use of silver nanoparticles (AgNP) has evolved from being historically claimed as a
potion for curing human ailments (Woods, 2011) to being used as an antimicrobial agent
in paint (Kumar et al., 2008; Kaiser et al., 2013) and as a disinfectant for medical use
(Roe et al., 2008; Prabhu & Poulose, 2012). Different morphologies of silver result in
changes to the surface plasmon resonance of the nanoparticles and thus its colour also
changes in solution and on various substrates (Mock, 2002). One type of material being
studied increasingly for its ability to mediate the growth of AgNP is cellulose (Ifuku et
al., 2009). Despite the presence of reducing groups at the end of cellulose chains, in many
publications, a strong reducing agent such as sodium borohydride (Lokanathan et al.,
2014) is used to produce Ag (0). As the science progressed, milder reducing and
stabilizing agents such as ascorbic acid were used to prevent aggregation of single crystal
small particles into larger heterogeneous clusters (Umer et al., 2014). Cellulose-based
reactions were also developed and optimized by using environmentally friendly methods
of producing AgNP: using food-based reducing agents and using oxidized cellulose in
one-pot reactions (Ifuku et al., 2009; Christensen, 2011). More recently, it was observed
that cellulose, under certain forms, is capable of reducing metal cations (Trombino et al.,
2012). Although the reducing ends of cellulose are often claimed to be responsible for
metal ion reduction, non-reducing sugars have also been documented to reduce silver
ions (Filippo et al., 2010). It has been explained that when cellulose hydroxyl groups
oxidize, the reduction of Ag (I) to Ag(0) occurs (Garza-Navarro et al., 2013). As
illustrated in Fig. 4.1, we demonstrate a way of producing AgNP using cationic cellulose

57

nanocrystals (cCNC) and silver nitrate (AgNO3) without any added reducing or
stabilizing agent.

Fig. 4.1. The reaction scheme of CNC with QUAB 151 (a quaternary ammonium
epoxide) to produce cCNC and AgNP
4.3. Materials and methods
4.3.1. Materials
The source of cellulose for this work is Canadian softwood kraft pulp provided by
FPInnovations. Cellulose nanocrystals (CNC) were isolated by acid hydrolysis using
sulphuric acid (ACP Chemicals). The AgNO3 was provided by Fluka Analytical and
glycidyltrimethylammonium chloride (QUAB 151) was provided by Sigma Aldrich.
AgNP standards (10 nm and 40 nm) were provided by Sigma Aldrich.
4.3.2. Methods and characterization techniques
Conductivity was monitored during AgNO3 titration by using the Metrohm Titrando
Autotitrator. Atomic force microscopy (AFM) was performed on the Nanoscope IIIa
MultiMode with Extender (Veeco Metrology Group, Santa Barbara, CA) in tapping
mode, Nanoscope Analysis 1.4 was used to process the AFM images. Transmission
electron microscopy (TEM) was performed on the Tecnai 12 electron microscope at 120
kV. A separate analysis was performed by TEM using the Tecnai G2 F20 Cryo-STEM to
58

provide electron diffraction patterns and elemental analysis. X-ray diffraction (XRD) was
performed on the Bruker 2. X-ray photoelectron spectroscopy (XPS) was performed
using the Thermo Scientific K-Alpha X-ray Photoelectron Spectrometer. Induced coupled
plasma atomic emission spectroscopy (ICP-AES) was performed using the Thermo ICAP
6500 Dual View spectrometer.
4.3.3. Preparation of cationic cellulose nanocrystals (cCNC)
The wood pulp was treated with 64% sulphuric acid (1:10 weight ratio) at 45C for 1
hour to produce CNC and placed in dialysis bags. The dialysis bags were placed in
deionized water and mixed-bed ion-exchange resin was added. The mixture was allowed
to stir for 1 week while the water was changed every day. Glycidyltrimethylammonium
chloride (QUAB 151) was added at 12 mmol/g of purified CNC and the mixture was
covered with aluminium foil to prevent exposure to light. This mixture was allowed to
stir at 62C for 24 hours to produce cCNC and, again, dialyzed to remove excess reagent
and molecular products. In the case of wood pulp, no acid hydrolysis was performed and
the pulp was suspended in water containing QUAB 151 (see Fig. 4.1) at 12 mmol/g of
cellulose. The reaction was identical to the one performed on CNC.
4.3.4. AgNP formation
At a CNC suspension concentration of 0.3% w/w, the conductivity of the mixture was
monitored using the Metrohm Titrando Autotitrator while AgNO3 (0.01 M) was being
added dropwise. The end-point was marked at the inflexion point where the slope
representing the change in conductivity just turns positive

59

Fig. 4.2. Conductometric titration curve of AgNO3 with cCNC; an equivalence point at
7.1 mL (using a 0.01 M standard), cCNC (50 mL, 0.3% w/w suspension) was calculated
to contain 0.47 mmol/g of cationic charge groups
4.3.5. Analysis
The crystallinity of cellulose at each step as well as the crystallinity of the AgNP were
first measured by XRD. The cellulose and the AgNP were deposited on carbon film
copper grids without staining and imaged by using TEM. The cCNC structure was
imaged by AFM on freshly cleaved mica because the particle thickness was too low to
provide good contrast by electron microscopy. Electron diffraction was used in
conjunction with the TEM instrument to analyse AgNP. To confirm the elemental makeup
of the silver nanoparticles without cellulose, the suspended samples were first passed
through a 0.2 m mesh nylon membrane syringe filter and then analysed by ICP-AES,
then compared to two separate AgNP standards (10 nm and 40 nm). The sample from the
filtrate was then heated to dryness inside a vacuum furnace for analysis by XPS at 10-9
bar. Although not discussed in this chapter, many of the analytical techniques presented
herein are explained in detail in Appedix I.

60

4.4. Results and discussion

The reactant, cellulose nanocrystals (CNC), is a rod-shaped material derived from plant
cellulose by acid hydrolysis. From AFM (Fig. 4.3), the height profile of these rods ranges
from 6 to 15 nm, which is in good agreement with the results reported in literature (Aulin
et al., 2009). Once the CNC reacts with QUAB 151, a decrease in height to
approximately 1 nm is observed. The two samples were prepared for AFM on freshly
cleaved mica at equal concentration for this AFM analysis. It is conjectured that cationic
cellulose chains are solubilized causing fragments of cCNC to come apart, which results
in many detached layers of the original particles at only a fraction of the height of the
original material. The reduced height of cCNC indicates that indeed polysaccharides have
been solubilized during the reaction process with QUAB 151.

Fig. 4.3. AFM tapping mode image of unmodified CNC (left) and cCNC (right); the
height profile distribution of the two samples is plotted as two separate histograms
(bottom)

61

The reduction of metal cations to metal nanoparticles is indicated by a lowering of Ag (I)

ion concentration in an aqueous solution. By using a conductivity probe, the titration of
AgNO3 into a suspension of cCNC (0.3% w/w) shows a decrease in conductivity
depending on colloidal stability and the amount of cationic reagents and CNC used. The
end-point is signalled by an inflexion point where the slope of the conductivity becomes
positive. The end of the titration was observed between 0.4-0.5 mmol of silver nitrate per
gram of cCNC. It has also been observed that, if the AgNO3 concentration is too high or
if the charge group concentration is too low, then a lowering of the conductance will not
be observed during titration. The titration curve would be flat and eventually start to rise
past the equivalence point.
Electron microscopy was performed without staining at 120 kV. Unlike the instrument
used for EDX (section 3.6), this electron microscope operates at lower voltage, which is
more suitable for viewing a metal-organic hybrid system (Kaushik et al., 2014). With this
setup, good contrast was achieved for silver nanoparticles, but cellulose was poorly
resolved. As for the size of the nanoparticles, a computerized count of the dark spheres
shows that no AgNP particle is larger than 22 nm (Fig. 4.4). Based on numerous studies,
smaller AgNPs of approximately 10 nm are more effective at preventing the growth of
microbes (Jeong et al., 2014; Lu et al., 2013). Based on various forms of analysis, the
particles formed have been identified as silver nanoparticles.

62

Fig. 4.4. The particle size distribution of AgNP from over 10 TEM images; (right) an
example of TEM imaging where AgNP is depicted as dark spheres
An analysis of powder samples by means of XRD in Fig. 4.5 reveals that cationic
cellulose nanocrystals (cCNC) can be used to reduce silver (I) to silver metal. The XRD
patterns (Fig. 4.5 - left) clearly indicate that cationic wood pulp and unmodified CNC in
0.01 M NaCl all produce AgCl due to the reaction between AgNO3 and chlorine anions.
The precipitation of AgNO3 with NaCl does not serve as a reaction comparison with
cCNC and AgNO3. The purpose of performing the precipitation reaction is to analyze
AgNP on cCNC all the while analyzing AgCl on CNC as an exercise in sample analysis.
The epoxy group from the cationic reagent reacts with the hydroxyl groups of cellulose to
form a covalent bond; cellulose peaks are seen at 17 to 24 representing the 110 and 200
reflection planes of cellulose. AgCl peaks at 27, 32, 46, 54, 57, 67, and 77 are seen
for cationic wood pulp and unmodified CNC, which are indexed based on the facecentered cubic structure of AgCl (Fig. 5.5, curves 1-2). It is important to note that a small
amount of AgNP peaks can also be seen for the CNC sample spectrum. For the sample of
cCNC titrated with AgNO3, peaks at 38, 44, 65, and 78 suggest the presence of AgNP
Bragg reflections, which are indexed based on the face-centered cubic structure of silver
(Fig. 5.5, curve 3). Additional information on reflection planes and XRD as an analytical
technique can be found in Appendix I. It can be postulated that the formation of AgCl

63

and the formation of AgNP are competitive reactions governed by the reducing agent in a
reaction mixture containing silver and chlorine ions. In the case of modified charged
particles such as cCNC, favourable interaction with water along with the presence of
dissolved carbohydrates, as concluded from a decrease in diameter observed by AFM,
permit a better reduction of Ag (I) to Ag (0). Along with the XRD data, electron
diffraction also shows the diffraction pattern of AgNP (Fig. 5.5 - right).

Fig. 4.5. (Left) The XRD patterns of: 1) cationic wood pulp and AgCl 2) CNC and AgCl
3) cCNC with AgNP; (Right) The 220 reflection of AgNP is shown for sample 3
After removing as much organic material as possible, the concentration of silver was
determined by ICP-AES. The larger cCNC rods had to be removed in order to prevent
obstruction and damage to the instrument in the presence of nitric acid, which also means
that an unknown amount of silver nanoparticles adsorbed to cellulose would be inevitably
lost during sample preparation. The yield reported by ICP-AES is approximately 70%
based on a determination of the silver concentration, though this value may not be
indicative of the full conversion of AgNO3 to AgNP given the loss in yield during the
removal of cCNC.

64

Elemental analysis of AgNP by EDX (Fig. 4.6) and by XPS (Fig. 4.7) indicates a
noticeable amount of silver, trace amounts of chlorine, and a layer of organic material
around the nanoparticles. This organic material possesses C-C bonds as well as C-O
bonds, which suggests the presence of carbohydrates at the surface of silver. This is
consistent with the hypothesis that each AgNP is stabilized by the dissolved cationic
cellulose chains. Given that chloride is the counter-ion for the quaternary amine
functional group, it is likely that this halogen cannot be easily removed during dialysis.
EDX was performed by using TEM at 200 kV directing the electron beam to one
particular area of the sample grid. For the analysis, several dark spherical particles with
diameters ranging from 10 to 15 nm assumed to be silver nanoparticles were aimed at.
Using this method, very little chlorine was detected.
Pairing the latter with analysis by XPS, which is much more sensitive, allows for
additional analytical validity of the elemental makeup of the samples. However, for XPS,
entire powder regions need to be selected to perform a run, which means that CNC
particles would be inevitably analysed as well. Although much of the cellulose was
removed by syringe filtration (0.2m pore-size), a sizeable amount still remains in the
sample. Nevertheless, the two chlorine peaks are shown as one poorly resolved bump that
is at the limit of detection. As for the carbon peaks at 285-287 eV, the data was collected
after adventitious carbon charge correction was performed. This technique is further
discussed in Appendix I.

65

Fig. 4.6. The EDX elemental analysis spectrum of silver nanoparticles viewed using TEM
at 200kV; the Ag peaks are prominent, but the Cl peaks are weak and slightly overlapped
by secondary Ag peaks; this technique is paired with XPS for a complete analysis

Fig. 4.7. The XPS spectra of AgNP (with 3-point moving average) from cationic cellulose
assisted reduction compared to a silver chloride standard (left and middle); the C-C and
C-O peaks are also compared before and after etching of the surface by using Ar+ gas
4.5. Mechanism
The molecular structure of cCNC is observed to change upon oxidation, which also
resulted in the reduction of Ag (I) to Ag (0). After reacting native CNC with QUAB 151
(quaternary ammonium), a hydrogel is formed. In NMR spectroscopy, this can suppress
structural features. Based on good correlation between 1H and 13C liquid NMR (Fig. 8 and
Fig. 9), protons from pyranose-form C1-H and from the quaternary ammonium group

66

gave very weak signals until AgNO3 was added to initiate the oxidation of cCNC. The
clear signals seen using D2O as solvent also proves the solubilization of cCNC.

Fig. 4.8 - The 1H-NMR of cCNC in both reduced and oxidized forms

Fig. 4.9 - The 13C-NMR of cCNC in both reduced and oxidized forms

The acyclic C1-H was only observed by 1H NMR (8.4 ppm) because the signal to noise
was high enough to resolve this peak, which is interpreted as a hemiacetal C-H signal.
Upon oxidation, the peak at 8.4 ppm reduced in size because the hemiacetal was
oxidized. Hydrolysis of cCNC resulted in several signals appearing between 140-160
ppm for 13C-NMR. The region of 140-160 ppm belongs to carboxylate groups. No
67

carboxylic acid signals were seen using 1H-NMR possibly due to proton exchange. For
this reason, conductometric titration was performed to confirm the presence of carboxylic
acid groups (0.50 mmol/g) taking presence of carboxylic acid groups originally present
(0.20 mmol/g) into consideration (total: 0.70mmol/g). The increased amount of weak acid
groups signifies a 100% conversion of silver (I) to silver (0). Assuming the literature
molecular weight value of CNC isolated from the acid hydrolysis of Whatman paper
(70,000 g/mol), the total number of cellulose chains can be determined. Multiplying this
value by 2 affords the number of reducing ends present. Comparing the number of
reducing ends with the amount of carboxylic acid groups present upon oxidation reveals
16 times more carboxylic acid groups than reducing ends originally in the sample. This
helps to elucidate the mechanism and suggests the possibility of chain cleavage during
cationization with the epoxide.

Peaks can be seen at 5.2-5.4 ppm and they are associated with the signals from -C1-H.
NMR provides evidence of chain cleavage because cellulose is structurally bonded -1-4
rather than -1-4. This change can occur upon hydrolysis and signals between 90-110
ppm observed by 13C-NMR also suggests a mixture of - anomers. All peak positions
for the carbohydrate backbone are unchanged. However, peak height did increase for C6
of oxidized cCNC in 1H-NMR, though not for 13C-NMR. The C6 feature can be obscured
due to the hydrogel formation of cCNC. Aside from microscopy and elemental analysis,
by using two separate high-field NMR instruments to corroborate data, the structural
information obtained clearly shows that hydrolysis occurs and is responsible for
providing additional hemiacetal groups to reduce silver ions. Signals were not seen above
8.4 ppm for 1H-NMR or above 160 ppm for 13C-NMR.
68

4.6. Conclusion
By modifying CNC using a cationic epoxide, the resulting mixture is capable of reducing
Ag (I) to AgNP. In this system, two products can form: AgCl, which is the precipitation
product, and AgNP, which is the reduction product. Changes to the reactants greatly alter
the outcome of the chemical reaction. It has been found that cCNC possesses all of the
qualities needed to obtain AgNP. In this work, the exact reduction mechanism has not
been determined. In future works, the method by which cCNC reduces metals should be
elucidated in order to deepen the understanding of this phenomenon. It can be posited
that the polyol reduction mechanism and the reducing ends of cellulose may work
synergistically. In this case, the two mechanisms combined are responsible for cellulose's
ability to act as a reducing agent. The modified cellulose can replace harsh reducing
agents in many chemical reactions.

69

4.7. References
Aulin C.; Ahola S.; Josefsson P.; Nishino T.; Hirose Y.; sterberg M.; Wgberg L. (2009)
Nanoscale Cellulose Films with Different Crystallinities and MesostructuresTheir
Surface Properties and Interaction with Water, Langmuir, 25 (13), 7675-7685.
Christensen, L. (2011) Biosynthesis of silver nanoparticles using murraya koenigii (curry
leaf): An investigation on the effect of broth concentration in reduction mechanism and
particle size, Adv Mat Lett, 2 (6), 429-434.
Filippo E.; Serra A.; Buccolieri A.; Manno D.J. (2010) Green synthesis of silver
nanoparticles with sucrose and maltose: Morphological and structural characterization,
Non-Cryst Solids, 356 6-8 344-350.
Garza-Navarro M. A.; Aguirre-Rosales J. A.; Llanas-Vzquez E. E.; Moreno-Cortez I. E.;
Torres-Castro A.; Gonzlez-Gonzlez V. (2013) Totally Ecofriendly Synthesis of Silver
Nanoparticles from Aqueous Dissolutions of Polysaccharides, Int J Polym Sci, Article ID
436021, 8pp.
He J.; Kunitake T.; Watanabe T. (2005) Porous and nonporous Ag nanostructures
fabricated using cellulose fiber as a template, Chem Commun, 795-796.
Ifuku S.; Tsuji M.; Morimoto M.; Saimoto H.; Yano H. (2009) Synthesis of Silver
Nanoparticles Templated by TEMPO-Mediated Oxidized Bacterial Cellulose Nanofibers,
Biomacromolecules, 10 (9), 2714-2717.
Jeong Y.; Lim D. W.; Choi J. (2014) Assessment of size-dependent antimicrobial and
cytotoxic properties of silver nanoparticles, Adv Mater Sci Eng, Article ID 763807, 6
pages.
Kaiser J-P.; Diener L.; Wick P. (2013) Nanoparticles in paints: A new strategy to protect
faades and surfaces, J Phys: Conf Ser, 429, 012036.
Kaushik M.; Chen W.C.; van de Ven T.G.M.; Moores A. (2014) An improved
methodology for imaging cellulose nanocrystals by transmission electron microscopy,
Nordic Pulp & Paper Research J, 29 (1), 77-84.
Kumar A.; Vemula P.K.; Ajayan P.M.; John G. (2008) Silver-nanoparticle-embedded
antimicrobial paints based on vegetable oil, Nat Mater, 7 (3), 236-241.
Lokanathan A.R.; Uddin K. M. A.; Rojas O. J.; Laine J. (2014) Cellulose NanocrystalMediated Synthesis of Silver Nanoparticles: Role of Sulfate Groups in Nucleation
Phenomena, Biomacromolecules, 15 (1), 373-379.

70

Lu Z.; Rong K.; Li J.; Yang H.; Chen R. (2013) Size-dependent antibacterial activities of
silver nanoparticles against oral anaerobic pathogenic bacteria, J Mater Sci Mater Med.
24 (6), 1465-1471.
Mock J. J.; Barbic M.; Smith D. R.; Schultz D. A.; Schultz S. (2002) Shape effects in
plasmon resonance of individual colloidal silver nanoparticles, J Chem Phys, 116, 67556759.
Prabhu, S.; Poulose E.K. (2012) Silver nanoparticles: mechanism of antimicrobial action,
synthesis, medical applications, and toxicity effects, Int Nano Lett, 2:32.
Raveendran, P.; Fu, J.; Wallen, S.L. (2003) Completely "green" synthesis and
stabilization of metal nanoparticles, J Am Chem Soc, 125, 13940-13941.
Roe D.; Karandikar B.; Bonn-Savage N.; Gibbins B.; Roullet J.B. (2008) Antimicrobial
surface functionalization of plastic catheters by silver nanoparticles, J Antimicrob
Chemother, 61 (4), 869-876.
Travan A.; Pelillo C.; Donati I.; Marsich E.; Benincasa M.; Scarpa T.; Semeraro S.; Turco
G.; Gennaro R.; Paoletti S. (2009) Non-cytotoxic Silver Nanoparticle-Polysaccharide
Nanocomposites with Antimicrobial Activity. Biomacromolecules, 10 (6), 1429-1435.
Trombino, S.; Cassano, R.; Ferrarelli, T. in Cellulose and Dextran Antioxidant Polymers
for Biomedical Applications, in Antioxidant Polymers: Synthesis, Properties, and
Applications, Eds. G. Cirilo and F. Iemma; John Wiley & Sons, Inc.: Hoboken, NJ, USA,
2012.
Umer A.; Naveed S.; Ramzan N.; Rafique M.S.; Imran M. (2014) A green method for the
synthesis of Copper Nanoparticles using L-ascorbic acid, Matria (Rio J), 19 (3), 197203.
Woods, D.J. (2011) Potion or Poison? Colloidal silver, Primary health Care, 4 (3), 251252.

71

Bridging section between chapters 4 and 5

The previous chapter dealt with a competing reaction between reduction of silver cations
(Ag+) to form silver nanoparticles (AgNP) and precipitation of Ag+ with chloride anions
(Cl-) to form silver chloride (AgCl) nanoparticles. It was found that, by modifying
cellulose nanocrystals (CNC) with cationic charge groups to form cationic CNC (cCNC),
a smaller particle diameter is observed. Based on this evidence, short chains of cCNC
was assumed to have solubilized into aqueous solution. This resulted in a greater
exposure of hydroxyl (-OH) groups and reducing chain ends (CHO). Given that there are
much more -OH groups than CHO groups in cellulose, it is hypothesized that the polyol
reduction mechanism is greatly responsible for the reduction of Ag+ to Ag0. It was shown
that unmodified CNC, in a solution of NaCl and AgNO3 (Ag+ source), resulted in the
precipitation product AgCl. It was also shown that the use of cationic wood pulp also
resulted in the formation of AgCl.

Based on the results from chapter 4, lower molecular weight carbohydrates

(monosaccharides and disaccharides) were explored to reduce Ag+ to Ag0 in order to
continue the project. In chapter 5, a study involving molecular and higher molecular
weight carbohydrates with and without CHO groups allows for the assessment of the
polyol reduction and the aldehyde reducing sugar mechanism.

72

Chapter 5:
Synthesis and Stabilization of Silver Nanoparticles with Carbohydrate Shells
5.1. Abstract
Silver nanoparticles (AgNP) were synthesized by the reduction of silver nitrate (AgNO3)
using various types of carbohydrates in the absence of light. By means of AgNO3 blankcorrected UV-VIS, the characteristic broad peak of AgNP at approximately 450 nm was
identified for all carbohydrates used, regardless of the presence of reducing ends.
Analysis using high-resolution TEM revealed the formation of AgNP. By means of high
magnification, a 'sugar shell' was observed on the surface of AgNP. The size of AgNP
ranges from 5-50 nm while the sugar shell ranges from 1-10 nm depending on the
carbohydrate used in the synthetic process. Electron diffraction x-ray spectroscopy
(EDX) revealed the elemental composition and crystal diffraction pattern of the structures
observed. The silver core is crystalline and consists of silver atoms while the sugar shell
is an amorphous organic material consisting of carbon and oxygen atoms.

73

5.2. Introduction
The synthesis of silver nanoparticles (AgNP) by the reduction of a silver salt using
naturally occurring carbohydrates has been reported in many publications found in the
scientific literature (Shervani & Yamamoto, 2011; Mehta et al., 2010). In the past, strong
reducing agents such as sodium borohydride were used (Lokanathan et al., 2014), but the
progression of green chemistry brought about new means of synthesizing AgNP.
Naturally occurring reducing agents have been greatly studied because of their ecofriendliness. It has been found that plant antioxidants are capable of reducing silver ions
(Rodrguez-Len et al., 2013) and that many types of carbohydrates also possess this
quality (Christensen et al. 2011; Wang et al. 2005). The ability of simple and more
complex sugars to reduce metal ions has been established by the presence of a hemiacetal
group, also called the reducing end (Varki, 2009). The mechanism of reduction has been
explained by the reaction of sugars with Tollen's reagent and Benedict's reagent where the
hemiacetal group can be oxidized to form a carboxylic acid. Aside from this mechanism,
a second pathway to the formation of metals has been gaining approval: the oxidation of
hydroxyl (-OH) groups (Raveendran et al., 2003; Donati et al., 2009). It has been
demonstrated in the literature that non-reducing sugars such as sucrose can also be used
to form AgNP (Filippo et al., 2010). The ability of carbohydrates to reduce metal ions to
form metal nanoparticles is not the only feature of these green reducing agents. It has also
been shown that upon reduction, carbohydrates also stabilize the metal nanoparticles to
prevent aggregation and coalescence (Huang & Yang, 2004). In our work, the green
synthesis of AgNP was performed using various carbohydrates. By means of UV-VIS
spectroscopy, transmission electron microscopy (TEM), and electron diffraction x-ray

74

spectroscopy (EDX), the metal nanoparticles formed were analyzed to study differences
between the carbohydrates.
5.3. Materials and methods
5.3.1. Materials
The source of sugars for this work were arabinose (Acros), galactose (Acros), and reagent
grade sucrose (Aldrich). The polysaccharides were agarose (Fluka Chemika) and chitosan
(Aldrich). The source of silver ion was AgNO3 (Fluka Chemika).

5.3.2. Methods and characterization techniques

Transmission electron microscopy (TEM) was performed using the Tecnai G2 F20 CryoSTEM to provide high-resolution images, electron diffraction patterns, and elemental
analysis. Analysis by UV-VIS was performed using the Cary 5000 UV-VIS
photospectrometer.
5.3.3. AgNP formation
At a carbohydrate concentration of 10% w/w in aqueous suspension, 5% w/w solid
AgNO3 was added to the mixture inside a vial covered with aluminium foil and shaken
vigorously by hand for 10 seconds. A blank containing only 5% w/w AgNO3 was also
prepared the same way. All subsequent analyses were performed with the vial covered in
aluminium foil to prevent photoreduction.

75

5.3.4. Analysis
The liquid samples were deposited on carbon film copper grids without staining and
imaged by using TEM. The elemental makeup and the crystal structure of the silver
nanoparticles were analysed using the same TEM instrument's built-in electron
diffraction x-ray spectroscopy analyser. The colloidal suspensions were analyzed by UVVIS spectroscopy without additional sample preparation.
5.4. Results and discussion
Analysis by UV-VIS spectroscopy requires no sample preparation because colloidal
suspensions could be introduced into the sample chamber directly. The AgNP suspended
in water was only briefly exposed to light during manipulations. Although it has been
documented by other research groups (Son et al., 2004) that the photoreduction of AgNO3
can occur and result in the formation of unstable silver nanoparticles, the blank solution
prepared never underwent photoreduction. The spectrum of the blank is not included in
Fig. 5.1 because no peak or broad signal was observed and all samples containing
carbohydrates were blank-corrected. It is important to note the dynamic changes that
were observed for the AgNP samples. Over time, aqueous suspensions became more
prominently coloured and their extinction changed, though the broad peak at
approximately 450 nm remained for all samples. In the case of arabinose and galactose,
the suspension grew more opaque and darkened quickly over time until dark metal
sediments formed in less than 2 hours. Compared to reducing sugars, the extinction of the
non-reducing sugar sucrose rose slower, and resulted in the formation of a silver mirror.
In literature and the UV-VIS spectrometer refer to both the absorption and the scatter of

76

light as 'absorption', it is actually the extinction that is measured. This point is further
discussed in Appendix I.

Fig. 5.1. The UV-VIS spectra of 5 different carbohydrates with their most intense peaks
shown at approximately 450 nm
The use of UV-VIS spectroscopy to assess the formation of AgNP is a quick and
convenient method based on the absorbance of light between 350 and 700 nm. The
presence of a broad peak at approximately 450 nm is in good agreement with values
found in the published literature (Shahverdi et al., 2007; Martinez-Castanon et al., 2008).
In all cases, regardless of the presence or concentration of hemiacetal groups, AgNP
formed while the AgNO3 blank was not reduced under the same conditions. It was
observed that galactose formed AgNP at a quicker rate while sucrose, a non-reducing
sugar, required more time to perform the same reduction. This can be attributed to a
reducing sugar having two pathways to metal cation reduction over non-reducing sugars,
77

which only have the polyol reduction mechanism and not the aldehyde reduction
mechanism. By performing a time-based analysis, the evolution of AgNP growth
mediated by galactose and sucrose are shown in Fig. 5.2. The UV-VIS peaks are broader
for galactose and narrower for sucrose, which is attributed to greater aggregation and
coalescence in the case of the galactose-stabilized silver colloidal suspension. After 90
minutes, sedimentation began for this reducing sugar. No sedimenation was observed in
the case of the sucrose-stabilized silver colloidal suspension.

Fig. 5.2. The reduction of silver cations to AgNP using galactose (reducing sugar)
proceeds on the order of minutes (A) and rapidly sediments after 120 minutes; the
reduction using sucrose (non-reducing sugar) proceeds on the order of hours (B) and did
not sediment
By using the Tecnai G2 F20 instrument, high-resolution TEM images were acquired.
Using the same instrument, electron diffraction patterns and EDX elemental makeup
information were also acquired. All AgNP varied in size between 5-50 nm. This size
dispersion can be attributed to the sample preparation. An important feature of this work
is characterised by the avoidance of harsh chemicals, ultrasonication, and heating. The
use of natural carbohydrates, a silver salt, and shaking by hand afforded AgNP of size
dispersion no greater 45 nm that are stable for many days. In chapter 4, the AgNP were
much smaller in size because titration was performed to introduce silver ions. In this

78

chapter, reagents were added to the mixture in solid form, which has an influence on
particle size.

As shown in Fig. 5.3, a reducing sugar such as galactose is able to form AgNP from
AgNO3 readily. A closer look at individual nanoparticles reveals a shell of lower electron
density of approximately 1 nm thickness. Sucrose, a non-reducing sugar was observed to
produce a much thicker sugar shell at 8 nm on average. This can be attributed to the
solubility of the sugar upon reduction of the silver cation. It is known that the formation
of AgNP also causes the oxidation of the carbohydrate. In the case of a reducing sugar,
aldehyde groups are converted to carboxylic acid while, for non-reducing sugars,
hydroxyl groups are converted to ketones and aldehydes. Given that the solubility of
carboxylic acids is greater than that of ketones and aldehydes, a thick sugar shell is less
likely to form in the case of reducing sugars. As shown in Fig. 5.3, the 1 nm thickness
shell can be attributed to either a monolayer or a bilayer of sugar molecules depending on
the amount of carboxylic acid present. In the case of non-reducing sugars, the formation
of a less soluble oxidation product is favourable for the formation of a thick shell. A
thinner shell also corroborates with the theory of higher instability leading to
sedimentation of AgNP when using reducing sugars. Presumably reducing sugars do not
produce an adsorption layer thick enough to cause steric stability and the charges of the
carboxylated sugars are too low to create an electrostatic energy barrier sufficiently high
to prevent slow aggregation. For the non-reducing sugars, the adsorption layer is
sufficiently thick to produce sterically stable core-shell AgNP.

79

Fig. 5.3. A core-shell structure of AgNP and carbohydrates is seen by high resolution
TEM using galactose (A) and sucrose (C); a low-resolution image of the same systems
reveal no core-shell structure when using galactose (B), though a faint shell can be seen
in the case of sucrose (D) given the added thickness
The TEM images of higher molecular weight carbohydrates (Fig. 5.4) reveal thick sugar
shells that also corroborates with the solubility argument. As shown in Fig. 5.4, agarose
and chitosan also form core-shell structures with AgNP. The shell thickness is on the
order of 5-10 nm. Unlike sucrose, though, this thickness represents a monolayer of high
molecular weight macromolecules.

80

Fig. 5.4. High-resolution TEM of Core-shell structures for higher molecular weight
carbohydrates: agarose (A) and chitosan (B)
By analyzing both the metal core and the sugar shell (Fig. 5.5), a silver-core sugar shell
structure is observed. The use of EDX reveals the presence of copper, silicon, carbon, and
oxygen in an empty area, which suggests the elemental makeup of the TEM grid can
interfere with the analysis. The thickness of each nanoparticle can vary, which makes a
subtraction of this spectrum with all other spectra an unreliable technique to remove
interfering patterns. Nevertheless, subtraction was performed for all EDX spectra to
lower the unwanted signals.

Fig. 5.5. The EDX spectra of AgNP stabilised by various carbohydrates with the electron
beam aimed at the silver core (A) and the sugar shell (B)
81

As it can be seen from Fig. 5.5, the core is made up of silver atoms and the electron beam
necessarily detects the sugar shell, which causes the detection of carbon-oxygen signals.
The shell is rich only in carbon and oxygen, which suggests the presence of an organic
material. In the absence of any other reagent, it can be deduced that the shell can only be
a carbohydrate. Furthermore, crystal diffraction in Fig. 5.6 revealed that the shell is
amorphous, which also suggests the formation of a sugar shell. It has also been observed
that residual water remains in the shell after sample preparation for TEM analysis, which
explains the greater amount of oxygen in the shell region. Core-shell metal-organic
systems have been documented by TEM in the past(Wu & Zhou, 2015; Schofield et al.,
2006; Wang et al., 2015), and AgNP stabilized by a sugar shell has been documented
previously (Vigneshwaran et al., 2006; Kennedy et al., 2014). However, our work
demonstrates the stability of silver colloidal suspensions, which depends greatly on the
thickness of the shell and is dictated by the type of carbohydrate being used.

Fig. 5.6. Electron diffraction pattern of the silver core (220) (A) and the amorphous sugar
shell (B); the diffraction patterns for all samples were closely similar; these images were
obtained using galactose stabilised AgNP
Although structural determination by NMR showed oxidation occurring at C1 of cationic
cellulose nanocrystals (cCNC) in chapter 4, the same cannot be said for the simple sugars

82

in this chapter. Assuming that carboxylic acids form upon oxidation at the anomeric
carbon, 13C NMR should reveal peaks around 150 ppm. However, experimental data
shows no change in the peak positions prior to and after oxidation. In other words,
despite the formation of AgNP, the carbohydrate reducing agents did not show any
change by NMR spectroscopy. It is posited that the carbohydrate shells around the AgNPs
are oxidized and would display a peak indicative of carboxylic acids. However, AgNPs
and their carbohydrate shells are not soluble in D2O and, therefore, are not in the analyte
solution. This is unlike cCNC where no carbohydrate shell forms. Nevertheless,
conductometric titration was performed and did reveal the large quantity of carboxylic
acid, consistent with a 1:1 stochiometric molar balance seen in chapter 4. This provides
further evidence that the carbon-rich shell contains the oxidation product of
carbohydrates.
In the literature (Chitvoranund, 2013), the Tollen's reagent, [Ag(NH3)2]NO3 has been
used to perform the silver mirror reaction in which silver ions adhere to the surface of
glassware to form a reflective layer. It is reported herein that, under the reaction
conditions specified, a silver mirror results from the reduction of AgNO3 using a nonreducing sugar. Moreover, the use of a reducing sugar does not produce the silver mirror
due to rapid sedimentation.
5.5. Conclusion
The use of carbohydrates as a means of reducing silver cations to AgNP has grown in
popularity in recent times. By limiting the use of toxic chemicals, it is possible to
synthesize metal nanoparticles that are stabilized by the reducing agent. Although the

83

reducing and stabilizing feature of carbohydrates have been explored by other groups,
observations of a sugar shell by TEM and EDX have not been documented prior to this
work. Future development in this field includes the analysis of a wider range of
carbohydrates and a kinetic elucidation on the reduction mechanism of the -OH groups
and the reducing ends.
5.6. References
Chitvoranund N.; Jiemsirilers S.; Kashima D.P. (2013) Effects of surface treatments on
adhesion of silver film on glass substrate fabricated by electroless plating J Aust Ceram
Soc, 49, 62-69.
Christensen, L. (2011) Biosynthesis of silver nanoparticles using murraya koenigii (curry
leaf): An investigation on the effect of broth concentration in reduction mechanism and
particle size. Adv Mat Lett, 2 (6), 429-434.
Donati I.; Travan A.; Pelillo C.; Scarpa T.; Coslovi A.; Bonifacio A.; Sergo V.; Paoletti S.
(2009) Polyol synthesis of silver nanoparticles: mechanism of reduction by alditol
bearing polysaccharides. Biomacromolecules, 10 (2), 210-213.
Filippo E.; Serra A.; Buccolieri A.; Manno D. (2010) Green synthesis of silver
nanoparticles with sucrose and maltose: Morphological and structural characterization. J
Non-Cryst Solids, 356 (6), 344-350.
Huang H.; Yang X. (2004) Synthesis of polysaccharide-stabilized gold and silver
nanoparticles: a green method. Carbohydr Res, 339 (15), 2627-2631.
Kennedy D. C.; Orts-Gil G.; Lai C.; Mller L.; Haase A.; Luch A.; Seeberger P. H. (2014)
Carbohydrate functionalization of silver nanoparticles modulates cytotoxicity and cellular
uptake. J. Nanobiotechnology, , 12 (1), 59.
Lokanathan A.R.; Uddin K. M. A.; Rojas O. J.; Laine J. (2014) Cellulose NanocrystalMediated Synthesis of Silver Nanoparticles: Role of Sulfate Groups in Nucleation
Phenomena. Biomacromolecules, 15 (1), 373-379.
Martinez-Castanon G. A.; Nino-Martinez N.; Martinez-Gutierrez F.; Martinez-Mendoza
J. R.; Ruiz, F. (2008) Synthesis and antibacterial activity of silver nanoparticles with
different sizes. J Nanopart Res, 10 (8), 1343-1348.

84

Mehta S. K.; Chaudhary S.; Gradzielski M. (2010) Time dependence of nucleation and
growth of silver nanoparticles generated by sugar reduction in micellar media. J. Colloid
Interface Sci., 343 (2), 447453.
Raveendran, P.; Fu, J.; Wallen, S.L. (2003) Completely "green" synthesis and
stabilization of metal nanoparticles. J Am Chem Soc, 125, 13940-13941.
Rodrguez-Len E.; Iniguez-Palomares R.; Navarro R.E.; Herrera-Urbina R.; Tanori J.;
Iniguez-Palomares C.; Maldonado A. (2013) Synthesis of silver nanoparticles using
reducing agents obtained from natural sources (Rumex hymenosepalus extracts).
Nanoscale Res Lett, 8, 318.
Schofield C.L.; Haines A.H.; Field R.A.; Russell D.A. (2006) Silver and gold
glyconanoparticles for colorimetric bioassays. Langmuir, 22 (15), 6707-6711.
Shahverdi A.R.; Fakhimi A.; Shahverdi H.R.; Minaian S. (2007) Synthesis and effect of
silver nanoparticles on the antibacterial activity of different antibiotics against
Staphylococcus aureus and Escherichia coli. Nanomedicine, 3 (2), 168-171.
Shervani Z.; Yamamoto Y. (2011) Carbohydrate-directed synthesis of silver and gold
nanoparticles: effect of the structure of carbohydrates and reducing agents on the size and
morphology of the composites, Carbohydr Res, 346 (5), 651658.
Son W.K.; Youk J.H.; Lee T.S.; Park, W.H. (2004) Preparation of antimicrobial ultrafine
cellulose acetate fibers with silver nanoparticles. Macromol Rapid Commun, 25 (18),
1632-1637.
Varki, A. Chapter 2: Structural Basis Of Glycan Diversity. In Essentials of glycobiology;
Cold Spring Harbor Laboratory Press: Cold Spring Harbor, NY, New York, 2009.
Vigneshwaran N.; Nachane R.P.; Balasubramanya R.H.; Varadarajan P.V. (2006) Effects
of surface treatments on adhesion of silver film on glass substrate fabricated by
electroless plating. Carbohydr Res, 341 (12), 20122018.
Wang C.; Tang F.; Wang X.; Li L. (2015) Self-Assembly of Fluorescent Hybrid Core
Shell Nanoparticles and Their Application. ACS Appl. Mater. Interfaces, 7 (24), 1365313658.
Wang H.; Qiao X.; Chen J.; Ding S. (2005) Preparation of silver nanoparticles by
chemical reduction method. Colloids Surf., A, 256 (2-3), 111-115.
Wu W.; Zhou S. (2010) Hybrid micro-/nanogels for optical sensing and intracellular
imaging. Nano Res, 1.

85

6. Conclusions, contributions, and suggestions for future work

6.1. Conclusions and contributions to original knowledge
The main objective of the present dissertation is to modify carbohydrates in bulk, on the
nano- and molecular scales to improve the properties of handsheets and the stability of
colloidal suspensions. As a superhydrophobic wood pulp, modified cellulose did not
undergo hornification or permanent lumen closure upon drying from forced wetting in
vacuo. On the nanoscale, cellulose nanocrystals (CNC) were modified by introducing
cationic charge with an epoxy quaternary amine. By performing this modification,
cationic CNC (cCNC) was shown to undergo mild disintegration. This resulted in a
greater exposure of nanocellulose surface area and is responsible for the formation of
silver nanoparticles (AgNP) when cCNC together with dissolved cCNC chains are in the
presence of silver nitrate (AgNO3). To continue the study on carbohydrate-mediated
AgNP synthesis, reducing and non-reducing sugars were employed to form silver core
carbohydrate shell nanoparticles. It was found that the use of a non-reducing sugar such
as sucrose afforded the most stable colloidal suspension.

The summary of the main findings of the work employing the chemical modification
methods mentioned, together with all features and results of the studies, are given in this
chapter of the thesis. The discussion is divided into four subsections, corresponding to the
four main chapters of the dissertation. Sections 6.1.1 and 6.1.2 summarize the study on
superhydrophobic cellulose pulp and its drying properties. Section 6.1.3 focuses on the
ability of cCNC to form and stabilize AgNP. Section 6.1.4. deals with the reduction and

86

stabilization of AgNP using carbohydrates of varied molecular weights that are either
reducing or non-reducing sugars.
6.1.1. Superhydrophobic cellulose handsheets
As discussed in chapter 2, cellulose is a natural biopolymer that contains many hydroxyl
(-OH) groups. Cellulose fibres can uptake water into their lumen and cell walls through
micro- and nano-structures. The -OH groups make cellulose a naturally hydrophilic
material. By means of vapour deposition, trichloromethylsilane (TCMS) forms a layer of
siloxane by reacting with -OH groups This chemical modification affords a modified
wood pulp with a contact angle of 150 when pressed into a handsheet, which makes this
material superhydrophobic.
In this work, never-dried native wood pulp was washed with anhydrous ethanol by
means of a procedure called solvent exchange. With water almost completely removed
and replaced by ethanol, the wood pulp was dried in an oven. This drying technique
resulted in a low-density foam-like dry pulp because ethanol has a lower surface tension
than water, which prevents the lumen collapse of cellulose fibres by capillary forces.
Fibres with this open lumen structure were then chemically modified by vapour
deposition with TCMS to yield a low-density superhydrophobic material called
Cellufoam.
By means of X-Ray tomography and SEM, it was found in a side-by-side comparison
with native fibres that the fibre cross-section of Cellufoam fibres is indeed an open
structure.

87

The reaction conditions were explored in this work and numerous conclusions were
made. Solvent exchange with ethanol affords an open lumen structure due to a lower
surface tension (22.11 mN/m). The density of wood pulp dried using this solvent was
found to be 0.06 g/cm3. Using surfactants at 0.01 % w/w, the density of dried wood pulp
was found to be 0.1 g/cm3. The vapour phase reaction with TCMS was also explored.
From the start of the reaction to 5 minutes in, a rapid surface coverage was observed in
which %coverage went from 0 to over 80% by surface area. The reaction rate slowed
after 5 minutes to nearly 100% coverage at 15 minutes. The reaction cannot easily
proceed at this point because a high amount of HCl by-product forms and adsorbs onto
the cellulose surface, which hydrolyses the material. The optimal conditions to produce
Cellufoam involve solvent exchange using ethanol and between 5-15 minutes vapour
phase reaction with TCMS. The resulting product is a superhydrophobic handsheet with a
density of 0.06 g/cm3.
6.1.2. Superhydrophobic cellulose handsheet: A material that expands upon drying
In chapter 3, as a continuation to the work on superhydrophobic cellulose material called
Cellufoam, this material was forcibly wetted in vacuo and subsequently pressed into a
wet handsheet. Under normal circumstances using native cellulose, the volume of a wet
handsheet is greater than the dry handsheet because materials have always been
documented to swell upon wetting and shrink upon drying. Instead of undergoing dryshrinkage, Cellufoam was shown to expand upon drying, which is characterised by a
negative dry-shrinkage coefficient of approximately -12. This phenomenon has never
been documented previously. The handsheet at 50% dry weight expanded by over 5 times
its original thickness when fully dried.

88

In this work, the dry-expansion of Cellufoam was measured when originally produced
and after a maximum of three cycles. It was found that the water drop contact angle of
this material decreased after each cycle until the handsheet contact angle was just barely
above 90. This was rationalised by the mechanical damage that the pulp undergoes
during preparation in which the superhydrophobic pulp is stirred in water under vacuum
to force the uptake of water and subsequently pressed under high pressure. The
mechanical damage incurred in each cycle is responsible for exposing previously
inaccessible -OH groups and for permanently deforming and rupturing cellulose fibres.
The handsheet expansion mechanism was explored in this work by observing the
expansion versus solid content graph. Two separate expansion mechanisms are
responsible for the total expansion of this handsheet. Firstly, the deformation of fibre
lumen is an unstable conformation. As a result, elastic rebound of the lumen structure
occurs first and causes the most prominent amount of sheet expansion. In a second step,
fibres that were bridged to each other by a capillary force due to the presence of water
can also undergo elastic rebound when the capillary force is reduced by lower water
content. This mechanism also explains the low reusability of Cellufoam. With mechanical
damage, the force of elastic rebound is reduced and fibres can no longer overcome
capillary forces.
6.1.3. Cationic cellulose nanocrystal assisted reduction of silver cations to silver
nanoparticles
As discussed in chapter 4, the commonly accepted principle by which silver cations (Ag+)
are reduced to silver metal (Ag0) is the presence of aldehyde (-CHO) groups in which the
reduction of Ag+ to Ag0 also causes the oxidation of -CHO groups to carboxylic acid (-

89

COOH) groups. This mechanism has been shown in cited literature to be a viable method
of producing Ag0. A second reduction pathway also exists and has been documented in
the literature whereby the numerous hydroxyl (-OH) groups are able to perform polyol
reduction, which also results in the reduction of metal cations.
In this work, cellulose nanocrystals (CNC), rod-like nanoparticles were modified
by coupling quaternary amine (-R4N+) groups onto the surface of CNC via an epoxide
ring-opening ether linkage reaction. The resulting material is a suspension of positively
charged nanorods called cationic cellulose nanocrystals (cCNC) in which chloride (Cl-)
acts as the counter-ion. As verified by AFM, cCNC nanorods possess smaller crosssections than CNC nanorods due to the partial dissolution of cCNC chains. This change
in dimensions along with the presence of dissolved cCNC chains has the effect of
increasing the surface area occupied by cCNC in aqueous suspension, which also leads to
an increase in accessible -OH groups as well as -CHO groups. As a result, cCNC is able
to reduce Ag+ to Ag0 despite the presence of Cl-, which could also have led to the
formation of silver chloride (AgCl), a precipitation product. It was found that a
suspension of CNC in the presence of either NaCl or un-reacted quaternary amine both
led to the formation of AgCl via precipitation. Furthermore, cationic wood pulp with Clcounter-ions also led to the precipitation product. It was concluded that two competing
reactions occur when cellulose, Ag+, and Cl- are all present: reduction resulting in the
formation of silver nanoparticles (AgNP) and precipitation resulting in the formation of
AgCl. The methods by which AgNP and AgCl can be distinguished include diffraction by
XRD and by EDX as well as elemental analysis by XPS and EDX. In the case of AgNP,
diffraction showed no AgCl diffraction pattern and elemental analysis revealed trace

90

amounts of chlorine, which is not quantifiable in the analysis. By the analytical means
mentioned, it was concluded that cCNC is capable of producing significantly more AgNP
than AgCl in this competitive reaction.
6.1.4. Synthesis and stabilization of silver nanoparticles with carbohydrate shells
The work on reducing metal cations to metal nanoparticles continued into chapter 5. The
difference lies in varying the reducing agent in order to explore the difference between
-CHO groups and -OH groups in terms of their abilities to reduce Ag+ to AgNP. For this
work, reducing sugars (arabinose and galactose), a non-reducing sugar (sucrose), and
weak reducing sugars of high molecular weight (various polysaccharides) were used to
cause the reduction of Ag+.
By using analytical techniques such as UV-VIS spectroscopy and TEM, it was
found that, although reducing sugars acted faster, non-reducing sugars also caused the
formation of AgNP. The greater reducing ability of arabinose, galactose, and
polysaccharides led to the rapid coalescence and sedimentation of silver metal while a
stable colloidal suspension of AgNP was observed when sucrose was employed. Highresolution TEM revealed that a core-shell structure formed spontaneously in all cases.
However, sucrose formed a thicker shell (8 nm) compared to arabinose and galactose (1
nm). This can be attributed to a multilayer of sucrose forming around the AgNP while a
monolayer or bilayer formed in the case of the reducing sugars. Polysaccharides also
formed shells around AgNP of approximately 5-10 nm; however, given their molecular
weight, this thick shell is most likely a monolayer. By elemental analysis, it was shown
that the silver core is composed of silver and the sugar shell contains high levels of

91

carbon and oxygen. Given the large amount of oxygen in the shell, it was also concluded
that water was present.
By allowing the suspensions a sufficient amount of time, the ones stabilized by
reducing sugars and polysaccharides eventually resulted in sedimentation, which is
evidenced by the lack of an absorbance signal when analyzed by UV-VIS spectroscopy
after 30 days. On the contrary, sucrose-stabilized AgNP resulted in the formation of a
silver mirror staining the walls of the glass vial used to store the colloidal suspension. It
has not been reported previously that the reduction of Ag+ to AgNP mediated by sucrose
can also lead to the formation of a silver mirror.
6.1.5. Suggestions for future work
In the present thesis, carbohydrates were explored on size scales ranging from bulk
cellulose to monosaccharides. In the case of superhydrophobic Cellufoam, its ability to
expand upon drying when pressed into a wet handsheet is the first evidence of a material
with dry-expansion properties. As discussed previously, dry-expansion has been
documented for the first time in this work, but other hydrophobic and superhydrophobic
polymers can mimic bulk cellulose fibres and be tailored to expand upon drying. The
discovery of such a material can be significant for industrial applications, especially as an
expanding filler in mouldable technology. As for the ability of carbohydrates to reduce
silver cations, reducing ability has often been studied by other groups, but the stability
and structure of the AgNP formed has not been greatly studied. Although the polyol
reduction mechanism has been documented in the past and non-reducing sugars have
been shown to reduce metal cations, the stability of sucrose-mediated AgNP was

92

unexplored. In this work, a core-shell AgNP structure was elucidated and sucrose was
shown to form the most stable colloidal suspension amongst all carbohydrates as
evidenced by the sedimentation of AgNP colloids stabilized by reducing sugars. Aside
from AgNP, other nanostructures and systems susceptible to colloidal instability may be
improved by the addition of a non-reducing sugar such as sucrose.

93

Bridge to the appendix

As mentioned in the Introduction, the Appendix contains a publication by Madhu
Kaushik, Wei Chung Chen (author of the present thesis), Dr. Theo G.M. van de Ven, and
Dr. Audrey Moores. The manuscript was published in the Nordic Pulp and Paper
Research Journal in 2014. With the guidance of Dr. van de Ven and Dr. Moores, major
portions of the written text were improved because both the first and second authors of
this publication had no prior experience during the writing of the draft text. The majority
of the draft text was edited by Dr. Moores to meet publication standard.
The introduction section was written collaboratively between Madhu Kaushik, the first
author, and Wei Chung Chen, the second author. Preparation and analysis of never-dried
and freeze-dried CNC were performed by Chen. Analysis on freeze-dried and spray-dried
CNC were performed by Kaushik. Additionally, Chen provided all TEM grids (carbon,
formvar, and silicon monoxide) as well as pH dependent analyses of never-dried and
freeze-dried CNC while Kaushik performed pH dependent work on spray-dried CNC. All
work involving high-resolution TEM, low-dose, and metal nanoparticle analysis were
also performed by Kaushik. Furthermore, significant portions of the particle counting
were facilitated using the software imageJ. This task was also performed by Kaushik. As
a result of the greater contribution, Kaushik was appointed first author and Chen second
author. The supplementary information section of the publication can be found online by
referring to the following reference:
Kaushik, M.; Chen, W.C.; van de Ven, T.G.M.; Moores, A. (2014) Imaging Cellulose
Nanocrystals by Transmission Electron Spectroscopy, Nordic Pulp and Paper Res J, 29
(1), 77-84. (Invited contribution to the special issue on Nanocellulose)

94

Appendix I
A-I-1. UV-VIS spectroscopy
When measuring the absorbance of an aqueous suspension using UV-VIS spectroscopy, a
proper understanding of the underlying physics is necessary to interpret different effects
caused by scattering, which results in peak broadening and lower extinction.

In the work presented, UV-VIS analysis was performed on silver nanoparticles (AgNP) in
a heterogeneous suspension with cellulose nanocrystals (CNC). It is known that the
optical properties of metal nanoparticles depend greatly on surface plasmons. This
concept, although greatly ignored in chapters 4 and 5, is essential to understand why
certain peak shapes and intensities were observed. Plasmons are described as the
oscillation of electron density within a confined space. The excitation of surface
plasmons by radiation at a given frequency (wavelength) is responsible for the
nanoparticles unique extinction, which is a combination of absorption and scatter. It has
been shown by Near et al. that the broadening of absorbance spectra is caused by greater
scatter contribution, which also implies a smaller absorbance contribution, which is
illustrated in Fig. A-I-1. Although UV-VIS spectroscopy measures absorption, the actual
value being measured is the extinction.

95

Fig. A-I-1. The contribution of absorption and scattering to the extinction (left); the
formation of a plasmonic dipole in spherical nanoparticles as determined by computation
(right); the colours of the dipole (right) indicate the electric field enhancement where the
middle slice exhibits maximal enhancement
In cases where excitation does occur, atoms can, for a short time period, experience
electronic excitation to a higher energy level. In mere nanoseconds, the excited atoms
drop back down to their ground state and release the same quantum of energy that was
absorbed. This is the physical description of the term scatter. Instead of releasing the
energy elastically during excitation, atoms can collide with each other. In this case, the
energy can be given up instead of re-emitted. The photonic energy is thus converted into
kinetic energy. This is now the physical description of absorption. When both effects are
combined, the extinction describes the curve seen by UV-VIS spectroscopy. It has been
shown that the greatest contributor to the extinction is absorption and peaks tend to be
fairly sharp when scatter does not play a large role. Surface plasmons are most prominent
at specific wavelengths and absorption bands can form at discrete regions of the
spectrum. It is when particles are larger or anisotropically shaped that different minor
absorption bands and peak broadening due to scatter can occur. In the case of a
suspension of AgNP and CNC, the AgNP will absorb due to surface plasmons at
approximately 450 nm. However, the presence of CNC is largely responsible for light
scattering and can even contribute to the aggregation of AgNP into larger clusters, thus
further broadening the signals and reducing the extinction.

96

A-I-2. X-ray diffraction

In order to properly analyze the silver nanoparticles (AgNP) shown in this thesis, several
different techniques were used to match the experimental data with the expected results
either theoretically or using experimentally data obtained by other groups. In other words,
only fingerprinting has been performed without a true explanation of what the raw data
actually mean. In this subsection, the fundamental principles of x-ray diffraction (XRD)
and electron diffraction will be addressed.
Observing the crystal structures of solid materials is a good method to prove the identity
of analytes. This is because no two solids are packed the same way. The atoms may only
be packed in a limited set of patterns, but the spacing between atoms is unique for any
one crystalline structure. By using a technique such as XRD, not only is the angle at
which signals appear important, but the height of the peak as well.
The two most important equations in XRD are Braggs equation and Scherrers equation.
Braggs equation (S2-1) is a general explanation of diffraction interference patterns where
an incident radiation of known wavelength () reflects off of a crystal plane at an angle
taking the spacing between two atoms (d) into consideration. Constructive and
destructive interference result from Braggs equation given that sin makes Braggs
equation an oscillating function. Using this equation, the d-spacing (space between
atoms) can be known.
n = 2dsin

(A-1)

Scherrers equation (A-2) takes the analysis further by including a shape factor (k) and
the full width at half maximum (FWHM) denoted . In the case of silver, a face-centric
97

cubic structure is expected and k is set to 0.94 throughout. Scherrers equation allows for
a prediction of the d-spacing as a function of , which makes the XRD diffractogram
even more comprehensive. In the case of ideal crystals free of defects, one would expect
perfectly matching peak heights at a given . However, the presence of impurities and
non-uniform strains can alter both the peak height and the for expected peaks. It could
also be specified that diffractograms use 2 to denote angles because the angle between
the incoming and the diffracted beam are considered.
d = k / cos

(A-2)

Fig. A-I-2. The crystal spacing (d) as well as the 2 angle are illustrated
Geometrically, the faces are assigned coordinates to denote different faces. The
coordinate system used is called Miller Indices. As it can be seen with cubic crystal
lattices, diffraction patterns exist for light scattering off of cubic faces (e.g. 100) as well
as through the cubic lattice on a slant (e.g. 110). In the case of a face-centric lattice such
as silver, reflection planes from cubic faces are expected to have smaller d-spacings than
the slanted planes because there is no body-centric atom. In other words, there is no atom
in the middle of the cube. The d-spacing on the cubic face is different from the d-spacing

98

at the cubic center. As a result, peak height plays a significant role in the identification in
crystal analytes when looking at XRD data.

Fig. A-I-3. The top, middle, and lower rows represent the cubic (sc), the body-centric
cubic (bcc), and the face-centric cubic (fcc), respectively
As mentioned, the presence of impurities or strains and deformations in the crystal
structure can alter the and values greatly. For this reason, searching theoretical
diffractograms of silver in the literature often does not match with the values obtained by
different research groups in experimental work. This technique alone cannot be used to
identify materials with 100% accuracy. However, a combination of approaches makes
XRD a valuable tool.
99

100

A-I-3. Electron diffraction

The wave-particle duality is the concept responsible for electron diffraction. Electrons
possess mass and also momentum. Referring to eq. A-3, the wavelength associated with a
particle () is inversely proportional to the momentum (p); Planks constant is represented
by the letter h.

h
p

(A-3)

It has been observed that when a thin material (<100 nm) is bombarded with high energy
electrons, a diffraction pattern results. Transmission electron microscopy (TEM) makes
use of high energy electrons for imaging and for electron diffraction. Instead of
measuring reflected radiation as demonstrated for XRD, electrons of even shorter
wavelength than x-rays bombard the thin and electron transparent material. In the case of
a single crystal material, the crystal lattice acts as a grating for transmitted electrons. This
results in constructive and destructive interference as shown by Braggs equation (A-1).
The diffraction patterns shown in chapters 4 and 5 of this thesis were obtained by
bombarding electrons onto AgNP surfaces observed to possess a lattice structure and
single crystal diffraction patterns resulted. Sites on AgNP that possess defects were not
reported, but were measured and multiple patterns observed. It can be said that AgNP can
grow as single crystals, but that defects can result. It is postulated that colloidal instability
leading to aggregation is responsible for crystal defects.

101

Fig. A-I-4. A diagram illustrating the working principles of electron diffraction

A-I-4. X-ray photospectroscopy (XPS)
A very sensitive technique used for elemental analysis in chapter 4 was used to determine
the presence of chlorine in silver nanoparticles. Furthermore, the technique was also used
to qualitatively determine the role of carbon in this heterogeneous system. It is therefore
important to discuss the working principle of this technique.

Fig. A-I-5. The working principle of XPS

The irradiation of solid analytes with x-rays in vacuo results in the ejection of a
quantifiable number of electrons at a given kinetic energy. This technique allows for the
ejection of electrons from the topmost 10 nm of a sample (Fig. A-I-5). Many instruments

102

are also equipped with argon gas ion etching capabilities. This technique is similar to
sandblasting, but on the atomic scale.

When analysing surfaces, atmospheric carbon causes the formation of adventitious

carbon, a surface impurity that slightly changes the kinetic energy of ejected electrons
due to surface charge. It has been observed that adventitious carbon surfaces can shift
spectral peaks by over 0.2 eV. In order to account for this effect, XPS spectra are
corrected by positioning the largest carbon (1s) peak at 284.8 eV and, thus, shifting the
entire spectrum. When analysing samples that contain carbon, the aforementioned
technique cannot be used because there may be several sizeable carbon peaks. In this
case, a standardised material is used for calibration.

A-I-5. Energy-dispersive X-ray photospectroscopy (EDX)

A similar technique, though less sensitive, is EDX. Rather than to irradiate samples with
x-rays, an electron gun is used to excite and eject core electrons. Given the instability of
electron holes in the core orbitals, electrons from higher energy orbitals rapidly fill the
hole. This electronic transition results in the emission of x-rays, which are captured by a
detector (Fig. A-I-6).

103

Fig. A-I-6. EDX works by ejecting core electrons and detecting the X-rays emitted by the
transition of electrons from high energy levels to fill these electron holes
The experimental setup used in this thesis involves the use of high energy electrons
emitted by the electron gun used for transmission electron microscopy (TEM). The
advantage of EDX over XPS is the ability to aim the electron beam at a specific region
with nanometer accuracy. The disadvantage of using EDX, as mentioned, is the lower
sensitivity of the measurements. The ejection of x-rays is in all directions and may not be
captured by the x-ray detector. Furthermore, the density of the analyte will result in
greatly varied sensitivity. Another inherent problem with EDX is the numerous peak
overlaps between elements. For the reasons listed, pairing XPS with EDX provides a
much more complete analysis.

A-II-6. Conductometric titration

In order to quantitatively determine the presence of weak and strong acids separately in
an aqueous solution or suspension, titration using a base titrant is a good technique.
During the titration, both the pH and the conductance change as a function of titrant
volume added. As shown in Fig. A-1-7, the pH measurement only shows an equivalence

104

point when all acids have been neutralized. Conversely, two separate acid regions can be
seen by conductometric titration due to the lower dissociation rate of protons for weak
acids.

Fig. A-I-7 The plot on the left shows the pH versus titrant volume titration curve; the
plot on the right shows the conductivity versus titration volume titration curve where
three distinct regions are represented by different slopes
The change in pH reflects the change in concentration of protons (H+). When acids are
fully neutralized by the addition of base, an equivalence point is reached and this is
marked by an inflection point. In aqueous samples where a mixture of acids are present or
when a polyprotic acid is under investigation, several equivalence points can overlap and
be difficult to deconvolute. In a system where sulphonate groups (strong acid) and
carboxylic acids (weak acid) are present, conductometric titration is a more appropriate
choice for quantitative analysis. The difference in conductivity change with respect to

105

titrant volume is shown in Fig. A-I-8. The titration methodology is very similar to pH
titration, in which a base is the titrant and the added volume is recorded. However, rather
than to monitor the change in pH, it is the conductance that is under investigation. For the
work in this thesis, the presence of carboxylic acid upon oxidation (chapters 4 and 5) is
measured. In order for this technique to be valid, titration before and after the oxidation
reaction was performed. The difference in carboxylic acid molar quantities is meaningful
because cellulose and other carbohydrates naturally possess carboxylic acid groups in
low concentration.

Fig. A-I-8 The slope of the strong acid is always steeper than that of the weak acid; the
weak acid can have a positive or negative slope depending on its acid constant; excess
base is characterized by a significant rise in conductivity
During a conductometric titration, ions from the analyte are replaced by ions from the
titrant. In order to reduce error, the conductivity of the two ions needs to be significantly
different. In the case of protons from acids and sodium ions (Na+) from the NaOH titrant,
the difference in conductivity is approximately 29 mS m-2 mol-1, which allows for welldefined changes in conductivity with respect to titrant volume added. Furthermore, a

106

strong acid such as sulphonic acid has highly dissociative protons while carboxylic acids
have less dissociative protons. When replacing these protons by sodium cations, the
slopes between the two types of acids are different, which allows for a quantifiable
determination of weak and strong acids. During acid-base neutralization, the
conductance decreases for two reasons: sodium is less conductive than protons and water
is less conductive than either protons or hydroxyl anions. Once all of the acid groups
have been neutralized, excess OH- from the titrant causes a rise in both conductance and
pH.

Structural analysis by infrared spectroscopy (IR) and nuclear magnetic resonance

spectroscopy (NMR) can also be used to qualitatively determine the presence of
carboxylic acid groups in the samples under investigation. However, if the analyte is an
aqueous suspension of hydrogels and insoluble fractions, then certain structural features
can be obscured. In the work presented, cationic cellulose nanocrystals (cCNC) formed a
hydrogel and several carbohydrates formed insoluble shells bound to silver nanoparticles
(AgNP). These suspensions are more effectively analyzed using conductometric titration
because the addition of strong base during the analysis helps solubilize the carbohydrates.
For this reason, results obtained by conductometric titration varied greatly from the
results obtained by NMR.

107

Appendix II:
An improved methodology for imaging cellulose nanocrystals by transmission electron
microscopy
A-II-1. Abstract
Although transmission electron microscopy (TEM) is a method of choice for the analysis
of cellulose nanocrystals (CNCs), it is still a challenging technique to use due to the low
electron density and aggregation issues associated with this material. This work presents
a robust and reproducible method to image CNCs by TEM, without the need for staining.
It is demonstrated that low pH and hydrophilic TEM grids favour high dispersion of the
sample for optimal imaging. Three distinct types of CNCs are analyzed: never-dried,
freeze-dried and spray-dried CNCs. Although a few reports accounted for mushroom
structures, we demonstrate herein that, regardless of the sample history, this nanomaterial
is always rod-shaped.

108

A-II-2. Introduction
Cellulose nanocrystals (CNCs) are derived from the renewable polymer, cellulose,
through acid hydrolysis by concentrated sulphuric acid, first reported by Rnby (Rnby,
1949). This material has well-defined size and morphology (Klemm et al., 2011), high
specific surface area (Morandi et al., 2009), high aspect ratio, high crystalline order and
chirality, superior mechanical strength, and controllable surface chemistry (Habibi et al.,
2010). In addition, CNCs are relatively cheap, renewable, biodegradable, non-toxic
(Kovacs et al., 2010; Kmmerer et al., 2011), and accessible industrially in large scale.
These properties of CNCs have made them applicable in diverse fields, for example,
production of iridescent and birefringent films (Revol et al., 1992), reinforcing fillers in
plastics and polymers (Dufresne, 2010; Habibi et al., 2010; Kloser & Gray, 2010),
support for nanoparticles (Cirtiu et al., 2011; Padalkar et al., 2010), and paper, food,
cosmetic and pharmaceutical industry (Habibi et al., 2010). The knowledge of exact size
and morphology of CNCs plays a very important role in the development of all these
applications. In addition, the development of several processes for the production of
CNCs (vide infra), calls for a systematic and robust method of analyzing these properties
(Klemm et al., 2011). Scanning electron microscopy (SEM), transmission electron
microscopy (TEM) and atomic force microscopy (AFM) are the classical techniques used
to determine the shape and size of these nanocrystals. In this context, TEM offers several
advantages: its resolution is well adapted for the imaging of CNCs, it is a milder
technique than SEM in terms of flux of electrons received by the sample, it allows
visualization of individual particles, and the sampling techniques are easy, reproducible,
and tunable depending on the selected TEM grids. However, despite the ubiquitous use of

109

TEM for CNC characterization, there are still challenges associated with its use. This
paper proposes a straightforward method of imaging CNCs by TEM, without staining. In
doing so, we explored the properties of CNCs in terms of drying process, of pH, and
functionalization, and demonstrated that the size and shape of CNCs are robust features.
CNCs, being organic matter, are not as electron dense as metal nanoparticles, and are
susceptible to burning upon long electron beam exposure. Hence, TEM imaging of CNCs
has often involved negative staining with uranyl acetate (Kvien et al., 2005), a relatively
expensive and toxic heavy metal salt (Nakakoshi et al., 2011). Staining poses aggregation
problem in the drying step. In addition, when characterizing CNC-containing hybrid
structures (for instance CNC-supported metal nanoparticles), staining interferes with
accurate image analysis (Cirtiu et al., 2011). A few unstained images have been reported
in the past, but they suffered from poor contrast and no systematic method has been
reported.
Herein, we report a method for taking TEM images of CNCs in detail, covering various
aspects of ascertaining high quality images, such as method of preparation of TEM grids,
dispersion of CNCs on the grid, types of TEM grids used, low dose versus high
resolution (HR) TEM, and preventing damage to the CNC sample under electron beam
while using a high resolution and/or high voltage TEM.
This work also looks into morphology differences depending on the drying method. The
size and morphology of this biopolymer depends on the source of cellulose and the
method of preparation, where the key conditions include: the type and concentration of
the acid used for hydrolysis, reaction time and temperature, and method of drying (Habibi
et al., 2010; Klemm et al., 2011). The usual methods of drying (Peng et al., 2012a) used

110

for the preparation of CNCs are: (1) oven drying, (2) freeze drying (FD), (3) supercritical
drying (SCD), and (4) spray-drying (SD). It has been reported that freeze-dried CNCs are
fibre/rod-like whereas spray-dried CNCs are spherical or mushroom shaped (Peng et al.,
2012a; Peng et al., 2012b; Peng et al., 2013); (Ramanen et al., 2012). We investigated in
detail TEM sampling and imaging for CNC samples with distinct drying history. We
compared never-dried samples (samples that were not dried after synthesis), spray-dried
and freeze-dried samples. We demonstrated that all samples possess the same nanorod
morphology, provided they are dispersed properly.
A-II-3. Materials and methods
A-II-3.1. Materials and equipment
Never-dried CNCs were prepared from wood pulps (Q90 and bHKP, FPInnovations).
FPInnovations also provided dried CNC samples: freeze-dried and spray-dried, the
freeze-drying or spray-drying process being performed by FPinnovations. No further
treatment was performed before imaging on these samples, unless specified otherwise.
For the manipulation of pH, hydrochloric acid (0.01M) was provided by Fluka
Analytical. To redisperse dried CNC samples in suspension, the Sonics & Materials Inc.
Vibra-Cell VC1500 ultrasonicator (220V, 15A, 1500W, 20 kHz) was used. The Metrohm
Titrando (836 base, 804 Ti stand, and 802 stirrer) autotitrator was used for attaining
various pH values. The transmission electron microscopy (TEM) samples were deposited
on 400 mesh copper grids coated with silicon monoxide, formvar, and carbon. All three
were supplied by Electron Microscopy Sciences. The analyses were performed on a
Tecnai 12 microscope (FEI electron optics) and a Lab6 filament at 120kV, and a Gatan
792 Bioscan 1k x 1k Wide Angle Multiscan CCD Camera (Gatan Inc.); and an FEI G2

111

F20 Cryo-S/TEM microscope (FEI, Inc) at 200kV, equipped with a Gatan Ultrascan 4k x
4k Digital (CCD) Camera System at different magnifications corresponding to different
pixel size (defocus level ranging from 2.5 to 4.5 m). High pressure reactions were
performed on Parr Instrument 5000 Series Multiple Reactor System.
A-II-3.2. Preparation of CNC for never-dried samples
Based on the method developed by Dong et al. (Dong et al., 1998), CNCs were prepared
from wood pulp. A 500 ml aqueous solution of H2SO4 (60% w/w) was preheated to 45C
and stirred with an overhead rotating paddle at 200 RPM. To this solution, a 100 ml
suspension of Q90 and bHKP wood pulp (obtained as wet pulp samples from the
manufacturer) at a 1:1 volume ratio were added. After 60 minutes, the suspension turned
beige in colour. To quench this reaction, 1 l of water was added. By centrifugation, the
supernatant was removed and more fresh water was added to repeat the washing. After
four washings by centrifugation, the sediments became more loosely packed. At this
stage, the sediments were collected and dialyzed with deionised water for 5 days under
constant stirring. The resulting suspension was cast into films and weighed to determine
the concentration of CNCs in water.
A-II-3.3. Samples preparation for TEM imaging
In the case of never-dried CNCs, the samples were already suspended in water. As for
freeze-dried and spray-dried CNCs, samples were weighed and mixed with a known
amount of deionised water. All three types of CNCs were dispersed uniformly in water by
ultrasonication at 60% output, for about 20-25 seconds, until turbidity was visibly
reduced and no macroscopic particles could be seen. Their concentrations ranged

112

between 0.1% w/w and 0.3% w/w. The samples were either used as is (pH 5-6) or the pH
of the samples was set to 3.5 by autotitration. One drop of each sample was then
deposited onto either a carbon, formvar, or silicon monoxide coated grid. After 1 minute,
the water drop was removed by dabbing with a tissue paper. CNC single rod size was
determined by manually measuring 1000 single rods with the help of imageJ software
(version 1.48e). For comparison purposes, never-dried CNCs at pH 3.5 were stained with
uranyl acetate and imaged. The staining was done according to the protocols already
available in the literature (Qua et al., 2011).
A-II-3.4. Synthesis of Pd nanoparticles onto CNCs (PdNPs@CNCs)
This method was adapted from Cirtiu (Cirtiu et al., 2011). A suspension of 500 mg of
CNCs in 50 ml HCl solution at pH 2 was mixed with 50 ml of saturated PdCl2 acidic
solution (pH adjusted to 2 with dilute HCl). The mixture was stirred for 10 min to
homogenize the suspension. The final concentration of PdCl2 was 0.75x10-3 M, while
the CNC concentration in the mixture was 0.5% w/w. Then, the mixture was placed under
4 bars of hydrogen gas for 2 hours at room temperature to obtain PdNPs@CNCs.
A-II-3.5. Low dose and high resolution transmission electron microscopy
The low dose and high resolution (HR-TEM) transmission electron microscopy were
carried out on an FEI G2 F20 Cryo-STEM microscope (FEI, Inc) at 200 kV. At this high
voltage, the CNCs are more prone to be damaged by the electron beam. Therefore, it is
advisable to use the microscope at low dose TEM for organic polymers. Low-dose
imaging is carried out by locating an area of interest on the grid surface and determining
the appropriate focus without significantly exposing the location to be imaged to high

113

electron beam. This microscope has preinstalled low dose software, which was used to
get the TEM image of freeze-dried CNCs on carbon grid at pH 5-6. An image with good
focus and contrast can also be achieved without using the low dose software. Using the
HR-TEM, the intensity of the electron beam was lowered to reduce the number of
electrons striking the sample and the objective aperture was reduced from 100 to 40, to
get a better contrast, where necessary, for some of the samples.
A-II-4. Results
For this work, we imaged CNCs while varying the following parameters: drying history,
pH of the suspension and TEM grids. CNC suspensions were synthesized in the lab,
sampled directly and referred to as never-dried. Freeze-dried and spray-dried CNCs
were ordered from FPinnovations and resuspended prior to sampling. All our TEM
observations are summarized in table A-II-1 based on Fig. A-II-1, Fig. A-II-2, Fig. A-II-4,
Fig. A-II-5, Fig. A-II-6.
A-II-4.1. Optimization of dispersion conditions for TEM imaging: pH and grid type
In a first step, optimization of imaging conditions was performed on never-dried CNCs
samples. Their pH values ranged between 5 and 6. At this pH and concentration, TEM
images featured individual rods and aggregated ones (Fig. A-II-1, Table A-II-1). In order
to improve dispersion, the pH was lowered to 3.5 using dilute hydrochloric acid (0.01N).
The larger aggregates came apart, smaller bundles were observed and individual CNCs
could be seen in larger proportion.
CNCs were also imaged using three different types of grid coating: carbon, formvar and
silicon monoxide. Imaging was impacted by this parameter as well. The formvar grids

114

have a hydrophobic surface. On this substrate, single rods came together in large
aggregates and were no longer observable individually. On the carbon grid, dispersion
improved. On the hydrophilic silicon monoxide coated grids, the dispersion was
maximized. Fig. A1 and Fig. A2 compare never-dried CNCs at pH 5-6 on the three types
of grids and clearly show this trend. Fig. A3 summarizes the size distribution, in length
and in width, of the CNC bundles/rod of never-dried CNCs at pH 3.5 on different types
of grid, based on manual counting of 200 particles of each type. Their average length and
width are provided in Table A-II-2. Both histograms and average values show a clear
trend: the more hydrophilic the grid, the smaller the particles measured, both in length
and in width. For instance, silicon monoxide grids featured a CNC length of 149 nm,
while this value was 211 for carbon coated ones and 250 for formvar. The trend was even
more pronounced for CNC width with values of 5.2, 8.8 and 19.1 nm for silicon
monoxide, carbon and formvar coating respectively. This is consistent with the fact that
bundles composed of a small number of CNC rods are more numerous in a hydrophobic
environment. This phenomenon is more visible when looking at the CNC width
compared to length, because high aspect ratio CNCs favorably aggregate laterally.
Table A-II-1. Summary of results from TEM images of CNCs depending on sample type,
grid type and pH

115

Fig. A-II-1. Never-dried sample on carbon grid: (left) at pH 5-6; (right) at pH 3.5

Fig. A-II-2. Never-dried CNCs at pH 5-6: on formvar grid (left); silicon monoxide grid
(right)
CNCs were also imaged using three different types of grid coating: carbon, formvar and
silicon monoxide. Imaging was impacted by this parameter as well. The formvar grids
have a hydrophobic surface. On this substrate, single rods came together in large
aggregates and were no longer observable individually. On the carbon grid, dispersion
improved. On the hydrophilic silicon monoxide coated grids, the dispersion was
maximized. Fig. A1, Fig. A2, and Fig. A4 compare CNCs at varied pH, drying conditions,
and on three types of grids. Bundles composed of a small number of CNC rods are more
numerous in a hydrophobic environment, especially when samples have never been dried.
This phenomenon is more visible when looking at the CNC width compared to length,
because high aspect ratio CNCs favourably aggregate laterally. To compare our

116

methodology with the earlier methods of staining prior to imaging the CNCs, we also
counted approximately 200 single rods after staining with uranyl the never-dried CNCs at
pH 3.5 on carbon grid. TEM images and a histogram for size distribution in this case
have been included in the supplementary info (Fig. S1, S2, S3 and S4). The average
width and length came out to be 9.1 3.2 nm and 248 72 nm, respectively. These
values are comparable, but both higher than the values obtained in this article without
staining (Table A-II-2). Also, in stained samples, we observed more large clusters than
unstained samples, confirming that staining favors aggregation.

Fig. A-II-3. Length and width distribution of CNCs bundles/rod of never-dried CNCs at
pH 3.5 on different type of grids, based on the counting of 200 particles on each type of
grid.
A-II-4.2. The impact of CNC drying history on TEM imaging
Fig. A4 compares the never-dried, freeze-dried and spray-dried samples on carbon grid, at
pH 5-6. Never-dried samples exhibited the best dispersion. Single rods were observed in
all three types of CNCs. Never-dried CNCs featured more single rods than the others. On
Fig. A4 we compared the typical width of the bundles observed in the three samples, as a
measure of aggregation. These values are characteristic of what we observed over
analysis of dozens of images. Spray-dried CNCs, with a typical CNCs bundle width of
168 nm featured a much enhanced aggregation behavior than both never-dried (29 nm)
117

and freeze-dried (33 nm). This conclusion is actually consistent with the fact that spraydried CNCs revealed to be more difficult to disperse than freeze-dried or never-dried
counterparts. In general, spray-dried CNC suspensions were allowed to be redispersed at
room temperature for more than a week before being imaged.
Increasing ultrasonication times prior to sampling did not improve dispersion of these
samples, demonstrating that drying history impacted irreversibly the dispersion properties
of CNCs. These results are also consistent with the work done by Beck et al. (Beck et al.,
2012) where they observe difference in aggregate sizes due to the different drying history
of CNCs.

Upon varying the sampling conditions and samples, we did observe one other kind of
structure, entirely different from the rods or rod bundles discussed above (Fig. A-II-5,
Table A-II-1). When imaging spray-dried samples, large circles of between 100 to 200
nm in size, and featuring a much higher contrast than the rods, were observed. This
morphology was exclusively observed in spray-dried samples, either on formvar grids, or
on carbon grids when the dispersion was not complete, namely when the suspensions
were prepared only a day before sampling and not a week. Interestingly, if silicon
monoxide coated grids were used instead of formvar coated grids, the spray-dried CNCs
were always observed to be rod-like. These results are consistent with the idea that the
circles are constituted of dense pockets of CNCs brought together during the drying
process on the grid. Formvar is hydrophobic and can encourage the formation of water
nano-droplets rich in CNCs, causing these morphologies.

118

Aggregation of CNCs, either caused by drying method (spray-dried vs other samples), by

poor resuspension, or by inappropriate grid coating (formvar vs silicon monoxide)
favored a similar mechanism. These structures have been observed by others and referred
to as mushrooms. Previous reports established that never-dried and freeze-dried CNCs
were fibre/rod-like whereas spray-dried CNCs were spherical or mushroom-shaped (Peng
et al., 2012b). While we did observe the mushroom structures, our results demonstrate
that such a structure is an artefact of sampling, caused by poor distribution on the TEM
grid. All CNC samples, regardless of drying history are clearly rod-shaped. This example
illustrates the importance of carefully preparing samples prior to TEM imaging.

Fig. A-II-4. Types of CNCs at pH 5-6, carbon grid: never-dried CNCs (left); freeze-dried
CNCs (middle); spray-dried CNCs (right)

Fig. A-II-5. Spray-dried CNCs (at pH 5-6) onto a formvar grid (left), carbon grid
(middle) and silicon monoxide (right).
Table A-II-2. Average length and width of CNC bundles/rod of never-dried CNC samples
at pH 3.5 on different type of grids, based on the counting of 200 particles on each type
of grid.

119

Fig. A-II-6. Freeze-dried CNCs (pH 5-6 and carbon grid): high resolution TEM (top) and
low dose TEM (bottom).
Comparisons have been made on never-dried, freeze-dried and spray-dried samples at pH
5-6. Never-dried samples exhibited the best dispersion. Single rods were observed in all
three types of CNCs. Never-dried CNCs featured more single rods than the others. On
Fig. A4 we compared the typical width of the bundles observed in the three samples, as a
measure of aggregation. These values are characteristic of what we observed over
analysis of dozens of images. Spray-dried CNCs, with a typical CNCs bundle width of
168 nm featured a much enhanced aggregation behaviour than both never-dried (29 nm)
and freeze-dried (33 nm). This conclusion is actually consistent with the fact that spraydried CNCs revealed to be more difficult to disperse than freeze-dried or never-dried
counterparts. In general, spray-dried CNC suspensions were allowed to be re-dispersed at
room temperature for more than a week before being imaged.
Increasing ultrasonication times prior to sampling did not improve dispersion of these
samples, demonstrating that drying history impacted irreversibly the dispersion properties
of CNCs. These results are also consistent with the work done by Beck et al. (Beck et al.,
2012) where they observe difference in aggregate sizes due to the different drying history
of CNCs.

120

Upon varying the sampling conditions and samples, we did observe one other kind of
structure, entirely different from the rods or rod bundles discussed above (Fig A4, Table
A-II-1). When imaging spray-dried samples, large spheres of between 100 to 200 nm in
size, and featuring a much higher contrast than the rods, were observed. This morphology
was exclusively observed in spray-dried samples, either on formvar grids, or on carbon
grids when the dispersion was not complete, namely when the suspensions were prepared
only a day before sampling and not a week. Interestingly, if silicon monoxide coated
grids were used instead of formvar coated grids, the spray-dried CNCs were always
observed to be rod-like. These results are consistent with the idea that the circles are
constituted of dense pockets of CNCs brought together during the drying process on the
grid. Formvar is hydrophobic and can encourage the formation of water nano-droplets
rich in CNCs, causing these morphologies.
Aggregation of CNCs, either caused by drying method (spray-dried vs other samples), by
poor re-suspension, or by inappropriate grid coating (formvar vs silicon monoxide)
favoured a similar mechanism. These structures have been observed by others and
referred to as mushrooms. Previous reports established that never-dried and freeze-dried
CNCs were fibre/rod-like whereas spray-dried CNCs were spherical or mushroomshaped (Peng et al., 2012b). While we did observe the mushroom structures, our results
demonstrate that such a structure is an artefact of sampling, caused by poor distribution
on the TEM grid. All CNC samples, regardless of drying history are clearly rod-shaped.
This example illustrates the importance of carefully preparing samples prior to TEM
imaging.

121

A-II-4.3. Comparison of TEM images of CNCs under high resolution and low dose
microscopy
It is a common practice to carry out a low dose TEM when the sample is an organic
polymer, so that the electron beam does not destroy the sample. CNCs may burn under
the high density electron beam while performing HR-TEM. There are two ways to save
the sample from getting destroyed: (1) use of low dose TEM using preinstalled software
in the microscope, and, (2) playing with the intensity of the electron beam on HR-TEM;
so that the sample is prevented from burning. As seen in Fig. A6, obtaining good images
from both methods was possible, as long as the electron beam did not damage the sample.
A-II-4.4. Morphology of modified/decorated CNCs with palladium nanoparticles

Fig. A-II-7. TEM image of modified CNCs, PdNPs@CNCs

At last, samples were prepared where CNCs were used as support for palladium
nanoparticles. Staining with heavy metal salts is not appropriate for the characterization
of such hybrid systems, because it would hinder the metal nanoparticles (Cirtiu et al.,
2011). Fig. A7 shows a TEM image of PdNPs@CNCs without any staining. Here, it was

122

important to take the image at lesser acquisition times and a large defocus value. The
presence of Pd could indeed induce damage of CNCs under the electron beam. Under
these conditions, both the CNCs and the Pd NPs were clearly visible. A close examination
of this image revealed that, as published before (Cirtiu et al., 2011), Pd NPs are mostly
present onto CNCs. It is important to note, however, that because of the need to defocus
the image to observe the CNCs, the size of Pd NPs, as seen on the image, is overestimated. Some particles appear white, as a consequence of the same effect.
A-II-5. Discussion
Transmission electron microscopy is a convenient technique for imaging CNCs by
passing electrons across a grid on which a dried sample rests. In the case of dry materials,
a reliable and representative image can be obtained, one that describes the system
accurately. However, when the aim is to study systems such as colloidal suspensions, it
becomes important to take aggregation phenomena into consideration as the specimen
dries. As shown above, CNCs were optimally imaged when the sample had never been
dried and deposited onto a hydrophilic grid, ideally silicon monoxide-coated. The more
common and affordable carbon-coated grids also provided reliable results in the majority
of cases. Besides, these results showed that never-dried samples were the easiest to
disperse on the grid, followed by freeze-dried and then spray-dried. Freeze-drying is less
likely to strongly bind individual CNC nanorods together than spray-drying. In any case
and regardless of the history of the CNCs, the material itself was always a rod-like, high
aspect ratio nanoparticle. As a general trend, the population of individual rods decreased
with increasing pH, increasing hydrophobicity of the grid coating and increasing particle
packing upon drying (Table A-II-1). Never-dried CNCs were better-dispersed and single

123

rods were more numerous in these samples. For freeze-dried CNCs, single rods, small
clusters and large aggregates were present. For freeze-dried samples, aggregation
depended on the concentration of suspensions used while freeze drying, as discussed by
Cervin et al (Cervin et al., 2011). Thanks to the high quality and resolution of the images
observed, we were able to size individual rods of CNCs (Fig. A8). 1000 single rods were
counted for never-dried CNCs at pH 3.5 on a carbon grid. We measured them to be 9.7
4.4 nm in width and 216 82 nm in length, consistent with reported values (Habibi et al.,
2010).

Fig. A-II-8. Histograms of CNCs single rods length (top) and width (bottom) measured
on 1000 particles for never-dried particles on carbon grids at 3.5 pH.
We also demonstrated that never-dried, freeze-dried and spray-dried are all rod shaped. If
dispersed properly, never-dried, spray-dried as well as freeze-dried CNCs appear very
similar. The mushroom-structure was occasionally observed, but we showed it is an
artefact of drying; as improved dispersion sufficed to break them apart and reveal the
presence of rods. This result is also in accordance with the TEM studies done with carbon
nanotubes (Boccaccini et al., 2006; Wang et al., 2000) and various types of inorganic
nanoparticles, such as ZnO (Hu et al., 2010) and Au nanoparticles (Westcott et al., 1998),
where the importance of dispersion of the sample is stressed while imaging these

124

nanoparticles. The type of morphology observed by TEM depends on how well the
sample has been dispersed in the suspension, which in turn depends upon the drying
history of the CNCs, interaction between the CNCs and the surface of the grid, and the
rate of evaporation.
In this work, we demonstrated that that negative staining with heavy metals can be
avoided by a careful tuning of TEM parameters. We provided above clear indications and
details. We also showed that good images could be obtained with voltages of 120 kV or
200 kV, a low voltage being recommended for sample conservation during long exposure.
An increased potential did not allow for greater contrast or sharpness. The most important
part of the procedure is to prepare the samples with diligence and thoroughness.
Appropriate dispersion of the CNC sample is critical to get high quality and reproducible
images, which is, in turn, important to carry out statistical measurement on size and
morphology of the CNCs. Silicon monoxide grids are better for sample preparation as
they are hydrophilic and prevent aggregation on the grid itself. Nevertheless, this work
showed that the carbon grids could be used with success as well.
A-II-6. Conclusions
This work emphasizes the importance of a well-established methodology for electron
microscopy imaging of CNCs. These results clearly demonstrate the importance of
controlling the sampling method, especially during sample drying on the TEM grid. We
provide in this article a clear methodology to perform This work emphasizes the
importance of a well-established methodology for electron microscopy imaging of CNCs.
These results clearly demonstrate the importance of controlling the sampling method,

125

especially during sample drying on the TEM grid. We provide in this article a clear
methodology to perform TEM of CNCs in a reliable and reproducible fashion by
controlling the pH, the grid properties and the sample suspension. This work revealed that
never-dried, freeze-dried and spray-dried all exist as rod structures, with a size of 9.7
4.4 nm in width and 216 82 nm in length, in the case of never-dried samples. An
important conclusion to draw from this work is that expensive and toxic heavy metal
stains are unnecessary for imaging. Low-dose and high resolution TEM were also
compared and discussed.TEM of CNCs in a reliable and reproducible fashion by
controlling the pH, the grid properties and the sample suspension. This work revealed that
never-dried, freeze-dried and spray-dried all exist as rod structures, with a size of 9.7
4.4 nm in width and 216 82 nm in length, in the case of never-dried samples. An
important conclusion to draw from this work is that expensive and toxic heavy metal
stains are unnecessary for imaging. Low-dose and high resolution TEM were also
compared and discussed.

A-II-7. References
Beck, S.; Bouchard, J.; Berry, R. (2012) Dispersibility in water of dried nanocrystalline
cellulose, Biomacromolecules, 13 (5), 1486-1494.
Boccaccini, A.R.; Cho, J.; Roether, J.A.; Thomas, B.J.C.; Jane Minay, E.; Shaffer, M.S.P.
(2006) Electrophoretic deposition of carbon nanotubes, Carbon, 44 (15), 3149-3160.
Cervin, N.T.; Aulin, C.; Larsson, P.T.; Wgberg, L. (2011) Ultra porous nanocellulose
aerogels as separation medium for mixtures of oil/water liquids, Cellulose, 19 (2), 401410.
Cirtiu, C. M.; Dunlop-Brire, A. F.; Moores, A. (2011) Cellulose nanocrystallites as an
efficient support for nanoparticles of palladium: Application for catalytic hydrogenation
and heck coupling under mild conditions, Green Chemistry, 13 (2), 288.

126

Dong, X.; Revol, J.-F.; Gray, D. (1998) Effect of microcrystallite preparation conditions
on the formation of colloid crystals of cellulose, Cellulose, 5 (1), 19-32.
Dufresne, A. (2010) Processing of polymer nanocomposites reinforced with
polysaccharide nanocrystals, Molecules, 15 (6), 4111-4128.
Habibi, Y.; Lucia, L. A.; Rojas, O. J. (2010) Cellulose nanocrystals: Chemistry, selfassembly, and applications, Chemical Reviews, 110, 3479-3500.
Hu, X.L.; Masuda, Y.; Ohji, T.; Kato, K. (2010) Rapid low-temperature synthesis of
porous ZnO nanoparticle film by self-hydrolysis technique, Key Engineering Materials,
445, 123-126.
Klemm, D.; Kramer, F.; Moritz, S.; Lindstrm, T.; Ankerfors, M.; Gray, D.; Dorris, A.
(2011) Nanocelluloses: A new family of nature-based materials, Angewandte Chemie
International Edition, 50 (24), 5438-5466.
Kloser, E.; Gray, D.G. (2010) Surface grafting of cellulose nanocrystals with
poly(ethylene oxide) in aqueous media, Langmuir, 26 (16), 13450-13456.
Kovacs, T.; Naish, V.; O'Connor, B.; Blaise, C.; Gagn, F.; Hall, L.; Trudeau, V.; Martel,
P. (2010) An ecotoxicological characterization of nanocrystalline cellulose (NCC),
Nanotoxicology, 4 (3), 255-270.
Kmmerer, K.; Menz, J.; Schubert, T.; Thielemans, W. (2011) Biodegradability of organic
nanoparticles in the aqueous environment, Chemosphere, 82 (10), 1387-1392.
Kvien, I.; Tanem, B. S.; Oksman, K. (2005) Characterization of cellulose whiskers and
their nanocomposites by atomic force and electron microscopy, Biomacromolecules, 6
(6), 3160-3165.
Morandi, G.; Heath, L.; Thielemans, W. (2009) Cellulose nanocrystals grafted with
polystyrene chains through surface-initiated atom transfer radical polymerization (siatrp), Langmuir, 25 (14), 8280-8286.
Nakakoshi, M.; Nishioka, H.; Katayama, E. (2011) New versatile staining reagents for
biological transmission electron microscopy that substitute for uranyl acetate, Journal of
electron microscopy, 60 (6), 401-407.
Padalkar, S.; Capadona, J.R.; Rowan, S.J.; Weder, C.; Won, Y.H.; Stanciu, L.A.; Moon,
R.J. (2010) Natural biopolymers: Novel templates for the synthesis of nanostructures,
Langmuir, 26 (11), 8497-8502.
Peng, Y.; Gardner, D.J.; Han, Y. (2012a) Drying cellulose nanofibrils: In search of a
suitable method, Cellulose, 19 (1), 91-102.

127

Peng, Y.; Han, Y.; Gardner, D.J. (2012b) Spray-drying cellulose nanofibrils: Effect of
drying process parameters on particle morphology and size distribution, Wood and Fiber
Science, 44 (4), 448-461.
Peng, Y.; Gardner, D.J.; Han, Y.; Kiziltas, A.; Cai, Z.; Tshabalala, M.A. (2013) Influence
of drying method on the material properties of nanocellulose i: Thermostability and
crystallinity, Cellulose, 20 (5), 2379-2392.
Qua, E.H.; Hornsby, P.R.; Sharma, H.S.S.; Lyons, G. (2011) Preparation and
characterisation of cellulose nanofibres, Journal of Materials Science, 46 (18), 60296045.
Ramanen, P.; Penttila, P.A.; Svedstrom, K.; Maunu, S.L.; Serimaa, R. (2012) The effect
of drying method on the properties and nanoscale structure of cellulose whiskers,
Cellulose, 19 (3), 901-912.
Rnby, B.G. (1949) Aqueous colloidal solutions of cellulose micelles, Acta Chemica
Scandinavica, 3 (1), 649-650.
Revol, J.F.; Bradford, H.; Giasson, J.; Marchessault, R.H.; Gray, D.G. (1992) Helicoidal
self-ordering of cellulose microfibrils in aqueous suspension, International Journal of
Biological Macromolecules, 14 (3), 170-172.
Wang, Z.L.; Poncharal, P.; de Heer, W.A. (2000) Nanomeasurements of individual carbon
nanotubes by in situ tem, Pure Appl Chem, 72 (1-2), 209-219.
Westcott, S.L.; Oldenburg, S.J.; Lee, T.R.; Halas, N.J. (1998) Formation and adsorption
of clusters of gold nanoparticles onto functionalized silica nanoparticle surfaces,
Langmuir, 14, 5396-5401.

128

129