N

G
LM

-Z

JU

FUNCTIONAL
POLYMERS

JI

Reading Materials for Students

Department of Polymer Science & Engineering

Zhejiang University
2011.3.

Chapter 1
Intrinsically Electrically Conducting Polymers

1.1 History and introduction

1.2 Synthesis of conducting polymers
1.3 Characterization of conducting polymers

1.4.2 Sensors and electromechanical devices

1.4.3 Batteries
1.4.4 Electrochromic cells
1.4.5 Controlled-release applications

N
G
LM

1.4.6 Radar applications

-Z

1.4.1 Corrosion protection

JU

1.4 Devices and applications of the oxidized and/or reduced (conducting) forms of conjugated
polymers

1.4.7 Infrared polarizers

1.5 Devices and applications of the neutral (non-conducting) form of conjugated polymers
1.5.1 Third-order non-linear optical polymers
1.5.2 LEDs

1.6 Goals for the future and speculation

1.1 History and Introduction

JI

On a historical note, polyaniline was first made as far back as 1862 by H. Letheby. Known as
aniline black, this material was formed by oxidation of aniline under mild conditions. Aniline
black was an important material for dyeing and printing.

Conducting-polymers research has roots back to the 1960s where Pohl, Katon and others first
synthesized and characterized semiconducting polymerss and conjugated polymers. The discovery
of the high conductivity of poly(sulfurnitride) (SN)x, a polymeric inorganic explosive and its
interesting electrical properties was a step towards conducting polymers as they are known today.
The beginning of conducting-polymer research began nearly a quarter of a century ago, when
films of polyacetylene were found to exhibit profound increases in electrical conductivity when
exposed to iodine vapor. This was the first report of polymers with high electrical conductivity.
The procedure for synthesizing polyacetylene was based upon a route discovered in 1974 by
Shirikawa and coworkers through serendipitous addition of 1000 times the normal amount of
catalyst during the polymerization of acetylene. Over the past two decades, there have been a
number of excellent reviews on conducting polymers. Conducting-polymer research is evolving
rapidly enough that yearly reviews are almost necessary. Today, there are hundreds of articles on
2

conducting polymers published every year. There is now a journal called Synthetic Metals,
which is almost exclusively dedicated to reporting on various aspects of conducting-polymer
research.
Definitions
Conducting polymers (their conducting forms) are usually classified as the cation salts of highly
conjugated polymers. The cation salts are obtained by electrochemical oxidation and
electrochemical polymerization or chemical oxidation (removal of an electron).35 It is also
possible to obtain the anion salts of the same highly conjugated polymers (which are also
conducting but much less stable than the cation counterparts) by either electrochemical reduction
or by treatment with reagents such as solutions of sodium naphthalide.

-Z

JU

In general, a conjugated backbone and/or a backbone that has a low enough oxidation potential
is necessary but not sufficient for the electrically conducting form of a polymer to remain stable in
the presence of air and water or end-use conditions, such as inside an automobile or home. For
clarity, the following definitions and abbreviations are given and will be referred to for the
remainder of this review.

N
G
LM

Oxidized (conducting) is the form of the polymer that has had electrons removed from the
backbone, resulting in (radical) cations. Neutral (insulating polymer) is the form of the polymer in
its uncharged state. This can be referred to as a reduced form of the oxidized polymer. For the
sake of clarity in this article, the phrase neutral or insulating will be used.
Reduced (conducting) is the form of the polymer that has had electrons added to the backbone
resulting in (radical) anions. The difference between the oxidation potential and the reduction
potential is roughly equal to the electronic band gap. It is not unusual that this form of conducting
polymer has not been isolated, since reduction potentials are usually less than 1.0 V vs the
saturated calomel electrode (SCE). It may be possible to isolate both forms of the polymer with
some of the low band gap materials being synthesized and as stronger electron acceptors are
incorporated into conducting-polymer backbones.

Fig. 1.1 shows some structures of conducting polymers; for poly(aniline) (PANI) see Fig. 1.2.

JI

Poly(acetylene)

Poly(para-phenylene)
Poly(thiophene)

Poly(pyrrole)
Poly(3,4-ethylenedioxythiophene)

Poly(para-phenylene vinylene)

Fig.1.1. Structures of conjugated polymers in their respective neutral forms.

Polyaniline is usually obtained by protonation of what is called the emeraidine base form,
shown in Fig. 2. The protonation reaction does not change the number of electrons in the polymer
backbone. However, starting at the leucoemeraldine form of polyaniline, one would obtain the
3

emeraldine salt (conducting) form of polyaniline by an oxidation reaction. Protonic doping has
also been observed in the case of alkoxy-substituted PVV.

1.2. Synthesis of Conducting Polymers

New conducting-polymer structures are reported every month. Conducting polymers are either
made directly by electro- or oxidative-polymerization or polymerized and then oxidized
chemically or electrochemically. Research areas in conducting polymers focus on making the
conducting form of PANI and other types of conducting polymer soluble, new structures, adapting
synthetic routes for similar monomers (such as adapting the Wessling route for furanylene
vinylene), new substituents (such as putting amino groups on PPV) and on semi- and
total-regiospecific polymerization.

N
G
LM

-Z

JU

Examples of partial regiospecific polymerization include polymerizing undecylbithiophene: see

Fig. 3. The structure of the monomer guarantees at least 50% of the polymer will have head-to-tail
linkages. The bulky nature of the undecyl groups will limit the amount of head-to-head linkages.

Fig. 1.2. Structures of polyaniline.

JI

An example of total regiospecific polymerization, apart from the polymerization of symmetric

monomers (such as ethylenedioxythiophene- and bis-thiophene-substituted disubstituted-benzenes)
is the use of special catalysts by McCollough et al., which give almost quantitative head-to-tail
regiochemistry (see Fig. 1.4).
Regiochemistry of conducting polymers has a remarkable influence on the properties of the
final material. The aregic, syndioregic and isoregic polymers of poly(3-undecyl thiophene) are
constitutional isomers of each other. It has also been shown that conducting films of isoregic
poly(3-undecylthiophene) have conductivities one to two orders of magnitude higher than the
aregic counterpart. The electronic structures are also different, as seen by the red shift in the max,
going from aregic to isoregic; they exhibit remarkable solvatochromism. The isoregic sample also
gives better film properties, most likely due to the different ring tilt of the isoregic system, which
in turn gives better conductivities.

JU

Fig.1.3. Regiospecific polymerization of undecylbithiophene.

Fig. 1.4. Near-quantitative regiospecific polymerization of alkyl thiophenes.

N
G
LM

-Z

Some of the goals in conducting-polymer synthesis are: synthesizing new and novel structures,
increasing the order of the polymer backbone (and also the conductivity), good processability,
easier synthesis, more defined three-dimensional structure, stability in both conducting and
non-conducting states, solubility in certain solvents such as water, and many other
application-unique properties. Since the conductivity of conducting polymers cannot be predicted,
the only way to evaluate new structures is to synthesize and measure. Granted, one can make some
general statements about proposed structures, particularly stability, which is related to the
oxidation (or reduction) potential of the polymer. Also adding substituents appears to impart
increased solubility, but sometimes at the expense of order in the film, film properties and,
therefore, conductivity. As conducting polymers are organic chemicals, there is a certain amount
of synthetic flexibility which allows the use of feedback from device and end-use requirements to
tailor properties of conducting polymers. This flexibility will be a key factor in successful
commercialization.

JI

A brief summary of some specific synthetic efforts is listed. There is considerable effort to
synthesize low band-gap polymers. Conducting-polymer wires and molecular circuits have also
been discussed. New methods for making polyacetylene via precursor polymers and metathesis
and dehydrohalogenation have been reported. As have methods for making polyazomethines,
derivatives of polydiacetylenes, vinyl polymers with pendant oligothiophenes, silyl-substituted
PPV, dimethoxy aniline, polynaphthothiophene, bis(dialkylamino) PPV, planarized PPP, soluble
PPP, polydiisoproylidinecyclobutene, so and polythiophenepolyketone.

1.3. Characterization of Conducting Polymers

Like most other polymers, conducting polymers can be characterized by a variety of analytical
techniques. Many examples exist in the literature, some of which include: cyclic voltammetry for
understanding redox processes in conducting polymers and evaluating potential battery and
electrochromic window material candidates; optical characterization of conducting polymers for
electrochromic window and non-linear optical materials; nuclear magnetic resonance for structure
confirmation, chain orientation and molecular motion; gel permeation chromatography for
molecular weight; Raman analysis, for vibrational assignments; differential scanning calorimetry
and thermogravimetric analysis for evidence of glass and melting transitions and decomposition
5

temperatures, dependence of conductivity on temperature and electric field and magnetic

susceptibility, to understand the conductivity mechanism; electroluminescence to screen for
potential use in light-emitting diodes (LEDs); various types of X-ray analysis to understand the
crystal structure and Rutherford backscattering to obtain elemental depth profiles, in order to gain
insight into the conductivity mechanism.

1.4. Devices and Applications of the Oxidized and/or Reduced (Conductind)

Forms of Conjugated Polymers
In this area of potential applications, conducting polymers are used as replacements for metals
because the conducting polymers have potentially unique and or superior properties, or because
the metals are toxic or damage the environment.
1.4.1. Corrosion protection

N
G
LM

-Z

JU

Current methods of corrosion protection (particularly marine coatings) do not last very long and
are coming under increased scrutiny by the Environmental Protection Agency. For example, the
use of chromium and cadmium for anti-corrosion coatings will soon be banned. A mechanism for
corrosion protection involves the use of a sacrificial electrode, such as a zinc coating, which will
corrode (oxidize) in the place of the substrate. However, the coatings do not last very long. The
oxidized zinc metal is dissolved by water or moisture. For this reason there are extreme
environmental concerns, since toxic metals are being released. Barrier coatings, such as epoxy, are
employed extensively, but they are not very durable/robust once a pit or hole in the coating has
been formed. The corrosive species then attacks the underlying metal and, thereby, increases the
exposed surface, accelerating the corrosion process.

JI

The corrosion inhibiting properties of conducting polymers was suggested by MacDiarmid in

1985. Initial studies on the protection of metal surfaces against corrosion by conducting polymers
were reported in the literature that same year. Much of the work on corrosion protection has
focused on PANI, but it has also been extended to other conjugated polymers. A major type of
corrosion occurs by oxidation of a metallic surface by a salt-water medium to produce oxides and
hydroxides. As these form, soluble species are produced, the surface pits increasing its surface
area and the rate of decomposition accelerates. One way to provide corrosion protection is to coat
the metal with a barrier to prevent the reactive species from reaching the surface. Galvanization
with zinc (or other metals with low enough oxidation potentials) prevents corrosion via the
creation of an interfacial potential at the metal-zinc interface. The zinc will corrode preferentially.
While the reactive species may come in contact with the meral, the increased oxidation potential
causes the metal to be unreactive. Corrosion is, therefore, inhibited.

The prior work utilizing PANI as a corrosion protection coating shows that it works quite well.
In fact, exposed metal surfaces adjacent to conducting-polymer coatings (scratches or edges) are
unreactive to corrosion, as reported by Thompson and coworkers. The corrosion protection
properties of PANI on aluminum have also been studied.
At present, some conducting polymers (in their neutral, non-conducting states) are soluble in
organic solvents. Various types of surfactant counter-ion have been used with PANI to make the
conducting form of polyaniline soluble in organic solvents.
There is a huge commercial potential for the use of conducting polymers as corrosion inhibiting
coatings. Some estimates indicate that corrosion costs US industries tens of billions of dollars per
6

year.
There are several proposed mechanisms for corrosion protection, one or more of which could be
occurring at any time. The first is a simple galvanic process by which the polymer has a lower
oxidation potential than the metal it is protecting; the polymer is preferentially oxidized. Since
oxidized polymers are usually insoluble and, therefore, do not dissolve away as zinc does,
corrosion protection with conducting polymers should last longer. Another proposed mechanism is
that the polymer reacts with the surface of the metal, requiring that the polymer have an oxidation
potential higher than that of the metal. The surface of the metal reacts with the polymer and forms
a passivating layer which inhibits further corrosion by either setting up a barrier or by changing
the surface potential or both.

N
G
LM

-Z

JU

One possible disadvantage to using polyaniline is that the corrosion protection ability is pH
dependent. In acidic media, polyaniline-coated mild steel coupons corrode 100 times slower than
counterparts, whereas in pH 7 media, the PANI-coated material corrodes two times slower. Since
the pH of seawater is around 8.0 to 9.4 depending upon the season and location, it is unclear and
unproven that PAN1 will provide any additional practical corrosion protection for ocean-going
vessels. This could be explained by the pH dependence of the structure of PANI. At low pH, the
conducting emeraldine salt is the predominant form; at high pH, the non-conducting emeraldine
base is the predominant form. It appears that the conducting form is required for the formation of
the passivation layer. In summary, the amount of corrosion protection is controlled by the type of
polyaniline (emeraldine base vs emeraldine salt) and the characteristics of the corrosion
environment (acidic medium, aqueous sodium chloride, sea water) and also by adhesion to the
substrate. Studies on the marine application of corrosion protection capability of conducting
polymers will need to be performed in solutions isotonic with seawater and/or a salt fog according
to ASTM methods. For corrosion protection, it may be necessary to develop conducting polymers
that do not have the pH dependence of conductivity that PANI has.
1.4.2. Sensors and electromechanical devices

JI

Since conducting polymers change properties by incorporation of ions and solvents (the
property change easiest to measure is conductivity), it is possible to develop and market
ion-specific sensors based upon conducting polymers. Conducting polymers could permit the
incorporation of sensors into clothing. There are some challenges, such as background noise due
to water absorption, lifetime, selectivity and sensitivity. Conducting polymers also change volume
depending upon their oxidation state. Therefore, it is possible for conducting polymers to convert
electrical energy into mechanical work. Conducting-polymer actuators were proposed by
Baughmann and coworkers. Oxidation-induced strain of polyaniline- and polypyrrole-based
actuators has been reported. The first self-contained actuators were reported by MacDiarmid et
al. There are many interesting possibilities for conducting-polymer actuators, but a great deal of
work needs to be done.

1.4.3. Batteries
Batteries were one of the first areas where conducting polymers promised to have a commercial
impact. The general design for a battery is shown in Fig. 6. Batteries have several key components:
the electrodes allow for collection of current and transmission of power; the cathode material
becomes reduced when the anode material is oxidized and vice versa; the electrolyte provides
physical separation between cathode and anode and provides a source of cations and anions to
7

balance the redox reactions. All of these components are critical.

Aside from picking the best conducting polymer available, there are many other issues, not
conducting polymer related, that affect battery performance, such as electrolyte stability and
stability of the counter half-cell reaction (which is at least as important as the conducting polymer
electrode), and compatibility between the electrolyte and the materials.

JU

There was a great deal of initial excitement about conducting polymers as active materials in
batteries. Owing to their low densities, it was thought that batteries with power densities much
higher than those of the ordinary lead/acid battery could be readily obtained. Since the charge on a
polymer backbone is distributed over three or four repeat units, the charge capacity per unit weight
for conducting polymers is marginally better (not orders of magnitude) than for that of metals.
Conducting-polymer batteries were investigated by BASF/VARTA and Allied Signal. Bridgestone
has marketed a button-sized battery using polyaniline and lithium.

N
G
LM

-Z

Conducting polymers do still have potential use in lithium-based high power density batteries,
which use the high potential difference between lithium and the polymer to achieve higher power
densities, although stability and shelf-life are still issues. Also, as more and more of the world
utilizes cellular phones, lap-top computers and cordless drills, the importance of batteries that will
handle many deep cycles (at least 60% depth of discharge) becomes increasingly apparent.
Conducting-polymer-based batteries show promise, but much work needs to be done.

Fig. 1.5. General battery design.

1.4.4. Electrochromic cells

JI

Electrochromic cells are used to go from opaque to transmissive at selected regions of the
electromagnetic spectrum. Batteries and electrochromic cells have many common critical issues
for commercial viability. Electrochromic cells and rechargeable batteries require that the cathodic
and anodic reactions balance nearly exactly (cyclic voltammetry is a good comparison tool for
materials). Operation of an electrochromic window is shown in Fig.1.6.

Fig. 1.6. Electrochromic window operation.

The electrochromic window is similar to a battery with some additional requirements: at least
one of the electrodes must be transparent to the given region of the electromagnetic spectrum; the
cathode material (which colors upon being oxidized) must be electrochemically reversible, the
ion-conducting electrolyte must not only provide physical separation between cathode and anode,
8

a source of cations and anions to balance redox reactions, but must also be transparent to the given
region of the spectrum and the anode material (which colors upon being reduced) must also be
electrochemically reversible.
The ion-conducting electrolyte in electrochromic cells is usually an inorganic salt dissolved in a
solvent such as propylene carbonate with a polymer such as poly(methylmethacrylate) added as a
stiffener. The ion-conducting electrolyte acts as a source and sink for ions as the various redox
processes take place and maintains ionic contact between the materials.
Conducting polymers have also found application in electrochromic cells which attenuate
various regions of the electromagnetic spectrum. Polyaniline has been used in conjunction with
tungsten trioxide (WO3) for materials in an electrochromic windows.

N
G
LM

-Z

JU

Aside from the battery issues mentioned in Section 1.4.3, electrochromic cells have some
additional requirements. Although thinner layers (for optical windows) are usually sufficient,
retention of extinction coefficients and contrast ratios are critical. Many electrochromic cells need
to last more than 10,000 cycles, and have switching times of a few minutes. In this case,
spectroelectrochemistry is a good evaluation tool for conducting-polymer materials.
Spectroelectrochemistry measures both the electrical and optical response of the material in
question. Again, it must be emphasized that spectroelectrochemistry is not a substitute for devices;
one usually performs spectroelectrochemistry in a great excess of electrolyte and, therefore, the
counter half-cell reactions are often ill-defined.
The conducting polymer must retain its extinction coefficients in the transmitting and opaque
state for at least 10,000 cycles in a device. Furthermore, the absorbing and transmitting states must
be complementary to the counter half-cell material; they must both be colored and both be
transmissive at the complementary redox states (for example one colored while oxidized and the
other colored while reduced or neutral).

JI

There is a great deal of data that indicate that conducting polymers are good candidates for
materials in electrochromic cells. In particular, poly(thiophene), poly(pyrrole), and poly(aniline)
have been cycled tens of thousand of times. However, these data, although they are good for
evaluating differences between polymers, may not accurately reflect the performance of the
conducting polymer in a sealed, self-contained device. The reason behind this is that the counter
half-cell reaction in a device must be a well-defined complementary reaction to the polymer
half-cell reaction and that devices often require deep cycles, i.e. near complete
oxidation/reduction. Surveying the literature, devices usually start to degrade at ten to 100 times
fewer cycles than conducting polymers studied by cyclic voltammetry because of these differences.
This does not necessarily mean that the polymer is degrading, as the counter half-cell reaction, or
the limited amount of electrolyte may control the lifetime of the device. Furthermore, cyclic
voltammetry measures the retention of charge capacity in a polymer film, and not the
electrochromic contrast ratio. It appears from some literature that the charge capacity at least
tracks the contrast ratio. However, whereas a 20% drop in charge capacity might be acceptable in
a battery, a 20% decrease in contrast ratio for electrochromic cells may be unacceptable. In
devices that are based upon retention of surface conductivity, it appears that conductivity is lost
long before a significant amount of charge capacity decreases. The evaluation of conducting
polymers for battery and electrochromic applications should be similar to the actual use conditions
(depth of discharge, cycle time, and transmission to extinction ratio).
9

Some very recent results by Reynolds and coworkers have demonstrated over 10,000 deep
cycles (> 95% depth of discharge), using derivatives of PEDOT. Retention of 60% of
electrochromic activity after 16,000 deep cycles (> 95% depth of discharge) using EDOT (counter
electrode) devices, and retention of approximately 60% of electrochromic activity after 10,000
deep cycles using devices with complementary EDOT polymer derivatives has been observed.
4.5. Controlled-release applications
Another application for conducting polymers is controlled-release devices. Ions can be
selectively released, as well as biologically active ions such as adenosine 5-triphosphate (ATP)
and Heparin. A conducting polymer (polymer A) with a given oxidation potential is
electrodeposited onto a substrate with a mobile counter-ion. Another polymer layer (polymer B
with a higher oxidation potential than polymer A) is electrodeposited (directly on top of polymer
A) using an immobile counter-ion (polyanion). During complete reduction (a), it is

JU

difficult/impossible for the polymeric anion (Y) to move into the electrolyte solution and,
therefore, cations (M+) from the solution must move into the outer polymer layer, but the mobile

N
G
LM

-Z

anions (X) from the inner layer move into the electrolyte. During selective oxidation of polymer
A (b), the mobile anions (X) move from the solution to the inner layer. During oxidation of
polymer B (c), the cations (M+) move back into solution. This potential-dependent ion transport is
an interesting way to deliver ionic drugs to certain biological systems. One can deliver exclusively
anions by cycling back and forth through step (b), exclusively cations by cycling back and forth
through step (c), or deliver anions and receive cations by cycling through step (a).

JI

Fig. 1.7 shows a general schematic diagram showing selective ion transport of an electroactive
bilayer.

Fig. 1.7. Selective ion transport of an electroactive biiayer.

It is very important for the inner polymer layer to have a lower oxidation potential than the
outer layer. If this is not the case, then the inner layer, that is the layer connected to the electrode,
will act as an electrical insulator and prevent the oxidation of the outer layer at potentials below
the oxidation potential of the inner layer. Once the oxidation potential of the inner layer is reached,
a pulse charge will occur, making selective ion transport difficult (rectification). Furthermore, a
biologically compatible counter half-cell is necessary for a practical device. This might not present
10

too much of a problem if the drug delivery system is used once only. If repeated uses of the same
device are needed, then issues such as reversibility (like those for batteries and electrochromics)
will need to be resolved.
1.4.6. Radar applications

JU

Radio direction and ranging (RADAR) uses electromagnetic waves that bounce off a particular
target and are collected by a receiver which analyzes the signal and determines the range direction
and speed of the object in question. Reflections occur whenever there is a sharp impedance
difference between the medium (usually air) and the object. Impedance differences are most
notable between metals and air. Metals tend to re-radiate (reflect) the incoming signal.
Conducting-polymer camouflage works a little differently, in that it reflects back in a way that is
different from the object it covers and absorbs microwave radiation because it has a more
continuously variable impedance. A conducting-polymer textile used for camouflage has no sharp
edges or wings and tends to appear indistinguishable from the surrounding hills and trees, and
absorbs more than 50% of incident microwave radiation.

N
G
LM

-Z

Microwave (100 MHz to 12 GHz) properties of conducting polymers have been studied as have
the millimeter wave (24-40 GHz) properties of polypyrrole-coated fibers. Conducting polymers as
radar absorbers in antennas, Salisbury screens, camouflage, and other types of shielding are of
interest to the military.

1.5. Devices and Applications of the Neutral (Non-Conducting) Form of

Conjugated Polymers
5.1. Third-order non-linear optical polymers

The understanding of third-order optical non-linearity, particularly the comprehension of

structure-property relationships, can be best described as limited. One relationship that is known
well, however, is that materials with highly delocalized p-electrons, conducting polymers, will
show measurable non-resonant third-order optical effects.

JI

Owing to the p-electron polarizability in conjugated polymers, the effect manifests itself in third
harmonic generation and/or the change of the index of refraction dependence with light intensity.
These are called third-order non-linear optical polymers. Thus, a light beam in such a material can
be modulated by a second light beam. Also, devices could be designed such that they become
reflective at very high light intensities, which would make them useful for eye protection from
lasers. There was a flurry of initial activity in this area in the mid- and late-1980s. The third-order,
or 3, properties of many materials, PA, PP, PT, poly(acenequinone)s PPV, and alk
oxy-substituted PPV modified with buckminsterfullerene have been studied extensively.

A major challenge was that the value of non-resonant 3 that is required for a practical
demonstration device was 100 to 1000 times greater than the highest reported value for
non-resonant 3. Only recently have devices that demonstrate all-optical switching at I .6 pm using
polydiacetylene (opaque to visible light) been reported. A figure of merit that was widely accepted
was the intensity dependence of refractive index n2 divided by optical absorption a, which was
developed for laser eye protection applications. This figure of merit quantifies the light intensity
needed to achieve reflection of the incoming beam vs the optical loss (optical clarity). Eye
protection devices built with conjugated polymers today do not provide sufficient protection, since
the intensity at which reflection occurs would be well above the damage threshold for the human
11

eye, and most devices are too opaque to visible light, but progress is being made.
However, it should be noted that the understanding of 3 processes in materials and polymers
has just got beyond its infancy. Perhaps, when greater understanding comes, materials (hopefully
conducting polymers because of the synthetic degrees of freedom inherent in conjugated polymer
systems) can be designed to give 3 values that would yield practical devices.
1.5.2. LEDs
A significant event occurred when Friend and coworkers published electroluminescence studies
on the neutral (non-conducting) form of poly(para-phenylene vinylene). This work has opened up
a new avenue of research and, more importantly, a potential market for the material.

JI

N
G
LM

-Z

JU

A general design for LEDs is shown in Fig. 1.8. The electron-injecting electrode is usually a
low work function (easily oxidized) metal and the hole-injecting electrode is a high work function
metal, or indium tin oxide, or a conducting polymer with an oxidation potential higher than that of
the active layer. Some general trends that have been observed are that as one uses a lower work
function metal (less stable because it is easily oxidized) the efficiency increases but the lifetime
decreases; if a higher work function metal is used the efficiency decreases but the lifetime
increases. Studies have been reported in which a layer of neutral conjugated polymer with a
reduction potential closer to zero than that of the active conducting-polymer layer increases the
efficiency.

Fig.1.8. General design for LEDs.

A simplistic overview of the function of an LED is as follows. An electron is injected into the
polymer from the cathode while a hole is injected from the anode. One can envision there to be
oxidized polymer on one side of the polymer film and reduced polymer on the other side. The hole
and the electron then migrate towards the center of the film, and when they meet each other, they
recombine and give off light. The frequency of light emitted is roughly equal to the difference
between the oxidation and reduction potential of the polymer (the electrochemical band gap), and,
therefore, is related to the electronic band gap. Polymers with different band gaps have different
values for the difference between oxidation and reduction potentials and give off different
wavelengths of light. LEDs have been constructed using PPV derivatives, PT derivatives, and PPP
derivatives. Several articles on conducting-polymer LEDs and the effect of various additives,
electrode modifications, tuning emissions, the effect of impurities and discussions of hole
12

tunneling, photoexcitation and unusual symmetric bias have been published. The efficiency of
LEDs is constantly being improved, a long with novel developments such as flexible LEDs,
polarized-light emitting LEDs, and light emitting electrochemical cells. The emission of red, green,
blue and white light have all been demonstrated, as have brightnesses on the order of 400 cd m1,
which are similar in brightnesses to fluorescent lights or computer displays.

-Z

JU

A challenge in the operation of LEDs is the fact that it appears that the mobility of rhe hole is
higher than the mobility of the electron. The barrier height (like a resistance) between the polymer
and each of the electrodes must be low and roughly equal, so that the hole and the electron
recombine near the middle of the conducting-polymer layer to insure good operation. Another
limit is the competition between radiative and nonradiative decay, which for PPV-type systems is
around 25%, which does not limit device performance at this juncture. Yet another challenge is
the fact that polymer that initially perform well in LEDs contain electron-rich double bonds. These
double bonds are fairly easy to oxidize and are most likely one of the major causes for device
degradation. Thermal and photo-oxidation of the double bonds in LED polymers have been
discussed in detail. It is interesting to note that this degradation mechanism has also been observed
and shown to degrade device performance in second-order non-linear optical polymers.

N
G
LM

The efficiency, lifetime and brightness depend upon a variety of factors. One major challenge is
balancing the electron mobility to that of the hole mobility. This is done by adding electron
transport layers and hole transport layers such as trisubstituted amines. These layers reduce the
barrier height and encourage the holes and electrons to combine near the center of the film.
Furthermore, electron and hole transport layers could permit the use of more stable metals without
compromising efficiency.
Conducting polymers enable a wide variety of structures to be synthesized, and, therefore, many
different wavelengths of light are possible. Whereas conducting polymers will not replace
fluorescent lightbulbs (which have efficiencies around 70%), because conducting polymer LEDs
are easily patterned, operate at low DC voltages, and have uniform areas of light, there is a
significant potential market for low level backlighting, and alphanumeric displays.

1.6. GOALS FOR THE FUTURE AND SPECULATION

JI

Funding for research in the conducting polymers peaked some time around the mid- 198Os,
then fell in the late-1980siearly-199Os, and is currently, in the view of this author, experiencing a
level amount of support. It appears that the number of researchers in the academic arena has
remained roughly constant, where the number of industrial researchers has decreased significantly.
New and exciting applications for conducting polymers occur almost biennially, so there is no
lack of possibilities. Conducting-polymer research is, however, at a critical juncture between
laboratory curiosity and commercial viability. Although conducting-polymer research will not
fade away entirely, the amount of financial support will depend upon evolutionary development
and commercialization.
The current climate of frugality and the waning of public interest in basic science has made it
difficult for all but a few researchers to sustain robust programs in the area of conducting
polymers. Most sponsored work comes in the area of government grants and/or cost sharing
contracts. There are few industrial sponsors of purely synthetic work (synthesis of new conducting
polymers) in the world. The number of industrial sponsors will increase as revenue (from
consumers) generating conducting polymer businesses are created.
13

JU

The key to conducting-polymer researchs survival into the next millennium will depend upon
entrepreneurial companies. These are companies with a few employees, usually former/current
researchers in conducting polymers. These small companies could find and fill niche applications
for conducting polymers that larger companies can not or will not do. Furthermore, because small
companies can survive for several years on government contracts (such as National Science
Foundation, DARPA, Office of Naval Research, Air Force Office of Scientific Research) these
companies could afford to be more long-term future oriented than their larger counterparts;
projects could continue over 5 to 15 year spans without being stopped for failing to produce a
product. However, the long-term health of conducting polymers requires more weal & generating
industries based upon conducting polymers. A company that receives revenue from conducting
polymers is very likely to sponsor many types of conducting-polymer research. There are several
examples of the many entrepreneurial companies already working in the area of conducting
polymers. These companies, and companies like them, are the linchpin to commercialization of
conducting polymers.

-Z

Much of the progress being made in conducting-polymer research today is evolutionary. One
could argue that the development of new applications, such as conducting-polymer LEDs, is
revolutionary in nature; however, there are still many challenges to be met that will probably
require years of evolutionary progress.

N
G
LM

Towards this end, I have summarized a wish list of material/device properties that, if met,
could be sufficient for immediate commercial applications.
Conducting polymers that are environmentally stable and solution-processable with
reproducible conductivities around 10,000 S cm1 per pound to be used for weight-limited
replacement of metals (perhaps on ships, missiles, aircraft. or on some selected electronic
devices).
In areas where space/volume is limited (in tightly packed circuit boards) conductivities need to

JI

be at least as good as copper (50,000 S cm1). One property of conducting polymers that might
actually work to their advantage is the fact that the conductivity of most conducting polymers
increases with temperature. It is not clear that polymers with such high conductivities will have
the same temperature dependence of conductivity as most conducting polymers. If this is true,
however, circuits designed with high conductivity polymers will tend to stabilize themselves, i.e.
as the circuit heats up due to current flow, the resistance will decrease and the circuit would settle
in to some steady state value.

Another area where conducting polymers could have an impact is in the area of electrochromic
cells. For many commercial applications (such as smart windows for homes) conducting polymers
meet or nearly meet the requirement of transmission, switching speed (seconds are preferable, but
minutes are certainly acceptable), and contrast ratio. What is needed are devices that withstand
100,000 or more deep cycles, especially if the smart window is on a commercial building. Cyclic
voltammetry is a key indicator as to how many cycles are possible with a given polymer, but
actual devices require well-defined redox reactions at both electrodes, since reactions at the
counter-electrode in cyclic voltammetry are often ignored.
Some other devices of interest are dependent upon conductivity differences between the
oxidized state and the neutral state. This criterion is even more stringent than those for absorbance
or charge retention, since conductivity depends on both the material and the macroscopic
14

properties

of

the

polymer

film.

Stable,

processable,

reproducible

(manufacturable)

electrodeposited conductivities of 1000 S cm or greater for film thicknesses of about 10 pm

would make a significant impact. It has been observed that solution-cast conductivities are much
higher that electrodeposited conductivities. In the device, an initial reading (conductivity) may be
high from the solution cast film, but once the film has be neutralized and reoxidized, the
conductivities are usually much lower, due probably to the ions and some residual solvent moving
in and out of the film, 17%IX0 Once the film has been neutralized and oxidized a few times, the
conductivity in the conducting state remains constant for several hundred cycles.

JU

For LEDs, it is generally accepted that the efficiencies (up to 4% to date) of polymer base
LEDs give acceptable brightness. Unfortunately, the heat build-up from polymer-based LEDs
severely limits the lifetime of the device, and means that efficiencies will have to be improved up
to the theoretical maximum of 25%. Further research/advanced development in manufacturing
techniques that will give inexpensive and stable large-area devices is also required. These
challenges are significant and will require years of development to overcome.

-Z

Another area where conducting polymers have great potential is in corrosion protection. If
superior performance of conducting polymers can be demonstrated under use conditions (which
vary greatly depending upon use and environment), there is a potential billion dollar industry.

N
G
LM

In summary, there are a variety of applications that have commercial potential. Some of the
challenges are related more towards device operation, and not necessarily to the material
properties of conducting polymers. These device-based challenges, nevertheless, need to be
overcome. It will take perseverance, ingenuity and dedication (luck is, of course, always helpful)
to overcome these challenges and make a self-sustaining conducting-polymer industry.

JI

Ref. John D. Stenger-Smith, Prog. Polym. Sci, 23, 57-79(1998)

15

Chapter 2
Ion-Conducting Polymers

2.1. Introduction
2.2. Nonflowing Polymeric Liquids
2.3. Ionic Migrations in Polymers
2.4. Carrier Generation in Polymers

2.7. Applications of Ion-Conducting Polymers

2.1. Introduction

-Z

2.6. Superion-Conducting Polymers

JU

2.5. Ion Transport Number in Polymers

N
G
LM

Electrolyte solutions are ionic conductors. The dissociation of electrolytes into fast-migrating
carrier ions causes electrolyte solutions to have a high ionic conductivityThis also makes
electrochemlcal reactions possible when a certain potential difference is applied between two
appropriate electrodes in the solutions. A major potential gradient exists at the electrode-solution
interface. The objective of this chapter is to describe what happens when macromolecules are used
as solventsfor the electrolyte instead of low molecular weight solvents.

JI

The high ionic mobility (10-4 cm2 V-1 s-1) in electrolyte solutions is dependent on the low
viscosity of the solvents. Fast ion transport cannot be expected in crystalline solids and glassy
solids unless these solids have a special structure that will enable ion transport. In contrast, under
certain conditions, macromolecules can exist in a characteristic elastomeric state. Inorganic
glasses and low molecular weight organic glasses begin to flow above the glass transition
temperature (Tg). For macromolecules In the elastomeric state above Tg, polymer segments are
locally mobile (i.e., exhibit micro-Brownian motion), similar to a liquid, whereas long-range
motion (macro-Brownian motion) is restricted by entanglement of the polymer chains, existence
of crystallltes, or chemically produced cross-links. In other words, the macroscopic viscosity or
rigidity of elastomers is high, whereas their microscopic viscosity is low, similar to a liquid.
Certain kinds of polar polymers can dissolve inorganic salts up to a high concentration. In these
solid solutions, a part of the dissolved salt is expected to dissociate into ions, just as in liquid
electrolyte solutions. If the micro-Brownian motion of polymer segments allows ion transport, the
system may be described as a polymer electrolyte (i.e., solvent-free ion-conducting polymer).
Such polymer electrolytes contain dissociation ions in the amorphous elastomeric phase, and the
ionic conduction in this phase is predominant. Angell called polymer electrolytes super-Tg
materials, which means that their application range lies above Tg.
Studies of polymer electrolytes began with the pioneering works of Wright et al. and Armand et
al. Since then, several studies have been devoted to the fundamental understanding of ionic
conduction in polymer electrolytes and to further their application as solid electrolytes in solid
16

electrochemical devices, especially in rechargeable lithium solid-state batteries. In this chapter, we

summarize the correlation between the characteristics of polymer electrolytes and their ionic
conductivity in terms of carrier transport and generation processes. Our recent studies on the use
of polymer electrolytes as media for electrochemical reactions are also presented.

2.2. Nonflowing Polymeric Liquids

Most linear polymers become viscous liquids on heating. Low molecular weight amorphous
polymers that have a Tg well below room temperature become liquid at room temperature. We
have chosen non stereoregular poly(propylene oxide) (PPO) as a model for ionic conduction in
amorphous polymers. PPOs with molecular weights of 10-10, which are liquid at room
temperature and can dissolve alkali metal salts in high concentrations to form viscous electrolyte
solutions. The ionic conductivity of PPOs (Mw = 400-3000)-LiClO4 increases with increasing salt

JI

N
G
LM

-Z

JU

concentration, and reaches 105 to 106 S cm1 at 25C, as shown in Fig.1.This means that LiClO4
dissociates in PPO to generate carrier ions. Simultaneously, the dissolution of LiClO4 in PPO
remarkably increases the viscosity. This behavior becomes more pronounced as the molecular
weight of PPO is increased. Large viscosity increases are not found in low molecular weight
solvents. PPO contains polar groups, and LiClO4 dissolved in PPO seems to dissociate into carrier
ions by interaction of Li+ with the polar ether groups. If one considers the molecular motion or
PPO, there will be both a macroscopic motion as in low molecular weight solvents, and
microscopic motion of polymer segments, which is independent of macroscopic motion. If the
latter motion mediates the ionic migration, solid polymer electrolytes may be attained when higher
molecular weight polymers are used. In the viscous PPO-LiClO4 solutions, the viscosity is
affected both by dissolution of the salt and by temperature; however, the two effects should be
distinguished when considering the ionic conductivity.

Fig. l. Ionic conductivity of poly(propylene oxide) solutions of LiClO4 as a function of salt concentration at 25C.
Molecular weight of PPO dids: , 420; , 1160 ,202O; , 2880.

The PPO solutions of electrolytes can become solid polymer electrolytes by making the
polymer structure cross-linked or by incorporating (by copolymerization) another polymer
segment with a high-melting point or a high Tg. Examples of the former type are network
polymers made from PPO triol with diisocyanate or diacid dichloride in which the electrolytes are
dissolved. Examples of the latter type are block copolymers consisting of PPO and poly(urethane
urea) segments. Figure 2.2 shows the ionic conductivity as a function of salt concentration for the
polymer electrolytes consisting of PPO network polymer and LiClO4. The room temperature ionic
conductivity was approximately 107S cm1. This conductivity was comparable with that of the
17

JU

liquid polymer electrolyte solutions (see Fig.1); accordingly, it is more or less Independent of the
macroBrownian motion. Thus, the micro-Brownian motion of polymer segments is the prime
motive force of ionic transport in polymer electrolytes.

-Z

Fig.2. Ionic conductivity as a function ot salt concentration for polymer electrolytes consisting of poly(propylene
oxide) networks of LiClO4.

N
G
LM

Arrhenius plots of the ionic conductivity of amorphous polymer electrolytes, such as

PPO-based electrolytes, frequently do not lie on a simple straight line, but rather, on a positively
curved line (Fig.3). Such curves are well represented by a Williams-Landel-Ferry (WLF) type
equation:
log(T)/ (Tg) = C1(TTg)/[C2 + (TTg)]

JI

where (T) is the ionic conductivity at temperature T. The WLF parameters representing the ionic
conductivity data in Fig.2.3 are summarized in Table 1. The WLF equation is an empirical one that
interprets the temperature dependence of dielectric or mechanical relaxation times for the local
main-chain motion of many amorphous polymers, independently of their chemical structures. A
theoretical basis of this equation has been proposed by Cohen and Turnbull (free volume theory)
and by Adam and Gibbs (configurational entropy theory). That the temperature dependence of the
ionic conductivity obeys the WLF equation suggests that the change in ionic conductivity with
temperature is mainly caused by the charge in ionic mobility which, in turn, is influenced by the
local segmental motion.

18

Fig.3. Temperature dependence of ionic conductivity for polymer electrolytes consisting of poly(propylene oxide)
networks and LiClO4.

JU

Table 1. WLF parameters for ionic conductivity of polymer electrolytes from PPO network and LiClO4.

-Z

The ionic conductivity of many amorphous polymer electrolytes frequently exhibits a maximum
at a certain salt concentration (see Fig.2.2). As can be seen in Tablethe Tg of the polymer
electrolytes rises considerably as the salt concentration is increased, This behavior is quite general
in salt-containing polymers. It is believed that the number of carrier ions increases with the salt
concentration, but that there Is also a simultaneous decrease in ionic mobility owing to the

N
G
LM

increase in Tg for the most important term in the WLF equation is TTg. The conductivity
maximum is thus the consequence of these two opposite effects on ionic conductivity.
The foregoing interpretation is based on the assumption that the ionic conduction in polymer
electrolytes takes place in the amorphous phase. In contrast with PPO, poly(ethylene oxide) PEO,
which is a prototype matrixfor forming polymer electrolytes, is a highly cryslaulne polymer. Also,
PEO forrms crystalline complexes with alkali metal salts, which have considerably higher melting

JI

points (about 180C) than that of pure PEO (about 60C). Thus, the PEO-alkali metal salt
electrolytes have, in general, a multiphase nature, consisting of a salt-rich crystalline phase, a pure
PEO crystalline phase, and an amorphous phase with the dissolved salt. Ionic conduction takes
place primarily in the amorphous phase. The phase diagram is affected by many factors, such as
the salt species and its concentration, the temperature, thermal history, and preparative methods.
As a result, the ionic conductivity is complicated and is dependent on the change in the phase
diagram and temperature and time. This has been a serious problem when attempts are made to
attain high and stable ionic conductivity in PEO-based electrolytes. Figure 2.4 shows the
temperature dependence of ionic conductivity for polymer electrolytes consisting of a PEO
network polymer and LiClO4. The cross-linked structure caused a considerable decrease in the
degree of crystallinity of the PEO-LiClO4 polymer electrolytes. This decrease allows a higher
ionic conductivity to be attained; close to 105 S cm1 at 30C. This conductivity is higher than
that of crystalline linear PEO-LiClO4 electrolytes by three orders of magnitude. The sudden
decrease in conductivity for the low LiClO4 concentration electrolytes at temperatures lower than
10C (see Fig. 2.4) also corresponds to the crystallization of PEO segments. The decrease in ionic
conductivity that accompanies crystallization has been interpreted as follows: The crystallization
of PEO segments causes exclusion of the incorporated salts from the crystallites. Thus, the salt
concentration in the amorphous region becomes higher than that of the overall value. The
enhanced concentration in the amorphous region causes a considerable increase in Tg, which

19

N
G
LM

-Z

JU

reduces the ionic mobility. Furthermore, the interface between the crystallites and amorphous
region functions as a deep trap for ionic migration because the motion of the segments linked to
the crystallites is considerably suppressed. The crystallites also sterically hinder the ionic
migration. These three effects are considered to be responsible for the lower ionic conductivity of
crystalline PEO electrolytes, compared with that of amorphous ones.

Fig.4. Temperature dependence of ionic conductivity for polymer electrolytes consisting of poly(ethylene oxide)

JI

networks and LiClO4.

Fig.5. Temperature dependence for the reciprocal of the dielectric relaxation time (T) at temperature T for
poly(propylene oxide) networks, and its fit to the WLF equation.

2.3. Ionic Migrations in Polymers

The correlation between the ionic migration and the local segmental motion of polymer chains
may be subjected to closer examination. According to the free volume theory, the free volume in
amorphous polymers that occurs above Tg increase in accordance with the temperature difference
T-Tg. Because of the enlarged free volume above Tg, the polymer chain becomes locally mobile.
The mode of the micro-Brownian motion is considered to be cooperative, involving several repeat
20

units of the polymer chains. The dielectric relaxation time for the backbone motion of PPO and
PEO, which are typical matrices for polymer electrolytes, has been reported to be 109 to 108 s at
room temperature, and its temperature dependence obeys the WLF equation. These features are
shown in Fig. 2.5. The temperature dependence of the inverse of the dielectric relaxation time (T),
owing to the backbone motion of the PPO network polymer, obeys the WLF equation shown in
this figure. How small ions migrate in these rubbery media is an interesting question. The
percentage change in the conductivity with temperature is comparable with that in the dielectric or
mechanical relaxation time of the backbone motion for the PPO and PEO-based polymer
electrolytes, when Tg is used as reference temperature. A typical result is shown in Fig.2.6, in
which the ratio of ionic conductivity at T, (T), to that at Tg, (Tg), and the ratio of mechanical
relaxatlon time at T to that at Tg, the so-called shift factor aT, are plotted against T-Tg for the
PPO-based block copolymers containing dissolved LiClO4.

JI

N
G
LM

-Z

JU

PPO electrolytes cross-linked by an azobenzene derivative have been prepared to gain an

insight into the ionic transport process. The ionic conductivity, the thermal cis-trans-isomerization
rate of the azochromophores, and the dielectric relaxation time have been studied for the salt-free
network polymers and the PPO-LiClO4 electrolytes in the rubbery state. The ionic conductivity for
the electrolytes showed a WLF-type temperature profile and was influenced by the increase in Tg
caused by the increases in both LiClO4 concentration and cross-linking density. The temperature
dependence of the dielectric relaxation time also showed WLF behavior, and its magnitude
increased with increasing cross-linking density. In contrast, the thermal isomerization rate for the
azochromophores in the salt-free network polymers and the PPO-LiClO4 electrolytes was hardly
changed by the increase in Tg owing to the increases in both LiClO4 concentration and
cross-linking density. These results suggested that the average hole size (free volume) in the
matrix was larger than the average hole size required to allow thermal isomerization and smaller
than the average hole size for ionic transport.

Fig.6. Correlation between logaT or log[(T)/(Tg) and T-Tg for poly(propylene oxide)-segmented poly(urethane
urea)] with dissolved salt at two concentration: [LiClO4]/[PO unit] = 0.062; (T)/(Tg); aT; [LiClO4]/[PO unit]
= 0.040; (T)/(Tg); aT.

Figure 7 shows the WLF plots of ionic conductivity for amorphous PEO networks containing
11 different alkali metal salts at the same concentration. Although the conductivity at a constant

21

-Z

JU

temperature differed at most by two orders of magnitude, the WLF plots were represented by one
master curve irrespective of the salt species, and the WLF parameters obtained were comparable
with the universal values for the relaxation times of many amorphous polymers.

Fig.7. WLF plots of ionic conductivity for polymer electrolytes consisting of poly(ethylene oxide) networks and

N
G
LM

alkali metal salts using a reference temperature T0 = Tg + 50 K.

The foregoing results at first seem to be rather peculiar when one realizes that the volume of
carrier ion may be far smaller than that of the moving unit of polymer chain involved in the
relaxation process. This may seem more reasonable when one considers that the ionic transport in
polymer electrolytes does not occur by itself, but that the segmental motion with associated carrier
ions causes the ionic transport. In other words, ions in the polymers are solvated by the polymer
chain, and these solvated ions move by the assistance of the segmental motion.

JI

In the low molecular weight solvents we have the concept of solvated ion radius. If we apply
the same concept to the ions in polymers, the solvated ion radius may be considerably larger than
the naked ion radius. The size may be closely correlated with that involved in the relaxation
process, because in the polymeric solvents, polar groups that interact with ions are covalently
linked together, and independent motion of one polar group in the polymer chain is impossible.
Because the moving PEO unit with associated carrier ions is far larger than the volume of naked
ions, and its size is nearly constant and independent of the ionic radii, the master curve in Fig.7
may be accounted for irrespective of the nature of the incorporated salt.

Carrier ions in polymer electrolytes are thus ions solvated by polar groups of polymers. The
polymer segments with associated ions constantly rearrange because the polymer electrolytes are
in a rubbery state. This rearrangement changes the local position of the carrier ions. For ions to
migrate over a long distance, such as from one electrode to the other, the ions must be passed from
one polymer segment to another. The repeated association of the carrier ions with the polymer
segments, the segmental motion with associated ions, and the dissociation of the ions from the
polymer segments seem to be responsible for ionic transport in polymer electrolytes. The
segmental motion with associated carrier ions appears to be the rate-determining step for ionic
conduction. Thus, the WLF profile of the ionic conductivity indicates a large temperature
dependence for transporting small ions that is comparable with the change in the segmental
22

mobility with temperature. These characteristics are in strong contrast to the temperature
dependence of the diffusion coefficient of gas molecules in elastomeric polymers, for which the
temperature dependence is very small, even though the volume of gas molecules, such as helium
and hydrogen, is larger than or comparable with that of naked alkali metal ions. The strong
interaction of ions with polymers thus makes the solvated ion radius very large. The mobility of
ions migrating in the PPO and PEO network electrolytes was roughly 10-7 cm2V-1s-1 at room
temperature, as determined by transient ionic current methods.

N
G
LM

-Z

JU

These considerations lead us to the following predictions: first, the polymers that have a Tg as
low as possible, in addition to salt-solubilizing properties, are effective for the attainment of high
ionic conductivity in polymer electrolytes; second, polymers that give a small solvated ion
radius will favor the attainment of high ionic conductivities, having a small temperature
dependence. Much effort has been focused on the former prediction. The flexible phosphazene
polymers having ether groups in their side chain have been prepared. Their Tg was lower than that
of linear PEO, and a high ionic conductivity, of about 10-5 S cm-1 at room temperature, was
attained. The introduction of dimethylsiloxane units has been tried by several laboratories. In
poly[dimethylsiloxane-co-oligo(ethylene oxide)] electrolytes, the Tg of the copolymer decreased
with increasing proportion of dimethylsiloxane units. However, at the same time, the
salt-solubilizing property of the copolymer was lowered. As a result, the ionic conductivity
attained at a certain dimethylsiloxane composition. On the other hand, the network polymer
electrolytes from poly[dimethylsiloxane--poly(ethylene oxide)] formed a microhetero-geneous
structure from the constituent segments, an the incorporated LiClO4 preferentially interacted with
the poly(ethylene oxide) segments. As a result, the segmental motion of poly(ethylene oxide)
appeared to contribute to the ionic migration, whereas that of poly(dimethylsiloxane) did not.
Thus, the reduction of Tg of polymer electrolytes must be carefully controlled, such as the
salt-solubilizing property of the polymer is not lost, and a microheterogeneous structure is not
formed.
Some

experiments

have

been

performed

to

demonstrate

the

second

prediction.

JI

Poly[-methoxy-oligo(ethylene oxide)-L-glutamate] (PMnEOG, in which n is the number of EO

units) was selected as a matrix for salts. The main chain of the poly(-amino acid) polymer forms
a rod-like helix structure, on the side of which flexible oligo(ethylene oxide) chains may be
attached. In this polymer the segmental motion of the side chain is independent of that of the main
chain. Thus, if the side chain motion mediates the ionic transport, the temperature dependence of
the ionic conductivity might become larger as the side chain is lengthened. Figure 8 shows the
WLF plots for the ionic conductivity of PM2EOG and PM3EOG with dissolved LiClO4, for which
the reference temperature was 50C above Tg of the side chain. It is clearly seen that the
temperature dependence is larger in the PM3EOG electrolytes than in the PM2EOG electrolytes.
This result implies that the ionic transport was mediated by the side-chain motion and the solvated
ion radius was smaller for the shorter side chain. The concept of the solvated ion radius in
polymers was presented for the first time in this article, and further experiments demonstrating the
importance of this concept are now in progress.
Most of the polymer electrolytes that have been studied are solid solutions of salts in polymers.
There is a possibility that both the cation and anion are mobile in such electrolytes. The ionic
transport number in polymer electrolytes is a very important parameter in terms of the conduction
mechanism of ions in polymers and of their practical application. Cationic transport numbers have
23

N
G
LM

-Z

JU

been measured in the polymer electrolytes, especially in those of PEO using various methods,
including nuclear magnetic resonance (NMR), complex impedance measurements, tracer diffusion
experiments, limiting current measurements, the Tubandt method, potentiometric measurements,
transient ionic current measurements, and complex impedance measurements, in combination with
polarization experiments. The data obtained indicate that transport numbers may vary from zero to
nearly unity; it has become apparent that both cations and anions are generally mobile in polymer
electrolytes. Electrolytes in which anionic or cationic sites are fixed to the polymer backbone have
also been prepared. In these polymer electrolytes, the cationic or anionic transport number is
necessarily unity, because the opposite charge is fixed to the polymer backbone.

Fig.8. WLF plots of ionic conductivity for polymer electrolytes consisting of poly[-methoyoligo(ethylene
oxide)-L-glutamate] (PMnEOG: n = number of EO units = 2 and 3) and LiClO4 using a reference temperature T0 =

JI

Tg + 50 K.

Fig.9. Ionic conductivity at constant reduced temperature as a function of the lattice energies UL of incorporated
salts for polymer electrolytes based on PEO networks.

2.4. Carrier Generation in Polymers

The ionic conductivity of polymer electrolytes is affected not only by the ionic mobilitybut
also by the number of carrier ions. It is unlikely that salts incorporated at such high concentrations
24

dissociate completely in these media of relatively low polarlty. Thus, a part of an incorporated salt
may dissociate to free ions, which function as carrier ions, but we cannot deny the possibility of
the presence of aggregated ions, so-called ionic multlplets. The energetics for the formation of
ionic aggregates has been studied as afunction of the dielectric constant of the solvent and of the
salt concentration in electrolyte solutions. If the incorporated salts are partially dissociated, the
number of carrier ions will be influenced by the incorporated salt species, its concentration,
temperature, and the polymer structure.

N
G
LM

-Z

JU

The ionic conductivity for the amorphous PEO and PPO electrolytes, based on network
polymers at a constant salt concentration, has been investigated as a function of the lattice eneegy
of the incorporated salts at the iso-free-volume state. Figure 9 shows the ionic conductivity of the
PEO network electrolytes as afunction of the lattice energy of salts. The ionic conductivity
decreased with increased with increase in the lattice energy. This characteristic was also observed
in PPO electrolytes. Thus, a part of the incorporated salts seems to dissociate to carrier ions, and
the degree of dissociatlon is suppressed with increasing lattice energy of the salt. At the
iso-free-volume state, the increase in ionic conductivity of the PEO network electrolytes as a
function of salt concentration was smaller than that expected for complete dissociation (Fig.10).
This indicates that not all of the incorporated salt functions as carrier ions. The formation of ion
pairs and other aggregates seems to occur at the higher concentrations. We cannot now deny the
possibility of long-range ion interaction that reduces the ionic mobility at constant-reduced
temperatures with increasing salt concentration.

JI

The temperature dependence of the number of carrier ions is another complicated problem.
Direct measurements of the carrier ions number of or their mobility are very difficult to make. The
mobilities of both cations and anions are required, however, to make a quantitative estimate of the
number of carrier ions. An estimation of the ionic mobility has been made on the cross-linked PPO
and PEO electrolytes containing LiClO4. The ionic mobility was calculated from the current peak
time, observed in the current-time curve of the electrolytes with ion-blocking electrodes after
applying a constant voltage in one direction for an appropriate period and reversal of the applied
voltage polarity. The current peak time was assumed to be consistent with the migration time of
carrier ions from one electrode to the other. The theoretical background of this method has been
developed using a simple model. However, the initial conditions and the situations of the actual
measurements are very different from those of the model system. Therefore, the mobility values
obtained should be considered only apparent values. Nevertheless, the following results were
obtained: the temperature dependence of the ionic mobility obeyed the WLF equation; the change
in the mobility with temperature was comparable with that in the ionic conductivityand the
number of ionic carriers, which was calculated from the data of ionic conductivity and mobility,
was far smaller than that based on tpe assumption ofcomplete dlssoclatlon. The temperature
dependence of the number of carrler ions obeyed the Arrhenlus equationthe activation energy
was very small in the PEO electrolytes, but that in the PPO electrolytes was not negligible (about
0.3 eV). That the Increase in the ionic conductivity in PPO electrolytes, as a function of salt
concentration at the iso-free-volume state, is frequently larger than that expected for cornplete
dissociation might be explained by the relatively large temperature dependence of the number of
carrier ions in this kind of electrolyte. This is in sharp contrast with the PEO electrolytes.
Sodium-23 NMR measurements of the sodium salt-containing polymer electrolytes showed the
coexistence of two separate lineshape components; one was attributed to bound Na+ species and
25

-Z

JU

the other to mobile Na+ ions. The ratio of the nonbound species to the bound ones showed a small
temperature dependence, but this dependence was far smaller than that in the ionic conductivity. It
is plausible to suppose that the temperature dependence of ionic conductivity is dominated by that
of ionic mobility; however, there might be some examples for which the temperature dependence
of the number of carrier ions is not negligible.

Fig.10. Ionic conductivity as a function of salt concentration for electrolytes consisting of PEO networks and
LiClO4 at constant-reduced temperature.

JI

N
G
LM

Various polymers have been tested as matrices for polymer electrolytes: polyethers (such as
PEO and PPO, as already discussed), polyesters [poly(ethylene succinate), poly(-propiolactone),
and poly(ethylene sebacate);polyiminespoly(ethylenimine) and poly(N-substituted-ethylenimine),
and poly(alkylene sulfides). The conductivity results show that the carrier generation in these
polymeric electrolytes cannot be simply correlated with the macroscopic dielectric constant of the
polymer. For ion dissociation in polymeric media, solvation of an ion with polar groups in the
polymers may be an essential process. Tpere are two possible ways in which the polar groups may
solvate the ions: one involves solvatlon by polar groups within a given chain (intrapolymer
solvation); and the other involves solvation of a given ion by polar groups from different chains
(interpolymer solvation). The former may involve the cooperative interaction of the neighboring
polar groups with an ion.

Fig.11. Comparison of ionic conductivity between polymer electrolytes based on poly(ethylene oxide) and
poly(propylene oxide) networks containing LiClO4.

26

JU

Figure 11 shows the temperature dependence of ionic conductivity for the PEO and PPO
network electrolytes containing LiCiO4. Because both of the polymer electrolytes were completely
amorphous in this temperature range, a direct comparison of the ionic conductivity. The ionic
conductivity of the PEO electrolytes was about fiive to ten times higher than that of the PPO
electrolytes, whereas the apparent ionic mobility or Tg did not differ appreciably from one another.
this result suggested that the ionic dissociation was facilitated in the PEO electrolytes. The
cooperative interaction of the neighbonng polar groups with an ion seems to be assured in the PEO
electrolytes. In the crystalline PEO complex, it is believed that PEO hss T2GT2G conformation,
which enables the cooperative interactio. The present PEO complexes were in the elastomerlc
state, allowing the conformation of the PEO chain to change continuously. However, PEO will
have the T2G conformation occasionally ana will then be able to solvate ions more enectlvely. It
has been reportet that PEO has a strong preference for the gauche state about the constituent
OC-CO bond. The side methyl group in PPO may sterically hinder this conformation.
Further results that support the importance of the interaction have been obtained for
poly(ethylene succinate) and poly(ethylene sebacate) electrolytes. The Tg of the former

N
G
LM

-Z

electrolytes was higher than that of the latter by 30-40C, whereas the conducitivity of the former
electrolyes was also higher. The polar groups in poly(ethylene succinate) are present in relatively
high density, which may assist the cooperative interaction. In contrast, the polar groups in
poly(ethylene sebacate) are far separated by octamethlyene units. Thusthis polymer does not
favorthe cooperative interaction because of its entropic instability. We now consider that the
cooperative interaction of neighboring polar groups with an ion is more important than the
intermolecular interaction. Whether or not neighboring polar groups can coordinate to ions
depends on the conformational energetics

2.5. Ion Transport Number in Polymers

JI

Because most of the polymer electrolytes are solid solutions of salts in polymers, both cations and
anions have a chance to move. It is of great importance to clarify the ionic transport number in
terms of the understanding of the ion-conduction mechanism and of their practical applications.
The ionic transport number has been measured in certain kinds of polymer electrolytes, especially
in PEO electrolytes, using various methods, including PFG-NMR measurements, tracer diffusion
experiments, complex impedance measurements, limiting current measurements, Tubabdt method,
EMF measurements on concentration cells, potentiostatic polarization experiments, and transient
ionic current measurements. The data obtained are somewhat distributed. However, it has become
plausible that both cation and anion are mobile in the polymer electrolytes consisting of PEO and
a lithium salt, and that the anion has a somewhat higher mobility than the canon.
The complex impedance measurements over a wide range of frequency (10-4 to 106 Hz) have been
widely adopted for the estimation of the ionic transport number. The theoretical background for
this method has been analyzed in detailby MacDonald. This method was applied to the
electrolytes with two mobile ions sandwiched between two electrodes, which are blocking
electrodes for one ion and nonblocking electrodes for the other. For the actual measurements, this
method requires the whole range of impedance responses including the bulk electrolyte impedance,
the charge transfer impedance, and the diffusion impedance. One nontrivial problemn with
applying this method to polymer electrolytes seems that the low-frequency diffsion arc appears
completely only at an elevated temperature (>100C) even in the impedance measurements down

27

to 10-4 Hz. The potentlostatic polarization measurements give a more simple and convenient
method. In the polarization experiments on the electrolyte sandwiched between cation-reversible
electrodes, the initial current after considered to be due to both migration of cation and anion, and
the final current after complete cell polarization, during which the anion diffusion created by a
concentration gradient that exactly opposed the migration of the ion under the influence of the
applied potential, was considered to be the migration of only the cation. Thus, the ratio of the final
current to the initial current was assumed to give the cation transport number. This may be true
when the charge transfer resistance is far smaller than the bulk resistance, because the foregoing
consideration does not include the contribution ofthis resistanceic the polarization current.

-Z

JU

While taking these facts into consideration, this section deals with the estimation of Li+
transport nunmber by the combination of the complex impedance and potentiostatic polarization
measurements on the polymer electolyte sandwiched between two lithium electrodes. As the
polymer electrolytes we used three types of amorphous network polymers. The first was the PEO
network polymer in which LiClO4 was dissolved. The second was the polyelectrolytetype PEO
network polymer, in which Li+ ions were introduced as counterions of the anion sites fixed to the
polymer backbone. The third was the network polymer from PEO-grafted polydimethylslloxane in
which LiClO4 was dissolved. The results on the transport number in these polymer eletrolytes
well reflected the differences in the polymer structures.

JI

N
G
LM

The three types of amorphous network polymer electrolytes were synthesized, according to the
procedure shown in Fig.12.

Fig.12. Synthetic procedure for three kinds of polymer electrolytes.

Figure 13 shows typical complex impedance spetra of the three types of polymer electrolytes
sandwiched between lithium electrodes at an oscillation level of 0.5 V. The profiles of the spectra
were two neighboring arcs. The low-frequency arcs (right-hand side arcs) and the high-frequency
arcs (left-hand side arcs) corresponded to the loci of the charge transfer impedance and the bulk
electrolyte impedance, respectively. The bulk resistance (Rb) and the charge transfer resistance (Re)
are shown in Fig.13. The ionic conductivities at 50C for the polymer electrolytes (1), (2), and (3)
were aproximately 10-6 to 10-5 S cm-1 (depending on the LiClO4 concentration (see Table 1), 10-7
S cm-1, and 10-6 S cm-1, respectively. The Re values, corresponding to the electrode reaction, were
about 104, irrespective of the kinds ofpolymer electrolytes. These values were similar to those
found in the poly(propylene oxide) networks with dissolved lithium salts and to those found in the
28

poly(ethylene succinate) with dissolved lithium salts, but were considerably higher than those
found in poly(-propiolactone)-LiClO4 complexes.

2.6. Superion-Conduction Polymers

N
G
LM

-Z

JU

Solid polymer electrolytes (SPE) have been attracting interest in terms of applications to battery
or display devices. They are mostly based on amorphous polymers that have interactions with
cations, such as poly(oxyethylene) derivatives. However, their ionic conductivities are limited to
below 10-4 S cm-1 at ambient temperature; also, the temperature dependencies of their ionic
conductivities are so large that their applications at low temperatures are limited.

Fig. 13. Complex impedance spectra of polymer electrolytes with lithium electrodes: (a) (1) [Li+]/[O]=0.007; (b)
(3) [Li+]/[O]=0.067; (c) (2) [Li+]/[O]=0.009.

Amonium compounds, such as pyridinium salts and imidazorium salts, form molten salts with
aluminum chloride at room temperature. The molten salts are ionic liquids and consist of only ions,
and these materials have high ionic conductivity of about 10-2 S cm-1 at room temperature because
a high carrier-ion density and high mobility are achieved in the molten state.

JI

This section describes the preparation of new polymer electrolytes consisting of polymer blends
of quarternized salts such as polypyridinium salt and aluminum (Schemes 1 and 2). The
polypyridinium here constitutes the molten salt chloride, instead of the pyridinium salt and enables
the polymer complexes to form thin films, resulting from the enormous increase in viscosity of the
molten salt.

29

JU
-Z
N
G
LM

Fig. 14. Ionic conductivity of molten salts as a function of the mole percentage of AlCl3 at r.t.

Figure 14 shows the ionic conductivity as a function of the mole percent of AlCl3 for the
molten salts. The molten salts were formed with the AlCl3 composition ranging from 30 to 70
mol, although both pyridinium salts and aluminum chloride are crystalline solids at room
temperature. These molten salts have high ionic conductivities of about 10-2 S cm-1 over the entire
molar ratios. The conductivity of the molten salts depends on the anion species of the pyridinium
salts, and the chloride molten salts have higher conductivities than do the bromide molten salts.

JI

The polymer complexes containing 15 mol% of the PPyBr component formed homogeneous
solids at room temperature, which could be made into films. Figures 15 and 16 show the
conductivity of these polymercomplexes as a function of the mole percentage of PPyBr and the
temperature dependence of the ionic conductivity, respectively. The polymer complexes shown in
Figs. 15 and 16 contain PyBr and AlCl3 with a molar ratio of 2:1. As the PPyBr component
increased from zero to 15 mol%, the viscosity of the resulting polymer complexes increased
considerably. However, the conductivities of polymer complexes gradually decrease with an
increasing PpyBr component, and a conductivity higher than 10-4 S cm-1 was obtained in all the
polymer complexes. This result indicates that the ionic motions in the present complexes are
decoupled with the segmental motion of Ppy.
Polycation salts systems based on polyblend or polythienyl salts were prepared by synthetic
routes (Scheme 3).

30

Scheme 3 Preparation of polycations.

N
G
LM

-Z

JU

Ionic conductivities of these polycation salts systems are shown in Figs.16-18, which indicate
higher ion conductivities than those of poly(ethylene oxide)-based polymers. The temperature
dependence of the ionic conductivity for these polymers are also small in comparison with those
of poly(ethylene oxide)-based polymers. Figure 19 summarizes the ion conductivities of these
polycation-based SPE to compare their ion conductivities.

Fig.19. Relation between content of polycations and ionic conductivity for molten salt-polycation complexes:
o, PBCl; --o--, PBBr; , PBACl; , PMVCl; ----, PMVABr; , PpyBr; --PpyBr;

JI

, PBPCl; ----, PSI.

It is seen in Fig. 19 that the systems based on PpyCl or PPyBr have higher and less
temperature-dependent ion conductivities amon these polycation-based SPE, whereas PMVC or
PMVB, which contain a pyridiniu cation in the main chain, have a lower or higher temperature
dependency. These results clearly relate to the Tg of these polycation-based SPE.
Figure 20 shows the ion conductivities of a polycation-based SPE that does not contain any
monomeric cations; in other words, it is completely made up of polymeric polycation systems.
The PpyCl/AlCl3 system show ion conductivities in the range of 10-7 to 10-9 S cm-1 at ambient
temperature, and their temperature dependence obeys linear Arrhenius plots. On the other hand,
the PSI/AlCl3 system indicates an ion conductivity as high as 10-3 to 10-5 S cm-1at ambient
temperature. Presumably, the flexible thieny cation in the main chain may increase the ion
conductivity. This high ion conductivity for all the polymeric SPE is the first ever attained.

31

complexes.

2.7. Applications of Ion-Conducting Polymers

JU

Fig.20. Temperature dependence of ionic conductivity for (a) PBBR or PpyCl-AlCl3 complexes; (b) PSI-AlCl3

N
G
LM

-Z

In general, polymer electrolytes can be processed into thin films and have a wide potential
window and good electrode compatibility. The application of polymer electrolytes to lithium
rechargeable batteries, making use of these properties, has been vigorously studied. For instance,
the joint project between groups in France and Canada has aimed at thin-film lithium batteries that
can be used in electri vehicles. As a solid electrolyte, the polymer electrolytes from linear PEO
and LiClO4 or LiCF3SO3 were used, and as a positive electrode, the composite materials
containing TiS2 or V6O13 were used. Polymer electrolytes enable the thickness of a unit cell to be
reduced to less than 100 m. at operation temperatures between 80C and 100C, the use of the
positive electrode materials was sufficiently high when the discharge reduction was performed at a
current density of 0.5-1.0 mA cm-2. The durability for the charge-discharge cycles cxtended over
250 cycles, and the energy density of 100 W h kg-1 was achieved.

JI

Another unique and interesting property of polymer electrolytes, whivh is not obtained with
inorganic solid electrolytes, is the ability to include various kinds of electroactive molecules in
them by dissolution or by chemical modification. The combination of this property with their high
ionic conductivity will enable us to use polymer electrolytes as media for electrochemical
reactions of electroactive molecules, just as we use ordinary electrolyte solutions for this purpose.
The first example of the use of a polymer electrolyt as a medium for an electrochemical reaction
is the electrochemical synthesis of conducting polymers. Electrochemical polymerization of
pyrrole, using ion-conducting polymers as a solid electrolyte, produced polypyrrole-polymer
electrolytr bilayer composites in situ. Figure 28 shows a schematic representation of the
electrochemical polymerization of pyrrole (Py) by using a polymer electrolyte. The polymer
electrolytes used were PEO network polymers in which several kind salts were dissolved. The
polymer electrolyte film in which pyrrole had been incorporated was sandwiched between two
electrodes, and the polymerization was carried out under galvanostatic condictions. When the
electrodes were removed from the polymer electrolyte after polymerization, polypyrrole (Ppy+X-)
grew on the surface of the polymer electrolyte in contact with the anode (see Fig.21). As a result,
polypyrrole-polymer electrolyte bilayer composites were obtained. The electronic conductivity of
the surface polypyrrole reached 101-102 S cm-1 at room temperature and was almost independent
of the quantity of electricity used in the polymerization. Because the amount of dopant salt
32

-Z

JU

consumed during the polymerization was negligible, these bilayer composites were junctions
between an electronic conductor and an ionic conductor. The electronic spectra of the bilayer
composites changed in accordance with the redox sates of the polypyrrole layer, in the solid state
(Fig.22).the heavily doped, lightly doped, and undoped states were blue, red-brown
andyellow-green, respectively. The spectral change coincided well with that of polypyrrole
synthesized and measured in an electrolyte solution. The bilayer composites have improved not
only mechanical properties, compared with polypyrrole itself, but also have electrochemical
activity in the solid state. The latter property might open a route for the application of these bilayer
composites to solid electrochemical devices.

N
G
LM

Fig.21.Schematic representation of the electrochemical polymerization of pyrrole (Py) using a solid polymer

JI

electrolyte, in which Ppy+X- is an X- ion doped polypyrrole.

Fig.22. Change in electronic spectra of polypyrole-polymer electrlyte bilayer composite film corresponding to
redox reaction of polypyrrole in a solid state.

33

Fig.23. Cyclic voltammograms of ferrocene in a polymer electrolyte consisting of poly(ethylene oxide) network
and LiClO4 at 80C.

JI

N
G
LM

-Z

JU

Whether or not reversiable redox reactions occur in polymer electrolytes when redox-active
molecules are incorporated in them is the next question to be considered here. Ferrocene, a typical
redox-active molecule, can be dissolved in the PEO network electrolytes. Figure 23 shows cyclic
voltammograms of ferrocene in the polymer electrolyte, using an Ag quasi reference electrode.
Both oxidative and reductive waves are clearly seen. Because a redox current of the polymer
electrolyte in the absence of ferroence was not observed in this potential range, the redox waves
seen in Fig. 23 could be assigned to the reversible redox reactions of the incorporated ferrocene.
Thus, the redox current was in proportion to the redox reaction rate of ferrocene. The mechanism
of the electron-transfer reaction of ferrocene is unclear. For electrodes covered with thin films,
so-called modified electrodes, in which redox-active molecules are incorporated in high
concentrations, electron self-exchange reactions can occur. In polymer electrodes containing
ferrocene, there may be two possible mechanisms for the electron-transfer reaction: one is the
diffusion of ferrocene from the bulk to the electrode, fellowed by the electron transfer reaction at
the electrode; the other possibility is that the redox reaction at the electrode is propagated by the
electron transfer reaction between the redox-active centers (the electron self-exchange reaction).
Figure 24a shows the temperature dependences of the ionicconductivity and the oxidation peak
current. Both of the temperature dependences showed a WLF profile, and the oxidation peak
current was proportional to the square root of the sweep rate (see Fig.24b). From these results, the
redox reaction of ferrocene was proved to be diffusion-controlled. Which mechanism dominates
the redox reactions in the polymer electrolytes, as functions of species, concentrations of
redox-active molecules, and their manner of incorporation, is a very interesting problem. In any
event, polymer electrolyes can be a medium of reversible redox reactions of electroactive
molecules in the absence of low molecular weight solvents.

34

JU

Fig.24. (a) Temperature dependences of ionic conductivity and oxidation peak current of ferrocene for polymer
electrolytes based on PEO net works. (b) Relation between oxidation peak current and square root of sweep rate at

-Z

80C.

JI

N
G
LM

Ref. Naoya Ogata, in Fuctional Monomers and Polymers, second Ed. edited By Kiichi Takemoto,
Raphael m Ottenbrite, and Mikihard Kamachi. Marcel Dekker, Inc. New York. 1997.

35

Chapter 3
Magnetic Polymers

3.1. Introduction
3.2. Classification of Magnetism
3.3. Historical Development of Polymers for Magnetism

3.1. Introduction

JU

3.4. Organic Polyradicals for Magnetism

N
G
LM

-Z

It is easy to find a wide variety of applications for magnetism in human life when we take a glance
at our surroundings. One of the first applications of magnets began with the mariners compass.
Since the invention of the compass, magnets have been widely used by humans: effective
development in the applications for magnets has provided generators, electric motors, magnetic
tapes, and so on, all of which contribute to much of our wealth and happiness. Highly magnetic
compounds (i.e., ferro- and ferromagnetic) are ascribed to the spin-containing molecules in which
the magnetic sites are the d or f orbitals of the transition, lanthanide, or actinide metals.

Synthetic organic polymers have been practically applied to magnetic materials, such as a
magnetic gum and magnetic tapes. These materials were made by dispersing an inorganic
magnetic substance in a synthetic rubber, or by kinking ferrite onto a polymer film, but these
polymers are not themselves magnetic substances. Generally, organic compounds have no
unpaired electrons, except for the free radicals and carbenes, which are unstable at room
temperature. Accordingly, organic compounds have not been of interest as magnetic materials,
although there have been a few papers that pointed out the theoretical possibility of ferromagnetic
organic compounds.

JI

Organic polymers containing paramagnetic species have been of interest for use as conducting
materials, redox resins, and antioxidants. To my knowledge, no systematic study of magnetic
properties of these polymers had been carried out before we began investigating polymers with
stable radicals in their side chains. In 1984, several polycarbenes were prepared by the in situ
photolysis of polydiazo derivatives. The magnetic susceptibility measurements of these
polycarbenes showed that the polycarbenes have the highest possible spin multiplicities in the
ground state. For example, tetra carbenes have very large magnetic moments with nonet (S = 8/2).
Moreover, in 1986, the first organic ferromagnet was reported by Ovchinnikov and his colleagues
in the Soviet Union. These molecular-based magnetic properties were ascribed to interactions of
the p-electrons of C, O, and N. This is different from magnetism in conventional atom-based and
d-f electron inorganic materials, and revealed the possibility of magnetic materials associated only
with organic compounds. The magnetic properties can be modulated by flexible organic synthetic
methods, which may expand the range of the potential application. These magnetic polymers
suggest the possibilities of new types of functional polymers. Accordingly, much more attention
has been paid to the preparation of high-spin polymers, and some of them have been
36

ferromagnetic.
This chapter surveys the research trends pertaining to polymers for magnetism.

3.2. Classification of Magnetism

Some substances are attracted to a magnetic field, whereas others are repelled from it. When a
substance is placed in a magnetic field H, its magnetic induction B can be expressed by Eq. (1)
BH + 4M (1)
where M is the magnetlzation.
The magnetic susceptibility () of a material placed in a magnetic field is expressed as the ratio
of the magnetization to the magnitude of the magnetic field.
= M/H

(2)

-Z

JU

The sign of the magnetic susceptibility usually depends on whether the electrons in a particular
molecule are paired or unpaired. Accordingly, we can divide all matter into two categories relative
to their magnetic properties: diamagnetic and paramagnetic. Paramagnetism is characterized by
unpaired electron spins in a molecule or ion. Diamagnetism is generated by paired electrons
moving in a closed orbital. In a diamagnetic molecule, in which the magnetic induction B is

N
G
LM

smaller than the magnetic field H, the value of is negative. In a paramagnetic molecule, in which
the value of B is larger than that of H the value of is positive.

Figure 1. Schematic two-dimensional representation of three types of spin orientation: J is the exchange integral.
(a) Paramagnetic; (b) ferromagnetic; (c) antiferromagnetic; (d) ferrimagnetic.

JI

In paramagnetic species, four types of magnetization are possible as a result of magnetic

interaction among randomly oriented spins. The first, in which spins are oriented randomly
because there is no magnetic interaction among spins, is known as paramagnetism. The second, in
which randomly oriented spins of a paramagnetic substance orient themselves parallel to one
another, is known as ferromagnetism. The third, in which spins that have a magnetic moment of
the same magnitude orient themselves antiparallel to one another, is known as antiferromagnetism.
The fourth, in which the two kinds of spins having magnetic moments of different magnitude
orient themselves antiparallel alternately to one another, is known as ferrimagnetism. Schematic
representations of these generallied types of spin orientations for a two-dimensional lattice are
shown in Fig.1.
The magnetic susceptibility () of a paramagnetic sample is inversely proportional to
temperature (Curie law):
= C/T (3)
37

where C is the Curie constant and T is temperature in units of Kelvin.

The Curie constant C is dependent on the number of unpaired electrons and the g value of the
compound. The Curie constant for the spin-only magnetic susceptibility is
C = N2g2S(S + 1)/3k (4)
where is the enective magnetic moment, S is the spin angular momentum, and k is the
Boltzmann constant.
A weak magnetic interaction between neighboring spins in a material may be approximated as a
perturbation to Eq. (3). The influence of this interaction may be described by replacing the
temperature term in Eq.(3) with a (T - ) term in Eq.(5).
= N2g2S(S + 1)/3k(T - )

(5)

JU

where is the Weiss constant, in units of Kelvin.

-Z

The magnetic behavior of paramagnetic molecules and ions is shown in Fig.2. A plot of the
inverse of the magnetic susceptibility against temperature provides information on the magnetic
interaction among electron spins. Paramagnetic species having no magnetic interaction yield a
straight line (Curie law), and those that have magnetic interaction deviate from the straight line,
which obeys the Curie-Weiss law. The slope of these lines gives the Curie constant from which the
g value is calculated. The intercept of the line with the temperature axis gives both the sign and

JI

N
G
LM

magnitude of the Weiss constant. A positive value for may result from a ferromagnetic spin
interaction, and the negative value for from an antiferromagnetic spin interaction. Figure 2
shows generallied plots of the magnetic behavior in a sample that obeys the Curie law or the
Curie-Wiss law.

Figure 2. A plot of the inverse of magnetic susceptibility versus temperatures: = C/T-; = 0,

paramagnetic; >0, ferromagnetic; <0, antiferromagnetic.

38

Figure 3. Plot of effective magnetic moment versus temperature.

Another convenient measure to distinguish magnetic properties of a substance is the

JI

N
G
LM

-Z

JU

temperature dependence of the effective magnetive moment [eff = 2.828(T)1/2]. The magnitude
of eff is dependent on the number of unpaired electrons and the g value. For a system obeying the
Curie law, a plot of eff as a function of temperature yields a horizontal straight line, which means
that eff is independent of the temperature. The presence of spin interaction is denoted by a
departure from the horizontal line in the low-temperature region. Typical plots of the effective
magnetic moment are shown in Fig.3. When the interaction among spins is large enough to form a
ferromagnetic spin ordering, there appears to be a macroscopic magnetization at zero applied to
the magnetic field below the temperature required for the long-range spin ordering, which is
referred to as the Curie temperature (Tc). A hysteresis curve is observed at temperatures lower than
Tc (Fig. 4). Usually, no spontaneous magnetization occurs when the spins are antiparallel to one
another. Accordingly, a hysteresis loop is observed in the ferromagnet and ferrimagnet.

Figure 4. A hysteresis vurve for a sample with ferromagnetic ordering.

3.3. Historical Development of Polymers for Magnetism

In 1967, Itoh and Wasserman et al. independently found that the electronic ground state of
bis(phenylmethylenyl)-m-phenylene, Structure (1) is the quintet state. On the basis of this finding,
Mataga proposed that the polymers, Structures (2) and (3), with the comjugated p-electron system,
would show a ferromagnetic spin alignment owing to the topological nature of the molecular
orbits. To confirm this idea, Itoh and Iwamura et al. prepared bis(phenylmethylenyl)-m-phenylene
with no net spin multiplicity owing to intramolecular spin alignment, and suggested that such a
high-spin state is relevant to the design of organic ferromagnets. However, the temperature
dependence of the magnetic susceptibility gave a straight-line with a negative Weiss temperature
39

N
G
LM

-Z

JU

22 K in the Curie-Weiss law. This finding indicates that intermolecular spin alignment is
antiferromagnetic. To obtain information on the relation between intermolecular spin alignment
and the relative orientation, the electronic spin-spin interaction between two diphenylcarbene units
incorporated in the [2,2]paracyclophane skeleton was investigated by electron spin resonance
(ESR) spectroscopy. The pseudo-ortho- and pseudo-para-bis(phenylmethylenyl)[2,2]paracyclophanes are in the ground quintet state, whereas the pseudo-meta-isomer is a singlet.
These findings show that intermolecular ferromagnetic interaction is possible in organic free
radicals. In 1977, Ovchinnikov pointed out that theoretically organic polymers such as Structures
(4) and (5) might be ferromagnetic owing to an exchange interaction. Buchachenko independently
proposed preparing polyradlcals possessing ferromagnetism and generating microscopic
ferromagnetism of organic radicals. However, no experimental data on magnetic polymers was
reported before 1980. Ovchinnlkov and co-workers experimentally pointed out the possibility of
organic ferromagnetic polymers, although unsolved problems remained for ferromagnetism. Since
then, chemists have challenged the preparation of organic ferromagnetic polymers. Ferromagnetic
organic and organometallic compounds with ferromagnetic interaction have been prepared on the
basis of theoretical concepts.

JI

3.4. Organic Polyradicals for Magnetism

Organic polymers for magnetism are classified into three categories: (1) polymers with organic
radicals in the side chains, (2) main-chain conjugated polymers with organic radicals, and (3)
two-dimensional polyradlcals.
A. Polymer Radicals in the Side Chains
In the course of ESR studies on polymers containing 2,2,6,6-tetramethylpiperidine-1-oxyl as
side chains, we found that an electron exchange interaction takes place between paramagnetic side
chains. This phenomenon suggests the possibility that organic polymers containing paramagnetic
species might be a new type of magnetic material owing to the magnetically long-range ordering
of unpaired electrons through a spin-spin interaction. On the basis of this idea, we investigated the
temperature dependence of the magnetic susceptibility of polymethacrylates containing
2,2,4,4-tetramethylpiperidine-1-oxyl or verdazyl as side chains: Structure (6) or (7). Weak
antiferromagnetic interaction took place between the unpaired electrons of the polymers. Since the
40

magnetic interaction between unpaired electrons in the corresponding monomer was weaker than
that in the polymer, it was concluded that there was a polymer effect on the magnetic behavior of
free radicals. However, the interaction was too weak to be considered a magnetic material.

B. Main-Chain-Conjugated Polyradicals

JI

N
G
LM

-Z

JU

Ovchinnikov suggested that polyacetylene derivatives with p-conjugated stable radicals as side
chains could be potential organic ferromagnets. In 1986, Ovchinnikov et al. obtained black and
insoluble polymers by the polymerization of 1,4-bis(2,2,6,6-tetramethyl-4-oxy-4-piperidyl-1-oxyl)
-butadiene:Structure (8). The Infrared (IR) spectra of the polymers suggest that the polymer was
composed of a l,2-addition unit and a l,4-addition unit as shown in structures (9) and (10). The
magnetic behavior was investigated by the measurement of temperature dependence of the
magnetic susceptibility and the field dependence of magnetization, which indicated that some
parts of polymers were ferromagnetic. Because of the poor yield of the ferromagnetic fraction and
incomplete characterization of mapnetic properties, the presence of pure organic ferromagnetism
was controversial. However, a study on a structurally analogous polymerwhich exhibited a
similar ferromagnetic behavior, supported the existence of organic ferromagnetism.

p-Conjugated polyradical macromolecules, in which side chains of p-conjugated stable radical

moieties are accumulated along a p-conjugated main chain, are expected to be potential candidates
for ferromagnetic molecules, because they accumulate radical molecules along one conjugated
polymer chain. Accordingly, new p-conjugated radical derivatives of poly(phenylacetylene) and
poly(phenylenevinylene) were prepared and their magnetic behaviors investigated by ESR and
magnetic susceptibility measurements. The polymers used for this study are shown in Table 1.
Magnetic ordering of unpaired electrons through these p-cojugated systerns could be faintly
41

obseved by the measurement of magnetic susceptibility. For example, the magnetization curve for
poly[(3,5-di-t-butyl-4-hydroxyphenyl)acetylene] (11) almost coincided with the Brillouin curve of
S1/2, while the curve for poly[p-ethynylphenyl]hydrogalvinoxyl] (12) deviated slightly from
that of S = 1/2, which means very weak interaction between spins through bonding in the
polyacetylene derivatives. Torsion in the main chain of polyphenylacetylene or twisted phenoxy
radical moieties would inhibit an exchange interaction throughout the p-conjugated main chain.
Tabata et al. prepared high-spin polymers by thermal anneating of polyphenylacetylene and its

N
G
LM

-Z

JU

derivatives at 120C. The concentration of radicals increased several orders of magnitude after the
thermal annealing, and the magnetic behavior suggested the presence of spin-glass (see Fig.4).

JI

To obtain a planar-conjugated main chain, poly(phenylenevlnylene) (PPV) was suggested as a

potential candidate for an effective backbone for p-conjugated polyradicals because of its
developed conjugation, coplanarity, and solubility after substitution on the phenylene ring. Nishide
et al. synthesized poly(1,4- and 1,2-phenylenevinylene) (13-16) bearing phenoxy and nitroxy
radicals, the unpaired electrons of which are delocalized to the p-conjugated main chain, as one of
the polyradicals with a PPV skeleton, and their magnetic behaviors were investigated.
Measurements of magnetizatlon and static magnetic susceptibility of Structures (13)-(16)
indicated ferromagnetic interaction at low temperature. The ferromagnetic interaction is expected
to be enhanced with the spin concentration the degree of polymerization, and the spin density
distribution onto the main chain.

Miura et al.synthesized poly(1,3-phenyleneethynylene) with pendant nitronyl nitroxide (17) or

t-butylnitroxyl (18,19) and investigated their magnetic properties. Although the polyradlcals
possess high-spin concentratlons and satisfy the p-topologic symmetry requirement for exhibiting
ferromagnetic interactions among p-spins, the magnetic susceptibility measurements showed that
there was no appreciable amount of ferromagnetic interaction among the spins in the polyradicals.
In addition, an ESR study of the model compound showed that the nitronyl nitroxide free radical
is not delocalized to the conjugated main chain. Consequently, no ferromagnetic interaction is
ascribable for the following two reasons: (1) the spin is less delocalized from the nitronyl

42

nitroxide sites to the m-phenylenediethylene complexes, and (2) the distances between one spin
center and the neighboring ones are too great to observed ferromagnetic coupling among the spins.

CTwo-Dimensional Polyradicals

JU

In 1986, Torrance et al. reported that some parts of the black, insoluble polymers (20), obtained
by the reaction of triaminobenzene with iodine, showed ferromagnetic behavior in measurements
of magnetization dependence on temperature and magnetic field.

JI

N
G
LM

-Z

In 1989, Ohtani et al. found that ferromagnetic polymers (21) were formed by dehydrogenation
of triarylmethane resins by irradiation of UV light or laser light in the presence of photooxidizing
agents. This finding led to the idea that conjugated polymer chains are necessary for the
occurrence of ferromagnetism. More recently, Murata et al. found that carbon powder prepared by
the pyrolysis of a mixture of COPNA resin and activated carbon showed an apparent saturation
magnetization and a hysteresis loop.

Tanaka et al. prepared a polymer with an indigo unit in the main chain (22), and reported that
the insoluble polymer obtained contained a fraction that was attracted to a permanent magnet and
showed a definitive hysteresis loop at room temperature. Recently, polymers obtained from
43

DPolymers that Contain Paramagnetic ions

JU

catalyst-free, 3-dipolar cycloaddition of 1,3-bis-(3-sydnone) and N,N-(1,4-phenylene)bismaleimide (23) have exhibited a magnetic behavior characteristic to ferromagnets, even at
room temperature. Significant magnetic hysteresis was observed and the magnetization at 300 K
reached a saturation value of 0.15 G/g. The structure of the polymer was not clearly determined,
but the tentative structure is shown in Structure (23).

-Z

Polymers that contain paramagnetlc ions are classified into two categories: (1) polymers with
paramagnetic metalloporphyrins as side chains, and (2) polymers containing paramagnetic metal
ions in the main chain.
1. Polymers with Paramagneticmetalloporphyrins in Their Side Chains

JI

N
G
LM

Usually, magnetism is caused by spin-alignment owing to an intermolecular exchange interaction

of electrons in d- or f-orbitals of metal ions. Accordingly, magnetic behavior of organic polymers
containing paramagnetic metal ions was investigated as a new strategy for developing polymer
ferromagnetism.

We preparpd polymers containing tetraphenylporphyrin (TPP) moieties, which were considered

to function as inhibitors of radical polymerizations, and found that TPP moieties in the pdymers
were forced to interact with each other because they are so close. Accordingly, we introduced
paramagnetic metal ions [Cu(II), Ag(II), VO(II), and Co(II)] into the TPP moieties of the polymers
and investigated the magnetic behavior. The magnetic susceptibility (m) of these polymers (24-29)
showed that the paramagnetic species, bound to polyacrylate, polyacrylamide, or
polymethacrylamide, interact antiferromagnetically, and this interaction is much larger than that of
44

JU

the corresponding monomer units. Polymer structural effects on the magnetic behavior were
clearly observed. For example, an antiferromagnetic polymer was obtained with
tetraphenylporphinatosilver(II). However, for polymers in which the TPP moieties were directly
bound to the main chain, there was no antiferromagnetic interaction. Accordingly, the origin of the
antiferromagnetic behavior is probably ascribable the superchange interaction through the C=O
group between the Ag(II) ions. We attempted to expand to polyacrylates that contained other
paramagnetic ions. Rare earth metal ions [Er(III) and Yb(III)] were chosen instead of Ag(II) ions,
because these ions have electrons in the f-orbital and, accordingly, are considered to effect
magnetic interactions. The termperature dependence of cM of AOTPPEr(III) showed a
straight-line through the origin, indicating that there is no magnetic interaction between the
AOTPPEr(III)OH molecules. A similar straight line was also observed for poly[AOTPPEr(III)OH],
but a detailed study showed some differences from AOTPPEr(III) below 5 K: a weak
ferromagnetic interaction was found; however, it was much weaker than expected. Copolymers
containing both TPPCu(II) and TPPVO(II) as side chains were prepared by the radical
copolymerization of AOTPPCu(II) and AOTPPVO(II), the metal ions of which have d9 and d1
orbitals, respectively. The temperature dependence of the magnetic susceptibility follows the

N
G
LM

-Z

Curie-Weiss law with = 50 K, indicating the existence of the ferromagnetic interaction between
TPPCu(II) and TPPVO(II) moieties. Because poly[AOTPPCu(II)] and poly[AOTPPVO(II)] are
antiferromagnetic, the ferromagnetic interaction in the copolymer might be ascribed to an
interaction between TPPCu(II) and TPPVO(II) moieties located casually in the position suitable
for the occurrence of long-range ferromagnetic interaction.

JI

Miller et al. found that a compound obtained by the reaction of TPPMn(II) and
tetracyanoethylene (TCNE) in toluene showed ferromagnetic behavior with magnetic
temperature-dependent susceptibility, in which the Curie-Weiss temperature is 61 K. The structure
(30) of the compound is considered as follows:

2. Polymers with Paramagnetlc Metal Ions in Their Main Chains

Kahn et al. found a ferromagnetic interaction between Cu(II) and VO(II) ions in a
heterobinuclear complex Cu(II)VO(II)(fsa)2enCH3OH, where (fsa)2en-4 stands for the
heterobichelating ligand derived from a Schiff base N,N-(2-hydroxy-3-carboxybenzylidene)
-1,2-diaminoethane. Moreover, Kahn et al. found that ferrimagnetic chains with magnetic
transitions can be designed in which two kinds of magnetic centers regularly alternate. For
example, the inorganic polymer (31) with MnCu(pbaOH)(H2O)3, in which pbaOH is
2-hydroxy-1,3-propylenebis(oxamato), exhibits a maynetic transition at 4.3 K.

45

JU

Since this finding, polymers composed of 2-substituted 4,4,5,5-tetramethylimidazoline-1-oxyl-3

-oxide and paramagnetic ions were prepared as new magnetic polymers, shown in the following
structures, (32-35) with M=Cu(II), Ni(II), or Mn(II).

JI

N
G
LM

-Z

Gatteschi et al. prepared linear chain polymers that were composed of M(hfac)2(NIT)R, where
M is Cu(II), Ni(II), or Mn(II); (NIT)R is 2-alkyl-4,4,5,5-tetramethylimidazoline-1-oxyl-3-oxide;
and hfac is hexafluoroacetylacetonate. Ferromagnetic ordering was observed for those polymers
composed of Mn(hfac)2(NIT)Et (36) and Ni(hfac)2(NIT)Prn, respectively. The critical temperature
for each polymer was 8.6 and 5.3 K, respectively.

Recently, Kahn et al. prepared a molecular-based magnet (37) with a three-dimensional

in
which
rad
is
structure
composed
of
(rad)2Mn2[Mn(opba)]3(DMSO)2H2O,
2-(4-N-methylpyridinium)-4,4,5,5-tetramethylimidazoline-1-oxyl-3-oxide; opba is o-phenylenebis
-(oxamato); and DMSO is dimethyl sulfoxlde.The compound has three kinds of spin carriers:
Mn(II), Cu(II), and radical cations. The X-ray crystal structure of this compound revealed that it
was composed of nearly perpendicular graphite-like networks, with edge-sharing Mn(II)6Cu(II)6
hexagons (Fig.5), and that the Cu ions of the two nearly perpendicular networks are fully
46

-Z

JU

interlocked with 2-(4-N-methylpyridinium)-4,4,5,5-tetramethylimidazoline-1-oxyl-3-oxide. Mn(II)

and Cu(II) ions of the Mn(II)6Cu(II)6 hexagons are antiferromagnetically coupled through the
oxamato bridges to form a two-dimensional ferrimagnetic network. These are ferromagnetlcally
coupled through radical cations which bridge the Cu(II) ions through the nitronyl nitroxide groups.
The magnetic temperature dependence revealed that the compound behaves as a magnet below
22.5 K.

JI

N
G
LM

Ref. Mikiharu Kamachi, in Fuctional Monomers and Polymers, second Ed. edited By Kiichi
Takemoto, Raphael m Ottenbrite, and Mikihard Kamachi. Marcel Dekker, Inc. New York. 1997.

47

Chapter 4
Polymeric

Controlled

Release

System,

Polymeric

Prodrugs and Pulsatile Release System

4.1. Polymers for Controlled Release

4.2. Polymeric Prodrugs

4.1.0. Introduction
Need for Controlled Release Systems

-Z

4.1. Polymers for Controlled Release

JU

4.3. Pulsatile Release System

N
G
LM

Controlled drug delivery technology represents one of the most rapidly advancing areas of
science in which chemists and chemical engineers are contributing to human health care. Such
delivery systems offer numerous advantages compared to conventional dosage forms including
improved efficacy, reduced toxicity, and improved patient compliance and convenience. Such
systems often use synthetic polymers as carriers for the drugs. By so doing, treatments that would
not otherwise be possible are now in conventional use. Although the introduction of the first
clinical controlled release systems occurred less than 25 years ago, 1997 sales of advanced drug
delivery systems in the United States alone were approximately $14 billion dollars.1 In this paper,
we examine the breadth, the mechanisms, and rationale for controlled drug delivery and then focus
our attention on some of the major families of synthetic polymers being used in the field.

JI

This review examines the chemical issues involved in the design of synthetic polymers used in
the controlled release of drugs. Before considering the variety and the evolution of these
polymeric structures, it is necessary to examine the motivation for achieving controlled release.
This field of pharmaceutical technology has grown and diversified rapidly in recent years.
Understanding the derivation of the methods of controlled release and the range of new polymers
can be a barrier to involvement from the nonspecialist.

All controlled release systems aim to improve the effectiveness of drug therapy.1,2 This
improvement can take the form of increasing therapeutic activity compared to the intensity of side
effects, reducing the number of drug administrations required during treatment, or eliminating the
need for specialized drug administration (e.g., repeated injections). Two types of control over drug
release can be achieved, temporal and distribution control.
Methods of Controlled Release
In temporal control, drug delivery systems aim to deliver the drug over an extended duration or
at a specific time during treatment. Controlled release over an extended duration is highly
beneficial for drugs that are rapidly metabolized and eliminated from the body after administration.
An example of this benefit is shown schematically in Figure 1 in which the concentration of drug
48

N
G
LM

-Z

JU

at the site of activity within the body is compared after immediate release from 4 injections
administered at 6 hourly intervals and after extended release from a controlled release system.
Drug concentrations may fluctuate widely during the 24 h period when the drug is administered
via bolus injection, and for only a portion of the treatment period is the drug concentration in the
therapeutic window (i.e., the drug concentration that produces beneficial effects without harmful
side effects). With the controlled release system, the rate of drug release matches the rate of drug
elimination and, therefore, the drug concentration is within the therapeutic window for the vast
majority of the 24 h period. Clinically, temporal control can produce a significant improvement in
drug therapy. For example, when an opioid pain killer is administered to a patient with terminal
cancer, any time that the drug concentration is below therapeutic concentrations the patient
experiences pain. A temporally controlled release system would ensure that the maximum possible
benefit is derived from the drug.

Figure 1. Drug concentrations at site of therapeutic action after delivery as a conventional injection (thin line) and

JI

as a temporal controlled release system (bold line).

Figure 2. Drug delivery from an ideal distribution controlled release system. Bold line: Drug concentrations at site
of therapeutic action. Thin line: Systemic levels at which side effects occur.

In distribution control, drug delivery systems aim to target the release of the drug to the precise
site of activity within the body. The benefit of this type of control is shown schematically in
Figure 2 in which drug concentrations at the site of activity and side effect production are
compared. There are two principle situations in which distribution control can be beneficial. The
first is when the natural distribution causes drug molecules to encounter tissues and cause major
side effects that prohibit further treatment. This situation is often the cause of chemotherapy
failure when bone marrow cell death prevents the patient from undergoing a complete drug
treatment. The second situation is when the natural distribution of the drug does not allow drug
molecules to reach their molecular site of action. For example, a drug molecule that acts on a
49

receptor in the brain will not be active if it is distributed by the patients blood system but cannot
cross the blood-brain barrier.

Mechanisms of Controlled Drug Release Using Polymers

JU

A large number of classes of drugs can benefit from temporal or distribution controlled release.
These classes include chemotherapeutic drugs, immunosuppressants, 6 antiinflammatory agents,
antibiotics, opioid antagonists, steroids, hormones, anesthetics, and vaccines. Recently, the need to
develop new controlled release strategies has been intensified by advances in the design of peptide
drugs and emergence of gene therapy. These biotechnologyderived agents may dominate the next
generation of drug design. However, their clinical success may be dependent on the design of
controlled release devices that ensure that the drugs reach their target cells precisely at the
required time. A discussion of the pharmacological and clinical motivations for controlling the
release of the specific drug classes referred to above is beyond the limit of this article; however, a
number of excellent reviews are available.16,17 In addition, it should be noted that controlled
release technology is not confined to pharmaceutical applications but has also proven beneficial in
agricultural18 and cosmetic industries.

N
G
LM

-Z

A diverse range of mechanisms has been developed to achieve both temporal and distribution
controlled release of drugs using polymers. This diversity is a necessary consequence of different
drugs imposing various restrictions on the type of delivery system employed. For example, a drug
that is to be released over an extended period in a patients stomach where the pH is acidic and
environmental conditions fluctuate widely will require a controlled release system very different
from that of a drug that is to be delivered in a pulsatile manner within the blood system. An
important consideration in designing polymers for any controlled release mechanism is the fate of
the polymer after drug release. Polymers that are naturally excreted from the body are desirable
for many controlled release applications.20,21 These polymers may be excreted directly via the
kidneys or may be biodegraded into smaller molecules that are then excreted. Nondegradable
polymers are acceptable in applications in which the delivery system can be recovered after drug
release (e.g., removal of patch or insert) or for oral applications in which the polymer passes
through the gastrointestinal tract.

JI

From a polymer chemistry perspective, it is important to appreciate that different mechanisms

of controlled release require polymers with a variety of physicochemical properties. This
requirement has stimulated the evolution of the new polymers that will be discussed in section IV.
Before consideration of these polymers, the major mechanisms of controlled release and
polymeric characteristics that are required to carry out these mechanisms will be briefly described.
Temporal Controlled: Most drug molecules need to be dissolved in the aqueous environment of
the patient and freely diffuse within that media before they can act on their target receptors.
Polymeric devices that achieve temporal controlled release protect drug molecules from this
aqueous living environment for preprogrammed periods of time. This protection can involve
delaying the dissolution of drug molecules, inhibiting the diffusion of the drug out of the device,
or controlling the flow of drug solutions.22 These mechanisms are shown in Figure 3.
Mathematical descriptions of release mechanisms have been described previously.

Polymers employed to delay drug dissolution aim to slow the rate at which drug molecules are
exposed to water from the aqueous environment surrounding the drug delivery system. This may
be achieved by a polymer coating or matrix that dissolves at a slower rate than the drug.
50

N
G
LM

-Z

JU

In diffusion-controlled release, drug molecule diffusion within an aqueous solution is inhibited

by the insoluble polymer matrix in which drug molecules must travel through tortuous pathways
to exit the device. Polymer chains such as those in a cross-linked hydrogel form the diffusion
barrier. The barrier to diffusion can be decreased by swelling of the hydrogel, for example, which
creates voids in the gel structure. Such hydrogels may also benefit from bioadhesive
characteristics which allow them to reside within the gastrointestinal tract for extended time
periods. Polymers used for diffusion-controlled release can be fabricated as either matrices in
which the drug is uniformly distributed or as a rate-limiting membrane that protects the drug
reservoir from the living environment.

Figure 3. Examples of mechanisms of temporal controlled release.

JI

Devices that control the flow of drug solutions sometimes utilize osmotic potential gradients
across semipermeable polymer barriers to generate pressurized chambers containing aqueous
solutions of the drug. This pressure is relieved by the flow of the solution out of the delivery
device. The rate of flow is controlled because flow is restricted to fluid transport through a
micrometer scale to larger diameter pore or pores.

Many temporal controlled release devices use the above mechanisms to provide sustained
release of drug at a constant rate. Another form of temporal controlled release is responsive drug
delivery in which drug is released in a pulsatile manner only when required by the body.24 Much
work in this area has as its eventual goal the delivery of insulin to diabetics. Insulin requirements
fluctuate throughout the day as patient food intake and activity change blood glucose levels.
Current insulin formulations require repeated injections daily and careful control of glucose intake.
Responsive drug delivery hopes to revolutionize insulin therapy with the design of systems that
release insulin in response to increased blood glucose levels. In general, responsive drug delivery
systems have two components: a sensor that detects the environmental parameter that stimulates

51

drug release and a delivery device that releases drug. For diabetes treatment, responsive drug
delivery systems have been proposed that use the enzyme glucose oxidase as the sensor.25 When
blood sugar levels rise, glucose oxidase converts glucose to gluconic acid resulting in lowered pH.
This pH decrease is then used as the signal for insulin release. Release is achieved by pH-sensitive
polymers that either swell or degrade in acidic environments.

JU

The concept of responsive drug delivery can be used for any drug therapy in which a sensor and
delivery device can be coupled. Signals that have been employed to trigger drug release have been
reviewed by Langer27 and include the following: magnetic signals in which magnetic beads are
distributed within a polymer matrix and cause a rearrangement of that matrix when a magnetic
field is applied; electrical signals in which pore size, permeability, and other factors are controlled
by electrically stimulated polymer swelling; ultrasonic signals in which the intensity, frequency,
and duration of ultrasound increase release for both nondegradable and biodegradable polymeric
systems; pH systems in which ionizable groups within polymer gels control polymer chain
interactions; and temperature systems in which thermosensitive hydrogels swell and collapse in
response to temperature variations.

N
G
LM

-Z

Distribution Controlled: The simplest method of achieving distribution control is to implant the
drug delivery system directly at the site. This method has been successfully described in the
delivery of chemotherapeutic agents to malignant gliomas using poly(anhydrides) by Brem et al.4
During treatment, polymer disks containing carmustine are implanted in cavities created after
surgical removal of the tumor. This distribution control is highly beneficial given that 90% of
malignant gliomas recur within 1 in. of the original tumor site. In general, direct implantation is
suitable for distribution control only if the site of drug action is accessible without risk to the
patient and the drug is unable to leave this site, e.g., the drug is unable to pass through the
blood-brain barrier.

JI

For the majority of diseases that require distribution controlled release of drug, a targeting
mechanism must be employed that allows the delivery system to find the desired target [17].
Polymers are used in two types of delivery systems for these applications, colloidal carriers and
polymer-drug conjugates. In colloidal formulations, the polymer encapsulates drug within microor nanoparticles.28 In polymer-drug conjugates, the drug is covalently coupled to the polymer. In
these forms of distribution controlled release, the polymer acts as a carrier but is not responsible
for targeting the delivery device [29]. Biological molecules such as immunoglobulins and
carbohydrates are frequently utilized as targeting moieties. However, there are several examples of
targeting in which distribution control is an inherent property of the polymeric carrier. Polymer
surfactants such as block copolymers of poly(ethylene glycol) and poly(propylene oxide), also
referred to as pluronics, alter the distribution of colloidal carriers around the body.30,31 The
change in distribution depends on the ability of the surfactant polymer to change protein
adsorption on the particle surfaces (section IV.B.2). In another case, the polymer drugconjugate
contains a spacer molecule that is sitespecifically cleaved. One application of this targeting
approach is the delivery of drugs to the colon, and site-specific cleavage is ensured by the
presence of linkages that are only degraded by bacteria present in that section of the
gastrointestinal tract.

4.1.1. Biodegradable Hydrophobic Polymers

52

A. Overview
Classification of polymers in controlled release applications can be difficult due to the inherent
diversity of structures. However, it is beneficial to attempt this classification because it can
highlight common properties within groups of polymers. In broad terms, polymers may be
classified as either biodegradable or nonbiodegradable. Biodegradable systems have garnered
much of the recent attention and development in drug delivery systems because nonbiodegradable
systems need retrieval or further manipulation after introduction into the body.

JU

In the realm of degradable polymers, there exists another level of classification based upon the
mechanism of erosion. The term degradation specifically refers to bond cleavage, whereas
erosion refers to depletion of material. Degradation is a chemical process; erosion is a physical
phenomena reliant on dissolution and diffusion processes. Two mechanisms of polymer erosion
can be identified, surface and bulk erosion. In practical terms, these two mechanisms represent
extremes. For most biodegradable polymers both mechanisms will occur, but the relative extent of
surface or bulk erosion varies radically with the chemical structure of the polymer backbone.

JI

N
G
LM

-Z

Surface erosion occurs when the rate of erosion exceeds the rate of water permeation into the
bulk of the polymer. This is often considered to be a desirable mechanism of erosion in drug
delivery because the kinetics of erosion, and hence the rate of drug release, are highly
reproducible. Furthermore, the magnitude of erosion may be changed by simply changing the
surface area of the drug delivery device. The slow rate of water permeation into surface eroding
devices has a further beneficial effect of protecting water labile drugs up to the time of drug
release. Examples of surface eroding polymers discussed in this review are the poly(anhydrides)
and the poly(ortho esters). Both of these classes of biodegradable polymers possess highly labile
groups that ensure rapid hydrolysis of polymer chains encountering water molecules. Water
permeation is retarded by designing the polymers with hydrophobic monomer units. Alternatively,
hydrophobic excipients can be added to stabilize the polymer bulk. In ideal surface erosion, the
erosion rate is directly proportional to external surface area. Surface erosion can lead to zero-order
drug release provided that diffusional release is limited and the overall shape remains constant.

Bulk erosion occurs when water molecules are able to permeate into the bulk of the polymer
matrix at a quicker rate than erosion. As a consequence, polymer molecules in the bulk may be
hydrolyzed and the kinetics of polymer degradation/erosion are more complex than for surface
eroding polymers. The majority of biodegradable polymers used in controlled drug delivery
undergo bulk erosion, including the very important poly(ester) materials. While the more limited
predictability of erosion and the lack of protection of drug molecules are inherent disadvantages to
bulk eroding devices, these properties do not inhibit their successful employment as drug delivery
devices. In addition, many new applications in controlled release use nano- or microparticle
formulations that possess massive surface areas resulting in bulk and surface eroding materials
possessing similar erosion kinetics.
Within the scope of biodegradable systems, natural polymers, particularly those in the
poly(saccharide) family (e.g., starch, cellulose, and chitosan), are being investigated.33 They are
53

referred to as biopolymers, and synthesis of this class of polymers is limited to the manipulation of
bulk material to enhance their viability. Due to the physicochemical limitations of natural
materials, there is significant exploration of synthetic materials which can be readily tailored to
offer properties for specific applications. For example, degradation of synthetic polymer can be
limited to 1 week or 1 month, depending on the desired range of therapeutic effect. The ability to
design biomaterials with specified release, mechanical, and processing properties has opened
opportunities for synthetic chemists in the controlled release arena.

N
G
LM

-Z

JU

Historically, homopolymers such as the poly(esters) (section III.B) were first in the discovery
process for synthetic biomaterials due to their availability. As properties are defined and utilized
from homopolymer systems, copolymer systems emerge that combine and merge desired function
for more effective systems. Biodegradable materials possess chemical functionalities that are
unstable within living environments, e.g., anhydride, ester, or amide bonds. The most common
routes of biodegradation in vivo are hydrolysis and enzymatic cleavage resulting in scission of the
polymer backbone. However, for some polymers, cleavage of a side chain results in a
watersoluble polymeric product that can be excreted. Biodegradation is frequently a desirable
property for controlled release applications because metabolism and excretion of the polymer
results in complete removal. In the presence of enzymes, rates of biodegradation are enhanced.
The role of degradative enzymes is to both facilitate the mechanism as well as increase the
degradation rate. Table 1 provides an overview of polymeric systems used in controlled release as
a function of chemical makeup of the polymer backbone with corresponding references.
Table 1 includes a diverse range of functionalities: from nonbiodegradable systems based on
C-C backbones to potentially degradable heteroatom-containing polymer backbones. In the
remainder of this review, a number of polymer backbones that may confer biodegradibility will be
detailed in the text. This restriction certainly does not reduce the impact and significance of C-C
backbones for controlled release applications but is simply a mechanism to focus on an important
subset of materials.
B. Poly(esters)

JI

Poly(esters) are the best characterized and most widely studied biodegradable system. The
synthesis of poly(esters) has received as much attention as the degradation of these materials. A
patent for the use of poly(lactic acid) (PLA) as a resorbable suture material was first filed in
1967.34 The mechanism of degradation in poly(ester) materials is classified as bulk degradation
with random hydrolytic scission of the polymer backbone.

Poly(esters) have been extensively employed in drug delivery applications and comprehensively
reviewed.35-38 The predominant synthetic pathway for production of poly(esters) is from
ring-opening polymerization of the corresponding cyclic lactone monomer. The more prominent
poly(esters) and their starting materials are shown in Figure 4. Polymerization of the cyclic
lactone alone is usually too slow to produce high molecular weight material (>20 000 amu). The
rate of ring opening for the cyclic lactone can be increased by activation of a Zn- or Sn-based
catalyst with the carbonyl ester. However, the introduction of a catalyst invites concerns over
traces of potentially cytotoxic material. Thus, stannous octoate SnII(CO2CH(nBu)(Et))2 is
commonly used because it has FDA approval as a food stabilizer.35 Alternatively, resorbable
Fe(II) salts have been utilized as initiators for lactide polymerization above 150 C.39 Zinc
powder and CaH2 have also been evaluated as potential nontoxic catalysts for copolymer
54

formation of poly(lactic acid) (PLA) with poly(ethylene oxide) (PEG).

1. Poly(lactic acid), Poly(glycolic acid), and Their Copolymers

N
G
LM

-Z

JU

Poly(esters) based on poly(lactic acid) (PLA), poly(glycolic acid) (PGA), and their copolymers,
poly(lacticacid-co-glycolic acid) (PLGA), are some of the best defined biomaterials with regard to
design and performance. Lactic acid contains an asymmetric R-carbon which is typically
described as the D or L form in classical stereochemical terms and sometimes as the R and S form,
respectively. For homopolymers, the enantiomeric forms are poly(D-lactic acid) (PDLA) and
poly(L-lactic acid) (PLLA). The physicochemical properties of optically active PDLA and PLLA
are nearly the same, whereas the racemic PLA has very different characteristics.41 For example,
racemic PLA and PLLA have Tgs of 57 and 56 C, respectively, but PLLA is highly crystalline
with a Tm of 170 C and racemic PLA is completely amorphous.

Figure 4. Ring-opening polymerization of selected cyclic lactones to give the following: (a) poly(_-caprolactone)
(PCL); (b) poly(glycolic acid) (PGA); (c) poly(L-lactic acid) (PLA).

The stereochemical dependence of PLA from D- and L-lactide (DD and LL cyclic dimers of
lactic acid enantiomers) has also been studied as a function of the catalyst.42,43 Diad and tacticity
concerns for the polymerization with Sn- and Zn-based initiators have shown a preference for
DD/LL and LL/DD heterotactic additions. Such stereoregular concerns are known to affect the
mechanical, thermal, and biological properties of PLA.

JI

Because the naturally occurring lactic acid is L (or S), PLLA is considered more biocompatible.
The polymers are derived from monomers that are natural metabolites of the body; thus
degradation of these materials yields the corresponding hydroxy acid, making them safe for in
vivo use. Biocompatibility of the monomer is the foundation for biocompatibility of degradable
polymer systems. To this end, the degradation products often define the biocompatibility of a
polymersnot necessarily the polymer itself. Even though PLGA is extensively used and represents
the gold standard of degradable polymers, increased local acidity due to the degradation can lead
to irritation at the site of the polymer employment. Introduction of basic salts has been
investigated as a technique to control the pH in local environment of PLGA implants.
From a physical level of understanding, poly(esters) undergo bulk degradation. PLA
homopolymers degrade slower than PGA homopolymers on the basis of crystallinity as well as
steric inhibition by the pendent methyl group of PLA to hydrolytic attack. However, the
complexity of PLA, PGA, and PLGA degradation has been demonstrated by Vert45 and does not
conform to a simple model. Vert and coworkers have demonstrated that a size dependence for
hydrolytic degradation exists for PLA systems. Other research efforts suggest that PLA-derived
microparticles will degrade faster than nanoparticles derived from PLA.46,47 This is modeled on
55

a diffusionreaction phenomena. An autocatalytic effect at the interior of larger devices is thought

to contribute to the initial heterogeneous degradation of larger devices as acidic byproducts cannot
readily diffuse out from the interior as is the case for smaller constructs. Extensive degradation
studies have also been reported for PLA, poly(caprolactone) (PCL), and their copolymers both in
vitro [48] and in vivo.

2. Poly(ethylene glycol) Block Copolymers

JU

Studies in hydrolytic degradation for poly(esters) have focused on understanding the effects of
changes in polymer chain composition. A distinguishable effect based on end group composition
for poly(ester) degradation demonstrated that terminal carboxyl groups have a catalytic effect on
hydrolysis for PGA.50 The ability to tailor rates of protein release from PLGA microspheres was
derived from the understanding of end-group effects.51 The commercial developmental process
for formulating poly(esters) with selected drug candidates has been reviewed.52 The
aforementioned review highlights the development of poly(ester) matrices containing human
growth hormone that sustained levels of a therapeutic protein in humans for 1 month from a single
dose.

-Z

Poly(ethylene glycol) (PEG) is also referred to as poly(ethylene oxide) (PEO) at high molecular
weights. Biocompatibility is one of the most noted advantages of this material. Typically, PEG
with molecular weights of 4000 amu is 98% excreted in man.

N
G
LM

One of the emerging uses for inclusion of PEG in a controlled release system arises from its
protein resistivity.54 The hydrophilic nature of PEG is such that water hydrogen bonds tightly
with the polymer chain and thus excludes, or inhibits, protein adsorption. Many research groups
are investigating attachment of PEG chains to therapeutic proteins; PEG chains at the surface
allow for longer circulation of the protein in the body by prolonging biological events such as
endocytosis, phagocytosis, liver uptake and clearance, and other adsorptive processes.

JI

PLA-PEG copolymer systems (Figure 5) possess surfactant properties because the PEG block is
very hydrophilic and the PLA block is hydrophobic. Therefore, when PLA-PEG is employed in a
fabrication process that uses an aqueous external phase, e.g., particle fabrication by the double
emulsion technique, PEG enriches the surface. The inclusion of PEG in copolymer systems
imparts extremely beneficial surface properties within the body because of the ability to repel
proteins within aqueous environments.54 This repulsion inhibits the adsorption of proteins to the
polymer surface and, therefore, prevents many polymer-cell interactions. For example,
nanoparticles made from diblock PLA-PEG copolymer have increased blood circulation times
(decreased clearance) in vivo above that of particles made from PLA alone.60 Further studies
demonstrated that PLA-PEG nanoparticles were inert toward proteins of the coagulation
system.61 Cannizzaro et al. have demonstrated that the PLA-PEG structure may act as the
foundation for more complex biodegradable materials. They synthesized a PLA-PEG polymer to
which a biotin molecule was grafted to the free end of the PEG chain. The new polymer was
designed to simplify the engineering of the polymer surface via the use of avidin-biotin
interaction.

56

Figure 5. Synthesis of PLA-PEG copolymers: (a) PLA/PEG; (b) PLA/PEG/PLA; (c) a multiblock copolymer of
L-lactide and ethylene oxide.

N
G
LM

-Z

JU

PEG can be made with a range of terminal functionalities which lends to its easy incorporation
into copolymer systems. PEG is commonly terminated with chain-end hydroxyl groups which
provide a ready handle for synthetic modification. Diblock PLA/PEG and triblock PLA/PEG/PLA
systems have been synthesized and characterized with various PLA contents.40,63-66 The free
hydroxyl groups of PEG are ring-opening initiators for lactide in forming the diblock or triblock
materials (Figure 5a,b). Recently, Chen et al. have synthesized PLA-PEG multiblock copolymers
from L-lactide and ethylene oxide, the monomer precursors for PLA and PEG, respectively
(Figure 5c).67 This approach is different in two respects: (i) use of bimetallic catalysts which
proceed by anionic mechanisms; (ii) multiblock polymers are generated.
Han and Hubbell further demonstrated the synthetic utility for PLA-PEG systems by
introducing acrylate moieties to form cross-linked systems.68 Similarly, Jeong et al. prepared
thermosensitive PLA-PEO hydrogels that exhibit temperature-dependent gel-sol transition for use
as injectable drug delivery systems.
C. Poly(ortho esters)

The motivation for designing poly(ortho esters) for drug delivery was the need to develop
biodegradable polymers that inhibited drug release by diffusion mechanisms and allowed drug
release only after the hydrolysis of polymer chains at the surface of the device.

JI

Most research on poly(ortho esters) has focused on the synthesis of polymers by the addition of
polyols to diketene acetals. For example, Heller et al. have described the synthesis and application
of the 3,9-diethylidene-2,4,8,10-tetraoxaspiro[5.5]undecane (DETOSU)-based poly(ortho
esters).71 The basic structure is formed by the addition of the DETOSU monomer to a diol to
form the chemical structure shown in Figure 6.

57

Figure 6. Degradation of the 3,9-bis(ethylidene-2,4,8,10-tetraoxaspiro[5.5]undecane) (DETOSU)-based

poly(ortho esters).

-Z

JU

The DETOSU-based poly(ortho esters) contain acid labile ortho ester linkages in their backbone
structure.Within aqueous environments, the ortho ester groups are hydrolyzed to form
pentaerythritol dipropionate and diol monomers as breakdown products (Figure 6). The
pentaerythritol dipropionate is further hydrolyzed to pentaerythritol and acetic acid.

N
G
LM

Acid-catalyzed hydrolysis of these polymers can be controlled by introducing acidic or basic

excipients into matrixes. Rates of hydrolysis can be increased by the addition of acidic excipients,
such as suberic acid, as demonstrated by the zero-order release of 5-fluorouracil over a 15 day
period.72 Alternatively, basic excipients stabilize the bulk of the matrix but diffuse out of the
surface region, thereby facilitating surface-only erosion. This approach has been employed in the
temporal controlled release of tetracycline over a period of weeks in the treatment of periodontal
disease.

JI

Recently, a number of changes in diol structure have been attempted to avoid the need for
acidic excipients. These new poly(ortho ester) structures address the problem of acidic excipient
diffusion from matrices which leads to unpredictable degradation kinetics. Ng et al. described the
synthesis of selfcatalyzed poly(ortho esters) that contain glycolide sequences that can be
hydrolytically degraded without excipient catalysis. Once degraded, these sequences then catalyze
ortho ester bond breakage, hence forming a self-catalyzing system. The synthesis of these
polymers is shown in Figure 7.

Figure 7. Synthesis of a poly(ortho ester) containing glycolic acid dimer.

A useful feature of the DETOSU systems is the ability to control the mechanical properties by
58

changing the diol monomer ratios within the final polymeric structure. For example, Heller et al.
have shown that the glass transition temperature of polymers containing a rigid diol monomer
(transcyclohexanedimethanol) and a flexible monomer (1,6-hexanediol) could be varied between
20 and 105 by increasing the proportion of the rigid diol.70 This control can also be achieved
with the glycolidecontaining polymers.
A number of applications have been described for cross-linked poly(ortho esters) formed by the
substitution of 1,2,6-hexanetriol for 1,2-hexanediol, for example. The triol monomer allows
cross-linked materials to be formed that are semisolid materials.75 It has been envisaged that these
materials could be injected into the patient as a viscous liquid at slightly elevated temperatures
that form nondeformable depot implants upon cooling.

N
G
LM

D. Poly(anhydrides)

-Z

JU

A further modification of diol structure has been introduced to allow acid-catalyzed hydrolysis
to be highly pH sensitive for applications requiring response. This modification involves the
formation of a tertiary amine-containing polymer by incorporating N-methyldiethanolamine.76
This polymer has been proposed as a material for the pulsatile delivery of insulin in which the
drug delivery system includes the enzyme glucose oxidase. High glucose levels decrease
environmental pH due to the activity of glucose oxidase. In turn, lowered pH increases the rate of
poly(ortho ester) hydrolysis thereby increasing insulin release and creating a negative feedback
mechanism.

To obtain a device that erodes heterogeneously, the polymer should be hydrophobic yet contain
watersensitive linkages. One type of polymer system that meets this requirement is the
poly(anhydrides). Poly(anhydrides) undergo hydrolytic bond cleavage to form water-soluble
degradation products that can dissolve in an aqueous environment, thus resulting in polymer
erosion. Poly(anhydrides) are believed to undergo predominantly surface erosion due to the high
water lability of the anhydride bonds on the surface and the hydrophobicity which prevents water
penetration into the bulk.77 This process is similar to the slow disappearance of a bar of soap over
time. The decrease in the device thickness throughout the erosion process, maintenance of the
structural integrity, and the nearly zero-order degradation kinetics suggest that heterogeneous
surface erosion predominates.

JI

The majority of poly(anhydrides) are prepared by melt-condensation polymerization. Starting

with a dicarboxylic acid monomer, a prepolymer of a mixed anhydride is formed with acetic
anhydride. The final polymer is obtained by heating the prepolymer under vacuum to remove the
acetic anhydride byproduct. The most widely studied poly(anhydrides) are based on sebacic acid
(SA), p-(carboxyphenoxy)propane (CPP), and p-(carboxyphenoxy)hexane (CPH) (Figure 8).

Figure 8. Structures of widely used aromatic poly(anhydrides) based on monomers of p-(carboxyphenoxy)propane (CPP) and p-(carboxyphenoxy)hexane (CPH) and aliphatic poly(anhydrides) based on sebacic acid (SA).

Degradation rates of these polymers can be controlled by variations in polymer composition.

59

The more hydrophobic the monomer, the more stable the anhydride bond is to hydrolysis.
Aliphatic poly(anhydrides) (e.g., SA) degrade within days whereas aromatic poly(anhydrides)
(e.g., CPH) degrade over several years.
Poly(anhydrides) are best formed into drug-loaded devices by compression-molding or
microencapsulation because of their high melting temperatures. A wide variety of drug and
proteins, such as insulin, enzymes, and growth factors, have been incorporated into
poly(anhydride) matrixes and their in vitro and in vivo release characteristics evaluated. Leong et
al. demonstrated that reaction of the poly(anhydrides) with drug molecules containing nucleophilic
groups did not occur during fabrication using solvent-casting techniques or when low
temperatures are maintained during compression molding.

JU

The biocompatibility of copolymers of SA and CPP has been well established. Evaluation of the
toxicity of poly(anhydrides) show that they possess excellent in vivo biocompatibility. Recent
clinical trials have demonstrated that an intracranial device of SA/CPP copolymers improves the
therapeutic efficacy of an antitumor agent, bischloronitrosourea, for patients suffering from a
lethal type of brain cancer.

N
G
LM

-Z

Another type of poly(anhydride) is poly(fatty acid dimer-sebacic acid). These are synthesized
from hydrophobic dimers of erucic acid and sebacic acid (Figure 9). They undergo surface erosion
as indicated by the presence of an erosion zone, independence of erosion rate on device thickness,
and low water contact in the polymer interior during erosion. In vitro and in vivo elimination of
the polymers is dependent upon monomer solubility. For example, the elimination time for
polymers based on watersoluble monomers (shorter side chains) was 7-14 days, whereas for
polymers based on monomers with low water-solubility (longer alkyl side chains) elimination took
nearly 8 weeks.

Fig. 9. Poly(fatty acid dimer-sebacic acids) synthesized from hydrophobic dimers of erucic acid and sebacic acid.

JI

As the polymers degrade, most of the fatty acid dimers deposit on the surface of the polymer
matrixes to effectively act as a diffusion barrier for the release of low molecular weight
compounds (e.g., drugs or monomers) from polymer devices.
1. Poly(anhydride-imides)
Poly(anhydrides) have been modified by inclusion of amino acids such as glycine and alanine
into the polymer backbone (Figure 10) to increase the mechanical properties (e.g., Youngs
modulus and compressive strength) of the poly(anhydrides).

Figure 10. Poly(anhydrides) monomers that yield poly(anhydride-imides) include amino acids.
60

The amino acids are incorporated by imide bonds at the amino terminus, leaving the terminal
carboxylic acids available for activation by acetic anhydride. 87 The amino acid-containing units
can then be copolymerized with activated monomers such as SA and/or CPH.
The poly(anhydride-imides) appear to undergo predominantly surface erosion similar to the
poly- (anhydrides).88,89 IR and NMR spectral data verify the visual evidence that degradation of
the poly- (anhydride-imides) happens in several stages. Water is first absorbed into the matrix,
with hydrolysis of the anhydride bonds occurring at the forefront of the inner erosion zone.
Hydrolysis of the polymer backbone continues until the monomer units are solubilized in water.
Finally, monomer units were removed by diffusion through the polymer matrix.
2. Poly(anhydride-esters)

JU

Other modifications of poly(anhydrides) include poly(anhydride-esters), which include two

different types of hydrolytically cleavable bonds in the polymer backbone. In one example, low

-Z

molecular weight carboxylic acid-terminated prepolymers of poly(-caprolactone) were coupled

via anhydride linkages.91 The intent of this research was to design polymers that displayed
two-stage degradation profiles: anhydride bonds rapidly hydrolyzed to the poly(ester) prepolymers
which degraded much more slowly.

N
G
LM

In another example, carboxylic acid-terminated monomers that contain ester bonds are activated
and then polymerized using the same chemistry described for the poly(anhydrides). A unique
aspect of these poly(anhydride-esters) is that hydrolytic degradation of the polymer backbone
yields a therapeutically useful compound, salicylic acid (Figure 11).

Figure 11. Poly(anhydride-esters) that degrade into salicylic acid, an antiinflammatory agent.

JI

As stated previously (section III.B.1), biocompatibility of polymers degradation products

typically define the biocompatibility of the polymers themselves. This work is the first example
where the polymers degradation products are potentially beneficial.
E. Poly(amides)

The most interesting class of poly(amides) for controlled release are the poly(amino acids). The
synthetic ability to manipulate amino acid sequences has seen its maturity over the last two
decades with new techniques and strategies continually being introduced. An excellent review of
the history of amino acid-derived polymers is given by Nathan and Kohn.93 Poly(amino acids)
have been used predominantly to deliver low molecular weight drugs, are usually tolerated when
implanted in animals,94 and are metabolized to relatively nontoxic products.95 These results
suggest good biocompatibility, but their mildly antigenic nature makes their widespread use
uncertain. Another concern with poly(amino acids) is the intrinsic hydrolytic stability of the amide
bond which must rely upon enzymes for bond cleavage. The dependence on enzymes generally
61

results in poor controlled release in vivo.

N
G
LM

-Z

JU

The expense and difficulty in production of elaborate polypeptide sequences has limited the
composition to homopolymers, predominantly poly(glutamic acid) and poly(aspartic acid).
Poly(amino acids) are generally hydrophilic with degradation rates dependent upon hydrophilicity
of the amino acids. Amino acids are attractive due to the functionality they can provide a polymer.
For example, poly(lactic acid-colysine) (PLAL) was synthesized using a stannous octoate catalyst
from lactide and a lysine-containing monomer analogous to lactide (Figure 12). Inclusion of the
amino acid lysine provides an amino group that allows for further modification of the PLAL
system.98 Recently, peptide sequences that promote cell adhesion have been attached to PLAL.

Figure 12. Poly(lactic acid-co-amino acid) (PLAL) polymer system.

The use of N-carboxyanhydride-activated amino acids was the first efficient method for
production of amino acid homopolymers. Hrkach et al. have recently exploited the PLAL system
by reaction with lysine N-carboxyanhydride derivatives to increase the systems functionality with
a poly(lysine) graft.100 PLAL has been formulated into microspheres that exhibit deep lung
delivery from porous particles.101

JI

Poly(amino acids) can be modified to enhance release or bioavailability of drug by attaching the
drug molecule to the polymer via carboxylate bonds. An example is the attachment of the
chloroformate derivative of norethindrone (a steroid) to poly[N-(3-hydroxypropyl)-L-glutamine].
The polymer conjugates were designed as insoluble particles for prolonged drug release and act by
penetrating cells and then releasing drug by the action of lysosomal enzymes. Enzymatic
degradation of synthetic polypeptides and poly(amino acids) along with an evaluation of their
utility as drug delivery systems has been reviewed.
1. Poly(iminocarbonates)
Poly(amino acids) are highly insoluble, nonprocessible, and antigenic when the polymers
contain three or more amino acids.105 To circumvent these problems, pseudo-poly(amino acids)
synthesized from tyrosine dipeptide were investigated.106 These degradable polymers are derived
from the polymerization of desaminotyrosyl tyrosine alkyl esters. The general structure of these
polymers is shown in Figure 13. Tyrosine-derived poly(carbonates) are readily processible
polymers that support the growth and attachment of cells and have also shown a high degree of
tissue compatibility.107 Tyrosine-derived poly(carbonates) are characterized by their relatively
high strength and stiffness exceeding poly(esters) such as poly(ortho esters) but not poly(lactic
62

acid) or poly(glycolic acid).108,109 The postulated mechanism of in vitro degradation involves

hydrolysis of the pendent ester bonds and the imino-carbonate bonds of the backbone.110
Degradation rates are comparable to the degradation rate of poly(L-lactic acid), occurring over a
period of months. Poly(iminocarbonates) are currently being investigated for use in small bone
fixation devices as bone screws and pins.

Figure 13. Degradable polymers derived from the polymerization of desaminotyrosyl tyrosine alkyl esters.

F. Phosphorus-Containing Polymers
1. Poly(phosphazenes)

-Z

JU

Poly(phosphazenes) provide an interesting contrast to the development of poly(ester),

poly(ortho ester), and poly(anhydride) systems because properties and biodegradation kinetics are
generated by structural changes in the side-chain structure rather than the polymer backbone.112
Biomedical poly(phosphazenes) are synthesized
by molecular substitution of
poly(dichlorophosphazene) as shown in Figure 14.

N
G
LM

Generally, it is difficult to perform substitution reactions on polymers because of the lowered

reactivity of the side groups. Yet due to the high reactivity of uncrosslinked
poly(dichlorophosphazenes) side groups, these polymers can readily undergoes halogen
replacement. Poly(phosphazenes) are of particular interest because of their unique inorganic
phosphorus-nitrogen backbone and remarkable synthetic versatility. Comprehensive reviews for
these polymers have recently been published by Scopelianos and Allcock. The poly(phosphazenes)
provide covalent and coordinate drug binding sites and breakdown into nontoxic products such as
phosphate, ammonia, amino acids, and ethanol.

JI

Biodegradable poly(phosphazenes) that are insoluble in water prior to hydrolysis have been
employed in the temporal controlled release of many drug classes including nonsteroidal
anti-inflammatory agents and peptides. For these types of applications, poly(organophosphazenes)
have been synthesized that possess amino acid side groups. When these polymers degrade, they
form amino acid, ethanol, phosphate, and ammonium salts. The mechanical properties and rates of
degradation have been controlled by appropriate selection of amino acid sidechain structures.117
The versatility of these polymers has been demonstrated by the formation of 200 nm diameter
poly(organophosphazene) nanoparticles that present covalently coupled poly(ethylene glycol)
(PEG) at their surfaces.118 In a development that parallels the synthesis of the self-catalyzed
poly(ortho esters) (section III.C), Schacht et al. incorporated hydrolysissensitive ester groups that
generate pendant carboxylic acid groups which can catalyze the degradation of the inorganic
backbone.
A number of approaches have been proposed to generate cross-linked poly(phosphazene) for
temporal controlled release. Poly[bis(carboxylatophenoxy)-phosphazene] was cross-linked in the
presence of Ca2+ ions to produce an ionically stabilized system. This polymer allowed drug
molecules to be encapsulated into poly(phosphazene) microspheres under mild environmental
conditions. pH-sensitive hydrogels have been synthesized by the formation of poly(phosphazenes)
with oxybenzoate and methoxyethoxyethoxy side groups (Figure 15). Swelling at different pH

63

values was controlled by varying the ratios of the two side groups.

JU

Figure 14. Formation of poly(phosphazenes) and examples of backbone modification.

2. Poly(phosphoesters)

-Z

Figure 16. Formation of poly(phosphoester-urethanes).

N
G
LM

Leong et al. have incorporated phosphoester groups into poly(urethanes).122 For years,
poly(urethanes) have been used as blood-contacting biomaterials because of range of physical
properties that can be obtainedsfrom hard and brittle to soft and tacky. Poly(urethanes) were
designed to be inert biomaterials, but for some applications controlled biodegradation is desirable.
Leong introduced a phosphoester linkage into the poly(urethanes) to provide biodegradable
materials that maintain the mechanical properties inherent in the poly(urethanes).
Poly(phosphoester-urethanes) are obtained by reaction of diisocyanates and polyols (e.g., PEG)
with phosphates added as chain extenders (Figure 16).
Hydrolysis of the poly(phosphoester-urethanes) yields phosphates, amines, alcohols, and carbon
dioxide. Phosphoester bonds are readily cleaved under physiological conditions.

JI

In addition, the pentavalency of the phosphorus provides a site for future functionalization. For
example, Leong et al. observed that the release kinetics of poly(phosphoester-urethanes) were
influenced by the side chains attached via the phosphoester of the polymer backbone. The release
mechanism was found to be a combination of diffusion, swelling, and degradation.
Conclusions

This review has focused on some of the more widely studied synthetic biodegradable polymers
considered or used for controlled release applications. Yet many of the future challenges we face,
such as gene therapy delivery, may require degradable polymer systems modeled with unique
requirements for specific applications. At present, there remains a scarcity of materials that can be
evaluated for biomaterial applications even though there are many research groups actively
designing new materials. One unique approach to polymer design is the use of combinatorial
methods to design arrays of new polymeric materials. For example, Kohn et al. created a
permutationally designed library of over 100 copolymers. 124 For nonchemists needing polymers
for biomedical applications, combinatorial methods provide a large selection of materials to
evaluate for their required applications.
The ability to impart bioadhesivity, cell specificity, active transport, or other specific
64

characteristics into a biocompatible polymer represents an important synthetic challenge.

Biodegradable polymers have had a remarkable impact on the science of controlled drug delivery
and promise to have an even greater impact in human health care.

4.1.3. Hydrogels
A. Introduction

-Z

JU

On the occasion of the publication of the 50th volume of our illustrious journal, it is important to
recall the major changes of the pharmaceutical yield in the last 50 years. In this changing scientific
and educational world, the contributions of the pharmaceutical industry are leading to major new
solutions of signicant medical problems. No longer is the treatment of diabetes, osteoporosis,
asthma, cardiac problems, cancer and other diseases based only on conventional pharmaceutical
formulations. In fact, Professor Kenneth Dill of the University of California in San Francisco
pointed out in a recent article in `Nature' that `a key problem of biology and medicine this century
has been to reduce the problems of disease to problems of molecular science. Many of the
associated methodological advances in biomedical sciences are the result of earlier investments in
the basic sciences. Dill eloquently closes by stating that `breakthroughs in molecular science have
opened a flood-gate of new opportunities for curing disease'.

N
G
LM

A significant such opportunity has appeared in the pharmaceutical sciences over the past 20
years with the examination of advanced drug delivery formulations. These formulations do not
simply release the drug, peptide or protein at some characteristic rate, but do so in a way that the
pharmaceutical scientist and molecular designer wants. For example, insulin maybe delivered only
when needed, calcitonin may be directed to bypass the stomach and be delivered only in the upper
small intestine, and large molecular weight, genetically-engineered molecules are delivered across
tissues at acceptable rates.

JI

These recent developments are the subject of this review, which addresses the use of
water-swollen, crosslinked biomedical materials as carriers for the development of novel
pharmaceutical formulations and for the delivery of drugs, peptides and proteins, as targeting
agents for site specific delivery, or as components for the preparation of protein or enzyme
conjugates. The network structure and the thermodynamic nature of the components of these
networks play a key role in their diffusional behavior, molecular mesh size changes (especially in
environmentallyresponsive hydrogels), and the associated molecular stability of the incorporated
bioactive agents. We wish to present new and promising techniques for the production of drug and
protein delivery formulations that have been developed in our or other laboratories.

Overview Hydrogels are three-dimensional, hydrophilic, polymeric networks capable of

imbibing large amounts of water or biological fluids. The networks are composed of
homopolymers or copolymers, and are insoluble due to the presence of chemical crosslinks
(tie-points, junctions), or physical crosslinks, such as entanglements or crystallites [4-9]. The latter
provide the network structure and physical integrity. These hydrogels exhibit a thermodynamic
compatibility with water which allows them to swell in aqueous media.
There are numerous applications of these hydrogels, in particular in the medical and
pharmaceutical sectors [13-15]. Hydrogels resemble natural living tissue more than any other class
of synthetic biomaterials. This is due to their high water contents and soft consistency which is
similar to natural tissue [13]. Furthermore, the high water content of the materials contributes to
their biocompatibility. Thus, hydrogels can be used as contact lenses, membranes for biosensors,
65

linings for artificial hearts, materials for artificial skin, and drug delivery devices [13-17].
Classification Hydrogels can be classified as neutral or ionic, based on the nature of the side
groups. According to their mechanical and structural characteristics, they can be classified as
affine or phantom networks. Additionally, they can be homopolymer or copolymer networks,
based on the method of preparation. Finally, they can be classified based on the physical structure
of the networks as amorphous, semicrystalline, hydrogen-bonded structures, supermolecular
structures and hydrocolloidal aggregates.

B. Monomers and hydrogel structure

JU

Hydrogels may also show a swelling behavior dependent on the external environment. These
polymers are physiologically- responsive hydrogels [23], where polymer complexes can be broken
or the network can be swollen as a result of the changing external environment. These systems
tend to show drastic changes in their swelling ratio as a result. Some of the factors affecting the
swelling of physiologically-responsive hydrogels include pH, ionic strength, temperature and
electromagnetic radiation [23].

N
G
LM

-Z

Monomers A convenient way to classify hydrogels is based on the nature of the side groups;
they can be either neutral or ionic. The chemical nature and number of these pendent groups can
be precisely controlled by the choice of the chemical entities used in the polymer synthesis. A
summary of monomers most commonly used in the preparation of polymeric materials in the
pharmaceutical eld is given in Table 1.

The literature on the synthesis of new polymer materials has exploded since the first biomedical
application of poly(hydroxyethyl methacrylate), known also as PHEMA, by Wichterle and Lim
[24] in 1961. Instead of utilizing `off-the-shelf' polymeric materials designed for use in consumer
applications and adapting them for medical purposes, researchers are trying to intentionally design
materials that would solve specic drug delivery problems [25]. Therefore, while novel materials
based on polymers such as PHEMA, poly(N-isopropylacrylamide) (PNIPAAm), and poly(vinyl
alcohol) (PVA), are synthesized primarily by new and innovative preparation techniques, a
number of new monomers have been prepared for production of polymers with desired properties
as well.

JI

Hydrogels are also used as carriers that can interact with the mucosa lining in the
gastrointestinal (GI) tract, colon, vagina, nose and other parts of the body due to their ability to
prolong their residence time at the delivery location [26]. The interaction between such carriers
and the glycoproteins in the mucosa is thought to occur primarily via hydrogen bonding.
Therefore, materials containing a high density of carboxyl and hydroxy groups appear to be
promising for this type of application. Monomers most often used for the synthesis of
mucoadhesive polymers include acrylic and methacrylic acid (MAA). The idea of adhesion
promoters diffusing across the polymer/mucin interface has also been introduced [27]. Chains of
polymerized ethylene glycol, either freely loaded in the carrier or grafted to the polymer surface,
have been utilized as adhesion promoters [27]. The stealth properties of poly(ethylene glycol),
known also as PEG, have also been used to reduce the uptake of particulate carriers by the
reticuloendothelial system [28]. PEG has also been shown to both lengthen the biological half-life
and reduce the immunogenicity of high molecular weight substances, such as adenosine
deaminase (ADA) and asparaginase [29].

66

JU
-Z

Network structure The suitability of a hydrogel as a drug delivery device and its performance
in a particular application depend to a large extent on its bulk structure. A number of excellent
reviews discuss this topic in great detail. The most important parameters used to characterize the

N
G
LM

network structure of hydrogels are the polymer volume fraction in the swollen state (2,s),
molecular weight of the polymer chain between two neighboring crosslinking points (Mc), and the
corresponding mesh size ().

The polymer volume fraction in the swollen state is a measure of the amount of fluid imbibed
and retained by the hydrogel. The molecular weight between two consecutive crosslinks, which
can be either of a chemical or physical nature, is a measure of the degree of crosslinking of the
polymer. It is important to note that due to the random nature of the polymerization process itself,
only average values of Mc can be calculated. The correlation distance between two adjacent

JI

crosslinks, , provides a measure of the space available between the macromolecular chains
available for the drug diffusion; again, it can be reported only as an average value. These
parameters, which are related to one another, can be determined theoretically or through the use of
a variety of experimental techniques. Two methods that are prominent among the growing number
of techniques utilized to elucidate the structure of hydrogels due to their frequent use, are the
equilibrium swelling theory and the rubber elasticity theory.
Equilibrium swelling theory

The structure of hydrogels that do not contain ionic moieties can be analyzed by the
Flory-Rehner theory [10]. This thermodynamic theory states that a crosslinked polymer gel, which
is immersed in a fluid and allowed to reach equilibrium with its surroundings, is subject only to
two opposing forces, the thermodynamic force of mixing and the retractive force of the polymer
chains. At equilibrium, these two forces are equal. Eq. (1) describes the physical situation in terms
of the Gibbs free energy.

Here, Gelastic is the contribution due to the elastic retractive forces developed inside the gel,
and Gmixing is the result of the spontaneous mixing of the fluid molecules with the polymer
67

chains. The term Gmixing is a measure of the compatibility of the polymer with the molecules
of the surrounding fluid. This compatibility is usually expressed by the polymer-solvent
interaction parameter, 1 [12].
Differentiation of Eq. (1) with respect to the number of solvent molecules, while keeping the
temperature and pressure constant, results in Eq. (2), where is the chemical potential of the
penetrating solvent.

JU

In Eq. (2), 1 is the chemical potential of the solvent in the polymer gel and 1,0 is the chemical
potential of the pure solvent. At equilibrium, the difference between the chemical potentials of the
solvent outside and inside the gel must be zero. Therefore, the changes of the chemical potential
due to mixing and elastic forces must balance each other. The change of chemical potential due to
mixing can be expressed using heat and entropy of mixing.

N
G
LM

-Z

The change of chemical potential due to the elastic retractive forces of the polymer chains can
be determined from the theory of rubber elasticity [12,35]. Upon equating these two contributions,
an expression can be written for the determination of the molecular weight between two adjacent
crosslinks of a neutral hydrogel prepared in the absence of a solvent (Eq. (3))

Here, Mn, is the molecular weight of the polymer chains prepared under identical conditions, but
in the absence of the crosslinking agent, is the specific volume of the polymer, and V1 is the
molar volume of water.

JI

Peppas and Merrill [36] modified the original Flory-Rehner theory for hydrogels prepared in the
presence of water. The presence of water effectively modifies the change of chemical potential
due to the elastic forces. This term must now account for the volume fraction density of the chains
during crosslinking. Eq. (4) predicts the molecular weight between crosslinks in a neutral hydrogel
prepared in the presence of water.

Here, 2,r is the polymer volume fraction in the relaxed state, which is defined as the state of the
polymer immediately after crosslinking, but before swelling.
The presence of ionic moieties in hydrogels makes the theoretical treatment of swelling much
more complex. In addition to the Gmixing and Gelastic in Eq. (1), there is an additional
contribution to the total change in Gibbs free energy due to the ionic nature of the polymer
network, Gionic (Eq. (5))

Upon differentiating Eq. (5) with respect to the number of moles of solvent keeping T and P
constant, an expression similar to Eq. (2) for the chemical potential can be derived in Eq. (6)

Here, the ionic is the change of chemical potential due to the ionic character of the hydrogel.
68

JU

Expressions for the ionic contribution to the chemical potential have been also developed [37-39].
They exhibit strong dependencies on the ionic strength of the surrounding media and on the nature
of the ions present in the solvent. Eqs. (7) and (8) are expressions that have been derived for the
swelling of anionic and cationic hydrogels prepared in the presence of a solvent, respectively.

Rubber elasticity theory

-Z

In these expressions, I is the ionic strength, and Ka and Kb are the dissociation constants for the
acid and base, respectively.

N
G
LM

Hydrogels resemble natural rubbers in their remarkable roperty to elastically respond to applied
stresses. A hydrogel subjected to a relatively small deformation, less than 0%, will fully recover to
its original dimension in a rapid fashion. This elastic behavior of hydrogels can be used to
elucidate their structure by utilizing the rubber elasticity theory originally developed by Treloar
[35] and Flory [40,41] for vulcanized rubbers and modified to polymers by Flory [12]. However,
the original theory or rubber elasticity does not apply to hydrogels prepared in the presence of a
solvent. Such expressions were developed by Silliman [42] and later modified by Peppas and
Merrill [43].

JI

Here, we present only the form of rubber elasticity theory used to analyze the structure of
hydrogels prepared in the presence of a solvent, and leave it up to the reader to consult the original
reference for detailed derivations.

In Eq. (9), is the stress applied to the polymer sample, is the density of the polymer, R is the
universal gas constant, T is the absolute experimental temperature, and Mc is the desired
molecular weight between crosslinks.
In order to perform an analysis of the structure of hydrogels using the rubber elasticity theory,
experiments need to be performed using a tensile testing system. Interestingly, the rubber
elasticity theory has not only been used to analyze chemically, but also physically, crosslinked
hydrogels, as well as hydrogels exhibiting temporary crosslinks due to hydrogen bonding.
Calculation of the mesh size
The primary mechanism of release of many drugs from hydrogels is diffusion, occurring
through the space available between macromolecular chains. This space is often regarded as the
`pore'. Depending upon the size of these pores, hydrogels can be conveniently classified as (1),
macro-porous; (2), micro-porous; and (3), non-porous. A structural parameter that is often used in
69

describing the size of the pores is the correlation length, , which is defined as the linear distance
between two adjacent crosslinks, and can be calculated using the following Eq. (10)
(10)

Here, is the elongation ratio of the polymer chains in any direction and r02 is the
root-mean-square, unperturbed, end-to-end distance of the polymer chains between two
neighboring crosslinks. For isotropically swollen hydrogels, the elongation ratio, , can be related
to the swollen polymer volume fraction, 2,s, using Eq. (11).
(11)

(12)

N
G
LM

(13)

-Z

JU

The unperturbed end-to-end distance of the polymer chain between two adjacent crosslinks can be
calculated using Eq. (12), where Cn is the Flory characteristic ratio, l is the length of the bond
along the polymer backbone (for vinyl polymers 1.54), and N is the number of links/chain that
can be calculated by Eq. (13).

In Eq. (13), Mr is the molecular weight of the repeating units from which the polymer chain is
composed. Finally, when one combines Eqs. (10)-(13), the correlation distance between two
adjacent crosslinks in a swollen hydrogel can be obtained (Eq. (14))

(14)

A detailed theoretical characterization of the network structure of the polymer carrier in terms of
the correlation length, , in combination with diffusion studies of model drugs and proteins,
provides an invaluable insight into the very complex structure of polymer networks and aids in the
design of drug delivery carriers [49].

JI

C. Physical, chemical and toxicological properties of hydrogels

Factors affecting swelling of hydrogels

The crosslinking ratio is one of the most important factors that affects the swelling of hydrogels. It
is defined as the ratio of moles of crosslinking agent to the moles of polymer repeating units. The
higher the crosslinking ratio, the more crosslinking agent is incorporated in the hydrogel structure.
Highly crosslinked hydrogels have a tighter structure, and will swell less compared to the same
hydrogels with lower crosslinking ratios. Crosslinking hinders the mobility of the polymer chain,
hence lowering the swelling ratio.
The chemical structure of the polymer may also affect the swelling ratio of the hydrogels.
Hydrogels containing hydrophilic groups swell to a higher degree compared to those containing
hydrophobic groups. Hydrophobic groups collapse in the presence of water, thus minimizing their
exposure to the water molecule. As a result, the hydrogels will swell much less compared to
hydrogels containing hydrophilic groups.
70

Swelling of environmentally-sensitive hydrogels can be affected by specific stimuli. Swelling of

temperature-sensitive hydrogels can be affected by changes in the temperature of the swelling
media. Ionic strength and pH affect the swelling of ionic strength- and pH-sensitive hydrogels,
respectively. There are many other specific stimuli that can affect the swelling of other
environmentally-responsive hydrogels.
Dynamics of swelling
The swelling kinetics of hydrogels can be classified as diffusion-controlled (Fickian) and
relaxation-controlled (non-Fickian) swelling. When water diffusion into the hydrogel occurs much
faster than the relaxation of the polymer chains, the swelling kinetics is diffusion-controlled. A
nice mathematical analysis of the dynamics of swelling is presented by Peppas and Colombo.
Mechanical properties

-Z

JU

Mechanical properties of hydrogels are very important for pharmaceutical applications. For
example, the integrity of the drug delivery device during the lifetime of the application is very
important to obtain FDA approval, unless the device is designed as a biodegradable system. A
drug delivery system designed to protect a sensitive therapeutic agent, such as protein, must
maintain its integrity to be able to protect the protein until it is released out of the system.

N
G
LM

Changing the degree of crosslinking has been utilized to achieve the desired mechanical
property of the hydrogel. Increasing the degree of crosslinking of the system will result in a
stronger gel. However, a higher degree of crosslinking creates a more brittle structure. Hence,
there is an optimum degree of crosslinking to achieve a relatively strong and yet elastic hydrogel.
Copolymerization has also been utilized to achieve the desired mechanical properties of hydrogels.
Incorporating a comonomer that will contribute to H-bonding can increase the strength of the
hydrogel.
Cytotoxicity and in-vivo toxicity

JI

Cell culture methods, also known as cytotoxicity tests, can be used to evaluate the toxicity of
hydrogels. Three common assays to evaluate the toxicity of hydrogels include extract dilution,
direct contact and agar diffusion. Most of the problems with toxicity associated with hydrogel
carriers are the unreacted monomers, oligomers and initiators that leach out during application.
Therefore, an understanding the toxicity of the various monomers used as the building blocks of
the hydrogels is very important.

The relationship between chemical structures and the cytotoxicity of acrylate and methacrylate
monomers has been studied extensively [51]. Several measures have been taken to solve this
problem, including modifying the kinetics of polymerization in order to achieve a higher
conversion, and extensive washing of the resulting hydrogel. The formation of hydrogels without
any initiators has been explored to eliminate the problem of the residual initiator. The most
commonly used technique has been gamma irradiation. Hydrogels of PVA have been also made
without the presence of initiators by using thermal cycle to induce crystallization [8]. The crystals
formed act as physical crosslinks. These crystals will be able to absorb the load applied to the
hydrogels.
D. Stimuli-sensitive swelling-controlled release systems
Environmentally-sensitive hydrogels have the ability to respond to changes in their external
environment. They can exhibit dramatic changes in their swelling behavior, network structure,
71

permeability or mechanical strength in response to changes in the pH or ionic strength of the

surrounding fluid, or temperature [57]. Other hydrogels have the ability to respond to applied
electrical or magnetic fields, or to changes in the concentration of glucose [57]. Because of their
nature, these materials can be used in a wide variety of applications, such as separation
membranes, biosensors, artificial muscles, chemical valves and drug delivery devices [57].
pH-sensitive hydrogels
Hydrogels exhibiting pH-dependent swelling behavior can be swollen from ionic networks.
These ionic networks contain either acidic or basic pendant groups. In aqueous media of
appropriate pH and ionic strength, the pendant groups can ionize, developing fixed charges on the
gel. As a result of the electrostatic repulsions, the uptake of solvent in the network is increased.

N
G
LM

-Z

JU

Ionic hydrogels are swollen polymer networks containing pendent groups, such as carboxylic or
sulfonic acid, which show sudden or gradual changes in their dynamic and equilibrium swelling
behavior as a result of changing the external pH. In these gels, ionization occurs when the pH of
the environment is above the pKa of the ionizable group. As the degree of ionization increases
(increased system pH), the number of fixed charges increases, resulting in increased electrostatic
repulsions between the chains. This, in turn, results in an increased hydrophilicity of the network,
and greater swelling ratios. Conversely, cationic materials contain pendent groups such as amines.
These groups ionize in media which are at a pH below the pKb of the ionizable species. Thus, in a
low pH environment, ionization increases, causing increased electrostatic repulsions. The
hydrogel becomes increasingly hydrophilic and will swell to an increased level.

There are many advantages to using ionic over neutral networks in drug delivery. Their
characteristics can be exploited for applications in a wide variety of biomedical applications, such
as dental adhesives and restorations, controlled release devices, prodrugs and adjuvants, and
biocompatible materials. The swelling of polyelectrolyte gels is significantly affected by the ionic
strength of the swelling agent. As the ionic strength of the swelling agent increases, the
concentration of ions within the gel must increase in order to satisfy the Donnan equilibrium. The
swelling force is reduced due to increased gelcounter ion interaction and a decrease in the osmotic
swelling forces.

JI

Many researchers have studied the dynamic swelling of pH-sensitive networks. The early work
of Katchalsky established that the collapse and expansion of PMAA gels occurred reversibly by
simply adjusting the pH of the fluid. Ohmine and Tanaka observed the sudden collapse of ionic
networks in response to sudden changes in the ionic strength of the swelling medium. Studies by
Khare and Peppas examined the swelling kinetics of poly(MAA) or poly(acrylic acid) with
poly(hydroxy ethyl methacrylate). They observed pH- and ionic strength-dependent swelling
kinetics in these gels.

Hydrogels composed of lightly crosslinked N-isopropylacrylamide (NIPAAm) and MAA were

synthesized and characterized for their sensitivity to external conditions and their ability to control
the release of two antithrombotic agents, heparin and streptokinase. PNIPAAm is noted for its
sharp change in swelling behavior across the lower critical solubility temperatures of the polymer,
while PMAA shows pH-sensitive swelling due to ionization of the pendant carboxylic groups in
the polymer. Hydrogel copolymers ofNIPAAm and MAA with appropriate compositions were
designed to sense small changes in blood stream pH and temperature to deliver antithrombotic
agents, such as streptokinase or heparin, to the site of a blood clot [73].
72

Experiments were performed to show that hydrogels with certain compositions could show both
temperature- and pH sensitivity, and that these changes could control the release of heparin or
streptokinase. Equilibrium and pulsatile swelling studies were performed on all polymers to
determine to what extent the hydrogels would respond to changes in environmental temperature
and pH, and how fast that response would be.

JU

Antithrombotic agents were loaded into the hydrogels by partitioning, and released into
buffered solutions as a function of pulsatile changes in pH, temperature and a combination of
temperature and pH. The results show that crosslinked polymers containing at least 75% NIPAAm,
the rest being MAA, showed a sharp temperature-dependent swelling behavior, while gels even
with minimal MAA content were sensitive to solution pH, swelling to a large extent in basic
buffer solutions. The drug release studies indicated that pH- or temperature-control of the drug
release was marginal, while the combined effects of temperature and pH over small ranges caused
the drug release to be patterned after the swelling state of the hydrogel. Because the response time
of the gels was long, drug release was often not immediate upon external stimulation, but lagged
behind the pH or temperature profile.

-Z

We have also synthesized hydrogels based on polyacrylamide, pH-sensitive hydrogel

(poly(acrylamide (AAm)-coacrylic acid)), and temperature-sensitive hydrogels, e.g.
poly(AAm-co-diethylAAm). The swelling behavior of these networks was characterized over a

N
G
LM

temperature range of 16-45C, and a pH range from 2 to 9. Only the diethylAAm-containing

hydrogel demonstrated temperature-sensitivity, with swelling ratios decreasing with increasing
temperature over the temperature range studied. In pH buffer solutions, acrylic copolymers
showed relatively small swelling ratios in low pH buffers, and extremely high degrees of swelling
in high pH regions.

These systems were functionalized with an enzyme (trypsin) by binding the enzyme into a
polymer network during a polymerization reaction. The activity of trypsin was then recorded
under various temperature and pH conditions, using an assay of a substrate affected by trypsin.
Trypsin activity was found to be largely a function of the swelling behavior of the hydrogel
networks, with the largest difference being between the high and low temperature regions, which
inuence the swelling characteristics of the poly(diethylAAm-co-AAm) hydrogel.

JI

Temperature-sensitive hydrogels
Temperature-sensitive hydrogels have gained considerable attention in the pharmaceutical field
due to the ability of the hydrogels to swell or deswell as a result of changing the temperature of
the surrounding fluid. Numerous researchers studied various applications of these hydrogels, such
as on-off drug release regulations, biosensors and intelligent cell culture dishes.
Thermosensitive hydrogels can be classified as positive or negative temperature-sensitive
systems. A positive temperature-sensitive hydrogel has an upper critical solution temperature
(UCST). Such hydrogels contract upon cooling below the UCST. Negative temperature-sensitive
hydrogels have a lower critical solution temperature (LCST). These hydrogels contract upon
heating above the LCST.
Some of the earliest work with temperature-sensitive hydrogels was done by the group of
Tanaka. PNIPAAm is the best example of a negative temperature-sensitive hydrogel. Hitotsu et al.
worked with crosslinked PNIPAAm and determined that the LCST of the PNIPAAm gels was
34.3C. They also found that the LCST could be increased by mixing small amounts of ionic
73

copolymers in the gels. Beltran et al. also worked with PNIPAAm gels containing ionic
comonomers. They observed results similar to those achieved by Tanaka.
Hoffman [78] proposed the application of PNIPAAm and its copolymers for
temperature-modulated drug release by bulk squeezing and surface regulation. In the bulk
squeezing system, the drug that is distributed evenly inside the matrix is squeezed out of the
system due to the deswelling of the hydrogel as a result of increasing the temperature of the
environment above the volume phase transition temperature. In the surface regulation system, the
swelling ratio of the skin layer is increased as the temperature of the system is lowered below the
volume phase transition temperature, and hence, the drug molecules will be able to diffuse through
the skin layer.

JU

Chen and Hoffman [79] prepared P(NIPAAm-g-AA) gels which exhibited temperature- and
pH-sensitive behavior. These gels were able to respond rapidly to both temperature and pH
changes. The temperature- and pH-dependent swelling behavior was more pronounced in the graft
copolymers than in random copolymers containing similar amounts of pH- and
temperature-sensitive components.

N
G
LM

-Z

The group of Okano [80-82] developed an ingenious method to prepare comb-type graft
hydrogels of PNIPAAm. The main chain of the crosslinked PNIPAAm contained small molecular
weight grafts of PNIPAAm. Under conditions of gel collapse (above the LCST), hydrophobic
regions were developed in the pores of the gel, resulting in a rapid collapse. These materials had
the ability to collapse from a fully swollen conformation in less than 20 min, while comparable
gels that did not contain graft chains required up to a month to fully collapse.

JI

The ability of PNIPAAm and its copolymers to exhibit a hydrophilic nature below the LCST
and a hydrophobic nature above the LCST has attracted many researchers to create surfaces for
cell culture systems. Most cells attach and grow on hydrophobic surfaces, such as polystyrene.
However, cells do not attach to a hydrophilic surface, and hence, will not grow. Currently, the
method to detach the grown cell is by using enzyme or mechanical means. These methods have
been proven to cause damages to the cells. By applying a coating of the temperature-sensitive
polymer on top of the cell culture dishes, the cells can be detached easily by their own mechanism
by simply changing the temperature. By changing the temperature from above the LCST to below
it, the nature of the coating changes from hydrophobic to hydrophilic, and hence, the cells will
detach from the hydrophilic surface.

Inoue et al. synthesized hydrogels grafted with oligomers with two different LCSTs. The
oligomers of choice were carboxy-terminated oligo NIPAAm, oligo(N-vinylcaprolactam) (VCL)
and a random co-oligomer of NIPAAm and AAm. The resulting hydrogels showed two volume
phase transition temperatures corresponding to the LCSTs of the grafted side chains of oligo
NIPAAm and oligo VCL. They explored the possibility of applying this hydrogel to store and
retrieve information as a function of temperature.

Conjugates of PNIPAAm with various enzymes have also been reported [89,90]. Hoffman and
associates synthesized conjugates of oligomer of PNIPAAm and trypsin. These conjugates are
soluble in solution and can catalyze enzymatic reactions. They can then be separated from the
solution by thermal precipitation. The recovery of the conjugates by thermal precipitation was
highly efficient (more than 95%) even after 14 cycles through the LCST. The enzyme conjugates
were found to be more stable than the native trypsin both in solution and precipitated states.
74

Jeong et al. [91] synthesized a new biodegradable triblock copolymer of poly(ethylene

glycol-b-(dl-lactic acid-coglycolic acid)-b-ethylene glycol) (PEG-PLGA-PEG). This copolymer
shows sol-to-gel or gel-to-sol transitions as temperature increases monotonically. This
thermoreversible copolymer is very important for drug delivery applications because it can be
injected as a free-flowing solution (sol) at room temperature. Upon injection into the body, the
copolymer becomes a gel at body temperature.

JU

Another promising application of temperature-sensitive gels is controlled drug delivery for

highly sensitive therapeutic agents. Kim and associates [92] developed a hydrogel of PNIPAAm
and PAA which was able to effectively release the protein drug, calcitonin, in response to
temperature and pH changes. Similar work was done by the group of Peppas [93]. They prepared
block copolymers of PNIPAAm and PMAA which had the ability to respond to both temperature
and pH. Using these materials, they were effectively able to modulate the release kinetics of
streptokinase.

JI

N
G
LM

-Z

Micro- or nano-particulate devices composed of hydrogels or containing hydrogel components

may have a great advantage over disk- or film-shaped macrogels hitherto developed, because they
may have an improved swelling kinetics of hydrogels involving thermosensitivity of drug release
property and a wide variety of biomedical applications due to their small size. A typical example
of such devices was reported very recently [94]. Fig. 1 illustrates microcapsules showing the
thermosensitive, controlled release in question. A key structural feature of this microcapsules is
their composite coat, consisting of nanoparticles with crosslinked poly(NIPAAm) shells dispersed
in a thermo-insensitive ethylcellulose matrix. As shown in Fig. 1, at low temperatures, the
permeability of the membranes is low because of absence of void formation by the swollen
PNIPPAm shells in the microcapsule membrane, leading to a low release rate of drugs. At high
temperatures, the PNIPAAm shells should collapse. Thus, numerous voids can be formed in the
membranes providing high drug release rates. This microcapsule demonstrated thermally on-off
switching release rate changes within a few min. As such, miniaturization of the hydrogel, as well
as the device, make it possible to achieve an exceptionally rapid, sharp thermosensitive response.

Fig. 1. Schematic diagram showing the ideal structure of microcapsules with thermally-modulated onoff pulsatile
drug release.

Other stimuli-sensitive hydrogels

Several stimuli, other than pH and temperature, can trigger drug release from a depot. These
include physical stimuli, such as light [95], magnetic eld [96], electric current [97] and
ultrasound [98], which can be applied to the systems externally, and chemical stimuli, like ionic
75

species [99], certain chemical substances and biological compounds [100]. In some cases, their
effective applications can be found in engineering, rather than pharmaceutical, fields.

-Z

JU

Versatile stimuli-sensitive controlled release systems can be fabricated, provided that the
hydrogels are well designed to alter their configuration in response to these stimuli based on
almost infinitely available mechanisms. Meanwhile, the elucidation of the release mechanisms
relying on more complicated mass transport phenomena over conventional diffusion-regulating
systems is a practically important issue in the substantial application of stimuli-sensitive
controlled release devices. In this context, Lavon and Kost [97] examined mass transport
enhancement in non-erodible polymeric controlled release systems by ultrasound for a better
understanding of the ultrasound-enhancing drug release phenomenon. They suggested that the
enhancing effect of ultrasound on drug release from the systems is due to the contribution of a
convective term, generated by cavitation, without any destructive effect of the morphology of the
polymers. Hsu and Block [98] studied electrokinetic phenomena in three types of anionic gels
(agarose, agarose-carbomer 934P, and agarose-xanthan gum) under an applied electric current for
electrically-modulated drug delivery. It was shown that electrical current strength and gellant
content could influence both syneresis and drug migration. These findings reported by the two
research groups may give a practically useful insight into the application of hydrogels to
transdermal delivery assisted by electroportion, iontophoresis, or sonophoresis.

N
G
LM

Several classes of hydrogels that respond to specific molecules have been proposed to date.
Calcium-responsive bioerodible drug delivery systems were devised by Goldbart and Kost [99].
The system composed of a starch-cellulose matrix containing a-amylase, the activity of which is
regulated by calcium. The principle relying on calcium-responsiveness is based on the mechanism
that a-amylase in its non-active form is incorporated into the matrix composed of starch, and the
matrix thus responds to calcium which causes the non-active a-amylase to become active. As the
unstabilized calcium shows the calcium concentration-dependent activity, the release of drugs
incorporated in the matrix accompanying the degradation of the starch matrix is regulated in a
calcium concentration-dependent manner.

JI

Very recently, two fascinating stimuli-sensitive hydrogelbased drug delivery systems were
proposed in quick succession. Miyazaki et al. [100,101] designed a novel reversibly
antigen-responsive hydrogel based on a unique idea. This hydrogel was an AAm-based
semi-interpenetrating polymer network (IPN), consisting of an antibody-grafted linear PAAm and
the crosslinked PAAm grafted with the corresponding antigen. In the absence of a free antigen, the
hydrogel can shrink due to the intra-chain antigen-antibody binding in the polymer network, while
it swells in the presence of the free antigen because of dissociation of the intra-chain binding by
exchange of the grafted antigen for free antigen. This swelling/shrinking process was shown to be
reversible. Due to this property, the hydrogel membrane allowed the antigen-responsive change of
hemoglobin permeation through the membrane in response to stepwise changes in the antigen
concentration. A striking feature of the antigen-sensitive hydrogel would be the specificity of
molecular recognition. Using this hydrogel, a sensing device with a broad application for
immunoassay and antigen sensing is expected to be fabricated. A series of studies on the
fabrication of a novel microbial infection-responsive drug release system was accomplished by
Suzuki and associates. They prepared a PVA-based hydrogel with specially designed
thrombin-sensitive peptide linkers. In this system, a remarkably increased thrombin-like activity in
microbial-infected wound exudates is utilized as a biological signal for microbial infection.
76

Antibiotics covalently attached to the PVA-based hydrogel via the thrombinsensitive peptide
linkers can be released from the hydrogel only in the presence of infection, because of
thrombin-sensi-tive cleavage of the peptide linkers. The hydrogel can be used for a wound
dressing with microbial infection-responsive controlled release of antibiotics.
The ultimate goal of stimuli-sensitive controlled release is the ability to integrate the systems to
respond to more than two stimuli. Such a multi-stimuli-sensitive device has been fabricated by
Kaetsu et al. [106]. It was realized by the idea based on the simultaneous installation of discrete
hydrogels showing different stimuli-response into the device.
E. Applications of hydrogels in drug delivery

JU

A number of strategies have been proposed to achieve drug delivery systems for efcient therapy.
Among them, hydrogels have attracted considerable attention as excellent candidates for
controlled release devices, bioadhesive devices, or targetable devices of therapeutic agents.

N
G
LM

-Z

Hydrogel-based delivery devices can be used for oral, rectal, ocular, epidermal and
subcutaneous application. Fig. 2 illustrates various sites that are available for the application of
hydrogels for drug delivery. Excellent reviews in relation to this topic are readily available.
Historical research trends on hydrogel formulations for pharmaceutical applications, as well as the
anatomy and physiology of each administration site, can be found in these reviews. Therefore, the
present paper will mainly survey recent reports published in the last few years.

Fig. 2. Tissue locations applicable for hydrogel-based drug delivery systems.

JI

Peroral drug delivery

Drug delivery through the oral route has been the most common method in the pharmaceutical
applications of hydrogels. In peroral administration, hydrogels can deliver drugs to four major
specific sites; mouth, stomach, small intestine and colon. By controlling their swelling properties
or bioadhesive characteristics in the presence of a biological fluid, hydrogels can be a useful
device for releasing drugs in a controlled manner at these desired sites. Additionally, they can also
adhere to certain specific regions in the oral pathway, leading to a locally increased drug
concentration, and thus, enhancing the drug absorption at the release site.
Drug delivery in the oral cavity
Drug delivery to the oral cavity can have versatile applications in local treatment of diseases of
the mouth, such as periodontal disease, stomatitis, fungal and viral infections, and oral cavity
cancers. Long-term adhesion of the drugcontaining hydrogel against copious salivary flow, which
bathes the oral cavity mucosa, is required to achieve this local drug delivery. For this purpose,
77

many types of bioadhesive hydrogel systems have been devised since the early 1980s. Some of
these are already on the market. For example, a bioadhesive tablet developed by Nagai et al. is
commercially available under the brand name Aftach. This product is composed of a double
layer, with a bioadhesive layer made of hydroxypropyl cellulose and poly(acrylic acid) and a
lactose non-adhesive backing layer. It is a local delivery system of triamcinolone acetonide for the
treatment of aphthous ulcers.

JU

A hydrogel-based ointment can also be utilized for the topical treatment of certain diseases in
the oral cavity. It can be used not only as a drug delivery device, but also as a liposome delivery
vehicle. The possible advantage of liposome delivery with this ointment is that the use of
liposomal formulations with encapsulated drug can lead to an increase of local, and a decrease of
systemic, drug concentration, because of the encapsulation of drugs with phospholipids. This may
provide more desirable properties for topical use, such as reduction of uncontrolled release of
drugs into the blood circulation and certain undesirable side effects, compared with the
conventional ointment-drug formulations.

-Z

Petelin et al. investigated the pharmaceutical performance of three different hydrogel-based

ointments as possible vehicles for liposome delivery into the oral cavity tissues by electron
paramagnetic resonance (EPR). The vehicles employed were Orabase (a sodium
carboxymethylcellulose, pectin and gelatin combination in a polyethylene-paraffin base),

N
G
LM

Carbopol
934P and
neutralized
poly(MAA-co-methyl
methacrylate
(MMA)).
Liposomecontaining mucoadhesive ointments were prepared by simply mixing multilamellar
liposomes with each ointment prediluted with phosphate-buffered saline of pH 7.4 in the volume
ratio of 1:4. An EPR study showed that P(MAA-co-MMA) was the most appropriate ointment in
terms of liposomal stability in the ointment, transport of liposomeentrapped molecules from the
ointment into the oral soft tissues, and washing-out time from oral mucosa or gingvia.
The oral cavity can also provide a useful location as a transport route for heavily metabolized
drugs, since the drugs absorbed from this route bypass first-pass hepatic metabolism. Kitano et al.
proposed a hydrogel ointment containing absorption enhancers for the buccal delivery of 17

JI

-estradiol (E2) to treat osteoporosis. It is well known that the oral administration of E2 results in
very low availability due to its high first-pass effect. Ethanol solution containing E2, and glyceryl
monolaurate as an absorption enhancer, and an aqueous solution of a commercial carboxyvinyl
polymer (Hiviswako 103) and triethanolamine were mixed together to produce the hydrogel
ointment. In-vivo studies using hamsters demonstrated that the buccal administration of E2 with
this formulation allowed the maintenance of the E2 plasma level at over 300 ng/ml per cm3 for 7 h,
while no primary morphological change of buccal membrane was observed 7 h after application.
Remunn-Lpez et al. reported new buccal bilayered tablets containing nifedipine and
propranolol hydrochloride intended for systemic drug administration. The tablets, comprising two
layers, a drug-containing mucoadhesive layer of chitosan with polycarbophil and a backing layer
of ethylcellulose, were obtained by direct compression. The double-layered structure design
provided a unidirectional drug delivery towards the mucosa, and avoided a loss of drug resulting
from wash-out with saliva flow. The striking feature of this device would be the utilization of an
in-situ crosslinking reaction between cationic chitosan and anionic polycarbophil, which
progressed upon penetration of the aqueous medium into the tablet. As a result of the crosslinking
effect, the tablets showed controlled swelling and prolonged drug release, and an adequate

78

adhesiveness could be obtained.

Drug delivery in the GI tract
The GI tract is unquestionably the most popular route of drug delivery because of the facility of
administration of drugs for compliant therapy, and its large surface area for systemic absorption. It
is, however, the most complex route, so that versatile approaches are needed to deliver drugs for
effective therapy.

N
G
LM

-Z

JU

Like buccal delivery, hydrogel-based devices can be designed to deliver drugs locally to the
specific sites in the GI tract. For example, Patel and Amiji proposed stomach-specific antibiotic
drug delivery systems for the treatment of Helicobacter pylori infection in peptic ulcer disease.
For localized antibiotic delivery in the acidic environment of the stomach, they developed cationic
hydrogels with pH-sensitive swelling and drug release properties. The hydrogels were composed
of freeze-dried chitosan-poly (ethylene oxide) (PEO) IPN. pH-dependent swelling properties and
the release of two common antibiotics, amoxicillin and metronidazole, entrapped in the
chitosan-PEO semi-IPN were evaluated in enzyme-free simulated gastric fluid (SGF; pH 1.2) and
simulated intestinal fluid (SIF; pH 7.2). The swelling ratio of the hydrogels after 1 h in SGF was
found to be 16.1, while that in SIF was only 8.60. Additionally, the freeze-dried chitosanPEO
semi-IPN demonstrated fast release of the entrapped antibiotics in SGF because of its highly
porous matrix structure resulting from freezedrying. More than 65 and 59% of the entrapped
amoxicillin and metronidazole, respectively were released from the hydrogels after 2 h in SGF.
The rapid swelling and drug release demonstrated by these hydrogel formulations may be
beneficial for site-specific antibiotic delivery in the stomach, because of the limitations of the
gastric emptying time. Amiji et al. also reported enzymatically degradable gelatin-PEO semi-IPN
with pH-sensitive swelling properties for oral drug delivery. In this case, the incorporation of
gelatin in the IPN made it possible to swell in the acidic pH of the gastric fluid, due to the
ionization of the basic amino acid residues of gelatin. The IPN was found to be degraded by
proteolytic enzymes, such as pepsin and pancreatin.

JI

Undoubtedly, peroral delivery of peptides and proteins to the GI tract is one of the most
challenging issues, and thus, under much investigation. However, there are many hurdles,
including protein inactivation by digestive enzymes in the GI tract, and poor epithelial
permeability of these drugs. However, certain hydrogels may overcome some of these problems
by appropriate molecular design or formulation approaches. For example, Akiyama et al. [118]
reported novel peroral dosage forms of hydrogel formulations with protease inhibitory activities

using Carbopol(C934P), a poly(acrylic acid) product, which has been shown to have an
inhibitory effect on the hydrolytic activity of trypsin, and its neutralized freeze-dried modification
(FNaC934P). They demonstrated that two-phase formulations, consisting of the rapid gel-forming
FNaC934P and the efficient enzyme-inhibiting, but more slowly swelling, C934P, had the most
profound effect on trypsin activity inhibition.

Recently, oral insulin delivery using pH-responsive complexation hydrogels was reported by
Lowman et al. The hydrogels used to protect the insulin in the harsh, acidic environment of the
stomach before releasing the drug in the small intestine were crosslinked copolymers of PMAA
with graft chains of polyethylene glycol (P(MAAg-EG)). The insulin-containing P(MAA-g-EG)
microparticles demonstrated strong dose-dependent hypoglycemic effects in in-vivo oral
administration studies using both healthy and diabetic rats. The blood glucose levels in these
79

animals were decreased signicantly for at least 8 h due to the absorption of insulin in the GI tract.
It is worth noting that these effects were observed without the addition of additives, such as
absorption enhancers or protease inhibitors.
Due to a lower proteolytic activity in comparison to that in the small intestine, the colonic
region has also been considered as a possible absorption site for orally administered peptides and
proteins. Several hydrogels are currently being investigated as potential devices for colon-specific
drug delivery. These include chemically or physically crosslinked polysaccharides, such as
dextran, amidated pectin, guar gum and inulin, and azocross-linked poly(acrylic acid). They are
designed to be highly swollen or degraded in the presence of colonic enzymes or microflora,
providing colon-specificity in drug delivery.
Rectal delivery

N
G
LM

-Z

JU

The rectal route has been used to deliver many types of drugs, although patient acceptability is
variable due to the discomfort arising from administered dosage forms. Its primary applications
have been for local treatment of diseases associated with the rectum, such as hemorrhoids.
Additionally, it is well known that drugs absorbed from the lower part of the rectum drain into the
systemic circulation directly. Thus, the rectal route is a useful administration route for drugs
suffering heavy first-pass metabolism. Conventional suppositories hitherto adapted as dosage
forms for rectal administration are solids at room temperature, and melt or soften at body
temperature. A problem associated with rectal administration using conventional suppositories is
that drugs diffusing out of the suppositories in an uncontrolled manner are unable to be
sufficiently retained at a specific position in the rectum, and sometimes migrate upwards to the
colon. This often leads to a variation of the bioavailability of certain drugs, in particular, for drugs
that undergo extensive first-pass elimination.

JI

In this context, hydrogels may offer a valuable way to overcome the problem in conventional
suppositories, provided that they are designed to exhibit a sufficient bioadhesive property
following their rectal administration. For example, Ryu et al. reported that increased
bioavailability of propranolol subject to extensive first-pass metabolism was observed by adding
certain mucoadhesive polymeric compounds to poloxamer-based thermally gelling suppositories.
Among the mucoadhesive polymeric compounds tested, polycarbophil and sodium alginate
provided the largest mucoadhesive force and the smallest intrarectal migration to the suppositories,
resulting in the largest bioavailability of propranolol (82.3 and 84.7%, respectively). Miyazaki et
al. investigated the potential application of xyloglucan gels with a thermal gelling property as
vehicles for rectal drug delivery. Xyloglucan processed by the researchers has the sol-gel
transition temperature of around 22-27.8C, and thus, it can be a gel at body temperature; on the
other hand, it can be easily administered since it can behave as a liquid at room temperature.
In-vivo rectal administration of xyloglucan gels containing indomethacin using rabbits showed a
wellcontrolled drug plasma concentration-time profile without reduced bioavailability, when
compared to commercial indomethacin suppositories.

Avoiding rectal irritation caused by vehicles is another important issue in rectal drug delivery.
Both Ryu's and Miyazaki's [130] products, described above, revealed no evidence of mucosal
irritation after rectal administration. A significantly reduced irritation by rectal hydrogels prepared
with water-soluble dietary fibers, xanthan gum and locust bean gum, was also reported by
Watanabe et al. [131].
80

Ocular delivery
In ocular drug delivery, many physiological constraints prevent a successful drug delivery to the
eye due to its protective mechanisms, such as effective tear drainage, blinking and low
permeability of the cornea. Thus, conventional eye drops containing a drug solution tend to be
eliminated rapidly from the eye, and the drugs administered exhibit limited absorption, leading to
poor ophthalmic bioavailability. Additionally, their short-term retention often results in a frequent
dosing regimen to achieve the therapeutic efficacy for a sufficiently long duration. These
challenges have motivated researchers to develop drug delivery systems that provide a prolonged
ocular residence time of drugs.

-Z

JU

Certain dosage forms, such as suspensions and ointments, can be retained in the eye, although
these sometimes give patients an unpleasant feeling because of the characteristics of solids and
semi-solids. Due to their elastic properties, hydrogels can also represent an ocular
drainage-resistant device. In addition, they may offer better feeling, with less of a gritty sensation
to patients. In particular, in-situ-forming hydrogels are attractive as an ocular drug delivery system
because of their facility in dosing as a liquid, and their long-term retention property as a gel after
dosing.

N
G
LM

Cohen et al. developed an in-situ-gelling system of alginate with high guluronic acid contents
for the ophthalmic delivery of pilocarpine. This system significantly extended the duration of the
pressure-reducing effect of pilocarpine to 10 h, compared to 3 h when pilocarpine nitrate was
dosed as a solution. Rheological evaluation of Gelrite, deacetylated gellan gum which gels upon
instillation in the eye due to the presence of cations, was carried out by Carlfors et al. Their study
indicated that a high rate of the sol/gel transition of Gelrite in-situ gels results in long precorneal
contact times.

Chetoni et al. reported silicone rubber/hydrogel composite ophthalmic inserts. Poly(acrylic acid)
or poly(MAA) IPN was grafted on the surface of the inserts to achieve a mucoadhesive property.
The ocular retention of IPN-grafted inserts was signicantly higher with respect to ungrafted ones.
An in-vivo study using rabbits showed a prolonged release of oxytetracycline from the inserts for
several days.
Transdermal delivery

JI

Drug delivery to the skin has been traditionally conducted for topical use of dermatological
drugs to treat skin diseases, or for disinfection of the skin itself. In recent years, a transdermal
route has been considered as a possible site for the systemic delivery of drugs. The possible
benefits of transdermal drug delivery include that drugs can be delivered for a long duration at a
constant rate, that drug delivery can be easily interrupted on demand by simply removing the
devices, and that drugs can bypass hepatic first-pass metabolism. Furthermore, because of their
high water content, swollen hydrogels can provide a better feeling for the skin in comparison to
conventional ointments and patches. Versatile hydrogel-based devices for transdermal delivery
have been proposed so far. Sun et al. devised composite membranes comprising of crosslinked
PHEMA with a nonwoven polyester support. Depending on the preparation conditions, the
composite membranes can be tailored to give a permeation flux ranging from 4 to 68 mg/cm2 per
h for nitroglycerin. A Carbopol 934-based formulation containing phosphatidylcholine liposomes
(liposome-gel) was prepared by Kim et al. In their study, the skin absorption behavior of
hydrocortisone-containing liposome-gel was assessed. Gayet and Fortier [137] reported hydrogels
81

obtained from the copolymerization of bovine serum albumin (BSA) and PEG. Due to their high
water content over 96%, allowing the release of hydrophilic and hydrophobic drugs, their use as
controlled release devices in the field of wound dressing was proposed as the potential application
of the BSA-PEG hydrogels. Comprehensive studies on in-situ photopolymerizable hydrogels
made from terminally diacrylated ABA block copolymers of lactic acid oligomers (A) and PEG (B)
for barriers and local drug delivery in the control of wound healing have been carried out by
Hubbell.

Subcutaneous delivery

JU

Recent research trends in transdermal applications are focusing on electrically-assisted delivery,

using iontophoresis and electroporation. Several hydrogel-based formulations are being
investigated as vehicles for transdermal iontophoresis to obtain the enhanced permeation of
luteinizing hormone releasing hormone [140], sodium nonivamide acetate [141], nicotine [142]
and enoxacin [143]. On the other hand, a methyl cellulose-based hydrogel was used as a viscous
ultrasonic coupling medium for transdermal sonophoresis assisted with an AC current, resulting in
an enhanced permeation of insulin and vasopressin across human skin in vitro.

N
G
LM

-Z

As described through Sections A-D, hydrogels posses a wide variety of possible pharmaceutical
applications. Among them, their substantial application may be found in implantable therapeutics.
Subcutaneously inserted exogenous materials may more or less evoke potentially undesirable
body responses, such as inflammation, carcinogenecity and immunogenecity. Therefore,
biocompatibility is a prerequisite that makes materials implantable. Due to their high water
content, hydrogels are generally considered as biocompatible materials. They also provide several
promising properties: (1), minimal mechanical irritation upon in-vivo implantation, due to their
soft, elastic properties; (2), prevention of protein adsorption and cell adhesion arising from the low
interfacial tension between water and hydrogels; (3), broad acceptability for individual drugs with
different hydrophilicities and molecular sizes; and (4), unique possibilities (crosslinking density
and swelling) to manipulate the release of incorporated drugs. Some of these may offer an
advantage for the delivery of certain delicate drugs, such as peptides and proteins.
Giammona et al. developed new hydrogels originating from the chemical reticulation of

JI

,-polyasparthydrazide (PAHy) by glutaraldehyde. PAHy is a new watersoluble macromolecule,

synthesized from a polysuccinimide by reaction with hydrazine. Histological analysis revealed
that this hydrogel was inert when implanted subcutaneously into rats.
Several hydrogel formulations for the subcutaneous delivery of anticancer drugs have been also
proposed. For example, crosslinked PHEMA with good biocompatibility was applied to cystabine
(Ara-C) and methotrexate. Poly(AAm-co-monomethyl or monopropyl itaconate) developed by
Blanco's group was employed for the controlled release of Ara-C [149] and 5-fluorouracil.
Current studies on implantable hydrogels have been directed towards the development of
biodegradable systems requiring no follow-up surgical removal once the drug supply is depleted.
A bioerodible hydrogel based on a semi-IPN structure composed of a poly(1-caprolactone) and
PEG macromer terminated with acrylate groups was devised by Cho et al. [152]. Long-term
constant release over 45 days of clonazepam entrapped in the semi-IPN was achieved in vivo.
Recently, two types of novel degradable PEG hydrogels for the controlled release of proteins were
developed by Zhao and Harris [153]. One type is prepared by a polycondensation reaction
between difunctional PEG acids and branched PEG polyols. Upon hydrolysis of the resulting ester
82

linkages, these gels degrade into only PEG and PEG derivatives. The other is PEG-based
hydrogels having functional groups in which protein drugs can be covalently attached to the gel
network via ester linkage. Thus, the release of the protein drugs immobilized would be controlled
by the hydrolysis of the ester linkage between the gel and the protein, followed by the diffusion of
the protein out of the gel, and by the degradation of the gel. Extensive research efforts on
degradable dextran hydrogels have been carried out by Hennink and his coworkers. These
hydrogels are based on acrylate derivatives of dextran. In their studies, the application of the
hydrogels to the controlled release of protein was thoroughly investigated. Biodegradable
crosslinked dextran hydrogels containing PEG (PEG-Dex) were reported by Moriyama and Yui
[159]. Insulin release from these hydrogels was regulated by the surface degradation of PEG-Dex
microdomain-structured.
F. The future

JU

Recent advances in the development of neutral and ionic hydrogels for drug delivery applications
have concentrated on several aspects of their synthesis, characterization and behavior. Major
questions that have been addressed or are presently researched in our work include:

-Z

synthetic methods of preparation of hydrophilic polymers with desirable functional groups.

synthetic methods of preparation of multifunctional or multiarm structures including branched
or grafted copolymers and star polymers.

N
G
LM

understanding of the criticality and the swelling/syneresis behavior of novel anionic or cationic
polymers.
development of ultrapure polymers, such as crosslinkedfree PVA gels produced by
freezing-thawing of aqueous solutions.
synthesis and characterization of biomimetic hydrogels.

understanding of the relaxational behavior during dynamic swelling.

modeling of any associated dissolution or biodegradation.

JI

New promising methods of delivery of chemotherapeutic agents using hydrogels have been
recently reported. For example, biorecognition of various sugar-containing copolymers can be
used for the release of chemotherapeutic agents. Kopecek and associates have used
poly(N-2-hydroxypropyl methacrylamide) carriers for the release of a wide range of such agents.
In the last few years, there have been new creative methods of preparation of hydrophilic
polymers and hydrogels that may be used in the future in drug delivery applications. For example,
we have synthesized self-organized nanostructures based on triblock copolymers that may have
applications in controlled drug delivery. Novel biodegradable polymers include polyrotaxanes,
which are considered to be particularly exciting molecular assemblies for drug delivery.
Dendrimers and star polymers are exciting new materials because of the large number of
functional groups available in a very small volume. Merrill [176] has offered an exceptional
review of PEO star polymers and their applications in the biomedical and pharmaceutical fields,
whereas Keys and Peppas [177] have prepared gels of controlled structure and large biological
functionality by irradiation of PEO star polymers. Such gels may be promising materials as
carriers for drug delivery if combined with the techniques of molecular imprinting. Indeed, there
have been several reports of the use of crosslinked polymers as templates for drug imprinting and

83

subsequent release. Still, this field is relatively new and its applications may not be immediately
available.
Thus, new synthetic methods have been used to prepare homo- and co-polymeric hydrogels for
a wide range of drug, peptide, and protein delivery applications. Random copolymers with
balanced hydrophobicity/hydrophilicity can offer desirable release rates and dissolution profiles,
but graft, block and comb-like copolymers offer additional advantages, especially when they
contain temperature- or pHsensitive pendent groups. Several interesting applications of such
systems in the treatment of diabetes, osteoporosis, cancer or thrombosis have been discussed.
Other hydrogels with great promise as drug delivery vehicles include neutral gels of PEO or PVA,
and gels of star molecules and other complex structures.
Ref.1. Robert S. Langer, Chem. Rev. 99(1999)3181-3198.

I. Introduction

-Z

4.2. Polymeric Prodrug

JU

Ref.2. N.A. Peppas et al. European Journal of Pharmaceutics and Biopharmaceutics, 50 (2000)
27-46.

N
G
LM

In the last decades polymer chemists have been actively involved in designing polymer materials
for biomedical applications. One field of application that has attracted polymer chemists attention
from the late 1960s onwards is the need for advanced drug delivery systems to improve drug
efficacy. Polymer materials were designed and proposed as matrices or depot systems for
injectable or implantable systems or devices. One particular approach towards an improved use of
drugs for therapeutic applications is the design of polymeric prodrugs or polymerdrug conjugates.
It was already early in the 1950s and 1960s that polymer chemists started to link drugs onto
polymers to improve their efficiency. At that time however, they were mainly concentrating on the
chemistry itself and almost any class of polymers was covalently combined with any class of
drugs. The biological aspects for the design of polymeric prodrugs were hardly taken into account.

JI

It was for the first time in 1975 that a rational model for pharmacologically active polymers was
proposed. Prof. H. Ringsdorf was the first to recognise the immense potential of polymeric
prodrugs, if only polymer chemists and biologists would work together in the field. The proposed
model consists mainly of five components: the polymeric backbone, the drug, the spacer, the
targeting group and the solubilising agent.

The polymeric carrier can be either an inert or a biodegradable polymer. The drug can be fixed
directly or via a spacer group onto the polymer backbone. The proper selection of this spacer
opens the possibility of controlling the site and the rate of release of the active drug from the
conjugate by hydrolytic or enzymatic cleavage. The most challenging aspect of this model is the
possibility of altering the body distribution and cell uptake by attaching cell-specific or
nonspecific uptake enhancers (homing devices).

This model, although still oversimplified, has been an important mark in the history of
polymeric prodrug design. It made clear that a more rational design was needed based on
information arising from biological work. This remarkable paper has also catalysed the interest of
biologists and pharmacists in synthetic polymers. As more information becomes available from
cell biology and molecular biology, polymer chemists are trying to design tailor-made polymeric
carriers that better fulfil the specified requirements.
84

In time it has been shown that there is a clear relationship between the structural elements of the
Ringsdorf model and the properties of the synthetised polymeric prodrugs based on it. In fact, the
properties that polymer chemists want to reach with the design of their polymerdrug conjugates,
are translated in the components of the model. Some of these properties and their relationship with
the components in the model will now be discussed in more detail in this paper (Fig. 1).

JU

Fig. 1. The Ringsdorf model.

II. Properties of polymeric prodrugs

A. Prolongation of action of the drug

N
G
LM

-Z

The profile of plasma concentration of drugs is an important determinant of their quantitative

access to peripheral targets. The plasma profile is usually measured as the area under the curve
(AUC). In general, slow renal elimination and metabolic inactivation promote better access of
drugs to remote targets, although this can also cause elevated toxicity. Many drugs in routine use
are membrane permeable because their sites of action are intracellular and such drugs typically
exhibit high volumes of distribution and rapid plasma clearance.
By linking a drug onto a polymer, a conjugate is obtained with a higher hydrodynamic volume.
This results in a slower renal excretion, longer blood circulation and an endocytotic cell uptake.
To select polymers as candidate drug carriers a number of requirements should be fulfilled:
availability of suitable functional groups for covalent coupling with drugs;
biocompatibility: preferably nontoxic, nonimmunogenic;

biodegradability or a molecular weight below the renal excretion limit;

availability.

JI

A number of reviews cover what has been done over the past 20 years in the field of soluble
polymers as potential drug carriers. The polymers selected for preparing macromolecular prodrugs
can be categorised according to: (a) the chemical nature (vinylic or acrylic polymers,
polysaccharides, poly(-amino acids)); (b) the back bone stability (biodegradable polymers, stable
polymers); (c) the origin (natural polymers, synthetic polymers); and (d) the molecular weight
(oligomers, polymers).
Vinyl polymers can be easily prepared by radical polymerisation of the corresponding vinyl
monomer. They are interesting drug carrier candidates. Since copolymerisation of selected
monomers results in polymers with a variable composition, different polymer properties can be
achieved. In a way, the candidate carriers can be tailor-made to fulfil the requirements for the
design of the polymerdrug conjugate. However, vinyl polymers are not biodegradable. Hence, in
order to avoid undesirable storage, the molecular weight should at least be below the renal
filtration limit (4050 kDa).
At

present,

the

most

intensively

studied

vinyl

polymers

are

copolymers

of
85

JU

N-(2-hydroxypropyl)- methacrylamide (HPMA). HPMA homopolymer was originally developed

in Czechoslovakia as a plasma expander. It is hydrophilic and non-toxic in rats. HPMA copolymer
with adriamycin as antitumor agent linked onto it with the peptidyl linker GlyPheLeuGly (PK1)
(Fig. 2), was developed by Duncan and Kopecek.

Fig. 2. Structure of PHMPA copolymer containing adriamycin (PK1).

N
G
LM

-Z

It was demonstrated that HPMAadriamycin conjugates are remarkably less toxic than the free
drug and accumulate within solid tumor models. PK1 reached phase II clinical trial for the
treatment of breast, colon and non-small-cell lung cancer. The only polymeric prodrug with
targeting moiety, that entered early clinical trials for the treatment of primary and secundary liver
cancer is PK2. This conjugate is a HPMA copolymer with adriamycin as antitumor agent and
N-acylated galactosamine as targeting group. It is developed for targeting to the liver by
facilitating the interaction with hepatocyte asiaglycoprotein receptors.

JI

Poly(styrene-co-maleic acid/anhydride) (SMA) is a vinyl polymer introduced by Maeda and

co-workers. It was used to synthetise the prodrug, SMANCS (Fig. 3), a conjugate of a
low-molecular-weight styrene maleic anhydride copolymer (SMA, 1.6 kDa) and the antitumor
protein neocarcinostatin (NCS). It is already marketed in Japan for the treatment of hepatocellular
carcinoma.

Fig. 3. Diagrammatic representation of the reaction between SMA and NCS to produce the conjugate SMANCS.

Synthetic poly(-amino acids) like poly(l-lysine),poly(l-glutamic acid), poly((N-hydroxyalkyl)glutamines) can be made by ring-opening polymerization of the N-carboxyanhydride
monomers .These polymers have functionalities in their side groups (amine, hydroxyl, carboxyl)
that allow covalent coupling with drug molecules (Fig. 4). Generally, poly(L-amino acids) are
biodegradable, whereas their D-enantiomers are not.
In our research group poly(N-(2-hydroxyethyl-lglutamine)) (PHEG) is already used for many
years in the development of polymeric prodrugs. PHEG was originally designed by Neri as a
plasma expander . It is non-toxic, biocompatible and degradable by lysosomal enzymes. The

86

synthesis and evaluation of PHEG-based drug conjugates are described in a large number of
articles.

Fig. 4. Structure of poly(N-(2-hydroxyethyl-L-glutamine)).

-Z

JU

Polysaccharides are another interesting class of drug carriers. Much attention has been directed
to the use of dextran. Sezaki and co-workers prepared dextranmitomycin conjugates with either
aminocaproic acid or 6-bromohexanoic acid as spacer (Fig. 5). The pharmacokinetics of these
conjugates proved to be dependent on the molecular weight and the electrical charge of the
polymer derivative.

N
G
LM

Fig. 5. Structure of dextran.

Dextrans are a family of polysaccharides mainly composed of 1,6-linked -D-glucose units.

Dextran with a molecular weigth below 100,000 is not immunogenic and is clinically used as
blood substitute. Dextran was claimed to be biodegradable. However, it was demonstrated by
Vercauteren that the in vitro degradation of dextrans in presence of lysosomal glucosidases or
endodextranases is rather slow. Moreover, it was shown that chemical modification of the dextran
further reduces its biodegradability.

Proteins such as serum albumin have also frequently been used for preparing polymeric
prodrugs. An interesting example is the work of Meyer and co-workers who used mannosylated
serum albumin as carrier for antiviral drugs. A disadvantage of proteins is their complexity in
chemical composition, which complicates the identification of the final conjugates.

JI

Poly(ethylene glycol) (PEG) has been used to modify a number of therapeutically interesting
proteins. PEG is a polymer with many useful properties, it is soluble in water and in organic
solvents, it is not toxic and not immunogenic. It is approved by the FDA to be used in nose sprays,
food and cosmetics. It has been clearly demonstrated by Abuchowski et al. that grafting of PEG
onto proteins reduces their immunogenicity, improves their resistance to proteolytic degradation
and improves their thermostability.
Micelle-forming block copolymers have been introduced by Kataoka and co-workers.
Conjugates of adriamycin with poly(ethylene glycol)poly(aspartamide) block copolymers tend to
form micelles. (Fig. 6). The hydrophilic PEG chains form the outer shell and the hydrophobic
poly(aspartic acid)doxorubicin components form the inner core. It was demonstrated that these
systems have a very high in vivo antitumor activity and show a reduced non-specific accumulation
in heart, lung and liver.
B. Controlled drug release
After administration, it is necessary that the macromolecular prodrug is stable during circulation
87

in the bloodstream but the cytotoxic drug should be released from the macromolecular drug
conjugate intracellulary in the lysosomes (lysosomotropic drug delivery) and/or intratumorally
(tumoritropic drug delivery). This controlled release from polymeric drug conjugates by
enzymatic or hydrolytic cleavage can only be achieved by proper selection of linkage between
drug and polymeric carrier.

JU

In the development of spacers, the most interest has been focussed on pH-sensitive spacers (pHcontrolled drug release) and oligopeptide spacers (enzyme-assisted drug release).

pH Controlled drug release

-Z

Fig. 6. Adriamycin-conjugated poly(ethylene glycol)poly(aspartic acid) block copolymer.

N
G
LM

When the macromolecular drug conjugate is taken up by the cell through endocytosis, the
conjugate is predestined to be exposed to the acidic pH of the lysosome. Also in or near the tumor
tissue the pH is slightly acidic in comparison with healthy tissue. This relatively low pH can be
exploited to design pH-sensitive spacers. Two types of acid-sensitive spacer have been frequently
studied and reviewed: the hydrazon linkage and the N-cis-aconityl spacer (Fig. 7).
The acid labile spacers were first explored by Shen and Ryser. They reported on the synthesis
of daunorubicin-linked aminoethyl polyacrylamide beads and poly(D-lysine) via a N-cis-aconityl
spacer. The pH-controlled hydrolysis of the cis-aconityl was demonstrated by measuring the half
lives at pH 5 ranging from 1 to 4 h, depending on the conjugate and the used buffer. The
cis-aconityl spacer was readily hydrolysed at pH 4 but release was not noticeable at pH 6.

JI

Another extensively studied acid labile spacer is the carboxylic hydrazon linkage. Kaneko and
co-workers synthetised a series of conjugates with hydrazon spacers in order to study the
relationship between the acid-sensitivity and cytotoxicity of adriamycinimmuno conjugates. The
immunoconjugate with propanoyl-hydrazonspacer showed the highest in vitro and in vivo
antitumor activity. The research group of Kratz developed and evaluated a series of transferrin and
albumin conjugates with anthracyclines. Coessens and co-workers linked the antibiotic
streptomycin to dextran and to poly-(N-(2-hydroxyethyl)-L-glutamine) via a carboxylic hydrazon
linkage. Release of streptomycin was demonstrated at lysosomal pH.

88

Fig. 7. The hydrazon and N-cis-aconityl spacer.

Drug release by lysosomal and/or tumor-associated enzymes

-Z

JU

After the cell uptake of the polymeric prodrug through endocytosis and after fusion of the
endosome with the lysosome, the drug conjugate is not only exposed to the acid environment but
also to the degrading nature of the lysosomal enzymes. When the lysosomal hydrolases degrade
the spacermost likely an oligopeptide spacerthe drug is released inside the cell. The
lysozymes are not only present in normal cells but are often overexpressed in tumor tissues.
Several lysosomal proteases such as cathepsins B, D and metalloproteinases, play a very important
part in tumor growth and formation of metastases.

N
G
LM

If the substrate is a specific oligopeptide for lysosomal proteases, the cytostatic drug can be
released by these enzymes in or near the tumor tissue. Subsequently, the tumor cells can be
selectively destroyed. For the design of a specific polymer drug conjugate, the site and the rate of
the cleavage will then depend on the amino acid composition of the oligopeptide.
Jatzkewitz was the first to use a dipeptide GlyLeu to couple mescaline to
poly(vinylpyrrolidone-co -acrylic acid). The attachment of the drug onto the polymer resulted in a
drastic increase in the biological half-life of mescaline. Trouet and co-workers synthetised
albumindaunorubicin conjugates with oligopeptide spacers.

JI

A major contribution to the understanding of lysosomal release was made by Duncan and
Kopecek. The release of model drug p-nitroaniline from HMPA copolymers, catalysed by
chymotrypsin, was studied by Kopecek and Chytry. The oligopeptide sequence was later
optimized for lysosomal thiol proteases, such as cathepsin B. Kopecek and Duncan also evaluated
a series of HPMAdoxorubicin conjugates with different oligopeptide spacers. In vitro release
studies in media containing lysosomal enzymes clearly demonstrated that drug release can be
tailored by the length and composition of the peptidyl spacer. As a result of a study with cathepsin
B, the oligopeptide sequence GlyPheLeuGly was incorporated in the HPMAdoxorubicin
copolymers with and without N-acetylated galactosamine as targeting moiety (PK2 and PK1,
respectively), which are currently being used in clinical trials.

The influence of the oligopeptide composition of the spacer on the drug release was also
demonstrated by De Marre and co-workers for poly((2-hydroxyethyl)-L-glutamine)
PHEGpeptide mitomycin C conjugates. Conjugates with a spacer having glycine as C-terminal
amino acid are less hydrolytical stable in aqueous buffer or serum than those having a more
hydrophobic terminal amino acid such as leucine or phenylalanine. Tetrapeptide spacers were
more susceptible to cleavage by lysosomal enzymes than tripeptides (Table 1).

89

C. The enhanced permeability and retention effect

Fig. 8. The EPR effect

JI

N
G
LM

-Z

JU

It is known by several studies that macromolecules (natural and synthetic) and macromolecular
prodrugs are taken up by solid tumors. Rapid pinocytosis by tumors cells was originally suggested
as an explaination for this passive tumor uptake. In 1986 Matsumura and Maeda proposed that this
passive targeting can be ascribed to the combination of the poor tissue drainage and an increased
tumor vascular permeability. It is termed the enhanced permeability and retention effect (EPR
effect). Due to permeability enhancing factors, such as vascular endothelial growth factor (VEGF)
and bradykinin, the endothelium of the tumor vasculature becomes discontinous. This leads to the
extravasation of the macromolecules from the bloodstream towards the tumor tissue. Additionally,
the lack of effective lymphatic drainage prevents the macromolecules or macromolecular prodrugs
from being removed and subsequently this results in an extravascular retention of the
macromolecules or the macromolecular drug conjugates. This so-called enhanced permeability
and retention effect (EPR effect) was also observed in inflamed tissue. The EPR effect is now
generally accepted and considered as a major rational for using polymeric prodrugs (Fig. 8).

D. Alteration of the body distribution and the cell uptake by active targeting

Antibody conjugates

The use of monoclonal antibodies to direct drug conjugates is based on the fact that surfaces of
tumors contain a wide variety of proteins, some of which are specific to the tumor type. The
monoclonal antibodies used as targeting group selectively seek out the tumor cells by binding to
such tumor-specific antigens. As a result, the drug conjugate should bind very specifically these
tumor cells. Frequently, however, the quantity of drug that can be selectively targeted is limited by
the number of antigens available. Hence, in cancer therapy the targeted-drug approach has been
most successful for extremely potent agents such as the plant toxins, which in conjugation with
antibodies have been termed the immunotoxins. Further problems associated with the use of
monoclonal antibodies as targeting moiety are lack of tumor selectivity, tumor access and
immunogenecity.

90

One antibody-based targeting strategy is antibodydirected enzyme prodrug therapy (ADEPT).

An enzyme, capable of converting a non-toxic prodrug into a potent cytotoxic drug, is covalently
attached to a tumor selective monoclonal antibody. Following localisation of the antibody enzyme
conjugate at the tumor site and clearance of residual conjugate from the bloodstream, the prodrug
is administered. This prodrug can be converted by the enzyme into a potent cytotoxic drug at the
tumor site, so minimising non-specific toxicity (Fig. 9).

JU

Fig. 9. Schematic representation of the ADEPT concept.

-Z

One major advantage over conventional antibodytargeting is the inherent amplification stage,
meaning that for every successful enzyme-targeting event a very large number of prodrug
molecules can be activated. Initial results have been promising, though dogged with such
problems as poor water-solubility of prodrugs, and the approach is currently being refined for
further development.

N
G
LM

Macromolecular glycoconjugates as carrier systems

JI

Sugar-specific receptors are plasma membranecomponents (either glycoproteins and glycolipids,

called lectins; Goldstein et al., 1980) of many mammalian cells. The first membrane lectin was
characterized on hepatocytes by Ashwell and Harford. Endogenous lectins, generally multivalent
in their binding and recognition capacities, are found on numerous normal and malignant cells
(Table 2).

The possibility of using hepatic lectins, such as the asialoglycoprotein receptor (ASGP-R),
recognizing galactose as targets for drug delivery is particularly attractive and has been studied
intensively as a possible target for the treatment of various liver diseases such as hepatitis,
parasitic infections and liver metastasis.
The ASGP-R (Schwartz, 1984) is easily accessible to the vascular circulation, being situated
predominantly on the blood-facing surfaces of hepatocytes. Moreover, it is present in relatively
large numbers, between 100,000 and 500,000 per cell.
The feasibility of liver targeting is well documented. As an example, Duncan and Kopecek
prepared a series of co-polymers using methacryloylated galactose units as co-monomer, resulting
in materials containing pendent galactose units. It was found that these conjugates are rapidly
91

cleared from the blood and accumulate into the liver Duncan et al., more specifically in the
hepatocytes. Galactose was also introduced on the HPMAdoxorubicin derivative (PK2). OHare
demonstrated that these derivatives associate in vitro with human hepatoma cell lines. After
intraveneous administration of PK2 70% was taken up by the liver, making this conjugate possibly
useful in the treatment of liver cancer. PK2 is now in clinical trial phase I.

JU

Vansteenkiste et al. (1991) prepared dextran and poly-HEA conjugates with pendent mono- or
triantennary glycosides. Subcellular distribution experiments indicated that mono-galactosylated
dextran is accumulated within the lysosomal compartment of liver hepatocytes. In contrast, the
tris-galactose-substituted polymer shows a greater affinity for the galactose-specific receptor in
vivo and also shows a high level of association with the cell surface of hepatocytes. This can be
explained by the so-called clustering effect. Binding to the ASGP-R depends strongly on the
structure of the oligosaccharide ligand: mono-, biand triantennary sugar units bind with increasing
affinity.

-Z

Drug delivery to macrophages (e.g. Kupffer cells) offers a second potentially attractive goal in
the development of targeted treatment of various malfunctions, notably parasitic disorders such as
Leishmaniasis or enzyme deficiencies such as Gauchers syndrome. Moreover, since macrophages
are part of the immune system, they can be activated and rendered tumoricidal by
immunostimulating agents (e.g. N-acetylmuramyldipeptide, MDP).

N
G
LM

Mannosylated carriers can also fulfil an important role not only in active drug targeting but also
in receptor blocking. It was demonstrated that mannosylated dextrans were useful as transient
receptor blockers in vivo for a 791T/36-ricin toxin A immunotoxin. The circulation half-life of the
immunotoxin was prolonged by a factor 34 up to 40 min following co-injection of an excess of
mannosylated dextran. The liver disposition of the immunotoxin was markedly reduced from 43 to
18% of the recovered dose. The influence of the molecular size as well as the sugar loading of the
competing polysaccharide was demonstrated to be small.
Targeting to angiogenic vessels

JI

Angiogenesis is a fundamental process by which new blood vessels are formed. It is essential in
reproduction, development and wound repair. However, many diseases are also driven by
persistent unregulated angiogenesis, like artritis and several eye diseases. Angiogenesis is also an
important process for tumor growth and metastasis of solid tumors. Endothelial cells in angiogenic
vessels of solid tumors show an increased expression of several cell surface proteins that stimulate
cell invasion and proliferation. These proteins include receptors for different angiogenic growth
factors such as the vascular endothelial growth factor (VEGF) and they also include the 3
integrin receptor. The 3 integrin is highly expressed in most growing tumor vasculature but has
very low expression in normal vasculature and most other normal or benign human tissue.

One type of integrin receptor binding peptides is the RGD (arginineglycineaspartic acid)
containing peptides. These peptides bind to the integrin receptors with high affinity and can
therefore be used as targeting moieties for drug delivery. Moreover, it is known that peptides
containing the RGD sequence inhibit experimental metastasis. Arap and co-workers showed that
the coupling of cyclic RGD and NGR peptides (CDCRGDCFC and CNGRCVSGCAGRC) to the
anticancer drug adriamycin resulted in an increased efficacy of the drug against human breast
cancer xenografts in mice. The research group of Mayumi prepared conjugates of RGD-peptides
(RGD and RGDS) and poly(ethylene glycol) (PEG). The inhibitory effect of these conjugates,
92

examined on experimental metastasis in mice, was demonstrated to be superior to the free RGD
peptides.
E. Immunoprotective therapy
Recently it was found that the use of polymeric antitumor drug derivatives may play an
important role in the protection of the cancer patients immune system. One of the mechanisms
that induces the programmed cell death or apoptosis of cancer cells, is the FasFas ligand
interaction.

JU

Fas and Fas ligand (FasL) are both transmembrane proteins (Nagata, 1997). Both receptor and
ligand are expressed either constitutively or after activation on most of the cells of the immune
system. Fas is also expressed on cancer cells. Interaction between Fas and FasL triggers a cascade
of signals, that eventually results in apoptosis. It has been reported however that several tumor cell
lines can express FasL. Hence, they are able to kill cells of the immune system expressing Fas.
This mechanism is called the Fas counterattack. The counterattack of the tumor cells not only
prevents the eradication of the cancer cells, but participates also in the destruction of the immune
system.

N
G
LM

-Z

The counterattack mechanism is often favoured by non-functioning down-regulation or loss of

the cancer cell Fas receptors. Moreover, it has been reported that treatment with antitumor drugs
promotes the induction of Fas ligands on the cancer cells.
There is a strong indication that treatment with macromolecular drug derivatives can overcome
this. Rihova and co-workers found a strong expression of FasL on the SW620 human metastatic
colorectal cancer cell line when it was exposed to doxorubicin or mitomycin C (MMC). However,
when the cell line was exposed to polymeric derivatives of these drugs, no increase of the FasL
was noticed on the SW620, even not when higher concentrations were used. The drug derivatives
used in this experiment wereMMCbound via a GFAL-spacer onto PEG-grafted

JI

poly(N5-(2-hydroxyethyl-l-glutamine)) (PHEG) and doxorubicin coupled via a GFLGspacer onto

poly-(N-(2-hydroxypropyl)methacrylamide) (PHPMA) with or without anti-CD71 mAbs as
targeting group. These results suggest that the expression of Fas ligands on cancer cells is different
when they are exposed to free antitumor drugs or to their macromolecular derivatives. This is an
important outcome that might indicate that polymeric prodrugs are able to protect the patients
immune system.
F. Polymeric prodrugs in clinical use or in clinical trial
The interdisciplinary research of the last decades has resulted in a number of polymer based
products which are now on the market or have entered clinical trial.
One approach of particular note involving soluble macromolecular drug carriers is SMANCS.
In the clinical formulation, neocarcinostatin (NCS) (molecular weight: 10,700) is conjugated to
two chains of a styrenemaleic anhydride copolymer (SMA) (molecular weight average 1500,
polydispersity <1.2). The SMA copolymer is itself derivatised with an alkyl group (usually butyl)
which determines the overall hydrophobicity of the conjugate.
Aqueous SMANCS formulations have been tested in pilot studies in patients with solid tumors
of the ovary, lung, stomach, adrenal gland and in the brain. Formulations based on
SMANCS/Lipiodol have been shown to be efffective both as a diagnostic tool and for therapeutic
use in solid tumors where the formulations are given arterially via a catheter. The prognosis of the
93

patient receiving intra-arterial SMANCS/Lipiodol is a 90% chance of survival for at least 5 years
after treatment, if the patient has no active liver cirrhosis and the tumor has not spread to more
than two segments of the liver. With conventional therapy the survival time is about 6 months.
SMANCS is marketed in Japan by Yamanouchi for the treatment of hepatocellular carcinoma.
PEG-modified adenosine deaminase (ADAGEN) and PEGL-asparaginase (ONCASPAR)
were the first PEG modified enzymes that were on the market in the early 1090s. PEGADA is
used for the treatment of ADA-deficient Severe Combined Immunodeficiency Syndrome.
PEGL-asparaginase is used to treat lymphocytic leukaemia and malignant lymphosarcoma.
Both native enzymes have a short plasma half-life and PEG-modification resulted in a prolonged
plasma clearance. Further more, the PEG-enzymes display a marked reduction in immunogenicity.
Both products are in clincal use today.

JU

PK1 and PK2 are both derivatives of HPMA copolymer with the antitumor agent doxorubicin
linked onto it via the peptidyl spacer GlyPheLeuGly. PK2 also contains galactose as a
targeting group to facilitate liver targeting.

N
G
LM

-Z

PK1 is currently undergoing Phase II evaluation for treatment of breast, colon and
non-small-cell lung cancer. Phase I results revealed that PK1 displayed greatly reduced toxicity
with maintained antitumor efficacy compared with free doxorubicin. The maximum tolerated dose
of PK1 is about four times higher than the usual clinical dose of free doxorubicin. PK2 has entered
Phase I clinical testing.
The micelle forming conjugates of adriamycin with poly(ethylene glycol)poly(aspartamide)
block copolymers already showed excellent in vivo antitumor activities. These micellar systems
are very attractive, since they can also be used to entrap a drug within their hydrophobic core as
well as providing the opportunity for covalent conjugation. In 2001 these micellar aggregates have
entered Phase I clinical trials.
Ref. E. Schacht, International Journal of Pharmaceutics 277 (2004) 119131.

4.3. Pulsatile Release System

Introduction

I.

JI

Research on site-specific and temporal control of drug delivery systems is receiving a major
impetus towards the development of new and/or improved drug therapies. Due to progress in
biotechnology and genetic engineering, numerous kinds of bioactive peptides and proteins have
been recently produced for use as novel therapeutic drugs. The development of delivery
formulations for these bioactive drugs has been recognized as one of the most important
cofundamentals for a successful therapeutic effect in clinical medicine because these bioactive
molecules are normally metabolized rapidly in the body. Furthermore, the administration of large
amounts of surroundbioactive compounds can cause some undesired side-effects [1,2], which
could be minimized by the use of a regulated release system.

Pulsed or pulsatile drug release is defined as thethe rapid and transient release of a certain
amount of drug molecules within a short time-period immediately after a predetermined
off-release period. Pulsatile release is commonly found in the body, for example during hormone
release, in which a baseline release is combined with pulsed, one-shot type release within a short
time range [3,4]. For this mode of delivery it is assumed that constant plasma drug levels are not
94

N
G
LM

-Z

JU

preferred and an optimal therapeutic effect comes from a periodically fluctuating drug
concentration. As will be discussed in the following section, insulin is one good example of a
hormone experiences pulsatile release in the body. Basal release of insulin stimulates the synthesis
of proteins and glycogen in muscle and adipose tissues. In ddition, pulsatile insulin release is
observed during hysand after the intake of foods to regulate blood lectropoglucose levels in the
body. Gastrointestinal hormones elong to another group of compounds with pulsatile elease
characteristics stimulated by the existence of ood in the gastrointestinal tract. In this case,
hormone release stimulates the release of digestive nzymes from the pancreas and stomach. The
menstrual cycle in adult women is regulated by the ulsatile release of several hormones, including
FSH, H, estrogen, and progesterone. Many biological unctions in the body are thus regulated by
the emporal and pulsatile release of hormones. Thus, it ay be necessary to administer most protein
drugs in a manner which more closely follows the circadian hythms in the body [1]. In this respect,
pulsatile elease system are required. A continuous dose of ormones generally induces down
regulation of ormone receptors on the target cellular membranes nd shows undesired effects in the
body. From this oint of view, a temporal release system is important o regulate hormone functions
in the body. As an xception to this rule, one example has been reported f the sustained release of a
hormone that induced a holly different therapeutic effect. Okada and co-rkers [5,6] prepared
biodegradable microspheres ade from poly(lactic / glycolic acid) (PLGA) and oly(lactic acid)
(PLA). The hydrophobic PLGA olecules formed a micelle-like structure surrounding the
leuprorelin acetate, in which the carboxyl end groups of PLGA molecules interacted with such
basic amino acid residues of leuprorelin acetate, as histidine and arginine residues. The drug
molecules were distributed within the biodegradable microspheres and these drugs underwent
sustained release controlled by the degradation of the polymer matrix. The sustained release of
this hormone resulted in a suppressed steroidogenesis for more than 1 month in vivo. When PLA
was used as the matrix sustained release was extended to 3 months. This formulation is useful in
the therapy of prostate cancer and endometriosis and is currently in clinical use in the USA.

JI

To prevent the down regulation of hormone receptors and to achieve efficient therapeutic
effects, the pulsatile drug release system is recognized as one of the most important technologies
necessary for an intelligent drug delivery that is able to regulate drug release in response to the
external chemical, phys ical, and biological stimuli. For example, electropo ration is used to
achieve a pulsatile delivery of drugs. Riviere et al. [7] used electroporation followed by
iontophoresis to deliver LHRH through the skin. They observed a more effective (nearly two-fold
increase), pulsed delivery of LHRH via temporal application of an electric pulse, compared to the
administration exclusively via iontophoresis, in which only a small amount of drug was introduced
to the body. The electroporation also induced the reduction of onset time. Thus, the construction
of a pulsatile drug release system is a feasible choice for the treatment of patients.

In this article, several approaches for the delivery of drugs in pulsatile manner, mainly using
polymeric carriers, will be reviewed. Our recent research using Ca2+-alginate gel beads for the
temporal delivery of macromolecular drugs is also presented here.
II. Time-Controlled Pulstile Release Systems

As is often seen in human beings, hormone release is regulated with a certain rhythm,
depending on the hormone type, in the order of several hours-, days, or months. Similarly, the
action of a certain drug should coincide with the proper site and time for optimal effect. Therefore,

95

the development of a time-controlled release system is desired for the treatment of patients.
Niwa et al. [8] prepared a novel capsule made from ethyl cellulose for the time-controlled
release of drugs in the colon. Initially, the ethyl cellulose capsule was prepared using a gelatin
capsule as a parent mold and then by coating the gelatin capsule with ethyl cellulose, followed by
the dissolution of the gelatin in water. The thickness of the ethyl Cellulose capsule body was
varied and the effect of wall thickness on the release of drugs in the capsules was investigated.

JU

Ethyl cellulose capsules contained a large number of mechanically made micropores (400 m) at
the bottom. Also located in the bottom of the capsule body was a swellable layer, consisting of
low-substituted hydroxypropylcellulose. Above the swellable layer was the drug reservoir, which
contained a mixture of the model drug, fluorescein, and a bulking agent, such as lactose or starch.
The capsule was thus capped and sealed with a concentrated ethyl cellulose solution. After
administration of drug-containing capsule, water molecules penetrated the capsule through the
micropores in the bottom of capsule body. Hydration and swelling of the hydroxypropylcellulose
induced an increase in the internal osmotic pressure, which resulted in theexplosion of the
capsule and a burst-like drug release was observed. By altering the capsule thickness, the lag time

N
G
LM

-Z

of the drug release could be controlled, ranging from 2 h for a 44.1-m thick capsule to 6 h for a
76.7-mm thick capsule. An in vivo examination was further carried out using beagle dogs, which
received capsules of varying thickness via oral administration. The drug concentration was
monitored from blood samples taken from the jugular vein at scheduled time intervals. In vivo
drug release revealed that the time-controlled release of drugs correlated with the capsule
thickness. Although further research is required, e.g. the number of pores and/or the quantification
of environmental water content in the GI tract could be optimised, this method holds promise for
pulsed colon-specific delivery of drugs.

JI

A similar approach for the pulsatile release of drugs was reported, in which a hydrostatic
pressure was generated inside the capsules. To control the lag times of this pulsed drug release
system, Jimoh et al.[9] utilized hollow biodegradable capsules with a thinner membrane at one end.
The poly(lactic acid) patients. (PLA) capsule was the effervescent agents, citric acid /sodium
bicarbonate were entrapped within. A schematic representation of the capsule formulation is
shown in Fig. 1. As water penetrated into the capsule through the thin poly(lactide-co-glycolide)
(PLGA) membrane side, it generated an effervescent reaction caused by the citric acid /sodium
bicarbonate mixture. Generated carbon dioxide gas accumu lated in the capsule and finally
ruptured the thin membrane, which was not resistant to the increased internal pressure. The burst
time, i.e. lag time of drug release, can be modulated by the dimensions of capsule formulation
(thickness of the membrane, hollow size, and the amount of effervescent agents). The capsule
body, made from PLA with a molecular weight of 50 000, was molded on a stainless die and
plunger at 700 psi and room temperature. The capsule has one open end covered with a thin PLGA
membrane of varying thickness. Core composition was found to be a key factor in regulating the
lag time (burst-like release). By altering the sodium carbonatecitric acid to glucose ratio, the
capsule burst time could be modulated from 2 h up to 168 h (7days) in vitro. There was a linear
relationship between the amounts of glucose in the core agents and in vitro burst time. Thus, the
lag (burst) time could be predicted from the glucose percentage in the core agents and the capsule
diameter obtained in in vitro experiments. These capsules were then utilized for the controlled
release of follicle stimulat ing hormone (FSH) in female rabbits. Several types of capsule
formulations were implanted subcutaneously in the back of the neck of each rabbit to prevent the
96

-Z

JU

animals from reaching the samples. Blood samples were taken from the ear veins or arteries and
FSH in the blood was monitored using the serum. Even though the membrane thickness varied, a
maximum of the serum FSH level was observed at the same time-point. This suggests that burst
time is unchanged regardless of membrane thicknesses. Thus, as long as the capsules are of the
same size and contain the same core ingredient formulations, the burst time is expected to be the
same. Furthermore, the capsule material is biodegradable and, therefore, there is no need for the
retrieval of empty capsules after complete release of drugs. This type of capsule may be useful for
the pulsatile release of hormones with multiple administration, immunization, and animal
reproduction management.

N
G
LM

Fig. 1. Scheme of the pulsed release biodegradable capsules.

Control of dissolution rate of the coating is another approach to achieve pulsed drug delivery.
Pozzi et al. [10] proposed the TIME CLOCK system for oral dosage, which should enable a fast
and complete release of drugs after a predetermined lag time. A tablet core was made containing
the drug molecule (samarium oxide for scintigraphic evaluation and salbutamol sulfate for the
phase I testing on healthy volunteers) and bulking agents (lactose, poly(vinylpyrrolidone),
cornstarch, and magnesium stearate). This core was coated with a hydrophobic dispersion of
carnauba wax, bees wax, poly(oxythe ethylene) sorbitan monooleate, and hydroxypropyl methyl
cellulose in water. By altering the coating thicknesses, the lag time could be proportionally
modulated. In vitro tests indicated that the drug core was dissolved immediately after direct
immersion in water and release was completed within 30 min, while a rapid release was observed

JI

after a certain lag time for the TIME CLOCK system with the hydrophobic coating. In vivo tests
revealed that drug disintegration was modulated by coating thickness of the drug core, as well as
the food intake before drug administration (i.e. viscosity in the intestinal tract). This approach may
also be used to control the release onset time. Since the drug core is formulated with soluble
ingredients, shell dissolution (or dis integration) becomes the key factor controlling lag time.
Furthermore, drug release is independent of normal physiological conditions, such as pH,
digestive state, and anatomical position at the time of release. This approach may be applicable,
not only for oral administration, but also for implantable drug formulations as well.

Noda and co-workers [1114] carried out a series experiments designed to develop sigmoidal
releas-ing systems for novel oral controlled release formu lations. The system consisted of two
components: the drug core, which is dissolved or able to swell in an aqueous medium, and the
shell layer, which has to control the lag time of drug release. Poly(ethylene glycol) (PEG) of a
molecular weight of 6000 and hydrogenated caster oil in various concentrations were used for
outer shell materials. The coat thickness, as well as the mixture composition, affected the outer

97

-Z

JU

shell strength and water penetration time. With increasing PEG content, the water penetration rate
increased, while a dramatic reduction in shell strength was observed. Thus, the lag time could be
controlled by the composition of the outer shell components and its thickness. The core
components consisted of the drug molecules and the swellable substances, which caused the
instantaneous rupture of the tablet, to achieve a pulsatile release of drugs. The authors used
carboxymethyl cellulose as a disintegration agent. The lag time was controlled by the total amount
of shell components, as well as the PEG content in the shell. With decreasing PEG Content in the
shell, the lag time could be extended. Fluid composition and pH did not affect to the lag time and
an identical lag time, as well as instantaneous drug release was observed within a short time
period in all fluid tested [12]. Ishino et al. [13] further investigated the applicability of the
proposed pulsatile release system in in vivo experiments using beagle dogs. In this case,
poly(vinyl chloride) was added to the shell component mixture of hydrogenated caster oil and
PEG. The tablet core contained a model drug and carboxymethyl cellulose as the swelling agent.
The formulated tablets showed a pulsed release of drug after a certain lag time (71 h) in vitro.
Tablets were administered orally to beagle dogs and the plasma drug levels were evaluated at
scheduled times. The plasma drug level increased after a certain lag time. A lag time can be seen
between 4 and 8 h for in vivo tests. The results suggested that the desired pulsatile release could
be achieved using the prepared tablets.

JI

N
G
LM

Gazzaniga et al. described on oral colonv specific delivery system which showed programmed
lag times. The drug core was coated with various amounts of low-viscosity hydroxypropylmethyl
celof lulose (HPMC), which acted as a retarding agent. In vitro release tests were performed using
the USP XXII paddle method. A burst release immediatelyafter the immersion of drug core was
observed and release was completed within 15 min. On the con trary, a burst release after a certain
lag time was achieved with HPMC-coated drugs. Lag times were prolonged with increasing
HPMC amounts, rising from 45 to 160 min with a weight gain from 35 to 115% HPMC,
respectively. The lag time was lengthened proportionally to the amount of coated HPMC. Thus, it
could be concluded that the lag time was regulated only by the change in the amount of HPMC.
However, higher amounts of HPMC coating resulted in a slower release after the lag time, which
continued for more than 1 h. Although the desired release pattern after the lag phase is dependent
upon the mechanism of drug action, it may also be necessary to modify the core formulation to
show either a burst release within a short time period or sustained release.
Liposomes have been used as drug delivery cariers because of their structural similarr to cell
membranes. However, uptake by the reticuloencontent adsorption of plasma proteins on the lipid
bilaye have been major drawbacks and have thus limited the application of these formulations. To
overcome the drawbacks of liposomes as a drug carrier system, Langer and co-workers
incorporated drugpulsatile loaded liposomes into microcapsules of alginate hydrogels. The
hydrogel matrix was designed to protect the liposomes from degradation and/or disgenated persion
in vivo, as well as possibly regulate the release rate of the incorporated drug molecules. To
achieve a pulsatile release of drug molecules the liposomes inside the microcapsules were coated
with phospholipase A (Fig. 2a). Phospholipase A2 was shown to gather at the water /liposome
interfaces and remove an acyl group from the phospholipid in the liposome, resulting in the
destabilization of the liposome. Destabilized liposomes release their drug molecules from the
interior, thus allowing drug release to be regulated by the rate determining microcapsule
membrane. Langer and co-workers [16,17] first incorporated FITC-labeled bovine serum albumin
98

(FITCBSA) as a model drug into hydrogenated lecithin or phosphatidylcholine / cholesterol

based multi-lamellar liposomes with an entrapment efficiency of 2050%. The liposome size

N
G
LM

-Z

JU

(diameter) was 2.82 m. Liposomes were then incubated with phospholipase A2 for 15 min at
20C to allow for adsorption onto the liposome surfaces. The liposome pellets were re-suspended
in a sodium alginate solution and the suspension was sprayed into a calcium chloride solution to
form gel beads. The calcium alginate gel beads containing drug-incorporated liposomes were then
stabilized by complex formation with poly(lysine). Fig. 2b shows the release profile of FITCBSA
from microcapsules. In the case of soy lecithin, BSA release from microcapsules was dependent
upon the phospholipase A2 concentration which coated the liposome surface. Without
phospholipase A there is only a 10% release of BSA for up to 63 days of observation. By
increasing the amount of phospholipase A2 from 1 to 10 Units, it could be observed that the
release process began earlier, on day 20 instead of 30 and was complete after 1015 days.
Phospholipase C and D cleave on either side of the phosphodiester linkage. When either enzyme
was used instead of phospholipase A2, a different release profile could be observed. A zero-order
release was obtained for liposome coated with phospholipase C, while only 10% release within 50
days was observed for phospholipase D [17]. Thus, release can be modulated with the type of
enzymes used to destabilize liposomes in the microcapsules. In vivo studies were performed, in
which horseradish peroxidase (HRP) release was measured by injecting the microcapsules
subcutaneously in male mice. Within 15 days, approximately 30% of the HRP was released by
diffusion, followed by pulsed release of remaining HRP within 5 days. Thereafter, HRP lost its

JI

enzymatic activity at 37C within a few days, while microencapsulated HRP retained 40% of its
biological activity even after 30 days of incubation at 37C. Thus, the microencapsulation of
enzymes into liposomes inside the alginate microcapsules should be effective not only for
maintaining biological activity but also for achieving pulsed release profiles. The molecular
weight of the poly(lysine) coating on the calcium alginate gels also affected the release behavior
of drug molecules. When lower molecular weight poly(lysine) was used the onset time of the drug
release was retarded. This is probably due to the easier penetration of poly(lysine) molecules with
a lower molecular weight inside the calcium alginate gel beads, which stabilized polyion
complex-formed capsule layers. Histological studies of the implanted sites indicated that there is
little inflammation after the injection of drug-incorporated microcapsules for up to 54 days
following the injection. The authors concluded that this is indicative of microcapsules
biocompatibility. Although the release profile seemed to be pulsed, drug release extended for
several days after the lag time. Even though the lag time can be controlled by the rate of liposome
destabilization using phospholipase A2 and the stability of microcapsule membranes by
poly(lysine) complex formation, release after the lag time should be more rapid (in the order of
several hours or even minutes) to achieve an actual pulsatile release profile. Thus, this system may
need some modification. It May be a promising approach for the delivery of drugs for
immunization purposes using a cocktail of several types of microcapsules with different contents,
as well as preparation methods.
Hellers group systematically investigated enzyme protective coatings for the use in triggered
drug delivery. The basic method of formulation is as follows: Drug molecules are incorporated
within the biodegradable matrix. The matrix is coated with enzymatically degradable materials.
The modified matrix is then placed within a microporous membrane capsule with the
antibody-protected enzymes. When the concentration of target molecule is increased, activation of
99

the enzyme could occur. The activated enzyme would then degrade the coating layers of the
biodegradable matrix, followed by drug release during matrix degradation. To realize this
mechanism, cellulose acetate phthalate was used as a degradable matrix and naltrexone as the drug
molecule. Drug-incorporated matrix microspheres were then coated with trilaurin, which is
digested by the enzyme, lipase.
Drug-containing microspheres of 250500 m in diameter were prepared by emulsification.
Drug release from the microspheres into 0.01 M phosphate-buffered saline (pH7.4) was very fast
and complete within 1 h. By coating the microspheres with trilaurin using a spray-coating method,
drug release was completely suppressed and less than 10% of incorporated drug was released after
28 days of examination. They then investigated enzyme-induced drug release. The microspheres
were placed in a dialysis membrane tube and suspended in the release medium at a temperature of

JI

N
G
LM

-Z

JU

37C. At time zero, varying concentrations of morphine-conjugated lipase were added to

suspension in the dialysis membrane tube. When control microspheres were suspended in the
release medium without lipase, drug release was suppressed up to 10% during the 24-h release
period In contrast, drug release from microspheres in the presence of lipase was significant. An
immediate effect was apparent after enzyme addition and approximately 30% of the naltrexone
was released with the first 30 min. Drug release was completed almost within 24 h. As the authors
described in this report, the strategy of this device design is not limited to the treatment of narcotic
addiction, but could be extended to other therapies, as well.

Fig. 2. (a) Schematic representation of liposome-loaded alginatepoly(L-lysine) microcapsules. (b) In vitro release
profiles of FITCBSA from microcapsules with different phospholipase A concentrations (s) 10 Units, (d) 1 Unit,
(h) no enzyme coating.

Yui et al. recently reported the development of a new drug delivery device, which exhibited a
pulsatile peptide release. The drug formulation consisted of a poly(ethylene glycol) (PEG)-grafted
dextran gel combined with an ungrafted dextran gel in a multi-layer structure located within an
impermeable silicone tube (3 mm I.D.) exhibiting an open end at both sides. Glycidyl
methacrylate was introenzymatically duced to the dextran hydroxyls and, thus, dextran was
cross-linked by UV light irradiation. The peptide drug, insulin, was thought to accumulate within
the PEG graft domains via two-phase partitioning becreased, beween the two aqueous dextran and
PEG systems. In act, when insulin was introduced to the PEG-grafted extran, only 10% of the
insulin was released during a 72-h incubation time, while a significant release was apparent for

100

insulin in a dextran solution alone. Thus, insulin in the PEG-grafted dextran was not released
spontaneously by simple immersion of the gels in the release medium.

-Z

JU

Insulin release was also examined in the presence of dextranase, an enzyme that specifically
digests dextran. Since the multi-layered gel formulations were located within the silicone tubing,
the degraphate-dation fronts were limited to both ends of the tube and a surface erosion type of
degradation could be observed. Insulin release from the multi-layered gels was examined in the
presence of 0.1 U/ml dexof tranase in releasing medium. Dextran degradation was proportional to
time regardless of the PEG chains and a pulsed-type insulin release was observed. When dextran
gels without PEG chains were degraded enzymatically, no insulin was released. This result
strongly suggests that the insulin moleconcentrations cules accumulated within the PEG-grafted
dextran gel layers. Insulin release was thus characterised by a degradation-controlled mechanism.
Conformational analysis of insulin using circular dichroism indicated that the insulin in
PEG-grafted dextran showed an almost identical conformation, even after 60 h of incubation. In
comparison, insulin in PEG-free dextran showed significant conformational changes. Thus, the
released insulin should have a higher bioactivity. Although this type of formulation could
guarantee the pulsatile release of peptide drugs, dextranase does not exist in the human body.
Therefore, it would be necessary to use another matrix material which can be degraded in the
human body before this system could be used for the treatment of patients.

N
G
LM

Yui et al. [21] further developed an aqueous, two-phase partitioning system using hyaluronic
acid (HA) and PEG graft chains. Similar to the PEG-grafted dextran, insulin molecules
preferentially accumulated within the PEG domains of the gel, rather than HA domains. Minimum
leakage of insulin was observed from a PEG-grafted HA solution prepared with 7 wt% PEG grafts,
while almost an 80% leakage of insulin was observed using a HA solution and PEG-grafted HA
with more than 22 wt% PEG graft chains after a 60-h incubation time. Negligible conformational
changes in the insulin structure were detected by circular dichroism. Since HA can be degraded
enzymatically in the human body, this formulation should be able to release insulin through the
matrix under in vivo conditions. Thus, the PEG-grafted HA system may be a feasible matrix for
injectable, degradation controlled drug release.

III. Stimuli-Induced Pulsatile Release

JI

A. Temperature-induced pulsatile release

Thermoresponsive hydrogel systems
Thermoresponsive hydrogels have been investigated as possible drug delivery carriers for
stimuli- responsive drug delivery systems. PIPAAm cross-linked gels have shown
thermoresponsive, discontinuous swelling / deswelling phases: swelling, for example, at
temperatures below 32C, while shrinking above this temperature. A sudden temperature increase
above the transition temperature of these gels resulted in the formation of a dense, shrunken layer
on the gel surface (skin layer), which hindered water permeation from inside the gel into the
environment. Drug release from the PIPAAm hydrogels at temperatures below 32C was
governed by diffusion, while above this temperature drug release was stopped completely, due to
the skin layer formation on the gel surface (onoff drug release regulation).
Swellingdeswelling kinetics of conventional cross-linked hydrogels are normally reciprocal of
the square of gel dimension [27]. The mobility of the cross-linked chains in the gel is affected by
the surrounding chains and the swelling/deswelling phases of the gel are governed by the
101

collective diffusion of the network chains. Thus, to accelerate structural changes of the gel in
response to external stimuli, several approaches have been developed which form porous structure
within the gel and decrease gel size.

-Z

JU

We have recently reported a new method to accelerate gel swelling / deswelling kinetics based
on the molecular design of the gel structure. Free mobile linear PIPAAm chains were grafted
within the cross-linked PIPAAm hydrogels as shown in Fig. 3. These novel graft-type PIPAAm
gels had the same transition temperature as the conventional cross-linked PIPAAm gels and
existed in the swollen state below the transition temperature, while above this temperature they
shrank. The molecular weight of the graft chains had a significant effect on the gel deswelling
kinetics and also on the drug release profiles, especially for large molecular weight drugs. A dense
skin layer was formed on the conventional PIPAAm gels upon temperature change above the
transition temperature, which limited the complete shrinkage of the gel. In contrast, the
PIPAAm-grafted gels showed rapid deswelling kinetics without the formation of a skin layer on
the gel surface. This is probably due to the rapid dehydra- tion of the graft chains formed by
hydrophobic aggregation on the three-dimensional cross-linked gel chains. The hydrophobic cores
might accelerate the entire dehydration and shrinking processes of the gel. Cylinder-shaped gels
showed fast and large changes in length when subjected to repetitive temperature cycles between

N
G
LM

20 and 40C [35]. Thus results indicated the feasibility of our concept that the introduction of
freely mobile, linear PIPAAm chains into PIPAAm cross-linked gels is effective for fast gel
deswelling kinetics.

JI

The drug release profiles from the graft-type PIPAAm gels were monitored after the immersion
of drug-containing gels in a suitable release medium (Fig. 4). Low molecular weight sodium
salicylate(Fig. 4a) was released from conventional PIPAAm gels immediately after a temperature
increase, after which the release was terminated due to the formation of a dense impermeable skin
layer on the surface. In comparison, 65% of the sodium salicylate was released in one burst from
free PIPAAm-grafted hydrogels with a graft molecular weight of 9000(IGG9000) following the
temperature increase Graft-type gels with a molecular weight of 4000 (IGG4000) showed
oscillating drug release profiles. The release of a higher molecular weight drug from PIPAAm
graft type-gels (IGG9000), for example dextran, with a molecular weight of 9300, was shown to
burst after a temperature increase to 40, while release was suppressed from IGG4000 (Fig. 4b).
The difference in drug release profiles for two graft-type gels is probably due to the different
strengths of aggregation forces between the formed hydrophobic cores within the graft-type gels.
That is, the larger molecular weight graft chains formed more hydrophobic cores within the gels
upon temperature increase, which induced rapid gel deswelling. In contrast, aggregation forces
between graft chains in IGG4000 were relatively weak, thus leading to the formation of skin layer
102

on the gel surface, which limited the drug diffusion from gel interior. Thus, by regulating the graft
chain lengths, the released amount as well as release pattern of various molecular weight drugs
could be modulated.

N
G
LM

-Z

JU

A similarly rapid deswelling phase was achieved by incorporating poly(ethylene glycol) (PEG)
graft chains into PIPAAm cross-linked hydrogels [36]. The introduction of PEG chains did not
alter the transition temperature, as was seen in the gels with hydrophilic comonomers, such as
acrylamide and acrylic acid. This is due to the structural independence of the PEG chains from the
cross-linked PIPAAm main chains. In this case, however, deswelling mechanism is different from
PIPAAm graft-type hydrogels. During the shrinking process, the graft PEG chains formed
hydrophilic channels for water molecules, most likely due to a phase separation within the
shrinking gels. Therefore, a rapid deswelling was achieved. The majority of the drugs in the gels
were released through the PEG formed channels with water molecules.

As mentioned above, accelerated deswelling of cross-linked hydrogels is achieved by the

introduction of graft chains independent from gel main chains. These gels could be activated by
external stimuli and, thus, utilized for drug release at targeted areas.
Thermoresponsive polymeric micelle systems

JI

As Kataoka et al. [37] comprehensively reviewed, the properties and biological interests of
polymeric micelles make them a most noteworthy candidate as drug carrier for the treatment of
cancer. The polymeric micelle is composed of amphiphilic block copolymers exhibiting a
hydrophobic core with a hydrophilic corona. Due to this unique structural characteristic, polymer
micelles exhibit stealth characteristics and are not detected by the body defense system
(reticuloendothelial system; RES). Thus, passive targeting could be achieved through a enhanced
permeation retention (EPR) effect [38] of the tumor sites.

We used an end-functionalized PIPAAm to prepare block copolymers with hydrophobic

polymers, such as poly(butyl methacrylate) (PBMA), polystyrene (PSt) [39,40] or poly(lactic acid)
(PLA) [41,42]. Block copolymers formed micellar structure (with coreshell structure) in aqueous
solution below PIPAAms transition temperature (Fig.5). The shell was made from
thermoresponsive PIPAAm, while the core consisted of hydrophobic polymer aggregates. The
PIPAAm corona showed an alteration in its hydration / dehydration properties with changing
temperature. The hydrated corona acted as an inert material toward all biological entities, such as
proteins and cells below the PIPAAms LCST. However, upon temperature increase above 32
hydrated PIPAAm chains became hydrophobic, due to the dehydration of polymer chains, thus
resulting in aggregation and precipitation.

103

JU

JI

N
G
LM

-Z

The core acted as a reservoir for hydrophobic drug molecules. The hydrophobic anticancer drug,
adriamycin, was incorporated into the either PBMA or PSt micelle cores via dialysis. The loaded
micelles were then evaluated with regard to their thermoresponsive drug release profiles [39,40].
The micelle formation was confirmed by dynamic light scattering measurements and LCST
changes. When PIPAAmPSt was used for micelle formation, the PSt core was relatively stable
against temperature changes. Therefore, drug release was suppressed at all temperatures examined
[40]. In contrast, PIPAAmPBMA micelles showed different thermo-sensitive drug release celles
showed different thermo-sensitive drug release profiles with temperature. At temperatures below
PIPAAms LCST, drug release was at a minimum, with a value less than 10%. However, upon
temperature increase above the PIPAAms LCST, accelerated release of adriamycin was observed
[39,40]. The application of a temperature gradient induced an onoff drug release regulation from
PIPAAmPBMA micelles between 4 and 37 (Fig. 6).

The thermoresponsive polymeric micelles were then added to endothelial cell culture media and
the effect of thermoresponsive drug release on the cell survival rate was investigated. Addition of
0.1g/ml free adriamycin induced a 10% cell death rate during the experimental period, while

104

almost all cells treated with adriamycin-containing polymeric micelles are viable at 29. This
indicated that the PIPAAm micelles displayed no cytotoxicity and no drug release from micelle
core was evident below PIPAAms LCST. By increasing culture temperature to 37
, how ever,
PIPAAm dehydration and precipitation was induced, which destabilized the PBMA core structure.
An enhanced interaction of hydrophobized micelles with the cells was observed, resulting in
adriamycin release and cell death. Thus, a higher cytotoxic effect could be elicited from the
thermoresponsive micelle system than from that of free adriamycin.

-Z

JU

When the hydrophobic core forming polymers are biodegradable, the micelle structure is
destabilized viacore segment degradation. A PIPAAmpoly(lactic acid) block copolymer was
prepared via ring opening polymerization of d,l-lactide, initiated by hydroxyl-terminal PIPAAm
and using stannous octoate as a catalyst [41]. Block copolymer formation was confirmed by the
difference in elution profiles using gel permeation chromatography. Polymer micelles were
prepared by dialysis of the polymer solution against water. Atomic force microscopic observation
revealed that colloidal spheres with average diameter of 40 nm were formed by this procedure.
Thermal transition phenomena were completely reversible, with micelles maintaining their
original size and shape throughout the temperature cycle below and above the LCST.

N
G
LM

In order to effectively utilize this system in the body, the hydrophilic /hydrophobic transition
tem perature of the outer shell polymer should be just above the body temperature. To achieve this,
poly(IPAAm-co-N,N-dimethylacrylamide (DMAAm)) with a hydroxyl terminal end and a
transition temperature at 44.6 was prepared in phosphate buffered saline [42]. The end
hydroxyl end groups were used to polymerize d,l-lactide to form block copolymers. The
copolymer micelles showed an LCST of 40
and a uniform diam eter of 69 nm . T he m icelle
s
transition temperature could be useful for active targeting in vivo, when combined with an
induced hyperthermia at 42.5.

JI

In vitro release of adriamycin was examined from polymer micelles at different temperatures.
At 37 , the physiological temperature, there was a minimum drug release of 3%. At 42.5
,
however, a gradual increase in drug release was observed. At ,
37 poly(IPA A m -co-DMAAm)
shell forming polymer segments existed in a hydrated state and expanded into the aqueous
environment. However, these chains dehydrated and were shrunken onto the hydrophobic core.
The aggregation of the thermoresponsive polymer segments induced perturbation of core forming
chain, resulting in drug release. When compared to the PIPAAmPBMA micelles, the rate of drug
release was significantly low. It is considered that structural distortion of the inner core took place
at 42.5 follow ed by a collapse of the tw o -phase structure or by a decrease in hydrophobicity
after mixing of the inner core and the outer shell above the LCST. As shown in Fig. 7, in vitro
cytotoxicity tests indicated that only 1 mg/ml adriamycin-loaded polymeric micelles showed a
higher cytotoxic effect in cultured endothelial cells only at 42.5
. M inim um cytotoxi city w as
evident for cells exposed to 10g/ml adriamycin-loaded micelles at 37. Thus, a cytotoxic effect
was only observed at 42.5, at w hich tem perature the therm oresponsive polym er shell becom es
dehydrated and exhibits hydrophobic characteristics.
Although there are still several parameters to be clarified, such as the targeting efficiency and
drug release behavior at specific sites in vivo, the above results clearly indicate that
thermoresponsive polymer micelles can be utilized for the thermoresponsive drug delivery to
tumor sites in conjunction with an induced hyperthermia.

105

Fig.7. In vitro cytotoxicity of adriamycin loaded micelles (circle) and free adriamycin (square for a, and triangle

B. Chemical stimuli-induced pulsatile rele

Glucose-responsive insulin release devices

-Z

mg/ ml, (b) 1 mg/ ml.

JU

for b) below (378C, open plot) and above (42.58C, closed plot) transition temperature. (a) Drug concentration: 10

N
G
LM

A decrease in or the absence of insulin secretion from pancreatic islets is the cause of diabetes
mellitus. Diabetes mellitus patients suffer long term from a gradual decline in the efficiency of
various organs, such as the occasional loss of eyesight. In very severe cases the condition may
lead to death. Thus, the injection of insulin at the proper time is necessary for the treatment of
diabetes patients. However, self-injection is painful and has sometimes led to the development of a
hypoglycemic coma, due to an overdose of insulin. In other cases, an insufficient amount of
injected insulin has led to hyperglycemia and an insufficient therapeutic effect. Thus, a great
demand has arisen for the precise, effective delivery of insulin to insure normal blood glucose
levels. In this section, a novel approach for the release of insulin is reviewed using
stimuli-responsive hydrogels which respond to glucose concentration changes.

JI

Several systems have already been developed which are able to respond to glucose
concentration changes. Glucose oxidase (GOD) catalyzes glucose oxidation. Utilizing this reaction,
Ishihara et al. [43] prepared two types of gel membrane systems to regulate insulin permeability.
They prepared GOD and nicotinamide-immobilized gel membranes, separately. When glucose
was oxidized by the immobilized GOD, the resulting hydrogen peroxide oxidized the
nicotinamide units inducing positive charges. The gel membranes became hydrophilic and
permeability increased. Horbett group [44,45] prepared GOD-immobilized hydrogels from
2-hydroxyethyl methacrylate and N,N-dimethylaminoethyl methacrylate. Glucose in the medium
was transformed to gluconic acid through a reaction mediated by hydrogel-immobilized GOD.
Thus, the protonation of hydrogel amino groups took place and hydrogel membrane swelling was
induced. Using this type of membrane, insulin permeability through the hydrogel was 2.45.5
times higher at 400 mg/dl glucose than that at 0 mg/ dl.

Kim and co-workers [46,47] have been working on the development of self-regulating insulin
delivery systems using microcapsules containing the glucose-binding lectin, concanavalin A (Con
A) and Con A-bound glucosylated insulin. Glucosylated insulin bound to Con A was released
through exchange with external glucose, due to the difference in their binding constants. Although
this system is a promising approach for the treatment of diabetes mellitus patients, direct injection
106

of microcapsules into the peritoneal cavity may cause undesirable side-effects arising from the
immune response toward Con A, if the microcapsules were to break and the Con A was to be
exposed to the immune system. Furthermore, it would be necessary to refill the glucosylated
insulin after exhausting initially bound insulin molecules. Thus, incorporation of these
microcapsules within biocompatible pouch membranes was considered.

-Z

JU

Obaidat and Park [48,49] prepared a copolymer of acrylamide and allyl glucose. The side chain
glucose units in the copolymer were bound to Con A. These hydrogels showed a
glucose-responsive, solgel phase transition dependent upon the external glucose concentration.
The nonlinear dependence of this solgel phase transition with regard to the glucose concentration
was not only due to the increased binding affinity of allyl glucose to Con A compared to native
glucose, but the cooperative interaction between glucose containing copolymer and Con A as well.
They further investigated the release kinetics of insulin using two chamber cells separated with
glucose responsive hydrogels [48]. The insulin concentration in the receptor chamber increased
with the glucose concentration, although the slow response was apparent. Thus optimization of the
system will be necessary to achieve a simultaneous response and release of insulin sensitive to
varying glucose concentrations.

N
G
LM

The above-mentioned systems used proteins of a natural origin, such as GOD and Con A. The
exposure of these proteins to the body may cause an undesirable immune response once applied.
Therefore, it may be necessary to separate these systems fore, it may be necessary to separate
these systems from the environment with semipermeable membranes, which are permeable to
glucose and insulin, but not to Con A, GOD, and the soluble polymers. Thus, a different approach
using a totally synthetic polymer system has been investigated. Okano, Kataoka and their
co-workers [5055] focused on the boronic acid moiety, which is known to form reversible bonds
with polyol compounds, including glucose. Water-soluble copolymers, containing phenylboronic
acid side chains, have formed reversible complex gels with polyol compounds, such as poly(vinyl
alcohol) (PVA) [50,51]. Such complexes dissociated after the addition of glucose in a
concentration dependent manner.

JI

Utilizing this polyol-binding characteristic of boronic acid, microgel beads (containing

phenylboronic acid) with a diameter of 100400 m were prepared [53]. Gluconic acid-modified
insulin was bound to the boronic acid moiety of the beads. The bound insulin was released after
substitution with external glucose molecules. For example, when a glucose-containing phosphate
buffer solution was passed through a mini-column packed with insulin-bound phenylboronic acid
containing microgels, insulin was released in a pulsatile fashion responding to glucose
concentration changes between 80 and 200 mg/dl (anything above this range was termed
hyperglycemic). After an extended release of insulin, each pulse (i.e. the amount of insulin
released in a pulse) was gradually decreased. However, a superior characteristic of this system is
the rechargeability of insulin onto the boronic acid moieties in the column packed beads. Thus, it
may possess a high potential for clinical application. The only disadvantage is that the system
works at pH 8.5 and, therefore, much higher than physiological conditions.

To overcome this disadvantage, Shiino et al. [55] incorporated amino groups in the vicinity of
the boronic acid in the gels. The lone electron pair of the amino nitrogen atom was coordinated
with the boron atom in such a position as to allow the formation of stable complexes with the
polyol at a pH lower than 8.5. Thus, glucose-sensitive insulin release was achieved at pH 7.4 using

107

amine-containing phenylboronic acid copolymer gel beads [55]. If the daily amount of insulin
required for diabetes patients was 1 mg, only 1.36 cm insulin-conjugated gel beads would be
necessary to maintain blood glucose levels. In comparison, more than 8000 cm of the
microcapsules containing ConA conjugated insulin, as reported by Kim et al. [47] in the
above-section, would be required to obtain similar therapeutic effect. This amount grossly exceeds
the actual application volume permitted to the human body. Thus the amine-containing
phenylboronic beads not only display the realistic dosage capacity required for clinical use, but
have also demonstrated the feasibility of recharging the system once insulin release has been
exhausted.

N
G
LM

-Z

JU

Kataoka et al. [56] recently developed glucose and thermo-responsive hydrogels using
acrylamidophenylboronic acid and N-iso-propylacrylamide (IPAAm). The obtained gels,
containing 10 mol% phenylboronic acid moieties, showed a transition temperature of 228C in the
absence of glucose. Below this temperature the gels existed in a swollen state. The introduction of
glucose to the medium altered the transition temperature of the gels in such a way that the
transition temperature increased with increasing glucose concentration to reach 36 at 5.0 g/ l
glucose concentration. Boronic acid was in equilibrium between the undissociated (uncharged)
and the dissociated (anionically charged) form. With increasing glucose concentration, the
equilibrium shifted to increase the amount of dissociated boronate groups and the gels became
more hydrophilic. Thus, the data indicated that the gel swelling / deswelling was regulated by
glucose concentration at a fixed temperature.

JI

Insulin was loaded into the gels and the time course of insulin release was evaluated with regard
to changing glucose concentrations in the medium at pH 9.0 and 28
. A t 0 1.0 g/ l glucose, the
insulin release over 24 h was suppressed to below 10 and 20%, respectively. However, a more
rapid insulin release was observed at a glucose concentration of 3.0 g/ l, in which 80% of insulin
was released within the first 10 h. At glucose concentration lower than 1.0 g/ l, gels existed in a
shrunken state. Above this concentration, the gels became hydrophilic, due to the increased
anionic boronate units within the gels. Electrostatic repulsion may be an additional factor which
influenced the increase in swelling at higher glucose concentrations in the medium. The result
clearly indicated that there is a threshold concentration of glucose necessary to induce loaded
insulin release from the hydrated, swollen gels. Using these glucose responsive hydrogels,
repeated insulin release was achieved varying the external glucose concentration at constant
temperature, as shown in Fig. 8. When glucose concentration decreased, a dense shrunken
polymer layer was formed on the gel surface, which stopped insulin diffusion from the gels.
Although the system needs to be optimized to achieve release at a suitable temperature, pH range
and response rate, it may be a promising approach for the treatment of diabetes mellitus using
totally synthetic copolymer systems.

108

Inflammation-induced pulsatile release

N
G
LM

-Z

JU

When human beings receive physical or chemical stress, such as injury, broken bones, etc.,
inflammation reactions take place at the injured sites. At the inflammatory sites,
inflammation-responsive phagocytic cells, such as macrophages and polymorphonuclear cells,
play a role in the healing process of the injury. During inflammation, hydroxylradicals (OH) are
produced from these inflammation-responsive cells. Yui and co-workers [57,58] focused on the
inflammatory-induced hydroxyl radicals and designed drug delivery systems, which responded to
the hydroxyl radicals and degraded in a limited manner. They used hyaluronic acid (HA), a linear
mucopolysaccharide composed of repeating disaccharide subunits of N-acetyl-D-glucosamine and
D-guluronic acid. In the body, HA is mainly degraded either by a specific enzyme, hyaluronidase,
or hydroxyl radicals. Degradation of HA via the hyaluronidase is very low in a normal state of
health. Degradation via hydroxyl radicals however, is usually dominant and rapid when HA is
injected at inflammatory sites. Thus, Yui and co-workers [57,58] prepared cross-linked HA with
ethyleneglycol diglycidylether or polyglycerol polyglycidylether. These HA gels degraded only
when the hydroxyl radicals were generated through the Fenton reaction between Fe2+ ions and
hydrogen peroxide in vitro. Thus, a surface erosion type of degradation was achieved. When
microspheres were incorporated in the HA hydrogels as a model drug, these microspheres were
released only when hydroxyl radicals induced HA gel degradation. The microsphere release was
regulated by the surface erosion type of degradation. Furthermore, in vivo tests of HA hydro-gel
degradation showed that HA gels were degraded only when inflammation at the implanted sited
was induced by surgical incision. Control HA gels implanted in the animals were relatively stable
over a period of 100 days. Thus, it is possible to treat patients with inflammatory diseases, such as
rheumatoid arthritis, using anti-inflammatory drug-incorporated HA gels as new implantable drug
delivery systems.
Drug release from intelligent gels responding to antibody concentration

JI

There are numerous kinds of bioactive compounds which exist in the body. Recently, novel gels
were developed which responded to the change in concentration of bioactive compounds to alter
their swelling / deswelling characteristics. Miyata and coworkers [59,60] focused on the
introduction of stimuli-responsive cross-linking structures into hydrogels. Special attention was
given to antigen-antibody complex formation as the cross-linking units in the gel, because specific
antigen recognition of an antibody can provide the basis for a new device fabrication. As Miyata
has reviewed the details of this system in Chapter 3 of this issue, this new approach will just be
introduced here. Utilizing the difference in association constants between polymerized antibodies
and naturally derived antibodies towards specific antigens, reversible gel swelling /deswelling and
drug permeation changes occurred. Thus, biological stimuli-responsive hydrogels were created.
C. Electric stimuli-responsive pulsatile release

The combination of developments in several technologies, such as microelectronics and

micromachining, as well as the potential need for chronotherapy, have currently assisted the
development of electronically assisted drug delivery technologies. These technologies include
iontophoresis, infusion pumps, and sonophoresis [61]. Several approaches have also been
presented in the literature describing the preparation of electric stimuli-responsive drug delivery
109

systems using hydrogels.

JU

Kishi et al. [62] developed an electric stimuli-induced drug release system using the
electrically-stimulated swelling / deswelling characteristics of polyelectrolyte hydrogels. They
utilized a chemomechanical system, which contained a drug model within the polyelectrolyte gel
structure. These gels exhibited reversible swelling / shrinking behavior in response to onoff
switching of an electric stimulus. Thus, drug molecules within the polyelectrolyte gels might be
squeezed out from the electric stimuli-induced gel contraction along with the solvent flow. To
realize this mechanism, poly(sodium acrylate) microparticulate gels containing pilocarpine as a
model drug were prepared. Pilocarpine release from microparticulate gel beads was investigated
by applying a direct current. During application of electric stimuli, pilocarpine release increased
with applied current-dependent manner. However, because the matrix itself showed higher
swelling in the medium, pilocarpine release was not terminated after removal of the electric
current. Thus, onoff release regulation of drugs could not be realized with this system.

N
G
LM

-Z

Kwon and co-workers [6365] used cross-linked poly(2-acrylamide-2-methylpropanesulfonic

acid-co-butyl methacrylate) (P(AMPS-co-BMA)) hydrogels for electric stimuli-induced drug
delivery systems. They incorporated positively charged edrophonium chloride as drug molecule
within negatively charged P(AMPS-co-BMA) hydrogels via an electrostatic interaction and
investigated the drug release from the hydrogels in an electric field. By applying an electric field,
ion-exchange between edrophonium ions and protons commenced at the cathode, resulting in a
rapid drug release from the hydrogels. This rapid drug release was attributed to the electrostatic
force, squeezing effect, and the electro-osmosis of the gel. By increasing the applied voltage, the
drug release rate was increased in a dose-dependent manner. Interestingly, complete onoff drug
release regulation was achieved with this system, as no drug release was apparent without the
application of an electric field. Hydration of the hydrogels did not affect drug release at all.
Contrary to the positively charged drug release from negatively charged hydrogels, the neutral
drug, hydrocortisone, was released from P(AMPS-co-BMA) hydrogels via a diffusion mechanism
[65]. The release rate of hydrocortisone showed a small pulse when an electric current was applied.
These pulses, however, may be attributed to the bulk squeezing caused by anisotropic deswelling
under the electric field. This hydrogel system may be a potential candidate for the delivery of
positively charged drugs in conjunction with iontophoresis.

JI

Kwon et al. [66] utilized a polyelectrolyte complex system to deliver bioactive compounds via
application of electric current. An insoluble polyelectrolyte complex was formed by the
combination of water-soluble cationic poly(allylamine) with anionic bioactive heparin. Upon the
application of an electric current, a rapid structural change of the polyelectrolyte complex
occurred, resulting in dissolution of component polyelectrolytes in a current dependent manner.
Dissociation and dissolution of the polymer complex was due to the neutralization of amino
groups along the poly(allylamine) backbone where the polyelectrolyte complex was attached to
cathode, as only local pH changes occurred near the electrodes. The release pattern of heparin
from the polyelectrolyte complex exhibited complete onoff profile in response to onoff stimuli
of electric current. However, a problem may arise from the simultaneous release of positively
charged polyamine when administered to living systems because the polycations may exhibit a
strong electrostatic interaction with many cell components, such as proteins. This dilemma should
be solved before in vivo application is attempted.

110

DEmanuele and Staniforth [67] investigated feed-back-controlled drug delivery based on the
electrophoresis of an ionic drug. In this system, drug release was regulated in either a positive or a
negative manner by appropriate control of the electric field between a pair of platinum electrodes
placed on either side of rate controlling polymer hydrogel membrane. Control of the electric field
was carried out by the microprocessor, which received signals from the biosensor element. In vitro
studies revealed that drug release through the cross-linked poly(2-hydroxyethyl methacrylate)
(PHEMA) membrane was positively or negatively controlled by the reversible polarity of the
applied electric field. However, before this artificial gland system may be ripe for in vivo
application, the development of a small and long acting (i.e. biologically inert) biosensor will play
a key role since no sensor tip has yet been produced which acts more than 12 weeks under
biological conditions.
D. Other physical /chemical stimuli for pulsatile release

JI

N
G
LM

-Z

JU

There are other stimuli-responsive pulsatile delivery approaches, for example, ultrasound and
magnetic fields. Ultrasound is mostly used as an enhancer for the improvement of drug
permeation through biological barriers, such as the skin, lungs, intestinal wall, and blood vessels.
There are several reports describing the effect of ultrasound on controlled drug delivery [6874].
Therefore, only a small portion of the research in the area of ultra-sound-induced pulsatile release
is described here. In many of these reports the matrix used for delivery was not composed of a
hydrogel, but from degradable polymers. Kost et al. [75] described an ultra-sound-enhanced
polymer degradation system. During polymer degradation, incorporated drug molecules were
released by repeated ultrasonic exposure. As degradation of biodegradable matrix was enhanced
by ultrasonic exposure, the rate of drug release also increased. Thus, pulsed drug delivery was
achieved by the onoff application of ultrasound. They further investigated the effect of clinically
used ultrasound on the degradation of polyanhydrides and reported the increased degradation of
matrix during ultrasonic irradiation [76]. Supersaxo et al. [77] also reported macromolecular drug
release from biodegradable poly(lactic acid) microspheres. Drug release from porous poly(lactic
acid) microspheres showed an initial burst followed by a sustained release for over several months.
When ultrasound was applied to this release system, pulsatile and reversible drug release was
observed. After the application of ultrasound, drug release was increased by nearly three-orders of
magnitude. Supersaxo et al. [77] speculated that ultrasonic exposure resulted in the enhancement
of water permeation within micropores of the polymer matrix, inducing drug dissolution into the
releasing medium.

Kwok and co-workers [78,79] prepared novel surface-modified hydrogels for

ultrasound-responsive pulsatile protein delivery devices. On the PEG-grafted PHEMA surface,
dodecyl isocyanate reacted to form self-assembling crystalline methylene chain layers, which
acted as a barrier membrane against peptide leakage from the hydrogel systems. Within the PEG
domains insulin was preferentially partitioned and became a reservoir for the protein [20,21]. With
the methylene chain coating around hydrogel surfaces, insulin leaching was not observed without
the influence of ultrasound. In sharp contrast, a pulsatile, transient insulin release was observed
after ultrasonic exposure for 1 min. During ultrasonic exposure, perturbation of the self-assembled
methylene chains occurred, leading to the release of insulin from the hydrogels. Bioavailability of
released insulin was also confirmed using cultured cells. After the development of new hydrogels,
pulsatile release of insulin could be extended for up to 4 or 5 days. Although the biological
response to the hydrogel surface is not clear, this system would be useful to deliver biologically
111

beneficial peptide drugs in a pulsatile manner.

JU

Another stimulus is the use of magnetic fields. Langers group [8082] investigated the
magnetically controlled drug delivery from polymer matrix incorporating a small magnet. In vitro
studies revealed that release rate of bovine serum albumin (BSA) increased 5- to 10-fold by
subjecting the matrix to an oscillating magnetic field [80]. Kost et al. [82] reported that the
exposure to an oscillating magnetic field induced insulin delivery to diabetic rats. Insulin was
incorporated into an ethylene-vinyl acetate copolymer membrane in which a small cylindrical
magnet was placed vertically. This device was then implanted subcutaneously into a diabetic rat.
Blood glucose levels decreased by 29.4% during exposure to an oscillatory magnetic field.
Histo-pathological analysis (after a 51-day implantation period) revealed that the polymer matrix
was surrounded by a relatively thin fibrous membrane, indicating the minimum tissue reaction.
Thus, magnetically controlled drug delivery systems may be effective for use as an artificial gland.
However, a complete system design is needed for clinical application.

IV. Pulsed Macromolecular Drug Release from Ca2+-Alginate Gel Beads

N
G
LM

-Z

Alginate is a naturally derived anionic polysaccharide, obtained mainly from algae, and is
widely utilized as a food additive. Alginate consists of two sugar moieties, 1,4-linked
D-mannuronic acid and L-guluronic acid, which exist either in blocks or random sequences
[8386]. Alginate is known to form complexes with divalent ions, such as Ca 2+, Ba2+, and Sr2+
[86], in aqueous solution. Depending on the composition of the two sugar residues and sequential
distribution within the molecules, the complexes form either precipitates or hydrogels. Guluronic
acid blocks are known to form a rigid buckled structure, the so-called egg-box array, in which
chelating calcium ions are nestled in the aqueous environment of the ordered gel structure, due to
the spatial arrangements of guluronic block oxygen atoms of carboxyl and hydroxyl groups. This
interaction is not only based upon electrostatic interactions, which neutralize acidic groups, but
also on the coordinating function of the calcium ions as a chelating center. Alginate forms stronger
gels with Ba2+ and Sr2+ [86], while no gelation is observed in the presence of Mg2+ and
monovalent cations (Na+ , K+ ions) [87].

JI

Since gelation occurs in an aqueous environment, alginate is a promising material as a food

additive, drug formulation , and useful even for encapsulation of living cells to protect them from
immune responses. Utilizing this stable complex formation with divalent cations, alginate gels
have been utilized for the investigation of cell encapsulation matrices, because living cells are
considered to be the ultimate system for the pulsatile release of biologically active compounds
[88,89]. Cell microencapsulated alginate beads were usually stabilized via polyion complex
formation using poly(L-lysine) or chitosan. For example, microencapsulated Islets of Langerhans
were transplanted into the peritoneal cavity of diabetic rodents and partially normalized blood
glucose levels could be observed, due to the glucose-responsive insulin secreted from the
encapsulated islets [89,90].
Among other drug delivery approaches, Kim et al. Among other drug delivery approaches, Kim
et al. [91] utilized calcium-alginate gels to release blue dextran of a molecular weight of 2 000000.
Dextran release was mainly governed by the pH of the dissolution solution. Retarded release was
observed at pH 1.2, while zero-order and a more rapid release was seen at pH 6.8. At a lower pH,
calcium ion-binding carboxylate anions were transformed to carboxyl groups and the gel itself
existed in a shrunken state, resulting in the retarded release of dextran. At pH 6.8, the alginate gels
112

began to swell caused by ion-exchanges between Na and Ca2+ ions, inducing the dissolution of the
alginate, as well as the release of dextran molecules. These beads could be utilized for the oral
delivery of peptides for possible vaccination purposes. Although a sustained release of high
molecular weight dextran from the calcium alginate gel beads was observed, the release behavior
was dependent upon the molecular weight and higher-order structure of the drug proteins. Thus,
optimization of not only the preparation condition of gels, but also of the release pattern in the
body is still required.

N
G
LM

-Z

JU

Formulation of delivery devices for protein and peptide drugs under aqueous conditions is
desirable to avoid the undesired decrease of bioactivities which may occur when using organic
solvents and/or heat during drug formulations. Since relatively stable alginate gels can be formed
in aqueous environments through chelation or complexation, alginate gel is one of the more
promising delivery matrixes for bioactive compounds. From this standpoint, we developed a novel
pulsed drug delivery system for macromolecular bioactive compounds using calcium alginate gel
beads [92,93]. The alginate used in our studies had a molecular weight of 100 000 and a ratio of
mannuronate to guluronate units of 1.360.3 in all cases. Fluorescein isothiocyanate (FITC)-labeled
dextrans of various molecular weights were used as model macromolecular drugs. An aqueous
solution of sodium alginate and FITCdextran was dropped into a 0.1-M CaCl2 aqueous solution
from a syringe connected with a needle tip and a syringe pump. The suspension was incubated in
the same solution for 2 days under light exclusion. Alginate beads with varying bead diameters
from 2 to 7.4 mm were used to examine release profiles of macromolecular model drugs. Release
experiments were performed following the JPXII protocol at 378C in Dulbeccos
phosphate-buffered saline solution (DPBS) at pH 7.4. Release of dextran was monitored by
fluorescence with excitation and emission wavelengths of 490 and 520 nm, respectively.

JI

Ca2+ -alginate gel beads dissociated (dissolved), due to the exchange of chelated calcium ions
with sodium ions in the DPBS medium [86]. Thus, alginate disintegration was monitored along
with the estimation of freed Ca2+ ions in the medium. Release of Ca2+ ions from calcium alginate
gel beads showed a two-phase release pattern. The first phase pattern of release followed a Fickian
diffusion, as the observed release was proportional to the square root of time. Alginate dissolution
was almost negligible at this stage. The second phase exhibited a burst-like release pattern, which
was accompanied by alginate disintegration. Polyguluronate units in the alginate molecule formed
a chelated structure with the divalent cations. This structure, the so-called egg-box junction,
contains interstices in which the cations may be coordinated [8385]. The egg-box junctions
between the alginate molecules are kinetically stable against dissociation [85], while the
polymannuronate units show normal polyelectrolyte characteristics of cation bonds. When
considering the ratio of mannuronate to guluronate (M/G=1.3) in the alginate molecules, the
amount of calcium ions in the first phase corresponded to the amount of mannuronate units in the
gels. This calculation was performed under the assumption that all guluronate units in the alginate
molecules participated in the egg-box junction formation. The calculated amount of calcium ions
in the second phase agreed with the amount of guluronate. Therefore, the relatively stable
association of polyguluronate sequences with Ca2+ ions served as stable cross-linking points
within the alginate gels. It may be also possible that calcium ions released in the first stage formed
new bonds to polyguluroante sequences. Release of calcium ions in the buckled egg-box junctions
in the gels induced the electrostatic repulsion of anionic alginate polymer chains, resulting in
swelling, and eventually facilitating gel erosion [94,95]. Based upon the above described results,
113

the release of macromolecules from calcium alginate gels could be controlled by the alginate gel
dissolution and exhibit a pulsatile pattern.

JI

N
G
LM

-Z

JU

Dextran release from calcium alginate gel beads was then evaluated [92,93]. The molecular
weight varied from 9400 to 14 5000. Dextran release from calcium alginate gel beads was
monitored by fluorescence of the solution taken at predetermined time intervals. Fig. 9 shows the
release of dextran with different molecular weights from calcium alginate gel beads. Dextran with
a molecular weight of 9400 showed a Fickian diffusion type of release in the initial 60% of the
release profile. With increasing molecular weight, initial dextran release was significantly
suppressed and eventually became negligible, as in the case of dextran with a molecular weight of
145000. All dextran incorporated into calcium alginate beads showed a burst-like drug release
after 75100 min. During this time period, calcium alginate gels showed disintegration and
complete dissolution. Therefore, the sudden increase in dextran release corresponded to the
destruction of the calcium alginate gel beads. When we compared dextran release of the low
molecular weight drug, Brilliant blue, from calcium alginate gel beads, the diffusion coefficients
for dextran were much lower than that determined for Brilliant blue, reaching values as low as one
tenth to one thousandth the order of dextran with regard to molecular weight. Thus, the release
mechanism of macromolecular dextran is totally different from that of lower molecular weight
drugs, in which case the release is mainly governed by Fickian diffusion through the cross-linked
networks. Burst-like dextran release from calcium alginate gel beads after a certain lag time is
dependent on the distribution and ratio of mannuronate and guluronate units along the alginate
molecule chains.

We then investigated the factors influencing the onset release time of macromolecular drugs
from calcium alginate gel beads. Two factors were focused upon: one was the alginate
concentration in preparation and the other was the calcium alginate gel bead diameter. The sodium
alginate solution concentration was varied from 1 to 4% during gel beads preparation.
Concentrations above 4% were too viscous to be extruded from syringe needle. The molecular
weight of incorporated dextran was 145000 in all cases. In this concentration range, all gels
showed a burst-like release of dextran with similar profiles. Onset release time varied with
increasing solution concentration during gel preparation, increasing from 30 min for a 1% solution
to 210 min for a 4% solution, respectively. The number of the apparent cross-linking points within
the formed calcium alginate gel beads increased with increasing alginate concentration in
preparation. This increase in the apparent cross-linking density resulted in the retardation of
114

alginate gel disintegration in PBS. Increased alginate gel density per unit volume was also thought
to affect the decreased mesh size within the gels and slow dextran release. Calcium alginate gel
bead diameter was varied from 2 to 7.4 mm without changing the solution concentration in
preparation (2% alginate solution). The onset release of dextran (145000 molecular weight)
increased with increasing gel bead diameter, exhibiting values of 1 h for 2-mm gel beads to 8 h for
beads with a 7.4-mm diameter [93]. Furthermore, the dextran release from a mixture of calcium
alginate gel beads of three different diameters (2, 3.8, and 5.5 mm) showed three pulses in the
time range of 60240 min, as shown in Fig. 10 [93]. These release onset times were in accordance
with the release times from the corresponding alginate gel beads in a separate experiment.

JI

N
G
LM

-Z

JU

Recently, we developed modified calcium alginate gel beads with smaller diameters to extend
onset release time to one or more days [96]. To this purpose, calcium alginate gel beads with 1
mm diameter were coated with a calcium-binding, water-soluble copolymer at a maximum coat
thickness of 125 mm. The copolymer used was poly(N,N-di-methylacrylamide-co-carboxy-npropylacrylamide) (poly(DMAAm-co-CNPAAm)) with a CNPAAm content of 22 mol%. The
copolymer contains carboxyl groups in the side chains, which acted as binding sites for calcium
ions. The calcium alginate gel beads with 1 mm diameter were immersed in an aqueous solution
of copolymer followed by submergence into a calcium chloride solution to chelate and reinforce
the coating layers. The procedure was repeated five times to form stable coating layers and the
resulting thickness of coating layers were varied from 25 to 125 m, depending on the
copolymer solution concentration. The release profile of macromolecular dextran was then
examined from the modified calcium alginate gel beads. Dextran release onset time for
unmodified calcium alginate gels was 2 h. With increasing coating thickness (from 25 to 125
m), the onset release times increased from 4 to 60 h (2.5 days). In all cases, initial release of
dextran was suppressed to levels below 10%, which escaped via diffusion, whereas the remaining
90% of the drug molecules were released within a short time period. Calcium ions were released
in a two-phase pattern, as was observed for unmodified calcium alginate gel beads, while the
initial diffusion rate decreased with increasing coating thickness. Alginate release also displayed a
burst-like pattern, as was seen for the release of dextran. We observed the broken shell remains
after release experiments. Thus, two factors were considered to regulate dextran release from the
modified alginate gel beads. One is the diffusion of sodium ions from the medium into the gel
interior where they are exchanged with calcium ions. In fact, when we monitored calcium ion
release from the modified alginate gels, initial release governed by Fickian diffusion decreased
and a prolonged release time was observed with increasing copolymer coat thickness. Retarded
ion-exchange resulted in the slow disintegration of alginate gels, which extended the onset time of
drug release. The other factor was the osmotic pressure, which accumulated inside the coated
shells. As ion-exchange proceeded, disintergration of alginate gels occurred. Alginate
disintegration induced hydration surrounding the carboxylate anions, supported by the electrical
repulsion of the negative charges. These phenomena induced the accumulation of an internal
osmotic pressure within the coated shells. When the osmotic pressure inside the gels overcame the
strength of the shells around the modified gels, ruption of shell membrane occurred and the
contents within the membrane diffused through the cracks of the shell. This process occurred
relatively fast and a burst-like release of the contents was observed. A schematic depiction of this
process is shown in Fig. 11.
By mixing modified alginate gel beads with a different coat thicknesses a pulsatile drug releases
115

could be demonstrated. The pulsed release times corresponded to the onset release times observed
for each modified bead size.

-Z

JU

These results suggested that the modification of calcium alginate gel beads with a thin layer of
water soluble copolymer containing carboxyl groups was able to extend the onset release time of
macromolecular drugs incorporated in the calcium alginate gel core. This approach has the
potential to extend the onset time of drug release to several hours or days and the administration of
a mixture of modified gel beads in one batch may achieve a successful therapeutic effect without
perturbing the patients.

JI

N
G
LM

Ref. Teruo Okano, Advanced Drug Delivery Reviews 54 (2002) 5377.

116

Chapter 5
Biomaterials: Where We Have Been and Where We Are
Going

5.1. Introduction
5.2. Traditional Biomaterials

5.4. The Future of Biomaterials

-Z

5.1. INTRODUCTION

JU

5.3. A New Generation of Biomaterials

N
G
LM

A commonly used definition of biomaterial is a nonviable material used in a medical device,

intended to interact with biological systems (1). If the words used in a medical device are
removed, this definition becomes more inclusive to the wide range of applications where we
interface synthetic materials and modified natural materials with biology. Table 1 lists some of
these applications, both medical and nonmedical. Biomaterials and medical devices have evolved
over the last 50 or so years to a $100 billion endeavor. The subject of biomaterials resides at a
multidimensional interface between chemistry, chemical engineering, materials science,
mechanics, surface science, bioengineering, biology, and medicine, with considerable input from
ethicists, government-regulated standards organizations, and entrepreneurs. The field has seen
consistent growth since its inception and a steady introduction of new ideas and productive
branches. Some examples of these outgrowths from biomaterials that have evolved into
identifiable fields of their own include controlled release, diagnostic arrays, and tissue
engineering.

JI

The numerous facets of the field of biomaterials make a comprehensive review article
impossible. However, in general terms, we can review where we have been, the state of the art
today, and where we might be in 10 or 20 years.

5.2.TRADITIONAL BIOMATERIALS

Although medical implant materials have been used for at least 2000 years (some historians trace
sutures back 32,000 years), most early medical implants were doomed to failure because
important concepts relating to infection, materials, and the biological reaction to materials were
not yet established. The modern era of medical implants might be traced back to an observation
made by British ophthalmologist Harold Ridley in the late 1940s. While examining Spitfire fighter
pilots who had shards of canopy plastic unintentionally implanted in their eyes from enemy
machine gun fire, he noted that these shards seemed to heal without ongoing reaction. He
concluded that the canopy plastic, poly(methyl methacrylate), might be appropriate for fashioning
implant lenses for replacing cataractous natural lenses. His first implantation of such a lens was in
1949. His observation and innovation led to the development of modern intraocular lenses (IOLs)
117

N
G
LM

-Z

JU

that are now being implanted in over 10 million human eyes each year and have revolutionized
treatment for those with cataracts. At about the same time that Harold Ridley was innovating IOLs,
Charnley was developing the hip implant, Vorhees invented the vascular graft, Kolff was
revolutionizing kidney dialysis, and Hufnagel invented the ball

JI

and cage heart valve. These pioneers, in an era before principles for medical materials were
established, proved feasibility, saved lives, and evolved the foundations that we build on today. By
the late 1960s engineers, chemists, and biologists, in collaboration with physicians, were
formalizing design principles and synthetic strategies for biomaterials. In particular, the idea that
the release of toxic leachables from biomaterials will adversely affect healing was
formalizedthis toxicology idea is implicit in todays definition of biocompatibility. As
developments took place in biology and materials science, biomaterials researchers were quick to
incorporate these newideas into biomaterials. Molecular biology concepts are now opening new
frontiers for biomaterials designa path to biomaterials that will work with the normal
physiology and seamlessly integrate into the human body.

The appreciation of the host response (healing) of implanted materials helps us understand the
performance of todays biomaterials and also aids us in envisioning how we might improve the
bodys response to biomaterials in the future. The following section reviews our understanding of
the reaction of a living organism to an implanted synthetic material.

Host Response to Biomaterials

When a tissue is injured, the normal healing response is initiated through a series of complex
events that include acute inflammation, the formation of granulation tissue, and eventual scar
formation. The immediate response is to flood the injured area with blood. Fibrinogen within the
blood is cleaved into fibrin to form a blood clot that promotes platelet adhesion and aggregation.
Cytokines and growth factors are released to recruit white blood cells, mainly neutrophils.
118

Monocytes are then called to the wound site where they differentiate into macrophages. The
macrophages are responsible for cleaning up the wound site, which may contain foreign material,
bacteria, and dead cells, and also for recruiting cells such as fibroblasts and endothelial cells,
which convert the fibrin clot into a highly vascularized granulation tissue. The formation of blood
vessels is essential to the healing wound. The granulation tissue is subsequently replaced by an
extracellular matrix (ECM) deposited primarily by fibroblasts. The degree of ECM remodeling
depends on the extent and location of the injury. In some cases, complete restoration of the tissue
architecture is possible; however, in most cases the granulation tissue is remodeled into scar
tissue.

N
G
LM

-Z

JU

A biomaterial implanted into the body, however, induces a different response, termed the
foreign body reaction. This response is illustrated in Figure 1. The biological response to materials
has been reviewed in detail by Anderson (2). In brief, a biomaterial elicits nonspecific protein
adsorption immediately upon implantation. Many different proteins adsorb to the surface in a
range of conformations from native to denatured. Non-specific protein adsorption, however, never
occurs in the normal physiological process of wound healing. Thus, nonspecific protein adsorption
may be an instigator in the foreign body reaction. A number of different cells, such as monocytes,
leukocytes, and platelets (cells that are key players in normal wound healing), adhere to these
biomaterial surfaces and as a result may lead to upregulation of cytokines and subsequent
proinflammatory processes. In addition, the implant is significantly larger than the adhered
macrophages, preventing them from phagocytosing the foreign body. Chronic inflammation at the
biomaterial interface ensues, and the frustrated macrophages fuse together to form multinucleated
foreign body giant cells that often persist for the lifetime of the implant (3). The end stage of the
foreign body reaction involves the walling off of the device by an avascular, collagenous fibrous
tissue that is typically 50200 m thick.

The Performance of Contemporary Biomaterial Implants

JI

Although many medical devices are implanted in humans each year that save lives and improve
the quality of life, there is substantial room for improvement of medical devices. In particular, the
fibrous capsule that forms around implants can lead to complications. Capsular contracture of
breast implants, insulating barriers around electrodes, scarring of heart valve sewing rings, and
tough, fibrous layers surrounding vascular grafts are examples of this poor healing. Furthermore,
posterior capsular opacification seen in IOLs, the poor luminal healing of porous vascular grafts,
and inflammation seen with some contact lenses may be manifestations of the interfacial
encapsulation reaction observed for implanted materials.
In general, todays biomaterials do not control the biological response induced by them. The
biological response is stochastic and unplanned. Clearly, an engineering approach that controls the
interfacial response with precision is a strategy worth considering. The next generation of
biomaterials may realize this vision.

5.3. A NEWGENERATION OF BIOMATERIALS

The next generation of biomaterials will be based upon knowledge of the biology of wound
healing and inflammation and will control bioreaction responses with precision. One approach to
achieve such healing biomaterials is to modify surfaces of existing devices to minimize the foreign
body reaction and to promote normal wound healing. This approach is elaborated on shortly. For
longer-term solutions, however, researchers must turn to biology to develop ways to help the body
119

N
G
LM

-Z

JU

mend itself and regenerate new tissue that can last for the patients lifetime. The only way to
replace function to damaged tissue is with similar tissue that contains the appropriate tissue
architecture of cells and ECM.Anumber of different approaches are being investigated including
drug and cell delivery. In both cases, a carrier is required to deliver the drug to a particular site
successfully and, in the case of cells, to provide a 3-D structure that facilitates tissue growth.
Many different types of biomaterials have been investigated for both drug and cell carriers. In
tissue engineering, a combination of drugs, cells, and a suitable carrier with tailored macroscopic
properties, a well-tuned degradation profile, and specific biological cues may be necessary to
promote normal tissue growth.

Fig. 1. The foreign body reaction as illustrated here is the normal reaction by higher organisms to an implanted
synthetic material.

JI

This section highlights the next generation of biomaterials, which include surface modification
of biomaterials to overcome nonspecific protein adsorption in vivo, precision immobilization of
signaling groups on surfaces, the development of synthetic materials with controlled and tailored
properties for drug and cell carriers, biologically inspired materials that mimic natural processes,
and finally the design of sophisticated 3-D architectures to produce well-defined patterns for
developing bioMEM devices and tissue-engineering scaffolds. Because of the fast growth of this
field, this section aims to highlight several exciting new research areas. We present some of the
most recent developments in the different subfields of biomaterials and provide references to
recent reviews for more in-depth discussions. This section aims to demonstrate the powerful
impact that biomaterials will have on the future of medicine.

Surface-Modified Biomaterials
The surface chemistry and topography of a biomaterial are important parameters that influence
protein adsorption, cell interaction, and the host response. For example, surface chemistry has
been shown to alter monocyte adhesion in vitro (4). However, in vivo the foreign body reaction
seems independent of simple surface chemistry. For instance, materials that are polymeric,
ceramic, or metallic based and that exhibit different surface properties, from hydrophilic to
120

hydrophobic or hard to soft, all elicit the same response in vivo (57). This observation has been
attributed to nonspecific protein adsorption. In contrast, all of natures processes employ specific
molecular (i.e., protein and saccharide) interactions. Where organisms seem insensitive to
relatively simple surface chemistries, the complex surfaces associated with biological recognition
are effective in controlling biological responses. Thus, to address this issue, efforts have focused
on modifying surfaces to control protein interaction and to decorate the surface appropriately with
signaling molecules. Specifically, a considerable amount of research has explored surfaces that are
resistant to protein adsorption, termed nonfouling surfaces, and more recently surfaces that control
protein adsorption.

-Z

JU

NONFOULING SURFACES Poly(ethylene glycol) (PEG), also known as poly (ethylene oxide)
(PEO), has been most widely studied for its protein and cell resistance. There have been many
approaches to attaching PEG to materials and rendering them nonfouling, such as through
covalent immobilization, adsorption, or interpenetration. Arecent novel approach takes advantage
of the highly adhesive nature of mussel adhesive protein (8). Specifically, PEG was covalently
attached to the mussel adhesive protein to form a copolymer that exhibited a nonfouling segment
and a sticky segment. The sticky segment adhered well to gold and titanium surfaces presenting
the PEG chains at the new interface. These modified surfaces were found to resist cell adhesion
for up to 2 weeks in vitro (8).

N
G
LM

Although PEG is effective in preventing nonspecific protein adsorption, its nonfouling character
is dependent on surface chain density and can be damaged by oxidants. As an alternative, Lopez et
al. (9) have investigated PEG-like surfaces prepared by radio frequency plasma deposition of
tetraethyleneglycol dimethylether (tetraglyme), which forms a highly nonfouling crosslinked
structure. In plasmadeposited tetraglyme, there are no chains longer than 3 tetraethyleneglycol
units, demonstrating that the conformation of PEG is not the controlling variable in its nonfouling
ability. These surfaces exhibit little protein adsorption with levels as low as 2 ng/cm2 (9). These
surfaces have also been shown to significantly reduce blood platelet (10) and monocyte (11, 12)
adhesion in vitro.

JI

A recent study investigated the in vivo response of untreated fluorinated ethylene propylene
copolymer (FEP) and FEP treated with plasma-polymerized tetraglyme when implanted
subcutaneously in the backs of mice (13). After 4 weeks, a fibrous capsule had formed around all
the implants. The capsule thickness was similar regardless of treatment. Interestingly, macrophage
adhesion was significantly higher in the tetraglyme-treated samples compared with the untreated
samples. This study further confirms the differences between in vitro and in vivo responses. It was
hypothesized that the tetraglyme surface binds very little protein in vivo, as seen in vitro, but that
the conformation, not quantity, of the adsorbed protein may promote the foreign body response.
For example, tetraglyme may prevent a protein from fully spreading, and therefore more active
sites (i.e., cell adhesion sites) become available. Alternatively, degradation of the tetraglyme
surface, and subsequent loss of the nonfouling character, may occur under in vivo conditions.

There are many nonfouling surfaces in addition to PEG. The most commonly studied are
phospholipid surfaces (14) and saccharide surfaces (15). A number of novel classes of nonfouling
surfaces have recently been reported (16). Theoretical and experimental studies on the nature of
the nonfouling phenomenon are expanding our understanding of the important parameters driving
protein adsorption (17, 18). In general, a number of different chemistries minimize protein

121

adsorption and, consequently, cell adhesion in vitro. However, the in vivo response is quite
different. To develop a long-term nonfouling surface for implanted devices will require better ex
vivo model systems that can capture the complex processes of the in vivo environment.
SURFACE SPECIFICITY Cell adhesion to adsorbed proteins is mediated through integrin and
other receptors located within the cell membrane. Upon adhering to the protein film, a cascade of
intracellular signaling events is triggered. Therefore, controlling protein adsorption on the surface
of biomaterials may be critical to controlling and directing cell responses to biomaterials. Several
approaches have been examined, from incorporating short oligopeptides that exhibit specific
binding domains to incorporating whole proteins on the surface of the biomaterial.

N
G
LM

-Z

JU

Several short oligopeptide sequences thatmake up the receptor-binding domains within large
proteins have been elucidated. A number of different short amino acid sequences have been
identified that mediate cell-specific adhesion and function, and these have been recently reviewed
(19). The most widely investigated celladhesive oligopeptide is arginine-glycine-aspartic acid (or
RGD), which is found in a number of different extracellular matrix proteins, such as fibronectin
(20), laminin (21), collagen (22), and vitronectin (23). Short oligopeptides are relatively
inexpensive, are easy to synthesize, and offer greater flexibility for surface modification compared
with the bulky and labile intact proteins. As a result, RGD has been applied to a number of
different surfaces to direct cell adhesion and render them bioactive. When a surface was modified
with a combination of nonfouling PEG (99%) andRGD(1%), protein adsorptionwas minimal (2
ng/cm2), but a sufficient number ofRGDsiteswas present to promote fibroblast cell adhesion (24).
Furthermore, the structure and the conformation of these oligopeptides are important in
modulating cell adhesion. For example, immobilization of a cyclic RGD peptide resulted in
increased human bone marrow stromal cell adhesion compared with linear RGD peptides (25).

Recently, an oligopeptide containing the integrin 21 binding domain found in collagen type I
was synthesized in such away that it adopted a triplehelical conformation similar to the native
collagen structure that is necessary for integrin binding (26). When the mimetic peptide was
immobilized on the surface, cell adhesion was similar compared with collagen type I, but
interestingly when the peptide was passively adsorbed on the surface, where a range of
orientations can exist, including disruption of the tertiary structure, cell adhesion decreased.

JI

In many cases, incorporation of the entire protein is desired to promote multiple

integrin-mediated responses (where often the exact roles are not yet understood). However,
nonspecific binding of a protein to a surface may render it nonfunctional owing to changes in
orientation or conformation, i.e., the active sites may be hidden or the protein may become
denatured. Surface chemistry influences the conformation of adsorbed fibronectin (27).
Self-assembled monolayers of alkanethiols with different terminal endgroups (e.g., CH3, OH,
COOH, and NH2) on gold provide a well-controlled method to explore such surface chemistry
effects. It was found that fibronectin adsorbed onto gold surfaces exhibiting the OH headgroup
chemistry increased cell adhesion compared with fibronectin adsorbed onto to COOH and NH2
and even more so to CH3. Integrin binding to fibronectin by cells using these receptors also
depended on the surface chemistry, but the dependence varied for each integrin studied,
suggesting that the orientation of the fibronectin is altered by surface chemistry (27). Osteopontin
is one protein of particular interest because of its potential role in mediating wound healing (28).
We have successfully immobilized osteopontin on poly (2-hydroxyethyl methacrylate) (pHEMA)
surfaces to promote endothelial cell adhesion (29) (Figure 2a). However, when collagen type I
122

JU

was first immobilized on the pHEMA surface, followed by the addition of osteopontin that was
allowed to affinity bind to the immobilized collagen, endothelial cell adhesion increased
significantly compared with cell adhesion to immobilized osteopontin, as shown in Figure 2b (29).
By allowing osteopontin to adopt a more native conformation by binding to collagen type I,
protein function may be preserved.

Fig. 2. Bovine aortic endothelial cells cultured for 1.5 h on poly(2-hydroxyethyl methacrylate) surfaces (a)

-Z

immobilized with osteopontin or (b) immobilized with collagen then incubated with osteopontin to allow the
natural binding of osteopontin to collagen.

N
G
LM

SURFACE MICROARCHITECTURE The microarchitecture of the implant, particularly at

the surface, is an important parameter that influences the host response. Nonporous implants result
in densely packed, well-organized fibrous capsules, whereas certain porous implants lead to a less
dense, more open and disorganized fibrous capsule (30, 31). The latter loose collagenous capsule
exhibits increased vascularization leading to enhanced diffusion of small molecules across the

JI

fibrous capsule to the implant (32). Porous membranes, with pores between 5 and 15 m,
resulted in significantly increased vascularization adjacent to the implant and within the fibrous
capsule compared with smooth surfaces, regardless of chemistry (31). In an effort to develop
porous materials that can be used to enhance the vascularity of implants,pHEMAhydrogels with
well-defined and controlled pore sizes have been synthesized by a sphere-template technique that
allows monodisperse and uniformly shaped pores (33). An example of these porous pHEMA
hydrogels is shown in Figure 3. By tightly controlling pore size and shape, we may better
understand this apparent influence of implant microarchitecture on the host response. Preliminary
results from our group show a strong vascularization effect of these porous materials (34).

Fig. 3. A scanning electron microscopy image of a porous poly(2-hydroxyethyl methacrylate) hydrogel prepared

123

by a sphere templating method with well-defined monodisperse pore sizes. Bar = 100 m.

The diameter of polymer fibers influences capsule thickness (3537). Fibers with diameters
greater than 5 m induced the classic foreign body response, whereas fiber diameters between 1
and 5 m exhibited little or no capsule (35). Fiber chemistry, not surface charge, appeared to
have an effect on fibrous capsule thickness, although not as dramatic as fiber diameter (36, 37).
Furthermore, blood vessel density surrounding the fibers was unaffected by fiber density for fiber
diameters between 1 and 15.9 m (36).

The Synthesis of New Biomaterials

-Z

JU

Because of the growing interest and need for new biomaterials, a great deal of research has been
devoted to synthesizing new materials from a range of natural and synthetic chemistries and 3-D
macrostructures to offer materials with a wide range of properties and/or controlled responses that
can be used for many different applications. The list of new materials developed specifically for
biological applications is far too long to describe here.We have chosen to focus on two types of
synthetic biomaterials: (a) hydrogels, which are attractive owing to their high water content and
tissue-like properties, and (b) smart materials, which can rapidly respond to changes in the in vivo
environment.

N
G
LM

HYDROGELS Hydrogels are crosslinked polymer networks that are insoluble but swellable in
aqueous medium. These materials offer an environment that resembles the highly hydrated state of
natural tissues, making them excellent candidates for tissue engineering and drug delivery. Several
types of crosslinking mechanisms have been investigated to develop hydrogels for tissue scaffolds
and drug delivery devices. These mechanisms include ionic (e.g., alginate), physical (e.g.,
pluronics and peptide self-assembling gels described in the section on Self-Assembled
Biomaterials, below), chemical (e.g., fibrin glue and multivinyl methacrylate and acrylate
derivatives), and hydrogen bonding (e.g., a new class of hydrogels described below). Several
synthetic and natural hydrogels have been recently reviewed (3840).

JI

Hydrogels that are chemically crosslinked via a radical-initiated chain polymerization of

multivinyl monomers or macromers have gained considerable interest because of their chemical
versatility and ease of fabrication. The polymerization reaction of methacrylate- or acrylate
[(meth)acrylate]-modified macromers can be initiated using photoinitiators or redox initiators
under physiological pH and temperature enabling in situ formation. Photoinitiators offer spatial
and temporal control over the polymerization process allowing for fabrication of complex 3-D
shapes, whereas redox initiators (e.g., ammonium persulfate and vitamin A) are typically milder
and potentially less toxic. Both synthetic (e.g., poly(ethylene glycol)-co-poly(lactic acid)
(PEG-co-PLA) (41), poly(vinyl alcohol) (PVA) (42), PVA-g-PLA (43) and natural polymers (e.g.,
hyaluronic acid (44, 45) and chondroitin sulfate (46)) have been modified with meth(acrylates) to
form crosslinkable macromers that form hydrogels with a range of different chemistries.
Alternatives to meth(acrylates) include novel hydrogels based on poly(propylene
fumarate-coethylene glycol) developed by Mikos and coworkers (47). The propylene fumarate
repeat unit contains a polymerizable vinyl group and a hydrolytically degradable ester group,
whereas ethylene glycol segments increase the hydrophilicity of the gels; together these facets
make this copolymer an attractive candidate for tissue engineering (48, 49). In general, crosslinked
hydrogels afford tight control over the final gel chemistry, macroscopic properties, and
124

degradation. Simple changes in the processing conditions (e.g., macromer concentration,

macromer molecular weight) can lead to a range of gel properties (42, 50), whereas simply mixing
different macromers in solution prior to polymerization can lead to networks that contain both
synthetic and natural components (46). Gel degradation is easily tailored by varying the length
and/or chemistry of the hydrolytic labile group, such as, for example, poly(lactic acid) (PLA)

-Z

JU

versus -caprolactone (41), by incorporating enzyme cleavable peptide sequences within the
crosslinker (51), or by employing natural polymers such as those mentioned above, which are also
sensitive to enzymatic degradation (45).

Fig. 5. A scanning electron microscopy image of a hydrogel foam prepared by solid colloidal gas aphrons of

N
G
LM

a PVA and arginine. Bar = 200 m.

JI

Photopolymerized hydrogels have been used more recently to successfully encapsulate

pancreatic islets (52), smooth muscle cells (53), osteoblasts (54), chondrocytes (5557), and
growth factors (53, 58) while maintaining cell viability and protein activity. In a recent study,
osteoconductive growth factors delivered from photopolymerizedPEGhydrogels were found to
enhance osteoblast mineralization (58). Degradable hydrogels based on PEG and PVAhave been
tuned to degrade at a rate that parallels the production ofECMby photoencapsulated chondrocytes.
This led to the development of a cartilaginous tissue throughout the hydrogel scaffold as
illustrated by the uniformly distributed glycosaminoglycans in Figure 4a (59). Furthermore, DNA
encoding for the green fluorescent protein (GFP) was coencapsulated with chondrocytes in
degradable photocrosslinked PEG hydrogels as a method for in situ gene delivery (60). Figure 4b
demonstrates successful in situ transfection of chondrocytes within the hydrogel.
A new class of hydrogels has been recently discovered by Nair et al. (61) that involves a unique
interaction between PVA and amino acids. Initial results based on nuclear magnetic resonance
(NMR) and infrared spectrometry suggest that hydrogen bonding is responsible for gel formation
(62). We have combined these unique hydrogels with a new approach to fabricate tissue scaffolds
using solid colloidal gas aphrons (CGAs) (63). CGAs or microfoams contain uniform air bubbles

that produce pores from 50500 m in size, suitable for tissue scaffolds. An example of a PVA
and arginine amino acid hydrogel scaffold prepared from CGAs is shown in Figure 5. This new
class of hydrogels offers several advantages over conventional gels, including the use of nontoxic
aqueous solvent, Food and Drug Administration (FDA) approved components, and short
preparation times.
SMART BIOMATERIALS Materials that can respond to changes in their surrounding
environment are attractive because these changes, particularly in vivo, can be exploited to control
125

parameters such as drug release, cell adhesiveness, mechanical properties, or permeability. Several
approaches have been examined for stimulants such as pH (64, 65), temperature (66, 67), and light
(68, 69).
The body employs changes in pH to facilitate a range of different processes. For example, along
the gastrointestinal track, food is broken down into nutritive substances in the stomach under
acidic pH 2 and subsequently absorbed in the small intestine (pH 7). Drug delivery by oral

JU

administration is attractive to many patients who require routine, periodic delivery of drugs.
However, for the drug to remain effective, it must survive the acidic pH of the stomach.
Researchers have developed pH-sensitive materials that take advantage of pH changes in the
gastrointestinal tract and carry drugs through the stomach for successful delivery in the small
intestine. pH responsive hydrogels prepared from poly(methacrylic acid) grafted with
poly(ethylene glycol) (PMAA-g-PEG) have been widely explored. These materials undergo
changes in swelling in response to pH. For example, the gel shrinks (trapping the drug cargo) at

-Z

low pH 2 owing to the formation of interpolymer complexes, whereas at physiological pH 7

N
G
LM

the gel can swell 325 times depending on its composition (releasing its cargo in the desired
compartment) (65). In vitro, these gels have low cytotoxicity (70) and have been shown to
improve protein transport across epithelial cell monolayers (71). As a practical demonstration of
this strategy, insulin loaded PMAA-g-PEG gels were delivered orally to diabetic mice. Their
glucose levels decreased significantly, suggesting that these materials can protect protein function
under acidic environments and in the presence of digestive enzymes (72).

JI

Another natural process that employs a drop inpHis intracellularwaste disposal. This process
involves trafficking of molecules to the lysosome for degradation. Many molecules are taken up
by the cells through receptor-mediated endocytosis, where they are entrapped in endosomes and
brought to the lysosome for destruction as depicted in Figure 6. Both endosomes and lysosomes
exist at pH values between 5.0 and 6.5, whereas the cytoplasm is at a pH of 7.4. The delivery of
therapeutics, e.g., DNA, RNA, and proteins, into a cell requires that the drug get into the cell
without being trafficked to the lysosome and degraded. To escape the degradative process, a
number of researchers have examined ways to take advantage of the pH gradient between the
endosome and the cytosol. A new class of polymers has been developed that targets disrupting the
endosomal membrane (64). These polymers have several unique features: (a) a
membrane-disruptive backbone (composed of dimethylaminoethyl methacrylate, butyl
methacrylate, and a styrene derivative); (b) PEG oligomers, which act as stabilizers at neutral pH,
are grafted to the membrane-disruptive backbone via an acid degradable linkage that enables the
release of the stabilizer under acidic environments; and (c) multiple specificities conjugated to the
PEG stabilizer for drug attachment and delivery allowing, for example, hexalysine to bind
oligonucleotides and lactose to target the ASGP receptor on hepatocytes (73). This new class of
polymers conjugated to a therapeutic drug forms a stable drug-polymer complex at physiological
pH. This complex can enter the cells via receptor-mediated endocytosis. Once in the endosome,
the acid-degradable acetal linkers are hydrolyzed, releasing both the stabilizing PEG chains and
the drug from the core membrane-disruptive polymer. As a result, the membrane-disruptive
polymer is free to disrupt the endosomal membrane and release the drug into the cytoplasm. When
these novel encrypted polymers were tested with fluorescein isothiocyanate (FITC) conjugated to
126

PEG, cytoplasmic delivery to hepatocyteswas observed by fluorescences. With a smaller

oligonucleotide as a model drug, delivery to the nucleus was possible (73).
A healthy human body maintains a constant temperature of 37C, and many researchers have
designed materials that can respond to changes in temperature from room temperature to body
temperature. Polymers that phase separate because of a change in temperature are often driven by
solution conformational changes of the hydrophobic and the hydrophilic moieties. Thermally
responsive biopolymers have been recently reviewed (66). Here, we briefly describe several
temperature-responsive polymers, with details deferred to the more focused reviews. The
usefulness of thermally responsive biopolymers is primarily based on their demonstration of a
polymer lower critical solution temperature (LCST). Two polymers that have been extensively
investigated are poly(N-isopropylacrylamide) (pNIPAAm) and elastin-like peptide (ELP) (see

-Z

JU

reviews 67, 74). For example, pNIPAAm has an LCST at 32C (75), whereas the LCST of ELP
can be controlled by varying the length of the ELP molecule and/or its amino acid composition to
obtain transition temperatures ranging from 30C to greater than 90C (76). The ELPs have been
used in tissue-engineering applications for chondrocyte encapsulation (77) and as a drug carrier
for cancer therapy (74). pNIPAAm and its derivatives have been used for a number of different
applications, including drug delivery (78), biological microelectromechanical systems (bioMEMs)
(79), and tissue engineering (80). A unique technique, termed cell sheet engineering, has arisen
from this thermally responsive polymer. At 37C, pNIPAAm becomes hydrophobic, promoting

JI

N
G
LM

protein adsorption and, subsequently, cell adhesion. By lowering the temperature below 32C,
cells cultured on pNIPAAm surfaces can be lifted off of the surface. This change from
hydrophobic to hydrophilic character over this transition results in the release of proteins and, thus,
adherent cells from the surface. This technique maintains intact cell-cell connections and their
associated extracellular matrix. Several approaches have been examined, including layering cell
sheets to reconstruct a homogenous tissue (81, 82) and layering several sheets made from various
types of cells to reconstruct multicellular tissues (83, 84). The detachment of the cell sheet was
accelerated when small amounts of PEG were grafted to the surface with pNIPAAm (85).

Fig. 6. A schematic of receptor-mediated endocytosis and the potential intracellular trafficking routes of
therapeutics.

Thermally responsive biopolymers can also be useful in controlling enzyme activity.

Polymer-enzyme conjugates have been constructed in which the polymer has been conjugated to
the enzyme near the site of enzymatic activity. At temperatures above the LCST, the polymer
conjugated to the enzyme assumes a collapsed state, blocking the active site and preventing
substrate interaction with the enzyme; however, at temperatures below its LCST, the polymer is
127

expanded, allowing substrate-enzyme interaction. This phenomenon has been demonstrated by

conjugating a short protein sequence derived from the active region of the enzyme endoglucanase,
which
hydrolyzes
amorphous
cellulose,
to
the
responsive
copolymer
N,N-dimethylacrylamide-co-N-4-phenylazo-phenylacrylamide (69). Interestingly, this copolymer
is both thermo- and photo-sensitive. Upon exposing to UV or visible light, the LCST shifts, giving
additional control over enzyme activity. Furthermore, this strategy can be adapted to other
enzymes by changing polymer size and the conjugation site.

N
G
LM

-Z

JU

Alternatives to environmental changes have also been investigated that are stimulated by a more
specific response. In 1985, Kost et al. (86) published a paper on a new hydrogel that swelled in
response to glucose. The crosslinked pHEMA was copolymerized with a monomer containing
pendant amine groups. The hydrogel system contained entrapped glucose oxidase. In the presence
of glucose, the enzyme produced gluconic acid, which lowered the pH, protonated the amines, and
induced the gel to swell. More recently, a unique approach to glucose sensing also based on
swellable gels was developed by Asher and coworkers (87). Specifically, the sensor employed an
anionic hydrogel fabricated from crosslinked polyacrylamide with tethered boronic acids
(4-amino-3-fluoro-phenylboronic acid, which are known to bind glucose reversibly) and stabilized
by PEG (87). Within the hydrogel was an embedded crystalline colloidal array (CCA) that
diffracts light. In this unique system, shown in Figure 7a, changes in glucose concentration result
in anionic gel swelling (high glucose concentration) or deswelling (low glucose concentration).
Changes in the gel volume cause a change in the CCA structure and hence a change in the
diffraction wavelength. This unique combination of elements allows the sensor to work effectively
at physiological pH and salt concentrations. The diffraction wavelengths are given in Figure 7b as
a function of glucose concentration. Such materials could be useful in providing an easy visual
recognition of glucose levels by sensing tear glucose levels.

Biologically Inspired Materials

JI

In an effort to design materials that will perform their intended functions in the presence of cells
and/or in vivo, we as engineers and scientists can look to biology and take advantage of biological
processes that have evolved over thousands of years. In this section, we present some of the most
current advances in biomaterials that have emerged by mimicking properties or processes that
occur in the body.

Figure 7 (a) A schematic of a glucose sensor fabricated from a polymerized crystalline colloidal array (CCA)
embedded in an ionic hydrogel containing a molecular recognition element for glucose. (b) Under physiological
pH, changes in glucose concentration affect gel swelling and subsequently affect the lattice structure of the CCA,

128

causing a shift in the diffraction wavelength. Glucose concentration (mM) is shown above the diffraction peaks.
Reproduced with permission from (87).

N
G
LM

-Z

JU

SELF-ASSEMBLED BIOMATERIALS One of the many fascinating aspects of biology is the

self-organization of molecules into hierarchal structures that perform a specific task. This
self-organization or self-assembly is based on the formation of weak noncovalent bonds, typically
hydrogen, ionic, or van derWaals bonds or hydrophobic interactions (88). For example,
amphiphilic molecules have hydrophobic and hydrophilic segments that self-assemble into distinct
structures, such as micelles, vesicles, and tubules that are nanometers in length. When dispersed in
an aqueous solvent, the hydrophobic segments in the amphiphilic molecules agglomerate, driving
out water and, as a result, producing a well-ordered structure that can be useful in a number of
different biomedical applications. A naturally occurring amphiphilic molecule is a phospholipid,
which largely composes the cell membrane. Alternatively, a polymer (or oligomer) of amino acids
can be synthesized to contain a number of different regions (hydrophobic, hydrophilic, charged,
etc.) that under certain conditions will self-assemble into a macroscopic hydrogel (89). In general,
the final three dimensional structure of the self-assembled material is dependent on the molecules
length and composition (9093). Thus, by systematically synthesizing a molecule, a desired 3-D
structure can be produced. These self-assembled biomaterials are promising new materials that
can be engineered for applications in nanotechnology and in tissue engineering for drug and cell
carriers. These materials may be particularly beneficial because the engineered peptides will take
on a 3-D shape that may be recognized by the in vivo environment, more so than a conventionally
synthesized material, and furthermore will breakdown into amino acids that the body is well
equipped to deal with.

JI

There have been significant investigations into the primary sequence of engineered peptides to
better understand the composition-structure relationships responsible for self-assembly. With such
an understanding, biomaterials can be tailored to achieve desired structures and functions. For
example, chemistry affects the packing of hydrophobic tails in peptide amphiphiles that results in
a variety of 3-D structures. In one study, changes to the hydrophobic tail length and amino acid
composition (where cysteine residues were replaced with alanine to prevent crosslinking of
cysteine thiols) did not alter a molecules ability to self-assemble into nanofibers, but did change
the morphology of the nanofibers (93). Changes in the amino acid sequence can lead to
morphological changes in self-assembled nanofibers by introducing kinks (94) or even branches
(95) into otherwise linear nanofibers.

In many cases, an external stimulus [e.g., pH (89), temperature (92), or salt concentration (91)]
is necessary to activate (or terminate) the self-assembly process. The amino acid sequence can
influence the point at which self-assembly occurs for a given external condition. Oligopeptides
with amino acid repeat units in the configuration of hydrophobic-negatively
chargedhydrophobic-positively charged can self-assemble in the presence of salt solutions (91).
For example, increasing the hydrophobicity of the oligopeptide results in a decrease in the critical
salt concentration required to induce self-assembly (91). In addition, the critical salt concentration
was affected by the overall length of the oligopeptide (91). Tuning the primary peptide sequence
to develop biomaterials that self-assemble in situ under physiological conditions has also been
investigated. For example, manipulating the amino acid sequence to give the oligopeptide a net
positive charge at neutral pH and low salt concentration prevented self-assembly, but when the
salt concentration was increased to physiological levels, the molecules self-assembled into a
129

hydrogel. Alternatively, certain oligopeptides self-assemble in the presence of divalent calcium at

physiological pH (93).

JU

The design of tissue-engineering constructs requires scaffolds with appropriate mechanical

stability and biological cues, e.g., adhesive proteins. It was found that the overall length (but not
hydrophobicity) of the oligopeptide (the hydrophobic- negatively chargedhydrophobic-positively
charged configuration described above) significantly influenced the shear modulus of the
self-assembled biomaterial (96). Oligopeptides containing an elastin-like repeat unit can be
crosslinked after self-assembly using bifunctional N-hydroxysuccinimidyl ester, which results in
crosslinking between two available amine groups, to further control the mechanical properties of
these materials (97). As a result, the tensile modulus was varied an order of magnitude by simply
adjusting the amount of crosslinker added to the gel (97). Self-assembled molecules with sequence
Ac-QQKFQFQFEQQAm can be further modified after gelation by tethering bioactive molecules
using transglutaminase as a catalyst to connect a peptide ligand, e.g., RGD, to the self-assembled
protein via a glutamine and/or lysine residue in the presence of calcium (98).

N
G
LM

-Z

A unique advantage of developing biomaterials from self-assembled oligopeptides is the ability

to incorporate biological cues within the material by simply integrating sequences directly into the
peptide backbone. Two peptide amphiphiles were designed to exhibit a hydrophobic alkyl tail and
a hydrophilic head group containing a biologically recognized amino acid sequence of RGD or
YIGSR (tyrosine-isoleucine-glycine-serine-arginine) derived from fibrinogen (20) and laminin
(99), respectively (100). Their chemical structures are shown in Figure 8a. At neutral pH,
molecule 1 has a net negative charge whereas molecule 2 has a net positive charge. Upon
combining, the two molecules self-assemble, and nanofibers spontaneously form at neutral pH
(Figure 8b).
The nanostructure of these self-assembled peptides offers a unique matrix structure containing
fibers that are three orders of magnitude smaller than typical fiber scaffolds used in
tissue-engineering applications [e.g., poly(lactic-co-glycolic acid) (PLGA) fibers typically range

JI

from 10100 min diameter]. Self-assembled peptides have been shown to support cell adhesion
for a number of different cell types (cell lines and primary cells) (101), as well as to support
neurite outgrowth by PC12 cells and primary neurons (102). In another study, 3T6 Swiss albino
mouse embryo fibroblasts were seeded onto a self-assembled layer containing a hydrophobic
moiety comprised of cholesterol and a PLA segment (103). A greater number of cells were present
on the self-assembled layers compared with a PLA control surface. Interestingly, the adhered cells
on the self-assembled layer were found to take up the cholesterol segment, suggesting a potential
mechanism for delivering a hydrophobic molecule to cells.

130

Fig. 8. Chemical structures of peptide-amphiphiles (a) that self-assemble into nanofibers (b) at neutral pH.
Reproduced with permission from (100).

N
G
LM

-Z

JU

Self-assembled peptide hydrogels have only recently been explored as cell carriers for tissue
engineering. The first study examined encapsulating chondrocytes in a peptide gel (with sequence
AcN-KLDLKLDLKLDL-CNH2) (104). Accumulation of glycosaminoglycans and the presence
of type II collagen (the two main components of cartilage) were observed in gels cultured up to 28
days. Furthermore, the equilibrium modulus and dynamic stiffness increased nearly 100-fold from
0 to 28 days, indicative of tissue development. In a separate study, an oligopeptide with the amino
acid sequence Ac-RADARADARADARADA-CONH2 (RAD), which self-assembles into a 3-D
peptide gel with interwoven fibers on the order of 1020 nm (102), was used as a
tissue-engineering scaffold (105). Specifically, liver progenitor cells were encapsulated in the
peptide gel (105). The encapsulated cells stopped proliferating and expressed the liver
developmental marker (C/EBP) as well as other hepatocyte markers such as albumin and
cytochromes P450 (CYP1A1 and CYP1A2), all indicative of differentiated liver cells.
Furthermore, when two different peptide sequences (n-AEAEAKAKAEAEAKAK-c and RAD
described above) were injected and formed in situ within the leg muscle of rats, no inflammation
or immune response was detected (102).

JI

BIOMIMETIC BIOMATERIALS Direct mimicry of biological processes represents an

important strategy in modern biomaterials. Hydroxyapatite is the natural mineral in bone. There is
now a huge amount of literature describing the use of hydroxyapatite and other forms of calcium
phosphate coupled with synthetic and other natural biomaterials to induce bone formation. Nacre,
the aragonite-rich lining of many seashells, also induces rapid bone formation. Nacre contains
about 5% protein that serves as a mortar to bind together a brick-like mineral structure. A
biomimetic approach to creating synthetic nacre-like structures has been published (106).
Normal tissues have a complex 3-D architecture important for the mechanics and functionality
of the biological organism. Biomimetically, we might emulate such structures with synthetic
biomaterials. Synthetic materials offer the ability to generate many different kinds of 3-D
structures with precise control over the final macroscopic properties and degradation profiles by
varying the chemistry and processing techniques. However, synthetic materials provide little
control over cell behavior and tissue response in vivo. Thus, to generate a biomaterial for a
specific cell type and/or tissue, efforts have focused on adding biological cues to synthetic
materials in an attempt to mimic the native tissue while simultaneously maintaining control over
the material properties.

131

-Z

JU

The primary focus of the biomimetic approach of emulating the extracellular matrix and tissue
architectures has been to incorporate short oligopeptides into synthetic materials to create
bioactive materials. The most studied sequence is RGD. Other sequences have also been
discovered that mediate cell adhesion, cellspecific adhesion, neurite extension, and degradation, to
name fewthese ideas have been reviewed elsewhere (19). A recent study examined cell
migration into a synthetically derived hydrogel matrix based on PEG (107). The PEG hydrogel
was fabricated to contain a matrix metalloproteinase (MMP)sensitive amino acid sequence as the
crosslink and RGD-tethered ligands throughout the bulk of the hydrogel. Interestingly, fibroblast
invasion (Figures 9a,b) into the gel was mediated by the presence of both the protease-sensitive
crosslinks and cell-adhesive moieties. When either was lacking, cell invasion was not observed.
This hydrogel was also placed in a rat calvarial critical-sized defect to examine bone ingrowth.
Intramembranous bone formation was observed throughout the defect area when the
MMP-sensitive gel was used in conjunction with human recombinant bone morphogenic protein-2
(Figures 9ce). The ability of cells to infiltrate a matrix loaded with a growth factor was essential
for bone ingrowth. This study demonstrates the importance of developing biomaterials that
express multiple bioactive elements. Such a combination of elements is likely necessary for
regeneration of new tissues.

JI

N
G
LM

ACELLULAR TISSUES Naturally derived tissues are attractive materials for cell seeding and
tissue-engineering applications because they are comprised of an extracellular matrix in native
conformation and composition. The tissue is decellularized and cell debris removed to reduce host
immune response. The resulting material retains many of the natural tissues mechanical and
biological properties. An acellular tissue commonly used in humans today to replace damaged
malfunctioning heart valves is a glutaraldehyde-fixed porcine heart valve. Glutaraldehyde
crosslinking increases the life span of the implant by minimizing enzymatic degradation; however,
it significantly alters the mechanics of the implant (108). In tissue-repair or tissue-engineering
applications, a resorbable tissue is desired. Acellular porcine small intestinal submucosa (SIS) has
been studied extensively as a naturally derived, resorbable material (109). Submucosa has been
used as a xenogeneic scaffold for tissue repair in both animal studies and human clinical trials.
Interestingly, upon implantation in the bladder, SIS exhibits no inflammation, has good
vascularization, and lacks the foreign body reaction typically observed for biomaterials.
Submucosa enables tissue ingrowth and has been used in a number of different applications as a
graft to promote tissue repair. In many cases, SIS is rapidly remodeled by host tissue and 90%
degraded within 4 weeks (110). Submucosa has been approved by the FDA for use in humans for
skin repair, softtissue support, and rotator-cuff repair. The success of SIS has been attributed to its
ECM proteins and glycosaminoglycans, cytokines, and growth factors that are retained within the
matrix, for example vascular endothelial growth factor (111), transforming growth factor 1
(TGF1), and basic fibroblast growth factor (112). Endogenous TGF1 in the SIS has been
implicated in the immunosuppressiveness of helper T (Th) cells (113). When compared with type
I collagen and PLGA subcutaneous implants, a marked increase in recruited marrow-derived cells
was observed for the SIS implants (114). This study further demonstrates the importance of
scaffold chemistry (and structure) and the presence of growth factors that, combined, are
important for tissue growth.

Patterned Biomaterials
The ability to pattern biomaterials has enabled researchers to create cost-effective diagnostic tools
132

by miniaturizing and combining laboratory procedures onto a single device. Patterning can also
assist us in better understanding many biological processes by mimicking in vivo structures. In
this section, we present some recent advances in bioMEMs technology and tissue-engineering
scaffold developments that exploit patterned biomaterials.

N
G
LM

-Z

JU

BioMEMs BioMEMs are technologies that can offer micrometer- and even nanometer-scaled
devices designed specifically for biological applications. These miniaturized, engineered
biosystems have many benefits in biology; in particular, experiments can be performed at the scale
of a single cell. Furthermore, these devices reduce cost by shortening reaction times and
minimizing reagent volumes. These biological devices have led to the idea of a total microanalysis
system, or labon-a-chip technology, in which multiple analyses can be performed in series and/or
in parallel on one device (115). Such a system has the potential to revolutionize medicine in both
diagnostics and in basic research. Before a total microanalysis system can be fully developed,
many specific fabrication and biointeraction challenges must be addressed. The promise these
systems hold for advancing medicine has stimulated a huge bioMEMs research endeavor. Because
bioMEMs is one of the fastest growing areas that rely on biomaterials, we present some of the
latest findings in bioMEMs technology. In particular, we discuss material processing using soft
lithography and new techniques in photolithography, as well as devices that take advantage of
smart polymers.We direct the reader to a number of recently published reviews on bioMEMs for a
more complete discussion (116124).

JI

SOFT LITHOGRAPHY Soft lithography has emerged in recent years as a popular

microfabrication technique that is inexpensive, facile, and bio-friendly (118). Soft lithography
begins with a patterned elastomeric material, typically poly(dimethylsiloxane) (PDMS), that is
fabricated using standard photolithography procedures and can generate patterns as small as 30
nm, as illustrated in Figure 10 (125). There are several merits of soft lithography: (a) Multiple
PDMS stamps can be made from one mask; (b) the PDMS stamps are often reusable (minimizing
clean room time); (c) the elastomeric and hydrophobic properties help establish and maintain good
contact with rough surfaces; (d) PDMS is transparent, allowing visualization through the stamp; (e)
the surface chemistry of the PDMS is easily modified; and finally, ( f ) PDMS is a nontoxic
material through which oxygen and carbon dioxide can permeate.

Fig. 10. Scanning electron micrograph of 30nmfeatures fabricated from composite PDMS masks and
phase-shifting photolithography. Reproduced with permission from (125)

A multitude of uses has emerged because of the versatility of these PDMS stamps. Figure 11
illustrates several recent applications of these materials. Microcontact printing of proteins has
133

N
G
LM

-Z

JU

enabled the facile transfer of proteins to surfaces either directly or indirectly using
microcontact-printed alkanethiols with defined geometries (126129) (Figure 11a). By spatially
controlling protein adsorption on a surface, researchers have been able to better understand cell
processes (130, 131). For example, Chen et al. (130) found that focal adhesions, which mediate
cell adhesion to ECM, do not depend on the total amount of ECM but rather on cell spreading.
Gray-scale photolithography (132) used PDMS stamps to provide a means to generate structures
of varying heights in one easy step (133, 134). One example is shown in Figure 11b (134) where
the channels were filled with a solution containing a gradient of UV-absorbing dye. Upon
irradiation with UV light (made possible by the PDMS transparency), the underlying photoresist
was exposed to a gradient of light intensities, resulting in structures of different heights. The last
example, illustrated in Figure 11c and experimentally demonstrated in Figure 11d, shows that
these materials can be used to examine subcellular processes (135). Because flow in
microchannels is laminar, a bioactive molecule can be introduced into one of the inlets, resulting
in localized flow through the channel. Thus, microchannels provide a means to expose only a
portion of a cell to a particular molecule. Figure 11d demonstrates the effects of treating a part of a
single bovine capillary endothelial cell with trypsin, an enzyme that digests the underlying ECM
proteins. The sequential micrographs demonstrate that only the part of the cell exposed to trypsin
detached from the surface, and when trypsin was replaced with medium containing serum, cell
reattachment occurred.
PHOTOLITHOGRAPHY When precision on the micron scale is desired (typical in biological
applications), simple photopolymerization reactions can be used to minimize the lengthy
preparatory procedures involved in designing a mold for soft lithography. An ultra-rapid
prototyping method for designing channels in a microfluidic device uses a solution of
photoreactive monomers, a photoinitiator, and a photomask. A device with channels as small as 25

m can be easily prepared in a matter of minutes (136).

JI

In an effort to design a versatile and simple patterning method, living radical polymerization
(LRP) has been investigated for designing microfluidic devices. In LRP, a photoiniferter [a
derivative of dithiocarbamate that acts as an initiator, transfer agent, and terminator in radical
polymerizations (137)] is bound to a surface from which a polymer chain can be grafted. The
grafting can be controlled by using a photomask in which grafting occurs only where UV light is
transmitted (138, 139). Furthermore, LRP enables 3-D micropatterning up to 1 mm. For example,
the microfluidics group at the University of Colorado has demonstrated the ability to form
bioMEM devices that have cells entrapped in 3-D PEG hydrogels that have been grafted to the
surface using LRP technology (140, 141). This device, shown in Figure 12, enables cell culture
without the need for individual wells and has potential for use in high-throughput screening to
develop and/or optimize tissue-engineering scaffolds and culture conditions in 3-D matrices.

SMART POLYMERS The future of bioMEMs technology relies on the development of

inexpensive devices that respond to either external or internal stimuli. Several approaches have
been investigated. For example, coating a surface with a thermoresponsive polymer such as
pNIPAAm allows control over the surface chemistry by simple changes in temperature, as
described in detail above. Cheng et al. (79) has fabricated a device with microheater arrays that
allows the surface chemistry of a small region to be reversibly switched on demand. The results
shown in Figure 13 demonstrate that cell adhesion can be localized to a certain region by simply
turning on one of the microheater array elements through an external power supply. Such a
134

technology allows for attaching several different types of cells in a given device. Huber et al. (142)
demonstrated that the initial adsorption of proteins on heated pNIPAAm surfaces is driven by
kinetics and is a function of the size of the protein. Thus, by controlling adsorption times, protein
separation was possible. In another study, genetically engineered, temperature-sensitive,
elastin-like polypeptides were used to selectively recover a protein, ELP fusion peptide, directly
from a cell lysate (143). The ELP polypeptides can selectively bind ELP fusion peptide above the
LCST of the polypeptide (allowing separation of the protein of interest), whereas below the LCST
the protein can be recovered for downstream processing, thus minimizing the lengthy purification
steps.

JI

N
G
LM

-Z

JU

BioMEMs that respond to internal stimuli are attractive because the device can be portable,
allowing analysis to be performed offsite and potentially by the patient him/herself. Many of these
advances have examined responsive polymers that act as valves. For example, Zhao et al. (144)
modified channels within a microfluidic device, using simple photolithography techniques, to
pattern surfaces with different free energies to serve as pressure-sensitive valves. In another study,
LRP was used to generate pH-sensitive microchannels that could be opened and closed by simple
changes in the pH (139). Strategies that use environmentally responsive polymers as pumps to
induce flow through the microchannels are also a subject of intense interest.

135

JU
-Z
N
G
LM
A
JI
Fig. 11. Schematics illustrating several applications of PDMS stamps: (a) Microcontact printing allows transfer of
materials, including biologically active molecules, e.g., proteins, to a surface in a well-defined pattern; (b) gray
scale photolithography provides a facile method with which to pattern structures of varying heights; and (c) partial
cell treatment uses laminar flow within microfluidic channels to treat part of a cell to a particular molecule. Panel
(d) illustrates the partial treatment of a bovine capillary endothelial cell with trypsin, an enzyme that digests the
underlying extracellular matrix proteins, causing cell detachment. Panel (d) reproduced with permission from
(135).

TISSUE ENGINEERING Tissue engineering requires a 3-D matrix that will not only guide
tissue formation into the desired shape but also provide adequate transport of nutrients and growth
factors to promote tissue growth. In an effort to develop complex scaffolds that better match
136

natural tissue structure, several approaches have been investigated to three dimensionally pattern
materials. For example, an indirect solid free-form (SSF) fabrication (145) has been developed
that combines direct SSF [that uses computational design to generate an exact external shape and
macroporous architecture with minimum features of 100 m (146)] with standard techniques to
generate porous tissue-engineering scaffolds (e.g., salt leaching). For example, PLA scaffolds
were fabricated that exhibited a 3-D structure similar to that of human trabecular bone with
macropores (500800 m) to micropores (50100 m) (145).

-Z

5.4. THE FUTURE OF BIOMATERIALS

JU

Efforts have also been explored to develop 3-D scaffolds that contain multiple cell types
wherein their spatial arrangement is controlled through 3-D patterning. Tan et al. (147) used
channels from a PDMS stamp as a 3-D template for encapsulating human umbilical vein
endothelial cells in a collagen type I matrix. A second cell type, human lung fibroblasts, was then
selectively patterned onto the collagen-cell matrices. Alternative and simpler methods have used
photopolymerization reactions to generate patterned 3-D hydrogel scaffolds. In one such study,
fluorescently labeled HepG2 cells were photoencapsulated in PEG hydrogels during patterning to
generate multiple layers of cells in defined geometries (148).

There are trends in biomaterials, many of them highlighted in this review, that suggest the future
of biomaterials. Medical implants and in vitro systems are addressed separately.

JI

N
G
LM

Implanted medical devices are essential to modern medical practice, and they save lives or
improve the quality of life for millions. However, they still heal like foreign objectsthey are
encapsulated and walled off.With the major investment by medical-device companies in
development, manufacture, marketing, surgeon acceptance, and regulatory approval, it is unlikely
that medical devices as we know them will change substantially in the next ten years. However, by
surface modification of existing medical devices, we might expect to see devices that heal in a
physiologically normal fashion (a vascularized reconstruction of the anatomy). Such a normal
healing will improve the performance of many devices that are now hampered by poor healing
(e.g., the problems associated with the foreign body capsule). For example, with integrative
healing characteristics coupled with advances in materials and robotics, implanted limbs may not
only replace the functionality of lost limbs but also provide some functionality even superior to
that possible in flesh and blood limbs.
Another area of research that will lead to improvements in the performance of implants is a
reduction in the propensity toward infection. This may be accomplished through antibacterial
surface treatments, nonfouling surfaces, and antibiotic controlled release strategies. Also, general
advances in controlled drug release will permit the development of devices with both device and
drug functionality. The potential of this approach is now being realized with drug eluting
cardiovascular stents that inhibit restenosis. Finally, advanced smart materials may be used for
tissue, organ, and system replacement. A polymeric artificial muscle or an insulin-releasing
system that intrinsically responds to blood glucose are examples of possible in vivo systems that
use smart materials concepts and also may substantially advance the treatment of diseases or
organ/tissue replacement.

Two technologies, tissue engineering and regenerative medicine, both in development now, will
ultimately revolutionize medical implants. Tissue engineering will allow replacement of many
tissues and organs with living, functional replacements. Tissue-engineering advances will rest
137

upon advances in biodegradable polymers, rapid prototyping, drug delivery, cell culture, stem cell
methodologies, angiogenesis, and biomimetic strategies for recreating extracellular, matrix-like
biomimetics. Also, advances in cell culture, gene delivery, cell and tissue storage, sterilization,
and surgery will be needed to fully realize the potential of tissue engineering. Regenerative
medicine, permitting in vivo regeneration of whole organs and tissues, will ultimately replace
tissue engineering. Technologies important to making regenerative medicine a reality include
advances in biology, gene delivery, and controlled release.

N
G
LM

-Z

JU

With respect to in vitro systems, regulatory issues are not so constraining, and we can expect to
see substantial advances in devices both for diagnostics and therapeutics. Smart materials,
nonfouling surfaces, and MEMS concepts will enter into these devices. A chip array with
complementary DNA to 100 different bacteria can rapidly recognize the origin of a bacterial
infection. An array device with the whole human genome on it will permit infants to be diagnosed
at birth with problems likely to crop up later in life. Preventative medicine, buttressed by this
foreknowledge of conditions that probably will develop later, can greatly improve the quality of
life. Possibly, aMEMSdevice containing a living culture of cells from a persons body might be
used to synthesize on demand specific pharmaceuticals tuned to the physiology of the individual.
The possibilities are almost limitless for developments using smart systems and micro- and
nanofabrication. These in vitro systems of the future will be grounded in an advanced
understanding of the way biological systems interact with synthetic surfaces. Careful ethical
deliberation will also enter into these new technologies that have the potential to alter the
expectations for humans.

CONCLUSIONS Biomaterials is a field with just over a half century of innovation and
development. The success of biomaterials is seen in the importance to modern medical therapies,
the economic potential of the market, and the steady growth of the field over more than 50 years.
Tissue engineering will change the face of biomaterials, but tissue engineering still makes use of
biomaterials. Regenerative medicine will reduce the need for synthetic materials in the body. Still
many applications will require synthetics, and we can envision the need for biomaterials well
through this century.

JI

Ref. Buddy D. Ratner and Stephanie J. Bryant. Annu. Rev. Biomed. Eng. 2004. 6:4175.

138

Chapter 6
Membrane Materials and Membrane Preparation

6.1. Introduction
6.2. Membrane Market
6.3. Membrane Preparation
6.4. Presently Available Membranes for Liquid Separation

6.6. Gas Separation with Membranes

-Z

6.1. Introduction

JU

6.5. Surface Modification of Membranes

N
G
LM

Membrane technology is presently an established part of several industrial processes. Well known
is its relevance in the food industry, in the manufacture of dairy products as well as in the
automotive industry for the recovery of electro-painting baths. Membranes make possible the
water supply for millions of people in the world and care for the survival of the unnumerable
people suffering from kidney disease. The chemical industry is a growing field in the application
of membranes, which, however, often requires membrane materials with exceptional stability. The
first part of the book will discuss the currently available membranes for different processes, which
are suitable for the chemical industry. Information on different methods of membrane preparation
will be given. Different materials will be compared, taking into account physical characteristics
and chemical stability.

6.2. Membrane Market

JI

The membranes and module sales in 1998 were estimated at more than US$ 4.4 billion
worldwide, shared by different applications (Fig. 1). If equipment and total membrane systems are
also considered, the estimate would be double. This amount according to data from 1996, about
45 % of the sales were in the United States and 29% in Europe and the Middle-East. The market in
Asia and South America is growing fast.

Fig.6.1. Membrane and module sale for different process applications.

Hemodialysis/hemofiltration alone had sales of over US$ 2200 million in 1998. Reverse
osmosis (RO), ultrafiltration (UF) and microfiltration (MF) together accounted for 1.8 billion
dollars in sales. About US$ 400 million membranes and modules are sold each year worldwide for
use in reverse osmosis. About 50 % of the RO market is controlled by Dow/FilmTec and
Hydranautics/Nitto. They are followed by Du Pont and Osmonics. Membranes are applied during
139

sea water desalination, municipal/brackish water treatment and in the industrial sectors. The
market is expected to grow at a rate of 10 %/year.
Ultrafiltration membranes and modules brought about US$ 500 million in sales in 1998 with an
expected growing rate of 10 % a year. Over 58 % of the sales are in the US. In contrast to RO, the
UF-market is shared by a large number of companies, but the leaders are Pall, Amicon/Millipore
and Koch. One of the largest industrial sectors for ultrafiltration is still the recovery of electrocoat
paints. UF-membranes are also in large scale responsible for supplying pure water for the
semiconductor industry. Growing demands of ultra-high purity chemicals in this sector could also
be supplied by UF with the availability of chemical resistant membranes. Oil/water separation is
now a large application for UF in industrial sectors such as metal cleaning and wool scouring and
is still growing with the implementation of new environmental legislation.

N
G
LM

-Z

JU

Microfiltration stands for US$ 900 million worldwide sales, mostly in the US, with growing
rates of 8% a year. Main applications are the production of sterilized water for the pharmaceutical
and biotechnology industry. In the semiconductor industry, MF is used to remove particles from
air and produce pure water.
Gas separation (GS) is a relatively young technology and accounts for about US$ 230
million/year, but is growing fast with a rate higher than 15 % a year. For the electrically driven
membrane processes the sales in 1998 were around US$ 180 million. For pervaporation (PV) in
1996 the market was about US$ 26 million, with a growing rate of 20 %.

6.3. Membrane Preparation

Different methods of polymer membrane preparation have been covered in several reviews.
Membranes can be classified, according to their morphology as shown in Fig. 6.2.

JI

Fig. 2: Membrane classification according to the morphology

Dense homogeneous polymer membranes are usually prepared (i) from solution by solvent
evaporation only or (ii) by extrusion of the melted polymer. However dense homogeneous
membranes only have a practical meaning when made of highly permeable polymers such as
silicone. Usually the permeant flow across the membrane is quite low, since a minimal thickness is
required to give the membrane mechanical stability.
Most of the presently available membranes are porous or consist of a dense top layer on a
porous structure. The preparation of membrane structures with controlled pore size involves
several techniques with relatively simple principles, but which are quite tricky.
Commercial membranes were already produced in Germany by Sartorius in the early 20s.
However they had only a limited application on a laboratory scale. The break-through of the
membrane technology came first in the 60s with the development of the asymmetric porous
membranes by Loeb and Sourirajan. The asymmetric membranes combine high permeant flow,

140

provided by a very thin selective top layer and a reasonable mechanical stability, resulting from
the underlying porous structure. An asymmetric structure characterizes most of the presently
commercially available membranes, which are now produced from a wide variety of polymers. By
far the most common method used in generation of asymmetric structures in membranes is the
phase inversion process. Other methods used to form pores in membranes will be discussed in
the following sections. Particularly in the case of microfiltration, several techniques other than
phase inversion are currently applied in the industry.

Phase inversion
The phase inversion process consists of the induction of phase separation in a previously
homogeneous polymer solution either by temperature change, by immersing the solution in a
non-solvent bath (wet process) or exposing it to a non-solvent atmosphere (dry process).

-Z

JU

In the thermal process, a low molecular weight component usually acts as a solvent at high
temperature and as a non-solvent at low temperature. It is then removed after formation of the
porous structure. Although the thermal process can be applied to a wide range of polymers, it is
especially interesting for those with poor solubility, such as polypropylene, which can be hardly
manufactured into a porous membrane by other phase inversion processes. An isotropic
microporous structure is usually formed.

JI

N
G
LM

The isothermal phase inversion is commercially more widespread. Usually the polymer solution
is immersed in a non-solvent bath (wet process) and a solvent-non-solvent exchange leads to
phase separation. The polymer rich phase forms the porous matrix, while the polymer poor phase
gives rise to the pores. The morphology is usually asymmetric, with a selective skin on the surface,
as shown in Fig. 6.3.

Fig. 6.3. Asymmetric porous membrane.

The pore structure is generated by phase separation. Phase separation in this case is mainly a
liquid-liquid demixing process, although solid-liquid demixing may also play an additional role in
systems containing a crystallizable polymer, such as cellulose acetate and poly (vinylidene
fluoride). After immersion in a non-solvent bath, the solventnon-solvent exchange brings the
initially thermodynamically stable system into a condition, for which the minimum Gibbs free
energy is attained by separating into two coexisting phases. The predominant mechanism of phase
inversion leading to pore formation and the thermodynamics involved are the subject of a fruitful
and sometimes controversial discussion in the literature, as well as the most probable paths in the
phase diagram. A simplified diagram is shown Fig. 6.4.

141

Fig. 6.4. Mechanism of phase separation during membrane formation.

N
G
LM

-Z

JU

Basically the mechanism of phase separation depends on the crossing point into the unstable
region. If the solvent-non-solvent exchange brings the system first to a metastable condition (Path
A), the nucleation and growth mechanism (NG) is favored. A dispersed phase consisting of
droplets of a polymer poor solution is formed in a concentrated matrix. If no additional
non-solvent influx or temperature change in the system were induced, the composition inside the
nuclei would be, from the beginning, that expected at the equilibrium and would, practically, not
change with time. Only the size of the droplets increases with time. If the demixing path crosses
the critical point, going directly into the unstable region (Path B), the spinodal decomposition (SD)
predominates. Aconcentration fluctuation appears in the initially homogeneous system and
progresses with increasing amplitude, leading to a separation in two co-continuous phases. Here
again the polymer poor phase will form the pores. The initial steps of phase separation, either by
NG and SD can be relatively well described according to theories of phase separation. However,
at later stages, both NG and SD usually progress to a phase coalescence and the final structure can
only be predicted with difficulty.

JI

At least as important as the starting mechanism of phase separation is the point where the
developing structure is fixed. Parallel to demixing, as the concentration of the polymer solution
changes, by solvent-non-solvent exchange, the mobility of the system decreases. Reasons for that
may vary from physically unfavorable polymer-solvent (or non-solvent) interaction, leading to
stronger polymer-polymer contacts, to vitrification of the polymer concentrated phase, as the
solvent concentration decreases, and also in some cases partial crystallization. If the system gels
and solidifies directly after the first steps of phase separation (for instance at t2), the membrane
will have a fine pore structure, which keeps the original characteristics given by the initial
demixing mechanism. If NG demixing stops during the initial stages, a morphology of closed cells
would be favored. At later NG stages, the nuclei would grow and touch each other forming
interconnected pores. The SD demixing would favor the formation of an interconnected pore
structure from the beginning.

An asymmetric structure is usually formed across the membrane since the solventnon-solvent
exchange may lead to different starting conditions for phase separation at layers far from the
surface. Besides the NG and SD demixing, other factors influence the morphology. The whole
membrane structure usually can be classified as spongelike or finger-like.
Finger-like cavities are formed in many cases, as the non-solvent enters the polymer solution.
This macrovoid structure may contribute to a lack of mechanical stability in membranes to be used
at high pressures. A combination of factors is responsible for the formation of macrovoids and this
topic has been well reviewed in the literature. For practical purposes, the predominance of a
sponge-like or a macrovoid structure can be induced in different ways. Basically, the sponge-like
142

structure is favored by
i)

increasing the polymer concentration of the casting solution

ii)

increasing the viscosity of the casting solution by adding a crosslinking agent

iii)

changing the solvent

iv)

adding solvent to the non-solvent bath

N
G
LM

-Z

JU

The growth of a polymer poor phase by SD or NG is an isotropical processes, which take

place as soon as the solvent-non-solvent contents supply the thermodynamic condition for
demixing. To understand the macrovoid formation, a quite interesting explanation was provided
by Koros as depicted in Fig. 6.5. For that, the coupling of the (NG or SD) demixing processes with
the rapidly moving front of non-solvent must be considered. If the non-solvent diffusion rate into
the polymer poor phase being formed exceeds the rate of outward solvent diffusion, the macrovoid
formation is favored. The diffusivity of water is usually expected to be one-to-two orders of
magnitude higher than the diffusivity of bulkier organic solvents. The main driving force for the
non-solvent (usually water) influx is the locally generated osmotic pressure. This could be
hypothetically approximately 100 bar with a difference of only 5 mol % non solvent concentration
between the initial nucleus and the approaching front. As water moves into a polymer poor
nucleus, its wall is deformed, expanding in the form of a tear. If the walls are fragile, the nucleus
may rupt giving rise to macrovoids with unskinned walls. If the walls are stronger, as in the case
of nuclei growing in a matrix with higher polymer concentration, the deformation can be
restrained or even totally inhibited, giving rise to a macrovoid free structure.

Fig. 6.5. Non-isotropic nucleus growth during macrovoid formation in membrane.

JI

Increasing the polymer concentration of the casting solution to suppress macrovoids has been
well registered in the literature for a wide spectrum of polymers such as cellulose acetate, aromatic
polyamide, and polyetherimide. Other factors such as the addition of crosslinking agents which
can improve the strength of the growing nucleus wall also contribute to a macrovoid free structure.
An example is the addition of amines to polyetherimide casting solutions.
Another way to suppress the macrovoid formation is to reduce the osmotic pressure between the
non-solvent moving front and the polymer poor phase inside the nuclei. This can be achieved by
adding non-solvent to the casting solution or adding solvent to the non-solvent bath. An example
is the addition of dioxane to an aqueous coagulation bath for a CA dioxane solution.

Changing the solvent may act in different ways. Solvents with higher diffusivity across the
nucleus walls would be able to leave the nucleus faster, while the non-solvent is added in, which
does not favour macrovoid formation. But even more effective in suppressing the macrovoid
formation are solvents which increase the solution viscosity or even promote a fast gelation,
making the nucleus wall stronger and resistant to deformation. Some examples of solvent
influence on membrane morphology are to be found in the literature for polyetherimide (Fig. 6.6)

143

and cellulose acetate.

Fig. 6: Polyetherimide (PEI) membranes prepared from different casting solutions: (left) 17.5 wt% PEI in
dimethylacetamide; (middle) 17.5 wt % PEI in 5.5 wt % tetraethoxysilane and 77 wt % dimethylacetamide; (right)
15.5 wt % PEI in 28 wt % THF and 56 wt % g-butyro lactone.

JU

6.4. Presently Available Membranes for Liquid Separation

JI

N
G
LM

-Z

6.4.1. Membranes for reverse osmosis

The most common membrane materials for reverse osmosis membranes are cellulose acetate,
polyamide and the thin film composites, prepared by interfacial polymerization on the surface of
a porous support. A review of composite membranes was published by Petersen. Cellulose acetate
(CA) was one of the first membrane materials, and it is still being successfully used, especially in
water treatment (in spiral wound modules). They usually allow quite high water flows with low
salt solubility. One advantage, when compared to the polyamide based membranes, is its chlorine
tolerance. Also because of the neutral surface, cellulose acetate membranes usually exhibit a more
stable performance than polyamide membranes in applications where the feed waters has a high
fouling potential, such as in municipal effluents and surface water supplies. However, CA
membranes are drastically less stable in organic solvents than polyamide. The recommended pH
range is between 3 and 7, they are less resistant to biological attack and the recommended
temperature is lower than 50C. The susceptibility to hydrolysis increases with temperature and is
an inverse function of the degree of acetylation. Aromatic polyamides have a much higher solvent
resistance and may be used in a wider pH range (pH 411). The main application is the treatment
of brackish water and seawater. They can be produced in very thin hollow fibers with large
surface area/volume. The membrane top layer is however quite thick (>0.2 m), which leads to
relatively low water flows. The main disadvantage is the very low chlorine tolerance.
Integral asymmetric membranes have a relatively low manufacturing cost. CAin particular,
dominates a significant part of the membrane market for water treatment due to its low cost.
However the possibility of expanding the application of reverse osmosis in separations which
demand membranes with higher performance came only with the advent of the thin film
composite (TFC) membranes. They consist of an ultrathin layer, usually of polyamide or
polyetherurea, which is polymerized in situ and crosslinked on an asymmetric porous support,
usually polysulfone. Since the dense selective layer is very thin, the membranes can operate at
higher flux and lower pressure. The chemical stability is very good, although the chlorine
tolerance is low. They are not biodegradable and can operate in a pH range of between 2 and 11.
The membrane preparation consists of immersing the porous support in an aqueous solution
containing a water-soluble monomer. After that the support is immersed in a solution of the
second monomer in a non-polar solvent. Both monomers are only allowed to react at the interface
between organic and aqueous solution, forming a thin polymer layer at the surface of the porous

144

N
G
LM

-Z

JU

support. As soon as the polymer layer is formed it acts as a barrier for the monomer transport and
avoids the continuity of the polycondensation. On the other hand, any defect on the polymer layer
is immediately repaired, since monomer transport and polycondensation is allowed at that point.
One of the most successful TFC membranes is the FT-30, developed by Cadotte in the North Star
Laboratories and presently commercialized by Dow. The reaction involved in the preparation of
the FT-30 is as follows:

Fig. 6.7. FILMTEC FT-30.

JI

The polyamide layer is formed on an asymmetric microporous polysulfone support cast on a

polyester support web. The polyester web gives the major structural support and the polysulfone
support with small surface pores with diameter of ca. 15 nm is the proper substrate for the
formation of a 0.2 m polyamide top layer (Fig. 7). The FT-30 has been optimized for different
applications, being commercialized as FILMTEC TW-30 (for municipal tapwater), BW-30 (for
brackish water) and SW-30 (for sea water conversion to potable water). Salt rejections higher than
99.5 % can be obtained with fluxes of 0.6 m3/m2 day or rejection. With 0.2% salt feed solution,
membranes work at 1.6 MPa with rejections above 98 % and fluxes of 1 m3/m2 day. This
represents a reduction up to 50 % in operating pressure for water treatment in comparison to
commercial cellulose acetate membranes. The rejection of other solutes by FT 30 is shown in Tab.
1. The maximum operating pressure of the FT 30 is about 7 MPa with free chlorine tolerance <
0.1 ppm.
Another very successful development for reverse osmosis is the energy saving membrane series
produced by Nitto Denko (Tab. 2). The membrane filtrating layer is also an aromatic polyamide.
Due to its irregular surface, the actual membrane area available for the permeation is much larger
than it would be in the case of a smooth surface on the same porous support. High fluxes are
therefore obtained.
Tab. 1: Rejection of different solutes by FT30

145

JU

2000 ppm solute, 1.6 MPa, 25 C, pH7 (unless otherwise noted)

N
G
LM

-Z

Tab. 2: Characteristics of NITTO DENKO RO membranes.

JI

6.4.2 Membranes for nanofiltration

While reverse osmosis and ultrafiltration were being established in several applications, there
was a lack of available membranes with cut offs between 400 and 4,000 g/mol. An increasing
interest in NF membranes developed in the last decade. One of the main applications is water
softening. The following membranes are recommended for use in water softening systems:
UOPFluid Systems modules 8231-LP(cellulose acetate blend) and 8921-UP (TFCS polyamide),
FILMTEC NF 70 and NF 40, Toray modules SCL-100 (modified cellulose acetate) and SU 600
(composite polyamide), Nitto Denko NTR-729 HF, Desalination Systems Desal-5 and DuPont
SM15.
However the improvement of solvent stability of available NF-membranes opens a wide range
of potential applications in the chemical and pharmaceutical industry.
FILMTEC/Dow has commercialized the NF55, NF70 and NF90 (water flow of NF55 > NF70 >
NF90) membranes, which work in the range of nanofiltration, being able to reject at least 95 %
magnesium sulfate. The top layer in this case is a fully aromatic crosslinked polyamide. The exact
composition is not known. The chlorine tolerance is lower than 0.1 ppm and the pH range in
146

JU

operation is 39. One procedure to prepare nanofiltration membranes is the interfacial

polymerization between a piperazine or an amine substituted piperidine or cyclohexane and a
polyfunctional acyl halide as described in US Pat 4769148 and 4859384. Another way to obtain
nanofiltration membranes is the modification of reverse osmosis membrane, as proposed by
Cadotte. The process involves contacting a crosslinked polyamide selective layer with a strong
mineral acid such as phosphoric acid at 100150 C, which is then followed by a treatment with a
rejection enhancing agent such as tannic acid or watersoluble polymers to selectively plug
microscopic leaks and defects. Another procedure to open polyamide RO-membranes consists of
contacting the membrane with ions to form a membrane ion complex, treating the membrane-ion
complex with an aqueous solution of an alkali metal permanganate to form manganese dioxide
crystals in the membrane and finally dissolving the crystals. In another procedure a reverse
osmosis membrane is treated with triethanolamine to open the pores. RO cellulose acetate
membranes can be opened by hydrolysis at very high and very low pH. However, it is difficult to
control this process.

N
G
LM

-Z

Nanofiltration membranes can also be obtained by coating ultrafiltration membranes with

different polymer solutions. Membranes with cut offs between 800 and 4500 g/mol and water
permeabilities of up to 10 l/h m2 bar could be obtained by coating PVDF membranes with
polyether-block-polyamide copolymers. A lower cut off was obtained by forming a polyamide
network dispersed in the blockcopolymer layer by reacting a polyether diamine and trimesoyl
chloride. Coatings of hydroxyalkyl derivatives of cellulose are used to prepare solvent resistant
membranes.

JI

Also in the range of nanofiltration the ESPA membrane, made of anionic aromatic polyamide,
has been commercialized by Hydranautics and the D- and H-Series membranes are produced by
Desal/Osmonics. The D-Series, which are thin-film nanofiltration membranes, have a molecular
weight cut off of 150-300 g/mol for uncharged organic molecules and can operate at pH values of
less than 1. They have been applied, for instance, by printed circuit board manufacturers to
recover soluble copper from acid rinse streams. Copper sulfate from acid rinse streams of a copper
refinery has also been successfully recovered with D-Series membranes. Another successful case
was the recovery of ammonium sulfate from ground water contaminated by a nickel refinery.
In a recent report the membranes Desal-5, from Osmonics, NF 45, from Dow, and PVD-1 from
Hydranautics, have been compared according to their performance in the separation of metal
sulfates and nitrates from their acids. The membrane characteristics, as provided by the
manufacturer and the retention of different ions are shown in Tab. 3. During the tests with
solutions containing multivalent salts, Desal-5 had the lowest flux and NF 45 the highest.
Tab. 3 Characteristics of some NF membranes in comparable conditions.

Cations of higher valency had a higher retention, especially in the case of NF 45, which might
be positively charged.
6.4.3 Membranes for ultrafiltration
147

Table 4 lists some commercially available membranes for ultrafiltration. The development of
UF membranes form different polymer materials is commented below.
Tab. 4: Some commercially available membranes for ultrafiltration.

JU

Polysulfone and polyethersulfone

UF-membranes are usually prepared by phase inversion. The most widely used polymer for the
preparation of UF-membranes is polysulfone (PSU) or polyethersulfone (PES).

JI

N
G
LM

-Z

The first developments of PSU membranes appeared in the 60s as an alternative to cellulosic
membranes. Since then several procedures have been described in the literature for PSU
membranes, in many cases using the high molecular weight polysulfone Udel P-3500
commercialized by Amoco. Agreat advantage when compared to cellulose acetate is its resistance
in extreme pH conditions, as well as its thermal stability. PSU has a Tg of 195 C and PES even
higher, at 230 C. Both PSU and PES are soluble in chloroform and dimethylformamide, and are
easily applied in conventional phase inversion processes. This high solubility is also the main
drawback of PSU as a membrane material, eliminating the use of polysulfone-supported
membranes in the processing of solvent-based feed solutions. It is also a problem coating the PSU
support with polymers, which are only soluble in organic solvents. Another disadvantage of PSU
and PES membranes is their hydrophobic character, which prevents spontaneous wetting with
aqueous media. Consequently, the membrane must be prevented from drying completely or the
membrane must be treated with a hydrophobic agent, glycerin, for example, before drying.
Another serious disadvantage of hydrophobic materials consists in the fact that they often possess
a powerful nonspecific adsorption capacity.This phenomenon, known as fouling, leads to a rapid
deterioration of the membrane permeability. Suggestions for hydrophilic membranes have already
been proposed that do not suffer from these disadvantages. Several procedures have been
proposed to make membrane surface more hydrophilic and they will be discussed later. One
effective way to make polysulfone membranes more hydrophilic is to prepare the membranes from
a mixture of sulfonated and non-sulfonated polysulfone. The sulfonation may be controlled to
limit the water solubility of the resulting polymer. Insolubilization can also be achieved by
crosslinking with additives such as polyols or polyphenols. Sulfonation has been a successful
alternative to the incorporation of other hydrophilic polymers, which are soluble in water.
Preparation of membranes from polymer blends with hydrophilic polymers has been well
described in the literature. Polyvinylpyrrolidone is one of the commonly applied polymers for this
purpose. Another common additive is poly(ethylene glycol). However hydrophilization of
membranes by using large quantities of water-soluble polymers has the disadvantage that the
hydrophilic nature of the membrane constantly decreases when they are used in aqueous media,
since the water-soluble polymer is washed out. Although not completely described, sulfonated
148

-Z

Polysulfone

JU

polysulfone probably coats polysulfone supports in the G-series membranes commercialized by

Desal. Sulfonated polysulfone seems to also play an important role in nanofiltration and reverse
osmosis membranes commercialized by Desal. According to Petersen the Desal-5 membrane
appears to consist of three layers: a microporous polysulfone, a sulfonated overlay and a top
highly ultrathin layer based on polypiperazineamide.

N
G
LM

Poly(ether sulfone)

JI

Poly (vinylidene fluoride) (PVDF) is quite interesting in the manufacture of UFmembranes due
to its chemical resistance. PVDF is resistant to most inorganic and orga nic acids and can be used
in a wide pH range. It is also stable in aromatic hydrocarbons, alcohols, tetrahydrofurane and
halogenated solvents. Furthermore it is resistant to oxidizing environments including ozone, which
is used in water sterilization. PVDF is semi-crystalline with a very low Tg (40C), which makes
it quite flexible and suitable for membrane application in temperatures ranging between 50 and
140 C, just prior to its melting temperature. Although stable in most of organic solvents, PVDF is
soluble in dimethyl formamide, dimethyl acetamide (DMAc), N-methyl pyrrolidone (NMP) and
dimethylsulfoxide, making membrane preparation by phase inversion possible. In an early patent
on PVDF membranes, solutions containing about 20% PVDF in DMAc were cast and immersed
in a methanol bath. Later a preparation was described using a casting solution in DMAc,
containing also ca. 17 % isopropanol and using an immersion bath with about 40 % water, 50 %
DMAc and 7% isopropanol. Membranes are also prepared from solutions in NMP containing
litium chloride and an immersion bath of methanol. US Pat 4203848 describes the preparation of
PVDF membranes by dissolving the polymer in boiling acetone and immersing it in a cold
water/acetone bath. Another interesting solvent is triethyl phosphate (TEP), a basic solvent, which
complexes with the acidic PVDF. The morphology of a PVDF membrane obtained from a solution
in DMAC is shown in Fig. 8. Like polysulfone, PVDF is highly hydrophobic and many attempts
to make them more hydrophilic have been described in the literature. One procedure is the
chemical treatment with a strongly alkaline solution either in the presence of an oxidizing agent or
with a polymerization initiator and monomers such as acrylic acid. The membrane surfaces have
also been grafted or coated with polyacrylamide, poly(acrylic acid) poly (vinyl alcohol) and
cellulose derivatives. Another possibility for improving the membrane properties is the use of
polymer blends. Blends of PVDF/ PVP, PVDF/poly (ethylene glycol) (PEG), PVDF/sulfonated
polystyren, PVDF/poly (vinyl acetate) and PVDF/poly (methyl methacrylate) have been used in
149

the preparation of microporous membranes.

-Z

JU

Fig. 6.8. Poly(vinylidene fluoride) membrane: (top) cross-section and (bottom) surface.

Polyetherimide (PEI)

N
G
LM

Polyetherimide (PEI) is an amorphous polymer with Tg near 200 C. It can be used at

temperatures higher than PVDF and it is known for its superior strength. The chemical stability,
although much higher than cellulosic polymers, is lower than that of PVDF. PEI can not be used in
contact with chloroform and dichloromethane. It is also attacked by tetrahydrofurane. The stability
at high pH is poorer than that of PVDF, PSU or PAN. The preparation of membranes from PEI
solutions leads to a large variety of porous asymmetric structures, which can be controlled by
changing the composition of the solvent mixture. The commercial polymer usually chosen is

JI

Ultem1000, manufactured by General Electric. Integral asymmetric membranes are quite

successful in gas separation, particularly in the case of helium recovery. More open asymmetric
PEI supports have been used for ultrafiltration or as support for composite membranes. Usually a
porosity higher than that of PVDF membranes is obtained with smaller average pore size as shown
in Fig. 9. Porous PEI membranes with a dense and thin top layer for for gas separation were
initially prepared from a solution in a mixture of dichloromethane, 1,1,2-trichloroethane, xylene
and acetic acid and coagulated in acetone. Another solvent mixture later allowed the preparation
of PEI membranes for gas separation coagulated in water. For that a mixture of tetrahydrofurane
(THF) and gammabutyrolactone (GBL) was used. Both THF and GBL are non-solvents for PEI,
but due to an effect of cosolvency a stable casting solution is obtained. Membranes with very thin
top layers are particularly formed with higher GBL contents. The addition of volatile non-solvents
such as butanol to the casting solution lead to the formation of even thinner top layers. After
coagulation in water a sponge-like structure is obtained. The preparation of hollow fibers from
PEI solutions in a mixture of NMPand GBLor DMAc and GBL is described by Kneifel and
Peinemann. Here the addition of GBL, a nonsolvent slightly increases the solution viscosity and
favors a sponge-like structure without finger-like cavities.
Blends of PEI with other polymers are also reported to improve membrane characteristics.
Blends of PEI and poly (ether sulfonamide) were obtained as an attempt to improve the membrane
hydrophilicity. In order to make both polymers compatible, 1,3-diamino propanol (DAP) was
150

N
G
LM

-Z

JU

added to the polymer solution. DAP reacts with PEI and increases the solution viscosity. As a
result, besides the compatibilization effect, the addition of DAP induces a sponge-like structure,
eliminating the finger-like cavities which are usually observed in membranes obtained from PEI
or PESA solutions in dimethyl acetamide. PEI is successfully used as a porous support for
composite membranes. However, in order to improve their resistance to compactation at extreme
conditions of high pressure and hydrocarbon atmosphere, an inorganic polymer was generated in
the casting solution by hydrolysis of alkoxy silanes and incorporated in the membrane structure.
Here again compatibility was necessary and was achieved with the introduction of amino silanes.
Also for the preparation of asymmetric porous membranes a blend of PEI polyimide with
phenylindane groups (Matrimid) was reported with the purpose of improving the PEI gas
permeation in hollow fibers.

Fig. 6.9. Poly(etherimide) membrane: (top) cross-section and (bottom) surface.

Polyacrylonitrile (PAN)

JI

Polyacrylonitrile (PAN) has been used in the preparation of UF-membranes for a long time due
to its superior resistance to hydrolysis and oxidation. PAN is highly crystalline and relatively
hydrophilic and is usually copolymerized with more hydrophilic monomers to improve
processability and to make it less brittle. Hollow fibers can be prepared from PAN dissolved in
nitric acid. Preparation of PAN membranes by phase inversion from solutions in DMAC, DMF or
NMP is also possible. An example is shown in Fig. 10. ASumitomo patent discloses the
preparation of membranes from copolymers containing 89 % acrylonitrile and 11 % ethyl acrylate
dissolved in DMF and formamide and coagulated in water. Amicroporous membrane is obtained.
In order to make the membranes suitable for reverse osmosis, they were submitted to a plasma
treatment in the presence of 4-vinyl pyridine.

Cellulose

Cellulose UF-membranes are used in applications where low fouling characteristics are required.
Cellulose has a very regular structure and is able to form strong intermolecular hydrogen-bonds
between the several hydroxy groups. As a result, cellulose is practically insoluble in almost all

151

N
G
LM

-Z

JU

solvents. The only exceptions are dilute solutions in DMAc or NMP with addition of lithium
chloride. Cellulose membranes are prepared by methods which basically involve precipitation
from a solution of chemically modified native cellulose (from cotton linters, etc). Until some years
ago the three main methods were celluphane, cuprophane and cuenophane. Celluphane
membranes are prepared by a viscose process, in which cellulose is regenerated from a cellulose
xanthate solution, as described in the US Pat 981368 and 991267. Cuprophane membranes are
prepared in a similar way, regenerating cellulose from its soluble cop- per complex formed by
reacting with ammoniacal copper sulphate, as described in US Pat 2067522. To obtain cuenophane
membranes, cellulose is regenerated after dissolving it in cupriethylen diamine. For regeneration
of cellulose from solution, a coagulation in strong alkali solutions is usually required. Today most
of the cellulose membranes are prepared by hydrolysis of asymmetric cellulose acetate membranes
also in strong alkali solutions. An alternative method for the preparation of cellulose membranes
has been recently proposed; acid hydrolysis of trimethyl silyl cellulose.

Fig. 10: Poly(acrylo nitrile) membrane: (top) cross-section and (bottom) surface.

6.4.4 Solvent resistant membranes for nano- and ultrafiltration

JI

Areason for the restrained application of membrane technology in the chemical industry, as
compared to other fields, is the availability of well established chemical resistant membranes,
which could work in harsh process conditions, eventually at extreme pH conditions or in processes
with organic solvents. Although several examples have been described in the patent literature in
the last decade and some are commercialized, reasonably low costs associated with a well
documented long term application are usually required to make them commercially attractive and
lower the risks of substituting conventional separation processes. The development of solvent
resistant membranes is a relevant research topic in the leading membrane companies. Some
solvent stable membranes are discussed here and resumed in Tab. 5. Membrane Products Kiryat
Weizmann Ltd developed the SelRO nanofiltration membranes with excellent solvent resistance
and are now commercialized by Koch Membrane Systems. The MPS-44, 50 and 60 are claimed to
have an excellent stability in alkanes, alcohols, acetates, ketones and aprotic solvents. The MPS 44
is a hydrophilic mem- brane suitable, for instance, for separation processes in solvent mixtures
containing water and organics. Solutes with molecular weights 250 g/ml can then be separated or
concentrated, while the composition of the solvent mixture does not change through the membrane.
The hydrophilic MPS 50 and MPS 60 are nanofiltration membranes for use in a pure organic
medium. Some applications include the recovery of antibiotics and pepitides from organic
solvents, recovery of catalysts from organic medium and recovery of hydrocarbons from cleaning
processes.
Tab. 5.: SelRO Nanofiltration Membranes.

152

N
G
LM

-Z

JU

Although the composition of the MPS-44, 50 and 60 membranes are not completely open, two
patents of Kiryat Weizmann describe interesting procedures for the preparation of solvent stable
membranes. In the former, porous polyacrylonitrile (PAN) membranes are crosslinked, for
instance, by immersing them in a solution containing 1% of metal alkoxides such as sodium
ethoxide or 10 % of NaOH. The membranes are heated at 110 C. UF-Membranes which are not
soluble or swellable in DMF, NMP or DMSO are then obtained. If a cut off in the range of
nanofiltration or reverse osmosis is required, the membrane is coated with a hydrophilic polymer,
which is later crosslinked, or with polyfunctional reactants, which react forming a crosslinked
coating. Some of the described coatings were based on polyethyleneimine and reactive dyes. In a
second patent, coatings of bromomethylated phenylene oxide were crosslinked with ammonia. An
earlier patent discloses the improvement of the solvent resistance of PAN membranes by a
reaction with hydroxylamine, followed by treatment with cyanuric chloride and NaOH. The
resulting membrane also has an improved resistance to compaction.
Carbon membranes are also described in the Kiryat Weizmann patent starting from the same
PAN membranes mentioned above. After immersion in an organic or inorganic base solution the
membrane is heated at 110130 C, below the glass transition temperature, to induce a partial
cross-linking and prevent plastic flow during heating at higher temperature. The membrane is then
heated further at 250 C for a few minutes and later in a non-reactive environment at 6001000 C
for carbonization.

Koch also commercializes SelRO membranes especially suitable for extreme pH conditions.
Some of them are listed in Tab. 6.
Such membranes have been applied in the separation of heavy metals from acids and highly
alcaline solutions, and in the recovery of alcaline solutions used in cleaning processes.

JI

D-Series nanofiltration membranes produced by Desal/Osmonics work at very low pH levels.

They have been used to recover heavy metals and clarify 35 % sulfuric acid feed streams or 25 %
phosphoric acid streams. They have also been applied to permeate boric acid and reject
radionuclides at a nuclear power station. Somicon is now announcing new membranes able to
withstand up to 15 % NaOH at 60C for nanofiltration (cut off 200-250 g/mol). The membranes
are being applied to alcaline degreaseing baths and in the food sector. Also membranes for pH
much below 1 with 90 wt % NaCl rejection are being supplied by Somicon /Nitto.
The ETNA membranes which were commercialized by Dow Danmark consisted of poly
(vinylidene fluoride) (PVDF) porous supports with cellulosic coatings. PVDF is soluble only in a
few organic solvents such as dimethyl acetamide. Cellulose is very stable in organic and aqueous
solvents. However, because of its low solubility, preparing cellulose membranes is not a trivial
task. Stengaard proposed the preparation of composite membranes by coating
chlorotrifluoroethylene/vinylidene
fluoride
(CTFE/VF)
or
PVDF
supports
with

153

JU

hydroxyethylcellulose and hydroxypropylcellulose, which are water soluble. In order to fixate

these materials on the membrane surfaces they have to be crosslinked and/or chemically bonded to
the support. PVDF and CTFE/VF copolymer can be made reactive in the following way. Under
highly alkaline conditions and elevated temperatures hydrogen fluoride/hydrogen chloride are set
free while reactive groups and double bonds are formed. The support is then treated with a
solution of hydroxyalkylcellulose in the presence of NaOH and a cross-linking agent such as
1,3-dichloro-2-propanol, which forms ether bonds with hydroxyalkylcellulose via the OH groups.
Astable hydrophilic layer is formed on the top of the PVDF support. The coating may just
hydrophilize the UF-membrane or close the pores bringing the cut-off in the range of
nanofiltration. An Etna membrane with 1000 molecular weight cut off was reportted to have a
pure water flux of 50100 l/m2 h at 40 C and was stable between pH 1 and 12 (Petersen).
Membranes of hydroxyalkylcellulose crosslinked with glutaraldehyde were discussed in the
literature as being chemical resistant with cut-offs around 600 g/mol.

N
G
LM

-Z

Tab. 6: pH Stable SelRO nanofiltration membranes

JI

Polyetherketons are engineering thermoplastics with an exceptional combination of heat and

chemical stability. However, their high insolubility in all common solvents, a property which
assures the successful application of the membrane in chemical processes, inhibits their
production by conventional solution casting methods. In order to bypass this difficulty, alternative
preparation methods have been reported in the patent literature. One possibility is the partial
sulfonation of PEEK with sulfuric acid and preparation of the membrane by coagulation from the
sulfuric acid solution. The final membrane characteristics are, however, affected since the
sulfonated PEEK is now soluble in common solvents. Furthermore, sulfonated PEEK swells in
aqueous soluti- on, which prejudices the performance of the membrane in water. Membranes have
been also prepared from blends of poly (aryl ether ketone) and poly(ether imide) (PEI) submitted
to solvent leaching (of PEI) to form the pores. Asymmetric membrane with low porosity is formed.
Ionics proposed the preparation of membranes by casting from solutions in strongly protic,
non-reactive acids such as methanesulfonic acid. Dow described the extrusion of PEEK with a
plasticizer, followed by coagulation in a non solvent bath and leaching of the plasticizer. Drawing
before, during and/or after leaching improves the flux of the membranes.
Poly(phenylene sulfide) (PPS) is another chemically stable polymer. Analogously to the
procedure for preparation of PEEK membrane, Dow proposed in a recent patent the preparation of
PPS membranes by extrusion followed by leaching. Polyimide membranes for ultrafiltration have
long been the subject of Nitto patents. Although unsoluble in many common solvents such as
alcohols, ketones, ethers and esters, solutions in dimethyl formamide may be manufactured into
membranes by casting/coagulation procedures. Polyimide membranes with high solvent resistance
are claimed by Bend Research, by casting from polyamic acid solutions.
154

Porous polybenzimidazole (PBI) membranes were also developed in an attempt to overcome the
temperature and chemical stability limitations of other membranes. Procedures for preparation of
PBI membranes by phase inversion are long known. The polymer can be dissolved in dimethyl
acetamide and coagulated in a bath containing a mixture of solvent and non-solvent to hinder the
excessive formation of finger-like cavities. Dense PBI membranes, treated with phosphoric acid
are now being considered for application in fuel cells.

-Z

JU

Polyphosphazenes are also valuable as a membrane material with high chemical stability. They
have been explored for ultra- and nanofiltration, pervaporation, gas separation and more recently
for application in fuel cells. Although commercial membranes are not available, UF phosphazene
membranes have been prepared by phase separation with cut offs between 70,000 and 500,000
g/mol and water fluxes up to 30 m3/m2 day MPa with mean levels of stability in acetone or hexane.
Nanofiltration phosphazene membranes on ceramic support have been successfully used for the
separation of organic dyes from isopropanol. The US Department of Energy, through its Office of
Industrial Technology is supporting a wide spectrum of research on polyphosphazene membranes
in the chemical and petrochemical industry. One topic of interest is, for instance, the separation of
water from hydrocarbons by pervaporation.

N
G
LM

A successful case also supported by the US Department of Energy was the treatment of waste
water containing pigments and solvents with a combined UF/RO process. The membrane system
was supplied by Zenon Environmental Systems. Ultrafiltration removed suspended solids and high
molecular weight particles and reverse osmosis removed smaller impurities, during the treatment
of waste water of PPG Industries, the worlds largest producer of automotive and industrial
coatings. The system reduced the amount of contaminated water requiring offsite disposal from
400,000 gallons to less than 20,000 gallons annually.

6.4.5 Membranes for microfiltration

Several of the polymers mentioned above for ultrafiltration can also be used for the preparation
of microfiltration membranes by phase inversion. However, for the range of pore size useful for
microfiltration, other procedures have also been successfully used for the preparation of
commercial membranes. Among these are stretching, tracketching, leaching and sintering. Some
commercially available MF membranes are shown in Tab. 7.

JI

Tab. 7: Some commercially available membranes for microfiltration.

Millipore

Chemical resistance:
1 Water solution pH 211
2 Alcanes, aromatic hydrocarbons, halogenated alcanes
3 Alcohols
155

4 Ketones, esters, organic acids

5 Concentrated strong acids and bases

JU

PALL

N
G
LM

-Z

Desal

JI

PALLGelman

156

Polypropylene and poly (tetrafluor ethylene)

Stretching is part of the preparation process both of the Celgard and the Gore-Tex membranes.
Cold drawing has been described already in 1969 for membrane preparation starting from
crystalline polymers. Another preparation method is solvent stretching, where the precursor film is
brought in contact with a swelling agent and stretched. The swelling agent is removed while the
film is maintained stretched to render the film microporous. Other processes use sequencial cold
and hot stretching steps.

N
G
LM

-Z

JU

The Celgard membrane is made of polypropylene, which is a low cost and quite inert polymer.
It is resistant under extreme pH conditions and is insoluble in most solvents at room temperature.
It swells however in very apolar solvents such as carbon tetrachloride. No solvent is required for
the preparation of the membrane. It involves the extrusion of PP films with high melt stress to
align the polymer chains and induce the formation of lamellar microcrystallites when cooling. The
film is then 50300% stretched just below the melting temperature. Under stress, the amorphous
phase between the crystallites deforms, giving rise to the slit-like pores of the Celgard membrane
(Fig. 11). The film is then cooled under tension. PP is highly hydrophobic and several surface
treatments have been proposed to improve the hydrophilicity of PPmembranes. Incorporation of
surfactants is used to make Celgard membranes more hydrophilic.

Fig. 11: Celgardmembrane.

JI

Another commercial membrane prepared by stretching is Gore-Tex (Fig. 12). The polymer here
is poly(tetrafluor ethylene), what makes the membrane extremely inert and thus convenient for
processing even harsh streams. Processing PTFE is only possible by paste extrusion. In paste
forming the polymer is mixed with a lubricant such as odorless mineral spirits naphtha or kerosene.
The lubricant component is removed by heating up to 327 C. Above this temperature, sintering
would lead to a dense PTFE film. After lubricant removal, the PTFE film is submitted to an
uniaxial or biaxial stretching, giving rise to an interconnected pore structure. The process was
proposed by Gore and the resulting porous film is today a successful product in the membrane and
textile industry. For uniaxial stretching the unsintered film from the paste extrusion isfed to a
machine with heating roles, where one role is driven faster than the previous one to input stress
and induce the pore formation. The difference in speed determines the amount of stretch.
Additionally in the Gore patent a biaxial stretching is performed using a pantograph.

157

Fig. 12: Goretex membrane.

N
G
LM

-Z

JU

A special characteristic of the Gore membrane is that, since PTFE is very hydrophobic, (liquid)
water must not be allowed to wet the membrane and its transport is hindered. On the other hand,
water vapor can freely pass through the micropores, making the film suitable for transpirating
impermeable cloths. However due to their inertness, PTFE membranes are also interesting during
the processing of aggressive streams. If the goal is the filtration of aqueous wastes the membranes
should be modified to beco-me hydrophilic. Asolution is disclosed in a later Gore patent, mixing
PTFE with silica and a surfactant to form the paste for the extrusion. To improve the adhesion of
the filler to the PTFE matrix, the membrane, in the last step of preparation, is heated with a
solution of dimethyl octadecyl chlorosilane in toluene.

Polycarbonate and poly (ethylene terephtalate)

JI

Dense films of polycarbonate or poly(ethylene terephtalate) can be transformed into porous

microfiltration membranes with very narrow pore size distribution (Fig. 13), by exposing them to
fission fragments from radioactive decay with subsequent etching in alcaline solutions. The
number of pores can be controlled by the length of exposure to the fission segments. The
maximum pore density is limited by the fact that membranes become excessively brittle and
radioactive at very high doses. The size of the pores is controlled by the etching conditions and the
length of time in the etching bath. The shape of the nuclear pores (cylindric or conic) is
determined by V, the relation between the rate of etching of the bulk membrane material and the
rate of etching of the material along the high energy particle track. For instance, in poly (ethylene
terephthalate) films tracked by argon ions, V= 10100 and the pores are less cylindrical than in
films tracked by xenon ions, for which V = 1001000.

Fig. 6.13. Poly(ethylene terephthalate) membrane.

158

6.5. Surface Modification of Membranes

Although membrane preparation has been reported concerning a large variety of polymers, it is
often not possible to combine the best characteristics needed for the application, using just one
polymer. It is frequently observed for instance, that polymers with the best solvent resistance or
those which provide the most convenient pore structure are too hydrophobic to have an acceptable
performance in the filtration of aqueous solutions. On the other hand, chemical modification of the
polymer chain prior to membrane formation usually drastically changes the resulting pore
structure. Therefore several procedures have been researched to chemically modify the surface of
a previously formed porous membrane in order to increase the hydrophilic character or to allow
functionalization and incorporation of polymer segments. These may improve selectivity, raise
biocompatibility or bond catalytically active groups for membrane reactors. Chemical
modification of polymer surfaces has been reviewed.

N
G
LM

-Z

JU

6.5.1. Chemical oxidation

In many cases hydrophilicity is the main goal of membrane surface modification. Surface
oxidation is the simplest way to attain this goal. Several methods are available, including glow and
corona discharge and surface flaming, which are quite old but still very effective methods
particularly in the case of polyolefins. Additionally one of the earliest surface treatments, still
effective, is the exposure to oxidizing chemicals such as chromic acid, nitric acid or potassium
permanganate, which lead to the formation of carbonyl, hydroxyl and carboxylic groups on the
surface of polyethylene, polypropylene and polyesters.
One procedure for the modification of polypropylene membranes consists of reacting the
membrane in an aqueous solution of heated potassium peroxy disulphate to produce
oxygen-centered radicals which are responsible for introducing hydroxyl groups. If the process is
performed in the presence of monomers such as acrylamide, grafting takes place.

JI

6.5.2 Plasma treatment

Plasma is a complex gaseous state of matter, consisting of free radicals, electrons, photons, ions,
etc. Plasma can be generated by continuous electrical discharge in either an inert or a reactive gas.
For membrane application, plasma can be used to improve the characteristics both of porous
supports and of polymer films for gas separation. Several examples have been reviewed by
Kramer et al.. Porous membranes can be submitted to plasma treatment to achieve the following
effects: (i) crosslinking of the top layer and reduction of pore size; (ii) introduction of functional
groups to the surface or (iii) grafting and deposition of a thin selective layer on a porous substrate.
In the former instance, plasma treatment with inert gases such as argon or helium leads to
ablation of the substrate material by the excited plasma molecules and then redeposition of the
substrate material as a highly crosslinked layer on the surface. If the time of exposure is limited, a
controlled reduction of pore size is obtained and a microfiltration membrane can be transformed
into an ultrafiltration or even reverse osmosis membrane. Argon plasma was used to reduce the
pores of polyethylene hollow fibers. It has been reported that helium plasma modifies the surface
of porous polyacrylonitrile membranes, making them suitable for reverse osmosis.

A second application is the incorporation of functional groups. Plasma treatment with air,
oxygen and water vapor introduces oxygen-containing functional groups on the surface. Nitrogen,
ammonia and alkyl amine plasmas introduce nitrogen-containing functional groups. Treatment of
PAN and polysulfone membranes with helium/water plasma, as reported by Belfort and Ulbricht,

159

JU

turns them hydrophilic and minimizes fouling. It is interesting to note that amino groups are for
instance interesting to bind heparin and retard blood coagulation in membranes used in medical
applications. Ammonia plasma was also used to improve flux and selectivity of UF-polysulfone
membranes. Nitrogen and oxygen plasma have been used to improve the hydrophilicity of poly
(vinyl chloride) membranes.
Hydrophilization can also be obtained by plasma-induced graft polymerization with the
incorporation of hydrophilic monomers. The plasma is then generated from gaseous organic
monomers which polymerize and eventually crosslink on the membrane surface. The conditions
may also be adjusted to lead to a dense selective polymer coating. One advantage of the plasma
treatment is the absence of solvents or any hazardous liquid. Aweak point is the complexity of the
coating, which is difficult to predict. Several examples are reported in the literature. Ulbricht and
Belfort polymerized hydroxy ethyl methacrylate on polysulfone membranes. Acrylic acid was
deposited on commercial membranes to improve solute rejection during the ultrafiltration of
bleach effluents.

N
G
LM

-Z

6.5.3. Classical organic reactions

The well-defined surface functionalization using classical organic reactions plays an important
role in membrane development. In order to be susceptible to reaction, the polymer chain should
contain double bonds, hydroxyl groups or benzene rings. An example is the modification of
polysulfone by reaction with different chemicals to increase hydrophilicity. The surface
modification of polysulfone membranes has been reported by several authors.

6.5.4. Polymer grafting

JI

The covalent bonding of polymer segments and chains to porous supports can be achieved by
polymer graft (Tab. 8), which is, to a great degree based on the free radical reaction of vinyl or
acrylic monomers. Reactive sites on the polymer support, usually in the form of unpaired electrons,
can be created by i) UV-irradiation in the presence of initiators such as benzophenone and the
chosen monomer; ii) reactive sites are also thermally created as result of the decomposition of
organic peroxides. The third alternative is iii) the generation of unpaired electrons by exposure to
high-energy radiation, such as gamma- or electron-irradiation.
Grafting of hydrophilic vinyl monomers such as hydroxyethyl methacrylate on polysulfone and
PAN membranes under UV exposure to make them less susceptible to foulinghas been described
in the literature.
Tab. 8: Different monomers and initiation methods for polymer grafting.

160

JU
-Z
N
G
LM

PE = polyethylene, LLDPE = linear low density polyethylene; PP = polypropylene, PS = polystyrene; PET = poly
(ethylene terephthalate); PSU = polysulfone; TFB = poly (tetrafluoro ethylene-hexafluoropropylene)

JI

The incorporation of positive charges has decreased the fouling susceptibility of membranes
even more effectively. This is the principle of the aromatic polyamide membrane series
commercialized by Hydranauts as low-fouling composite membranes (LFC). Cationic
charge-modified polyamide membranes are also commercially available from CUNO under the
trademark ZETAPOR. Pall Corp. sells cationic chargemodified polyamide membranes under the
trademark N66 POSIDYNE. There are different ways to make the membrane positively charged.
A patent from Millipore describes the surface modification of hydrophobic membranes by
contacting them with a solution of polyamine epichlorohydrin containing quaternary or ternary
ammonium functional groups and acrylate monomers which can polymerize under UV irradiation
onto the surface. The CUNO membrane is based on the procedure disclosed in the US Pat.
4473475 for modifying polyamide membranes. Here a charge modifying agent such as
tetraethylene pentamine is bonded to the hydrophilic sites of the membrane through a polyepoxide
crosslinking agent. The membrane surface modifying polymers, disclosed in the Pall patent, are
cationic, water-soluble, quaternary ammonium, thermosetting polymers such as the
epoxy-functional polyamine epichlorohydrin.

6.6. Gas Separation with Membranes

6.6.1. Introduction
The separation of gas mixtures with membranes has emerged from being a laboratory curiosity
to becoming a rapidly growing, commercially viable alternative to traditional methods of gas
separation within the last two decades. Membrane gas separation has become one of the most
significant new unit operations to emerge in the chemical industry in the last 25 years. The gas
separation membrane module business for 2000 is estimated at $ 125 million with an annual
161

growth rate of 8 %. Table 9 shows commercial applications and some of the major suppliers of
membrane gas separation units.
Tab. 9: Gas membrane applications and suppliers (adapted from Economics of gas separation membranes, R.W.
Spillman, in: Membranes separation technology. Principles and applications, Ed.: R.D. Noble, S.A. Stern, 1995,

N
G
LM

-Z

JU

Elsevier Science.

JI

Tab. 9 shows established applications in the field of membrane gas separation. One of the new
and currently small applications as shown in Tab. 9 is natural gas dehydration. Problems related to
this separation will be discussed in the last part of this chapter (basic process design
considerations). Besides the well-established applications there are a number of emerging
membrane gas separations. These are, for example, natural gas hydrocarbon dewpointing,
olefin/paraffin separation and separation of hydrocarbon isomeres. These will be addressed in the
following material section. The purpose of this chapter is to provide an overview of state of the art
and emerging materials for gas separation membranes, to give some key features of integral
asymmetric and composite membranes and finally to explain the influence of basic process
parameters.

Materials and transport mechanisms

Organic polymers are the dominating materials for gas separation membranes. Many polymers
exhibit a sufficient gas selectivity and they can be easily processed into membranes.
Palladium alloys are the only inorganic materials which are currently used for gas separation
(ultra-pure hydrogen generation) on a commercial scale. However, during the last decade
inorganic materials have been developed with exciting unmatched selectivities for certain gas
mixtures and some of the inorganic membranes described in the scientific literature seem to be on
the brink of commercialization. Table 10 shows relevant membrane materials for gas separation.
Tab. 10: Materials for gas separating membranes

162

Organic polymers

JU

Anumber of excellent books and reviews have been published on the subject of polymeric gas
separating membranes, which are recommended to the interested reader. It is the purpose of this
chapter to supply the reader with a basic background which is important in the understanding of
the transport mechanism of gases through polymers, to introduce those polymers which are
currently of commercial importance and finally to give an outlook on interesting developments in
this field.

-Z

Background

N
G
LM

The simplest model used to explain and predict gas permeation through non-porous polymers is
the solution-diffusion model. In this model it is assumed that the gas at the high pressure side of
the membrane dissolves in the polymer and diffuses down a concentration gradient to the low
pressure side, where the gas is desorbed. It is further assumed that sorption and desorption at the
interfaces is fast compared to the diffusion rate in the polymer. The gas phase on the high and low
pressure side is in equilibrium with the polymer interface. The combination of Henrys law
(solubility) and Ficks law (diffusion) leads the to the equation
(1)

which can be simplified to

(2)

JI

where D is the diffusion coefficient of the gas in the polymer, S is the gas solubility, pis the
pressure difference between the high and low pressure side, l is the membrane thickness and P is
the permeability coefficient1. As can be seen from (1) and (2) the permeability coefficient P is the
product of D (a kinetic term) and S (a thermodynamic term).
(3)

The selectivity of a polymer to gas A relative to another gas B can be expressed in terms of an
ideal selectivity AB defined by the relation

(4)

The ratio DA/DB can be viewed as mobility selectivity and the ratio SA/SB as solubility

163

N
G
LM

-Z

JU

selectivity. For a given gas pair mobility and solubility selectivity depend on the chemical and
physical properties of the polymeric material. Excellent reviews on relationships between polymer
structure and transport properties of gases have been given by Stern, Paul and Freeman. Some
general rules are useful for a first understanding. The diffusion coefficient D always decreases
with increasing size of the molecule. The extent of this decrease is generally dependant on the
flexibility of the polymer backbone. The more rigid the polymer structure the higher the mobility
selectivity will be for a given gas pair. The mobility selectivity is dominant for most glassy
polymers. Hence, the transport of smaller molecules is favored. On the other hand the solubility of
gases generally increases with molecular size, because the intermolecular forces between gas and
polymer increase. Most rubbers show a low mobility selectivity due to their flexible polymer
chain but their ability to separate gases is dominated by their solubility selectivity. Thus large
organic vapor molecules can permeate much faster through some rubbers than smaller gases like
oxygen or nitrogen. This is depicted for silicone rubber in Fig. 14.

Fig.6.14. Diffusion and solubility coefficients of different gases in silicone rubber at 30C vs. the critical volume.

JI

It can be seen that the diffusion coefficient of the large pentane molecule is 3.6 times smaller
than the diffusion coefficient of oxygen. However, the solubility of pentane is about 200 times
larger than the solubility of oxygen. This solubility selectivity outnumbers the reverse diffusion
selectivity. As a result silicone rubber is much more permeable for pentane than for oxygen.
Contrary to rubbers glassy polymers usually show a preferred permeability to smaller molecules.
The mobility selectivity is much higher than the reverse solubility selectivity. This is shown in Fig.
15 with the polyimide Matrimid (Ciba Geigy).
NOTE: The permeability coefficient is most commonly given in Barrer, defined as 10-10 cm3cm/cm2s cm Hg and
named after R. M. Barrer. The correspondend SI unit is mol m/m2 s Pa (1Barrer= 0.3310-15 mol m/m2 s Pa).

164

Fig. 15: Diffusion and solubility coefficients of different gases in Matrimid at 30C vs. the critical volume.

Polymers for commercial gas separation membranes

JU

During the last two decades dozens of new polymers have been described in the literature,
which have been developed for gas separation. The largest group among these are probably
polyimides. In spite of these efforts less than 10 polymers are used for industrial gas separations.
Nearly all of these are technical polymers developed for totally different applications. Designer
polymers were either too expensive or their advantage over commercial polymers were not
sensational enough or they did not show the expected performance in real applications. The latter
is especially true for modified (fluorinated) polyimides. Some of the 6FDA-based polyimides
showed tremendous separation abilities in the laboratory but failed in real life due to plastizisation
or physical aging. Table 11 gives a list of polymers which are of practical importance for gas
separation.

N
G
LM

-Z

Tab. 11: Gas permeabilities of gas separation polymers.

JI

Cellulose acetate, polysulfone and polyimides are by far the most important polymers for gas
separation membranes. When we look at the volume streams treated, the old-fashioned cellulose
acetate is probably still the dominating polymer. The company Kvaerner Process Systems alone,
which has acquired Grace Membrane Systems, has sold or operates membrane plants with
CA-membranes for carbon dioxide separation for a total stream of more than 5 Mio m3 (STP)/day.
The only polymer in Tab. 11, which is especially designed for gas separation, is the brominated
polycarbonate (tetra-bromopolycarbonate). It was shown by Paul and coworkers that this
relatively simple modification led to an impressive increase in the oxygen/nitrogen separation
factor without loss of permeability (see Fig. 16)

Fig. 6.16. Selectivity vs. permeability for O /N for various polycarbonates at 35C and 1 bar. (reproduced by
2

permission of Elsevier Scientific Publishers).

165

Ultra-high free volume polymers

N
G
LM

-Z

JU

Because the so-called ultra-high free volume polymers aroused much interest during the last 10
years, they will be briefly described in this introductory chapter. The publication of the physical
properties of poly (1-trimethylsilyl-1-propyne) (PTMSP) in 1983 aroused much interest in the
field of membrane research. Up to this time it has been believed that the rubbery poly (dimethyl
siloxane) has by far the highest gas permeability of all known polymers. Very surprisingly, the
glassy PTMSP showed gas permeabilities more than 10 times higher than PDMS. This could be
attributed to its very high excess-free volume and the interconnectivity of the free volume
elements. Since then a number of high free volume polymers exhibiting extraordinaryly high gas
permeabilities have been synthesized. Two of these are under extensive investigation and are
currently being studied for gas separation on a pilot scale. These are Du Pont s 2,2bistrifluoromethyl-4,5-difluoro-1,3-dioxole/tetrafluorethylene copolymer (Teflon AF 2400) and
poly (4-methyl-2-pentyne) (PMP). All three polymers PTMSP, PMP and Teflon AF2400 are
glassy with glass transition above 230C and have a very high fractional free volume (FFV).
Figure 6.17 shows the chemical structure and fractional free volume of these three polymers.

Fig. 6.17. Chemical structure and fractional free volume of PTMSP, PMP and Teflon AF24000.

Table 12 shows the oxygen permeability and oxygen/nitrogen selectivity of PTMSP, PMP and
Teflon AF2400 in comparison with other polymers.
Tab. 12: Oxygen permeability and oxygen/nitrogen selectivity of high free volume polymers in comparison with

JI

conventional polymers.

Although the PMP was already synthesized in 1982, its high gas permeabilities were first
published 1996 by Pinnau et al. at MTR, Menlo Park. MTR is currently evaluating the
performance of PMP membranes for hydrocarbon separation. An attractive application of
membranes in this field is natural gas hydrocarbon dewpointing. This means the separation of
higher hydrocarbons like butane present in natural gas from methane. Table 13 shows the
performance of PTMSP and PMP membranes for the separation of a methane/butane mixture.
Tab. 13: Mixed gas permeation properties of PTMSP and PMP. Feed: 2% butane in methane, feed pressure: 10 bar,
166

permeate pressure: atmospheric, temperature: 25 C From: I. Pinnau et al. In: Polymer membranes for gas and
vapor separation, ACS Symposium Series 733 (1999), 5667.

N
G
LM

-Z

JU

Although the permeability and selectivity of PMP is significantly lower compared to PTMSP ,
its performance for hydrocarbon separation is still superior to all other known polymers. The
advantage of PMP lies in its much better chemical stability. In contrast to PTMSP it is not soluble
in linear saturated hydrocarbons. With a mixed gasselectivity of butane over methane of 14, the
PMP shows a much better performance than poly (dimethyl-siloxane), for which a selectivity of
about 5 has been determined under similar conditions. PMP has a poor pure gas selectivity for
butane/methane separation, but in the presence of butane the methane permeability drops by a
factor of 5. This effect is well known for the gas permeation through nanoporous solids, the
condensable gas is selectively adsorbed on the pore walls, thus hindering the passage of the
smaller molecules. It is believed that in a similar way the extraordinarily high excess free volume
of the super glassy polymers can be occupied by condensable gases.
It will be quite interesting to observe which class of polymers will finally be applied for the
attractive field of hydrocarbon dewpointing of natural gas. The superglasses like PMPshow the
highest selectivities and fluxes. However, due to their double bonds their chemical stability
remains uncertain and they might be prone to physical aging, which is the irreversible absorption
of components with high boiling points. Rubbery polymers like PDMS on the other hand are
stable under natural gas conditions. However they loose selectivity under high partial pressure of
higher hydrocarbons.

JI

The third polymer listed in figure 4 has a very different structure in comparison with the
polyacetylenes. The Teflon AF2400 is a perfluorinated random copolymer composed of 13 mol %
tetrafluoroethylene and 87 mol % 2,2-bis(trifluoromethyl)-4,5-difluoro- 1,3-dioxole. Its
extraordinarily high gas permeability was first described by Nemser and Roman. Composite
membranes fabricated from this polymer are currently being tested on a pilot scale by Compact
Membrane Systems, Wilmington. An attractive application seems to be the production of
oxygen-enriched or oxygen-depleted air for mobile diesel engines and the separation of
supercritical carbondioxide.

Inorganic materials for gas separation membranes

The current market for inorganic membranes for gas separation is extremely small. One of the
few commercial applications are small scale palladium membrane systems to produce ultrapure
hydrogen for specialized applications. They are marketed by Johnson Matthey and Company. It is
not believed that the market share of inorganic membranes will increase significantly in the near
future. The main obstacle is their high price and some principle difficulties during reproducible
large-scale production. On the other hand fascinating research results have been published in the
recent past such as a propene/ propane mixed gas selectivity of more than 40 with carbon
molecular sieve membranes or unmatched selectivities for carbondioxide/methane separation with
ceramic membranes. The interested reader is referred to some good reviews on inorganic
membrane materials.
167

In this chapter three examples of inorganic membranes will be discussed, which might find
industrial applications in the future.
6.6.2. Nanoporous carbon membranes
Nanoporous carbon membranes were one of the highlights in membrane development for gas
separation during the last decade. They can be produced by different methods. The most advanced
membranes of this kind have been produced by Air Products first published 1993. Air Products
called this membrane Selective Surface Flow (SSF) membrane.

-Z

JU

It consists of a thin layer (23m) of a nanoporous carbon matrix (57 pore diameter)
supported on the bore side of a macroporous (< 1m pore diameter) alumina tube. The membrane
is produced by: (a) coating the bore side of the tubular support with a thin uniform layer of a
poly(vinylidene chloride-acrylate) terpolymer latex containing small polymer beads in aqueous
emulsion, (b) drying the coat under N2 at 50 C, (c) heating under a dry N2 purge to 600 C for
carbonizing the polymer, and (d) finally passivating the nascent carbon film by heating in an
oxidizing atmosphere at 200300 C. The resulting membranes have quite defined pores in the 5
to 8 range. They are especially well suited for the separation of hydrogen/hydrocarbon mixtures.
The remarkable point is that they exhibit a preferred permeability for higher hydrocarbons over
hydrogen. Table 14 gives some permeability data from the original paper.

N
G
LM

Table 14: Gas separation properties of nanoporous carbon membrane.

JI

As can be seen from the table the pure gas selectivities of the nanoporous carbon membrane are
quite low, e.g. 1.19 for butane/hydrogen. However, for the mixture given in Table 14 the
butane/hydrogen selectivity increases to 94. The reason for this is that the butane is selectively
absorbed over hydrogen at the carbon pore wall and because the pores are so small the pathway
for hydrogen is blocked. This effect of selective surface flow and pore blocking was first observed
by Barrer and coworkers. Due to its unmatched selectivity the nanoporous carbon membrane looks
very attractive for hydrogen enrichment of refinery off-gases with low hydrogen content, e.g. FCC
(fluidized catalytic cracker) off-gases. It is much more attractive than hydrogen sele membranes,
because the hydrogen remains on the high-pressure side of the membrane and can be fed into a
pressure swing unit for further purification. The drawback of the nanoporous carbon membrane is
that water vapor and pentane and higher hydrocarbons should be removed before the membrane
separation because they absorb very strongly in the membrane pores. Air Products
SSF-membrane is now being field-tested at different refinery sites.

6.6.3. Perovskite-type oxide membranes for oxygen separation

It has been known for a long time that certain dense ceramic materials are good conductors of
oxygen at elevated temperatures. Oxygen transport through an ionic conductor is a result of
oxygen ionic conduction mechanisms that involve oxygen defects such as lattice vacancies. One
of the best known ceramic oxygen conductors is yttriadoped zirconia, which is widely used in high
168

N
G
LM

-Z

JU

temperature oxygen sensors. Oxygen is transported through these materials as O2- ion. Hence,
when oxygen permeates through these materials there must be a flow of electrons in the opposite
direction. Oxygen conducting ceramics like doped zirconia are good oxygen conductors but poor
electronic conductors. The electronic conductivity of yttria stabilized zirconia is three to four
orders of magnitude lower than its ionic conductivity. The oxygen can be pumped through the
material by an external electrical field. However, a simple calculation reveals that this is not
economic for oxygen separation due to the high electricity consumption. The situation changes
when the ceramic material is a good conductor for both oxygen ions and electrons. These
materials are referred to as ionic-electronic mixed conductors. With these a high oxygen flux can
be obtained without an external electrical field. Fig. 6.18 shows schematically the two types of
oxygen ion transport membranes.

Fig. 6.18. Types of oxygen ion transport membranes.

One of the first papers which stimulated large interest in this research was published in 1985 by
Teraoka et al. One of the figures of this pioneering paper is given here (Fig. 6.19).

JI

It may be deduced from Fig. 19 that, with a specific composition of the ceramic, an oxygen flux
of up to 2.4 cm3/min cm2 could be obtained at a oxygen partial pressure difference across the
membrane of 0.21 bar. The membrane thickness was 1 mm. If one uses these data to calculate the
permeability a tremendous value of 2.5 million Barrer is obtained. This is, of course, not totally
correct because the oxygen flux through these membranes is not direct proportional to the oxygen
partial pressure difference and is also not inversely proportional to the membrane thickness. At
layers below 0.3 mm surface reactions become rate limiting. But even 0.5 mm thick membranes
exhibit an oxygen flux at high temperatures which is orders of magnitude higher than the flux
through the best polymeric membranes. The promising prospect of these membranes is not in the
first place the production of oxygen, but their application in membrane reactors for the partial
oxidation of natural gas, which is schematically shown in Fig. 6.20.

169

JU

N
G
LM

-Z

Fig. 19: Temperature dependence of the rate of oxygen permeation through perovskite membranes.

Fig. 6.20. Ion transport membrane mediated partial oxidation of methane.

JI

The mixed conducting membrane eliminates the cryogenic air separation plant and it forms a
safety barrier between the natural gas and air. The membrane becomes more productive in a
configuration like this, because the slow oxygen desorption at the permeate side is enhanced by
the chemical reaction. The big challenge for the future is the economic scale up of membrane
production and the gas tight potting for temperatures above 700 C.
The perovskite type ceramic membranes have attracted much attention from major chemical
and petrochemical companies in the USA. Companies currently involved in the development of
the mixed conducting ceramic membranes include AirProducts, Praxair, BP and Amoco. Many
believe that the mixed conducting membrane technology will represent a major breakthrough in
industrial application of inorganic membranes.
6.6.4. Mixed matrix membranes
Molecular sieves such as zeolites or carbon molecular sieves show a much higher selectivity for
many gas mixtures than polymeric membranes due to their very defined pore sizes. For example it
can be calculated from reported sorption and diffusivity data that zeolite 4A has an
O2-permeability of 0.77 Barrer and an O2/N2 selectivity of approximately 37 at 35C. The
preparation of defect-free zeolite layers on a large scale is extremely difficult and it seems
170

-Z

JU

doubtful that this will ever be achieved at a competitive price. However, the combination of the
superior gas selectivities of molecular sieves with the processibility of polymeric membranes have
attracted many researchers. The hybrid membranes consisting of inorganic molecular sieves and
polymers are often referred to as mixed matrix membranes.
It was believed for a long time that the incorporation of a molecular sieve in a polymer matrix
does not change the polymer selectivity under steady state conditions. Hennepe, from the
university of Twente, proved for the first time that the incorporation of silicalite in PDMS
increased the ethanol/water selectivity significantly under steady state conditions in pervaporation
experiments. Later it was shown by Jia et al. that using the same approach (silicalite in PDMS) the
gas selectivity could be also changed due to a molecular sieving effect. However, the effects were
too small to be of any interest for practical applications. One of the fundamental questions of this
concept is how the permeability of the polymer should match with the molecular sieve
permeability. Koros proposed to use the following equation for an ideal modelling of the mixed
matrix permeability:

N
G
LM

where Peff is the effective permeability, P is the volume fraction of the dispersed phase, and the
subscripts d and c refer to the dispersed and continuous phases, respectively. This equation was
first derived by Maxwell to calculate the electric conductivity of a metal in which small spheres of
a second metal are dispersed. If this equation is used to calculate the selectivity of a mixed matrix
membrane we obtained

where c is the selectivity of the continuous phase (the polymer) and the indices 1 and 2 refer to
gas 1 and 2.

JI

This equation looks complex at first sight but it allows the plotting of the selectivity of a mixed
matrix membrane versus the polymer permeability, when the other parameters are fixed. An
example is given in Fig. 6.21. In this example a 4A zeolite with an estimated O2/N2 selectivity of
37 and an O2 permeability of 0.77 Barrer has been dispersed in polymer matrix with a O2/N2
selectivity of 7. The volume fraction of the zeolite is 0.60. The plot reveals that a maximum
selectivity of the mixed matrix membrane is obtained in this case, when the polymer permeability
is slightly higher than the zeolite permeability. When the polymer permeability becomes too high
the selectivity of the mixed matrix membrane approaches the polymer selectivity. Hence the above
equation gives a theoretical estimation of the selectivity of a mixed matrix membrane and it gives
an idea of how the permeability of molecular sieve and polymer should match. The practical
challenge is to improve the compatibility between inorganic molecular sieves and glassy polymers
in order to eliminate gas diffusion pathways at the interface between polymer and zeolite. Once
this problem is solved using nano-sized zeolite crystals the mixed matrix membranes might
become an attractive alternative to polymeric membranes for commercial gas separation.

171

Fig. 6.21. Selectivity of a mixed matrix membrane vs. polymer permeability, zeolite O /N selectivity 37, zeolite
2

JU

oxygen permeability 0.77 Barrer, polymer selectivity 7, zeolite volume fraction 0.60.

N
G
LM

-Z

6.6.5. Process design

In this chapter some fundamental equations are given, which allow a first design of a one stage
membrane separation unit. Questions to be answered are: what is the maximum enrichment, which
can be achieved with a membrane of a given selectivity? How is the separation performance
influenced by feed and permeate pressure? It will be explained why for some applications a high
selective membrane will be outperformed by a membrane with a lower selectivity.

JI

For the following we look at a simple gas separation unit with two components, which is
illustrated in Fig. 22. One parameter, which does of course determine the gas enrichment, is the
membrane selectivity, which is a membrane property and defined here as = P2/P1with P2 and
P1 as permeability coefficients for gas 2 and 1. Equally important are two process parameters the
stage cutand the pressure ratio. The stage cut is defined as ratio permeate flow/feed flow and
the pressure ratio is the ratio of total feed pressure to total permeate pressure. For the sake of
simplicity we start with a stage cut close to zero, i.e. there is no concentration difference between
feed and retentate. The maximum enrichment of the faster component 2 can now determined
easily. For the maximum enrichment the maximum driving force is needed, i.e. the permeate
pressure can be neglected when compared to the feed pressure. The flux of component 1 is
proportional to its volume fraction on the feed side, for component 2 we have, as an additional
factor the membrane selectivity.

Fig. 6.22. Model gas separation unit with two components.

172

The concentration of component 2 on the permeate side must be equal to the flux J2 of
component 2 divided by the total flux J1 + J2. Combining equations 1 and 2 then yields

(3)

This simple equation gives the maximum possible enrichment of one gas of a two component

JU

mixture when separated by a membrane with a selectivity of . The equation becomes more
complex when the permeate pressure cannot be neglected. Following the simple solution-diffusion
model the gas fluxes for gas 1 and 2 through the membrane are

and

N
G
LM

(5)

-Z

(4)

where P1 and P2 are the permeabilities of component s 1 and 2, l is the membrane thickness, and p1',
p2'' and p1'', p2'' are the partial pressures of the two gases in the feed and permeate streams,
respectively. The total gas pressure is equal to the sum of the partial pressures, i.e.

(10)

JI

Combining equations (49) then yields the expression

Equation 10 gives the concentration of the faster permeating gas in the permeate stream as a
function of the membranes selectivity and the pressure ratio across the membrane. It breaks down
into two limiting cases. At high driving forces when the pressure ratio is much higher than
selectivity (>>) equation 10 reduces to equation 3. We call this a selectivity controlled region,
because the enrichment is now independent of the pressure ratio. When on the other hand, the
pressure ratio becomes much smaller than the selectivity (<<), equation 10 reduces to
(11)

The enrichment is now independent of the membranes selectivity. Hence, this is the pressure
ratio limited region. There is, of course, an intermediate region between these two limiting cases
where both the presure ratio and the membrane selectivity affect the degree of separation. This is
illustrated in Fig. 6.23, in which the calculated permeate concentration is plotted versus pressure
173

ratio for a membrane with a selectivity of 200.

JU

For numerous technical applications the pressure ratio does not exceed 10 or 20. An example is
the separation of organic vapors, where a typical pressure ratio is about 10. Fig. 24 shows a plot of
permeate concentration versus selectivity at a pressure ratio of 10. The plot reveals that an incrase
of the selectivity above 40 does not increase the enrichment significantly. The highest permeate
concentration achievable in this example is 5% as predicted by equation 11.

-Z

Fig. 6.23. Calculated permeate concentration for a membrane with a selectivity of 200 as a function of pressure
ratio. The feed concentration is 0.5 %.

JI

N
G
LM

The process might not only benefit from a high selective membrane, but a too selective
membrane might even be a disadvantage. This will be demonstrated with a simple model
calculation concerning natural gas dehydration. Natural gas dehydration is one of the emerging
applications of membrane-based gas separation. The permeation of water vapor through polymers
has one peculiarity. When plotting gas permeability versus selectivity for different polymers a
general tendency exists: the more selective a polymer is the lower is its permeability, as shown in
the famous Robeson plots. Fig. 25 displays a plot of water vapor/nitrogen selectivity versus water
permeability for a number of polymers. The afore-mentioned tendency does not hold here. On the
contrary, many very selective polymers also have a very high permeability1. Numerous polymers
are available with water vapor selectivities of 5000 and more.

Fig. 6.24. calculated permeate concentration versus membrane selectivity for a pressure ratio of 10,

feed concentration 0.5 %.

174

JU

Fig. 6.25. H2O/N2-selectivity versus permeability for various polymers.

NOTE 1: When looking at the permeabilities in Fig.6.25 one has to keep in mind that the water vapor permeability

-Z

often strongly depends on its vapor pressure. Most permeability data have been generated near saturation pressure.

N
G
LM

For the following calculation a two component water vapor/methane mixture has been assumed
with 0.2 % water vapor. The membrane unit shall reduce this water content by one order of
magnitude; i.e. the retentate water concentration has been set to 0.02 %. The simple equation 10
cannot be applied, the equations first derived by Weller and Steiner have been used for the
calculation of methane loss and the equation of Saltonstall for calculation of membrane area. The
methane loss is simply defined by methane permeate stream divided by methane feed stream times
100. In Fig. 6.26 the methane loss is plotted versus membrane selectivity for two different
pressure ratios.

JI

The figure reveals that the methane loss is smaller at the higher pressure ratio. For the pressure
ratio of 80 methane loss starts at 5% for a selectivity of 100 and drops to 2,75 % for a selectivity
of 5000. But even with a very selective membrane it is never smaller than 2,7 %. At the high
pressure ratio of 1000 the methane loss drops to a favorable 0.17 %. However, a pressure ratio of
1000 is quite unrealistic. Additional compression of the feed is out of question. The permeate
pressure could be reduced by vacuum pumps; but this idea is not liked by the gas companies
because of cost and safety concerns. If a membrane dehydration system is operated at a pressure
ratio of 80, the difference between methane loss from a membrane with a selectivity of 500 (3.2 %
methane loss) and a membrane with a selectivity of 5000 (2.75 % methane loss) is quite small.
Hence, as far as methane loss is concerned there is a small benefit in using the high-selective
membrane. The picture changes when the required membrane area is taken into account, which is
demonstrated in Fig. 6.27. Membrane area has been calculated for a feed stream of 1000 m3
(STP)/h, the water vapor flux through the membrane has been fixed at 1.110-2cm3/cm2 s cmHg.
Membrane selectivity was adjusted by variation of methane flux.
As the figure shows, the membrane area required increases strongly with selectivity. There is an
8 x increase of membrane area for a selectivity of 500 and 5000. This example illustrates that it is
not always a good strategy to look for the most selective membrane but that a less selective
membrane may do a better job.

175

N
G
LM

-Z

JU

Fig. 6.26. Methane loss vs selectivity for pressure ratios 80 and 1000, volume fraction of methane in feed 0.2 %.

Fig. 6.27. Membrane area versus selectivity at pressure ratios of 80 and 1000, water vapor flux fixed at 1.110-2

JI

cm3/cm2 s cmHg, water vapor feed concentration 0.2 %, retentate concentration 0.02 %, feed pressure 8 MPa.

176