Polyimides and Other High Temperature Polymers - Synthesis Characterization and Applications

Because of their many desirable attributes, polyimides
constitute an important class of materials and they have

found applications in many and varied arenas ranging from
microelectronics to aerospace to membrane technology to
adhesives.
The book is divided into four parts as follows: Part 1:

Polyimides and
Synthesis, Properties and Bulk Characterization; Part 2:
Surface and Adhesion Aspects; Part 3: Applications; Part
4: General Papers. The topics covered include: synthesis of Other High
various kinds of polyimides with different characteristics;
structure-property relationships; photo-definable polyimi-
des and other materials; space-survivability of polyimides; Temperature
Polyimides and Other High Temperature Polymers:

Synthesis, Characterization and Applications,
heavy ion irradiation of polyimides; surface modification
and adhesion aspects of polyimides; metallization of po-
lyimides; polyimide foam nanocomposites; ion-containing
polyimides; zeolite modified polyimides; cobalt-containing
Polymers:
polyimides; application of polyimides in aerospace, micro-
electronics, purification of bio-fuels, membrane techno-
logy, moisture absorption and sensing, low Earth orbit, Synthesis,
and others; organosilicon films and lifetime assessment of
high-temperature fibers.
This book represents the cumulative wisdom of many

Characterization
world-renowned scientists and technologists and reflects
the current state-of-the-knowledge in the domain of po-
lyimides and other high temperature polymers. All signals
and Applications,
indicate that interest in these materials will continue
unabated and as their new and improved versions are deve-
loped, more new applications will emerge. Volume 5
In essence, this book provides a repository of current
information on this topic and should be of interest and
Edited by
value to anyone interested in polyimides and other high
temperature polymers.
K.L. Mittal
K.L. Mittal
Volume 5
(Ed.)
9 789004 170803
VSP
brill.nl A.S.
POHTPS5.indd 1 18-03-2009 14:12:57

Polyimides and Other High-Temperature Polymers:
Synthesis, Characterization and Applications
Volume 5
This page intentionally left blank
Polyimides and Other
High-Temperature Polymers:
Synthesis, Characterization
and Applications
Volume 5
Edited by
K. L. Mittal
LEIDEN • BOSTON
2009
CRC Press
Taylor & Francis Group
6000 Broken Sound Parkway NW, Suite 300
Boca Raton, FL 33487-2742
© 2009 by Koninklijke Brill NV Leiden The Netherlands
CRC Press is an imprint of Taylor & Francis Group, an Informa business
No claim to original U.S. Government works

Version Date: 20120525
International Standard Book Number-13: 978-9-04-742411-6 (eBook - PDF)
This book contains information obtained from authentic and highly regarded sources. Reasonable efforts
have been made to publish reliable data and information, but the author and publisher cannot assume
responsibility for the validity of all materials or the consequences of their use. The authors and publishers
have attempted to trace the copyright holders of all material reproduced in this publication and apologize
to copyright holders if permission to publish in this form has not been obtained. If any copyright material
has not been acknowledged please write and let us know so we may rectify in any future reprint.
Except as permitted under U.S. Copyright Law, no part of this book may be reprinted, reproduced, trans-
mitted, or utilized in any form by any electronic, mechanical, or other means, now known or hereafter
invented, including photocopying, microfilming, and recording, or in any information storage or retrieval
system, without written permission from the publishers.
For permission to photocopy or use material electronically from this work, please access www.copyright.
com (http://www.copyright.com/) or contact the Copyright Clearance Center, Inc. (CCC), 222 Rosewood
Drive, Danvers, MA 01923, 978-750-8400. CCC is a not-for-profit organization that provides licenses and
registration for a variety of users. For organizations that have been granted a photocopy license by the
CCC, a separate system of payment has been arranged.
Trademark Notice: Product or corporate names may be trademarks or registered trademarks, and are
used only for identification and explanation without intent to infringe.
Visit the Taylor & Francis Web site at
http://www.taylorandfrancis.com
and the CRC Press Web site at
http://www.crcpress.com
Contents
Preface ix
Part 1: Synthesis, Properties and Bulk Characterization
Preparation, Characterization and Selective Solubility of Diels–Alder
Addition Polyimides
Y.-L. Liu and S.-Y. Chen 3
Soluble Polyimides Based on Aromatic Diamines Bearing Long-Chain
Alkyl Groups
Y. Tsuda 17
Synthesis and Properties of Novel Bismaleimides, Polyaspartimides and
Organosoluble Polyimides
M. Sarojadevi, R. Hariharan and N. Amutha 43
Synthesis, Characterization and Polymerization of Imide-Containing
Phthalonitrile Polymers
P. Selvakumar, K. Padmini and M. Sarojadevi 71
Structure–Property Relationships in Processable Heat Resistant
Poly(amide–imide–imide)s for High-Temperature Applications
M. Sarojadevi and S. Bhuvana 93
Structural Characterization and Mechanical and Thermal Properties of
New Cyanate/Epoxy and Cyanate/BMI Blends and Composites
M. Sarojadevi, L. S. Jayakumari, V. Thulasiraman, G. Anuradha and
P. S. Sampath 119
Rigid-Rod Polybenzimidazoles (PBIs): A Concise Review of Their
Synthesis, Properties, Processing and Applications
T. D. Dang, N. Venkat and J. E. Mark 145
A Novel Negative-Working Photodefinable Poly(benzoxazole) Using
Acid Catalyzed Cross-linking
T. Minegishi and N. Yamazaki 171
Low Temperature Curable, Positive-Tone, Photo-Definable
Poly(benzoxazole)
M. Ohe, K. Katoh, T. Kawamura, H. Matsutani and T. Hattori 181
Multiple Relaxation Processes in Swift Heavy Ion Irradiated Kapton-H
Polyimide
J. K. Quamara, M. Garg and G. Goyal 193
vi Contents
TSDC Spectroscopy Investigation of Polyetherimide

B. K. Kaushik, G. Goyal and R. Singh 213
Space-Survivability and Characterization of Main-Chain and Side-Chain
POSS–Kapton® Polyimides
S. J. Tomczak, M. E. Wright, A. J. Guenthner, B. J. Petteys, T. K. Minton,
A. Brunsvold, V. Vij, L. M. McGrath and J. M. Mabry 227
Part 2: Surface and Adhesion Aspects

Fluorination of PMDA-ODA Polyimide Using Hyperthermal Atomic
Fluorine Beams
M. Tagawa, K. Yokota, K.-i. Maeda, A. Yoshigoe, Y. Teraoka,
K. Akamatsu and H. Nawafune 249
Preparation of Highly Reflective and Conductive PI/Ag Composite Films
by an Ion-Exchange Self-Metallization Technique
S. Qi, Z. Wu, D. Wu and F. W. Harris 261
Adhesion and Tribological Properties of Sputtered Polymer Thin Films
from a Polyimide Target
S. Iwamori, A. Uemura and K. Kezuka 275
Adhesion and Mechanical Properties of Poly(tetrafluoroethylene)–
Poly(vinyl alcohol) Composite Thin Films Prepared
by Spin Coating Method
S. Iwamori, M. Mizoguchi and I. Nishiyama 285
Part 3: Applications
Polyimide–Aerogel Hybrid Foam Composites for Advanced
Applications
T. M. Smith, M. K. Williams, J. E. Fesmire, J. P. Sass and E. S. Weiser 295
Polyimide Foam Nanocomposites from Powder Precursors
C. R. Misiego, C. I. Cano and R. B. Pipes 307
High-Performance Composite Membranes Based on Sulfonated
Polyarylenethioethersulfone and Sulfonated Polybenzimidazole
for Fuel Cell Applications
Z. Bai, S. B. Juhl, N. Venkat and T. D. Dang 323
New Ion-Containing Polyimides for the Purification of Bio-fuels by
a Membrane Separation Process
A. Jonquières, M. Awkal, R. Clément and P. Lochon 339
Zeolite-Modified Polyimides as Absorption Materials for Molecular
Contamination
S. Virji, S. Halper, J. Ying, R. Anderson and R. Villahermosa 353
Contents vii
Cobalt-Containing Polyimides for Moisture Absorption and Sensing

S. R. Halper, S. Virji and R. M. Villahermosa 369
Part 4: General Papers
Durability of Si-Containing Polyimide in a Simulated Atomic Oxygen
Environment for Low Earth Orbit Applications
K. Yokota, M. Tagawa, M. Iwata, E. Miyazaki, J. Ishizawa, M. Suzuki
and R. Yokota 379
Thermo-Chemical Behaviour of Organosilicon Films Deposited by
a Cold Plasma Polymerization Process
B. Mutel and P. Supiot 391
Aspects of Lifetime Assessment of High-Temperature Fibers and
Lifetime Enhancement
T. Bahners, U. Schloßer, T. Textor and E. Schollmeyer 409
Preface
This volume chronicles the proceedings of the Fifth International Symposium

on Polyimides and Other High-Temperature Polymers: Synthesis, Characterization
and Applications held under the aegis of MST Conferences in Orlando, FL, No-
vember 5–7, 2007. The earlier four symposia in this series were also organized by
MST Conferences in 1999, 2001, 2003 and 2005, respectively, the proceedings of
which were properly documented in the form of hard-bound books [1–4].
For quite some time there has been considerable interest and brisk R&D activity
in the arena of polyimides and other high-temperature polymers. This is ascribed
to the fact that these materials have found application for a variety of purposes
in a host of technologies and industries, ranging from microelectronics to adhe-
sives to nanocomposites to membranes. Also currently there is a high tempo of
R&D activity in molecular tailoring of new and improved materials to overcome
certain deficiencies and drawbacks of existing materials. As novel and improved
high-temperature materials are developed, more new and exciting application vis-
tas will emerge.
This symposium was arranged with tho express intent of providing a forum
for discussion of latest developments in this field. The technical program of the
symposium contained 46 papers reflecting both overviews and original research
contributions. The presenters represented many and varied disciplines and organi-
zations and hailed from many corners of the globe. The various aspects (synthesis,
characterization and applications with emphasis on new and exciting developments)
of such materials were accorded due coverage.
Now turning to this volume it contains 25 papers, others are not included for a va-
riety of reasons, which were properly reviewed, revised (some twice or thrice) and
edited before acceptance and inclusion in this volume. So it should be recorded that
this book is not a mere collection of papers — which is normally the case with many
proceedings volumes — rather it represents the highest standard of publication and
is of archival value. The book is divided into four parts as follows: Part 1: Synthesis,
Properties and Bulk Characterization; Part 2: Surface and Adhesion Aspects; Part 3:
Applications; and Part 4: General Papers. The topics covered include: synthesis of
various kinds of polyimides with different characteristics; structure–property rela-
Polyimides and Other High-Temperature Polymers: Synthesis, Characterization and Applications, Vol. 5
© Koninklijke Brill NV, Leiden, 2009
x Preface
tionships; photo-definable polyimides and other materials; space-survivability of

polyimides; heavy ion irradiation of polyimides; surface modification and adhe-
sion aspects of polyimides; metallization of polyimides; polyimide foam nanocom-
posites; ion-containing polyimides; zeolite-modified polyimides; cobalt-containing
polyimides; application of polyimides in aerospace, microelectronics, purification
of bio-fuels, membrane technology, moisture absorption and sensing, low Earth or-
bit, and others; organosilicon films; and life-time assessment of high-temperature
fibers.
This volume and its predecessors [1–4] containing about 2200 pages provide
a repository of current information and reflect the cumulative wisdom of a legion
of researchers in this exciting and technologically highly important arena. All sig-
nals indicate that the interest in these materials will continue unabated — rather
will intensify — as their novel variants with improved characteristics will find new
applications. Also the information covered in this set of five volumes should help
spur further activity in this field. In essence, this and previous volumes in this series
should be of interest and value to anyone with a desire or need to learn the latest
regarding this fascinating class of materials.
Acknowledgements
First, as always, it is a pleasure to extend my sincere thanks to my colleague and
friend, Dr. Robert H. Lacombe, for taking care of the organizational aspects of this
symposium. Thanks are due to all the contributors to this book for their interest, en-
thusiasm, patience and cooperation without which this book would not have been in
the hands of the readers. Reviewers are thanked for their many valuable comments.
Finally, my appreciation goes to the staff of VSP/Brill (publisher) for giving this
book a body form.
K. L. Mittal
P.O. Box 1280
Hopewell Junction, NY 12533
USA
References
1. K. L. Mittal (Ed.), Polyimides and Other High-Temperature Polymers: Synthesis, Characterization
and Applications, Vol. 1. VSP, Utrecht (2001).
and Applications, Vol. 4. VSP/Brill, Leiden (2007).
Part 1
Synthesis, Properties and Bulk
Characterization
Preparation, Characterization and Selective Solubility of
Diels–Alder Addition Polyimides
Ying-Ling Liu ∗ and Shi-Yi Chen

Department of Chemical Engineering and R&D Center for Membrane Technology Chung Yuan
Christian University, Chungli, Taoyuan 32023, Taiwan
Abstract
α, ω-Functionalized polyimides with maleimide (PI-MI) and furan (PI-F) terminal groups are prepared
for using as macro-monomers in the synthesis of high-molecular-weight polyimides (PI-DA) through the
Diels–Alder (DA) reaction between maleimide and furan groups. The chemical structures of the prepared
polyimides were characterized with FT-IR and 1 H NMR. PI-DA polymers exhibit chain-breaking behavior
under heat treatment through retro-DA reaction to bring about solubility alternation for the polymer. Chlo-
roform and dichloromethane were used as selective solvents for the PI-DA polyimides before and after heat
treatment. Thus, a new class of polyimides with selective solubility was developed.
Keywords
Diels–Alder polymers, polyimides, addition polymerization
1. Introduction
Polyimides possess attractive properties and thus are used for wide applications
[1, 2]. However, the poor solubility of aromatic polyimides in organic solvents
limits their processibility and applications [3]. A large number of investigations
on the design and preparation of organo-soluble polyimides have been reported.
A common approach to improve the solubility of polyimides is incorporation of
fluoro-containing [4–6], and asymmetric [7–10] and bulky [11–13] groups in the
repeat units of polyimides. However, design and synthesis of new monomers, which
are usually expensive and involve complicated synthesis routes, is needed for such
approaches. Another way to enhance the solubility of polyimides is to reduce their
molecular weights. Unfortunately, reduction in the molecular weights of polyimides
might also result in relatively poor thermal and mechanical properties. On the
other hand, thermoplastic polyimides, which are melt-processable, have also been
reported [14–17]. Cross-linkable groups are introduced into such polyimides to im-
* To whom correspondence should be addressed. Tel.: +886-3-2654130; Fax: +886-3-2654199; e-mail:

ylliu@cycu.edu.tw
4 Y.-L. Liu and S.-Y. Chen
prove their heat-distortion temperature and solvent resistance by cross-linking. The

final polyimides are insoluble and are not re-processable.
High molecular weight polymers could be obtained using α, ω-functionalized
oligomers as macro-monomers. Homo- and co-polyimides were obtained via this
synthesis route by means of conventional addition and condensation reactions [18–
22]. On the other hand, the Diels–Alder (DA) reaction, which involves the addition
reaction between a diene and a dienophile groups, has also been applied to polymer
synthesis [23–33]. Meador and coworkers [23, 24] reported the synthesis of high-
molecular-weight polyimides by means of the Diels–Alder addition reaction. The
adduct linkages from the DA reaction could transform to aromatic groups through
dehydration reaction under heat, so as to result in thermally stable polymers.
Block-copolymers were also obtained from DA-reactions using end-functionalized
oligomers as precursors [25–28].
Another characteristic of the DA reaction is thermal-reversibility. The DA adduct
linkages formed at low temperatures can be broken down through the reverse-
DA reaction, which usually occurs at higher temperatures. The reverse reaction is
named as retro-DA reaction. The reversibility of DA reaction is widely applied to
the design and preparation of thermally-reversible cross-linked polymers, which
have shown self-healing and re-working characteristics [29–35]. Watanabe and
Yoshie [33] also reported recyclable linear and cross-linked polymers linked with
furan/maleimide adducts. The polymers can repeatedly polymerize and depolymer-
ize through DA and retro-DA reactions, respectively.
In this work, recyclable linear polyimides were of interest. Telechelic polyimides
possessing furan (diene) and maleimide (dienophile) end groups were prepared
and used as precursors for preparation of recyclable linear polyimides by means
of DA reaction. Formation and breakage of polyimide chain linkages, which are
DA adducts of furan/maleimide groups, occur through the DA and the retro-DA re-
actions, respectively. In addition to the reversible characteristic of chain formation
and breakage, another interesting property observed with the polyimides is their
difference in solubility before and after DA reactions. The temperature-dependent
solubility of the polyimides extends their applications as microelectronic encapsu-
lants and insulators.
2. Experimental Section
2.1. Materials
4,4 -(Hexafluoroisopropylidene)diphthalic anhydride (6FDA), 4,4 -methylene di-

aniline (MDA), furfuryl amine (FA), and bismaleimide (BMI) were purchased from
Sigma-Aldrich and used as received. 4-Maleimidobenzoic acid (MBA) was pre-
pared according to the reported method [34]. Other reagents used in the reactions
were commercial reagent-grade products.
Preparation, Characterization and Selective Solubility of Diels–Alder Addition Polyimides 5
2.2. Characterization and Measurements

Fourier transform infrared (FT-IR) spectra were obtained with a Perkin Elmer
Spectrum One FT-IR spectrophotometer. Nuclear magnetic resonance (NMR) spec-
troscopy was performed with a Brüker MSL-300 NMR spectrophotometer using
DMSO-d6 as a solvent. Differential scanning calorimetry (DSC) was performed
with a Thermal Analysis DSC-Q10 instrument at a heating rate of 10◦ C/min under
nitrogen atmosphere. The gas flow rate was 40 ml/min. Thermogravimetric analy-
sis (TGA) was performed with a Thermal Analysis TGA-2050 thermogravimetric
analyzer at a heating rate of 10◦ C/min under nitrogen atmosphere.
Time-dependent reaction conversions of retro-DA reaction of polyimides were
calculated from the FT-IR spectra collected with a Perkin Elmer Spectrum One
spectrophotometer equipped with a programmable heating stage and high tem-
perature cell [35]. The intensity change of the absorption peak of DA adduct at
1605 cm−1 (C=C stretching in adduct groups) with respective to the band of C=O
group at 1715 cm−1 , which was taken as the internal standard for reference, was
utilized to monitor the occurrence of the DA reaction. Let N0 = [I1605 /I1715 ] at
t = 0, and Nt = [I1605 /I1715 ] at t = t. Let x = (N0 − Nt )/N0 , the conversion of the
adduct, and thus the integrated rate expression for a second-order kinetic reaction
will be:
1/(1 − x) = kt. (1)
2.3. Preparation of α, ω-Amine-Terminated Polyimide (PI-A)
The MDA (10.4 mmol) solution in dried N-methyl-2-pyrrolidone (NMP, 40 ml)
was stirred at 0◦ C. After addition of 6FDA (10.0 mmol), the solution was reacted at
room temperature for 8 h to form poly(amic acid) solution (PAA-A). After addition
of dried NMP (4 ml), acetic anhydride (1.73 ml), and triethylamine (25.8 ml), the
solution was reacted at room temperature for another 24 h. The reaction mixture
was poured into a large excess of methanol/water mixture (v/v = 1:1). The pre-
cipitate was collected with filtration, re-dissolved in NMP, and precipitated from
methanol/water mixture twice. The collected polymer was dried under vacuum at
150◦ C for 24 h. PI-A was obtained with a yield of 91%.
2.4. Preparation of α, ω-Maleimide-Terminated Polyimide (PI-MI)
PI-A (0.804 g) and MBA (0.451 g) were dissolved in 3 ml of NMP. CaCl2 (0.250 g),
triphenylphosphine (8.0 ml), pyridine (8.0 ml) were added to the solution. When
a homogeneous solution was obtained, the temperature was raised to 100◦ C and
kept for 3 h. After cooling to room temperature, the reaction mixture was poured
into a large excess of methanol. The precipitate was filtered, washed successively
with methanol and hot water, and then dried under vacuum at 150◦ C for 24 h.
The obtained polymer was continuously extracted with hot methanol for 4 h. After
drying, polyimide with maleimide end groups (PI-MI) was obtained with a yield of
90%.
2.5. Preparation of α, ω-Furan-Terminated Polyimide (PI-F)

The MDA (10.0 mmol) solution in dried NMP (40 ml) was stirred at 0◦ C. After ad-
dition of 6FDA (10.4 mmol) and FA (0.8 mmol), the resulting solution was reacted
at room temperature for 8 h to form poly(amic acid) solution (PAA-A). After adding
dried NMP (4 ml), acetic anhydride (1.73 ml), and triethylamine (25.8 ml), the so-
lution was reacted at room temperature for another 24 h. The reaction mixture was
poured into a large excess of methanol/water mixture (v/v = 1:1). The precipitate
was collected with filtration. The polymer was dissolved in NMP and precipitated
from methanol/water mixture. The collected polymer was dried under vacuum at
150◦ C for 24 h. PI-F was obtained with a yield of 85%.
2.6. Preparation of Polyimides Through the Diels–Alder Reaction
A BMI (0.143 g) solution in 3 ml of NMP was prepared. The solution was then
poured into another solution of PI-F (0.483 g) in NMP (3 ml). The mixture was
reacted at 50◦ C. At certain periods of reaction, samples were drawn from the re-
action solution and were precipitated from excess methanol. After repeating the
dissolution–precipitation process twice, the collected polymer was dried under vac-
uum at 150◦ C for 24 h to give the sample of PI-DA-I with a yield of 97%.
Polyimide based on PI-MI and PI-F was prepared in the same manner using PI-
MI as a macro-monomer. The obtained sample was coded as PI-DA-II with a yield
of 98%.
3. Results and Discussion

3.1. Preparation and Characterization of Polyimides
α, ω-Functionalized polyimides with amine (PI-A), maleimide (PI-MI), and furan
(PI-F) terminal groups were prepared for using as macro-monomers in the synthesis
of high-molecular-weight polyimides. The preparation route is shown in Scheme 1.
PI-A was prepared with excess diamine compounds in the reaction system. PI-F
was prepared using mono-amine compound (FA) as an end-capping agent. Reac-
tion of PI-A with MBA transforms the amine end-groups of PI-A into maleimide
groups through amidation reaction, as the reaction rate of amidation under the
catalytic conditions is much higher than that of the Michael addition reaction be-
tween the amine and maleimide groups [36]. Figure 1 shows the FT-IR spectra
of the prepared α, ω-functionalized polyimides. All samples exhibit obvious ab-
sorption peaks arising from the asymmetric and symmetric stretchings of C=O
groups in imide rings at about 1785 and 1725 cm−1 , respectively. The appearance
of these peaks demonstrates the success of polyimide preparation. Another charac-
teristic absorption of imide group arising from C–N stretching appears at around
1375 cm−1 for all samples. The amine end groups of PI-A did not exhibit obvious
absorption peaks in FT-IR analysis. However, the characteristic peaks of maleimide
groups and the furan absorptions were observed with PI-MI at 693 cm−1 (C–H
Scheme 1. Preparation of α, ω-functionalized polyimides with amine (PI-A), maleimide (PI-MI), and furan
(PI-F) terminal groups.
stretching) and with PI-F at 1075 (C–O–C stretching) and 1325 cm−1 (–CH2 – wag-
ging). Figure 2 shows the 1 H NMR spectra of the polyimides. The single peak at
δ = 4.03 ppm is associated with the protons in Ph–CH 2 –Ph– linkages. The aro-
matic protons in various chemical environments are characterized by the resonance
peaks at δ = 7.32–7.40 ppm (–CH2 –Ph–N–) and δ = 7.70–8.11 ppm (–imide–Ph–
C(CF3 )2 –). The maleimide groups of PI-MI were characterized by the resonance
peaks at δ = 7.13 and 7.16 ppm (–CH=CH– of maleimide group). Moreover, the
presence of furan groups in PI-F was also demonstrated by the existence of reso-
nance peaks at around δ = 6.36 ppm (–CH=CH– of furan group). The appearance
of the resonance peak at δ = 4.74 ppm, which represents the protons of the –CH 2 –
Scheme 2. Preparation of polyimides by means of Diels–Alder reaction between maleimide and furan
groups.
Figure 1. FT-IR spectra of α, ω-functionalized polyimides.

Figure 2. 1 H NMR spectra of α, ω-functionalized polyimides.
group linking the furan end groups and the polyimide chains, gives further support
to the proposed chemical structure of PI-F.
Figure 3 shows the TGA thermograms of the prepared polyimides. Both PI-A and
PI-F exhibited similar behaviors of weight loss under heating, since they possess
similar chemical structures. The end-groups do not play a significant role in the
thermal stability and degradation behavior of the polyimides. On the other hand, the
weight loss for PI-MI starting at around 250◦ C is due to the presence of amidation
agents and solvents remaining in the polymer. Complete removal of the residuals is
difficult even after extraction and vacuum drying [36, 37]. The high char yield of
PI-MI might be attributed to the self-addition of maleimide groups to bring about a
cross-linked structure to the sample during heating.
3.2. Polyimides from the Diels–Alder Reaction
α, ω-Functionalized polyimides were used as macro-monomers in polyimide syn-
thesis by means of the Diels–Alder reaction between furan and maleimide groups.
A mixture of PI-MI and PI-F exhibits an exothermic peak centered at about 38◦ C in
DSC heating scan, demonstrating the occurrence of Diels–Alder reaction between
maleimide and furan groups. The occurrence of the DA addition reaction was also
Figure 3. TGA thermograms of α, ω-functionalized polyimides in nitrogen under a heating rate of

10◦ C/min.
Figure 4. Reaction time-dependent inherent viscosity of PI-DA polyimides during the DA reaction.
monitored by measurements of the inherent viscosities of the resulting polyimides.

As shown in Fig. 4, the inherent viscosities of both PI-DA-I and PI-DA-II increase
with the increase in reaction time. The inherent viscosity of the PI-F/BMI mixture
was about 0.22 dl/g, which gradually increased to 0.26 dl/g in 60 h of DA reaction.
The viscosity increase rate slows down during the reaction period of 60–180 h.
On the other hand, PI-DA-II shows an increase in inherent viscosity from 0.31 to
Figure 5. FT-IR monitoring of the DA reaction between PI-MI and PI-F to form PI-DA-II polymer.
The intensity of absorption peak at 1605 cm−1 increases with increase in reaction time indicating the
formation of maleimide/furan adduct during reaction.
0.46 dl/g in 350 h of reaction. The formation of maleimide-furan adducts linking

the polyimide chains was evidenced by FT-IR analysis (Fig. 5). The absorption peak
at about 1605 cm−1 represents the absorption of C=C groups generated from DA
reaction, showing an intensity increase with increasing DA reaction time.
The thermal reversibility of DA reaction makes it possible to break down the
polyimide chains of PI-DA-I and PI-DA-II under heat. The occurrence of retro-
DA reaction was monitored with FT-IR. The PI-DA samples were kept at a fixed
temperature and their time-dependent FT-IR spectra were recorded. The decrease in
the intensity of absorption peak at 1605 cm−1 demonstrated the occurrence of retro-
DA reaction (Fig. 6). Assuming the retro-DA reaction is first-order [38], the reaction
rate constants and the activation energy of the retro-DA reaction were obtained from
the plots shown in Fig. 7. The reaction rate constants and the activation energies of
the retro-DA reactions of PI-DA polymers are reasonable and comparable to the
values reported in the literature [38] (Table 1), demonstrating that the formation
and de-bonding of polyimide chains are thermally reversible.
3.3. Selective Solubility of Polyimides
PI-DA-I and PI-DA-II were thermally treated at 140◦ C for 30 min to break down
the DA adduct linkages in polymer chains through retro-DA reaction. The resulting
Figure 6. FT-IR monitoring of the retro-DA reaction of PI-DA-I polymer at 140◦ C. The intensity of
absorption peak at 1605 cm−1 decreases with increase in reaction time indicating the occurrence of
retro-DA reaction.
polyimides are coded as PI-DA-I-R and PI-DA-II-R, respectively. Table 2 repre-

sents the solubility data for PI-DA-I and PI-DA-II polyimides before and after the
heat treatment. PI-DA-I and PI-DA-II are soluble in tetrahydrofuran (THF) and
aprotic polar solvents like N-dimethylacetamide (DMAc), N-dimethylformamide
(DMF), dimethylsulfoxide (DMSO), and NMP, and are insoluble in chloroform,
dichloromethane, and low-boiling-point solvents. After the heat treatment, the poly-
imide chains break down, consequently rendering the PI-DA-I-R and PI-DA-II-R
polymers soluble in chloroform and dichloromethane, as it can be seen in Fig. 8.
PI-DA polymers exhibit solubility alternation after heat treatment. This is to say,
chloroform and dichloromethane are selective solvents for PI-DA polymers both
before and after heat treatment.
The inherent viscosity of PI-DA-I-R is 0.22 dl/g, which is similar to the value
measured for the mixture of PI-F/BMI. It is demonstrated that after heat treatment
(a)
(b)
Figure 7. Kinetic plots for the retro-DA reaction of PI-DA-II polymer. (a) reaction conversion plots to
determine the reaction rate constants at various temperatures and (b) Arrhenius plot to determine the
activation energy.
Table 1.
Reaction rate constants and activation energies of the retro-DA reactions with PI-DA-I and PI-DA-II
polymers
Rate constants of retro-DA reaction at various temperatures Activation energy

(×10−6 s−1 ) (kJ/mol)
Temperature (◦ C) 110 120 130 140 150 160 170 180

PI-DA-I 12.7 19.7 28.6 45.1 53.9 – – – 50.2
PI-DA-II – – – 1.05 1.19 1.58 1.97 2.33 32.4
Table 2.
Solubility of PI-DA polymers before and after the heat treatment. X: insoluble; O: soluble
Solventsa Acetone THF Toluene NMP DMAc DMF DMSO CHCl3 CH2 Cl2
PI-DA-I X O X O O O O X X
PI-DA-I-R X O X O O O O O O
PI-DA-II X O X O O O O X X
PI-DA-II-R O O X O O O O O O
a Tetrahydrofuran (THF); 1-methyl-2-pyrrolidinone (NMP), N-dimethylacetamide (DMAc);
N-dimethylformamide (DMF); dimethylsulfoxide (DMSO).
Figure 8. Selective solubility of PI-DA-I and PI-DA-II in chloroform after heat treatment due to the
occurrence of retro-DA reaction.
the linkage breakages occur for PI-DA-I through retro-DA reaction. Similar be-
haviour is also observed with PI-DA-II, as the inherent viscosity of PI-DA-II-R
(0.31 dl/g) is the same as for the PI-F/PI-MI mixture. The cyclical occurrence of
polymer chain formation and breakage were demonstrated by changes in the inher-
ent viscosity of polymer solutions measured with the DA and retro-DA reactions.
4. Conclusions
Polyimides were prepared from α, ω-functionalized polyimide oligomers through
the Diels–Alder (DA) reaction. The polyimides exhibit chain-breaking behavior
under heat treatment through retro-DA reaction. Solubility alternation is observed
with the polyimides. Chloroform and dichloromethane are selective solvents for the
polyimides both before and after heat treatment. A new class of polyimides with
selective solubility was developed. The applications of these polyimides in surface
patterns and removable encapsulating insulators are under investigation.
Acknowledgements
Financial supports for this work from the National Science Council, Taiwan (NSC
95-2221-E-033-024) and the Ministry of Education, Taiwan, under the Centre-of-
Excellence Program on Membrane Technology are highly appreciated.
References
1. M. K. Ghosh and K. L. Mittal (Eds), Polyimides: Fundamentals and Applications. Mercel Dekker,
New York, NY (1996).
2. K. L. Mittal (Ed.), Polyimides and Other High Temperature Polymers, Vol. 4. VSP/Brill, Leiden
(2007).
3. J. de Abajo and J. G. de la Campa, Adv. Polym. Sci. 61, 140 (1998).
4. R. Hariharan and M. Sarojadevi, Polym. Int. 56, 22 (2007).
5. Z. M. Qiu, J. H. Wang, Q. Y. Zhang, S. B. Zhang, M. X. Ding and L. X. Gao, Polymer 47, 8444
(2006).
6. J. G. Liu, X. J. Zhao, H. S. Li, L. Fan and S. Y. Yang, High Perform. Polym. 18, 851 (2006).
7. W. Anarnarukan, S. Tanatyanon, D. Zhang, E. A. Aleman, D. A. Modarelli and F. W. Harris,
Polymer 47, 4936 (2006).
8. M. Zhang, Z. Wang, L. Gao and M. Ding, J. Polym. Sci. Part A: Polym. Chem. 44, 6836 (2006).
9. S. H. Hsiao and K. H. Lin, J. Polym. Sci. Part A: Polym. Chem. 43, 43 (2005).
10. Y. Shao, Y. F. Li, X. Zhao, X. L. Wang, T. Ma and F. C. Yang, J. Polym. Sci. Part A: Polym. Chem.
44, 6836 (2006).
11. C. H. Lin and C. H. Lin, J. Polym. Sci. Part A: Polym. Chem. 45, 2897 (2007).
12. K. Miyatake, T. Yasuda, M. Hirai, M. Nanasawa and M. Watanabe, J. Polym. Sci. Part A: Polym.
Chem. 45, 157 (2007).
13. K. Sakayori, Y. Shibasaki and M. Ueda, J. Polym. Sci. Part A: Polym. Chem. 44, 6385 (2006).
14. M. F. Grenier-Loustalot, F. Joubert and P. Grenier, J. Polym. Sci. Part A: Polym. Chem. 29, 1649
(1991).
15. M. F. Grenier-Loustalot, F. Joubert and P. Grenier, J. Polym. Sci. Part A: Polym. Chem. 31, 3049
(1993).
16. M. B. Saeed and M. S. Zhan, Eur. Polym. J. 42, 1844 (2006).
17. T. Sasaki, H. Moriuchi, S. Yano and R. Yokota, Polymer 46, 6968 (2005).
18. S. Andre, F. Guida-Pietrasanta, A. Rousseau, B. Boutevin and G. Caporiccio, J. Polym. Sci. Part
A: Polym. Chem. 42, 200 (2004).
19. S. Andre, F. Guida-Pietrasanta, A. Rousseau, B. Boutevin and G. Caporiccio, J. Polym. Sci. Part
A: Polym. Chem. 43, 2237 (2005).
20. N. Asano, K. Miyatake and M. Watanabe, J. Polym. Sci. Part A: Polym. Chem. 44, 2744 (2006).
21. X. Z. Jin and H. Ishii, J. Appl. Polym. Sci. 100, 4240 (2006).
22. S. Andre, F. Guida-Pietrasanta, A. Rousseau, B. Boutevin and G. Caporiccio, Macromol. Chem.
Phys. 205, 2420 (2004).
23. M. A. B. Meador, J. Polym. Sci. Chem. Ed. 26, 2907 (1998).
24. M. A. B. Meador, M. A. Meador and M. K. Ahn, Macromolecules 22, 4385 (1989).
25. J. R. Jones, C. L. Liotta, D. M. Collard and D. A. Schiraldi, Macromolecules 32, 5786 (1999).
26. E. Goiti, M. B. Huglin and J. M. Rego, Eur. Polym. J. 40, 219 (2004).
27. H. Durmaz, B. Colakoglu, U. Tunca and G. Hizal, J. Polym. Sci. Part A: Polym. Chem. 44, 1667
(2006).
28. I. Bicu and F. Mastata, J. Polym. Sci. Part A: Polym. Chem. 43, 6308 (2005).
29. X. Chen, M. A. Dam, K. Ono, A. K. Mal, H. Shen, S. R. Nutt, K. Sheran and F. Wudl, Science
295, 1698 (2002).
30. X. Chen, F. Wudl, A. K. Mal, H. Shen and S. R. Nutt, Macromolecules 36, 1802 (2003).
31. Y. L. Liu and Y. W. Chen, Macromol. Chem. Phys. 208, 224 (2007).
32. Y. L. Liu and C. Y. Hsieh, J. Polym. Sci. Part A: Polym. Chem. 44, 905 (2006).
33. M. Watanabe and N. Yoshie, Polymer 47, 4946 (2006).
34. Y. L. Liu and Y. H. Wang, J. Polym. Sci. Part A: Polym. Chem. 42, 3178 (2004).
35. Y. L. Liu, C. Y. Hsieh and Y. W. Chen, Polymer 47, 2581 (2006).
36. J. de Abajo and E. Santos, Angew. Makromol. Chem. 111, 17 (1983).
37. Y. L. Liu, S. H. Li, H. C. Lee and K. Y. Hsu, React. Funct. Polym. 66, 924 (2006).
38. X. Z. Wei, B. K. Zhu, L. Xiao, Y. Y. Xu and K. E. Geckeler, Macromol. Rapid Commun. 26, 1224
(2005).
Soluble Polyimides Based on Aromatic Diamines Bearing
Long-Chain Alkyl Groups
Yusuke Tsuda ∗
Department of Biochemistry and Applied Chemistry, Kurume National College of Technology,
1-1-1 Komorino, Kurume, Fukuoka 830-8555, Japan
Abstract
Although polyimides have been widely used in microelectronics, one of the disadvantages of these polymers
is that these are insoluble in common solvents. We have systematically investigated the synthesis and charac-
terization of soluble polyimides based on aromatic diamines bearing long-chain alkyl groups, and also those
containing dendritic moieties. In this paper, the effects for the enhancement of solubility based on aromatic
diamines such as alkyloxydiaminobenzene (AODB-X, X = carbon number of alkyl chain, 10–14), alkyl-
diaminobenzophenone (ADBP-9–14), diaminobenzoic acid alkylester (DBAE-8–14) and alkyldiaminoben-
zamide (ADBA-9–14) are discussed in detail and the conclusions regarding enhancement of solubility are
as follows: (1) the effect of diamines decreases in the order AODB > ADBP > ADBA > ADBE, (2) long-
chain alkyl groups bearing twelve carbon atoms are the most effective, (3) long-chain alkyl groups bearing
even numbers of carbon atoms are more effective, and (4) branched alkyl chains are more effective. The
polyimides having dendritic side chains were also investigated. For example, the terphenylene diamine
monomer having a first-generation monodendron, 3,4,5-tris(n-dodecyloxy)benzoate and the monomer hav-
ing a second-generation monodendron, 3,4,5-tris[-3 ,4 ,5 -tri(n-dodecyloxy) benzyloxy]benzoate were suc-
cessfully synthesized and the corresponding soluble polyimides were obtained by polycondensation with
tetracarboxylic dianhydride monomers such as benzophenone tetracarboxylic dianhydride (BTDA). These
polyimides were soluble in various polar solvents and also exhibited good thermal stability. Their appli-
cation as alignment layers for LCDs was also investigated, and it was found that these polyimides having
dendritic side chains were applicable for the vertically aligned nematic liquid crystal displays (VAN-LCDs).
Keywords
Polyimide, soluble polyimide, long-chain alkyl group, copolymerization, thermal stability, solubility
1. Introduction
Polyimides exhibit excellent thermal and mechanical properties, and have extensive
engineering and microelectronics applications [1–4]. Aromatic polyimides such as
polyimides based on pyromellitic dianhydride are prepared from aromatic diamines
and aromatic tetracarboxylic dianhydrides via poly(amic acid)s. Since conventional
* Tel.: (+81-942) 35-9408; Fax: (+81-942) 35-9329; e-mail: tsuda@kurume-nct.ac.jp
18 Y. Tsuda
Figure 1. Synthesis of conventional polyimides and soluble polyimides.
aromatic polyimides are insoluble, these polymers are usually processed as the
corresponding soluble poly(amic acid) precursors, and then either thermally or
chemically imidized. However, owing to the instability of poly(amic acid)s and
the liberation of water in the imidization process, problems can arise. Extensive re-
search has been recently carried out to improve the solubility of polyimides and suc-
cessful examples involve the incorporation of fluorine moieties [5–20], bulky side
groups [21–27], alicyclic structure [28], spiro linkage groups [29], cardo groups
[30, 31], pyridine moiety [32, 33], porphyrin moiety [34], acridine moiety [35] and
sulfonated structure [36–39]. And their applications include as functional materials
such as low-k materials [6, 17, 18, 28], colorless or light-colored polyimides [12,
13, 15], gas transport membranes [19, 20, 27], electrochromic materials [25], per-
vaporation membranes [10], materials for organic light-emitting diodes (OLEDs)
[30, 35] and membranes for polymer electrolyte fuel cells (PEFCs) [22, 36–39].
Soluble polyimides bearing long-chain alkyl groups have also been reported [40–
46], and their applications mainly involve their use as alignment layers for liquid
crystal displays (LCDs).
Our research group has systematically investigated the synthesis and charac-
terization of soluble polyimides based on alicyclic dianhydrides such as 2,3,5-
tricarboxycyclopentyl acetic dianhydride (TCA–AH) [47, 48], 5-(2,5-dioxotetra-
hydrofuryl)-3-methyl-3-cyclohexene-1,2-dicarboxylic anhydride (cyclohexene-
DA) [49], and 4-(2,5-dioxotetrahydrofuran-3-yl)-tetralin-1,2-dicarboxylic anhy-
dride (tetralin-DA) [49], and aromatic diamines bearing long-chain alkyl groups
such as alkyldiaminobenzophenone (ADBP-X, X = carbon numbers of alkyl chain)
[50], alkoxydiaminobenzene (AODB-X) [51], diaminobenzoic acid alkylester
(DBAE-X) [52] and alkyldiaminobenzamide (ADBA-X) [53]. Our recent paper
Soluble Polyimides Bearing Long-Chain Alkyl Groups 19
Figure 2. Representative methods for the enhancement of solubility of polyimides.
has described soluble polyimides having dendritic moieties on their side chain [54],
and the synthesis of soluble polyimides in ionic liquids was also investigated [55].
In this paper, the synthesis and characterization of a novel series of soluble
polyimides based on aromatic diamines bearing long-chain alkyl groups such as
ADBA-9–14, AODB-10–14, DBAE-8–14 and ADBA-9–14 are described, and their
structure–property relationships such as the effect of linkage group of long-chain
alkyl group, alkyl chain length, and branched alkyl group are discussed in detail.
The novel polyimides having dendritic moieties (dendronized polyimides) have
been synthesized based on the copolymerization of aromatic diamines having the
first- or second-generation monodendrons bearing long-chain alkyl groups, and
their properties are compared with those of soluble polyimides bearing long-chain
alkyl groups described above. The attachment of alkyl side chains to polyimides has
recently been used to increase pretilt angles generated by the polyimides in align-
ment layer applications of liquid crystal displays (LCDs) and, thus, these studies are
of great practical importance [41–46]. The application for the polyimide alignment
layers based on dendronized polyimides for the vertically aligned nematic liquid
crystal displays (VAN-LCDs) is finally described.
2. Experimental
2.1. Materials
The reagents for the synthesis of aromatic diamines were purchased from Tokyo
Chemical Industry Co., Ltd (TCI) or Wako Pure Chemical Industries, Ltd (Wako)
20 Y. Tsuda
Figure 3. Aromatic diamines bearing long-chain alkyl groups.
and used as received. 3,3 ,4,4 -Benzophenone tetracarboxylic dianhydride (BTDA),

4,4 -diaminodiphenylether (DDE), and 4,4 -diaminodiphenylmethane (DDM) were
purchased from TCI and purified as follows: BTDA, m.p. 228.3◦ C, recrystallized
from acetic anhydride; DDE, m.p. 193.6◦ C, recrystallized from ethanol; DDM, mp.
95.4◦ C, recrystallized from ethanol. N -Methyl-2-pyrrolidone (NMP) (Mitsubishi
Chemicals) was distilled under reduced pressure using 4 Å molecular sieves. Other
conventional reagents were purchased from TCI or Wako and used as received.
2.2. Measurements
The inherent viscosities of all polymers were measured using a Cannon Fenske
viscometer at a concentration of 0.5 g/dl in NMP at 30◦ C. Size exclusion chro-
matography (SEC) measurements were performed in NMP containing 10 mM LiBr
at 40◦ C with a TOSOH HLC-8020 equipped with a TSK-GEL ALPHA-M column.
Number average molecular weight (Mn ), weight average molecular weight (Mw )
and polydispersity (Mw /Mn ) were determined by TOSOH Multi Station GPC-8020
analysis system calibrated with a series of polystyrene standards having narrow
polydispersities. Differential scanning calorimetery (DSC) traces were obtained on
a Shimadzu DSC-60 under nitrogen (flow rate 30 ml/min) at a heating rate of
20◦ C/min using 30 mg of the film samples and the glass transition temperatures
(Tg ) were read at the midpoint of the heat capacity jump from the second heating
scan after cooling from 250◦ C. Thermomechanical analysis (TMA) was performed
on a Shimadzu TMA-50 under nitrogen (30 ml/min) at a heating rate of 10◦ C/min
with a 10 g load in the penetration mode using the film samples approximately
300 µm in thickness. Softening temperatures were taken as the onset temperature of
the probe displacement on the second TMA scan after cooling from 220◦ C. Ther-
mogravimetric analysis (TGA) was performed on a Shimadzu TGA-50 in air or
under nitrogen at a heating rate of 10◦ C/min. 1 H-NMR spectra were measured on a
JEOL JNM-AL400 FT NMR in CDCl3 or dimethyl sulfoxide-d6 with tetramethyl-

silane (TMS) as an internal standard. IR spectra were recorded on a JASCO IR
Report-100 spectrophotometer.
2.3. Synthesis of Aromatic Diamines Bearing Long-Chain Alkyl Groups
The synthesis of 3,5-diamino-N-undecylbenzamide (ADBA-11) is described as an
example. 1-Aminoundecane (7.0 g, 41 mmol) and triethylamine (5.4 g, 53 mmol)
were dissolved in 35 ml of THF. Then 3,5-dinitrobenzoyl chloride (9.4 g, 41 mmol)
in 35 ml of THF was added dropwise, and the reaction mixture was stirred at r.t.
for 3 h. The reaction mixture was then poured into a large amount of ice water.
Toluene was added to dissolve the precipitate and the organic layer was separated
and washed sequentially with H2 O, 10% aqueous NaOH, and H2 O. Subsequently,
it was dried over anhydrous sodium sulfate, filtered and the toluene was evapo-
rated. The solid obtained was recrystallized from ethanol to give 11.8 g (33 mmol,
79% yield) of white crystals (3,5-dinitro-N-undecylbenzamide): m.p. 101.1◦ C; 1 H
NMR (CDCl3 ): δ 0.88 (t, 3H, –CH3 , J = 6.80 Hz); 1.27 (m, 16H, –(CH2 )8 –CH3 );
1.68 (m, 2H, –NH–CH2 –CH2 –); 3.53 (m, 2H, –NH–CH2 –); 6.30 (s, 1H, –NH );
8.94 (d, 2H, 2,6-ArH , J = 2.0 Hz); 9.17 (t, 1H, 4-ArH , J = 1.8 Hz): IR (cm−1 )
(KBr): 3300 (NH), 1640 (C=O), 1560 and 1350 (NO2 ). Elem. Anal. Calcd for
C18 H27 N3 O5 : C, 59.16%; H, 7.45%; N, 11.50%. Found: C, 59.10%; H, 7.41%;
N, 11.44%. A mixture of 3,5-dinitro-N-undecylbenzamide (7.0 g, 19.2 mmol) and
10% palladium on carbon (Pd/C, 1.368 g) in a mixture of ethanol (25 ml) and THF
(180 ml) was stirred at room temperature for 12 h under 0.2–0.3 MPa of hydrogen
gas using a medium pressure hydrogenerator (Ishii Laboratory Works Co., Ltd, Os-
aka, Japan). The solution was filtered and the solvents were evaporated. The crude
diamine was recrystallized from ethanol to give 4.7 g (80% yield) of a pale brown
powder: m.p. 115.4◦ C; 1 H NMR (CDCl3 ): δ 0.88 (t, 3H, –CH3 , J = 7.20 Hz);
1.26 (m, 16H, –(CH2 )9 –CH3 ); 1.57 (m, 2H, –NH–CH2 –CH2 –); 3.39 (m, 2H,
–NH–CH2 –); 3.5–3.7 (broad s, 4H, –NH2 ); 5.98 (s, 1H, –NH ); 6.12 (t, 1H, 4-
ArH , J = 2.0 Hz); 6.45 (d, 2H, 2,6-ArH , J = 2.4 Hz); IR (cm−1 ) (KBr): 3350 and
3400 (NH and NH2 ), 1630 and 1590 (C=O). Elem. Anal. Calcd for C18 H31 N3 O:
C, 70.78%; H, 10.23%; N, 13.76%. Found: C, 70.59%; H, 10.18 %; N, 13.60%.
2.4. Preparation of Poly(amic acid)s and Polyimides
As a typical example, to a 30–100 ml flask was added the mixture of an equimo-
lar amount of tetracarboxylic dianhydride and aromatic diamine. The system was
purged with argon and NMP was added. Monomer concentration was kept at
20 wt%. The mixture was stirred at room temperature under argon for 12 h to allow
viscosity to increase. The obtained poly(amic acid) solution was diluted with NMP,
and polymer concentration was kept at 10 wt%. About one-third portion of the mix-
ture was poured into a large amount of water or methanol. Precipitated poly(amic
acid) was filtered, washed with excess methanol and dried at room temperature for
2 days. The residual two-third portion of the poly(amic acid) solution was used for
22 Y. Tsuda
the following polyimide preparation. To this poly(amic acid) solution were added
five molar ratio of pyridine and four molar ratio of acetic anhydride to one molar
ratio of tetracarboxylic dianhydride. The system was purged with argon and was
stirred at 120◦ C under argon for 4 h. Powdered polyimide was obtained by precip-
itation from a large amount of methanol, filtered, and washed with a large amount
of methanol, and dried at 100◦ C for 1 day. Diamine compositions of copolyimides
obtained were confirmed by 1 H NMR and a detailed description is given in the next
section. 1 H-NMR spectra, IR spectra and results of elemental analysis of represen-
tative polyimide are as follows.
Polyimide Based on BTDA/ADBA-12/DDE (100/50/50 mol%)
1 H NMR (DMSO-d ): δ 0.83 (m, CH ); 1.23 (m, –(CH ) –CH ); 1.53 (m, –NH–
6 3 2 9 3
CH2 –CH2 –); 3.3 (m, –NH–CH2 –); 7.26 (d, ArH in DDE, J = 7.6 Hz); 7.54 (d,
ArH in DDE, J = 8.0 Hz); 7.82 (m, ArH ); 8.17–8.30 (m, ArH ); 8.54 (m, NH ):
IR (cm−1 ) (KBr): 3300 (NH), 1715 and 1770 (C=O) , 1370 (C–N). Elem. Anal.
Calcd for C68 H58 N5 O6 [C38 H41 N3 O6 (BTDA/ADBA-12 unit) + C30 H17 N2 O6
(BTDA/DDE unit)]: C, 78.44%; H, 5.61%; N, 6.73%. Found: C, 77.25%; H, 5.23%;
N, 6.60%.

3.1. Monomer Synthesis
The synthesis routes for aromatic diamines bearing long-chain alkyl groups are il-
lustrated in Fig. 4. Alkyldiaminobenzophenones (ADBP-9–14) were prepared via
two steps using 3,5-dinitrobenzoyl chloride as the starting material. The Friedel–
Crafts reaction of 3,5-dinitrobenzoyl chloride with alkylbenzene catalyzed by alu-
minum chloride in nitrobenzene gave 3,5-dinitro-4 -alkylbenzophenones in good
yields. The reduction of 3,5-dinitro-4 -alkylbenzophenone was performed by cat-
alytic hydrogenation using palladium on carbon and hydrogen gas introduced by
3–5 l gas-bag. Although hydrazine hydrate/ethanol system is sometimes used for
the reduction of nitrocompounds, this system is not preferred because the carbonyl
group in 3,5-dinitro-4 -alkylbenzophenones reacts with hydrazine.
Alkyloxydiaminobenzenes (AODB-10–14) were prepared in two steps using
2,4-dinitrophenol as the starting material. The Williamson reaction using 2,4-
dinitrophenol and 1-bromoalkanes catalyzed by potassium carbonate in DMAc gave
1-alkyloxy-2,4-dinitrobenzenes in satisfactory yields. The reduction of 1-alkyloxy-
2,4-dinitrobenzenes was performed by catalytic hydrogenation using Pd/C and hy-
drogen gas at 0.2–0.3 MPa. Although the hydrazine hydrate/ethanol system can be
used for the reduction of nitro-compounds, the medium pressure system is prefer-
able due to better yields and purity of the products.
Diaminobenzoic acid alkylesters (DBAE-8–14) were prepared in two steps using
3,5-dinitrobenzoyl chloride as the starting material. The esterification reaction using
3,5-dinitrobenzoyl chloride and aliphatic alcohols having long-chain alkyl groups
Figure 4. Synthesis of aromatic diamines bearing long-chain alkyl groups.
catalyzed by triethyl amine in THF gave alkyl 3,5-dinitrobenzoate in satisfactory

yield. The reduction of alkyl 3,5-dinitrobenzoate was performed by catalytic hy-
drogenation using Pd/C as a catalyst and hydrazine hydrate/ethanol as a hydrogen
generator. The relatively mild hydrogenation using hydrazine hydrate/ethanol sys-
tem seemed to be preferable in the case of alkyl 3,5-dinitrobenzoate, because the
scissions of ester linkages were sometimes recognized besides the hydrogenation
of nitro-groups in the use of medium pressure hydrogenerator.
Alkyldiaminobenzamides (ADBA-9–14) were prepared in two steps using 3,5-
dinitrobenzoyl chloride as the starting material. The condensation reaction using
3,5-dinitrobenzoyl chloride and aliphatic amines having long-chain alkyl groups
catalyzed by triethyl amine in THF gave N-alkyl-3,5-diaminobenzamides in satis-
factory yields. The reduction of N-alkyl-3,5-diaminobenzamide was performed by
catalytic hydrogenation using Pd/C and hydrogen gas at 0.2–0.3 MPa in a medium
pressure hydrogenerator in satisfactory yield (60–80%).
The aromatic diamines containing first-generation dendritic moieties, N-(3,5-
diaminophenyl)-3,4,5-tris(alkoxy)benzamide (DPABA-X, X = 6, 12), were syn-
thesized following the previous literature (Fig. 5) [56]. 3,4,5-Trialkyloxybenzoyl
chloride, known as the building block for Percec-type dendrons [57, 58], was syn-
thesized from 3,4,5-trihydroxybenzoic acid methyl ester (gallic acid methyl ester)
followed by Williamson-etherification using alkylbromide catalyzed by potassium
carbonate, hydrolysis of ester groups by potassium hydroxide, then acid chlorina-
24 Y. Tsuda
tion using thionyl chloride. The condensation reaction using the above acid chloride
and 3,5-dinitroaniline catalyzed by triethyl amine gave the dinitro-precursor of
DPABA, and this was finally hydrogenated to DPABA.
4-[3,5-Bis(3-aminophenyl)phenyl]carbonylamino]phenyl 3,4,5-tris(n-dodecy-
loxy)benzyloxy benzoate (12G1-AG-Terphenyldiamine) and 4-[3,5-Bis(3-amino-
phenyl) phenyl]carbonylamino]phenyl 3,4,5-tris[3 ,4 ,5 -tris(n-dodecyloxy) benzy-
loxy]benzoate(12G2-AG-Terphenyl diamine) were synthesized by the method
Figure 5. Synthesis of diaminophenylalkoxybenzamide (DPABA).
Figure 6. Synthesis of dendritic terphenyl diamines (12G1 and G2-AG-terphenyldiamines).

shown in Fig. 6 using the first- and second-generation Percec-type monodendrons.

These synthesis routes include the condensation reactions with 3,5-dibromo ben-
zoic acid and 3 ,4 ,5 -tris(n-dodecyloxy)benzyloxy chloride with 4-aminophenol,
followed by Suzuki coupling reaction with 3-aminophenyl boronic acid.
3.2. Polymer Synthesis
The synthesis route for the polyimides and copolyimides based on BTDA, DDE and
aromatic diamines bearing long-chain alkyl groups is illustrated in Fig. 7. Two-step
polymerization consisting of poly(amic acid)s synthesis and chemical imidization
was performed. The poly(amic acid)s were obtained by reacting the mixture of di-
amines with an equimolar amount of BTDA at room temperature for 12 h under an
argon atmosphere. Polyimides were obtained by chemical imidization at 120◦ C in
the presence of pyridine as base catalyst and acetic anhydride as dehydrating agent.
These are the optimized synthesis conditions previously developed for the synthe-
sis of soluble polyimides in our laboratory [47–53]. BTDA that is highly reactive
and a common aromatic tetracarboxylic dianhydride was used as a dianhydride
monomer, and DDE that is highly reactive and a common aromatic diamine was
used as a diamine co-monomer. In the case of soluble polyimides, clear polyimide
solutions were eventually obtained. In other cases, clear poly(amic acid) solutions
were obtained, however, gelation or precipitation occured in the course of imidiza-
tion process.
Figure 7. Synthesis of polyimides and copolyimides based on aromatic diamines bearing long-chain
alkyl groups.
26 Y. Tsuda
3.3. Effect of Linkage Group

Experimental results of homopolymerization and copolymerization based on
BTDA/ADBP-12, AODB-12, DBAE-12, ADBA-12, DPABA-12/DDE are sum-
marized in Table 1. Although all polyamic(acid)s were soluble in NMP which is
a solvent used for polymerization, however, the solubility of homopolyimides and
copolyimides depended on polymer structures. BTDA/ADBP-12 homopolyimides
and BTDA/ADBP-12/DDE copolyimides and BTDA/ADBP-12/DDE copoly-
imides containing 40 mol% of ADBP or more were soluble in NMP. Thus, the
effect of long-chain alkyl group in ADBP for the enhancement of solubility was
confirmed. BTDA/AODB-12 homopolyimides and BTDA/AODB-12/DDE copoly-
imides containing 25 mol% or more of AODB-12 units were also soluble in NMP.
Judging from the results of copolymerization based on BTDA/ADBP-9–14/DDE
and BTDA/AODB-10–14/DDE [50, 51], it is recognized that AODB bearing alkyl
groups via an ether linkage were more effective for the enhancement of solubility
in comparison to ADBP.
On the other hand, all homopolyimides and copolyimide based on BTDA/DBAE-
8–14/DDE were insoluble in NMP probably due to the rigid ester linkage groups.
The experimental results of copolymerization based on BTDA/ADBA-12/DDE
are quite unique. Although BTDA/ADBA-12 homopolyimide was insoluble,
the copolymers, BTDA/ADBA-12/DDE (100/75/25) and BTDA/ADBA-12/DDE
(100/50/50) were soluble in NMP. The solubility of these copolyimides may be
improved by the randomizing effect based on copolymerization as well as the en-
tropy effect of long chain linear alkyl groups [50–53]. Bryant [59] also reported that
the copolyimides based on only conventional monomers such as 4,4 -oxydiphthalic
anhydride, BPDA, and 3,4 -diaminodiphenyl ether were soluble for the selected
monomer ratio by a similar randomizing effect. Based on the fact that all copoly-
imides BTDA/DBAE-8–14/DDE were insoluble in NMP [52], ADBA is more
effective for the enhancement of solubility in comparison to DBAE [53]. Figure 8
summarizes the effect of functional diamines, AODB-X, ADBP-X, ADBA-X and
DBAE-X bearing long-chain alkyl groups for the enhancement of solubility inves-
tigated in our laboratory. From the results shown in Table 1 and previous papers
[50–53], it is concluded that AODB-X bearing long-chain alkyl groups via flexible
ether linkage is the most effective and ADBP-X bearing alkyl groups via benzoyl
linkage is the second most effective. It is speculated that relatively rigid benzoyl
linkages of ADBP-X reduce the polymer solubility even though the entropy effect
of long-chain alkyl groups exists in the system. In the cases of ADBA-X bearing
alkyl groups via rigid amide linkages and DBAE-X bearing alkyl groups via rigid
ester linkages, the effects are less in comparison to AODB-X and ADBP-X. The
ranking of ADBA-X and DBAE-X can be judged from the copolymerization ex-
periments using DDE as a co-monomer, and it is concluded that ADBA-X is more
effective than DBAE-X (Table 1). However, it seems that the ranking of ADBA-X
and DBAE-X is reverse, because it is generally known that amide linkage is more
rigid than ester linkage. It may be assumed that amide linkage is more compatible
Table 1.
Properties of polyimides and copolyimides based on BTDA/diamines bearing long-chain alkyl groups (long-chain DA)/DDE
Diaminea Polyimide
Poly(amic acid)
Long-chain-DA DDE ηinh b (dl/g) Solubility ηinh b Tg c 10% Weight loss temperatured Molecular weighte
(mol%) (mol%) in NMP (dl/g) (◦ C)
in air (◦ C) in N2 (◦ C) Mn Mw Mw /Mn
Soluble Polyimides Bearing Long-Chain Alkyl Groups

ADBP-12
0 100 0.68 insoluble
50 50 0.49 soluble 0.37 264 467 500
75 25 0.49 soluble 0.46 261 469 481
100 0 0.34 soluble 0.37 254 468 464
AODB-12
25 75 0.39 soluble 0.29 262 460 456
50 50 0.21 soluble 0.23 264 456 457
75 25 0.14 soluble 0.19 284 447 452
100 0 0.14 soluble 0.16 277 436 441
DBAE-12
ADBA-12
50 50 0.66 soluble 0.57 247f 474 468 43 700 97 000 2.2
75 25 0.59 soluble 0.36 260f 452 435 27 900 54 200 1.9
27
28
Table 1.
(Continued.)
Diaminea Polyimide
Poly(amic acid)
Long-chain-DA DDE ηinh b (dl/g) Solubility ηinh b Tg c 10% Weight loss temperatured Molecular weighte
in air (◦ C) in N2 (◦ C) Mn Mw Mw /Mn
DPABA-12
Y. Tsuda
50 50 0.83 soluble 0.65 253, 241f 453 446 45 300 119 100 2.6
75 25 0.60 soluble 0.39 325f 400 441 31 500 77 200 2.5
100 0 0.53 soluble 0.37 247f 352 429 25 600 55 300 2.2
a Equimolar amount of BTDA (3.3 ,4,4 -benzophenonetetracarboxylic dianhydride) was used as a tetracarboxylic dianhydride. Reaction conditions; r.t.,
12 h for poly(amic acid), pyridine (5 mol)/Ac2 O (4 mol), 120◦ C.
b Measured at 0.5 g/dl in NMP at 30◦ C.
c Measured by DSC at a heating rate of 20◦ C/min in N on second heating.
2
d Measured by TGA at a heating rate of 10◦ C/min.
e Determined by SEC in NMP containing 10 mM LiBr using a series of polystyrenes standards having narrow polydispersities.
f Softening temperature, measured by TMA at a heating rate of 10◦ C/min.
with the amide solvent, NMP and this phenomenon compensates for the reduction
of solubility due to rigid amide linkage.
3.4. Effect of Alkyl Chain Length
This effect was clarified by carrying out copolymerization experiments based
on BTDA/AODB-10–14/DDE [51], and the results are summarized in Fig. 9.
BTDA/AODB-10/DDE and BTDA/AODB-14/DDE copolyimides containing
75 mol% of AODB units were soluble in NMP, while these copolyimides containing
less than 75 mol% of AODB units were insoluble in NMP. BTDA/AODB-11/DDE
and BTDA/AODB-13/DDE copolyimides containing 50 mol% or more of AODB
units were soluble in NMP, while these copolyimides containing less than 50 mol%
of AODB units were insoluble in NMP. AODB-12 was found to be the most effec-
Figure 8. Effect of aromatic diamines bearing long-chain alkyl groups on polyimide solubility.
Figure 9. Effect of alkyl chain length of aromatic diamines bearing long-chain alkyl groups on poly-
imide solubility (summary of the data from [51]).
30 Y. Tsuda
Figure 10. Solubility of polyimides based on BTDA and aromatic diamines bearing long-chain alkyl
groups.
tive AODB unit for soluble polyimides, and BTDA/AODB-12/DDE copolyimides

containing 25 mol% or more of AODB-12 units were soluble in NMP. The effect
of chain length of AODB can be summarized as follows: AODB-12 > AODB-11,
13 > AODB-10, 14. It can be speculated that the entropy effect of long-chain lin-
ear alkyl groups increases up to 12 methylene units, however, the crystallization of
long-chain linear alkyl groups becomes dominant in the case of more than twelve
methylene units.
3.5. Odd–Even Effect of Long-Chain Alkyl Groups on Solubility of Polyimides
An interesting phenomenon was observed while polyimides based on ADBP were
investigated [50]. ADBP with even numbers of alkyl chains such as ADBP-10, 12,
14 produced NMP soluble polyimides in combination with BTDA, while the poly-
imides based on BTDA and ADBP having odd numbers of alkyl chains such as
ADBP-9, 11, 13 were insoluble in NMP (Fig. 10). The odd–even dependence of
physical properties is sometimes observed among the polymers containing poly-
methylene units. For example, the odd–even dependence of phase transition para-
meters for main chain liquid crystalline polymers containing flexible spacers has
been well known [60]. Furthermore, Jin et al. reported that thermotropic main
chain polyesters with even number polymethylene spacers were soluble only in
the mixture of phenol and 1,1,2,2-tetrachloroethane (TCE), while those with odd
number of methylene units were soluble in TCE [61]. The above solubility results
of BTDA/ADBP polyimides are also of great interest because of unique odd–even
effect on solubility of polymers with polymethylene units. It can be assumed that

the conformation around C–C bonds of the long-chain linear alkyl groups and the
alignment of benzene ring attached to these alkyl groups and carbonyl group affect
the solubility of these polyimides.
3.6. Effect of Branched Alkyl Group
To improve the solubility of rigid polyimides based on DBAE, branched alkyl

chains were introduced instead of linear long-chain alkyl groups [52]. Experimen-
tal results are summarized in Table 2. DBAE having branched alkyl chains such as
nonan-5-yl 3,5-diaminobenzoate (DBAE-9-branch-A) and 2,6-dimethylheptane-4-
yl 3,5-diaminobenzoate (DBAE-9-branch-B) were introduced in these polyimides,
and the homopolyimides based on BTDA/DBAE-9-branch-A and BTDA/DBAE-
9-branch-B, and copolyimides containing more than 50% of DBAE-9-branch-A or
DBAE-9-branch-B were soluble in NMP. Thus, it was found that the introduction
of branched alkyl chains enhances solubility.
3.7. Dendronized Soluble Polyimides
Based on the effect of branched alkyl groups for the enhancement of solubility in
comparison to linear alkyl groups from the above research [52], fan-shaped three
long-chain alkyl groups were introduced into the polyimide backbone via benza-
mide linkage by the copolymerization of DPABA-6, 12. Experimental results of
homopolymerization and copolymerization based on DPABA are summarized in
Table 1. The homopolyimides based on BTDA/DPABA-6, 12 and copolyimides
containing more than 50% of DPABA-6, 12 were soluble in NMP. It is speculated
that radially extended long-chain alkyl groups behave as a branched alkyl chain and
eventually enhance the solubility in a similar manner as DBAE having branched
alkyl chains enhances the solubility.
The polymerization of 12G1-terphenyldiamine in NMP was investigated us-
ing BTDA, which is a representative aromatic tetracarboxylic dianhydride, and
co-diamine monomer, DDE which is a conventional aromatic diamine (Table 3).
Although a viscous poly(amic acid) solution from the above monomers was ob-
tained, precipitation occurred during the imidization process. It was speculated
that the hydrocarbon moiety of dendron reduces the solubility of polyimides in
NMP; therefore, a polar aromatic solvent, m-cresol, was used to improve the sol-
ubility of dendron moieties. Thus, a soluble copolyimide based on BTDA/12G1-
terphenyldiamine/DDE (100/50/50) was obtained. This copolyimide has medium
molecular weight (ηinh = 0.56 dl/g) and good film forming ability. On the other
hand, homopolyimide based on BTDA/12G1-terpnenyl diamine was insoluble in
NMP. The solubility of above copolyimides may be improved by the random-
izing effect based on copolymerization as well as the entropy effect of long-
chain linear alkyl groups. The dendronized copolyimides based on BTDA/12G2-
terphenyldiamine/DDE were not soluble in even m-cresol, and pyridine, an aro-
32
Table 2.
Properties of polyimides and copolyimides based on BTDA/DBAE-X/DDE
Diaminea Poly(amic acid) Polyimide

ηinh b (dl/g) Diamine compositionc 10% Weight loss temperaturee
DBAE-X DDE Solubility ηinh b Tg d
DBAE-X (mol%) DDE (mol%) in air (◦ C) in N2 (◦ C)
DBAE-8–14
25 75 0.68–0.80 insoluble
50 50 0.57–0.65 insoluble
75 25 0.40–0.50 insoluble
100 0 0.39–0.43 insoluble
DBAE-9-branch-A
Y. Tsuda
50 50 1.02 soluble 0.83 278 346 351
75 25 1.13 soluble 0.82 76 24 280 340 333
100 0 0.90 soluble 0.88 286 334 331
DBAE-9-branch-B
50 50 0.99 soluble 0.93 292 348 340
75 25 0.88 soluble 0.86 76 24 278 330 330
100 0 0.89 soluble 0.84 100 0 296 323 327
a Equimolar molar amount of BTDA was used as a tetracarboxylic dianhydride.
b Measured at 0.5 dl/g in NMP at 30◦ C.
c Calculated from the intensity ratio of methine proton based on DBAE-9-branch-A(B) and aromatic protons based on DDE in 1 H-NMR measurement.
d Measured by DSC at a heating rate of 20◦ C/min in N on second heating.
2
e Measured by TGA at a heating rate of 10◦ C/min.
Table 3.
Properties of polyimides and copolyimides based on dendritic diamine monomers
Dianhydridea, b Diaminec Comonomerd Solvent Poly(amic acid) Polyimide Tg j Td (◦ C)k

(mol%) (mol%) ηinh (dl/g)i (◦ C)
Solubility ηinh (dl/g)i in air in N2
BTDA 12G1 NMPe 0.31 insoluble

BTDA 12G1 (50) DDE (50) NMP 1.00 insoluble
BTDA 12G1 m-cresolf insoluble
BTDA 12G1 (75) DDE (25) m-cresol insoluble
BTDA 12G1 (50) DDE (50) m-cresol soluble 0.56 not observed 455 449
BTDA 12G1 (25) DDE (75) m-cresol soluble 0.78 not observed 442 448
BTDA DDE m-cresol insoluble
BTDA 12G2 pyridineg soluble 0.06 249 365 381
BTDA 12G2(75) DDE (25) pyridine soluble 0.12 not observed 362 384
BTDA 12G2 (50) DDE (50) pyridine soluble 0.22 269 349 386
BTDA 12G2 (25) DDE (75) pyridine insoluble
6FDA 12G1 NMP 0.21 soluble 0.19 311 436 440
6FDA 12G2 NMP/THFh soluble 0.12 271 441 375
a BTDA: 3.3 ,4,4 -benzophenonetetracarboxylic dianhydride.

b 6FDA: 4,4 -hexafluoroisopropylidenediphthalic anhydride.
c See Fig. 3.
d DDE: 4,4 -diaminodiphenylether.
e Reaction condition: r.t., 12 h for poly(amic acid), pyridine (5 mol)/Ac O (4 mol), 120◦ C, 4 h for polyimides.
2
f Reaction condition: 80◦ C, 2 h → 160◦ C, 8 h.
g Reaction condition: 40◦ C, 12 h → Ac O (4 mol), 120◦ C, 4 h.
2
h Reaction condition: r.t., 12 h → pyridine (5 mol)/Ac O (4 mol), 120◦ C, 4 h.
2
i Measured at 0.5 dl/g in solution at 30◦ C.
j Measured by DSC at a heating rate of 20◦ C/min in N on second heating.
2
k 10% weight loss temperature measured by TGA at a heating rate of 10◦ C/min.
33
34 Y. Tsuda
matic and more polar solvent, was necessary to dissolve these, probably due to the
large aromatic moieties of second-generation dendrons.
It is well known that 6FDA produces a soluble polyimide in combination with
most aromatic diamines because of the effect of two trifluoro methyl groups. The
polyimide based on 6FDA/12G1-terphenyldiamine was readily soluble in NMP,
however 12G2-terphenyldiamine monomer itself was insoluble in NMP. The poly-
imide based on 6FDA/12G2-terphenyldiamine was obtained in NMP/THF (1/1,
volume ratio) (Table 3). It is interesting that the solubility of this dendronized PI was
dramatically changed and it was insoluble in polar aprotic solvents such as NMP,
DMF and DMSO, but easily soluble in less polar solvents such as dichloromethane,
chloroform, toluene and THF (Fig. 11). This solubility change is probably affected
by hydrocarbon moieties of dendritic side chains of polyimides. The detailed ex-
planation concerning dendronized polyimides will be reported elsewhere.
3.8. Polymer Characterization
In the case of soluble polyimides, NMR measurements are convenient because so-
lution samples can be prepared, and also provide more quantitative data. Structures
of polyimides were confirmed by 1 H NMR in case these polyimides were soluble
in common deuterium solvents such as DMSO-d6 . For example, Fig. 12 shows the
1 H-NMR spectrum of the copolyimide based on ADBA-12/DDE (50/50) which is
soluble in DMSO-d6 and the peaks support this polymer structure. The intensity
ratio of CH3 protons of long-chain alkyl groups and the aromatic proton HA or
HB is approximately 3/4, meaning that copolymer composition corresponds to the
monomers initial ratio.
Imidization ratios of polyimides are generally determined by FT-IR measure-
ments, by comparing absorption intensities of amic acid carbonyl groups with
those of imide carbonyl groups. Generally, a broad signal due to the NH protons
of poly(amic acid) appears around 12 ppm in DMSO-d6 , while this signal disap-
pears in the corresponding polyimide. The imidization ratios of these polyimides
can be calculated from the reduction in intensity ratio of the NH proton signals in
poly(amic acid)s and these values for the polyimides prepared in our laboratory are
sufficiently high, near 100%.
As an index of molecular weight, the measurement of inherent viscosities (ηinh )
and SEC measurement have been carried out in our laboratory. For examples, ηinh
values for the soluble polyimides in Table 1 are in the range of 0.16–0.65 dl/g.
The weight average molecular weights of the polyimides based on ADBA-12
and DPABA-12 determined by SEC measurements are in the range of 55 300 to
119 100. These values indicated that the molecular weights of these polyimides
were considered to be medium or rather low values for polyimides, however, all
polyimides show good film formation ability. In almost all cases, the molecular
weights increased with the percentage of DDE, i.e., highly reactive diamine. The
representative SEC traces are shown in Fig. 13, indicating that copolyimides based
on BTDA/ADBA-11/DDE have typical monomodal molecular weight distribution,
Figure 11. Solubility of dendronized polyimides based on 6FDA and 12G2-AG-terphenyldiamine (!; soluble, ×; insoluble).
35
36 Y. Tsuda
Figure 12. 1 H-NMR spectrum of a copolyimide based on BTDA/ADBA-12/DDE (100/50/50). 1 H

NMR (DMSO-d6 ): δ = 0.83 (m, CH3 ); 1.23 (m, –(CH2 )9 –CH3 ); 1.53 (m, –NH–CH2 –CH2 –); 3.3
(m, –NH–CH2 –); 7.23 (d, ArH in DDE, J = 8.0 Hz); 7.54 (d, ArH in DDE, J = 8.4 Hz); 7.83 (m,
ArH ); 8.17–8.30 (m, ArH ); 8.63 (m, NH ).
and their polydispersity is in the range of 1.9–2.2, which are typical values for poly-
condensation polymers.
The glass transition temperatures (Tg ) of these polyimides were determined by
DSC measurements and TMA measurements, and the thermal stabilities were eval-
uated in terms of 10% weight-loss temperatures (Td10 ) in TGA measurements. The
data are summarized in Tables 1–3, and representative DSC trace, TMA trace, and
TGA traces are shown in Figs 14, 15 and 16, respectively.
The Tg ’s of these polyimides sometimes were not recognized by DSC measure-
ments, probably due to the rigid imide linkages. In these cases, TMA measurements
were performed to determine the Tg . Many publications have described that the
softening temperature (Ts ) obtained from TMA measurements corresponds to the
apparent Tg of polymers [62, 63]. As can be seen from Tables 1–3, the Tg values
of these polyimides are in the range from 241–325◦ C, and are 100–150◦ C lower
than those of the conventional fully aromatic polyimides, however, are 100–150◦ C
higher than the commodity thermoplastics. Consequently, these soluble polyimides
based on ADBA and DPABA can be ranked as heat resistant polymers.
The Td10 values of these polyimides bearing long-chain alkyl groups in Tables
1–3 are in the range 323–474◦ C in air and 327–500◦ C under nitrogen, showing
Figure 13. Representative SEC traces of soluble polyimides based on aromatic diamines bearing
long-chain alkyl groups. BTDA/ADBA-11/DDE (100/50/50): Mn , 49 500; Mw , 118 800; Mw /Mn ,
2.4. BTDA/ADBA-11/DDE (100/75/25): Mn , 30 700; Mw , 67 900; Mw /Mn , 2.2.
Figure 14. Representative DSC trace of soluble polyimides based on aromatic diamines bearing
long-chain alkyl groups (data from [53]).
similar values as observed in soluble polyimides obtained from our laboratory (ca.
400–500◦ C). As the incorporation of DDE resulted in a reduction of aliphatic com-
ponents of the polyimides, the Td10 of these polyimides tends to increase with the
increment of the DDE component (Table 1). These Td10 values of soluble poly-
imides obtained in our laboratory are 100–200◦ C lower than those of wholly aro-
matic polyimides; however, these polyimides still can be ranked as heat resistant
polymers.
38 Y. Tsuda
Figure 15. Representative TMA trace of soluble polyimides based on aromatic diamines bearing
Figure 16. Representative TGA traces of soluble polyimides based on aromatic diamines bearing
3.9. Alignment Layer Application for VAN-LCDs

The alignment layer application for VAN-LCDs using polyimides with dendritic
side chain was performed at Cheil Ind. Inc., Korea. The polyimide alignment layers
containing 8 mol% of 12G1-terphenyldiamine were utilized for the vertical align-
ment mode (VA-mode). Figure 17 shows a V–T (voltage–transmittance) curve of
Figure 17. Voltage–transmittance curves of LCD test cells using dendronized and conventional poly-
imides.
these test cells, and shows a dramatic change of T . The pretilt angles of LC mole-
cules were over 89◦ in these test cells, which are suitable for VAN-LCDs. It is
speculated that an extremely bulky and hydrophilic dendron moiety affects the gen-
eration of pretilt angles between the surface of polyimide and liquid crystalline
molecules (Fig. 18). A detailed discussion of this application will be published
elsewhere with experimental data.
4. Conclusion
The synthesis and characterization of a novel series of soluble polyimides based on
aromatic diamines bearing long-chain alkyl groups such as alkyloxydiaminoben-
zene (AODB-10–14, number: carbon number of alkyl chain), alkyldiaminoben-
zophenone (ADBP-9–14) diaminobenzoic acid alkylester (DBAE-8–14), and alkyl-
diaminobenzamide (ADBA-9–14) were carried out, and the following conclusions
with respect to enhancing the solubility are: (1) the effect of functional diamines
decreases in the order AODB > ADBP > ADBA > ADBE; (2) long chain alkyl
groups bearing twelve carbon atoms are the most effective; (3) long chain alkyl
groups bearing even numbers of carbon atoms are more effective; and (4) branched
alkyl chains are more effective.
40 Y. Tsuda
Figure 18. Vertical alignment of LC molecules using dendronized polyimides as alignment layers.
Polyimides having dendritic side chains were also investigated. Polyimides and
copolyimides based on BTDA, N-(3,5-diaminophenyl)-3,4,5-tris(alkoxy)benza-
mide (DPABA-X, X = 6, 12), and DDE were examined, and those containing
50 mol% or more DPABA were soluble. It is speculated that the three long-
chain alkyl groups in DPABA-6,12 enhance the solubility of polyimides. The ter-
phenylene diamine monomer having a first-generation monodendron, 3,4,5-tris(n-
dodecyloxy)benzoate and the monomer having a second-generation monodendron,
3,4,5-tris[-3 ,4 ,5 -tri(n-dodecyloxy)benzyloxy]benzoate were successfully synthe-
sized and corresponding soluble polyimides were obtained by polycondensation
with tetracarboxylic dianhydride monomers such as benzophenone tertracarboxylic
dianhydride (BTDA) in aromatic polar solvents such as m-cresol and pyridine.
These polyimides and copolyimides were soluble in various polar solvents and ex-
hibited good thermal stability in both air and under nitrogen. Their application as
alignment layers for LCDs was investigated, and it was found that these polyimides
having dendritic side chains were applicable for the vertically aligned nematic liq-
uid crystal displays (VAN-LCDs).
Acknowledgements
The author thanks Drs. Takaaki Matsuda and Tsutomu Ishi-I of Kurume National
College of Technology for advice. The author also thanks many students of Kurume
National College of Technology for their help with the experiments. Financial sup-
port from the Technopolis Foundation of Kurume-Tosu and Cheil Industries Inc. is
gratefully acknowledged.
References
1. K. L. Mittal (Ed.), Polyimides; Synthesis, Characterization and Applications, Vols 1 and 2. Plenum
Press, New York, NY (1984).
2. M. K. Ghosh and K. L. Mittal (Eds), Polyimides; Fundamentals and Applications. Marcel Dekker,
3. K. L. Mittal (Ed.), Polyimides and Other High-Temperature Polymers, Vol. 3. VSP/Brill, Leiden
(2005).
4. K. L. Mittal (Ed.), Polyimides and Other High-Temperature Polymers, Vol. 4. VSP/Brill, Leiden
(2007).
5. D. Yao and P. R. Sundrarajan, Eur. Polym. J. 42, 302 (2006).
6. H. Li, J. Liu, K. Wang, L. Fan and S. Yang, Polymer 47, 1443 (2006).
7. C.-P. Yang, Y.-Y. Su and M.-Y. Hsu, Polym. J. 38, 132 (2006).
8. J. K. Park, D. H. Lee, B. J. Song, J. B. Oh and H. K. Kim, J. Polym. Sci., Part A: Polym. Chem.
44, 1326 (2006).
9. H.-S. Li, J.-G. Liu, J.-M. Rui, L. Fan, S.-Y. Yang and H.-S. Li, J. Polym. Sci., Part A: Polym.
Chem. 44, 2665 (2006).
10. L. Wang, Z. Zhao, J. Li and C. Chen, Eur. Polym. J. 42, 1266 (2006).
11. X. Z. Jin and H. Ishii, J. Appl. Polym. Sci. 100, 4240 (2006).
12. C.-P. Yang, Y.-P. Chen, E.-M. Woo and S.-H. Li, Polym. J. 38, 457 (2006).
13. C.-P. Yang, Y.-Y. Su and H.-C. Chiang, React. Funct. Polym. 66, 689 (2006).
14. Z. Qiu, J. Wang, Q. Zhang, S. Zhang, M. Ding, L. Gao, Z. Qiu, J. Wang and Q. Zhang, Polymer
47, 8444 (2006).
15. C.-P. Yang, Y.-Y. Su and Y.-C. Chen, J. Appl. Polym. Sci. 102, 4101 (2006).
16. Y. Shao, Y.-F. Li, X. Zhao, X.-L. Wang, T. Ma and F.-C. Yang, J. Polym. Sci., Part A: Polym.
Chem. 44, 6836 (2006).
17. R. Hariharan and M. Sarojadevi, Polym. Int. 56, 22 (2007).
18. X.-J. Zhao, J.-G. Liu, J.-M. Rui, L. Fan and S.-Y. Yang, J. Appl. Polym. Sci. 103, 1442 (2007).
19. W. Li, S. Zhang, G. Chen, Q. Zhang, W. Li, G. Chen and Q. Zhang, Polymer 48, 3082 (2007).
20. Z. Qiu, S. Zhang, W. Li, Z. Qiu and W. Li, J. Appl. Polym. Sci. 104, 2395 (2007).
21. B.-K. Chen, S.-Y. Tsay and Y.-J. Tsai, Polym. Int. 55, 93 (2006).
22. C.-H. Lee, H. B. Park, Y. S. Chung, Y. M. Lee, H. B. Park and B. D. Freeman, Macromolecules
39, 755 (2006).
23. D.-J. Liaw, C.-C. Huang and W.-H. Chen, Macromol. Chem. Phys. 207, 434 (2006).
24. X. Liu, J. Tang, J. Wang and Y. Gu, J. Appl. Polym. Sci. 101, 2255 (2006).
25. S.-H. Hsiao, Y.-M. Chang, H.-W. Chen and G.-S. Liou, J. Polym. Sci., Part A: Polym. Chem. 44,
4579 (2006).
26. R. Hariharan and M. Sarojadevi, J. Appl. Polym. Sci. 102, 4127 (2006).
27. Z. Qiu, G. Chen, Q. Zhang, S. Zhang, Z. Qiu, G. Chen and Q. Zhang, Eur. Polym. J. 43, 194
(2007).
28. A. S. Mathews, I. Kim and S.-C. Ha, J. Appl. Polym. Sci. 102, 3316 (2006).
29. Q. Zhang, G. Chen, S. Zhang, Q. Zhang and G. Chen, Polymer 48, 2250 (2007).
30. S. Cao, Y. Jin, S. Xu, M. Yang, S. Xu and F. Bai, Polym. Adv. Technol. 17, 556 (2006).
31. C.-P. Yang, Y.-Y. Chen and S.-H. Hsiao, J. Appl. Polym. Sci. 104, 1104 (2007).
42 Y. Tsuda
32. X. Wang, Y. Li, S. Zhang, T. Ma, Y. Shao and X. Zhao, Eur. Polym. J. 42, 1229 (2006).
33. X. Wang, Y. Li, C. Gong, S. Zhang and T. Ma, J. Appl. Polym. Sci. 104, 212 (2007).
34. W. Anannarukan, S. Tantayanon, D. Zhang, F. W. Harris, E. A. Aleman and D. A. Modarelli,
Polymer 47, 4936 (2006).
35. S. Xu, S. Cao, S. Xu and M. Yang, Polymer 48, 2241 (2007).
36. S. Chen, K. Tanaka, H. Kita, K. Okamoto and Y. Yin, Polymer 47, 2660 (2006).
37. Z. Hu, Y. Yin, S. Chen, O. Yamada, K. Tanaka, H. Kita and K. Okamoto, J. Polym. Sci., Part A:
Polym. Chem. 44, 2862 (2006).
38. Z. Qiu, S. Wu, S. Zhang, Z. Qiu, S. Wu, Z. Li, Z. Li, W. Xing and C. Liu, Macromolecules 39,
6425 (2006).
39. Y. Li, R. Jin, Z. Wang, Z. Cui, W. Xing, L. Gao, Y. Li and Z. Cui, J. Polym. Sci., Part A: Polym.
Chem. 45, 222 (2007).
40. J. R. Sasthav and F. W. Harris, Polymer 36, 4911 (1995).
41. K.-W. Lee, S.-H. Paek, A. Lien, C. Durning and H. Fukuro, Macromolecules 29, 8894 (1996).
42. C. Lee, T.-H. Woo and M. Lee, Mol. Cryst. Liq. Cryst. 316, 205 (1998).
43. B. S. Ban, Y. N. Rin and Y. B. Kim, Liquid Crystals 27, 125 (2000).
44. J.-T. Jung, M.-H. Yi, S.-K. Kwon and K.-Y. Choi, Mol. Cryst. Liq. Cryst. 333, 1 (1999).
45. L. Li, J. Yin, Y. Sui, H.-J. Xu, J.-H. Fang, Z.-K. Zhu and Z.-G. Wang, J. Polym. Sci., Part A:
Polym. Chem. 38, 1943 (2000).
46. S. J. Lee, J. C. Jung, S. W. Lee and M. Ree, J. Polym. Sci., Part A: Polym. Chem. 42, 3130 (2004).
47. Y. Tsuda, Y. Tanaka, K. Kamata, N. Hiyoshi, S. Mataka, Y. Matsuki, M. Nishikawa, S. Kawamura
and N. Bessho, Polym. J. 29, 574 (1997).
48. Y. Tsuda, K. Etou, N. Hiyoshi, M. Nishikawa, Y. Matsuki and N. Bessho, Polym. J. 30, 222 (1998).
49. Y. Tsuda, R. Kuwahara, K. Fukuda, K. Ueno and J.-M. Oh, Polym. J. 37, 126 (2005).
50. Y. Tsuda, T. Kawauchi, N. Hiyoshi and S. Mataka, Polym. J. 32, 594 (2000).
51. Y. Tsuda, K. Kanegae and S. Yasukouchi, Polym. J. 32, 941 (2000).
52. Y. Tsuda, M. Kojima and J.-M. Oh, Polym. J. 38, 1043 (2006).
53. Y. Tsuda, M. Kojima, T. Matsuda and J.-M. Oh, Polym. J. 40, 354 (2008).
54. Y. Tsuda, R. Kuwahara and J.-M. Oh, Trans. Mater. Res. Soc., Japan 29, 267 (2004).
55. Y. Tsuda, T. Yoshida and T. Kakoi, Polym. J. 38, 88 (2005).
56. J. J. van Gorp, J. A. J. M. Vekemans and E. W. Meijer, Mol. Cryst. Liq. Cryst. 397, 191 (2003).
57. V. S. K. Balagurusamy, G. Ungar, V. Percec and G. Johansson, J. Am. Chem. Soc. 119, 1539
(1997).
58. V. Percec, W.-D. Cho, M. Moeller, S. A. Prokhorova, G. Ungar and D. J. P. Yeardley, J. Am. Chem.
Soc. 122, 4299 (2000).
59. R. G. Bryant, Polym. Prep., Am. Chem. Soc., Div. Polym. Chem. 35, 517 (1994).
60. C. K. Ober, J.-I. Jin, Q. Zhou and R. W. Lenz, Adv. Polym. Sci. 59, 104 (1984).
61. J.-I. Jin, E.-J. Choi, S.-C. Ryu and R. W. Lenz, Polym. J. 18, 63 (1986).
62. S.-H. Hsiao and T.-L. Huang, J. Polym. Res. 11, 9 (2004).
63. T. Ozawa, J. Thermal Analysis 40, 1379 (1993).
Synthesis and Properties of Novel Bismaleimides,
Polyaspartimides and Organosoluble Polyimides
M. Sarojadevi ∗ , R. Hariharan and N. Amutha

Department of Chemistry, Anna University, Chennai-600025, India
Abstract
Aromatic polyimides have been widely used in microelectronics, aerospace and other industries. However,
many commercially available polyimides have restricted applications due to their insolubility, intractabil-
ity and processing requirements; therefore, many efforts have been made in synthesizing soluble, tractable
polyimides without appreciable sacrifice in their desired properties. One of the successful approaches has
been to introduce pendant phenyl and alkyl groups into the polymer. Bismaleimides are another approach to
improve processibility. Hence, in the present study, new diamine monomers were synthesized by treating a
substituted aldehyde with 2,6-dimethyl aniline in the presence of dry hydrochloric acid to yield the diamine
bis(4-amino-3,5-dimethyl phenyl) X phenyl methane (where X = Cl, Br, F, –OCH2 C6 H5 , –OCH3 ) and
bis(4-amino-3,5-dimethyl phenyl) anthramethane. A series of organosoluble polyimides were prepared from
the diamine with various aromatic tetra carboxylic acid dianhydrides via one-step and two-step methods. All
the polyimides were soluble in aprotic solvents such as NMP, DMF, DMAc and DMSO. The polyimides
show excellent thermal stability and good mechanical properties. The glass transition temperature of the var-
ious polyimides ranges from 210 to 325◦ C and the temperatures at which 10% weight loss occurs are in the
range 460–496◦ C. All polyimides formed tough transparent films with tensile strength 79–118 MPa, % elon-
gation 8–12% and modulus 1.5–2.3 GPa. These polyimide films have low dielectric constant (3.29–3.80 at
1 kHz and 3.14–3.76 at 10 kHz) and low moisture uptake 0.52–1.10%. Bismaleimides were also synthesized
from the prepared diamines and maleic anhydride via bismaleamic acid as an intermediate followed by cy-
clodehydration. The structure was confirmed by FT-IR, 1 H-NMR and elemental analyses. Further, a series of
polyaspartimides were synthesized by addition reaction of bismaleimides (BMDA) with various diamines.
The polymers were characterized by IR and elemental analysis. The Tg of polyaspartimides is in the range
of 125–278◦ C and T 10% weight loss is in the range of 398–476◦ C.
Keywords
Novel aromatic diamines, polyimides, polyaspartimides, mechanical and thermal properties
* To whom correspondence should be addressed. Tel.: (+91) 44 22203158; Fax: (+91) 44 22200660;
e-mail: msrde2000@yahoo.com
44 M. Sarojadevi et al.
1. Introduction
Polyimides are high-performance polymers, which have remarkable temperature
resistance (high glass transition temperature), mechanical and electrical properties
(low dielectric constant) and good dimensional stability [1–5]. They have wide ap-
plications in the electronics industry as flexible circuitry substrates, stress buffers,
interlayer dielectrics and passivation layers. Optically transparent polyimide films
have been used in solar radiation protectors, as orientation layers in liquid crys-
tal display devices, as optical waveguides for communication interconnects, and as
optical half-wave plates for planar light-wave circuits. However, polyimides have
some drawbacks such as insolubility, infusibility, and evolution of volatiles during
the ring formation, leading to poor processibility. To increase the utility of these
high-performance materials, extensive research is being carried out all over the
world to develop a potential material to overcome these drawbacks.
One approach involves how to synthesize organosoluble polyimides without loss
of their thermal properties. Several modifications of the chemical structure have
been made to improve the solubility and processability of polyimides with reason-
able thermal stability. Incorporation of flexible segments and bulky pendant groups
and the introduction of non-coplanar and asymmetrical moieties are some of the
methods that have been successful in decreasing crystallinity and intermolecular
interactions in order to increase solubility and tractability [6–18]. Solubility of poly-
imides can also be improved by incorporation of less symmetric structures such as
ortho- and meta-catenated aromatic units in the main chain [19–22]. The solubility
of polyimides is expected to be enhanced by the halogen atoms in the polymer be-
cause of polar interaction and also an asymmetrical structure when they are in the
meta/para-positions.
Another approach involves synthesizing thermosetting bismaleimides. They are
the leading class of polyimides with excellent thermal and mechanical properties,
which made them popular for use in the electronics industry and also in advanced
composites. They also exhibit excellent hot wet strength retention and fatigue re-
sistance [23]. Bismaleimides are defined as low molecular weight difunctional
monomers, which contain imide moieties in their backbone structure and have ter-
minal reactive groups (C=C), which undergo homo and/or copolymerization by
thermal treatment or in the presence of a catalyst. The double bonds present in
bismaleimides are highly electron-deficient, which can be self-polymerized ther-
mally to give a highly cross-linked network, exhibiting good thermal, chemical and
mechanical properties, superior to the epoxy resin [24, 25]. They have wide ap-
plications in electronic industries such as multilayered printed circuit boards for
computers [26–29] and in aerospace industries [30]. Bismaleimides bridge the tem-
perature performance gap between epoxy- and condensation-type polyimides. They
have become one of the important high performance thermosetting engineering
plastics. They have easy processing similar to epoxies, unlike the condensation-type
polyimides. Further, they can be fabricated into composite sheets by casting tech-
niques and also using fiber as reinforcement. One of the major drawbacks is their
Synthesis and Properties of Novel BMCMs, Polyaspartimides and Organosoluble Polyimides 45
brittleness (due to their high cross-link density), which restricts their utility. To over-
come these drawbacks, any dinucleophilic reagent can be added to the bismaleimide
(thereby reducing the cross-link density). Polyaspartimides can be prepared by
reacting a diamine with a bismaleimide. The main aim of bismaleimide/diamine
concept was to increase the molecular weight between cross-links, and thus to im-
prove the ductility of the bismaleimides.
Hence, in this study, a set of new diamines containing different substituents
(Cl, Br, F, –OCH2 C6 H5 and –OCH3 ) at the meta/para positions in the phenyl
ring, namely bis(4-amino-3,5-dimethyl phenyl) (X) phenyl methane (where
X = 3 -chloro, 3 -bromo, 3 -benzyloxy, 4 -chloro, 4 -fluoro and 4 -methoxy) [30]
and bis(4-amino-3,5-dimethyl phenyl) anthramethane were synthesized and sub-
sequently polycondensed with aromatic dianhydrides to produce a series of poly-
imides. The polymers were analyzed for solubility, moisture absorption and thermal
and dielectric properties. Using the same amines, organosoluble bismaleimides
and thermoplastic polyaspartimides were synthesized. The presence of Cl, Br, F,
–OCH2 C6 H5 and –OCH3 groups would increase solubility by enhancing the free
volume and polar interactions with solvents [31–36].
2. Experimental
2.1. Materials
m-Bromo benzaldehyde, m-chloro benzaldehyde, p-chloro benzaldehyde, p-fluoro
benzaldehyde, p-methoxy benzaldehyde, m-benzyloxy benzaldehyde, maleic an-
hydride, p-phenylene diamine, 4,4 -diamino diphenyl ether and 9-anthraldehyde
were purchased from E-Merk, India/Germany and used without any purifica-
tion; 2,6-dimethyl aniline (Aldrich), pyridine diamine (Spectrochem, India) and
hydrochloric acid (HCl; Spectrochem) were used as received; pyromellitic di-
anhydride (PMDA; E-Merk, purified by sublimation), 3,3 ,4,4 -benzophenone
tetracarboxylic acid dianhydride (BPTDA; Sigma) and oxydiphathalic anhy-
dride (ODPA; E-Merk) were recrystallised from acetic anhydride. N-methyl-
2-pyrrolidone (Aldrich) and N,N-dimethylformamide (DMF; E-Merk) were purified
by refluxing with calcium hydride, distilled under reduced pressure, and stored over
4-Å molecular sieves.
2.2. Synthesis of Diamines
The diamines, namely bis(4-amino-3,5-dimethyl phenyl)3 -chloro phenyl methane
(BACM), bis(4-amino-3,5-dimethyl phenyl)3 -bromo phenyl methane (BABM),
bis(4-amino-3,5-dimethyl phenyl)4 -fluoro phenyl methane (BAPFM), bis(4-
amino-3,5-dimethyl phenyl)4 -chloro phenyl methane (BAPCM), bis(4-amino-
3,5-dimethyl phenyl)3 -benzyloxy phenyl methane (BABOM), bis(4-amino-3,5-
dimethyl phenyl) anisyl methane (BAAIM) and bis(4-amino-3,5-dimethyl phenyl)
anthramethane (BAARM) were prepared adopting a common procedure since the
starting compounds differ only with respect to the substituents. The systematic
procedure for the preparation of bis(4-amino-3,5-dimethyl phenyl)4 -chloro phenyl

methane is given below.
2.2.1. Bis(4-amino-3,5-dimethyl phenyl) (X) Phenyl Methane
A three-necked flask equipped with a reflux condenser, a N2 inlet, and a bent in-
let for dry HCl was charged with 2,6-dimethyl aniline (0.04 mol, 4.84 g). Dry HCl
(0.045 mol, 5 ml) was passed into the reaction vessel for 30 min. The solid obtained
was melted by heating around 100◦ C. To this solution, the respective substituted
benzaldehyde/anthraldehyde (0.018 mol) was added dropwise. After complete ad-
dition, the reaction mixture was stirred at room temperature for 1 h. Then, the
temperature was slowly raised and maintained at 120◦ C for 6 h. The reaction mix-
ture was cooled and the pH was brought to 7 by adding sodium hydroxide solution,
the solid product obtained was filtered, washed repeatedly with methanol, and dried
in a vacuum oven at 60◦ C for 12 h.
2.2.2. Bis(4-amino-3,5-dimethyl phenyl) Anisyl Methane (BAAIM)
A three-necked flask equipped with a reflux condenser, a nitrogen inlet and bent
inlet for dry HCl was charged with 2,6-dimethyl aniline (12.1 g, 0.1 mol). Dry
HCl gas free from moisture (0.11 mol, 12.2 ml) was passed into the flask for about
90 min. The hydrochloride salt thus formed was dissolved in required volume of
DMF. To this solution, p-methoxy benzaldehyde (6.8 g, 0.05 mol) was added drop-
wise. After complete addition, the reaction mixture was stirred at room temperature
for 4 h. Then, the temperature was slowly raised and maintained at 110◦ C for
another five hours. The resulting solution was then cooled and neutralized with
sodium carbonate. The solid product obtained was filtered, washed repeatedly with
methanol and dried in a vacuum oven at 70◦ C for 12 h. The yield obtained was about
87%. IR (KBr): 3393 cm−1 and 3432 cm−1 (asymmetric and symmetric stretch-
ing vibrations of N–H group, 1622 cm1 (N–H bending vibration), 2963 cm−1 and
2925 cm−1 (C–H stretch of CH3 ). 1 H-NMR (CDCl3 , 200 MHz, δ) 2.1 (s, 12H)
CH3 , 5.1 (s, 4H) NH2 , 3.8 (s, 3H) OCH3 , 6.8 (s, 4H) phenyl protons. Elemental
analysis data calculated for C24 H28 N2 O, C–79.79%, H–7.82%, N–7.77%; found
C–79.86%, H–7.80%, N–7.74%. Melting point 177◦ C.
2.2.3. Bis(4-amino-3,5-dimethyl phenyl) Anthramethane (BAARM)
The diamine was prepared by using the procedure which was already reported by us
[17] as shown in Scheme 1. A three-necked flask equipped with a reflux condenser,
a nitrogen inlet, and a bent inlet for dry HCl was charged with 2,6-dimethylaniline
(24.2 g, 0.2 mol). Dry HCl gas (22.7 ml, 0.25 mol) was passed into the reaction ves-
sel for 90 min. The solid obtained was dissolved in the required quantity of DMF. To
this solution, anthraldehyde solution (20.6 g, 0.1 mol) in DMF was added dropwise.
After complete addition, the reaction mixture was stirred at room temperature for
4 h. Then, the temperature was slowly raised and maintained at 110◦ C for another
5 h. The reaction mixture was cooled and the pH was brought to 7 using sodium car-
bonate. The solid product obtained was filtered, washed repeatedly with methanol
and dried in a vacuum oven for 24 h at 65◦ C. Yield 83%, IR (KBr) cm−1 : 3451
X Cl Br H H OCH2 C6 H5 H
X H H F Cl H OCH3
Monomer BACM BABM BAPFM BAPCM BABOM BAAIM
Scheme 1. Synthesis of diamines.
and 3380 asymmetric and symmetric stretch of NH; 1613 N–H bending vibrations;
2960, 2914 C–H stretch of CH3 group; the absence of a band at 1664 cm−1 due
to C=O stretch of anthraldehyde shows the completion of the reaction. 1 H-NMR
(CDCl3 ), 2.16 d (s, 12H) CH3 , 4.8 d (s, 4H) NH2 , 6.9 d (s, 4H) phenyl protons,
7.2–8.4 d (m, 9H), elemental analysis as calculated for (C31 H30 N2 )n : C–86.5%,
H–6.98%, N–6.51%. Found: C–86.4%, H–6.96%, N–6.50%, MP 182◦ C.
2.3. Synthesis of Polyimides
2.3.1. One-Step Method
PMDA (0.01 mol, 2.18 g) was added immediately to a solution of diamine BAPCM
(0.01 mol, 3.64 g) in 15 ml of m-cresol and stirred at room temperature for 2 h.
Then 1 ml of isoquinoline (catalyst) and 20 ml of toluene were added and re-
fluxed at 180◦ C for 5 h while the water formed was azeotropically removed. The
reaction mixture was cooled and poured into methanol, and the polymer was col-
lected, washed in hot methanol, and dried in vacuum at 140◦ C for 12 h. All other
polyimides were synthesized by the same procedure using different diamines and
dianhydrides.
2.3.2. Two-Step Method
(a) Chemical imidization. PMDA (0.01 mol, 2.18 g) was added to a solution
(0.01 mol, 3.64 g) of BAPCM in 12 ml of DMF and then stirred at room temperature
for 12 h. Then 8 ml of acetic anhydride and 5 ml of pyridine were added, stirred
for 3 h at 100◦ C and the polymer precipitated in methanol and dried in vacuum at
100◦ C for 12 h.
(b) Thermal imidization. A three-necked flask equipped with a nitrogen inlet and
a reflux condenser was charged with the solution of 8.64 g (0.02 mol) of the pre-
pared diamine (BAPCM) in DMF. Under nitrogen atmosphere 4.36 g (0.02 mol) of
PMDA was added at once to avoid side reactions. The reaction mixture was stirred
at room temperature for 15 h; a viscous poly (amic acid) solution was formed. The
reaction temperature was slowly raised to 140◦ C and maintained at this temperature
for 6 h, then at 160◦ C for 6 h and finally at 180◦ C for 5 h to complete imidization.
38 ml of toluene was added and refluxed and the water formed during the reac-
tion was removed by Dean-Stark apparatus. The reaction mixture was cooled and
poured into methanol and the solid was collected by filtration, and dried in vacuum
at 70◦ C for 24 h.
(c) Film casting. Polyimide films were prepared by spreading poly (amic acid)
(prepared by stirring diamine and dianhydride in DMAc solution at room temper-
ature for 18 h) on a 9-cm glass culture dish and heating at 80◦ C overnight. The
semidried film was further dried and transformed into polyimide film by heating
at 120◦ C for 45 min, 150◦ C for 30 min, 180◦ C for 30 min, 250◦ C for 30 min and
300◦ C for 60 min.
2.4. Synthesis of Bismaleimides
Various bis(4-maleimido-3,5-dimethyl phenyl) substituted (X) phenyl methane
(X = 3 -chloro, 3 -bromo, 4 -fluoro, 4 -chloro, 3 -benzyloxy and 4 -methoxy)
and bis(4-maleimido-3,5-dimethyl phenyl) anthramethane were prepared from the
corresponding substituted diamines and maleic anhydride. A three-necked flask
equipped with an addition funnel and a nitrogen inlet was charged with a solution
of synthesized diamine (0.05 mol) in THF. Maleic anhydride (9.81 g, 0.1 mol) dis-
solved in THF was added dropwise to the above mixture. The solution was stirred
for 4 h at room temperature and at 70◦ C for additional 6 h. The precipitate of bis-
maleiamic acid was collected, washed with fresh THF to remove excess reactant,
and then it was dried in vacuum at 60◦ C for 12 h. IR (KBr, cm−1 ), 3320 (N–H),
1720 (C=O stretching), 1612 (N–H bending). 1 H-NMR–(CDCl3 ), d = 10.44–
10.50 (s, 2H, COOH). 0.05 mol of bismaleamic acid in 20 ml of DMF was charged
into a 250-ml round-bottomed flask. The solution was maintained at 75◦ C for 3 h
with stirring. Acetic anhydride (25 ml) and 0.25 g of sodium acetate were added to
the preheated solution. The temperature of the reaction mixture was raised to 90◦ C
and then stirred for 6 h. The reaction mixture was poured into cold water to yield
bismaleimide. All other bismaleimides were synthesized by adopting the similar
procedure.
2.5. Synthesis of Polyaspartimides
A 150-ml three-necked flask fitted with a N2 inlet, a reflux condenser, and a ther-
mometer was charged with (0.01 mol) of BMI in 15 ml of m-cresol. The solution
was stirred at room temperature, and when the BMI was completely dissolved,
1.08 g (0.01 mol) of diamine (p-phenylene diamine or 2,6-diaminopyridine or oxy-
dianiline) and 0.5 ml of glacial acetic acid were added into the flask. The reaction
mixture was stirred at 100◦ C for 96 h. The resulting viscous solution was poured
into excess ethanol with vigorous stirring and the polymer thus precipitated out
from the non-solvent was filtered, washed thrice with hot ethanol, and dried under
vacuum at 60◦ C overnight. All other polymers were prepared by following the same
procedure mentioned above.
2.6. Measurements
Fourier Transform Infrared (FT-IR) spectra were recorded on a Nicolet spectrom-

eter with KBr pellet. 1 H-NMR and 13 C-NMR spectra were recorded using JEOL
Ex-400 spectrometer with CDCl3 as the solvent. The 1 H chemical shifts were cali-
brated using tetramethylsilane (TMS). Elemental analysis was performed on a Carlo
Erba EA 1108 microanalyzer; DSC and TGA were performed with a Perkin–Elmer
7 series thermal analyzer system at a heating rate of 10◦ C/min. The inherent viscos-
ity was measured with a Ubbelohde viscometer using a 0.5 g/dl solution in NMP at
30◦ C. Solubility was noted visually by mixing 0.05 g of the polyimide sample (in
both the one- and two-step (thermal imidization) methods) with 8 ml of the solvent
at room temperature. X-ray diffraction measurements were performed on a Philips
PW 1730-10 X-ray diffractometer using Cu Kα radiation. Polyimides prepared as
described above were used for FT-IR, 1 H-NMR, DSC, TGA, elemental analysis,
inherent viscosity, and solubility measurements. Polyimide films prepared accord-
ing to the procedure described above were used for studying the mechanical and
electrical properties. The mechanical properties of the films were measured with
an Instron model 1130 tensile tester with a 5-kg load cell at a cross-head speed of
0.5 mm/min on strips with a thickness of approximately 0.15–0.20 mm, a width of
0.5 cm, and a gauge length of 2 cm. Dielectric properties were studied on polymer
films by the parallel plate capacitor method using an HP-4194A Impedance/Gain
phase analyzer at frequencies of 1 and 10 kHz. Gold electrodes were vacuum-
deposited on both surfaces of the dried films (5 cm × 0.5 cm × 0.15–0.20 mm).
Moisture absorption was measured by immersing the specimens in water at room
temperature for 3 days, and the weight difference was determined. The UV–vis
spectra of the polymers were recorded on a Shimadzu UV–visible spectrophotome-
ter.

3.1. Monomer Synthesis
The various diamines, namely bis(4-amino-3,5-dimethyl phenyl)3 -chloro phenyl
methane, bis(4-amino-3,5-dimethyl phenyl)3 -bromo phenyl methane, bis(4-
amino-3,5-dimethyl phenyl)4 -fluoro phenyl methane, bis(4-amino-3,5-dimethyl
phenyl)4 -chloro phenyl methane, bis(4-amino-3,5-dimethyl phenyl)3 -benzyloxy
phenyl methane bis(4-amino-3,5-dimethyl phenyl) anisyl methane and bis(4-
amino-3,5-dimethyl phenyl) anthramethane were prepared from the corresponding
substituted benzaldehyde and 2,6-dimethyl aniline as shown in Scheme 1 and
their structures were confirmed from elemental analysis, FT-IR, and 1 H-NMR and
13 C-NMR spectra. The experimental elemental analysis data obtained coincide well
with theoretical values confirming the proposed structure. The diamine monomers
show absorption bands around 3376 and 3487 cm−1 due to N–H symmetric and
asymmetric stretchings of NH2 group and the absence of band at 1664 cm−1 con-
firms the absence of aldehydic carbonyl group. All the monomers show absorptions
around 2970 and 2921 cm−1 due to C–H stretching vibration of CH3 group. The
1 H-NMR spectra also confirm the structure of the diamines. The signal at 4.8 ppm
corresponds to amino protons and that at 2.16 ppm corresponds to the twelve pro-
tons of the methyl group. The 13 C-NMR spectra of monomers show distinct signals
for all the different carbon atoms (representative spectrum is given in Fig. 1).
Figure 1. 13 C-NMR spectrum of BABOM.

3.2. Synthesis of Polyimides

Polyimides are generally prepared by one-step and/or two-step solution polycon-
densation methods (Scheme 2). In the present study, polyimides were synthesized
according to both methods for the sake of comparison. The one-step method is very
useful in the polymerization of non-reactive dianhydrides and diamines. Using this
Scheme 2. Preparation of polyimides.

Figure 2. FT-IR spectrum of polyimide PI1a (BACM–PMDA).
procedure in the present study, the diamine and dianhydride were reacted in a N2
atmosphere at elevated temperature (180◦ C) in m-cresol, using isoquinoline as the
catalyst. Sterically hindered monomers react rapidly at elevated temperatures to
give high-molecular-weight polyimides because an excellent solvation of a poly-
mer at high temperature may allow it to have the desired conformation for close
packing. All polymers obtained by this method had a good yield. The one-step
method does not have an intermediate poly(amic acid) stage, and so chain growth
and imidization occur simultaneously at elevated temperature. All the monomers
were sufficiently reactive to produce polyimides of high molecular weight, as ev-
ident from their inherent viscosity values. The monomer BAPFM, which is less
basic than normal aromatic amines because of inductive deactivation of the fluoro
group, was still sufficiently reactive to produce high-molecular-weight polyimides.
All the polyimides afforded good quality and creasable films. The spectroscopic
data for the polyimide samples prepared by the two-step method (thermal and chem-
ical imidization) were identical to those derived by the one-step method, so the
data are not presented separately. The polyimides were characterized by elemental
analysis, FT-IR and 1 H-NMR. The elemental analysis data for all the polyimides
were in good agreement with their calculated values. Figure 2 shows the FT-IR
spectrum of a representative polyimide (BACM–PMDA). The absorptions around
1778 and 1728 cm−1 were a result of asymmetric and symmetric stretchings of
the C=O group of the imide. The strong bands at 2960 and 2924 cm−1 were a re-
sult of C–H stretching of the CH3 group, an intense band at 1364 cm−1 was from
C=N–O stretching of the imide ring and the band at 1114 cm−1 was from imide
ring deformation. The 1 H-NMR spectrum of the representative polyimide (PI1a ) is
shown in Fig. 3. All the polyimides show a signal around 2.17 ppm for 12 methyl
protons as well as in the range 6.96–8.50 ppm. The various substituted diamines
Figure 3. 1 H-NMR spectrum of polyimide PI1a (BACM–PMDA).
were then reacted with required amount of maleic anhydride to give bismaleamic
acid (I). The bismaleamic acid was then cyclodehydrated using cyclodehydrating
agents (acetic anhydride and sodium acetate) as shown in Scheme 3. The struc-
ture was confirmed by elemental analysis, FT-IR, 1 H and 13 C-NMR. The elemental
analysis data were found to be in good agreement with the proposed structure. The
representative infrared spectrum of bismaleimide (BMCM, Fig. 4) shows a strong
absorption at 1715 cm−1 and a weak absorption at 1776 cm−1 due to asymmet-
ric and symmetric C=O stretching vibrations of the imide ring, respectively. The
absence of band in the region 3300–3500 cm−1 shows that bismaleamic acid was
completely converted into bismaleimide and the absence of band at 1545 cm−1 due
to N–H group of amide linkage also confirms a complete ring formation. The band
at 692 cm−1 is due to C=C of maleimide ring. The band at 1386 cm−1 is due to
C–N–C stretch and the band around 3100 is due to =C–H group of bismaleimide
ring. The 1 H-NMR spectrum (Fig. 5) of bismaleimide (BMCM) shows the distinct
signal at 7.26 ppm due to four olefinic protons and the absence of signal at 10.5 ppm
due to carboxylic protons shows a complete imidization and the signal at 2.18 ppm
is due to 12 protons of methyl group. All other aromatic protons are accountable
in the region of 6.9–7.08 ppm and the 13 C-NMR spectra show distinct signals for
all the carbon atoms in the region 17.8–169.8 ppm. The representative 13 C-NMR
Scheme 3. Synthesis of bismaleimides and polyaspartimides.
spectrum of bismaleimide (BMCM) is shown in Fig. 6. All other bismaleimides

show signals in the same region, ensuring the formation of bismaleimide.
3.3. Preparation of Polyaspartimides
The nucleophilic addition of diamines to aromatic bismaleimide, the Michael type
addition, is a well-known route to synthesize linear polyaspartimides. The synthe-
sized bismaleimides and an equivalent amount of aromatic diamines were stirred
together in m-cresol containing a catalytic amount of glacial acetic acid to promote
the polymerization as shown in Scheme 3. The structure of the polymer was con-
firmed by elemental analysis and FT-IR. The calculated and found values of all
Figure 4. FT-IR spectrum of bismaleimide (BMCM).
Figure 5. 1 H-NMR spectrum of bismaleimide (BMCM).
the polymers were in good agreement with the proposed structure suggesting the
formation of polyaspartimide. The yield obtained was also good. The structure of
polymers was confirmed by IR spectroscopy. The representative IR spectrum of
polyaspartimide (PAS1a ) is shown in Fig. 7. The bands around 3339–3365 cm−1
are due to N–H stretching vibration. The disappearance of band at 691 cm−1 due to
Figure 6. 13 C-NMR spectrum of bismaleimide (BMCM).
Figure 7. FT-IR spectrum of polyaspartimide PAS1a (BMCM + PPD).

maleimide C=C bond confirms completion of the addition reaction of these double
bonds. The bands around 1773–1780 cm−1 and 1709–1720 cm−1 are due to asym-
metric and symmetric stretching vibrations of carbonyl group. The bands around
1360–1382 cm−1 are due to C–N–C stretching vibration of the imide ring. The
methyl group shows the band around 2917–2926 cm−1 due to C–H stretching vi-
bration. The Cl, Br and F substituents show characteristic bands around 1089, 1072
and 1157 cm−1 , respectively. The spectroscopic data are well in agreement with the
expected structure, ensuring the formation of polyaspartimides.
3.4. Properties of Polyimides
3.4.1. Solubility
Most of the polyimides obtained via both one-step and two-step methods (chem-
ical imidization) are soluble in polar aprotic solvents such as NMP, DMF, DMAc
etc. (Table 1). They are soluble even in low boiling solvents such as chloroform
and THF. Polyimides obtained by chemical imidization exhibit similar solubility
behavior as polyimides prepared by the one-step method, so separate data are not
Table 1.
Solubility behavior of polyimides prepared by one-step and two-step (thermal imidization) methods
Code Polyimide NMP DMF DMSO DMAc THF CHCl3
PI1a BACM + PMDA ++ (±) ++ (±) ++ (−) ++ (±) ± (−) ++ (−)

PI1b BACM + BTDA ++ (++) ++ (++) ++ (±) ++ (++) ++ (±) ++ (±)
PI1c BACM + ODPA ++ (++) ++ (++) ++ (+) ++ (++) ++ (+) ++ (+)
PI2a BABM + PMDA ++ (±) ++ (±) ++ (−) ++ (±) ± (−) ++ (−)
PI2b BABM + BTDA ++ (++) ++ (++) ++ (±) ++ (++) ± (−) ++ (−)
PI2c BABM + ODPA ++ (++) ++ (++) ++ (++) ++ (++) ++ (+) ++ (+)
PI3a BAPFM + PMDA ++ (±) ++ (±) ++ (−) ++ (±) ++ (−) ++ (−)
PI3b BAPFM + BTDA ++ (++) ++ (++) ++ (++) ++ (++) ++ (++) ++ (++)
PI3c BAPFM + ODPA ++ (++) ++ (++) ++ (++) ++ (++) ++ (++) ++ (++)
PI4a BAPCM + PMDA ++ (±) ++ (±) ++ (−) ++ (±) ± (−) ++ (−)
PI4b BAPCM + BTDA ++ (++) ++ (++) ++ (±) ++ (++) ++ (±) ++ (±)
PI4c BAPCM + ODPA ++ (++) ++ (++) ++ (+) ++ (++) ++ (+) ++ (+)
PI5a BABOM + PMDA ++ (±) ++ (±) ± (−) ++ (±) ++ (−) ++ (−)
PI5b BABOM + BTDA ++ (++) ++ (++) ± (++) ++ (++) ++ (+) ++ (+)
PI5c BABOM + ODPA ++ (++) ++ (++) ++ (++) ++ (++) ++ (++) ++ (++)
PI6a BAAIM + PMDA ++ (++) ++ (++) ± (++) ++ (++) ++ (++) ++ (++)
PI6b BAAIM + BTDA ++ (++) ++ (++) ± (++) ++ (++) ± (−) ++ (++)
PI6c BAAIM + ODPA ++ (++) ++ (++) ± (++) ++ (++) ++ (++) ++ (++)
PI7a BAARM + PMDA ++ (++) ++ (++) ++ (++) ++ (++) − ++ (++)
PI7b BAARM + BTDA ++ (++) ++ (++) ++ (++) ++ (++) ± ++ (++)
PI7c BAARM + ODPA ++ (++) ++ (++) ++ (++) ++ (++) ++ ++ (++)
Data in parentheses are for polyimides prepared by the two-step method (thermal imidization),
++ completely soluble, + soluble on heating, ± sparingly soluble on heating, − insoluble.
given. The excellent solubility may be due to the presence of bulky pendant substi-
tuted phenyl group, which decreases polymer inter- and intra-chain interactions and
disturbs the co-planarity of the aromatic units to reduce the packing efficiency and
crystallinity [37]. The solubility of even those polyimides with rigid pyromellitic
unit in the test solvents is good indicating that the incorporation of bulky pen-
dant group is one of the best methods to improve solubility. The presence of halo
substitution in the pendant phenyl group also enhances the solubility by the polar
interactions with the solvents. The polyimides prepared from thermal imidization
(two-step method) exhibit very poor solubility (especially PMDA-derived poly-
imides), which may be due to partial cross-linking of polyimide during imidiza-
tion [38].
3.4.2. Inherent Viscosity
The inherent viscosity values of polyimides measured in NMP (Table 2) vary
depending on the type of bridging between the phthalimide group in the dianhy-
Table 2.
Inherent viscosity (η) and optical properties of polyimides
Code Polyimide η (dl/g) λ0 Transmit-

(nm) tance
One-step Thermal Chemical
(%)
imidization imidization
PI1a BACM + PMDA 1.33 − 1.02 361.3 83.7

PI1b BACM + BTDA 1.27 0.65 0.94 338.0 85.4
PI1c BACM + ODPA 1.22 0.60 0.90 312.0 92.0
PI2a BABM + PMDA 1.28 − 1.05 358.8 71.4
PI2b BABM + BTDA 1.19 0.68 0.97 338.2 78.5
PI2c BABM + ODPA 1.10 0.64 0.94 322.0 81.2
PI3a BAPFM + PMDA 1.24 − 0.99 341.7 83.3
PI3b BAPFM + BTDA 1.20 0.68 0.92 335.2 88.7
PI3c BAPFM + ODPA 1.09 0.54 0.90 302.8 92.6
PI4a BAPCM + PMDA 1.33 − 1.01 347.0 83.3
PI4b BAPCM + BTDA 1.27 0.67 0.94 329.4 88.7
PI4c BAPCM + ODPA 1.21 0.62 0.91 319.0 92.6
PI5a BABOM + PMDA 1.31 − 0.97 329.4 89.2
PI5b BABOM + BTDA 1.25 0.62 0.92 318.6 90.0
PI5c BABOM + ODPA 1.15 0.58 0.89 302.6 95.2
PI6a BAAIM + PMDA 0.63
PI6b BAAIM + BTDA 0.82
PI6c BAAIM + ODPA 0.45
PI7a BAARM + PMDA − 0.63 − − −
PI7b BAARM + BTDA − 0.82 − − −
PI7c BAARM + ODPA − 0.45 − − −
λ0 — cut-off wavelength.
dride part of the polyimide and on the stiffness of the polymer chain. The vis-
cosity values also vary with the extent of polymer chain mobility. The presence
of methyl substituent on the benzene ring increases the segmental rigidity of the
polymer backbone (PI1a –PI5c ) resulting in increased solution viscosity. The inher-
ent viscosity values of polyimides containing PMDA are higher in comparison to
BTDA and ODPA-based polyimides. The inherent viscosity of all polyimides is
sufficiently high to permit casting of flexible and tough films. The inherent vis-
cosity values of polyimides prepared by thermal imidization (two-step method,
0.54–0.68 dl/g) are in fact lower than those of polyimides prepared from one-step
method (1.09–1.33 dl/g) and chemical imidization (0.89-1.33 dl/g) possibly due to
the fact that the molecular weight of poly (amic acid) drops during thermal cure of
the polyimide due to reverse propagation reaction [39]. The reverse propagation re-
action does not occur during chemical imidization and, hence, the molecular weight
decrease experienced in thermal imidization does not occur in chemical imidization.
The inherent viscosity values of polyimides prepared by one-step method are gener-
ally high, in comparison with two-step method. This is because in one-step method,
the excellent solvation of the polymer at high temperature provides enough energy
(even for sterically hindered monomers) to obtain a more favorable conformation
for packing [3] which results in higher viscosity compared to chemical imidization.
3.4.3. Optical Properties
The wholly aromatic polyimides strongly absorb visible light because of their aro-
matic conjugated structures, and the intermolecular or intramolecular charge trans-
fer complexes formed between or within polymer chains usually produce polymers
that are pale yellow to deep brown. The coloration of polyimides always obstructs
their widespread applications for which colorlessness and transparency are of big
concern. The colour intensity of polyimides changes from deep to pale depending
on the type of anhydride used as: deep PMDA > BTDA > pale ODPA, when the di-
amine structure is fixed. The colour intensity of polyimide could also be elucidated
from the cut-off wavelength (λ0 ) observed in UV–visible absorption spectra.
The UV–visible spectra of polyimides obtained from various diamines and di-
anhydrides at a concentration of 5 × 10−3 mol/l in NMP solution and the cut-off
wavelength and percent transmittance at 500 nm from the spectra are presented in
Table 2. All polyimides have λ0 lower than 400 nm and all the polyimides exhibit
high optical transparency (71–95% transmittance). The ODPA-based polyimide
forms fairly transparent films, in contrast to other dianhydride-based polyimides.
These results are attributed to the reduced intermolecular charge-transfer complex
(CTC) formation between alternating electron donor (diamine) and electron accep-
tor (dianydride) moieties. The light colour of the PIs (with high transmittances)
can be explained from the decreased intermolecular interactions. The bulky phenyl
group and halo substitution are effective in decreasing CTC formation between
polymer chains through steric hindrance and the inductive effect (F > Cl > Br)
(by decreasing the electron donating property of diamine moieties). The decrease
in the extent of intermolecular CTC formation may also be due to changes in the
structure of the respective diamines. Among the polyimides, those containing ben-
zyloxy units exhibit higher % transmittance due to their efficiency in increasing the
free volume by steric hindrance [40]. Polyimides derived from PMDA show strong
absorption with a high cut-off wavelength due to the electronic conjugation in the
structure.
3.4.4. Moisture Absorption and Dielectric Properties
The dielectric constants of all the polyimides here are lower (Table 3) in comparison
to commercial polyimides such as Kapton films (3.94 at 10 kHz). The polyimide
films of 0.15–0.20 mm thickness were used in this study. The decreased dielectric
constant may be attributed to the pendant group and halo substituent present in it,
which resulted in less efficient chain packing and increased free volume.
The presence of fluorine substitution in the polyimides PI3a –PI3c decreases the
dielectric constant further, since strong electronegativity of fluorine may result in
very low polarizability of the C–F bonds, thereby decreasing the dielectric con-
stant [41]. The dielectric constant varies with respect to the dianhydride in the order
ODPA–PI > PMDA–PI > BTDA–PI. The lowest dielectric constant for BTDA–PI
Table 3.
Electrical and mechanical properties of polyimides
Polymer Polyimide Dielectric constant Moisture Tensile Elon- Tensile

code absorption strength gation modulus
1 kHz 10 kHz
(%) (MPa) (%) (GPa)
PI1a BACM + PMDA 3.51 3.48 0.84 99 9 2.1
PI1b BACM + BTDA 3.49 3.47 1.00 94 10 1.8
PI1c BACM + ODPA 3.67 3.54 0.86 92 11 1.5
PI2a BABM + PMDA 3.61 3.59 0.74 118 8 2.2
PI2b BABM + BTDA 3.62 3.60 1.10 95 9 2.1
PI2c BABM + ODPA 3.70 3.69 0.78 92 9 2.0
PI3a BAPFM + PMDA 3.31 3.16 0.56 111 8 2.3
PI3b BAPFM + BTDA 3.29 3.14 0.70 97 9 2.2
PI3c BAPFM + ODPA 3.42 3.29 0.52 95 12 2.1
PI4a BAPCM + PMDA 3.52 3.48 0.70 101 9 2.1
PI4b BAPCM + BTDA 3.46 3.44 1.00 96 11 1.9
PI4c BAPCM + ODPA 3.66 3.53 0.70 94 12 1.6
PI5a BABOM + PMDA 3.72 3.65 0.98 102 12 2.0
PI5b BABOM + BTDA 3.68 3.64 1.10 94 11 1.7
PI5c BABOM + ODPA 3.80 3.76 1.00 84 10 2.7
PI6a BAAIM + PMDA 2.58 − 0.68 92 13 1.7
PI6b BAAIM + BTDA 3.21 − 1.01 99 10 2.1
PI6c BAAIM + ODPA 3.31 − 0.66 94 11 1.8
PI7a BAARM + PMDA − − − − − −
PI7b BAARM + BTDA 2.92 − 0.81 86.2 3.2 4.15
PI7c BAARM + ODPA 2.98 − 0.69 66.2 4.23 4.10
may be attributed to the presence of carbonyl group with high electron-withdrawing

nature, decreasing the electron polarizability, hindering the electronic interaction
and hence reducing the dielectric constant [42].
On the other hand, the strong electron-attracting property resulting from the high
electronegativity of halogen atoms (F > Cl > Br) could induce additional perma-
nent dipole in the benzene ring connected to it. These dipoles may increase the
polarization and hence the dielectric constant. The experimental data indicate that
the effects of these additional permanent dipoles are smaller than the other effects,
i.e., chain–chain interaction, free volume and the electronic polarization in poly-
imide [43].
Polyimide materials usually show higher moisture uptake than the hydrocarbon
polymers because of the presence of polar imide group. The equilibrium moisture
absorption of polyimide films was measured by immersing the specimens at 25◦ C
in deionized water for 3 days and weight difference in the film before and after
immersion was determined. The moisture absorption of aromatic polyimides can
be up to 3.0–3.5% depending on the chemical structures of polymer and the rela-
tive humidity of the surrounding environment, which have significant influence on
the dielectric properties of polymer [24]. The moisture absorption of polyimides
might be related to several factors including chemical structure, introduction of
halo groups and other functional groups, the geometrical packing of the polymer
chains as well as film processing parameters and so forth in which the chemi-
cal structure/presence of functional groups in the polymer might be major factors.
The presence of bulky groups could loosen the packing of the polyimide backbone
chains, resulting in increase of free volume in the polymer that makes the poly-
mer to entrap a lot of water molecules. The presence of fluoro group introduces
hydrophobic feature, which inhibits the adsorption of water molecules on the sur-
face of the fluorinated polymers. The former effect is predominating in polyimides
with Cl, Br and benzyloxy substituents as is evident from the results obtained. How-
ever, in fluoro-substituted polyimides the latter effect is more predominant resulting
in lower moisture absorption than the other halo (Cl, Br) substituted polyimides.
Though substitution of Cl and/or Br in the polyimide is expected to reduce the mois-
ture absorption as well as dielectric constant, due to electron polarizability, they are
not sufficiently electronegative to really reduce dielectric constant and moisture ab-
sorption. In addition, other functional groups in the polymer chain also affect the
moisture uptake. It seems that ether bridged polyimides show lower moisture ab-
sorption than those linked with carbonyl group, probably C=O group would play
a major role in increasing the moisture absorption by forming hydrogen bonding
with water molecule. Similar observations have been reported earlier [44, 45]. The
order of moisture absorption by the different polyimides was found to be in the
order BTDA–PI > ODPA–PI > PMDA–PI.
3.4.5. Mechanical Properties
The tensile properties of polyimide films were determined using an Instron UTM
(Table 3). All the polyimides exhibit good tensile strength (84–118 MPa), % elonga-
tion (8–12%) and modulus (1.5–2.3 GPa), enhancing their utility in semiconductor
applications. All the polyimides exhibit distinct yield point in their stress–strain
curves and moderate elongation at break, indicating high toughness. The BTDA-
based polyimides and ODPA-based polyimides exhibit lower modulus than PMDA-
based polyimides due to the inherent rigidity of the pyromellitic unit. In general,
the modulus, which is the measure of mechanical hardness, decreases in the order
PMDA–PI > BTDA–PI > ODPA–PI. % Elongation, which is a measure of flexi-
bility and toughness decreases in the order ODPA–PI > BTDA–PI > PMDA–PI.
In contrast to the modulus behaviour, elongation at break for BTDA–PI is almost
comparable to that of ODPA–PI that exhibits high chain ordering. In addition,
ether-linked ODPA–PI with irregular chain–chain ordering shows strain at break
comparable to BTDA–PI. This result indicates that both high chain ordering and
limited chain flexibility directly affect the mechanical properties [46].
3.4.6. Thermal Properties
The thermal properties of polyimides were evaluated by DSC and TGA and the
data are presented in Table 4. Polymer series 1, 2, 3 and 4 undergo a single-stage
decomposition. All the polymers are stable up to 400◦ C. The methyl substituent is
expected to degrade at lower temperature and the backbone at a higher temperature.
However, it does not occur; the decomposition of methyl groups occurs along with
the main chain at about 400◦ C [44]. The temperature at which 10% weight loss of
the polymer occurs (T10 ) is very high, indicating high thermal stability of polymers.
The thermal stability of the polyimides with various dianhydrides decreases in the
order PMDA–PI > BTDA–PI > ODPA–PI. The PMDA-based polyimides exhibit
higher thermal stability in comparison to BTDA and ODPA-based polyimides due
to inherent rigidity of PMDA unit. The presence of flexible linkages like –O– and
C=O contributes to the decrease in the thermal stability. Polyimides 5a–c show a
two-stage decomposition. The first stage occurs around 340◦ C. The char yield of the
polyimides which correlates to the ability to sustain combustion is high (40–59.2%)
at 800◦ C. The Tg values of the polyimides are in the range of 210–325◦ C (Ta-
ble 4). This can be correlated to the structure of the amine component and stiffness
of the polymer backbone. The presence of methyl substituent ortho to the imide
ring enhances the segmental rigidity of the polymer, thereby enhancing the Tg . The
PMDA-based polyimides exhibit higher Tg values in comparison to BTDA and
ODPA-based polyimides, since the presence of rigid pyromellitic unit enhances
the rigidity of the polymer chain and hence the Tg . Except PI5c all the polyimides
prepared in the present study exhibit higher Tg values compared to a commercial
polyimide, namely Ultem 1000 (Tg = 215◦ C).
3.5. Properties of Polyaspartimides
3.5.1. Inherent Viscosity
The inherent viscosity of polyaspartimides was determined using Ubbelohde vis-
cometer at a concentration of 0.5 g/dl in NMP and the results are presented in
Table 5. All the polyaspartimides show viscosity values which are adequate for the
Table 4.
Thermal properties of polyimides
Code Polyimide Tg T10 Char yield

(◦ C) (◦ C) (%)
PI1a BACM + PMDA 324 532 42.95

PI1b BACM + BTDA 322 528 45.85
PI1c BACM + ODPA 305 510 40.36
PI2a BABM + PMDA 325 491 57.98
PI2b BABM + BTDA 320 484 59.26
PI2c BABM + ODPA 310 452 53.16
PI3a BAPFM + PMDA 250 537 46.95
PI3b BAPFM + BTDA 248 532 52.80
PI3c BAPFM + ODPA 236 505 42.26
PI4a BAPCM + PMDA 321 469 43.59
PI4b BAPCM + BTDA 319 458 48.85
PI4c BAPCM + ODPA 302 434 42.36
PI5a BABOM + PMDA 238 453 54.93
PI5b BABOM + BTDA 225 440 53.08
PI5c BABOM + ODPA 210 405 50.26
PI6a BAAIM + PMDA 294 496 57.00
PI6b BAAIM + BTDA 279 476 59.00
PI6c BAAIM + ODPA 265 460 53.00
PI7a BAARM + PMDA 276 482 52.00
PI7b BAARM + BTDA 272 476 56.00
PI7c BAARM + ODPA 279 396 50.00
T10 — Temperature at which 10% weight loss was recorded by TGA at a heating rate of 10◦ C/min.
Char yield — residual weight % at 800◦ C in nitrogen.
fabrication of composites and films. The viscosity values of polyaspartimides are in

the range 0.40–0.86 dl/g. The viscosity values of polyaspartimides containing oxy
linkage are low because of the presence of flexible linkage present in the polymer
backbone, which decreases the stiffness of the polymer, hence low viscosity [47].
Polyaspartimides containing p-phenylene moiety exhibit higher viscosity in com-
parison to polyaspartimides containing pyridine because of the rigidity of phenylene
moiety.
3.5.2. Cure Behaviour of Bismaleimide
BMCM/p-phenylene diamine blend curing characteristics were studied by DSC
and the results are shown in Fig. 8. In the case of pure BMI, a sharp endothermic
peak at 250◦ C was due to the melting, and an exothermic peak observed at 348◦ C
was due to the polymerization reaction between the two double bonds. In the case
of BMCM/diamine adduct, DSC analysis was first carried out up to 120◦ C; then
the sample was quenched and re-scanned. The DSC curves show only one broad
endothermic peak at 185◦ C, probably due to chain extension reaction between bis-
Table 5.
Inherent viscosity (η) and thermal properties of polyaspartimides
Code Polyaspartimide Tg T10 Char yield η

(◦ C) (◦ C) (%) (dl/g)
PAS1a BMCM + PPD 258 437 43.46 0.51

PAS1b BMCM + DAP 245 416 49.27 0.46
PAS1c BMCM + ODA 215 408 84.20 0.40
PAS2a BMBM + PPD 264 424 44.20 0.54
PAS2b BMBM + DAP 252 420 46.23 0.50
PAS2c BMBM + ODA 220 410 40.60 0.45
PAS3a BMPFM + PPD 248 422 47.96 0.49
PAS3b BMPFM + DAP 235 405 47.08 0.46
PAS3c BMPFM + ODA 210 390 46.65 0.41
PAS4a BMPCM + PPD 275 440 42.27 0.50
PAS4b BMPCM + DAP 262 425 48.77 0.45
PAS4c BMPCM + ODA 225 410 31.95 0.42
PAS5a BMBOM + PPD 252 441 48.12 0.56
PAS5b BMBOM + DAP 238 421 44.71 0.50
PAS5c BMBOM + ODA 205 403 43.14 0.40
PAS6a BMAIM + PPD 246 434 62.4 0.41
PAS6b BMAIM + DAP 278 476 56.0 0.50
PAS6c BMAIM + ODA 235 432 54.2 0.30
PAS7a BMARM + PPD 238 464 64.2 0.53
PAS7b BMARM + DAP 270 486 51.3 0.86
PAS7c BMARM + ODA 220 443 56.0 0.48
Figure 8. Curing behaviour of BMCM/p-phenylene diamine.
maleimides and melted diamine, and the endothermic peak was shifted to lower
temperature. Similar results were obtained by Wang and Hwang [48]. The above
results show that chain extension occurs at lower temperature and homopolymer-
ization at higher temperature.
3.5.3. Thermal Properties
DSC measurements were carried out at a heating rate of 20◦ C/min in nitrogen. The
Tg values of the polymers are in the range 205–275◦ C (Table 5), depending on
the stiffness of the polymer chain. The representative DSC thermograms of polyas-
partimides PAS3a , PAS3b and PAS3c are shown in Fig. 9. Thermal properties of
all the polyaspartimides are summarized in Table 5. Representative TGA curves
of polyaspartimides PAS3a and PAS5a are shown in Fig. 10. All polyaspartimides
show a similar pattern of decomposition and do not show significant weight loss
below 350◦ C in nitrogen. The ODA-based polyaspartimides show single-stage de-
composition with 10% weight loss occurring in the range 390–410◦ C. The methyl
substituent and N–H unit in the polymer chain tend to decompose along with the
main chain at lower temperature as is evident from the thermogram showing a broad
single-stage weight loss. All other polymers show two-stage decomposition. The
initial weight loss occurs due to NH and methyl group and the second weight loss
occurs due to the degradation of the polymer chain at higher temperature. The T10
values of these polyaspartimides are in the range of 390–486◦ C. Thermal stabil-
ity of a polymer is determined by the strength of the weakest bond present. The
strength of this bond varies depending on the extent of conjugation in the polymer.
In general, an electron-withdrawing substituent decreases the electronic conjuga-
tion and hence the thermal stability, which is evident from the T10 value. The char
Figure 9. DSC thermograms of polyaspartimides PAS3a , PAS3b and PAS3c .

Figure 10. TGA curves of polyaspartimides PAS3a and PAS5a .
yield of the polyaspartimides is in the range of 31.95–62.4% when heated to 800◦ C

in nitrogen, confirming that these polymers have good thermal stability and good
self-extinguishing property.
Incorporation of methyl substituent in the bismaleimide tends to restrict the seg-
mental rigidity of the polymer, thereby enhancing the Tg . The presence of bulky
pendant phenyl unit also tends to enhance Tg by restricting the segmental motion of
the polymer chain [49]. Polyaspartimides containing p-phenylene moiety or pyri-
dine moiety have high Tg due to increased rigidity of the polymer chain. However,
in the case of polyaspartimides containing ether linkages, a lower Tg was observed
because of the increased rotational movement caused by the flexible ether linkage
between two phthalimide units [40]. The introduction of the pendant phenyl group
with halo substituent resulted in only small changes in the Tg values obtained from
DSC analysis. The effect of Cl, Br, and F substituents on the Tg was studied. The
Tg values vary slightly and the variation is due to size effect, which directly influ-
ences the packing density, polymer aggregation and interaction of polymer chains
and hence the Tg [47]. However, the effect of meta and para substitutions on the Tg
shows that p-Cl imparts more aggregation compared to meta substituents (due to
higher increase in free volume) as is evident from the Tg values. Polyaspartimides
containing p-phenylene moiety show higher Tg in comparison to the polyaspar-
timides containing pyridine moiety, which may be due to the higher rigidity and
better packing of the polymer chain.
3.5.4. Solubility
The solubility of polyaspartimides was tested in various organic solvents and the
results are summarized in Table 6. The polyaspartimides containing p-phenylene
Table 6.
Solubility behaviour of polyaspartimides
Code Polyaspartimide NMP DMF DMSO DMAc THF CHCl3
PAS1a BMCM + PPD ++ ++ + ++ − −

PAS1b BMCM + DAP ++ ++ + ++ − −
PAS1c BMCM + ODA ++ ++ ++ ++ ++ ++
PAS2a BMBM + PPD ++ ++ ± ++ ± ±
PAS2b BMBM + DAP ++ ++ ++ ++ − −
PAS2c BMBM + ODA ++ ++ ++ ++ ++ ++
PAS3a BMPFM + PPD ++ ++ + ++ ± ±
PAS3b BMPFM + DAP ++ ++ ++ ++ ++ ++
PAS3c BMPFM + ODA ++ ++ ++ ++ ++ ++
PAS4a BMPCM + PPD ± ± ± ± − −
PAS4b BMPCM + DAP ++ ++ ± ++ − −
PAS4c BMPCM + ODA ++ ++ ++ ++ ++ ++
PAS5a BMBOM + PPD ++ ++ ++ ++ ++ ++
PAS5b BMBOM + DAP ++ ++ ++ ++ ++ ++
PAS5c BMBOM + ODA ++ ++ ++ ++ ++ ++
PAS6a BMAIM + PPD ++ ++ ++ ++ n −
PAS6b BMAIM + DAP ++ + ++ + n ±
PAS6c BMAIM + ODA ++ ++ ++ ++ n ±
PAS7a BMARM + PPD + + + + n −
PAS7b BMARM + DAP + + − + n −
PAS7c BMARM + ODA ++ ++ ++ ++ n ++
++ soluble at room temperature, + soluble on heating, ± sparingly soluble on heating, − insoluble,

n not studied.
diamine moiety exhibit limited solubility. They are soluble in polar aprotic solvents
such as DMF, DMAc and NMP. Though introduction of bulky group on the poly-
mer chain should impart solubility, they exhibit limited solubility due to symmetric
and rigid structures [50, 52]. Polyaspartimides containing oxy linkages exhibit ex-
cellent solubility even in common solvents such as CHCl3 and THF. This could
be attributed to the incorporation of bulky pendant groups and presence of flexi-
ble linkage in the polymer due to which there would be a considerable decrease
in the rigidity and a lowering of energy of internal rotation for the polymer chain,
thus reducing the crystallinity and improving the solubility [52]. Polyaspartimides
containing pyridine moiety exhibit higher solubility in comparison to p-phenylene
moiety, though the presence of pyridine group should enhance polymer rigidity. The
incorporation of bulky pendant group in the polymer chain decreases the intermole-
cular forces between the polymer chains, because of looser packing of polymer
chains. The effect of halo substitution on the solubility was also studied. Polyas-
partimides containing chloro and bromo substituents exhibit limited solubility in
comparison to fluoro substituent. The incorporation of fluorine atom enhances sol-
ubility, due to the electronic polarizability of C–F bonds and the polar interaction
of fluorine atom with solvents is much higher in comparison to Cl and Br. Polyas-
partimides containing benzyloxy unit exhibit higher solubility due to the presence
of bulky phenyl group which enhances free volume, thereby increasing solubility.
The crystallinity of the polymers was evaluated by wide-angle X-ray diffraction
measurements with 2θ measurements from 10 to 40◦ . Polyaspartimides contain-
ing p-phenylene moiety show crystalline pattern. This may be due to the presence
of highly symmetric p-phenylene moiety in the polymer backbone. All other poly-
mers exhibit amorphous pattern. This may be attributed to the presence of bulky
pendant groups, which enhance a looser chain packing, and the presence of flexible
ether linkages, which lower the energy of internal rotation of the polymer chain,
thus reducing the crystallinity and increasing the solubility [53–55]. This may be
attributed to the presence of bulky pendant groups, which enhances a looser chain
packing [56, 57].
4. Conclusions
A number of polyimides were prepared from diamines containing halo/benzyloxy
substituents with various aromatic tetracarboxylic acid dianhydrides. The poly-
imides obtained are soluble in polar aprotic solvents such as NMP, DMF, DMAc
etc. and in low boiling solvents such as chloroform and THF. The fluoro-substituted
polyimides have lower moisture absorption and dielectric constant than the other
halo (Cl, Br) and benzyloxy substituted polyimides. Polyimides containing ben-
zyloxy units were found to have maximum clarity and those containing PMDA
were strongly colored. All the polyimide films exhibit good tensile strength
(84–118 MPa), % elongation (8–12%), modulus (1.5–2.3 GPa) and high tough-
ness, enhancing their utility in semiconductor applications. The Tg values of the
polyimides are in the range of 210–325◦ C. The temperature at which 10% weight
loss of the polymer occurs (T10 ) is very high, indicating high thermal stability of
polymers. The char yield of the polyimides, which correlates to the ability to sus-
tain combustion, is high. A set of bismaleimides containing chloro, bromo, fluoro,
and benzyloxy moieties were also synthesized using the same diamines and were
characterized using FT-IR, 1 H-NMR, 13 C-NMR and elemental analysis. Polyaspar-
timides were successfully prepared by the Michael addition reaction of the syn-
thesized bismaleimides with various diamines and were characterized using FT-IR
and elemental analysis. The polymers were amorphous and their solubility in dif-
ferent organic solvents was significantly enhanced by the incorporation of pendant
groups and they have a wide gap between glass transition temperature and decom-
position temperature. The inherent viscosity values were found to be in the range
of 0.41–0.56 dl/g, which indicates that these materials can be considered as easily
processable polymeric materials. Thus, this series of polyaspartimides may find use
as membranes for gas separation, as well as in the microelectronics and composite
industries.
References
1. D.-J. Liaw, B.-Y. Liaw and J.-J. Chen, Polym. J. 42, 867 (2001).
2. M. K. Ghosh and K. L. Mittal (Eds), Polyimides: Fundamentals and Applications. Marcel Dekker,
3. D. Wilson, H. D. Stenzenberger and P. M. Hergenrother (Eds), Polyimides. Chapman & Hall, New
York, NY (1990).
4. C. Feger, M. M. Khojasteh and J. E. McGrath (Eds), Polyimides: Materials, Chemistry and Char-
acterization. Elsevier, Amsterdam (1989).
5. K. L. Mittal (Ed.), Polyimides: Synthesis, Characterization and Applications, Vols 1–2. Plenum,
6. A. Yamaguchi, S. Tamai and M. Otho, Polymer 37, 3683 (1996).
7. H. Yagci, C. Ostrowski and L. J. Mathias, J. Polym. Sci., Part A: Polym. Chem. 37, 1189 (1999).
8. S. H. Hsiao and L. R. Dai, J. Polym. Sci., Part A: Polym. Chem. 37, 665 (1999).
9. D. Ayala, A. E. Lozano, J. de Adajo and J. G. De le Campa, J. Polym. Sci., Part A: Polym. Chem.
37, 805 (1999).
10. G. S. Liou, M. Maruyama, M. Kakimoto and Y. Imai, J. Polym. Sci., Part A: Polym. Chem. 31,
2499 (1993).
11. S. H. Hsiao and C. T. Li, Macromolecules 31, 7213 (1998).
12. T. L. Grubb, V. L. Ulery, T. J. Smith, G. L. Tullos, H. Yagci, L. J. Mathias and M. Langsam,
Polymer 40, 42 (1999).
13. F. Li, S. Fang, J. J. Ge, P. S. Honigfor, J. C. Chen, F. W. Harris and S. I. D. Cheng, Polymer 40,
4517 (1999).
14. F. Li, J. J. Ge, P. S. Honigfort, S. Fan, J. C. Chen, F. W. Harris and S. I. D. Cheng, Polymer 40,
4987 (1999).
15. R. Hariharan and M. Sarojadevi, Polymer International 56, 22 (2007).
16. R. Hariharan and M. Sarojadevi, J. Appl. Polym. Sci. 102, 4127–4135 (2006).
17. R. Hariharan, S. Bhuvana, N. Amutha and M. Sarojadevi, High Performance Polymers 18, 893
(2006).
18. N. Amutha and M. Sarojadevi, J. Macromol. Sci., Part A: Pure Appl. Chem. 44, 1013–1022 (2007).
19. G. C. Eastmond and J. Paprotny, Macromolecules 28, 2140 (1995).
20. G. C. Eastmond, J. Paprotny and R. S. Irwin, Macromolecules 29, 1382 (1996).
21. F. Acosta, M. Inoki, M. Sawano, Y. Kasashima, K. Naruchi and M. Miure, Polymer 39, 6093
(1998).
22. S. H. Hsiao, C. P. Yang and K. Y. Chu, Macromolecules 30, 165 (1997).
23. H. D. Stenzenberger, M. Herzog, W. Romer, R. Schiblich and N. J. Reeves, Br. Polym. J. 15, 1
(1983).
24. A. Nagai, A. Takahashi, M. Wajima and K. Tsukanishi, Polym. J. 20, 125 (1988).
25. J. P. Pan, G. Y. Shiau and K. M. Chen, J. Appl. Polym. Sci. 44, 467 (1992).
26. L. T. Pappalardo, J. Appl. Polym. Sci. 21, 809 (1977).
27. J. T. Gotro and B. K. Appelt, IBM J. Res. Dev. 32, 616 (1988).
28. A. Takahasshi, A. Nagai, A. Mukoh, M. Wajima and K. Tsukanishi, IEEE Trans. Comp. Hybrids
Manuf. Technol. 13, 1115 (1990).
29. Z. D. Xiang and F. R. Jones, Composites Sci. Technol. 47, 209 (1993).
30. R. Hariharan and M. Sarojadevi, J. Appl. Polym. Sci. 108, 1126–1135 (2008).
31. C. S. Wu, Y. L. Liu and Y.-S. Chiu, Polym. J. 43, 1773 (2002).
32. A. Gungor, C. Smith, J. M. Wescott, S. Srinivasan and J. E. McGrath, Polym. Preprints (Am.
Chem. Soc. Div. Polym. Chem.) 32(1), 172 (1991).
33. J. M. Wescott, T. Yoon, L. Kiefer, D. Rodrigues, M. Konas, G. L. Wilkes and J. E. McGrath,

Polym. Preprints (Am. Chem. Soc. Div. Polym. Chem.) 34(1), 308 (1992).
34. R. Hariharan, S. Bhuvana, G. Anuradha and M. Sarojadevi, J. Polym. Res. 11, 239–245 (2004).
35. R. Hariharan, N. Amutha, S. Bhuvana and M. Sarojadevi, J. Macromol. Sci. Pure Appl. Chem. 41,
317–328 (2004).
36. R. Hariharan, S. Bhuvana and M. Sarojadevi, High Performance Polymers 18, 163–184 (2006).
37. C. P. Yang, S. H. Hsiao and H. W. Yang, Macromol. Chem. Phys. 20, 409 (2000).
38. C. P. Yang, S. H. Hsiao and K. H. Chen, Polymer 43, 5095 (2002).
39. C. P. Yang, S. H. Hsiao and J. Hsu, J. Polym. Sci., Part A: Polym. Chem. 40, 524 (2002).
40. Z. G. Qian, Z. Y. Ge, Z. X. Li, M. H. He, J. G. Li, Z. Z. Pang, L. Fang and S. Y. Yang, Polymer
43, 6057 (2002).
41. C. S. Wang and T. S. Leu, Polymer 41, 3581 (2000).
42. C. S. Wang and R. W. Yang, J. Appl. Polym. Sci. 66, 609 (1997).
43. K. Xie, J. Glut, H. W. Zhou, S. Y. Zhang, M. H. He and S. Y. Yang, Polymer 42, 7267 (2001).
44. C. P. Yang and Y. Y. Su, J. Polym. Sci., Part A: Polym. Chem. 42, 222 (2004).
45. C. P. Yang and T. W. Chen, Macromolecules 26, 4865 (1993).
46. M. Ree, T. J. Shin and S. W. Lee, Korean Polym. J. 9, 1 (2001).
47. R. Hariharan, S. Bhuvana, M. Anithamalbi and M. Sarojadevi, J. Appl. Polym. Sci. 93, 1846
(2004).
48. C.-S. Wang and H.-J. Hwang, J. Polym. Sci., Part A: Polym. Chem. 34, 1493 (1996).
49. S.-H. Hsiao and Y.-J. Chen, Eur. Polym. Mater. 38, 815 (2001).
50. C.-S. Wang and H.-J. Hwang, Polymer 37, 499, 58 (1996).
51. D.-J. Liaw, B.-Y. Liaw and J.-J. Chen, Polymer 42, 867 (2001).
52. Y. Oishi, M. Ishida, M. Kakimoto, Y. Imai and T. Kurosaki, J. Polym. Sci. Polym. Chem. 30, 1027
(1992).
53. M. Ree, K. J. R. Chen and G. Czornyj, Polym. Eng. Sci. 32, 924 (1992).
54. K. B. Kahen and D. L. Peterson, J. Appl. Phys. 68, 2087 (1990).
55. A. R. Katritzky, D. W. Zhu and K. S. Schanze, J. Polym. Sci. Polym. Chem. 32, 2187 (1993).
56. T. Komatsu, K. Fujita and K. Matusita, J. Appl. Phys. 68, 2091 (1990).
57. C. S. Wang and H. J. Hwang, Polymer J. 37, 499 (1996).
Synthesis, Characterization and Polymerization of
Imide-Containing Phthalonitrile Polymers
P. Selvakumar, K. Padmini and M. Sarojadevi ∗

Department of Chemistry, Anna University, Chennai-600 025, India
Abstract
A series of novel imide-containing phthalonitrile polymers with flexible aryl ether units have been syn-
thesized and characterized. Bisphenol monomers were synthesized by a multi-step synthesis involving
a condensation reaction between aromatic aldehydes and 2,6-dimethyl phenol. The bisphenols obtained were
reacted with 4-nitrophthalonitrile to form aryl ether linkage containing bisphthalonitriles. These products
were hydrolyzed to tetracarboxylic acids, which were subsequently converted into corresponding dian-
hydrides. The dianhydrides obtained were reacted with synthesized 4-(aminophenoxy) phthalonitrile by
thermal imidization leading to the formation of imide-containing phthalonitrile monomers. The synthesized
monomers were cured with 3.5 wt% of 4,4 -diaminodiphenylsulphone (DDS). Phthalonitrile end-capped
oligoimides with different molecular weights were prepared by the reaction of pyromellitic dianhydride,
diamine and 4-(aminophenoxy) phthalonitrile via thermal imidization. The average molecular weights of
the oligoimides were determined by GPC analysis. The structure and properties of all compounds synthe-
sized were confirmed by using elemental analysis, FT-IR, 1 H-NMR, 13 C-NMR, DSC, TGA and rheometric
studies. The cure temperatures are found to be in the range of 283–302◦ C, the temperatures for 5 and 10%
weight losses from TGA were in the range of 433 and 492◦ C in N2 and 424 and 478◦ C in air and the char
yield at 800◦ C was 40–51%.
Keywords
Phthalonitrile, ether and imide linkage, cure behavior, thermal stability, rheometric studies
1. Introduction
The demand for fiber-reinforced composites for advanced aerospace applications

has led to the search for high temperature polymers, which are easily processed and
exhibit high thermal and oxidative stability. Nowadays, epoxy and polyimides are
used for advanced composites. These resins have superior mechanical properties
and are lighter than metals but lack thermal stability to operate at high temperatures
and tend to oxidize and become brittle over time. Conventional epoxy-based com-
72 P. Selvakumar et al.
posites are limited to a maximum service temperature of 120◦ C. They also have
significantly high water absorption and require low temperature prepreg storage.
While aromatic polyimides have a greater thermal stability than epoxy resins, their
use has not been as extensive as epoxy resins because of their insolubility in or-
ganic solvents, their poor processability on account of the release of water during
imidisation, and evaporation of trapped solvent in the final resin. A major problem
is in processing blister and void-free components using these matrix resins due to
the volatilization of low boiling component that is generated during the step growth
synthesis [1–4].
Phthalonitrile polymers belong to the class of high-performance thermosetting
materials, possessing unusually high heat resistance. These resins have a wide range
of applications as composite matrices, adhesives, solar panels, and electrical con-
ductors [1, 2]. These resins have superior properties such as formation of void-free
composites, very low moisture absorption (<1%) [5, 6], excellent dimensional sta-
bility due to the absence of moisture-induced distortions, and high thermo-oxidative
stability. These polymers contain a substantial proportion of aromatic structures
which leads to high mechanical strength, high modulus and dimensional stabil-
ity. Such polymeric materials also contain flexible segments to permit conventional
processing and to prevent brittleness [7].
These polymers are prepared from phthalonitriles in which the linking group
(aromatic groups, aliphatic and unsaturated groups, aromatic and diamine groups,
sulfone and ketone groups) between the two ortho dinitrile groups separates the
dinitrile groups enough to permit polymerization. Phthalonitrile monomers are bis
(o-dicyanophenyl) compounds, can be cross-linked thermally by the formation of
triazine ring structure to produce void-free thermosets with high cross-link density.
The main approach for the development of high temperature polymers is the in-
corporation of thermally stable structural units such as aromatic or heteroaromatic
rings within the network of the polymeric system. Through the design of a poly-
meric system with these highly stable structural units and with flexible linkages,
the desired thermo-oxidative stability can be achieved, along with good process-
ability [8]. It has also been reported that the neat polymerization (without aromatic
diamine) of phthalonitrile resin is difficult even under extreme thermal conditions,
because these monomers do not contain active hydrogen. The slow rate of poly-
merization can be attributed to the rigidity of the linking sites, which reduces the
mobility of the reaction sites or to the absence of initiating agent. Aromatic di-
amines are considered to be suitable curing agents for phthalonitrile polymerization
(the aromatic diamines bring about polymerization/curing of phthalonitrile resins).
These diamines generate an active intermediate, which reacts with the excess of
phthalonitrile functionalities (phthalonitrile monomers are mixed with up to 3%
of aromatic diamines. In this mixture the diamine forms an active intermediate
which initiates the polymerization of phthalonitrile by reacting with the phthaloni-
trile functionalities available on the monomer). No volatiles are generated from this
reaction and they are usually compatible with the monomer [3–6]. The time and
Synthesis, Characterization and Polymerization of Imide-Containing Phthalonitrile Polymers 73
temperature needed for polymerization can be reduced by curing the phthaloni-

trile monomers in the presence of aromatic diamine curing agents [9]. Most of
the aromatic amines, particularly 4,4 -diaminodiphenylsulphone (DDS), have the
necessary thermal stability at elevated temperatures to enhance polymerization in
the present process (polymerization of phthalonitrile monomers takes place at very
high temperature (250–300◦ C). The aromatic diamines used for this purpose are
thermally stable to withstand the temperature of polymerization). Aliphatic amines,
however, are less desirable due to their thermo-oxidative instability. Polymerization
can be carried out by heating in air in the presence of amine catalyst for several
hours at high temperatures by addition polymerization [10–13]. The possible cross-
linking structures (the curing reaction of phthalocyanine monomers in the presence
of aromatic diamines leads to the formation of a network structure containg tri-
azine and isoindoline) include phthalocyanine, triazine and isoindoline depending
on the reaction conditions and co-reactants used. The network structure is primarily
from the formation of isoindoline. The trimerisation of nitrile groups leading to the
formation of triazine rings takes place only at high temperatures.
Phthalonitrile resins and networks are reasonably tough materials with excep-
tional thermo-oxidative stability. However, extensive curing times and temperatures
are necessary to achieve these properties. Furthermore, the processability of these
resins is limited as the crystallinity of the phthalonitrile monomers is high. Research
efforts are needed to improve the processability of phthalonitrile resins, by signifi-
cantly decreasing the temperature and time necessary to achieve networks with high
properties [14–18] (superior performance properties like tensile strength, modulus,
etc.).
It is, therefore, proposed to synthesize phthalonitrile endcapped oligoimides hav-
ing different molecular weights using aromatic dianhydrides, diamines and end
capping reagent 4-(aminophenoxy) phthalonitrile and subsequently curing into ther-
mally stable and void-free polymer systems.
2. Experimental
2.1. Materials
4-Nitrophthalonitrile was synthesized according to the procedure given in a pre-

vious study [19]. N -Methyl-2-pyrrolidinone (NMP), N ,N -dimethylformamide
(DMF) and toluene were purchased from SRL, India. These solvents were purified
by distillation under reduced pressure over calcium hydride and stored over 4 Å
molecular sieves. Phthalimide, acetic acid, p-aminophenol, 4,4 -diaminodiphenyl-
sulphone, 2,6-dimethyl phenol, benzaldehyde and anisaldehyde were purchased
from Lancaster, India. 2,4-Dimethoxy benzaldehyde, hydrochloric acid and anhy-
drous potassium carbonate were used as received from E-Merck, India. Thionyl-
chloride and acetic anhydride (Fisher, India) were used without further purification.
2.2. Measurements
Fourier Transform Infrared (FT-IR) spectra were recorded on a Perkin Elmer RX-1
spectrometer using KBr disk in the range from 4000 to 400 cm−1 . 1 H-NMR and
13 C-NMR spectra were obtained on a JEOL Ex-400 spectrometer (400 MHz) us-
ing DMSO-d6 as solvent and the chemical shifts were calibrated using 1% TMS
(Tetramethylsilane) as reference. Elemental analysis was carried out with a Perkin
Elmer model 2400 equipment. Viscosity measurements were performed on a TA
Instruments AR-2000 Rheometer, with an environmental test chamber for temper-
ature control. 40 mm diameter parallel plates were used in the test chamber of the
rheometer. The melt viscosity of the phthalonitrile prepolymers was monitored in
air at 290◦ C as a function of time. The viscosity measurement was used to de-
termine the optimum level of curative concentration and processing temperature.
DSC analysis was performed on a Q10 series TA Instruments differential scan-
ning calorimeter using 3 mg samples taken in aluminium pans at a heating rate
of 10◦ C/min and a flow rate of 40 ml/min in nitrogen. Thermogravimetric data
were obtained on a Q100 series TA Instruments equipment in flowing nitrogen/air
at a heating rate of 20◦ C/min.
2.3. Synthesis of Monomers
2.3.1. 4-Nitrophthalonitrile
This compound was synthesized and purified according to the reported proce-
dure [19]; m.p. 140–142◦ C (Lit. m.p. 141◦ C); IR (KBr) absorptions at 2235 cm−1
(C≡N), 1530 and 1295 cm−1 (asymmetric and symmetric stretching vibrations of
N=O). 1 H NMR (400 MHz, DMSO-d6 ) δ (ppm) = 9.02 (s, 1H), 8.67 (d, 1H), 8.42
(d, 1H). Elemental analysis for C8 H3 N3 O2 : calculated: C, 55.59%; H, 1.73%; N,
24.2%; Found: C, 56.08%; H, 1.92%; N, 23.01%
2.3.2. Synthesis of Bis(4-hydroxy-3,5-dimethyl) Phenyl Methane (HDPM) (Ia)
A 250 ml round-bottom flask was charged with 2,6-dimethylphenol (12.2 g;
0.1 mol), benzaldehyde (5.3 g; 0.05 mol) and hydrochloric acid (15 ml) in ex-
cess (Scheme 1). The reaction mixture was refluxed for about 6 h. The resultant
solution was cooled and neutralized using sodium hydroxide solution. The neutral-
ized solution was then precipitated using hexane. The product obtained was filtered
and dried in a vacuum oven: yield 14.2 g (86%); IR (KBr) 3585–3595 cm−1 (O–
H), 2920 cm−1 (CH3 ); 1 H NMR (400 MHz, DMSO-d6 ) δ (ppm) = 2.23 (s, 12H,
Hb ), 4.52 (s, 2H, Ha ), 5.35 (s, 1H, Hc ), 6.76 (s, 4H, Hd ), 7.12 (d, 2H, He ), 7.33
(dd, 3H, Hf ). 13 C NMR (400 MHz, DMSO-d6 ) δ (ppm) = C1 –152.2, C2 –15.2, C3 –
125.5, C4 –128.5, C5 –135.8, C6 –56.9, C7 –143.4, C8 –127.8, C9 –129.3, C10 –126.3;
elemental analysis: C, 83%; H, 7.2%; found: C, 82.6%; H, 7.12%.
2.3.3. Synthesis of 1,1-bis(4-(4-dicyanophenoxy)-3,5-dimethylphenyl Benzyl
Methane (DDBM) (IIa)
To a 250 ml three-necked flask were added HDPM (Ia) (8.3 g; 0.025 mol), 4-
nitrophthalonitrile (8.65 g; 0.05 mol), anhydrous potassium carbonate (10.35 g;
Scheme 1. Synthesis of bisphenols Ia–Ic.
0.075 mol) and 30 ml of dry DMSO as the solvent. The reaction mixture was stirred
at room temperature for 24 h under nitrogen atmosphere. The resultant solution was
slowly poured into DI water (800 ml) and stirred rapidly to precipitate the monomer.
The monomer was collected by vacuum filtration and washed with methanol to
remove the residual starting materials and DMSO. The purified bisphthalonitrile
monomer was collected by vacuum filtration and dried under ambient conditions:
yield 12.8 g; (83%); FT-IR (KBr) 2242 (C≡N), 1297 and 1076 cm−1 (C–O–C); 1 H
NMR (400 MHz, DMSO-d6 ) δ (ppm) = 2.1 (s, 12H, Ha ), 5.63 (s, 1H, Hb ), 6.89 (s,
4H, Hc ), 7.14 (d, 2H, Hd ), 7.25 (dd, 3H, He ), 7.75 (d, 2H, Hf ), 8.16 (s, 2H, Hg ), 8.73
(d, 2H, Hh ). 13 C NMR (400 MHz, DMSO-d6 ) δ (ppm) = C1 –115.8, C2 –109.0, C3 –
115.5, C4 –121.8, C5 –162.0, C6 –122.5, C7 –132.6, C8 –147.6, C9 –15.0, C10 –128.1,
C11 –126.8, C12 –136.3, C13 –56.9, C14 –143.4, C15 –128.3, C16 –129.3, C17 –126.3.
Elemental analysis: C, 80.13%; H, 4.79%; N, 9.58% Found: C, 80.17%; H, 4.91%;
N, 9.62% (Scheme 2).
2.3.4. Synthesis of 1,1-bis(4-(4-dicarboxyphenoxy)-3,5-dimethyl phenyl) Benzyl
Methane (DDPBM) (IIIa)
A mixture of 2,6-bis (ether dinitrile) (IIa) (7.72 g; 0.015 mol) and potassium hy-
droxide (24.3 g; 0.4327 mol) in 60 ml of ethanol and 60 ml of distilled water
was stirred at refluxing temperature until no further ammonia was generated. The
time taken to reach this stage was about 36 h. The resulting hot, clear solution
was filtered to remove any insoluble impurities. The hot filtrate was allowed to
cool, and the pH value was adjusted by adding dilute hydrochloric acid to near 3.
Scheme 2. DDBM (IIa).
Scheme 3. DDPBM (IIIa).
The precipitate formed was collected by filtration, washed repeatedly with water,
and dried under vacuum: yield 6.9 g (80%); FT-IR (KBr) 3400–3550 (O–H), 1758
(C=O) 1228 cm−1 (C–O–C); 1 H NMR (400 MHz, DMSO-d6 ) δ (ppm) = 10.6
(s, 4H, Ha ), 2.12 (s, 12H, Hb ), 5.62 (s, 1H, Hc ), 6.96, (s, 4H, Hd ), 7.01 (d, 2H,
He ), 7.23 (dd, 3H, Hf ), 7.67 (d, 2H, Hg ), 7.69 (d, 2H, Hh ), 7.8 (s, 2H, Hi ), 13 C
NMR (400 MHz, DMSO-d6 ) δ (ppm) = C1 –169.4, C2 –126.0, C3 –132.6, C4 –117.6,
C5 –162.1, C6 –122.6, C7 –129.9, C8 –147.6, C9 –15.0, C10 –128.1, C11 –126.8, C12 –
136.3, C13 –56.9, C14 –143.4, C15 –128.3, C16 –129.3, C17 –126.3. Elemental analy-
sis: C, 70.90%; H, 4.84%; N, 24.24%. Found: C, 70.23%; H, 4.79%; N, 24.58%
(Scheme 3).
2.3.5. Synthesis of 1,1-bis(4-(4-dicarboxyphenoxy)-3,5-dimethyl phenyl) Benzyl
Methane Dianhydride (DDBD) (IVa)
1,1-bis(4-(4-dicarboxyphenoxy)-3,5-dimethyl phenyl) benzyl methane dianhydride
was prepared as given in Scheme 4. The other two dianhydrides were prepared
by adopting a common procedure. A mixture of bis (3,4-dicarboxyphenoxy) ether
diacid (IIIa) (6.5 g; 0.01 mol), acetic acid (30 ml), and acetic anhydride (30 ml) was
stirred under reflux until a clear solution was obtained. The resulting hot, clear so-
lution was filtered to remove any insoluble impurities. On cooling, the crystallized
brown needles were collected, washed with dry toluene, and dried under vacuum:
yield 5.2 g (84%); IR absorptions at 1845 and 1785 (C=O) and 1194 cm−1 (C–
O–C); 1 H NMR (400 MHz, DMSO-d6 ) δ (ppm) = 2.25 (s, 12H, Hb ), 5.52 (s, 1H,
Hd ), 6.71 (s, 4H, Hc ), 7.08 (d, 2H, He ), 7.15 (dd, 3H, Hf ), 7.35 (d, 2H, Hg ), 7.75 (s,
Scheme 4. Synthesis of dianhydride compounds IVa–IVc.
2H, Ha ), 8.30 (d, 2H, Hh ); 13 C NMR (400 MHz, DMSO-d6 ) δ (ppm) = C1 –163.5,
C2 –129.4, C3 –118.2, C4 –162.1, C5 –122.6, C6 –129.9, C7 –122.8, C8 –147.6, C9 –
Scheme 5. DDBD (IVa).
Scheme 6. APPH.
15.6, C10 –128.1, C11 –126.8, C12 –136.3, C13 –58.7, C14 –143.4, C15 –128.3, C16 –
129.3, C17 –126.3. Elemental analysis: C, 75.0%; H, 4.48%; N, 20.51%. Found: C,
75.27%; H, 4.89%; N, 20.58% (Scheme 5).
2.3.6. Synthesis of 4-(aminophenoxy) Phthalonitrile (APPH)
This compound was synthesized by the nucleophilic nitro displacement reaction.
4-Aminophenol (5.4 g, 0.05 mol), potassium carbonate (6.9 g, 0.05 mol) and 4-
nitrophthalonitrile (8.65 g, 0.05 mol) were taken in a single-neck round-bottom
flask. Approximately 55 ml of dry DMF was added to the solution while stirring.
The solution was allowed to react for 24 h at room temperature under nitrogen
atmosphere. The 4-(aminophenoxy) phthalonitrile was precipitated by adding the
solution to ice cold water forming a dark brown precipitate. The solid product was
collected by vacuum filtration, allowed to air-dry overnight: yield 11.9 g (88%);
FT-IR (KBr) absorptions at 3340 (–NH), 2233 (C≡N), 1250 and 1070 cm−1 (C–
O–C); 1 H NMR (400 MHz, DMSO-d6 ) δ (ppm) = 5.23 (s, 1H), 6.78 (d, 2H), 6.85
(d, 2H), 7.37 (d, 1H), 7.71 (s, 1H), 8.12 (d, 1H); 13 C NMR (400 MHz, DMSO-d6 )
δ (ppm) = C1 –115.8, C2 –109.0, C3 –115.6, C4 –121.8, C5 –162.0, C6 –122.5, C7 –
132.6, C8 –147.0, C9 –118.3, C10 –116.0, C11 –141.5. Elemental analysis: calculated:
C, 71%; H, 3.82%; N, 17.8%; found: C, 69.80%; H, 4.15%; N, 16.93% (Scheme 6).
2.3.7. Synthesis of Imide-Containing Phthalonitrile (ICPH) (Va)
The ICPH (Va) (Scheme 7) was prepared as follows (Scheme 8). To a 100 ml
three-necked flask were added 2,6-bis(3,4-dicarboxyphenoxy) benzaldehyde di-
anhydride (6.3 g; 0.01 mol) and 30 ml of dry dimethylformamide (DMF). After
flushing the solution with nitrogen for 30 min, 4-(aminophenoxy) phthalonitrile
(4.7 g; 0.02 mol) was added to the reaction flask under ambient conditions. The
temperature of the reaction mixture was increased to 90◦ C and held at this temper-
Scheme 7. ICPH (Va).
ature for 1 h. Toluene (30 ml) was added and the solution was heated to reflux. The
water formed as a by-product was removed as azeotrope from the mixture using
a Dean-Stark trap. Total reflux time was 12 h. After removing the toluene by dis-
tillation and cooling, the solidified product was removed from the reaction vessel,
washed with ethanol, collected by filtration, and dried under vacuum: yield 8.9 g
(82%). Monomers Vb and Vc were prepared using appropriate reagents adopting
a similar procedure. FT-IR (KBr): 2242 (C≡N), 1257 cm−1 (C–O–C); 1 H NMR
(400 MHz, DMSO-d6 ) δ (ppm) = 2.20 (s, 12H, Hf ), 5.36 (s, 1H, Hh ), 6.63 (s, 4H,
Hg ), 6.90 (d, 4H, Hb ), 7.02 (d, 2H, Hd ), 7.12 (dd, 3H, Hj ), 7.35 (d, 2H, Hc ), 7.38
(s, 2H, Ha ), 7.45 (d, 2H, Hm ), 7.62 (d, 2H, Hl ), 7.66 (d, 4H, He ), 7.81 (s, 2H,
Hk ), 8.1 (d, 2H, Hd ); 13 C NMR (400 MHz, DMSO-d6 ) δ (ppm) = C1 –115.9, C2 –
114.5, C3 –116.8, C4 –131.6, C5 –141.5, C6 –132.9, C7 –133.4, C8 –132.6, C9 –128.1,
C10 –122.1, C11 –131.4, C12 –167.1, C13 –131.8, C14 –115.0, C15 –160.5, C16 –121.0,
C17 –127.3, C18 –125.2, C19 –147.6, C20 –128.1, C21 –16.2, C22 –126.3, C23 –136.3,
C24 –56.9, C25 –143.4, C26 –128.3, C27 –129.3, C28 –126.3. Elemental analysis: C,
76.0%; H, 3.96%; N, 7.93%. Found: C, 76.43%; H, 4.01%; N, 7.88%.
Monomer Vb (Scheme 9)
FT-IR (KBr): 2240 (C≡N), 1257 cm−1 (C–O–C); 1 H NMR (DMSO-d6 ): 2.32
(s, 12H, Hf ), 3.73 (s, 3H, Hk ), 5.32 (s, 1H, Hh ), 6.61 (s, 4H, Hg ), 6.65 (d, 2H,
Hj ), 6.93 (d, 2H, Hi ), 7.22 (d, 2H, He ), 7.38 (d, 4H, Hb ), 7.69 (d, 4H, Hc ),
7.73 (s, 2H, Hn ), 7.80 (s, 2H, Hl ), 7.92 (s, 2H, Ha ), 7.96 (d, 2H, Hm ), 8.22 (s,
2H, Hd ); 13 C NMR (400 MHz, DMSO-d6 ) δ (ppm) = C1 –115.9, C2 –114.5, C3 –
116.8, C4 –131.6, C5 –141.5, C6 –132.9, C7 –133.4, C8 –132.6, C9 –128.1, C10 –122.1,
C11 –131.4, C12 –167.1, C13 –131.8, C14 –115.0, C15 –160.5, C16 –121.0, C17 –127.3,
C18 –125.2, C19 –147.6, C20 –128.1, C21 –16.2, C22 –126.3, C23 –136.3, C24 –56.9,
C25 –143.4, C26 –128.5, C27 –117.3, C28 –158.3, C29 –58.2.
Monomer Vc (Scheme 10)
FT-IR (KBr pellet): 2237 (C≡N), 1257 cm−1 (C–O–C); 1 H NMR (DMSO-d6 ):
2.32 (s, 12H, Hf ), 3.76 (s, 6H, Hk ), 5.36 (s, 1H, Hh ), 6.61 (s, 4H, Hg ), 6.25 (d,
Scheme 8. Synthesis of imide-containing phthalonitrile monomers.
1H, Hj ), 6.83 (d, 1H, Hi ), 7.22 (d, 2H, He ), 7.38 (d, 4H, Hb ), 7.69 (d, 4H, Hc ),
7.73 (s, 2H, Ho ), 6.30 (s, 1H, Hl ), 7.92 (s, 2H, Ha ), 7.96 (d, 2H, Hn ), 7.94 (s, 2H,
Hm ), 8.22 (s, 2H, Hd ); 13 C NMR (400 MHz, DMSO-d6 ) δ (ppm) = C1 –115.9, C2 –
114.5, C3 –116.8, C4 –131.6, C5 –141.5, C6 –132.9, C7 –133.4, C8 –132.6, C9 –128.1,
C10 –122.1, C11 –131.4, C12 –167.1, C13 –131.8, C14 –115.0, C15 –160.5, C16 –121.0,
C17 –127.3, C18 –125.2, C19 –147.6, C20 –128.1, C21 –16.2, C22 –126.3, C23 –136.3,
Scheme 9. Monomer Vb.
Scheme 10. Monomer Vc.
C24 –47.9, C25 –123.4, C26 –56.7, C27 –159.3, C28 –102.8, C29 –161.2, C30 –108.7,
C31 –132.5.
2.3.8. Synthesis of Phthalonitrile Endcapped Oligoimides

Phthalonitrile endcapped oligoimides were synthesized as per the procedure given
in Scheme 11. To a 500 ml round-bottom flask were added pyromellitic dianhydride
(6.54 g, 0.3 mol), bisphenol diamine 1 (5.04 g, 0.015 mol) and 50 ml of dry dimethyl
formamide (DMF). After flushing the solution with nitrogen for 30 min, the tem-
perature of the reaction mixture was increased to 90◦ C and held at this temperature
for one hour. Toluene (35 ml) was added and the mixture was heated to reflux for
about 12 h. During the reaction, water formed as a by-product was removed by
azeotropic method using a Dean-Stark trap. The reaction mixture was cooled to
room temperature and 4-(aminophenoxy) phthalonitrile (3.52 g, 0.015 mol) was
added. The temperature was increased to 90◦ C and held at this temperature for 1 h.
The mixture was heated to reflux for about 12 h and again the by-product water,
was removed by azeotrope method. After removing the toluene and cooling, the
Scheme 11. Synthesis of phthalonitrile endcapped oligoimides.
mixture was poured into ethanol. The product was collected by filtration and dried:
yield 11.8 g; FT-IR (KBr): 2242 (C≡N), 1778, 1725 (C=O), 1243 cm−1 (C–O–
C).
2.3.9. Prepolymer and Polymer Synthesis
A typical prepolymer synthesis was accomplished by weighing 50 g of the ph-
thalonitrile monomer Va in a 100 ml reaction kettle, placing on a preheated hot
plate at 255◦ C and heating until the prepolymer melted and homogeneity occurred.
The curing agent (DDS) was added, stirred for 15 min in air when it dissolved in the
molten prepolymer. This molten mixture obtained was then quenched to room tem-
perature. This quenched product is the B-staged resin or a prepolymer, which has
the Tg in the range of 100–135◦ C and this amorphous solid is soluble in common or-
ganic solvents such as N ,N -dimethylformamide (DMF), N ,N -dimethylacetamide
(DMAc), N -methyl-2-pyrrolidone (NMP) [15, 19].
Figure 1. FT-IR spectra of compounds Ia–Va.

3.1. Synthesis of Monomers
As shown in Scheme 1 new bisphenols were prepared by the condensation reac-
tion between benzaldehyde, anisaldehyde, 2,4-dimethoxy benzaldehyde and 2,6-
dimethylphenol in the presence of HCl. The FT-IR spectrum of a representative
bisphenol is given in Fig. 1 (spectrum Ia). The broad absorption around 3499 cm−1
corresponds to O–H group and the absence of C=O stretching vibration due to the
aldehyde around 1730–1780 cm−1 confirms the formation of new bisphenol.
The synthesized bisphenol was reacted with 4-nitrophthalonitrile in the presence
of K2 CO3 to form bisphthalonitrile containing ether linkage. The FT-IR spectrum
of bisphthalonitrile is shown in Fig. 1 (spectrum IIa). The spectrum contains absorp-
tion bands around 3050–3081 cm−1 due to the aromatic C–H stretching, in addition
Figure 2. 1 H-NMR spectrum of bisphthalonitrile IIa.
to the sharp band at 2237–2240 cm−1 due to the nitrile groups (C≡N) and the in-
tense band at 1282–1288 cm−1 is due to the ether group. The 1 H-NMR spectrum of
bisphthalonitrile in DMSO-d6 is shown in Fig. 2. The singlet appearing at 2.13 ppm
is due to the methyl protons and that at 5.38 ppm corresponds to the methine proton.
The aromatic proton signals appear in the range between 6.85–8.71 ppm. The rep-
resentative 13 C-NMR spectrum of IIa monomer is shown in Fig. 3. Methyl carbons
appear at around 15.0–16.2 ppm and methine carbon appears in the range of 47.9–
58.7 ppm. The characteristic carbon of the nitrile functionality is found between
114.8–118.2 ppm. Aromatic carbons appear in the range of 109.0–162.0 ppm. The
synthesized bisphthalonitrile was converted into tetracarboxylic acid using KOH in
ethanol as solvent and it was converted into corresponding dianhydride using acetic
anhydride. The FT-IR spectrum of the dianhydride exhibits two characteristic cyclic
anhydride absorptions near 1845 and 1785 cm−1 , attributed to the asymmetric and
symmetric stretching vibrations of C=O, respectively. The 1 H-NMR spectrum of
the dianhydride (DDBD) shows a singlet at 2.25 and 5.52 ppm due to the methyl
protons and methine proton, respectively. The aromatic proton signals appear in
the range between 6.71–8.30 ppm. The end capping reagent 4-(aminophenoxy)
phthalonitrile was synthesized by the nucleophilic nitro-displacement reaction of
4-nitrophthalonitrile and p-aminophenol in NMP in the presence of potassium car-
bonate at room temperature. The FT-IR spectrum shows a band at 2235 cm−1 due
Figure 3. 13 C-NMR spectrum of monomer IIa.
to the cyano group (C≡N) and the band due to hydroxyl (O–H) stretching is absent.
But a sharp band appears in the region of 1253 and 1053 cm−1 due to the forma-
tion of C–O–C linkage. Two characteristic bands at 3375 and 3457 cm−1 are due to
the symmetrical and asymmetrical stretching vibrations, respectively, of the amino
group. The 1 H-NMR spectrum of 4-(aminophenoxy) phthalonitrile shows a broad
singlet at 5.2 ppm due to the two amino protons. The five aromatic protons show
four doublets and one singlet in the aromatic region, 6–8 ppm.
3.2. Synthesis of Imide-Containing Phthalonitrile Monomers, Va–Vc
The new imide-containing phthalonitrile monomers (ICPHs) were prepared by con-
densation of the dianhydride monomer (DDBD) with phthalonitrile end-capping
reagent 4-(aminophenoxy) phthalonitrile through a two-step method by chemi-
cal imidization. All the imide containing phthalonitrile monomers were obtained
in high yields, and elemental analysis data showed that the proposed structures
were correct since the calculated and observed values agreed well with each other.
Figure 1 (spectrum Va) shows the FT-IR spectrum of ICPH monomer Va. The for-
mation of imide linkage was confirmed by the characteristic imide absorption bands
at 1776 (asymmetrical C=O stretch), 1726 cm−1 (symmetrical C=O stretch), 1357
(C–N stretch), 1141, and 786 cm−1 (imide ring deformation). Figure 4 shows the
1 H-NMR spectrum of ICPH (Va). The singlet appearing at 2.20 ppm is due to the
Figure 4. 1 H-NMR spectrum of dianhydride Va.
methyl protons and that at 5.36 ppm corresponds to the methine proton. The aro-
matic proton signals appear in the range between 6.63–8.10 ppm. The results show
that the ICPH monomer had formed and its structure was confirmed. The structures
of all intermediate compounds and the monomers were also confirmed by elemental
analysis, which are in good agreement with calculated values.
3.3. Synthesis of Phthalonitrile Endcapped Oligoimides
Phthalonitrile endcapped oligoimides with different molecular weights were pre-
pared by the reaction of pyromellitic dianhydride, diamine and 4-(aminophenoxy)
phthalonitrile via thermal imidization. The molecular weight of the oligoimides
can be controlled by varying the ratio of diamine and dianhydride. The FT-IR spec-
trum of oligoimide 1 shows absorption band at 2233 cm−1 due to nitrile groups
(C≡N). The presence of strong bands at 1778, 1725, 1370 and 716 cm−1 confirms
the formation of imide ring. A strong absorption band around 1252 cm−1 may be
attributed to the C–O–C bond stretching vibration. The average molecular weight of
the oligoimides was determined by GPC using DMF as eluent. Molecular weights
of the prepared oligoimides are shown Table 2.
3.4. Prepolymer and Polymer Synthesis

The synthesized monomers Va–Vc were washed with sodium hydroxide solution
to remove residual impurities, dried, washed with isopropanol and dried in a vac-
uum oven at 100◦ C overnight prior to use. The melting point range was 242–246◦ C
for Va, 251–254◦ C for Vb and 256–261◦ C for Vc monomers. Polymerization of ph-
thalonitrile monomers into crosslinked networks involves two steps. In the first step,
phthalonitrile monomer (Vc) was mixed with 4,4 -diaminodiphenylsulphone (DDS)
in the molten state to prepare the prepolymer mixture (B-stage resin). The FT-IR
spectrum of the prepolymer is found to be very similar to that of the monomer ex-
cept that very weak bands at 1510 and 1342 cm−1 due to the triazine rings formed.
The second step involves the polymerization of B-stage resin by heating in an air
oven at various temperatures and periods of time. During this curing reaction, more
and more of nitrile groups trimerize to form triazine rings. The FT-IR spectrum
shows decreasing intensity of the nitrile absorption and increasing intensity of ab-
sorptions due to triazine ring at 1530 and 1365 cm−1 , confirming the formation of
crosslinked networks through the formation of triazine rings.
3.5. Cure Cycle Determination
3.5.1. Rheometric Studies
Rheometric studies on phthalonitrile prepolymers were performed with three dif-
ferent concentrations of curing agents (2.5, 3.0 and 3.5 wt%) to monitor the melt
viscosity during polymerization. An isothermal analysis was conducted to deter-
mine the amine concentration needed for curing reaction. Viscosity increases as
a function of time due to crosslinking, which depends upon the concentration and
reactivity of the curing agent. During the cross-linking process, the cyano groups
react in the presence of diamine to form triazine ring structures [14, 20, 21]. A plot
of viscosity change vs time at 290◦ C for phthalonitrile prepolymers containing 2.5–
3.5% by weight of DDS is shown in Fig. 5. It was observed that the viscosity
increases more gradually over one hour for the prepolymers containing the ini-
tial concentration of 2.50 and 3.0 wt% of amine. At a higher amine concentration
(3.5 wt%) viscosity increases at a faster rate. Based on these observations, 3.5 wt%
of amine content was considered as optimum level of curing agent concentration
for polymerization.
3.5.2. DSC Analysis
The polymerization of monomers Va–Vc was studied by DSC analysis. The DSC
thermograms of the monomers Va–Vc with 3.5 wt% diamine are shown in Fig. 6.
The DSC scans exhibit a small endotherm around 155◦ C and a large endotherm be-
tween 255–263◦ C, corresponding to the diamine and monomer melting transitions,
respectively. A broad exotherm immediately after the melting peak of the monomer
is observed for all three monomers Va–Vc. This exotherm corresponds to the reac-
tion of the nitrile group of the monomer in the presence of the aromatic diamine.
The cure reaction was initiated at 285◦ C and the exotherm completed at 302◦ C.
Figure 5. Rheometric studies on phthalonitrile monomer Va at 290◦ C with (a) 2.5 wt% DDS,
(b) 3.0 wt% DDS, (c) 3.5 wt% DDS.
Figure 6. DSC scans of Va, Vb and Vc with 3.5 wt% diamine (DDS).
Table 1.
Thermal properties of phthalonitrile polymers
Sample (DSC)a (TGA)b in N2 (TGA)c in air Char yield (%)d,e

Tcure (◦ C)a
T5 (◦ C) T10 (◦ C) T5 (◦ C) T10 (◦ C) N2 Air
Va 285–305 433 482 424 467 76 48

Vb 278–296 429 476 421 462 74 45
Vc 280–302 428 474 416 457 73 43
Va1 – 440 492 434 478 76 51
(DSC)a : DSC measurements were conducted at a heating rate 10◦ C/min.

(TGA)b and (TGA)c : temperatures at 5, 10% weight loss in N2 and air determined by TGA at
a heating rate of 20◦ C/min.
Char yield (%)d,e : residual weight at 800◦ C in N2 and air.
Va1 : additionally post-cured for 10 h at 370◦ C.
From the rheometric and DSC analysis results, a new cure cycle was developed
for the polymerization of phthalonitrile monomers [22, 23]. Initially the monomers
Va–Vc were heated at 255–260◦ C and degassed under reduced pressure to remove
residual solvent. The temperature was reduced and 3.5 wt% of diamine was added
with stirring. After homogeneity occurred (the mixture of monomers and diamines
became homogeneous by strring) the samples were cured at 250◦ C for 4 h, 280◦ C
for 6 h, 320◦ C for 10 h and post-cured at 340◦ C for 5 h and 370◦ C for 10 h.
3.6. Thermal Properties
The thermal and oxidative properties of the oligomides Va–Vc were evaluated in
the temperature range of 30◦ C to 800◦ C in a thermogravimertric analyzer at a heat-
ing rate of 20◦ C/min under nitrogen/air atmosphere. The results of thermal analysis
for all investigated resins are summarized in Table 1. Figure 7 shows the thermal
stability of Va–Vc in nitrogen atmosphere. All the three polymers (oligomides) are
thermally stable upto 430◦ C and begin to lose weight at higher temperatures with
a char yield of 60–72% at 800◦ C. The temperatures at 5% and 10% weight losses
are 433–440 and 482–492◦ C, respectively. As the temperature was increased from
340 to 370◦ C (curve a1 ) for additional 5 h, an improvement in the thermal stability
was observed. Therefore, the polymeric stability was found to be dependent on cure
cycle and postcure temperatures [14, 24]. Figure 8 shows the thermo-oxidative sta-
bility of Va–Vc in air atmosphere. None of the polymers exhibited obvious weight
loss before the scanning temperature reached 420◦ C in air atmosphere, indicating
that no thermal decomposition occurred. The temperatures at 5 and 10% weight
losses were in the ranges of 424–431 and 467–478◦ C, respectively. In addition, on
further heating to 800◦ C, the residual weight retentions were found to be 30.2–
42.7%, implying that these polymers possessed good thermo-oxidative stability.
When the post-cure temperature of Va was increased from 340 to 370◦ C for ad-
ditional 5 h, the degradation onset temperature increased to 430◦ C and a higher
Figure 7. TGA thermograms of polymers (nitrogen atmosphere) Va, Vb and Vc; additionally
post-cured for 10 h at 370◦ C (a1 ).
Figure 8. TGA thermograms of polymers (air atmosphere) Va, Vb and Vc; additionally post-cured for
10 h at 370◦ C (a1 ).
char yield was obtained indicating an increased thermo-oxidative stability. Once

the degradation process commences, a catastrophic breakdown occurs at a rate
dependent on the post-cure conditions. In essence, as the post-cure temperature
and exposure time were increased, the polymeric stability diminished at elevated
temperatures. From these results it is obvious that the polymers have high thermal
stability that could be attributed to the incorporation of imide units, phenyl groups
in the backbone and triazine rings.
The results of GPC and thermal analysis are summarized in Table 2. The molecu-
lar weights of the oligoimides are in the range of 2480–8952 depending on the ratio
Table 2.
GPC and thermal analysis results of oligoimide I
S. no Mole ratio Molecular DSC cure temp (◦ C) TGA

of a:b:c weight
Initial Peak Final T10 (%) wt loss in ◦ C Char yield (%)
Air N2 Air N2
1 1: 0.4: 0.6 2480 261 272 312 433 445 27.4 63.2
2 1: 0.5: 0.5 3818 273 289 325 448 457 29.8 65.4
3 1: 0.6: 0.4 4560 278 294 329 464 477 32.1 67.2
4 1: 0.7: 0.3 6120 285 299 336 476 488 33.9 69.1
5 1: 0.8: 0.2 7628 291 307 341 480 490 34.3 72.3
6 1: 0.9: 0.1 8952 298 314 348 482 491 34.8 72.9
a = diamine; b = dianhydride; c = 4-aminophenoxy phthalonitrile.
of diamine to dianhydride added [17]. The cure temperatures were in the range of
261–348◦ C. DSC cure temperature increased with increasing the average molecular
weight of the oligoimide. TGA analysis shows that the 10% weight loss tempera-
tures of oligoimides in N2 and air atmosphere are in the ranges of 445–491 and
433–482◦ C, respectively. The residual weight retentions in N2 and air atmosphere
are in the ranges of 63–72 and 27–34% respectively.
4. Conclusions
A series of imide linkage containing phthalonitrile monomers were prepared in high
yield from dianhydrides (IVa–IVc) and 4-(aminophenoxy) phthalonitrile. The ph-
thalonitrile oligoimides were prepared and the molecular weight ranged from 2480
to 8952. The imide linkage was formed before the curing, and hence can be fabri-
cated as a void-free composite material (polyimide-based composites are fabricated
by the two-step poly(amic acid) method. During imidization water molecules are
evolved and the solvent (DMF, DMAc or DMSO) used will be evaporating leav-
ing behind voids in the composites. In the present study the imidization takes place
prior to the fabrication of the composite and curing takes place through addition
reaction between the CN groups leading to the formation of a void-free composite.
DSC scans indicate that the polymerization reaction occurs between 283–302◦ C.
There was no Tg observed up to 450◦ C, confirming that there were no viscoelas-
tic transitions. Based on the rheometric curves, monomers were found to be cured
with 3.5 wt% of diamine (DDS). The rate of polymerization can be controlled and
optimized by the amount of curative used. TGA results exhibit excellent thermal
stability up to 450◦ C in N2 atmosphere with very high char yield (60–72 wt%).
Good thermo-oxidative stability was observed up to 440◦ C in air atmosphere with
30.2–42.7 wt% char yield. As the post-cure temperature increased to 370◦ C, en-
hancement of thermal stability was observed.
Acknowledgement
The authors sincerely acknowledge The Defence Research and Development Orga-
nization (DRDO-ER/IP), New Delhi, India for the financial support.
References
1. T. M. Keller and D. J. Mooney, SAMPE Symp. Proc. 34, 941 (1989).
2. T. M. Keller and C. M. Roland, US Patent 5,242,755 (1993).
3. T. M. Keller, Polym. Comm. 28, 337 (1987).
4. T. M. Keller, J. Polym. Sci., Chem. Ed. 26, 3199 (1988).
5. S. B. Sastri, J. P. Armistead and T. M. Keller, Polym. Composites 18, 48 (1997).
7. M. Laskoski, D. D. Dominguez and T. M. Keller, Polymer 48, 6234 (2007).
8. M. J. Sumner, R. Y. Weyers, A. C. Rosario, J. S. Riffle and U. Sorathia, Polymer 45, 5199 (2004).
9. T. M. Keller, SAMPE Symp. Proc. 31, 528 (1986).
11. T. M. Keller and T. R. Price, Polym. Comm. 25, 42 (1984).
12. T. M. Keller and T. R. Price, J. Polym. 26, 48 (1985).
14. T. M. Keller, Polymer 34, 952 (1993).
16. S. B. Sastri and T. M. Keller, J. Polym. Sci. Part A: Polym. Chem. 36, 1885 (1998).
17. T. M. Keller, SAMPE Symp. Proc. 31, 528 (1986).
18. T. M. Keller, US Patent 5,262,514 (1993).
19. G. Joseph and W. Onyebuagu, J. Org. Chem. 55, 2155 (1990).
20. D. D. Dominguez, H. N. Jones and T. M. Keller, Polym. Composites 25, 5 (2004).
21. D. D. Dominguez and T. M. Keller, Polymer 48, 97 (2007).
22. P. J. Burchill, J. Polym. Sci. Part A: Polym. Chem. 32, 1 (1994).
23. T. M. Keller, Chem. Mater. 6, 302 (1994).
24. M. Laskoski, D. D. Dominguez and T. M. Keller, J. Polym. Sci. Part A: Polym. Chem. 43, 4136
(2005).
Structure–Property Relationships in Processable Heat
Resistant Poly(amide–imide–imide)s for
High-Temperature Applications
M. Sarojadevi ∗ and S. Bhuvana

Abstract
Polyimides and poly(amide–imide)s are of high interest for many engineering applications due to their
excellent thermal and mechanical properties. The aerospace, automotive, and microelectronics industries
have developed many important applications. However, the processing of these thermoplastic polymers has
been greatly hindered because they lack softening or melting behaviour at usual processing temperatures,
and they tend to degrade before or at the softening temperature. Various attempts have been made to bring
down the Tg or melting temperature of aromatic polyamides to make them processable, either by introducing
kinked or flexible bridging units. Unfortunately, the loss of thermal stability and significant decrease in
mechanical properties on heating are usually a consequence of the reduced chain stiffness. In the present
study, a series of novel aromatic diamines bis(4-amino-3,5-dimethyl phenyl) (X) phenyl methane (X→3-
chloro, 3-bromo, 4-methoxy) from corresponding aldehydes and 2,6-dimethyl aniline; and bis(4-amino-
3,5-dimethyl phenyl) cyclohexane from cyclohexanone and 2,6-dimethyl aniline; bis(4-amino-3,5-dimethyl
phenyl) cycloheptanone from cycloheptanone and 2,6 dimethyl aniline; and bis(p-amino phenoxy)-methyl
phenyl silane from p-aminophenol and dichloro methyl phenyl silane were prepared. Tetrimide dicarboxylic
acids were prepared by reacting the prepared diamines with PMDA/BPDA and p-aminobenzoic acid. The
polymers were prepared by treating the tetrimide diacid with different aromatic diamines. The structures
of the monomers and polymers were identified by elemental analysis, FT-IR, 1 H-NMR and 13 C-NMR.
The polymers show excellent solubility. The polymers are amorphous and have high optical transparency.
They also show good thermal stability and their Tg is found to be in the range 268–372◦ C. They had 10%
weight losses at temperatures beyond 530◦ C and left more than 50% char yield even at 800◦ C in nitrogen
atmosphere. The thin films cast from DMAc exhibit tensile strengths of 96–101 MPa, elongations at break
of 9–14% and initial moduli of 1.9–2.3 GPa. The properties of the different polymers were compared and
discussed and correlated to their structures.
Keywords
Poly(amide–imide–imide)s, thermal and mechanical properties
94 M. Sarojadevi and S. Bhuvana
1. Introduction
The durability and reliability of materials used in aerospace components are of crit-
ical concern. Among the material requirements for these applications are a high
glass-transition temperature, Tg (at least 25◦ C higher than the intended use tem-
perature), good high-temperature stability in a variety of environments, and good
mechanical properties over a wide range of temperatures. In general, the stability
and Tg of most organic polymers limit their use, at best, to applications in which
temperatures are no higher than 370◦ C. From this standpoint, polyimides as a class
of materials are very promising due to an array of desirable characteristics pos-
sessed by these materials including excellent mechanical properties, excellent ther-
mal stability, wear resistance, radiation resistance, inertness to solvents, hydrolytic
stability, low dielectric constant and good adhesion strength. Indeed significant ad-
vances have been made in the field of polyimides in the past five decades and today
these versatile materials find use in a broad range of applications [1–6].
However, infusibility and limited solubility in organic solvents and their ten-
dency to degrade before or at the softening temperature [6, 7] are the characteris-
tic properties of wholly aromatic polyimides that makes them generally difficult,
or too expensive, to process, thus restricting their applications [8, 10]. Addition-
curing polyimides were investigated in the early 1970s in an attempt to improve the
processability of condensation polyimides without adversely affecting their stabil-
ity and high-temperature performance. Perhaps the most noteworthy development
in this area has been the Polymerization of Monomeric Reactant (PMR) family of
polyimides, in particular PMR-15, developed at the NASA Lewis Research Center
[11]. Because of its good high-temperature performance, PMR-15 is widely used in
composites and adhesives for a variety of aircraft engine applications. The demand
for PMR-15 has basically stagnated at the 1989 level, primarily for two reasons:
concern over the mutagenicity of methylene dianiline [12] and the high cost of fab-
ricating PMR-15 engine components due to processing limitations.
While norbornyl end-caps have been used successfully in a number of polyimide
systems, there are some drawbacks to their use. The thermal-oxidative stability of
the norbornene ring is poor due to the large amount of saturated carbons (nor-
bornene is alicyclic containing only saturated carbons and not aromatic) present
in this structure. In addition, processing problems can be encountered due to the
potential for formation of cyclopentadiene during cross-linking [13].
Another cross-linking approach that has been employed is the incorporation of
strained ring systems into the polyimide either as end-caps or in the polymer back-
bone. These moieties undergo ring opening at high temperatures to yield a reactive
intermediate, e.g., an o-quinodimethane or a biradical, which then polymerizes.
Stille and co-workers demonstrated that biphenylenes [14] could be used as re-
active end-caps for polyimide oligomers [15]. When heated above 300◦ C [16],
it undergoes a ring-opening reaction to produce a biradical that can couple, thus
leading to cross-linking (the radical sites in the various biradicals react and form
a bond (couple) leading to cross-linking). The activation temperatures for ring open-
Structure–Property Relationships in Processable Heat Resistant Poly(amide–imide–imide) 95
ing can be lowered through the use of Ni (O) catalysts. However, the cured films
of biphenylene-terminated polyimides have modest Tg s (261◦ C). High degrees of
polymerization could not be achieved with these polymers, even in the presence of
the Ni (O) (complexes in which the nickel atom is coordinated to the oxygen atom)
catalyst. In addition, graphite-reinforced composites prepared with this polyimide
have high void contents [15].
Other efforts to improve polyimide processability have dealt with modifications
to the polyimide backbone to inhibit crystal packing [17–21]. This is accomplished
by a number of approaches ranging from the introduction of bulky pendant groups,
e.g., phenyl or t-butyl, twisted or non-coplanar structures, bicyclics or multi-ring
systems. Bulky pendant groups have been used to increase free volume and asym-
metry in order to improve polymer solubility and melt processability. Introduction
of a single pendant methyl or t-butyl group onto poly (ether ketone) [22] led to
enhanced solubility of these polymers in chloroform, with little or no effect on Tg
or Td [23]. However, introduction of a second t-butyl group resulted in a threefold
decrease in solubility and 24◦ C increase in Tg , due to the more symmetric struc-
ture of this polymer. Thus, the Tg is raised by the restriction of segmental mobility
brought about by the incorporation of bulky groups, whereas solubility is increased
by decreasing the packing density and crystallinity [24, 25].
Monomers with twisted, kinked or crankshaft-like structures have also been used
to enhance polymer solubility and processability. Cage and bicyclic structures have
been used to increase polyimide solubility. For example, inclusion of diamantine
[26], or adamantine [27] units provided polyimides that were amorphous and solu-
ble in a variety of organic solvents, e.g., chloroform, m-cresol, o-chlorophenol and
N-methyl-2-pyrrolidone. In general, correspondingly substituted diamantane and
adamantane polyimides have similar thermal-oxidative stability (i.e., two different
polyimides, having diamantane in one and adamantine in the other, attached to sim-
ilar substrate have similar thermal-oxidative stability). The Tg s of the diamantane
systems are about 30◦ C higher than those of the adamantane polyimides due to the
more linear structure of diamantane.
Similar to the above modifications another interesting way to prepare process-
able and thermally stable polyimides is co-polymerization, i.e., preparation of
poly(amide–imide)s. (Poly(amide–imide)s have two different types of repeat units
namely amide and imide, and hence can be called as copolymers.) Aromatic
poly(amide–imide)s (PAIs) are another important class of high-performance poly-
mers having excellent resistance to high temperatures, good processing character-
istics and favourable balance of other physical and chemical properties [28]. The
present study deals with the synthesis and characterization of poly(amide–imide–
imide)s containing chlorophenyl, bromophenyl, anisyl or cycloaliphatic pendant
groups prepared by the polycondensation reaction.
2. Experimental
2.1. Materials
p-Amino benzoic acid (p-ABA, SRL, India), pyromellitic dianhydride (PMDA,
Spectrochem, India) and 3,3 ,4,4 -benzophenone tetracarboxylic acid dianhydride
(BPTDA, Spectrochem, India) were recrystallised using acetic anhydride before
use. 2,6-Dimethyl aniline and HCl were purchased from SRL, India. 3-Chloro ben-
zaldehyde, 3-bromo benzaldehyde and anisaldehyde were purchased from E. Merck
(Germany). Cyclohexanone, dichloro diamino diphenyl methane (DCDDM) and
cycloheptanone were purchased from Aldrich, USA and used as received. All
the diamines such as p-phenylene diamine (PPD), 4,4 -diamino diphenyl methane
(DDM) and diamino diphenyl ether (oxydianiline, ODA) were procured from
E. Merck, Germany and were used as received. N-methyl-2-pyrrolidone (NMP,
Spectrochem, India) and pyridine (Spectrochem, India) were purified by distillation
under reduced pressure and stored over 4 Å molecular sieves. Triphenylphosphite
(TPP, E. Merck, Germany) was used as received. Anhydrous calcium chloride
E. Merck (Germany), was dried under reduced pressure at 150◦ C for 6 h prior to
use.
2.2. Synthesis of Diamines
The diamines bis(4-amino-3,5-dimethyl phenyl)4-anisyl methane (BAAM), bis(4-
amino-3,5-dimethyl phenyl)3-chloro phenyl methane (BACM), bis(4-amino-3,5-
dimethyl phenyl)3-bromo phenyl methane(BABM), bis(4-amino-3,5-dimethyl
phenyl) cyclohexane (BAHX) and bis(4-amino-3,5-dimethyl phenyl) cycloheptane
(BAHT) were prepared from anisaldehyde, 3-chloro benzaldehyde, 3-bromo ben-
zaldehyde, cyclohexanone and cycloheptanone respectively, adopting a common
procedure given below (Schemes 1 and 2).
2,6-Dimethyl aniline (4.84 g, 0.04 mol) was charged into a 250 ml three necked
round bottomed flask equipped with a nitrogen inlet, an addition funnel and a re-
flux condenser. Hydrochloric acid (4.09 ml, 0.045 mol) was added dropwise to the
reaction vessel for 30 min. The solid substance obtained was melted by heating to
100◦ C. To this molten salt, the respective aldehyde or ketone (cyclohexanone, 2 ml,
0.02 mol) was added dropwise in about 30 min. After complete addition, the tem-
perature was raised to 120◦ C and maintained at this temperature for about 20 h. The
solution was then cooled and poured into water and then neutralized with aqueous
NaOH. The resulting oily layer was separated out and distilled to remove the ex-
cess 2,6-dimethyl aniline. The residual crude product was then recrystallised using
benzene to obtain bis(4-amino-3,5-dimethyl phenyl) cyclohexane (BAHX). Yield
88%.
2.3. Synthesis of Tetrimide Dicarboxylic Acids
A three-necked 150 ml round bottom flask equipped with nitrogen inlet and re-
flux condenser was charged with a solution of 2.912 g (8 mmol) of diamine (bis
Scheme 1. Synthesis of bis(4-amino-3,5-dimethyl phenyl) X phenyl methane.
Scheme 2. Synthesis of bis(4-amino-3,5-dimethyl phenyl) cycloalkane.
(4-amino 3,5-dimethyl phenyl)3 -chloro phenyl methane) and 2.192 g (16 mmol)
of p-ABA in 20 ml of NMP. To this 3.488 g (16 mmol) of benzophenone tetracar-
boxylic acid dianhydride/pyromellitic dianhydride (BTDA/PMDA) was added in
one portion. The mixture was stirred at room temperature for 2 h. The exchange
reaction of amic acid was carried out during this long stirring and the product with
lowest free energy (Diacid) was prepared. (The amino group in the diamine may
react initially with both the dianhydride and the –COOH group of benzoic acid
leading to the formation of a mixture of poly(amic acid)s. But subsequently during
the long stirring period the poly(amic acid) formed by reaction with the –COOH
of benzoic acid undergoes exchange reaction with the dianhydride leading to the
formation of the product with lowest free energy (when the dianhydride reacts with
the amine it undergoes ring opening, and hence this reaction is thermodynamically
more favourable and the product formed is the one with lower free energy than that
formed with benzoic acid).) About 35 ml of toluene was then added and the mix-
ture was refluxed for about 3 h. The water formed in the reaction was distilled off
azeotropically using toluene through a Dean–Stark trap. At the end of the reaction,
Table 1.
Notations of tetrimide diacids
S. no. Name/code Diamine X Dianhydride
1 Bis 4-(pyromellitimido, 3,5-dimethyl phenyl anisyl) benzoic BAAM PMDA

acid (BPAA)
2 Bis 4-(3,3 ,4,4 -benzophenone tetracarboximido, 3,5- BAAM BTDA
dimethyl phenyl anisyl) benzoic acid (BBAA)
3 Bis 4-(pyromellitimido, 3,5-dimethyl phenyl 3 chloro phenyl BACM PMDA
methane) benzoic acid (BPMCA)
4 Bis 4-(3,3 ,4,4 -benzophenone tetracarboximido, 3,5- BACM BTDA
dimethyl phenyl, 3 chloro phenyl methane) benzoic acid
(BBMCA)
5 Bis 4-(pyromellitimido, 3,5-dimethyl phenyl 3 bromo phenyl BABM PMDA
methane) benzoic acid (BPMBA)
6 Bis 4-(3,3 ,4,4 -benzophenone tetracarboximido, 3,5- BABM BTDA
dimethyl phenyl 3 bromo phenyl methane) benzoic acid
(BBMBA)
7 Bis 4-(pyromellitimido, 3,5-dimethyl phenyl cyclohexyl) BAHX PMDA
benzoic acid (BPCA)
8 Bis 4-(3,3 ,4,4 -benzophenone tetracarboximido, 3,5- BAHX BTDA
dimethyl phenyl cyclohexyl) benzoic acid (BBCA)
9 Bis 4-(pyromellitimido, 3,5-dimethyl phenyl cycloheptyl) BAHT PMDA
benzoic acid (BPCPA)
10 Bis 4-(3,3 ,4,4 -benzophenone tetracarboximido, 3,5- BAHT BTDA
dimethyl phenyl cycloheptyl) benzoic acid (BBCPA)
the residual toluene was distilled off under reduced pressure. After cooling, the ob-
tained solution was trickled into water and the precipitated product was collected
by filtration and dried in vacuum at 100◦ C for 12 h to obtain diacid. The purity of
the diacid was checked from the integral ratios of protons in the 1 H-NMR spectrum.
Ten different tetrimide diacids were prepared using all the five prepared diamines
with two dianhydrides (Table 1).
2.4. Synthesis of Poly(amide–imide–imide)s
A two-necked 100 ml round bottom flask equipped with reflux condenser and ad-
dition funnel was charged with a solution of 1.210 g (1 mmol) of diacid (BPMCA)
and 0.108 g (1 mmol) of p-phenylene diamine in 5 ml of NMP. To this about 0.30 g
of anhydrous calcium chloride, 0.6 ml of TPP and 1 ml of pyridine were added and
refluxed for 3 h. As the reaction proceeded the viscosity of the reaction mixture
increased and an additional 3 ml NMP was added. The obtained viscous solution
was trickled into excess methanol with vigorous stirring. The polymer thus obtained
was filtered, washed with methanol and dried under vacuum at 100◦ C for 12 h. All
Table 2.
Notations of poly(amide–imide–imide)s
Polymer Polymer Polymer Polymer Polymer Polymer

code code code
PAI3a BPAA + PPD PAI6c BBMCA + DDM PAI10a BBCA + PPD

PAI3b BPAA + ODA PAI6d BBMCA + DCDDM PAI10b BBCA + ODA
PAI3c BPAA + DDM PAI7a BPMBA + PPD PAI10c BBCA + DDM
PAI3d BPAA + DCDDM PAI7b BPMBA + ODA PAI10d BBCA + DCDDM
PAI4a BBAA + PPD PAI7c BPMBA + DDM PAI11a BPCPA + PPD
PAI4b BBAA + ODA PAI7d BPMBA + DCDDM PAI11b BPCPA + ODA
PAI4c BBAA + DDM PAI8a BBMBA + PPD PAI11c BPCPA + DDM
PAI4d BBAA + DCDDM PAI8b BBMBA + ODA PAI11d BPCPA + DCDDM
PAI5a BPMCA + PPD PAI8c BBMBA + DDM PAI12a BBCPA + PPD
PAI5b BPMCA + ODA PAI8d BBMBA + DCDDM PAI12b BBCPA + ODA
PAI5c BPMCA + DDM PAI9a BPCA + PPD PAI12c BBCPA + DDM
PAI5d BPMCA + DCDDM PAI9b BPCA + ODA PAI12d BBCPA + DCDDM
PAI6a BBMCA + PPD PAI9c BPCA + DDM – –
PAI6b BBMCA + ODA PAI9d BPCA + DCDDM – –
the other reported polymers were prepared using the ten different tetrimide diacids
adopting a similar procedure (Table 2).
2.5. Measurements
Infrared spectra were recorded on a Nicolet Fourier transform infrared spectrome-
ter. 1 H-NMR and 13 C-NMR spectra were obtained on a JEOL Ex-400 spectrometer.
Elemental analysis was carried out on a Perkin Elmer model 2400 equipment. Ther-
mogravimetric data were obtained on a DuPont 2100 equipment in flowing nitrogen
at a heating rate of 20◦ C/min. Differential scanning calorimetry (DSC) analysis was
performed on a DuPont model 910 Differential Scanning Calorimeter. UV mea-
surements were made in a Shimadzu UV-1601 spectrophotometer. The sample was
dissolved in NMP (5 × 10−3 mol/l) and the cut-off wavelength (λ0 ) was determined
and % transmittance at 500 nm was also determined. Solubility of the polymers was
tested in various solvents, viz., chloroform, THF, DMF, NMP, DMSO and DMAc.
About 2–3 mg of the polymer was treated with 6 ml of solvent and kept aside for
24 h with occasional shaking.

3.1. Synthesis of Bis(4-amino-3,5-dimethyl phenyl) Cycloalkanes and
Bis(4-amino-3,5-dimethyl phenyl) (X) Phenyl Methane (X = 3-chloro, 3-bromo,
4-methoxy)
Various diamines bis(4-amino-3,5-dimethyl phenyl) substituted (X) phenyl methane
(X = 4-methoxy, 3-chloro, 3-bromo) (BAAM, BABM and BAMM) were prepared
Table 3.
Physico-chemical characteristics of diamine monomers
Monomer code/ M.P. (◦ C) Yield (%) Elemental analysis (wt%)

formula
Carbon Hydrogen Nitrogen
Calcd. Found Calcd. Found Calcd. Found
C24 H28 N2 O 182 83 79.79 79.86 7.82 7.80 7.77 7.74

(BAAM)
C23 H25 N2 Cl 144 89 75.70 75.25 6.90 6.70 7.67 7.49
(BACM)
C23 H25 N2 Br 134 91 67.48 67.49 6.15 6.20 6.84 6.88
(BABM)
C22 H30 N2 152 88 81.44 81.68 9.32 9.51 9.24 9.04
(BAHX)
C23 H32 N2 158 85 81.61 81.79 9.53 9.69 8.86 8.64
(BAHT)
Figure 1. FT-IR spectrum of BAHT.
from the corresponding substituted benzaldehyde and 2,6-dimethyl aniline. The

monomers containing cycloalkane in the main chain, namely BAHX and BAHT,
were prepared by the reaction of the ammonium salt of 2,6-dimethyl aniline with
ketones (cyclohexanone, cycloheptanone) at high temperatures. The products were
obtained with comparably high yields (>80%). Schemes 1 and 2 describe the
synthesis of diamine monomers. The prepared diamines were characterized by el-
emental analysis, FT-IR and NMR techniques. The results of these analyses are
presented in Table 3. All the analysis clearly confirm the formation of the desired
products. The elemental analysis values correlate well with the calculated values.
The diamine monomers show absorption bands (Figs 1 and 2) around 3376–3391
and 3442–3487 cm−1 due to N–H symmetric and asymmetric stretching vibrations,
respectively, of NH2 group and the absorption band around 1619–1635 cm−1 is
due to N–H bending vibration. The monomers show absorptions around 2970 and
Figure 2. FT-IR spectrum of BAHX.
2921 cm−1 due to C–H stretching vibration of CH3 group. In addition to these
characteristic vibrations, all the monomers show absorption bands around 1400–
1460 cm−1 and/or 1540–1600 cm−1 corresponding to aromatic ring vibrations. The
band corresponding to C–X bond in monomers 2 and 3 (where X = Cl, Br) appear
in the region around 1083 and 1072 cm−1 . The 13 C-NMR spectra of monomers
show distinct signals for all the different carbon atoms. All the aliphatic carbons
appear below 55 ppm and the aromatic carbons appear between 120 and 150 ppm
and carbons in the monomers resonate between 17.5–18.2 ppm. All the spectro-
scopic data obtained agree well with the proposed structures. The representative
1 H and 13 C-NMR spectra of diamine monomer (BAHX and BACM) are shown in
Figs 3 and 4. The diamine monomers show a signal around 3.48–3.84 ppm (sin-
glet) corresponding to the NH2 protons. The methyl protons of these monomers
resonate around 2.17–2.25 ppm (singlet) and the aromatic protons resonate around
6.87–7.43 ppm.
3.2. Synthesis of Tetrimide Dicarboxylic Acids
Tetrimide dicarboxylic acids were synthesized (Scheme 3) by the reaction between
the prepared diamines, dianhydride and p-ABA in a 1:2:2 molar ratio at room tem-
perature in amide type solvent (such as NMP or DMAc), followed by intramolecular
cyclodehydration of the intermediate tetramic acid [29]. The addition of p-ABA, di-
anhydride, and diamine did not form the structure of tetramic acid completely in the
initial period, and some other diacids were produced. However, the exchange reac-
tion of amic acid was carried out during a long time of stirring [30], and the product
with the lowest free energy was prepared. From the molar ratio of monomers,
the tetrimide diacid is the most stable structure among the intermediates. There-
fore, a higher purity of tetrimide diacid was obtained after the cyclodehydration of
tetramic acid. The yields of the different diacids obtained were found to be from
(a)
(b)
Figure 3. 1 H (a) and 13 C-NMR (b) spectra of BAHX.
85 to 92%. The structures of diacids were confirmed by elemental analysis, FT-IR

(Table 4; Fig. 5) and 1 H-NMR (Fig. 6) spectroscopy.
The diacid monomers show absorption band around 3444–3501 cm−1 due to
O–H stretching of –COOH group. The absorptions around 1720 and 1778 cm−1
correspond to symmetric and asymmetric stretchings, respectively, of imide car-
bonyl group [21, 31]. The bands around 1368 cm−1 correspond to imide ring axial
vibration and those around 725 cm−1 are due to imide ring out-of-plane bending vi-
bration. The band around 1110 cm−1 corresponds to imide ring transverse vibration
(a)
(b)
Figure 4. 1 H (a) and 13 C-NMR (b) spectra of BACM.
[32, 33]. The diacids (BPAA–BBCPA) show absorption band around 2922 cm−1
corresponding to C–H stretch of methyl group. In addition to these bands, the
characteristic band of C–X (X = Cl, Br) in the diacids (BPMCA–BBMBA) ap-
pears in the region 1082–1072 cm−1 . In addition to these characteristic vibrations,
all the monomers show absorption bands around 1400–1460 cm−1 and/or 1540–
1600 cm−1 corresponding to aromatic ring vibrations.
The 1 H-NMR spectra of the diacid monomers (BPAA–BBMBA) show a signal
around 5.40–5.53 ppm (singlet) corresponding to the methyne protons. The methyl
protons of these monomers resonate around 2.13–2.31 ppm (singlet) and the aro-
X → BAAM, BACM, BABM, BAHX, BAHT
Scheme 3. Synthesis of tetrimide dicarboxylic acids.
matic protons resonate around 6.78–8.52 ppm. However, the protons for acid group
of diacid were not observed clearly for some diacids which may be due to the fast
exchange of proton with a trace amount of moisture associated with the solvent.
3.3. Synthesis of Polymers
A series of poly(amide–imide–imide)s were synthesized (Scheme 4) from prepared
diacids and various diamines by direct polycondensation with TPP/Py as the con-
densing agent in NMP in the presence of calcium chloride. Polymerization was
Table 4.
FT-IR spectral data of tetrimide diacids
Monomer Wavenumber (cm−1 )

code
O–H C=O C=O C–N–C C–N–C CH3 C–X
sym asym bending stretching
BPAA 3472 1720 1780 725, 1110 1370 2922 –

BBAA 3478 1727 1778 731, 1114 1363 2923 –
BPMCA 3469 1724 1779 725, 1107 1368 2922 1082
BBMCA 3466 1721 1776 728, 1106 1372 2920 1084
BPMBA 3468 1726 1777 731, 1115 1371 2922 1073
BBMBA 3469 1712 1780 725, 1116 1389 2924 1074
BPCA 3452 1714 1776 730, 1110 1384 2921 –
BBCA 3444 1726 1779 731, 1113 1360 2923 –
BPCPA 3458 1725 1779 726, 1108 1368 2921 –
BBCPA 3455 1721 1776 731, 1108 1365 2922 –
Figure 5. FT-IR spectrum of the tetrimide diacid BPCA.
carried out using the resultant solution of diacid synthesis. All the reactions went
on smoothly in homogeneous solution, except a few. In some cases, when the mole-
cular weight of the polymer formed in the solution reached high value, it started
forming a swollen gel, preventing the chain growth. To prevent this gelation and to
obtain high molecular weight polymer, additional quantity of solvent NMP was
added. Besides, an advantage of poly(amide–imide–imide) synthesis from high
molecular weight diacids is that a large amount of product can be obtained us-
ing a small amount of TPP. In other words, for a specified quantity of TPP used,
higher yields of PAIs are obtained using tetrimide diacids in comparison to use of
Figure 6. 1 H-NMR spectrum of BBMCA.
diamines. Therefore, new types of poly(amide–imide–imide)s would significantly

reduce the cost of production and thus lead to industrial production [35, 36]. All the
polymers were obtained in good yield. All the polymers obtained by this method
were characterized by FT-IR and 1 H-NMR. The representative FT-IR spectra of
poly(amide–imide–imide)s are shown in Fig. 7. The absorptions around 1726 and
1779 cm−1 correspond, respectively, to symmetric and asymmetric stretching vi-
brations of imide carbonyl group [37]. The bands around 1364 and 731 cm−1 are
due to C–N–C stretching and bending vibrations of the imide ring. The band around
1115 cm−1 corresponds to imide ring deformation. The polymers show absorption
band around 2922 cm−1 corresponding to C–H stretching of methyl group. The
presence of bands around 3364 cm−1 and 1650 cm−1 due to NH and C=O of amide
group and the absence of bands around 3449–3501 cm−1 due to O–H stretching
vibration confirms a complete conversion of poly(amic acid) to poly(amide–imide–
imide). The results support the formation of the desired product.
The 1 H-NMR spectrum of a typical polymer is shown in Fig. 8. The aromatic
protons of the polymers resonate at varying δ values between 6.90 and 8.53 ppm
depending on the polymer backbone structure. The aromatic protons in the dian-
Scheme 4. Synthesis of poly(amide–imide–imide)s.
hydride appear in the farthest downfield region due to the electron withdrawing
imide C=O group, present adjacent to it. The amide group protons appear around
10.6 ppm. The CH3 group which is ortho to imide group in the polymers appears
in the region around 2.13–2.18 ppm (singlet). Absence of signal at 10.38 ppm due
to carboxylate protons of tetrimide diacid confirms that polyimide ring closure had
taken place completely.
3.4. Solubility
The solubility of poly(amide–imide–imide)s was tested qualitatively in various or-
ganic solvents and the results are summarized in Table 5. All the PAIs were easily
(a)
(b)
Figure 7. FT-IR spectra of (a) PAI11a , (b) PAI12a .
soluble at room temperature in aprotic polar solvents such as NMP, DMAc, DMSO
as well as less polar solvents such as THF and CHCl3 . The good solubility of these
polymers is possibly governed by the structural modifications through the incorpo-
ration of the flexible and bulky diamine and dianhydride moieties. The introduction
of cyclo aliphatic pendant groups into the polymer backbone leads to an increased
solubility because of a decreased molecular order and intermolecular bonding. The
presence of halo substitution in the pendant phenyl group also enhances the solu-
bility due to the polar interactions with the solvents. The tetramethyl substituents
restrict the packing of macromolecular chains in tight structures through hydrogen
bonding between amide groups and, consequently, the solvent molecules can pene-
Figure 8. 1 H-NMR spectrum of poly(amide–imide–imide).
trate easily to solubilise the chains. They also make the polymer less polar in nature
and effectively enhance the solubility of the polymer.
3.5. Inherent Viscosity
Inherent viscosities of the prepared PAIs were obtained using a Ubbelohde vis-
cometer at 0.5 g/dl concentration in NMP at 30◦ C. The inherent viscosity values
(Table 6) of poly(amide–imide–imide)s vary depending on the type of diamine be-
tween the two amide groups, the dianhydride moieties and on the stiffness of the
polymer chain.
The inherent viscosities of the PAIs are in the range 0.76–1.09 dl/g indicating
the formation of high molecular weight polymers. This is because formation of
poly(amide–imide)s from diacids and aromatic diamine is promoted by TPP in
NMP at 100◦ C and the addition of inorganic salts (CaCl2 ) brings about an in-
crease in the chain length of the polymer probably due to an improved solubility
of the polymer in the reaction medium. However, the PMDA series polymers have
high viscosity than the BTDA series polymers. This is in accord with the general
Table 5.
Solubility behaviour of PAI1a –PAI5d
Code PAI NMP DMF DMSO DMAc THF CHCl3
PAI3a BPAA + PPD ++ ++ +h ++ – –

PAI3b BPAA + ODA ++ ++ ++ ++ ± +h
PAI3c BPAA + MDA ++ ++ ++ ++ – ±
PAI3d BPAA + DCMDA ++ ++ ++ ++ ± +h
PAI4a BBAA + PPD ++ ++ +h ++ – ±
PAI4b BBAA + ODA ++ ++ ++ ++ +h +h
PAI4c BBAA + MDA ++ ++ ++ ++ – ±
PAI4d BBAA + DCMDA ++ ++ ++ ++ +h +h
PAI5a BPMCA + PPD ++ ++ ++ ± – ±
PAI5b BPMCA + ODA ++ ++ +h ++ ± +h
PAI5c BPMCA + MDA ++ ++ ++ ++ ± ++
PAI5d BPMCA + DCMDA ++ ++ +h ++ +h +h
PAI6a BBMCA + PPD ++ ++ ++ ++ +h –
PAI6b BBMCA + ODA ++ ++ ++ ++ ++ ±
PAI6c BBMCA + MDA ++ ++ ++ ++ ± ±
PAI6d BBMCA + DCMDA ++ ++ ++ ++ ± ±
PAI7a BPMBA + PPD ++ ++ ++ ++ ± ±
PAI7b BPMBA + ODA ++ ++ ++ ++ +h ±
PAI7c BPMBA + MDA ++ ++ ++ ++ ++ ++
PAI7d BPMBA + DCMDA ++ ++ ++ ++ ++ ++
PAI8a BBMBA + PPD ++ ++ ++ ++ ± ++
PAI8b BBMBA + ODA ++ ++ ++ ++ ± ++
PAI8c BBMBA + MDA ++ ++ ++ ++ ++ ++
PAI8d BBMBA + DCMDA ++ ++ ++ ++ ++ ++
PAI9a BPCA + PPD ++ ++ ++ +h – –
PAI9b BPCA + ODA ++ ++ ± ++ ++ ++
PAI9c BPCA + MDA ++ ++ ++ ++ ++ ++
PAI9d BPCA + DCMDA ++ ++ ++ ++ ± ±
PAI10a BBCA + PPD ++ ++ ++ ++ ± ±
PA I10b BBCA + ODA ++ ++ +h ++ +h ++
PAI10c BBCA + MDA ++ ++ ++ ++ ± ±
PAI10d BBCA + DCMDA ++ ++ ++ ++ ++ ++
PAI11a BPCPA + PPD ++ ++ ++ ± – ±
PAI11b BPCPA + ODA ++ ++ +h ++ ± +h
PAI11c BPCPA + MDA ++ ++ ++ ++ ± ++
PAI11d BPCPA + DCMDA ++ ++ ++ ++ +h ++
PAI12a BBCPA + PPD ++ ++ ++ ± ± +h
PAI12b BBCPA + ODA ++ ++ ++ ++ ± ±
PAI12c BBCPA + MDA ++ ++ ++ ++ ++ ++
PAI12d BBCPA + DCMDA ++ ++ ++ ++ ++ ++
++ — completely soluble, ± — sparingly soluble, +h — soluble on heating, – insoluble.
observation that the highly rigid PMDA moiety in the polymer backbone leads to
increased chain–chain interaction or in other words increased close packing of the
Table 6.
Inherent viscosity, moisture absorption and optical properties of PAIs
Code Inherent viscosity Moisture λ0 Transmittance

η (dl/g) absorption (%) (nm) (%)
PAI3a , PAI3b 0.98, 0.75 –, 1.79 408, 392 64.0, 66.0

PAI3c, PAI3d 0.84, 1.09 1.82, 1.68 372, 366 72.0, 75.0
PAI4a , PAI4b 0.93, 0.74 2.28, 2.23 402, 390 65.2, 66.8
PAI4c , PAI4d 0.81, 1.03 2.21, 1.83 363, 358 75.6, 76.8
PAI5a , PAI5b 0.98, 0.80 2.22, 1.77 372, 368 72.2, 74.7
PAI5c , PAI5d 0.84, 0.91 1.98, 1.65 354, 347 79.1, 79.6
PAI6a , PAI6b 0.93, 0.74 2.30, 1.80 361, 353 75.8, 79.0
PAI6c , PAI6d 0.79, 0.82 2.06, 1.73 346, 338 79.8, 80.0
PAI7a , PAI7b 1.02, 0.79 2.34, 1.89 370, 364 73.1, 75.2
PAI7c , PAI7d 0.81, 0.94 2.02, 1.78 358, 350 76.9, 79.2
PAI8a , PAI8b 0.99, 0.82 2.41, 1.93 366, 350 75.1, 79.3
PAI8c , PAI8d 0.89, 0.91 2.22, 1.80 344, 336 79.6, 80.1
PAI9a, PAI9b 1.05, 0.78 2.02, 1.61 358, 347 77.0, 79.2
PAI9c , PAI9d 0.94, 0.99 1.73, 1.58 343, 332 79.6, 80.8
PAI10a , PAI10b 1.03, 0.76 2.10, 1.68 355, 346 77.7, 79.7
PAI10c , PAI10d 0.91, 0.97 1.80, 1.62 338, 330 80.1, 81.0
PAI11a , PAI11b 1.09, 0.81 2.05, 1.79 353, 342 79.1, 79.6
PAI11c , PAI11d 0.98, 1.01 1.83, 1.70 335, 328 80.6, 81.4
PAI12a , PAI12b 1.02, 0.77 2.17, 1.86 348, 334 79.2, 80.3
PAI12c , PAI12d 0.93, 0.99 1.98, 1.78 326, 319 81.7, 85.0
λ0 = Cut-off wavelength.
polymer chains. The viscosity values also vary with the extent of polymer chain
mobility. The presence of methyl substituent on the benzene ring increases the seg-
mental rigidity of the polymer backbone resulting in increased solution viscosity.
Hence, the PAIs having methyl sustituents show higher viscosity values.
3.6. Moisture Absorption
Poly(amide–imide–imide)s usually show higher moisture uptake than the hydro-
carbon polymers because of the presence of polar imide group and amide linkages.
Moisture absorption of PAIs was determined from the weight change of the vac-
uum dried polymer samples after they were exposed to air at room temperature for
8 to 10 h (Table 6). The moisture absorption of aromatic polyimides can be up to
3.0–3.5% depending on the chemical structure of polymer and the relative humid-
ity of the surrounding environment, which has significant influence on the dielectric
properties of the polymer [38]. The moisture absorption values of the PAIs are in
the range of 1.58–2.17%. The moisture absorption of the prepared polymers might
be related to several factors including the chemical structure, introduction of halo
groups and other functional groups, the geometrical packing of the polymer chains
etc., among which chemical structure and the presence of functional groups in the
polymer might be major factors. The presence of bulky groups could loosen the
packing of the poly(amide–imide–imide) backbone chains resulting in increase of
free volume in the polymer which makes the polymer to entrap a large number of
water molecules.
3.7. Optical Properties
In general, the colored nature of aromatic polymers is due to the presence of
electronic conjugation, intermolecular and intramolecular CTC interactions, or
both [39]. The coloration of polymers prevents their widespread applications for
which colorlessness and transparency are of big concern. The coloration of the
poly(amide–imide–imide)s was evaluated from the cut-off wavelength (λ0 ) ob-
served in the UV-visible absorption spectra. The representative UV-visible spectra
of PAIs were obtained from various tetrimide diacids and diamines at a concentra-
tion of 5 × 10−3 mol/l in NMP solution. The cut-off wavelength and percentage
transmittance at 500 nm from the spectra are sumerized in Table 6. The results
(Fig. 9) also show that the PMDA-based polymers have higher λ0 than the BTDA-
based polymers, which is due to the increased CTC formation by PMDA moiety
[39]. The PAIs exhibit shorter cut-off wavelength because of the presence of bulky
pendant group in them. The bulky pendant unit inhibits chain–chain interaction be-
tween the polymer chains by increased entropy mechanism and thereby it decreases
the CTC formation. Besides, the presence of alkyl groups in the polymer back-
bone eliminates the electronic polarization by breaking up the extended conjugation
along the chain, which results in color. From the results it has been observed that
the colorlessness of the polymer was affected by the diamine segments between the
two amide groups. Thus, when the diacid moiety is fixed the color intensity of the
PAIs is in the following order p-PPD > ODA > DDM > DCDDM. The DCDDM
based PAI shows the lowest λ0 value because of the presence of electron withdraw-
ing chlorine in the diamine segment. With respect to the commercially available
polyimides these polymers have comparable transmittance in the UV-visible region
revealing that the prepared poly(amide–imide–imide)s have lighter color.
3.8. WAXD Studies
The wide-angle X-ray diffraction (WAXD) patterns of poly(amide–imide–imide)s
were obtained. The WAXD patterns of the polymers are shown in Fig. 10. The
PAIs with bulky pendent groups show amorphous diffraction patterns and this can
be explained by the incorporation of bulky and asymmetric pendant groups in the
polymer backbone which could possibly disrupt the hydrogen bonds that could be
formed between the amide groups of the polymer chain, thereby restricting the close
packing of the polymer.
3.9. Thermal Properties
The thermal properties of the polymers were evaluated by DSC and thermogravi-
metric analysis (TGA) at a heating rate of 20◦ C/min under nitrogen atmosphere
Figure 9. UV-visible spectra of PAIs.
and are summarized in Table 7. The thermograms of the PAIs are shown in Fig. 11.
The 10% weight loss temperatures (Td10 ) in nitrogen atmosphere are in the range of
475–506◦ C. All the PAIs are stable up to the temperature of 400◦ C indicating a high
thermal stability of the poly(amide–imide–imide)s. The PMDA-based polymers ex-
hibit higher thermal stability in comparison to BTDA-based polymers due to inher-
ent rigidity of PMDA unit [40, 41]. The order of thermal stability of the PAIs with
variation in diamine is found to be as follows p-PPD > ODA > DDM > DCDDM.
The PAIs containing DDM show lowest thermal stability because of the presence
of highly thermo-oxidative –CH2 group in the polymer. The results also show that
(a) (b)
Figure 10. WAXD patterns of PAI5a (a) and PAI7a (b).
Table 7.
Thermal properties of PAIs
Code Tg (◦ C) T10 (◦ C) Char yield (%) Code Tg (◦ C) T10 (◦ C) Char yield (%)
PAI3a 321 560 61 PAI7a 319 542 62

PAI3b 290 548 66 PAI7b 300 538 60
PAI3c 288 546 64 PAI7c 286 531 55
PAI3d 285 545 67 PAI7d 276 522 58
PAI4a 318 558 62 PAI8a 315 538 60
PAI4b 282 540 67 PAI8b 298 536 58
PAI4c 280 536 65 PAI8c 287 528 54
PAI4d 275 525 68 PAI8d 271 520 57
PAI5a 316 552 62 PAI9a 309 510 62
PAI5b 290 547 58 PAI9b 294 498 60
PAI5c 274 538 52 PAI9c 282 489 54
PAI5d 268 532 54 PAI9d 276 486 59
PAI6a 312 548 61 PAI10a 305 496 58
PAI6b 287 546 60 PAI10b 292 492 53
PAI6c 270 534 64 PAI10c 286 487 51
PAI6d 262 528 63 PAI10d 272 482 51
T10 = 10% weight loss temperature.
the polymer with p-oriented phenylene has higher T10 value because of its higher
thermo-oxidative stability. The char yields of the PAIs are high (50–60%) at 800◦ C.
The char yields is an easy and important measurement which correlates to the abil-
ity of the material to sustain combustion. Fire induced formation of the char on top
of a polymer can protect the underlying material from the action of the flame. On
the other hand, the bulk polymer may undergo decomposition and form low molec-
ular weight volatile by-products that may, in fact, feed the combustion process. The
Figure 11. Thermograms of PAIs.
Figure 12. DSC curves of PAIs.
presence of a char layer can serve as a barrier for the diffusion of these flammable
by-products into the flame, which renders the system self-extinguishing [42]. The
glass transition temperatures of the prepared polymers are in the range 268–305◦ C.
The PAIs with bulky pendant groups and tetra methyl substitution show higher Tg
values because of the restricted polymer chain rotation. The DSC curves of a few
poly(amide–imide–imide)s are shown in Fig. 12.
4. Conclusions
A new class of poly(amide–imide–imide)s from a direct polycondensation of novel
tetrimide diacid and various diamines were prepared. These polymers show high
solubility in test solvents with high optical transparency and excellent thermal sta-
bility. Thus, this series of poly(amide–imide–imide)s exhibit good processibility
while retaining the thermal stability, indicating that they can be accepted as new
engineering plastic materials.
Acknowledgement
The authors acknowledge the financial support by RESPOND, Indian Space Re-
search Organisation, Thiruvananthauram, India.
References
6. D. Wilson, H. D. Stenzenberger and P. M. Hergenrother (Eds), Polyimides. Blackie, Glasgow, UK
(1991).
7. K. L. Mittal (Ed.), Polyimides: Synthesis, Characterization and Applications, Vols 1 and 2.
Plenum, New York, NY (1984).
8. J. G. Wirth, G. S. Kirshenbaum and R. B. Seymour (Eds), High Performance Polymers; Their
Origin and Development. Elsevier, Amsterdam (1986).
9. C. P. Yang and W. T. Chen, Macromolecules 26, 4865 (1993).
10. C. P. Yang and W. T. Chen, J. Polym. Sci., Part A: Polym. Chem. 32, 435 (1994).
11. T. T. Serafini, P. Delvigs and G. R. Lightsey, J. Appl. Polym. Sci. 16, 905–915 (1972).
12. M. A. Meador, P. J. Carvano and D. C. Malarik, in: Proc. ASM/ESD 6th Annual Advanced Com-
posites Conf., ASM Intl., Materials Park, OH, pp. 529–539 (1998).
13. J. N. Hay, J. D. Boyle, S. F. Parker and D. Wilson, Polymer 30, 1032–1040 (1989).
14. W. Vancraeynest and J. K. Stille, Macromolecules 13, 1032–1040 (1980).
15. J. P. Droske, U. M. Gaik and J. K. Stille, Macromolecules 17, 10–18 (1984).
16. N. Bilow, L. B. Keller, A. L. Landis, R. H. Boschan and A. Castillo, in: Proc. 23rd National
SAMPE Symp. 23, 791–805 (1978).
17. S. H. Hsiao, C. P. Yang, M. H. Chuang and S. J. Lin, J. Polym. Sci., Polym. Chem. 37, 4510 (1999).
18. S. H. Hsiao and K. Y. Chu, J. Polym. Sci. Polym. Chem. 35, 3385 (1997).
19. S. H. Hsiao and C. F. Chang, Macromol. Chem. Phys. 197, 1255 (1996).
20. G. Maglio, R. Palumbo and M. C. Vignola, Macromol. Chem. Phys. 196, 2775 (1995).
21. C. P. Yang and J. J. Cheng, J. Polym. Sci., Polym. Chem. 33, 2209 (1995).
22. M. J. Marks and J. K. Sekinger, Macromolecules 27, 4106–4114 (1994).
23. Y. Taguchi, H. Uyama and S. Kobayashi, J. Polym. Sci., Part A: Polym. Chem. 34, 561–565 (1996).
24. D. J. Liaw, B. Y. Liaw and J. J. Chen, J. Polym. Sci., Polym. Chem. 38, 797 (2000).
25. J. F. Espeso, J. G. delaCampa, A. E. Lozano and J. de Abajo, J. Polym. Sci., Polym. Chem. 38, 104
(2000).
26. Y. T. Chern and H. C. Shiue, Macromolecules 30, 5766–5772 (1997).
27. Y. T. Chern and H. C. Shiue, Macromolecules 30, 4646–4651 (1997).
28. C. J. Billerbeck and S. J. Henke, Engineering Thermoplastics. Marcel Dekker, New York, NY
(1985).
29. D. J. Liaw and B. Y. Liaw, Polymer 40, 3183 (1999).
30. D. J. Liaw and B. Y. Liaw, Macromol. Chem. Phys. 200, 1326 (1999).
31. S. Bhuvana and M. Sarojadevi, J. Polymer Res. 14, 261–267 (2007).
32. S. Bhuvana and M. Sarojadevi, Polymer Int. 56, 957–967 (2007).
33. S. Bhuvana, M. Madhumathi and M. Sarojadevi, Polymer Bull. 57, 61–72 (2006).
34. S. Bhuvana, R. Hariharan and M. Sarojadevi, J. Macromol. Sci. Pure Appl. Chem. 42, 1–12 (2005).
35. C. P. Yang and W.-T. Chen, J. Polym. Sci., Polym. Chem. Edn 32, 435–444 (1994).
36. C.-P. Yang, R.-S. Chen and C.-S. Wei, Eur. Polym. J. 38, 1721–1729 (2002).
37. C. P. Yang and S. H. Hsiao, J. Appl. Polym. Sci. 30, 2883 (1985).
38. K. Xie, J. G. Liu, H. W. Zhou, S. Y. Zhang, M. H. He and S. Y. Yang, Polymer 42, 7267 (2001).
39. C. P. Yang, R. S. Chen and K. S. Hung, Polymer 42, 4569 (2001).
40. R. Hariharan, S. Bhuvana, G. Anuradha and M. Sarojadevi, Polym. Int. 53, 1442 (2004).
41. R. Hariharan, S. Bhuvana, M. Anitha Malbi and M. Sarojadevi, J. Appl. Polym. Sci. 93, 1846–1853
(2004).
42. P. Carty, in: Polymeric Materials Encyclopedia, J. C. Salamone (Ed.), p. 2422. CRC Press, Boca
Raton, FL (1996).
Structural Characterization and Mechanical and
Thermal Properties of New Cyanate/Epoxy and
Cyanate/BMI Blends and Composites
M. Sarojadevi a,∗ , L. S. Jayakumari b , V. Thulasiraman a , G. Anuradha a

and P. S. Sampath b
a
b
Division of Rubber and Plastics Technology, Madras Institute of Technology,
Chrompet, Chennai-600044, India
Abstract
Different diols were prepared by the condensation of 2,6-dimethyl phenol/1-naphthol with different alde-
hydes, dichlorodiphenylsilane with 1,4-dihydroxy anthraquinone and vanillin with different diamines. All
the diols prepared and 1,4-dihydroxy anthraquinone were converted to dicyanate. The synthesized diols and
dicyanates were characterized by elemental analysis, FT-IR, 1 H-NMR and 13 C-NMR. The cure behaviour
of all the dicyanate monomers and their blends with epoxy/BMI was studied using DSC. The monomers
were then cured into thermoset polymers after heating at 180◦ C for 1 h and 220◦ C for 1 h. FT-IR, solid
state NMR, TGA and DSC were used to characterize the polymers. All the polymers exhibit a common
maximum decomposition temperature of 420 ± 5◦ C, which is characteristic of the cyanurate ring. However,
the percentage char yield is found to depend on the chemical structure. The Tg of the polycyanates depends
on the structure of the monomer. The dicyanates were blended with commercial epoxy/DDM system. The
Schiff base functionalized cyanate ester was blended with BMI at various proportions. The blends were
characterized by FT-IR. Composite laminates were fabricated from E-glass fiber plain-woven fabric and
epoxy resin or cyanate/epoxy or BMI blend. The mechanical properties such as tensile, flexural and impact
strengths and fracture toughness of the blends and composites were studied as per ASTM standards. The
blends and composites were evaluated by SEM analysis and DMA.
Keywords
Cyanate ester, epoxy blends, BMI blends, composites, mechanical properties
1. Introduction
The application of cyanate esters (Cy) as an aerospace material (as advanced com-
posites matrices) is a more recent development and this places them in direct com-
petition for the market share of epoxy and bismaleimide (BMI) resins. However,
* To whom correspondence should be addressed. E-mail: msrde2000@yahoo.com
being at an earlier stage of development than the more established products, they
are currently more expensive, and recent commercial developments have involved
the preparation of binary [1] (Cy/BMI, Cy/epoxy) or ternary [2] (Cy/BMI/epoxy)
blends. It is now generally accepted that cyanates and BMIs do not co-react di-
rectly [3]; instead they form interpenetrating networks. There are also reports on the
chemical reaction between cyanate and epoxy resin making them highly compati-
ble [4]. Hence attempts were made to toughen epoxy and BMI with novel cyanate
esters, thereby the utilization cost (processing, fabrication as composites) of the
cyanate esters will be reduced. Three new bisphenols were prepared and converted
to their respective cyanate esters. Two of the cyanate esters were blended with epoxy
resin and mechanical properties of blends and composites were studied. Similarly,
the third cyanate ester was blended with BMI and the mechanical properties of its
composites were studied.
2. Experimental
2.1. Materials
Vanillin, anisaldehyde, 1,4-dihydroxy anthraquinone, p-toluene sulphonic acid
were purchased from E-Merck, India. Triethylamine and cyanogen bromide (CNBr)
were obtained from Spectrochem, India and used as received, except triethyl
amine which was distilled from CaH2 prior to use. 4-Nonyl phenol was pur-
chased from Aldrich, India. 4,4 -dichloro diphenyl silane was purchased from Lan-
caster, India. 2-propanol, 2,6-dimethyl phenol, N,N-dimethyl formamide (DMF)
and toluene were purchased from Sisco Research Laboratories, India. Commer-
cial epoxy resin LY556 (diglycidyl ether of bisphenol-A) and 4,4 -diaminodiphenyl
methane (DDM) HT 972 were obtained from Huntsman India Ltd. Plain-woven
glass fabrics with an aerial density of 200 gsm (gram/square metre) were obtained
from Saint Gobain, India. Cobalt naphthenate was purchased from Qualigens, India.
Commercial BMI, 1,1 -(methylene di-1,4-phenylene bismaleimide), was purchased
from Abron Organics, Hyderabad, India. All reagents were used without any purifi-
cation, unless stated otherwise.
2.2. Measurements
FT-IR (Fourier Transform Infrared spectroscopy) spectra were obtained using a
Perkin Elmer spectrum RX1 spectrometer with KBr pellet. NMR spectra were
recorded on an AMX-400 NMR spectrometer (Bruker, Rheinstetten, Germany) at
400 MHz. The samples were dissolved in DMSO-d6 or CDCl3 and spectra were ac-
quired at 298 K using tetramethylsilane (TMS) as the internal standard. Solid-state
13 C-NMR spectra for the polycyanurates were obtained with a Bruker ARX 300
spectrometer operating in quadrupole detection (1/2 integer quadrupole spins, fast

efficient relaxation leading to line broadening). Elemental analysis was performed
on a Carlo Erba EA 1108 microanalyzer. Melting points were determined on an
Cy/Epoxy and Cy/BMI Blends and Composites 121
electrothermal melting point apparatus IA 6304 (Electrothermal, London, UK) us-

ing capillary tubes and were not corrected. Differential scanning calorimetry (DSC)
studies for the cyanate/epoxy blends were carried out using TA Instruments Q10
model at a heating rate of 10◦ C/min up to 300◦ C in flowing nitrogen (60 ml/min).
Thermogravimetric analysis (TGA) of the cyanate/epoxy blends was carried out
with TA Instruments SDT Q600 model at a heating rate of 20◦ C/min in N2 at-
mosphere.
The fibre volume fraction of the composites was evaluated by acid digestion
method as per ASTM D 3171 procedure B. The tensile, fracture toughness and
flexural properties of the blends and composites were investigated using a Uni-
versal Testing Machine (Model H50 K-S, Hounsfield Test Equipment Ltd, UK).
The crosshead speed was maintained at 1 mm/min. The gauge length of the speci-
men was 150 mm. The tensile properties were evaluated as per ASTM D 3039 for
composites and ASTM D 638 for blends. The flexural properties were measured
as per ASTM D 790 at a crosshead speed of 1 mm/min. The double cantilever
beam (DCB) test samples for mode-I fracture toughness (G1C ) measurements were
prepared as per ASTM D 5528 with a pre-initiated crack of 50 mm length. The
crosshead displacement rate was 1 mm/min. The impact strengths of the blends and
composites were studied as per ASTM D 256-88 using unnotched specimen with
an Izod Impact Tester (International Engineering Industries, Mumbai, India). The
dynamic mechanical properties of the cyanate ester/BMI and cyanate/epoxy blend
composites were obtained using a NETZSCH DMA 242 dynamic mechanical an-
alyzer in a three-point bending mode. Rectangular specimens of 55 mm length,
10 mm width and 3 mm thickness were used. Each specimen was heated from
room temperature to 400◦ C under different frequencies namely 2 Hz, 5 Hz, 10 Hz
and 20 Hz. The storage modulus E and the mechanical loss factor (damping) tan δ
were recorded as a function of temperature. The glass transition temperatures (Tg )
of the blends were obtained from the peaks of loss factor (tan δ) curves.
Morphological studies were conducted using scanning electron microscopy
(JEOL, JSM 840 A). Micrographs were taken on the surface, which was made by
breaking the specimen by an impact testing machine and then coating the ends
with gold powder. Photographs were taken on a Polaroid film at a magnification of
1000×.
2.3. Synthesis of Bis Aryl Hydroxy Diphenyl Silane (BAHPS)
1,4-Dihydroxy anthraquinone (48.0 g, 0.2 mol), anhydrous triethyl amine (20.2 g,
0.20 mol) and freshly distilled DMF (200 ml) were taken in a 500 ml three-necked
round bottomed flask which was dried, purged with nitrogen, and fitted with a me-
chanical stirrer and a condenser. The mixture was stirred for 30 min, and then heated
to 80◦ C. Dichloro diphenyl silane (25.2 g, 0.1 mol) was added dropwise to a well
stirred slurry mixture. The reaction mixture was refluxed and continuously stirred
for 20 h. After the completion of the reaction, the reaction mixture was poured into
500 ml of water, extracted with ethyl acetate and washed with water three times.
Scheme 1. Synthesis of bis aryl hydroxyl and cyanato diphenyl silane.
The solvent was removed by using a rotary evaporator and the brown colour solid
formed was collected and recrystallized in water/methanol solution (Scheme 1).
BAHPS — Yield 91%, m.p.: 143◦ C. Elemental analysis (%) for C40 H24 O8 Si:
calculated: C: 72.72, H: 3.66; found: C: 72.01; H: 3.04. FT-IR: 3332 cm−1 (–OH
stretch), 1620 cm−1 (C=O stretch), 960 cm−1 (Si–O–C stretch), 1437 cm−1 (Si–C
stretch). 1 H-NMR (DMSO-d6 , ppm): 7.54 (4H, d, Ha ), 7.33 (4H, t, Hb ), 7.37 (2H, t,
Hc ), 7.00 (2H, d, Hd ), 6.95 (2H, d, He ), 7.52–7.54 (m, Hf ), 7.74–7.77 (m, Hg ), 9.85
(2H, s, Hh ). 13 C-NMR (DMSO-d6 , ppm): 133 (C1 ), 132 (C2 ), 128.4 (C3 ), 130.2
(C4 ), 156.7 (C5 ), 122.6 (C6 ), 121.4 (C7 ), 152.1 (C8 ), 132.9 (C9 ), 133.3 (C10 ), 180.1
(C11 ), 136 (C12 ), 130.2 (C13 ), 132.4 (C14 ).
2.4. Synthesis of Bis(4-hydroxy-3,5-dimethylphenyl) Aryl Methane (BHAM)
A three-necked flask equipped with a condenser, a Dean–Stark water separator and
a nitrogen inlet was charged with 32.98 g (0.27 mol) of 2,6-dimethyl phenol, 19.2 g
(0.12 mol) of anisaldehyde, 1.63 g (0.012 mol) of p-toluene sulphonic acid mono
hydrate and 38 ml of toluene. The reaction mixture was refluxed at 132–134◦ C
for 5 h under nitrogen atmosphere. Water produced by the reaction was succes-
sively removed from the system using Dean–Stark apparatus. After completion of
the reaction, the solution was cooled to 80◦ C and 50 ml of 10% aqueous solution
of sodium hydroxide was added to the resulting solution. The solid formed was
filtered and recrystallized from methanol/water (3:1 v/v) to produce light brown
crystals (Scheme 2).
BHAM — Yield 90%. Elemental analysis (%) for C24 H26 O3 : calculated: C:
79.55, H: 7.18; found: C: 80.5, H: 7.47. FT-IR: 3511 cm−1 (OH stretch), 2923 cm−1
(CH3 stretch). 1 H-NMR: 4.8 ppm (OH) (Figs 1–3 represent the FT-IR and NMR
spectra of BHAM).
Scheme 2. Synthesis of BHAM and BCAM.
Figure 1. FT-IR spectrum of BHAM.
2.5. Synthesis of Bis(4 -hydroxy-3 -methoxybenzal)-4,4 -diaminodiphenyl

Methane (BHDM)
Synthesis of BHDM was carried out using the procedure reported elsewhere [5].
Yield: 91%, m.p.: 152◦ C. The FT-IR spectrum shows absorption at about 1605–
1630 cm−1 due to the –CH=N bond. The band at 3250–3500 cm−1 is assigned
to O–H stretching. The disappearance of sharp bands due to C=O and NH2 at
1680 cm−1 and 3334 cm−1 , respectively, confirms the completion of the reaction
(Scheme 3).
BHDM — Elemental analysis (%) for C29 H26 N2 O4 : calculated: C: 74.66, H:
5.61, N: 6.01; found: C: 6.01; H: 5.42; N: 5.83. FT-IR (KBr, cm−1 ): 3454 (ν O–H ),
1609 (ν C=N ), 1265 (asym ν C–O–C ), 1085 (sym ν C–O–C ), 2872 (aldehydic ν C–H ),
absence of –CHO and –NH2 bands at 1740 and 3400 cm−1 . 1 H-NMR (DMSO-d6 ,
ppm): 7.06 (4H, d, Ha ), 7.14 (4H, d, Hb ), 8.49 (2H, s, Hc ), 7.21 (2H, d, Hd ), 7.24
(2H, d, He ), 3.73 (6H, s, Hf ), 7.39 (2H, s, Hg ), 1.6 (2H, s, Hh ), 5.6 (2H, s, –OH).
Figure 2. 1 H-NMR spectrum of BHAM.
Figure 3. 13 C-NMR spectrum of BHAM.
13 C-NMR (DMSO-d6 , ppm): C1 –45.6, C2 –140.0, C3 –122.4, C4 –129.7, C5 –152.6,

C6 –160.0, C7 –137.2, C8 –131.4, C9 –116.0, C10 –155.5, C11 –∗, C12 –49.5,
C13 –150.3 (∗ — absence of characteristic carbon of OCN in dihydroxy compound).
Scheme 3. Synthesis of BHDM and BCDM.
2.6. Synthesis of Cyanate Ester Monomers
To a 250 ml three-necked round bottom flask fitted with an overhead stirrer, nitrogen
inlet and dropping funnel was added 75 ml of acetone (dried over fresh 4 Å molecu-
lar sieves for one day). After cooling to −10◦ C, 0.09 mol (9.53 g) of CNBr (caution:
toxic!) was added and the solution was further cooled to −15◦ C. A solution of
0.04 mol (18.64 g) of BHDM and 0.04 mol (5.54 ml) of triethyl amine in 25 ml of
acetone was prepared, transferred to a dropping funnel, and added dropwise with
rapid stirring over a period of 45 min while maintaining the reaction temperature at
−15◦ C. When the mixture had warmed to −2◦ C it was rapidly filtered and the qua-
ternary ammonium bromide salt (Et3 N · HBr) was removed. The filtrate was poured
into cold distilled water to precipitate the cyanate ester from solution. The precip-
itate was washed with aqueous hydrochloric acid solution to remove the base. The
crude product was further purified by recrystallization in methanol: water (1:1 v/v)
mixture. All other prepared bisphenols were converted into the corresponding di-
cyanate ester monomers in a similar way (BAHPS–BCAPS, BHAM–BCAM and
BHDM–BCDM).
BCAPS — Yield 91%, m.p.: 127◦ C. Elemental analysis (%) for C42 H22 O8 N2 Si:
calculated: C: 70.98, H: 3.12; N: 3.94; found: C: 70.31, H: 2.75, N: 3.63. FT-IR:
2259 cm−1 (OCN stretch), 1627 cm−1 (C=O stretch), 958 cm−1 (Si–O–C stretch),
1435 cm−1 (Si–C stretch). 1 H-NMR (DMSO-d6 , ppm): 7.59 (4H, d, Ha ), 7.39 (4H,
t, Hb ), 7.42 (2H, t, Hc ), 7.08 (2H, d, Hd ), 7.01 (2H, d, He ), 7.52–7.54 (m, Hf ), 7.74–
7.77 (m, Hg ). 13 C-NMR (DMSO-d6 , ppm), 133.2 (C1 ), 132.6 (C2 ), 128.6 (C3 ),
130.3 (C4 ), 156.9 (C5 ), 122.9 (C6 ), 121.5 (C7 ), 152.3 (C8 ), 133.1 (C9 ), 133.6 (C10 ),
180.3 (C11 ), 136.1 (C12 ), 130.2 (C13 ), 132.5 (C14 ), 108.6 (C15 ).
Figure 4. FT-IR spectrum of BCAM.
BCAM — Yield 90%, m.p.: −90◦ C. Elemental analysis (%) for C26 H24 O3 N2 :
calculated: C: 75.73, H: 5.83, N: 6.79; found: C: 74.52, H: 6.02, N: 6.59. FT-IR:
224 5cm−1 (OCN stretch), 2923 cm−1 (CH3 stretch). 13 C-NMR 108.1 ppm (OCN).
Figures 4–6 show the FT-IR and NMR spectra of BCAM.
BCDM — Elemental analysis (%) for C31 H24 N4 O4 : calculated: C: 72.09, H:
4.68, N: 10.84; found: C: 71.89, H: 4.72, N: 10.72. FT-IR (KBr, cm−1 ): 2234, 2270
(characteristic bands of –OCN group), 1605–1630 (–CH=N stretching), 1265, 1085
(C–O–C asymmetric and symmetric stretchings), 2966 (C–H stretching of –CH3 ).
1 H-NMR (DMSO-d , ppm, Scheme 4): 7.03 (4H, d, H ), 7.12 (4H, d, H ), 8.53 (2H,
6 a b
s, Hc ), 7.24 (2H, d, Hd ), 7.31 (2H, d, He ), 3.80 (6H, s, Hf ), 7.45 (2H, s, Hg ), 1.4
(2H, s, Hh ). 13 C-NMR (DMSO-d6 , ppm): C1 –45.8, C2 –140.1, C3 –122.6, C4 –129.8,
C5 –152.7, C6 –160.1, C7 –137.4, C8 –131.7, C9 –115.9, C10 –156.1, C11 –109.8, C12 –
49.4, C13 –150.2 (Fig. 7 shows the FT-IR spectrum of BCDM).
2.7. Solid-State NMR of Cured Polycyanurate (BCAM)
Solid state 13 C-NMR spectra (CP/MAS) (crossed polarization with magic angle
spinning) were obtained for all the polycyanurates. The representative spectrum
of polymer BCAM is shown in Fig. 8. From the spectrum it is observed that a
characteristic signal around 178 ppm confirms the formation of triazine ring as a
result of cyclotrimerization reaction [6].
2.8. Fabrication of Neat and Blended Resin Castings (Without Glass Fibre
Reinforcement)
In this study BCAM and BCAPS were blended with epoxy resin at different ratios.
Employing the compositions (0, 3, 6 and 9% of cyanate in epoxy for BCAM and 0,
3, 6, 9 and 15% of cyanate in epoxy for BCAPS) different formulations of cyanate
(BCAM/BCAPS), epoxy resin (EP), 4,4 -diaminodiphenyl methane (DDM) were
Figure 5. 1 H-NMR spectrum of BCAM.
Figure 6. 13 C-NMR spectrum of BCAM.
Scheme 4. Naming for NMR spectra (BCDM).

Figure 7. FT-IR spectrum of BCDM.
Figure 8. Solid-state NMR spectrum of BCAM.
made. The blends were thoroughly mixed and melted at 90◦ C in an oil bath and
degassed to remove air bubbles. The homogeneous liquid was then transferred to a
preheated open mould at 100◦ C coated with a silicone-based release agent to facil-
itate release of cured samples. All the castings (without glass fibre reinforcement)
were cured at 120◦ C for 2 h, at 180◦ C for 1 h, demoulded and further post-cured at
220◦ C for 1 h. Similar procedure was adopted for BCAPS.
2.9. Fabrication of Composites

In this study BCAM and BCAPS were blended with epoxy resin and BCDM was
blended with BMI at different ratios. The laminate composites were fabricated from
E-glass fibre plain-woven fabric with an aerial density of 200 g/m2 and epoxy
resin/cyanate-modified epoxy resin by hand lay-up technique (coating the resin on
the glass fibre by hand for every layer). The liquid epoxy resin was taken in a beaker,
heated to 90◦ C to lower the resin viscosity and the desired amount of cyanate was
added into the resin. The cyanate loading was varied between 0, 3, 6, 9 and 15%
by weight of epoxy resin. The mixture was degassed in a vacuum oven followed by
addition of DDM (curing agent) in 27% by weight of epoxy and stirred for 3 min at
90◦ C.
A steel mould plate was coated with a silicone release agent and then a layer
of the resin was applied using a brush. Then, the first layer of the fibres (160 ×
160 mm) was placed on the resin and consolidated using rollers. This process was
repeated to construct a 14-ply laminate. Necessary precautions were taken to keep
the fabric well aligned. A 15-µm-thick Teflon release film was inserted in the mid-
plane of the laminate as an initial crack for the mode-I fracture toughness test. The
fabricated sheet was then cured at 120◦ C for 1 h and 180◦ C for 1 h in a hydraulic
press. The sheet was then demoulded and post-cured at 220◦ C for 1 h. The sam-
ple codes for blends of BCAM and BCAPS are CY-2b and CY-5b, respectively.
The sample codes for composites of BCAM and BCAPS are CY-2c and CY-5c,
respectively.
A number of blend composites were made by employing formulations of cyanate
ester BCDM and BMI. The blends were thoroughly mixed together by dissolv-
ing in a low boiling solvent like acetone to obtain a homogeneous liquid. The
blends were catalysed by cobalt(II) naphthenate (300 ppm) and nonyl phenol (4 phr
(parts per hundred parts of resin)). The fabricated sheet was then cured at 100◦ C
(30 min) → 150◦ C (30 min) → 200◦ C (60 min) in a hydraulic press. The sheet was
then demoulded and post-cured at 250◦ C for 3 h.

3.1. Infrared Spectroscopy of Cyanate (BCAM/BCAPS) and Epoxy Blends
FT-IR spectroscopy was used to confirm the structure of neat epoxy resin and
cyanate-modified epoxy system. The disappearance of absorption band at
2260 cm−1 for –O–C≡N and appearance of new absorption bands at 1575 cm−1
and 1357 cm−1 for N–C=N and N–C–O confirm the formation of polycyanurate
ring after curing. The cyanate and epoxy resins when blended together react chem-
ically as reported [7]. The cyclotrimerization of the cyanate ester groups among
themselves and the formation of oxazolidinone and oxazoline rings by reaction
between cyanate and epoxy groups takes place in the blend as reported by other
authors [8]. Figures 9 and 10 represent the FT-IR spectra of BCAM and BCAPS
Figure 9. FT-IR spectra of (a) 3CY-2b, (b) 6CY-2b and (c) 9CY-2b.
cyanate/epoxy systems showing an absorption band at 1750 cm−1 , confirming the

formation of oxazolidinone ring. The intensities of absorptions at 1587 cm−1 and
1357 cm−1 due to cyanurate are found to decrease in the cyanate/epoxy blend,
further confirming the co-reaction.
3.2. SEM Analysis
The phase separation brought about by the inclusion of cyanate into the epoxy
system was studied by SEM analysis. The SEM micrographs of the fractured sur-
Figure 10. FT-IR spectrum of 3CY-5b.
faces of pure EP, 3CY-2b, 6CY-2b, 9CY-2b and 3CY-5b at 1000× magnification are
shown in Figs 11 and 12. The bright lines that emanate from the crack with the cut
edges were observed in pure epoxy system. The polymer ligament tear appears as
straight cut with a sharp edge. This shows that the cured unmodified resin is brittle,
whereas for the modified resins, the fractured surfaces show local shear deforma-
tion (the cracks appear as more diffused and irregular). Although a few definite cut
edges were seen, the local yielding and shearing seems to be prevalent.
3.3. Mechanical Properties of BCAM/BCAPS Epoxy Blends
The tensile, flexural and impact properties of pure epoxy and cyanate-modified
epoxy systems were measured as per ASTM standards and are presented in Ta-
ble 1. The tensile strength and tensile modulus values are found to increase with
increase in BCAM cyanate loading. The tensile strength values of 3CY-2b, 6CY-2b
and 9CY-2b are 55 MPa, 63 MPa, 76 MPa, respectively. Similarly for cyanate-
modified epoxy systems (CY-5b), the tensile properties are found to increase with
increase in cyanate loading [9]. Tensile strength values of CY-5b system are 59, 63,
66 and 70 MPa for 3, 6, 9 and 15% cyanate loadings. The reason for an increasing
trend in tensile properties may be attributed to an increase in the rigid anthraquinone
group which outweighs the effect of oxazolidinone group.
The introduction of 3%, 6% and 9% (by wt) of BCAM into epoxy resin increases
the flexural strength by 1.02, 1.7 and 1.9 times, respectively, for CY-2b blends. The
enhanced flexural property is due to the formation of a network structure between
(a) (b)
(c) (d)
Figure 11. SEM micrographs of neat epoxy and epoxy/cyanate blends. (a) Neat epoxy system, (b) 3%
BCAM in epoxy, (c) 6% BCAM in epoxy and (d) 9% BCAM in epoxy.
Figure 12. SEM micrograph of 3CY-5b system.
cyanate ester and epoxy matrix. The formation of aliphatic oxazolidinone imparts
more thermoplastic character and hence more resistance to bending stresses [10].
Table 1.
Mechanical properties of epoxy and cyanate-modified epoxy systems
Blend code Tensile strength Tensile modulus Flexural strength Impact strength
(MPa) (GPa) (MPa) (J/cm)
EP 52 2.4 89 0.67
3CY-2b 55 2.5 96 1.3
6CY-2b 63 2.6 148 2.0
9CY-2b 76 2.6 169 2.7
3CY-5b 59 – 112 1.6
6CY-5b 63 – 129 2.1
9CY-5b 66 – 141 3.6
15CY-5b 70 – 165 4.5
The flexural strengths of CY-5b are 112, 129, 141, 165 MPa for 3, 6, 9 and 15%
loadings, respectively. It was found that 15CY-5b has higher flexural strength. This
may be due to the presence of oxazolidinone and anthraquinone moieties which
resist the bending stress.
From Table 1, it is found that the impact strength increases with cyanate load-
ing. The enhancement in impact strength may be due to reduced cross-link density
and high percentage of flexible ether linkages in the network. The impact values of
pure EP, 3% CY-2b, 6% CY-2b and 9% CY-2b are 0.67, 1.3, 2.0 and 2.7 J/cm, re-
spectively. The impact strength values of siloxane containing cyanate ester systems
CY-5b (3%, 6%, 9% and 15%) are higher than that of the EP system. The impact
strength of 15CY-5b system is higher when compared to other blend systems (Ta-
ble 1). This is because of higher cyanate content leading to more oxazolidinone
linkages.
3.4. Cure Analysis by DSC
The cure characteristics of the dicyanate ester BCAM (CY-2b) and their 3%, 6% and
9% blends were monitored by DSC analysis. The temperatures corresponding to the
onset of cure Ti , the maximum cure temperature Tp and end of the cure temperature
Tf are presented in Table 2. From the table it is evident that the curing reaction peaks
at 200◦ C for BCAM dicyanate ester. The DSC curves of BCAM-modified epoxy
systems with 3%, 6% and 9% cyanate loadings during cure are shown in Fig. 13.
It can be seen from the figure that the cyanate-modified epoxy systems have a peak
temperature much lower than that of pure cyanate system. The peak temperatures
of cyanate-modified epoxy systems are 164, 165 and 169◦ C for 3CY-2b, 6CY-2b
and 9CY-2b, respectively.
The decrease in peak temperature with increasing cyanate concentration con-
firms that it is the co-reaction between epoxy and cyanate that accelerates the cure
rate and reduces the cure temperature [11]. The large exothermic peak obtained
from cyanate ester epoxy systems is due to the polycyclotrimerization of cyanate
Table 2.
DSC cure characteristics of neat epoxy and cyanate-modified epoxy systems
Composition Ti (◦ C) Tp (◦ C) Tf (◦ C)
BCAM 182 200 229

3CY-2b 126 164 242
6CY-2b 128 165 231
9CY-2b 131 169 236
EP 95 165 226
3CY-5b 91 159 222
6CY-5b 88 155 218
9CY-5b 86 152 212
15CY-5b 80 146 208
Figure 13. DSC curves of CY-2b systems.
monomers, reaction of cyanate with amine/epoxide and reaction of epoxy with cya-
nurate. Similar trend was observed for the BCAPS epoxy blends (3%, 6%, 9% and
15%) which was monitored by DSC analysis.
3.5. Thermal Stability
The thermal stability of the blends was studied by thermogravimetric analysis.
The incorporation of cyanate into epoxy resin improves the thermal stability and
enhances the degradation temperature depending on its content. The T10 values
for neat epoxy and BCAM-modified epoxy blends are 358◦ C and 390◦ C, respec-
Table 3.
TGA data on cured neat epoxy and cyanate-modified epoxy systems
Blend code T10 (◦ C) Char yield (%)
EP 358 18.0
3CY-2b 390 21.0
6CY-2b 390 23.8
9CY-2b 390 26.2
3CY-5b 358 20.5
6CY-5b 355 22.4
9CY-5b 350 26.7
15CY-5b 347 30.8
tively (Table 3). The char yield of pure epoxy system is 18%. The char yield of
BCAM-modified epoxy blend systems are 21.0, 23.8 and 26.2%, respectively. The
increasing trend of char yield with incorporation of cyanate ester is due to in-
crease in aromatic content. In siloxane containing cyanate-modified epoxy blends
(BCAPS–CY-5b) the T10 value is found to decrease with increase in cyanate load-
ing. When the cyanate loading is increased, the thermal stability of the siloxane
containing cyanate ester blends is low [12], because the first stage decomposition
occurs at lower temperature. The Si–O groups of the cyanate ester absorb more
thermal energy and their vibration can dissipate the thermal decomposition energy.
The decomposition leads to the formation of the silicone containing groups which
participate in the cross-linked carbonization in the second stage decomposition and
effectively retard the flame formation at higher temperature (Table 3).
3.6. Testing of BCAM/BCAPS Epoxy Composites (BCAM–CY-2c, BCAPS–CY-5c)
3.6.1. Fibre Volume Fraction
Prior to the study of mechanical properties of composites, the fibre volume fraction
and void content of the epoxy and cyanate/epoxy composites were determined by
ASTM 3171 procedure B. The fibre volume fraction of the composites was found
to be in the range 42 ± 2%.
3.6.2. Dynamic Mechanical Analysis
Dynamic mechanical analysis (DMA) is a powerful technique for studying the vis-
coelastic behaviour of polymer based materials. It is also the preferred method
for measuring the glass transition temperature (Tg ), particularly for polymers with
rigid backbones. Figure 14 represents the DMA curve of tan δ of pure EP, 3CY-2c,
6CY-2c and 9CY-2c against temperature at a frequency of 2 Hz. The peak represents
the Tg of the particular resin system. The Tg values of neat epoxy and cyanate-
modified epoxy resins are summarized in Table 4. The DMA curve with a single
Tg further indicates that the blend system is a single-component system [13]. The
temperature dependence of the storage modulus E , for the neat epoxy composite
and cyanate/epoxy composites are shown in Table 4. From Table 4 it is seen that,
Figure 14. Dynamic mechanical curves of pure epoxy and CY-2c cyanate-modified epoxy/glass fiber
composites.
Table 4.
Glass transition temperatures of epoxy and cyanate-modified epoxy/glass fibre composites
Sample Tg Storage modulus (GPa)

(◦ C)
50◦ C 100◦ C
EP 142 26 24
3CY-2c 140 30 29
6CY-2c 125 33 29
9CY-2c 147 29 29
3CY-5c 148 26 25
6CY-5c 150 28 27
9CY-5c 152 31 30
15CY-5c 157 32 30
6CY-2c (BCAM–CY-2c) composite exhibits a higher storage modulus (33 GPa) at

50◦ C than the neat composite (26.0 GPa), while the 9CY-2c composite exhibits
a lower modulus (29 GPa) [10]. In all the composites there is a decrease in stor-
age modulus on increasing the temperature (for EP at 50◦ C the storage modulus is
26.0 GPa and at 100◦ C it is 24 GPa) [14].
Figure 15 represents the DMA curves of tan δ of CY-5c cyanate ester with epoxy
blends at various compositions (3%, 6%, 9% and 15%). In the case of BCAPS–
CY-5c, the Tg value increases (though the height of tan δ peak decreases, the tem-
perature slightly shifts to higher value) as the cyanate loading increases. This may
Figure 15. DMA curves of epoxy/cyanate (BCAPS–CY-5c) blends.
be due to the aromatic character and higher triazine ring content. It is seen that on
increasing the cyanate percentage, storage modulus increases. The increasing trend
in storage modulus value may be attributed to the rigid anthraquinone moiety (an
increase in the rigid aromatic groups) in CY-5c blend systems.
3.6.3.1. Effect of Cyanate Loading on Tensile Properties. The mechanical prop-
erties of epoxy and (BCAM) CY-2c-modified epoxy composites are presented in
Table 5. With the incorporation of fibre into the polymer matrix the tensile strength
of the latter is found to increase by about 6 times. The decrease in tensile strength
of 6CY-2c-modified epoxy system may be due to the presence of pendant phenyl
groups which reduce the close packing of polymer molecules, thus reducing the
cross-link density and increasing the free volume [15].
From the table it is seen that the tensile strength of epoxy composite is 322 MPa
and for (BCAPS) CY-5c cyanate-modified epoxy composites the value is increased
to 343 MPa at 15% loading. It is observed that the cyanate-modified epoxy compos-
ites have higher tensile strength when compared to neat epoxy composite. A similar
increasing trend is observed with tensile modulus values on increasing the cyanate
content in the composites.
3.6.3.2. Effect of Cyanate Loading on Flexural Properties. It is observed that
flexural strengths are found to increase by 5 ± 3%, 11 ± 6% and 17 ± 2%, re-
spectively, for 3%, 6% and 9% of CY-2c-modified epoxy system (Table 5). The
introduction of 3, 6, 9 and 15% (by weight) of CY-5c cyanate ester into epoxy
resin increases the flexural strength of its composites from 409 (for unmodified
Table 5.
Mechanical properties of pure epoxy and cyanate-modified epoxy resin/glass fiber composites
Composition Tensile Tensile Flexural Flexural Fracture

strength modulus strength modulus toughness
(MPa) (GPa) (MPa) (GPa) (kJ/m2 )
EP 322 3.8 404 8.5 0.76

3CY-2c 302 5.1 435 16.1 0.77
6CY-2c 310 4.6 450 10.3 0.78
9CY-2c 324 4.6 471 8.4 0.79
3CY-5c 325 3.9 428 8.7 0.88
6CY-5c 332 4.0 437 9.0 0.91
9CY-5c 337 4.1 445 9.2 0.99
15 CY-5c 343 4.3 454 9.5 1.25
epoxy resin) to 428, 437, 445 and 454 MPa, respectively, for their composites. The
enhancement in flexural property may be due to the incorporation of cyanate es-
ter resin leading to the formation of aliphatic oxazolidinone ring which imparts
thermoplastic character to the resin system, enhancing resistance to the bending
stress [11].
3.6.3.3. Effect of Cyanate Loading on Interlaminar Fracture Toughness G1C . The
fracture toughness G1C results for pure epoxy composite and blend composites
were obtained by the compliance method [16] and the values are presented in
Table 5. From Table 5 it is obvious that the fracture toughness increases with in-
crease in cyanate loading which may be due to the presence of oxazolidinone and
cyanurate ring in the blend systems CY-2c and CY-5c [15, 17]. Apart from the
oxazolidinone and cyanate ring, the presence of bulky substituents also plays an
important role in deciding the fracture toughness of the resin blend systems (Ta-
ble 5). It is seen that the 15CY-5c has higher interlaminar fracture toughness than
all other systems. Though oxazolidinone and cyanurate are present in all the other
systems, the higher interlaminar fracture toughness of 15CY-5c may be due to the
presence of flexible Si–O linkage, pendant phenyl group, bulky substituent, all ren-
dering a synergic effect to the toughness.
3.7. Cyanate/BMI Composites (BCDM)
3.7.1. Fibre Volume Fraction
Prior to the study of mechanical properties of composites, the fibre volume fractions
of the cyanate/BMI composites were determined by ASTM D 3171 procedure B.
The fibre volume fractions of the composites were found to be in the range 41 ± 3%.
The mechanical properties of the cured blend composites were evaluated as a func-
tion of blend composition. Since the neat resin does not have any meaningful
strength, bismaleimides have almost exclusively served as their composites and
Table 6.
Mechanical properties of neat cyanate, BMI and blend composites
BT ratio Tensile Tensile Flexural Flexural Fracture Impact

Cy:BMI strength modulus strength modulus toughness strength
(MPa) (GPa) (MPa) (GPa) (J/m2 ) (J/m)
100:0 516 48 622 69 212 1412

3:97 422 32 524 56 104 1121
6:94 442 33 545 58 118 1175
9:91 461 35 554 59 136 1218
0:100 418 29 516 54 79 1039
not as pure resin alone in many structural and thermostructural applications in

aerospace [18]. The tough cyanate esters too find their way as composites (with
glass fibre reinforcement) in a majority of applications. It is generally accepted that
the trend in mechanical performance of the neat resin is not translated to the com-
posites as such, since the failure mechanisms are different in these two cases. Glass
fibre composites of the blends were molded under pressure employing the compo-
sitions given in Table 6.
The tensile strength increases up to 10% for the incorporation of 9% cyanate es-
ter in the BMI. The tensile modulus of the composites increases from 10–20% when
the cyanate content was increased from 3–9%. The tensile properties of the neat and
blend composites are shown in Table 6. With increasing cyanate content the tensile
properties show an increased value in the blends (tensile strength 422–461 MPa,
Tensile modulus 32–35 GPa). This may be attributed to the fact that curing is an
addition reaction leading to the formation of a network structure without the for-
mation of short polymer chains with low molecular weight. Therefore, the cured
resin structure in the final product is more compact. Moreover, many aromatic and
hybrid rings exist in the cured resin; all these features can be responsible for the
good tensile properties of the system [19].
The flexural strength of the composites increased from 1–7% when the cyanate
content increases from 3–9%. The flexural modulus shows 9% increase for 9% in-
corporation of cyanate ester in BMI. The flexural behaviour of neat and blended
composites is shown in Table 6. The trend suggests that it increases with increas-
ing amount of cyanate at least in the studied range (flexural strength 524–554 MPa,
flexural modulus 56–59 GPa). The enhancement in the value of the flexural strength
may be due to the interpenetrating polymer network formation between cyanate es-
ter and BMIs. Since the specimen is subjected to a combination of compressive and
tensile loads under three-point bending for the flexural test, the enhanced flexural
strength indirectly indicates an improvement in the compressive strength imparted
by the maleimide. The flexural strength and modulus for the AroCy B-30/glass fibre
composites were found to be 691 MPa and 36 GPa, respectively [20]. Hence, the
prepared Schiff base cyanate ester shows better flexural properties than the com-
mercial cyanate ester.
Mode-I interlaminar fracture toughness (GIC ) was measured to investigate the
effect of cyanate content on the resistance to crack propagation of the cured lami-
nates. The GIC , values were found to increase with increasing cyanate content. The
fracture toughness increased by 72% when the cyanate content was increased from
0 to 9%. This toughening mechanism can be explained by the formation of –O–
linked triazine IPN which is inherently tough and provides resistance to failure by
fracture (Table 6). The curing reaction creates a high concentration of s-triazine
rings linked by ether oxygen atoms to the benzene rings of bisphenol precursors.
These –O– linkages afford 67% high rotational freedom (freedom to move around,
flexibility) which is the basis for excellent toughness [4, 15]. Moreover, the decrease
in the inherently brittle BMI content and increase in tough cyanate ester content are
likely to enhance the mode-I fracture toughness. Hence, a modified BMI blend is
tougher than the unmodified BMI. For the commercial AroCy B-30 (Ciba-Geigy,
USA) the fracture toughness was observed to be 161.8 J/m2 .
As thermosets are usually brittle and are expected to have low impact strength,
toughness or impact property is the main property to be considered. Unnotched Izod
impact strength was determined to see the modification in the blend composites.
The results of the impact strength tests are shown in Table 6. The impact strength
of the Cy/BMI composites was improved significantly with the addition of cyanate
ester. In the range of 0–9% cyanate loading, the impact strength of the blend in-
creased from 0–17% BMI. The values of impact strength showed a maximum at
9% cyanate loading in Cy/BMI blend composites. This may be explained in terms
of decreased cross-link density of BMI due to its decreased content in the blends
and inherent toughness of cyanate esters (Table 6).
3.7.3. Dynamic Mechanical Analysis
The dynamic mechanical curves of the (temperature dependence of tan δ and stor-
age modulus E ) neat and blend composites are shown in Figs 16 and 17. The data
from DMA studies on neat and blend composites are shown in Table 7. The blends
as well as the neat cyanate and BMI composites show a single tan δ peak, indicating
that these cross-linked blends possess a single homogeneous phase.
The unmodified BMI has the highest storage modulus (E = 25 600 MPa). The
pure cyanate ester shows a storage modulus drop at 212◦ C (E = 12 560 MPa).
As the cyanate content increases the storage modulus and drop temperature (the
temperature at which the storage modulus drops suddenly) of the blends decreases
(Table 7). The relatively low cross-link density with rotational freedom due to –O–
linkages are responsible for the lower storage modulus of the blend composites.
The shape of the tan δ peak may be used as a convenient indicator of the mor-
phological state of the phases within the blend. Although only a single tan δ peak
was observed for these blend systems, it tends to become broader with increasing
cyanate loading. This is indicated by an increase in the full width at half maximum
for blend samples in the tan δ damping peaks. The full width at half maximum refers
Figure 16. DMA curves of neat cyanate ester composites.
Figure 17. DMA curves of neat BMI and cyanate/BMI blend composites.
to the homogeneity of the blend composites [21]. It also indicates the complexity of
the blend structure. Broadening of tan δ peak and increase of full width at half max-
imum suggests an increasing possibility for the existence of more than one network
structures with increasing cyanate incorporation.
The loss tangent is a sensitive indicator of cross-linking. The height and area of
the tan δ peak associated with α-transition could be related to the degree of cure and
cross-link density [22, 23]. As the samples are post-cured at higher temperatures
and for longer times, the blends achieve higher degree of cure and cross-link den-
Table 7.
Dynamic mechanical properties of cyanate/BMI homopoymer and blend composites
BT ratio Tg at tan δ Storage modulus Full width at

Cy:BMI 20 Hz drop temperature half maximum
(◦ C) (◦ C) (◦ C)
100:0 280 0.140 212 (12 560 MPa) 50

3:97 381 0.112 345 (24 750 MPa) 54
6:94 369 0.121 320 (23 450 MPa) 57
9:91 333 0.126 304 (22 870 MPa) 62
0:100 >400 0.063 383 (25 600 MPa) –
sity, which lower the segmental mobilities, leaving fewer relaxing species (mole-
cules resist rotation after cure), and yield the shorter and less obvious α-transition
peaks. The height of the tan δ peak is inversely proportional to the cross-link density
and directly proportional to the flexibility of the molecular chains and hence can be
related to the toughness of the system. As the cyanate content increases in the blend
the tan δ value increases from 0.112 to 0.126. These values are higher than that for
the unmodified BMI (0.063). This indicates that the cross-link density of the blends
decreases and hence the Tg decreases from 381 to 333◦ C. The lower Tg values may
reflect the increased flexibility of the blend, leading to a toughened system than the
unmodified BMI.
3.7.4. SEM Analysis
In order to investigate the fibre/matrix adhesion and morphology of the blend
composites SEM observations were performed on fractured composite samples
(Fig. 18). The samples were analyzed at an acceleration voltage of 20 kV. The frac-
tured surfaces of the systems reveal a homogeneous morphology due to the good
compatibility between the cyanate ester and BMI networks. The SEM micrograph
of the fractured surface of (two component) cyanate modified BMI was almost sim-
ilar to that of (single component) the neat systems. This reveals that there is no
separate phase domain. Further, a smooth fractured surface was observed due to
the brittle behaviour of high BMI content system. The fractured samples did not
show a rough surface (consisting of cracks in one regular direction and not in all
directions).
Both monomers are derived from DDM and their structural similarity (both con-
tain diphenyl methane in the backbone structure) is expected to facilitate good
miscibility and thereby reduce probability of phase separation in the cured net-
work [24].
4. Conclusions
Siloxane, pendant phenyl and Schiff base functionalized cyanate esters were pre-
pared and blended with epoxy and BMI. Siloxane and pendant phenyl cyanate
(a) (b) (c)
(d) (e)
Figure 18. SEM pictures of cyanate and cyanate/BMI composites. (a) Neat BMI, (b) 3% cyanate
loading, (c) 6% cyanate loading, (d) 9% cyanate loading and (e) neat cyanate ester.
esters were blended with epoxy resin and the mechanical properties such as ten-
sile and flexural properties and fracture toughness were determined for both the
blend castings (without glass fibre reinforcement) and composite systems. The re-
sults show that the incorporation of cyanate ester in epoxy increases the mechanical
and thermal properties of the blend systems. Similarly the Schiff base function-
alized cyanate ester was blended with BMI and the mechanical properties of its
composites have been studied. Upon modifying BMI with the prepared Schiff base
functionalized cyanate ester, the mechanical (tensile, flexural, fracture toughness)
and physico-chemical properties (phase compatibility) were improved without sig-
nificant sacrifice in the thermal properties of the BMI. Hence, these systems may
emerge as a new generation thermosetting polymers, encompassing several charac-
teristics required of an ideal high performance matrix.
References
1. C. P. R. Nair, T. Francis, T. M. Vijayan and K. J. Krishnan, J. Appl. Polym. Sci. 74, 2737 (1999).
2. J. T. Gotro, B. K. Appelt and K. I. Papathomas, Polym. Composites 8, 39 (1997).
3. J. M. Barton, I. Hamerton and J. R. Jones, Polym. Int. 29, 145 (1992).
4. T. Fang and D. A. Shimp, Prog. Polym. Sci. 20, 61 (1995).
5. G. Anuradha and M. Sarojadevi, High Perform. Polym. 18, 1003–1018 (2006).
6. M. Florence, G. Loustalot and C. Lartigau, J. Polym. Sci., Part A: Polym. Chem. 35, 3101–3115
(1997).
7. J. Bauer and M. Bauer, Acta Polymerica 39, 548–551 (1998).
8. R. H. Lin and J. H. Hsu, Polym. Int. 50, 1073–1081 (2001).
9. I. Hamerton, B. J. Howlin, P. Klewpatinond and S. Takeda, Polymer 43, 5737–5748 (2002).
10. C. H. Lin, C. N. Hsiao, C. H. Li and C. S. Wang, J. Polym. Sci., Part A: Polym. Chem. 42,
3986–3995 (2004).
11. D. Mathew, C. P. Reghunadhan Nair and K. N. Ninan, J. Appl. Polym. Sci. 74, 1675–1685 (1999).
12. D. J. Liaw, W. C. Ouyang, L. J. Li and M. H. Yang, J. Appl. Polym. Sci. 63, 369–376 (1997).
13. C. Marieta, E. Schulz, L. Irusta, N. Gabilondo, A. Tercjak and I. Mondragon, Composites Sci.
Technol. 65, 2189–2197 (2005).
14. J. Y. Shieh, S. P. Yang, M. F. Wu and C. S. Wang, J. Polym. Sci., Part A: Polym. Chem. 42,
2589–2600 (2004).
15. J. M. Barton, I. Hamerton, J. R. Jones and J. C. Stedman, Polymer 37, 4519–4528 (1996).
16. R. Velmurugan and S. Solaimurugan, Composites Sci. Technol. 67, 61–69 (2007).
17. R. W. Hillermeier and J. C. Seferis, J. Appl. Polym. Sci. 77, 556–567 (2000).
18. H. D. Stenzenberger, in: Polyimides, D. Wilson, H. D. Stenzenberger and P. M. Hergenrother
(Eds). Blackie, Glasgow (1990).
19. A. Gu, Composites Sci. Technol. 66, 1749 (2005).
20. J. N. Suman, J. Kathi and S. Tammishetti, Eur. Polym. J. 41, 2963 (2005).
21. S. Mallarino, J. F. Chailan and J. L. Vernet, Eur. Polym. J. 41, 1804 (2005).
22. I. M. Ward and D. W. Hadley, An Introduction to the Mechanical Properties of Solid Polymers.
Wiley, New York, NY (1993).
23. L. E. Nielsen, Mechanical Properties of Polymers and Composites, p. 174. Marcel Dekker, New
York, NY (1974).
24. C. P. Reghunadhan Nair, T. Francis, T. M. Vijayan and K. Krishnan, J. Appl. Polym. Sci. 74, 2737
(1999).
Rigid-Rod Polybenzimidazoles (PBIs): A Concise Review of
Their Synthesis, Properties, Processing and Applications
Thuy D. Dang a,∗ , Narayanan Venkat b and James E. Mark c

a
Nanostructured and Biological Materials Branch, AFRL/RXBN, Bldg. 654,
2941 Hobson Way, Wright-Patterson Air Force Base, OH 45433, USA
b
Multi-scale Composites and Polymers Division, University of Dayton Research Institute,
300 College Park Drive, Dayton OH 45469, USA
c
Polymer Research Center and Department of Chemistry University of Cincinnati,
1501 B Crosley, Cincinnati, OH 45221-0172, USA
Abstract
Polybenzimidazoles (PBIs) represent a unique class of polymers with exceptional thermo-mechanical prop-
erties and chemical resistance suitable for a variety of applications ranging from flame-resistant fabrics to
aerospace materials. While the commercial applications of the high temperature, high performance thermo-
plastic PBI, both as fibers and as molded materials with the trade name of Celazole® , are well-documented,
this review seeks to emphasize the synthesis, processing, properties and potential applications of the rigid-
rod version known as polybenzobisimidazoles. Various rigid-rod PBI-based molecular motifs for proven
as well as potential enhancement in the axial compressive properties of rigid-rod polymeric fibers are de-
scribed. In many instances, corresponding polybenzobisthiazole (PBZT) and polybenzobisoxazole (PBO)
rigid-rod polymer structures are also discussed, for comparison. Rigid-rod polybenzobisimidazoles with sul-
fonic acid pendants and their novel solubilization and processability as rod-like reinforcement in molecular
composites is also described. Derivatized rigid-rod polybenzobisimidazoles with alkylsulfonate pendants
are also discussed from the viewpoint of their high ionic conductivities for potential utilization in such areas
as polyelectrolytes for rechargeable battery applications.
Keywords
Polybenzobisimidazoles, rigid-rod PBIs, polymeric fibers, axial compressive properties, lateral stabiliza-
tion, derivatized rigid-rod PBIs, molecular composites, rigid-rod PBI polyelectrolytes
1. Introduction and Background
Polybenzimidazoles (PBIs) represent a unique class of aromatic heterocyclic poly-

mers with exceptional thermo-mechanical properties and unusual resistance to
organic solvents; the high performance commercial thermoplastic poly[2,2 -(m-
phenylene)-5,5 -bibenzimidazole] has been utilized in a variety of applications
* To whom correspondence should be addressed. E-mail: thuy.dang@wpafb.af.mil
146 T. D. Dang et al.
ranging from flame-resistant fabrics to aerospace materials. PBI in its molded forms
has been utilized, under the trade name of Celazole® , as sealing elements in high
temperature environments. High performance PBI fibers, stabilized with sulfuric
acid, were designed as non-flammable fabrics for PBI-based protective clothing,
firefighters’ clothing and gloves, astronauts’ flight suits as well as protective lay-
ers for aircraft seats [1]. The tensile properties of PBI fibers and films as well as
their thermal, electrical and dielectric properties have also been cataloged [2]. PBI
has also been investigated as reinforcement for enhancing the thermal stability and
the mechanical properties of thermoplastic microcellular foams, for potential use as
high temperature structural materials in US Air Force applications. Specific inter-
action between PBI and host matrices such as sulfonated polysulfones or aminated
polysulfones was utilized in the generation of highly miscible, compatible blends
from which microcellular foams with complex cell morphologies were fabricated
[3, 4]. In this context, a ternary molecular composite involving PBI, a carboxylated
polysulfone and a siloxane was also investigated from the viewpoint of generat-
ing novel microcellular foam structures [5, 6]. Water-soluble benzimidazole-based
polyelectrolytes containing both aliphatic and aromatic sulfonate pendant groups
by the alkylation of the polyanion have been described [7]. In the last few years,
a number of articles have appeared, describing the utilization of PBI–phosphoric
acid complexes as well as PBI with sulfonic acid groups as Proton Exchange Mem-
branes (PEMs) for fuel cells operations, especially at high temperatures [8–11].
The objective of the current paper is to provide a somewhat comprehensive
overview of the synthesis, processing, properties and applications of the rigid-
rod versions of PBI, referred to as polybenzobisimidazoles or rigid-rod PBIs. The
main difference between the thermoplastic PBI and the rigid-rod PBI lies in the
fact that the latter can form a liquid crystalline phase in solutions when a certain
critical concentration is exceeded, due to the alignment of the rigid-rod polymer
chains; oriented fibers spun from the ordered phase have the potential to exhibit re-
markable axial tensile strength and modulus, similar to the well-known rigid-rod
polymeric fibers such as poly(p-phenylenebenzobisoxazole) (PBO) and poly(p-
phenylenebenzobisthiazole) (PBZT) [12–15]. The chemical structures of all the
three rigid-rod aromatic heterocyclic polymers, generally known as the PBX class
of polymers, are shown below (Fig. 1). The common feature of these structures is
that they are primarily defined by para-catenation which stems from an angle of
nearly 180◦ between the two ends of the polymer repeat unit.
The following discussion sections deal with the influence of the unique features
of the benzimidazole heterocycle on the properties of the rigid-rod PBIs and the
impact of these polymers on some structural applications in which the rigid-rod
PBO and PBZT fibers have had only limited success. A molecular level approach
for enhancing the axial compressive properties of rigid-rod polymeric fibers will
be described with special emphasis on enhanced lateral (inter-chain) interactions
as well as cross-linking strategies in rigid-rod aromatic heterocyclic polymers. The
Rigid-Rod Polybenzimidazoles (PBIs) 147
Figure 1. Structures of rigid-rod aromatic heterocyclic PBX polymers.
last section will be devoted to some derivatized rigid-rod PBIs and their potential
applications.
2. Discussion
2.1. Parent Rigid-Rod PBI Structure
The p-oriented rigid-rod PBI, poly(1,7-dihydrobenzo[1,2-d:4,5-d]diimidazole-2,6-
diyl-1,4-phenylene) was first prepared from melt polycondensation of diphenyl
terephthalate with 1,2,4,5-tetraaminobenzene, resulting in a low molecular weight
polymer of limited solubility [16]. A higher molecular weight rigid-rod PBI with an
intrinsic viscosity of 5 dl/g in methanesulfonic acid was prepared from 1,3-diamino-
4,6-(p-toluenesulfamido)benzene and terephthalic acid from dilute solution poly-
merization in polyphosphoric acid (PPA) at 190◦ C [17]. The synthesis scheme is
shown in Fig. 2.
However, the p-configured rigid-rod PBI could only be synthesized in low con-
centrations in PPA from the isotropic phase, unlike its PBO and PBZT counterparts,
which could be synthesized in the anisotropic phase and spun into oriented fibers
with remarkable tensile properties [18–20]. Though rigid-rod PBI was among the
first synthesized ordered aromatic heterocyclic polymers, it could not be spun
into fibers with high tensile properties, unlike PBO and PBZT. It was assumed
that the problem related to the synthesis of high molecular weight, p-oriented
PBI was one of induced crystallization during the polymerization, causing the
growing polymer to prematurely precipitate out of solution. This is attributed to
the preferential formation of a stable hydrogen-bonded benzimidazole network in
the presence of water generated during the polymerization. In general, the pro-
nounced hydrophilicity of PBI-type polymers makes complete removal of water
extremely difficult even with prolonged heating under vacuum. This was later
corroborated by the unique packing characteristics, based on the crystal struc-
ture analysis, of rigid-rod PBI model compound, 1,7-dihydro-2,6-diphenylbenzo-
[1,2-d:4,5-d]diimidazole tetrahydrate [21]. The chemical structure of the model
compound is shown in Fig. 3. A crystal hydrate structure, bridging a pair of mole-
cules of the diimidazole model compound, was identified.
Figure 2. Synthesis of p-configured (rigid-rod) PBI via dilute solution polymerization.
Figure 3. Structure of rigid-rod PBI model compound, 1,7-dihydro-2,6-diphenylbenzo-[1,2-d:4,5-

d ]diimidazole.
2.2. Synthesis of Rigid-Rod PBIs with Pendants

While the parent PBI structure could not be polymerized in the PPA medium in
high concentrations, attachment of pendants effectively lowered the symmetry of
the PBI molecule, facilitating polymerization in high concentrations, resulting in
high molecular weight polymers. The polymerization reaction for the preparation
of PBI with a methyl pendant (MePBI) is depicted in Fig. 4. This was the first
example of a successful preparation of a p-oriented PBI in the liquid crystalline
phase.
An intrinsic viscosity as high as 14.7 dl/g was obtained, indicating high molec-
ular weight. Fibers could also be spun from the polymer anisotropic phase [22, 23]
in PPA, at 110◦ C, with good draw ratios.
2.3. Rigid-Rod Polymer Fibers–Advantages and Limitations
Rigid-rod polymeric fibers such as PBO, the commercial version of which is known
as Zylon™, exhibit exceptional axial tensile properties; the heat-treated PBO fibers
have a tensile modulus of 270 GPa and a tensile strength of 5.8 GPa. However,
the low axial compressive properties of PBO and other rigid-rod polymer fibers
[14] are considered their Achilles’ heel, severely limiting their ultimate potential
for structural applications requiring high tolerance to compressive loadings, as in
the case of aircraft wings. Zylon fibers have a low compressive strength of 0.3 GPa.
The polymer fiber failure under axial compression is attributed to microfibrillar
buckling as a result of formation and propagation of kink bands [24, 25]. The latter
has been observed to be a typical feature in the compression failure of oriented
polymers [15].
Figure 4. Synthesis of poly(2-methyl-1,4-phenylenebenzobisimidazole) (MePBI) rigid-rod polymer.
Figure 5. Rigid-rod polybenzobisthiazoles (PBZTs) with suitable pendant groups for high temperature
cross-linking.
Approaches to laterally stabilize the polymer chains via cross-linking or

by specific inter-molecular interaction were widely investigated as a means of
enhancing fiber axial compressive strength. Some examples of PBZT poly(p-
phenylenebenzobisthiazole) with suitable pendant groups to promote cross-linking
are shown in Fig. 5.
2.4. PBZTs with Pendant Groups for Cross-Linking
Fibers prepared from PBZT with a bromine pendant group were heat-treated at
450◦ C for 4 h. Cross-linking was promoted by loss of bromine followed by the
coupling of aryl radicals formed. While a two-fold enhancement of fiber compres-
sive strength resulted, it was generally found that the advantages of cross-linking
were offset by the formation of voids during the thermal elimination of the bromine
pendant [14, 26].
Homopolymer and copolymer PBZT compositions incorporating a methyl pen-
dant (MePBZT) were synthesized and spun into fibers for the evaluation of axial
tensile and compressive properties [22]. The fibers were heat-treated at a temper-
ature of 500◦ C for a short time to promote lateral stabilization via cross-linking,
which was indicated by the insolubility of the heat-treated fibers in methane-
sulfonic acid (MSA). While a cross-linked network mainly through the forma-
tion/combination of benzylic radicals was expected to dominate, possible products
from heat-treated MePBZT could also arise from phenyl–benzyl as well as phenyl–
phenyl coupling [27]. These are listed in Fig. 6.
Table 1 describes the mechanical properties of the fibers derived from both
MePBZT homopolymer and random copolymers. Under comparable heat-treatment
Figure 6. Possible structures generated in the MePBZT cross-linked network.
Table 1.
Fiber mechanical properties of MePBZT homopolymer and copolymers after cross-linking∗
Polymer Heat Tensile Tensile Elongation Compressive

(x:y) treatment modulus strength at break strength
(GPa) (GPa) (%) (GPa)∗∗
100:0 500◦ C/30 s 225 1.9 1.7 0.32–0.36

10:90 500◦ C/30 s 193 1.7 1.8 0.45–0.51
50:50 500◦ C/30 s 288 2.0 1.0 0.17–0.32
∗ No longer soluble in methanesulfonic acid (MSA).
∗∗ Measured by tensile recoil method.
conditions, both MePBZT homopolymer and 50/50 random copolymer have fiber
tensile properties comparable to those of PBZT fibers. Their compressive properties
are also somewhat comparable to those of PBZT fibers (0.31 GPa). In the case of a
random copolymer with 10 mol% of the PBZT unit with the methyl pendant group,
a relatively higher fiber axial compressive strength (0.45–0.51 GPa) was measured.
However, the fiber tensile modulus and tensile strength were significantly lower.
Presumably, the effect of cross-link density on the MePBZT copolymer fiber com-
pressive properties is also an important factor. However, no systematic correlations
were found between the methyl content in the copolymer and the fiber compressive
property. The lowest compressive strengths (Table 1) were measured for the 50/50
copolymer in the 0.17–0.32 GPa range for the fibers heat-treated at 500◦ C/30 s in
nitrogen.
2.5. PBZTs with Cross-Linkable Benzocyclobutene Moiety in the Backbone
The thermally-induced cross-linking of a benzocyclobutene moiety attached to

the backbone of a rigid-rod benzobisthiazole polymer (BCB–PBZT) was also ex-
plored from the viewpoint of enhancement of fiber compressive properties [22].
The chemical structure of the polybenzobisthiazole copolymer incorporating the
benzocyclobutene thermoset is shown in Fig. 7.
The thermally driven ring-opening to form the reactive o-xylylene intermedi-
ate and the subsequent cross-linking of p-disubstituted benzocyclobutene moiety
are known to occur around 330◦ C [28]. The mechanical properties of heat-treated
BCB–PBZT homopolymer and copolymer fibers are listed in Table 2. Heat-treated
BCB–PBZT fibers were no longer soluble in MSA, which was indicative of cross-
linking. Comparison of different fiber heat-treatment conditions (Table 2) revealed
that the homopolymer BCB–PBZT heat-treated at 330◦ C for 20 min had a fiber
compressive strength of 0.48 GPa which was nearly twice that of heat-treated PBZT
fibers. However, the same fiber heat-treated at 450◦ C/30 s, showed a poor com-
pressive strength of only 0.10 GPa [22]. While it would be tempting to conclude
that the shorter annealing time even at the higher temperature apparently resulted in
inadequate cross-linking, this was not exactly borne out by the poor axial fiber com-
pressive strengths of 0.12 GPa and 0.11 GPa exhibited by BCB–PBZT copolymer
(25 mol% BCB moiety in the backbone) which was heat-treated at 330◦ C at 10 min
and 20 min, respectively. Thus, no clear correlations seemed to exist in terms of
degree of fiber cross-linking and the fiber compressive strength.
Figure 7. Benzobisthiazole rigid-rod copolymer incorporating benzocyclobutene thermoset.

Table 2.
Fiber mechanical properties of BCB–PBZT rigid-rod polymer systems after thermally induced cross-
linking∗
Polymer Heat Tensile Tensile Elongation Compressive

(x:y) treatment modulus strength at break strength
(GPa) (GPa) (%) (GPa)∗∗
100:0 330◦ C/20 min 121 2.6 1.4 0.48

100:0 450◦ C/30 s 127 2.2 1.1 0.10
25:75 330◦ C/10 min 147 2.4 1.0 0.12
25:75 330◦ C/20 min 126 2.4 1.3 0.11
∗ No longer soluble in MSA.
∗∗ Measured by tensile recoil method.
2.6. Lateral Stabilization via Specific Inter-Chain Interaction —

Poly(2-methyl-1,4-phenylenebenzobisimidazole) (MePBI) Fibers
The mechanical properties of the polymeric fibers, spun from the anisotropic phase
(10 wt% MePBI in polyphosphoric acid) at high spin draw ratios as indicated ear-
lier (vide supra, Section 2.2), were evaluated and the results are shown in Table 3.
The fibers were heat-treated at 390◦ C under nitrogen and the mechanical properties
were compared with those of the as-spun MePBI fibers [23]. From Table 3, it is ev-
ident that there is significant improvement in both fiber tensile strength and tensile
modulus upon heat-treatment. The measured axial fiber compressive strength value
of 0.70–0.72 GPa for both as-spun and heat-treated fibers was quite impressive and
was nearly three-fold higher than the compressive strength of the state-of-the-art
PBO fibers. The relatively high compressive strength of MePBI was attributed to
increased inter-molecular association due to hydrogen bonding between the im-
idazole rings in adjacent MePBI chains. It was envisioned that crystalline defects
existed in the MePBI system, allowing for the association of several adjacent chains
through hydrogen bonding. As in the case of other rigid-rod polymer fibers, both
axial tensile and compressive failures in MePBI were attributed to fibrillar insta-
bility. Scanning electron microscopy revealed the presence of longitudinal cracks,
presumably due to inter-fibrillar rupture [23].
2.7. PBZT and PBO Structures with Dihydoxy Pendants
Polybenzobisthiazole structures with dihydroxy pendants were also generated, us-
ing 2,5-dihydroxyterephthalic acid as the monomer or as a comonomer, along with
terephthalic acid [29]. Figure 8 shows the general synthesis of a dihydroxy-PBZT
copolymer. It was envisaged that the hydroxyl groups would provide inter-chain
Table 3.
Fiber mechanical properties of poly(2-methyl-1,4-phenylenebenzobisimidazole) (MePBI)
Polymer Tensile Tensile Elongation Compressive

fibers modulus strength at break strength
(GPa) (GPa) (%) (GPa)∗
As-spun MePBI 100 1.7 2.5 0.70

Heat-treated MePBI∗∗ 130 2.4 1.9 0.72
∗ From the recoil compressive stress method.
∗∗ Heat-treated at 390◦ C under tension.
Figure 8. Synthesis of rigid-rod polybenzobisthiazoles with dihydroxy pendants.
Table 4.
Fiber∗ mechanical properties of dihydroxy-PBZT homo- and copolymers
Polymer Tensile Tensile Elongation Compressive

(x:y) modulus strength at break strength
(GPa) (GPa) (%) (GPa)
100:0 170 2.06 1.3 0.11–0.41

50:50 97 1.15 1.4 0.42
30:70 110 0.83 0.8 0.11–0.18
∗ Heat-treated at 435◦ C.
hydrogen bonding and lateral stabilization, enhancing the polymer fiber com-
pressive strength. Table 4 shows the mechanical properties of the fibers derived
from dihydroxy-PBZT homo- and copolymers. The measured compressive prop-
Figure 9. Intra-chain hydrogen bonded or pseudo-ladder structure of dihydroxy-PBZT.
Figure 10. Intra-molecular hydrogen bonded structure of dihydroxy-PBO.
erties ranged from poor to modest and, at best, slightly exceeded the compressive
strength of pristine PBZT fibers. The lack of any pronounced influence of the hy-
doxy pendant groups on the axial fiber compressive properties was attributed to
intra-molecular rather than inter-molecular hydrogen bonding, stabilizing what is
referred to as a ‘pseudo-ladder’ structure. The intra-chain hydrogen bonded struc-
ture of dihydroxy-PBZT is shown in Fig. 9.
Spectroscopic evidence was strongly indicative of a completely planar and
ribbon-like structure due to the intra-molecular hydrogen bonding between the
OH group and the N atom of the nearby benzothiazole (–C=N) unit. Dihydroxy-
PBO (DiOHPBO) structure was similarly synthesized and spun into fibers from the
anisotropic phase [30]. However, only a modest compressive strength of 0.31 GPa,
similar to that of PBO itself, was measured for the fibers. In this case too, the
lack of enhancement in the fiber compressive property, relative to PBO, can be
explained as being due to the fact that intra-molecular hydrogen bonding, as op-
posed to inter-molecular hydrogen bonding, tends to dominate in the dihydroxy-
PBO system (Fig. 10), leading to an absence of any lateral stabilization. The
intra-molecular hydrogen bond strength in DiOHPBO was estimated to be quite
significant (−11.7 kcal/mol), as determined from ab initio calculations [31, 32].
2.8. Versions of PBI with Dihydroxy Pendants
The synthesis schemes for dihydroxy-PBI (poly(2,5-dihydroxy-1,4-phenylene-
benzobisimidazole)) and poly(2,5-dihydroxy-1,4-phenylene-pyridobisimidazole),
also known as poly{2,6-diimidazo[4,5-b:4 ,5 -e]pyridynylene-1,4-(2,5-dihydroxy)
phenylene} (PIPD), are shown in Fig. 11. These two rigid-rod polymers were both
spun into fibers from their liquid crystalline phase using a dry-jet wet spinning
method. The measured axial compressive strengths for heat-treated dihydroxy-PBI
and PIPD fibers corresponded to 0.83 GPa and 0.86 GPa, respectively [33]. These
values are nearly three times higher than that of PBO fibers which have an average
compressive strength of 0.30 GPa.
Figure 11. Synthesis of dihydroxy-PBI (poly(2,5-dihydroxy-1,4-phenylenebenzo-bisimidazole)) and

PIPD (poly{2,6-diimidazo[4,5-b:4 ,5 -e]pyridynylene-1,4-(2,5-dihydroxy)phenylene}).
Figure 12. Inter-chain hydrogen bonded structure of PIPD.
The commercial version of PIPD, from Akzo Nobel, is known as ‘M5’ fibers.
The synthesis, characterization, fiber spinning, fiber morphology, crystal structure
and mechanical properties of ‘M5’ have been reported [34–36]. While the as-spun
fiber was described as a 2-D-crystal hydrate, the loss of water molecules upon heat-
treatment of the fiber at high temperatures resulted in a three-dimensional order
with a monoclinic unit cell. The highest axial compressive strength for a polymeric
material to date (1.68 GPa), measured for these heat-treated fibers, is attributed to
maximized lateral interaction due to an inter-molecular, bi-directional, honeycomb-
like network of hydrogen bonds [36]. It was determined that the inter-molecular
hydrogen bond in PIPD occurred between hydroxyl oxygen and the NH of the im-
idazole heterocycle in an adjacent chain even though the hydroxyl group could be
intramolecularly associated with –C=N of the heterocycle as well (Fig. 12) [32, 36].
The tensile strength and modulus for ‘M5’ were also reported to be exceptional, as
in the case of ‘Zylon’ PBO, with measured values of 5.3 GPa and 350 GPa, respec-
tively. The onset of thermal degradation of ‘M5’ was found to occur around 530◦ C.
Among the commercial rigid-rod polymer fibers, ‘M5’ is also known to have a
much higher hydrolytic resistance as well as stability to UV radiation, compared to
Zylon.
Rigid-rod polymer structural motifs discussed so far indicate that lateral stabi-
lization via specific interaction results in the enhancement of axial compressive
strength. As evidenced by the examples of MePBI, dihydroxy-PBI and PIPD for
enhanced fiber compressive strength, the key structural factor is the benzobisimida-
zole or pyridobisimidazole heterocycle in the rigid-rod polymer backbone. While
the proposed inter-molecular hydrogen bonding in MePBI is one of sheet-like
arrangements of hydrogen bonds [23], a bi-directional inter-molecular hydrogen
bonding scheme has been proposed for ‘M5’ chains. The structural similarity be-
tween PIPD and dihydroxy-PBI implies that bi-directional inter-molecular hydro-
gen bonding is also responsible for the high axial fiber compressive properties of
dihydoxy-PBI, measured in our laboratories.
2.9. Aromatic Rigid-Rod Benzobisazole Polymers with Pendant Benzazole Groups
Aromatic rigid-rod heterocyclic polymers were also synthesized with pendant hete-
rocycle groups. Examples of such structures [37] are provided in Fig. 13. A distinct
advantage of the presence of the heterocycle pendant is that it provides an ad-
ditional site for protonation by the acidic medium used in polymerization. This
enhances polymer solubility and provides greater homogeneity to the polymer dope
for the processing of uniform fibers. The fibers were spun from the liquid crys-
talline phase of the polymer in polyphosphoric acid. It was envisaged that the pres-
ence of heterocyclic pendants would result in enhanced lateral interaction between
chains, potentially leading to higher fiber compressive properties. A comparison
of the mechanical properties of heat-treated 2-benzothiazole pendant-PBZT fibers
with the control heat-treated PBZT fibers is shown in Table 5. Examination of the
reported data [37] reveals that the tensile strength and modulus for the heterocycle-
Figure 13. Rigid-rod benzobisazole polymers with pendant heterocyclic groups; (top, left): PBZT
with 2-benzothiazole pendant, (top, right): PBO with 2-benzothiazole pendant, (bottom, left): PBZT
with 2-benzoxazole pendant and (bottom, right): PBI with 2-benzothiazole pendant.
Table 5.
Mechanical properties of 2-benzothiazole-pendant PBZT fibers and control PBZT fibers
Rigid-rod polymer fiber∗ Tensile modulus Tensile strength Compressive strength

(GPa) (GPa) (GPa)
PBZT 380 3.1 0.34

2-Benzothiazole-pendant PBZT 200 2.4 0.83
∗ Heat-treated at 550◦ C under nitrogen.
derivatized PBZT are clearly lower than that of the control PBZT fibers. However,
the axial compressive strength of the 2-benzothiazole-pendant PBZT fibers is more
than double that of control PBZT fibers.
The preparation of a phenylbenzothiazole substituted-terphenyl diacid monomer,
via a multi-step synthesis sequence, has also been reported [38]. This was utilized
in the synthesis of a rigid-rod aromatic heterocyclic polymer containing benzo-
bisthiazole units. The structures of the monomer 4,4 -dicarboxy-2 -phenyl-3 ,6 -
[2-(4-phenylbenzothiazole)]p-terphenyl and the corresponding PBZT polymer are
shown in Fig. 14. The high molecular weight polymer was spun into fibers from
the anisotropic PPA dope and the fiber compressive strength was measured to be
0.86 GPa [33]. Presumably, the disruption of molecular rigidity due to the presence
of the twisted terphenyl system with bulky substituents can facilitate stronger lateral
interactions between the aromatic heterocyclic pendants, leading to an enhanced
fiber compressive strength. These constitute the few examples of aromatic hete-
rocyclic rigid-rod polymer structure motifs that could provide lateral stabilization
to enhance axial compressive strength, without requiring the presence of a benz-
imidazole moiety. There are also a few reports in the patent literature describing
the synthesis of rigid-rod polybenzobisthiazoles and polybenzobisoxazoles with
benzimidazole pendants and with phenyl, biphenyl and terphenyl moieties in the
backbone [39–41]. Such representative rigid-rod polymer structures designed for
improved fiber compressive properties are shown in Fig. 15. However, there are no
reports of fiber spinning of these high molecular weight polymers for the evaluation
of axial compressive properties.
2.9.1. Derivatized Rigid-Rod Poly(p-phenylenebenzobisimidazoles) and Their
Applications
The lack of solubility of aromatic heterocyclic rigid-rod polymers except in acidic
media such as methanesulfonic acid and polyphosphoric acid and as their Lewis
acid complexes in solvents such as nitroalkanes [42–44] tends to limit the broader
potential that might otherwise exist for the applications of these high performance
materials. This problem can be addressed, to some degree, by suitable modifica-
tion/functionalization of rigid-rod polymer structures. One of the earlier reports of
derivatization of a rigid-rod benzobisazole polymer is based on 2-benzothiazole-
pendant PBI (Fig. 13). The reaction involved the ring-opening polymerization of an
Figure 14. Terphenyl diacid monomer with phenylbenzothiazole pendants (top) and the corresponding
rigid-rod benzobisthiazole polymer (bottom).
epoxide, initiated by the anion derived from the imidazole structure, resulting in a
graft copolymer with oligoether side chains [45].
The introduction of sulfonic acid pendant groups in the PBX rigid-rod polymer
backbone was found to dramatically aid in the solubilization of the polymers, as
their trialkylammonium salts, in solvents such as methanol [46]. The new-found
solubility mechanism facilitated the formation of solvent cast rigid-rod polymer
films. Poly(2-sulfo-1,4-phenylenebenzobisimidazole) (SPBI) and poly(2-sulfo-1,4-
phenylenebenzobisthiazole) (SPBT) films were fabricated via their alcohol-soluble
triethylamine and tributylamine salts, respectively (Fig. 16).
2.9.2. Poly(2-sulfo-1,4-phenylenebenzobisimidazole) (SPBI)
The synthesis of SPBI from 4-carboxy-sulfobenzoic anhydride and 1,2,4,5-tetra-
aminobenzene tetrahydrochloride in polyphosphoric acid at elevated temperatures
has been described [47]. The general synthesis route for the preparation of SPBX
Figure 15. Various benzimidazole-pendant rigid-rod benzobisazole polymers (I) poly(2-benzimi-

dazolyl-1,4-phenylenebenzobisthiazole), (II) poly(2-benzimidazolyl-1,4-phenylenebenzobisoxazole),
(III) poly(2,2 -bisbenzimidazolyl-4,4 -biphenyl-benzobisthiazole) and (IV) poly(2 -phenyl-3 -[2-(4-
phenylbenzthiazole)]-6 -[2-(3-phenylbenzimidazole)]-p-terphenylbenzobisthiazole).
Figure 16. Sulfonated rigid-rod polybenzobisimidazole (top, left) and the corresponding polybenzo-
bisthiazole (bottom, left) and their alcohol-soluble trialkylammonium salts (top and bottom, right).
Figure 17. Synthesis route for SPBX rigid-rod polymer systems.
rigid-rod polymer systems [48] is shown in Fig. 17. A comparison of relative ther-
mal stabilities of SPBI and SPBT indicates that the PBI system has much higher
thermal as well as thermo-oxidative stabilities than the PBT system. For instance,
the onset temperature for weight loss for the sulfonic acid pendant degradation in
SPBI is ∼425◦ C, almost 100◦ C higher than that for the sulfonic acid degradation in
SPBT. Clearly, this is attributable to the strong protonation of the basic imidazole
functionality by the sulfonic acid, imparting a sulfonate salt-like thermal stability
to the pendant group in SPBI.
SPBT and SPBI were generated in high molecular weights with intrinsic vis-
cosities of 12 dl/g and 11 dl/g (30◦ C, methanesulfonic acid), respectively and were
spun into fibers at 100◦ C by a continuous dry-jet wet spinning technique from the
liquid crystalline phase of the polymer in polyphosphoric acid, using water as non-
solvent to remove the acid. A study of the fiber structure as well as an evaluation
of the fiber mechanical properties was also reported [48]. The fiber structure was
examined by Wide Angle X-ray diffraction (WAXD). The fiber X-ray patterns for
SPBT and SPBI are shown in Fig. 18. Two equatorial reflections (1 and 2) and three
meridional reflections (3, 4 and 5) are indicated in the figure. The equatorial reflec-
tions, corresponding to side-by-side and face-to-face packing, respectively, of the
polymer chains, have d-spacings of 8.35 Å and 3.25 Å in SPBT. The corresponding
equatorial reflections in SPBI have d-spacings of 7.7 Å and 3.37 Å. The first merid-
ional reflections, corresponding to d-spacings of 11.25 Å and 10.9 Å, respectively,
in SPBT and SPBI, are attributable to the polymer repeat unit distance in these
systems. The other two meridional reflections (indicated by 4 and 5 in the figure)
correspond to d-spacings of 3.65 Å and 2.26 Å in the case of SPBT and 3.81 Å and
2.36 Å in the case of SPBI. In general, the equatorial reflections are sharper in the
case of SPBT and more diffuse in the case of SPBI.
Figure 18. Wide Angle X-ray diffraction patterns of SPBT (left) and SPBI (right) fibers.
Figure 19. Polymerization scheme for the synthesis of rigid-rod polybenzobisazoles with two sulfonic
acid pendants.
Preliminary tensile testing of SPBI fibers indicated that there was considerable
improvement in tensile modulus, from 20 GPa to 40 GPa and the elongational strain
at break decreased from 5% to 4%, upon heat treatment.
Efforts were also made to synthesize rigid-chain benzobisazole polymers with
two sulfonic acid pendants, utilizing a new monomer, 2,5-disulfoterephthalic acid
[49]. The synthesis scheme for the preparation of poly(2,5-disulfo-p-phenylene-
benzobisthiazole) (DSPBT) and the corresponding benzimidazole polymer (DSPBI)
in polyphosphoric acid (10 wt% polymer concentration) is shown in Fig. 19. Both
DSPBT and DSPBI were found to precipitate out of the polymerization medium as
dark brown or dark red solids. The micro-analytical data obtained for both polymers
were found to correspond to the polymer repeat units as their tetrahydrates. They
were found to be partially soluble, at best, in methanesulfonic acid after several
days of stirring. DSPBT was found to be soluble in DMSO only after several days
of warming and exhibited a relatively low intrinsic viscosity of 1.35 dl/g. How-
ever, both DSPBT and DSPBI were readily soluble in 30% aqueous ammonium
hydroxide and as their tributylamine salts when a few drops of tributylamine were
added to their methanolic suspensions, to form deep yellow solutions. The corre-
sponding model compounds were found to be soluble in aqueous dilute NaOH and
1% NH4 OH and as their triethylammonium salts in methanol. The precipitation of
DSPBT and DSPBI in the polyphosphoric acid medium during polymerization and
their lack of solubility in methanesulfonic acid could be rationalized on the basis
of inter-chain protonation of the heterocycle (–C=N) by the two SO3 H pendant
groups. Presumably, the lateral ionic association between the polymer chains was
overcome by the addition of ammonium hydroxide or trialkylamines, thus resulting
in the preferential solubility of DSPBI and DSPBT in bases.
The thermal properties in air, by TGA, showed sulfonic acid pendant degradation
occurring around 378◦ C and 432◦ C for DSPBT and DSPBI, respectively, consis-
tent with earlier observations that the stronger protonation of the basic imidazole
heterocycle imparts a higher thermal stability to the sulfonic acid pendant in poly-
benzimidazole structures. The maximum weight loss in air for the second stage
decomposition, due to the backbone, was found to occur around 535◦ C and 568◦ C,
respectively, for DSPBT and DSPBI.
2.9.3. Rigid-Rod Molecular Composites Based on SPBI
A rigid-rod molecular composite is defined as the molecular-level dispersion of a
rigid-rod polymer such as PBZT or PBO or PBI in a flexible coil polymer matrix
such that the rods act as the reinforcing element. It has been the subject of extensive
research investigation over several years [50–53]. The major drawback of the rigid-
rod molecular composite system was that the molecular dispersion of the rod-like
polymer in a thermoplastic or a thermoset matrix could be achieved only when the
solution processing of the blend was carried out below the critical concentration Ccr
(3–4 wt%) of the rigid-rod polymer. Attempted processing of the blend above Ccr
resulted in phase separation, due to the aggregation of the liquid crystalline domains
of the rigid-rod, providing much less than desired mechanical reinforcement of the
matrix. Another major disadvantage encountered in rigid-rod polymer blends with
thermoplastics such as nylon and polyetheretherketone (PEEK) was the occurrence
of thermally induced phase separation above the glass transition temperature of the
matrix polymers during the consolidation of composites [54].
It was demonstrated that this problem could be alleviated to a great degree
via an acid–base interaction-mediated compatibilization between the rigid-rod and
matrix polymer components. It was shown that optically clear molecular compos-
ite films over a wide range of rod/coil compositions could be successfully cast
from methanol due to the miscibility of the acid component (SPBI) with a base
polymer component such as poly(4-vinylpyridine) (P4VP) or poly(2-vinylpyridine)

(P2VP), even above the Ccr of the rigid-rod polymer in solution [55, 56]. Opti-
cally clear rigid-rod thermoset composite films with SPBI as the acidic component
and phenylethynyl, nadimide and bisbenzoxazine thermosets with amine function-
alities as the basic component were also fabricated, in the mixed solvent system of
methanol and N,N-dimethylacetamide (DMAc) [57]. As an example, the solubiliza-
tion of SPBI in alcohol as its triethylammonium salt, followed by the attachment of
the basic thermoplastic to SPBI, via the process of ionic interchange, is depicted in
Fig. 20. Morphological characterization of the optically clear composite films, both
as cast and annealed/thermally cured, revealed homogeneous microstructures with
no observable phase separated domains, indicating high miscibility that can be as-
cribed to the favorable negative enthalpy of ionic association between the rigid-rod
and matrix components [56]. A dynamic mechanical study of compression molded
rigid-rod thermoplastic composites with low rigid-rod content showed significantly
enhanced thermo-mechanical properties relative to those of the pristine matrix.
Figure 20. Formation of rigid-rod thermoplastic molecular composite films via acid–base interaction.
The rigid-rod molecular composite approach, described above, was also ex-
tended to the preparation of polymer–ceramic hybrids, also known as ‘cer-
amers’ [58]. SPBI and SPBT, as their trialkylammonium salts in methanol, were
reacted with basic silane coupling reagents such as (N,N-diethyl-3-aminopropyl)tri-
methoxysilane and N-[3-(triethoxysilyl)propyl]-4,5-dihydroimidazole in the pres-
ence of tetraethoxysilane to promote the formation of homogeneous organic–
inorganic hybrid gels via sol–gel reaction.
2.9.4. SPBI Derivatized with Alkylsulfonate Groups
Water-soluble versions of the rigid-rod PBI were also reported [59]. For this pur-
pose, SPBI was further derivatized with alkylsulfonate groups by a deprotonation–
alkylation–ion exchange process [47], to evaluate the potential of the resulting
lithium-containing polyelectrolytes for high energy density rechargeable batter-
ies. The sequence of reactions leading to the formation of SPBI derivatized with
propanesulfonate pendants (SPBI-PS(Li+ )) is described in Fig. 21. While many
polyelectrolytes have room temperature ionic conductivities of ∼10−6 S/cm, prac-
tical applications require ionic conductivities in the range of 10−3 –10−4 S/cm.
The initial promise of a rigid-rod polymer electrolyte for such applications was
demonstrated with films fabricated from a propanesulfonate-derivatized PBI with
sodium as the counter-ion [60]. The lithium-doped propanesulfonated rigid-rod
PBI was cast into conductive films from water solution and was found to exhibit
ionic conductivities up to 8.3 × 10−3 S/cm [61]. The ionic conductivity of the
Figure 21. Functionalization of SPBI with ionomeric lithium propanesulfonate pendants (rigid-rod
polyelectrolyte).
films was anisotropic; the ionic conductivity measured transverse to the sample
surface was three to four orders of magnitude smaller than longitudinal conductiv-
ity. Ion-conducting films were also fabricated from a rigid-rod polyelectrolyte with
an articulated backbone formed by the incorporation of a m-phenylene copoly-
mer unit [62]. The chemical structure of the copolymer is shown in Fig. 22. The
solvent cast film of the polyelectrolyte showed a three-dimensionally isotropic con-
ductivity of 2.8 × 10−3 S/cm at room temperature. The above mentioned rod-like
polyelectrolytes display unusually high ionic conductivities and could be poten-
tially evaluated for a variety of rechargeable energy applications.
Ionic exchange between electrostatically stabilized poly(methyl methacrylate)
(PMMA) latex and the rigid-rod polyelectrolyte SPBI-PS(Li+ ) was also demon-
strated to be a flexible approach for the formation of polymer-based nano and
molecular composites. It was reported that with <1% incorporation of the rigid-
rod polymer electrolyte, significant increases in both the heat distortion temperature
and the rubbery modulus of the composite could be achieved [63].
The rigid-rod polyelectrolyte described above was also incorporated into a ther-
moplastic polymer matrix such as poly(vinyl pyrrolidone) (PVPy) [64]. The mole-
cular composite films were cast from deionized, distilled water. The miscibility in
this case was promoted by ion–dipole interaction [65, 66] between the rigid-rod
and the flexible coil components. This is depicted in Fig. 23. An examination of the
morphology of the various composite films revealed that at the lower rod content,
Figure 22. Structure of the polyelectrolyte derived from an articulated rigid-rod polymer backbone.
Figure 23. Rigid-rod/flexible coil polymer miscibility via ion–dipole interaction: SPBI-PS(Li+ ) and
PVPy system.
Table 6.
Glass transition temperatures (Tg s) as a function of dispersed rod content for ion–dipole (SPBI-
PS(Li+ )-PVPy) and acid–base (SPBI-P4VP) molecular composite systems
SPBI-PS(Li+ )/PVPy blend SPBI/P4VP blend

Rod/coil (w/w) Measured Tg ∗ (◦ C) Rod/coil (w/w) Measured Tg ∗∗ (◦ C)
0/100 175 0/100 153

5/95 188 5/95 176
10/90 215 10/90 210
∗ From tan δ measurements by DMA on solvent cast films.
∗∗ From tan δ measurements by DMA on compression molded samples.
a very homogeneous micro-structure is seen while at rod concentrations >40 wt%,

a plate-like morphology is observed. It is also noted, by visual inspection, that films
with rod dispersions of 40% and above in the matrix are opaque. This is in con-
trast to the SPBI/poly(4-vinylpyridine) (P4VP) composites in which optically clear
films resulted over the entire range of rod/coil compositions (5/95–80/20 rod/coil
w/w) due to the stronger acid–base association between the rod and the coil. The
glass transition temperatures (Tg s), as a function of dispersed rigid-rod content, in
the two blend systems, are shown in Table 6. The occurrence of significant en-
hancement in the Tg of the matrix upon incorporation of relatively small amounts
of the rigid-rod polymer is strongly indicative of miscibility brought about by the
synergy of rod-coil specific interaction. The findings related to above mentioned
molecular composites have implications for their utilization in structural as well as
in rechargeable energy applications such as batteries and fuel cells.
3. Summary
The synthesis, characterization, processing, properties and potential applications
of rigid-rod polybenzobisimidazoles have been reviewed. Rigid-rod PBI polymer
structures containing alkyl, hydroxyl as well as aromatic heterocyclic pendants
have been described; oriented fibers, spun from the liquid crystalline phase of the
polymers in acidic media, have been characterized from the viewpoint of their ax-
ial mechanical properties. Rigid-rod PBX homo as well as copolymers containing
cross-linkable moieties have also been reviewed as polymer systems with lateral
bonding for enhanced fiber axial compressive properties. Lateral stabilization via
inter-chain hydrogen bonding in rigid-rod PBIs has provided a class of materials
with increased fiber axial compressive properties, enhancing the prospects for the
utilization of rigid-rod polymer fibers in structural applications that also require
tolerance to compressive loading. While benzimidazole has remained the dominant
structural motif that has profoundly influenced fiber compressive strength, some
examples of rigid-rod polymer fibers having higher compressive properties, despite
lacking the presence of a benzimidazole moiety in their chemical structure, have

also been described.
While the solubility of rigid-rod polymers has been generally limited to strong
acidic solvents and as their Lewis acid complexes, an alcohol-soluble version of
rigid-rod PBI has also been described in this review, by the introduction of a
sulfonic acid pendant in the polymer structure. Sulfonated PBI, solubilized as its
trialkylammonium salt in alcohol, was blended with basic thermoplastics and ther-
mosets to fabricate rigid-rod molecular composites in which complete miscibility
of the components was achieved, via acid–base interaction, even above the criti-
cal solution concentration (Ccr ) of the rigid-rod polymer. Further derivatization of
sulfonated PBI via an imidazole deprotonation–alkylation sequence resulted in the
formation of water-soluble propanesulfonate salts. These rod-like polyelectrolytes
were found to exhibit unusually high ionic conductivities and thus are suitable for
a variety of rechargeable energy applications such as batteries, fuel cells and super-
capacitors. Formation of rigid-rod molecular composites, via ion–dipole interac-
tions, has also been demonstrated by utilizing rigid-rod polyelectrolytes and polar
thermoplastic matrix polymers. The ionic conductivity of the composites can be
potentially tailored by varying the composition of the polyelectrolyte membranes.
References
1. J. C. Salamone (Ed.), Polymeric Materials Encyclopedia, pp. 5620–5638. CRC Press, Boca Raton,
FL (1996).
2. W. L. Welsh, in: Polymer Data Handbook, J. E. Mark (Ed.), pp. 288–290. Oxford University Press
(1999).
3. H. Sun, N. Venkatasubramanian, M. D. Houtz, J. E. Mark, S. C. Tan, F. E. Arnold and C. Y.-C. Lee,
Colloid Polym. Sci. 282, 502–510 (2004).
4. H. Sun, J. E. Mark, S. C. Tan, N. Venkatasubramanian, M. L. Houtz, F. E. Arnold and C. Y.-C. Lee,
Polymer 46, 6623–6632 (2005).
5. H. Sun, J. E. Mark, N. Venkatasubramanian, M. D. Houtz, S. C. Tan, F. E. Arnold and C. Y.-C. Lee,
Macromol. Sci. — Pure Appl. Chem. A41, 981–1000 (2004).
6. H. Sun, N. Venkatasubramanian, J. E. Mark and F. E. Arnold, Polym. Mater. Sci. Eng. 88, 398–399
(2003).
7. M. B. Gieselman and J. R. Reynolds, Macromolecules 25, 4832–4834 (1992).
8. Q. Li, C. Pan, J. O. Jensen, P. Noye and N. J. Bjerrum, Chem. Mater. 19, 350–352 (2000).
9. J. Wainright, J.-T. Wang, D. Weng, R. F. Savinell and M. Litt, J. Electrochem. Soc. 142, L121
(1995).
10. J.-T. Wang, R. F. Savinell, J. Wainright, M. Litt and H. Yu, Electrochimica Acta 41, 193 (1996).
11. A. Schechter and R. F. Savinell, Solid State Ionics 181, 147 (2002).
12. J. F. Wolfe, B. H. Loo and F. E. Arnold, Macromolecules 14, 915 (1981).
13. W. W. Adams, R. K. Eby and D. E. Mclemore (Eds), Mater. Res. Soc. Symp. Proc., Vol. 134,
Materials Research Society, Pittsburgh, Pennsylvania (1989).
14. Y.-H. So, Prog. Polym. Sci. 25, 137–157 (2000).
15. X.-D. Hu, S. E. Jenkins, B. G. Min, M. B. Polk and S. Kumar, Macromol. Mater. Eng. 288,
823–843 (2003).
16. H. Vogel and C. S. Marvel, J. Polym. Sci. 50, 511 (1961).

17. R. F. Kovar and F. E. Arnold, J. Polym. Sci., Polym. Chem. Ed. 14, 2807 (1976).
18. H. Jiang, R. Eby, W. W. Adams and P. G. Lenhert, Mater. Res. Soc. Symp. Proc. 134, 341–350
(1989).
19. K. Tashiro, J. Yoshino, T. Kitagawa, H. Murase and K. Yabuki, Macromolecules 31, 5430–5440
(1998).
20. T. Kitagawa, H. Murase and K. Yabuki, J. Polym. Sci. Part B: Polym. Phys. 36, 39–48 (1998).
21. D. W. Tomlin, A. V. Fratini, M. Hunsaker and W. W. Adams, Polymer 41, 9003–9010 (2000).
22. T. D. Dang, C. S. Wang, W. E. Click, H. H. Chuah, T. T. Tsai, D. M. Husband and F. E. Arnold,
Polymer 38, 621–629 (1997).
23. S. Jenkins, K. I. Jacob, M. B. Polk, S. Kumar, T. D. Dang and F. E. Arnold, Macromolecules 33,
8731–8738 (2000).
24. S. B. Warner, Fiber Science. Prentice-Hall (1994).
25. W. Huh, S. Kumar and W. W. Adams, Polym. Eng. Sci. 43, 684–692 (2003).
26. W. J. Sweeney, J. Polym. Sci., Polym. Chem. Ed. 30, 1111–1122 (1992).
27. V. R. Mehta and S. Kumar, J. Appl. Polym. Sci. 73, 305–314 (1999).
28. K. A. Walker, L. J. Makoski, G. A. Deeter, G. E. Spilman, D. C. Martin and J. S. Moore, Polymer
35, 5012–5017 (1994).
29. L.-S. Tan, F. E. Arnold, T. D. Dang, H. H. Chuah and K. H. Wei, Polymer 35, 3091–3131 (1994).
30. T. D. Dang, H. H. Chuah, L.-S. Tan and F. E. Arnold, US Patent 5,106,940 (1992).
31. S. Trohalaki, A. T. Yeates and D. S. Dudis, in: Modeling the Hydrogen Bond, D. A. Smith (Ed.),
ACS Symp. Series No. 569, p. 269. American Chemical Society, Washington, DC (1994).
32. S. Jenkins, K. I. Jacob and S. Kumar, J. Polym. Sci., Polym. Phys. 38, 3053–3061 (2000).
33. T. D. Dang, N. Venkat and J. E. Mark, 2nd International Symposium on Advanced Materials and
Polymers for Aerospace and Defense Applications (SAMPADA-2008), Book of Abstracts, Paper
#S2-OR-6, December 4–8, Pune, India (2008).
34. D. J. Sikkema, Polymer 39, 5981–5986 (1998).
35. M. Lammers, E. A. Klop, M. G. Northolt and D. J. Sikkema, Polymer 39, 5999–6005 (1998).
36. E. A. Klop and M. Lammers, Polymer 39, 5987–5998 (1998).
37. T.-T. Tsai and F. E. Arnold, US Patent 4,892,921 (1990).
38. F. E. Arnold and J. L. Burkett, US Patent 4,892,953 (1990).
39. F. E. Arnold and L.-S. Tan, US Patent 5,081,256 (1992).
40. F. E. Arnold, L.-S. Tan and T. D. Dang, US Patent 5,128,443 (1992).
41. F. E. Arnold, L.-S. Tan and T. D. Dang, US Patent 5,140,092 (1992).
42. S. A. Jenekhe and P. O. Johnson, Macromolecules 23, 4419 (1990).
43. M. F. Roberts and S. A. Jenekhe, Chem. Mater. 5, 1744 (1993).
44. J. W. Connolly, D. S. Dudis, S. Kumar, L. T. Gelbaum and N. Venkatasubramanian, Chem. Mater.
8, 54 (1996).
45. R. Evers, T. D. Dang and D. R. Moore, J. Polym. Sci., Part A: Polym. Chem. 29, 121 (1991).
46. T. D. Dang, J. P. Chen and F. E. Arnold, US Patent 5,508,376 (1996).
47. T. D. Dang and F. E. Arnold, Mater. Res. Soc. Symp. Proc. 305, 49 (1993).
48. T. D. Dang, H. Koerner, M. J. Dalton, A. M. Iacobucci, N. Venkatasubramanian and F. E. Arnold,
Polymer Preprints (American Chemical Society) 44, 927–928 (2003).
49. N. Venkatasubramanian, T. D. Dang, T. A. Vance and F. E. Arnold, Polym. Mater. Sci. Eng. 84,
601–602 (2001).
50. T. E. Helminiak, C. L. Benner, G. E. Husman and F. E. Arnold, US Patent 4,207,407 (1980).
51. W. F. Hwang, D. R. Wiff, C. L. Benner and T. E. Helminiak, J. Macromol. Sci. Phys. 22, 231
(1983).
52. W. F. Hwang, D. R. Wiff and C. Verchoore, Polym. Eng. Sci. 23, 790 (1983).
53. C. S. Wang, I. J. Goldfarb and T. E. Helminiak, Polymer 29, 825 (1988).
54. F. E. Arnold, Jr and F. E. Arnold, in: Advances in Polymer Science: High Performance Polymers,
Vol. 117, pp. 257–295. Springer, Berlin (1994).
55. F. E. Arnold, N. Venkatasubramanian, T. D. Dang and D. R. Dean, US Patent 6,025,439 (2000).
56. N. Venkatasubramanian, D. R. Dean, T. D. Dang, G. E. Price and F. E. Arnold, Polymer 41,
3213–3226 (2000).
57. F. E. Arnold, N. Venkatasubramanian, T. D. Dang and D. R. Dean, US Patent 5,998,550 (1999).
58. T. D. Dang, J. P. Chen and F. E. Arnold, in: Hybrid Organic–Inorganic Composites, J. E. Mark,
C. Y.-C. Lee and P. A. Bianconi (Eds), ACS Symposium Series 585, pp. 280–290. American
Chemical Society, Washington, DC (1995).
59. T. D. Dang and F. E. Arnold, US Patent 5,312,876 (1994).
60. T. D. Dang, S. J. Bai, D. P. Heberer, F. E. Arnold and R. J. Spry, J. Polym. Sci., Polym. Phys. Ed.
B31, 1941 (1993).
61. R. J. Spry, M. D. Alexander, Jr, S. J. Bai, T. D. Dang, G. E. Price, D. R. Dean, B. Kumar,
J. S. Solomon and F. E. Arnold J. Polym. Sci., Polym. Phys. Ed. 35, 2925–2933 (1997).
62. S. J. Bai, J.-P. Sun, J.-W. Huang and T. D. Dang, Electrochem. Commun. 9, 517–521 (2007).
63. G. B. Rossi, G. Beaucage, T. D. Dang and R. A. Vaia, Nano Lett. 2, 319–323 (2002).
64. D. R. Dean, N. Venkatasubramanian, T. D. Dang, G. E. Price and F. E. Arnold, Polymer Preprints
(American Chemical Society) 41(2), 1354–1355 (2000).
65. G. Parker and M. Hara, Polymer 38, 2773 (1997).
66. Y. Feng, A. Schmidt and R. A. Weiss, Macromolecules 29, 3909 (1996).
A Novel Negative-Working Photodefinable Poly(benzoxazole)
Using Acid Catalyzed Cross-linking
Tomonori Minegishi a,∗ and Noriyuki Yamazaki b

a
New Applied Materials R&D Center, Hitachi Chemical Co., Ltd. 13-1 Higashi-cho 4-chome,
Hitachi-shi, Ibaraki 317-8555, Japan
b
Yamazaki Works, Hitachi Chemical Co., Ltd. 13-1 Higashi-cho 4-chome, Hitachi-shi,
Ibaraki 317-8555, Japan
Abstract
A novel thermally stable negative-working photodefinable material which employs a chemical amplifica-
tion system based on acid catalyzed cross-linking has been proposed. We chose methylol type cross-linkers
for the system since their photopolymerization proceeds efficiently even under mild acidic conditions. Fo-
cusing on the dissolution behavior study, we optimized the composition, which afforded high dissolution
contrast and fast dissolution rate for thick films fabrication. Photolithographic evaluations of the material
revealed that the sensitivity was less than 100 mJ/cm2 for a 7-µm-thick film, which is better than that of
reported positive systems. Subsequent detailed evaluations showed a high performance such as a fine linear
correlation between the mask size and the fabricated pattern and critical dimension uniformity (standard
deviation = 0.15). In addition, we demonstrated thicker film fabrication (>10 µm). Consequently, appro-
priate pattern profiles were obtained. Finally, this newly designed material showed sufficiently high film
properties, which are comparable to those of a conventional non-photodefinable polyimide.
Keywords
Poly(benzoxazole), chemical amplification, negative-tone, photopolymer, cross-linking
1. Introduction
Poly(benzoxazole) (PBO) is one of the thermally stable polymers, which is widely
used in electronics applications such as insulating, buffer coating, interlayer di-
electrics and so on. Nowadays TMAH (tetramethylammonium hydroxide) process-
able photodefinable PBOs are preferred in this field in order to reduce process
cost and to address environmental issues. For these aqueous developable materials,
the photodecomposition of naphthoquinonediazide (NQD) into indene carboxylic
acid is widely employed as a photochemical system [1–3] based on the origi-
nal design of novolac-based photoresist materials. Several types of photodefinable
* To whom correspondence should be addressed. Tel.: +81-294-23-8975; Fax: +81-294-22-3793; e-mail:

t-minegishi@hitachi-chem.co.jp
172 T. Minegishi and N. Yamazaki
Figure 1. A general chemical structure of the PBO precursor, poly(hydroxyamide).
PBOs other than the NQD type have been proposed [4–7]. We have also reported
a positive-working photodefinable PBO using chemical amplification system [8].
This proposed material including a polybenzoxazole precursor, photoacid genera-
tor (PAG) and acid-labile dissolution-inhibiting compound showed remarkably high
photolithographic properties on a 7-µm-thick film, which is the normal thickness for
buffer coatings. By the way, recent progress in semiconductor technologies with di-
versified packaging styles is placing new requirements on photodefinable PBOs and
related materials. For instance, some applications require the materials to have more
than 10-µm-thick films. Such a thick film application is one of the challenging tar-
gets since conventional positive systems might have a limitation in film thickness
in spite of using a highly sensitive chemical amplification. From this viewpoint,
we decided to design negative-working materials as well as positive ones. As for
a negative-working photosystem, use of a radical initiated polymerization is the
most popular and widely applied to numerous photodefinable polymers. However,
the PBO precursor, poly(hydroxyamide), has two phenolic OH groups in the repeat
unit, which are supposed to terminate radical reactions (Fig. 1). On the other hand,
recently several photosensitive PBOs as well as polyimides using cationic polymer-
ization have been reported [9–12]. These photosystems surely afford appropriate
photoimages, however, many of them have not shown their practicality such as thick
film (more than 5 µm) fabrication and detailed photolithographic properties.
In this study, we have examined several methylols as cross-linkers to design
a negative-working chemically amplified PBO. Then, we have evaluated its pho-
tolithographic and film properties. Moreover, detailed evaluations were conducted
to confirm its practicality.
2. Experimental
2.1. Materials
Poly(hydroxyamide) (PHA) was prepared from an aromatic dicarboxylic acid, aro-
matic hydroxydiamine, and an endcapping reagent by acid chlorination method.
PAGs, cross-linkers and solvents were used as received.
2.2. Evaluation of Model Composition
Each polymer solution was prepared by dissolving 91 wt% of PHA, 4.5 wt% of
a cross-linker and 4.5 wt% of a PAG in a mixture of γ -butyrolactone and propy-
lene glycol mono methyl ether acetate, and then filtered through a filter with 3 µm
pore-size. The polymer solution was spin-coated on a silicon wafer, which was
A Novel Negative-Working Photodefinable PBO Using Acid Catalyzed Cross-linking 173
then prebaked on a hot plate at 120◦ C for 3 min. The sample films were exposed
to 365 nm using a proximity mask aligner UX-1000SM-XJ01 (Ushio) through an
interference filter followed by post-exposure baking (PEB) conducted at 130◦ C for
3 min. 2.38 wt% TMAH was used as a developer. Dissolution rate was calculated
based on the change in film thickness before and after development as measured by
Dektak3 ST surface profiler (Veeco Instruments Inc.).
2.3. Evaluation of Photolithographic Properties
The photopolymer solution was prepared by dissolving the PHA, the cross-linker,
the photoacid generator and other additives such as the adhesion promoter in a
solvent, and then filtered through a filter with 3 µm pore-size. The photopoly-
mer solution was spin-coated on a silicon wafer, prebaked typically at 120◦ C for
3 min on a hot plate, and exposed to the radiation of an i-line stepper FPA-3000iW
(Canon) followed by PEB conducted at 135◦ C for 3 min. The puddle method with
a 2.38 wt% aqueous TMAH as the developer was applied for the development.
2.4. Evaluation of Film Properties
Cyclization of the spin-coated PHA to the corresponding PBO was accomplished
through curing at 350◦ C for 1 h under nitrogen atmosphere. Thermomechani-
cal analysis (TMA) and thermogravimetric analysis (TGA) were conducted using
TMA/SS6000 and TG/DTA6300 (both from Seiko Instruments Inc.), respectively.
Tensile properties were measured using Autograph AGS-100NH (Shimadzu).

3.1. Cross-linkers
Epoxide, oxetane and vinyl ether derivatives are popular cross-linkers used for ad-
hesives, paints, coating materials and so on. A large number of photopolymers
using their cationic polymerization have also been reported. However, such pho-
topolymerization usually requires a strong acid catalyst such as HSbF6 or HAsF6
for efficient reactions [13]. Since our target material was intended for use in semi-
conductor processes, it seemed desirable to avoid such a strong acid. Among the
reported cross-linkers, methylols attracted our attention mostly due to their high re-
activity in photodefinable polyimides or PBOs in stead of using weaker photoacids
such as aromatic sulfonates [9–11]. From this viewpoint, we decided to evaluate
several methylol-like cross-linkers for our system. Table 1 classifies them into 4
groups, ‘groups A–D’ based on the following two criteria: (1) with or without phe-
nol group and (2) reactive group is benzyl alcohol (PhCH2 OH or PhC(CH3 )2 OH)
or benzyl ether (PhCH2 OCH3 ). Our target material should have high sensitivity,
fine resolution and adjust to reasonably short-time development as a film of 10 µm
thickness or more. Therefore, we paid attention to both dissolution rate (DR) of the
unexposed area and dissolution contrast, which were defined by the following equa-
Table 1.
Classification of the cross-linkers tested
Group Phenol or non-phenol Reactive unit
A Phenol Alcohol
B Phenol Ether
C Non-phenol Alcohol
D Non-phenol Ether
tion. The composition affording both high dissolution contrast and fast DRunexp is
desirable.
DRunexp
Dissolution contrast = , (1)
DRexp
where DRunexp and DRexp are the dissolution rates of the unexposed and the ex-
posed areas, respectively. These two dissolution rates of the model formulation
containing the PHA, a cross-linker and a PAG were measured. Regarding the PAG,
we chose a 365-nm sensitive one since a typical process of the target application
uses 365 nm or longer wavelength as a light source. Figure 2 summarizes the dis-
solution contrast and DRunexp of each formulation tested. In terms of DRunexp ,
cross-linkers of groups A and B gave sufficiently fast rates. It is obvious that the
phenol group plays an important role. At first, we expected that cross-linkers bear-
ing benzyl alcohol moiety as a reactive group would afford high dissolution contrast
than benzyl ether in accordance with their reactivity. However, the obtained disso-
lution contrasts seem to have little dependence on their reactive units. This result
implies that the balance of their reactivity and solubility controls the dissolution
contrast. While we were able to find a promising cross-linker in the desirable re-
Figure 2. Dissolution contrast versus dissolution rate of unexposed area for photopolymer composi-
tions using various cross-linkers. (For groups A–D, see Table 1.)
gion, we did not reveal the key factors to controlling their reactivity. Therefore, we
needed further analyses based on their structures, molecular size, the number of
reactive units and so on.
3.2. Photolithography
On the basis of the above studies, we selected an appropriate cross-linker and

a PAG. Photolithographic properties of the best composition were evaluated on
about 10-µm-thick prebaked film. Typical process conditions and film thickness
change through the process are summarized in Table 2. According to the sensitiv-
ity curve demonstrated in Fig. 3, the minimum dose required to reach 95% of the
film retention was 50 mJ/cm2 . On the other hand, the cross-sectional images imply
insufficient photopolymerization at the bottom of the film in the case of 50 mJ/cm2
as shown in Fig. 4. From this viewpoint, it is reasonable to say that an appropriate
exposure dose should be 70–100 mJ/cm2 . The sensitivity of the material is higher
than our previously proposed positive system [8]. Moreover, this new material suc-
cessfully resolved 4-µm-wide lines.
Further study was conducted to ascertain the ability to make precise patterns
with this material. Figure 5 shows a good linear correlation between the mask size
and the obtained hole size. Table 3 shows an excellent critical dimension (CD)
uniformity of 7-µm square holes fabricated on 150-mm-diameter silicon wafers.
These properties are comparable to positive-working materials currently used in the
semiconductor manufacture. Figure 6 shows the dependence of fabricated hole size
on exposure dose. This material showed insignificant dependence on exposure dose
over a wide range, i.e., from 30 to 160 mJ/cm2 . The obtained hole sizes showed
only a slight deviation from the corresponding mask size.
Table 2.
Typical process conditions and film thickness during various processes
Process Conditions Film thickness (µm)
Coating 1000 rpm, 10 s 9.9

3000 rpm, 30 s
Prebaking 120◦ C, 3 min
Exposure i-line stepper 9.8
PEBa 130◦ C, 3 min
Developmentb 68 s × 2 9.6
Curing 350◦ C, 1 h, N2 7.7
a Post exposure bake.
b 2.38 wt% TMAH was used as a developer.
Figure 3. Sensitivity curve for the photodefinable chemically amplified PBO for 10-µm-thick pre-
baked film.
Figure 4. Cross-sectional SEM micrographs of 4-µm line on the cured film. The exposure doses were
50 mJ/cm2 , 70 mJ/cm2 and 100 mJ/cm2 for (a), (b) and (c), respectively.
Figure 5. Fabricated hole size versus mask size for the photodefinable chemically amplified PBO.
Table 3.
Critical dimension uniformity of 7-µm square holes for the photodefin-
able chemically amplified PBOa
Parameter Results after cure (µm)
Film thickness 7.2

Average hole size 6.76
Max 7.06
Min 6.49
Range 0.57
Standard deviation 0.15
a These data were collected from 17 patterns fabricated on 150-mm-
diameter silicon wafers.
Figure 6. Dependence of fabricated hole size on exposure dose.

Figure 7. Cross-sectional SEM micrograph of 20-µm line on 13-µm-thick cured film.
Table 4.
350◦ C cured film property for the proposed photodefinable chemically amplified PBO
Parameter Units Proposed materiala PIX-3400b
Tg ◦C 306 298
Td c ◦C 507 515
CTE 10−6 /K 56 43
Tensile strength MPa 113 112
Elongation % 43 60
Elastic modulus GPa 1.8 2.6
a Determined by TMA.
b Trade name of HD MicroSystems.
c 5% decomposition temperature.
By optimizing the formulation and process conditions, we have successfully

confirmed the processability of films thicker than 10 µm. Figure 7 demonstrates
a cross-sectional image of 20-µm-wide line fabricated on a 13-µm-thick film.
3.3. Film Properties
Since this proposed material will remain in semiconductor devices, mechanical
properties and thermal stability of the cured film are also important. Both prop-
erties were evaluated on 350◦ C cured film as summarized in Table 4. The typical
non-photodefinable polyimide PIX-3400, which is widely used in semiconductor
manufacture, was chosen as a reference. It is concluded that the proposed material
has sufficiently high properties, which are comparable to those of PIX-3400.
4. Conclusion
A novel thermally stable negative-working photodefinable material which employs
a chemical amplification system based on acid catalyzed cross-linking has been
proposed. We chose methylol type cross-linkers for the system since their pho-
topolymerization efficiently proceeds even under mild acidic conditions. Focusing
on the dissolution behavior study, we optimized the composition, which afforded

high dissolution contrast which can be applied to thick film process. Photolitho-
graphic evaluation revealed that its sensitivity was less than 100 mJ/cm2 for a
7-µm-thick film, which is better than that of reported positive systems. In addi-
tion to this, the material demonstrated its processability for thicker film fabrication,
more than 10 µm. Finally, the newly designed material showed sufficiently high film
properties comparable to those of the conventional non-photodefinable polyimide.
In conclusion, this material has adequate photolithographic and film properties for
practical use.
References
1. D. N. Khanna and W. H. Mueller, Polym. Eng. Sci. 29, 954–959 (1989).
2. R. Rubner, Adv. Mater. 2, 452–457 (1990).
3. R. Rubner, J. Photopolym. Sci. Technol. 17, 685–691 (2004).
4. S. L.-C. Hsu and W.-C. Chen, Polymer 43, 6743–6750 (2002).
5. K. Ebara, Y. Shibasaki and M. Ueda, J. Photopolym. Sci. Technol. 16, 287–292 (2003).
6. C. S. Hong, M. Jikei, R. Kikuchi and M. Kakimoto, Macromolecules 36, 3174–3179 (2003).
7. S. L.-C. Hsu, W.-C. Chen and P.-I. Lee, Polym. Bull. 50, 295–302 (2003).
8. T. Minegishi, H. Takusari and K. Kato, J. Photopolym. Sci. Technol. 17, 247–252 (2004).
9. M. Ueda and T. Nakayama, Macromolecules 29, 6427–6431 (1996).
10. K. Fukukawa and M. Ueda, Polym. J. 38, 405–418 (2006).
11. C. S. Hong, M. Jikei and M. Kakimoto, Polym. J. 35, 586–591 (2003).
12. H. S. Yu, T. Yamashita and K. Horie, Macromolecules 29, 1144–1150 (1996).
13. H. Ito, Adv. Polym. Sci. 172, 37–245 (2005).
Low Temperature Curable, Positive-Tone, Photo-Definable
Poly(benzoxazole)
Masayuki Ohe a,∗ , Kohji Katoh a , Tomoko Kawamura b , Hiroshi Matsutani b and
Takashi Hattori b
a
Hitachi Chemical DuPont MicroSystems Ltd., 13-1 Higashi-cho 4-chome, Hitachi-shi,
b
Hitachi Chemical Co., Ltd., 13-1 Higashi-cho 4-chome, Hitachi-shi,
Abstract
Due to their excellent thermal, mechanical and electrical properties, polyimides have been widely used as
stress buffer coatings to improve the reliability of semiconductor devices. In recent years, there has been
a trend away from non-photodefinable polyimides (non-PDPIs) towards photodefinable polyimides (PDPIs)
because of the ability of a PDPI to simplify the stress buffer layer fabrication process, resulting in process
cost reduction. Furthermore, aqueous-developable, positive-tone PDPIs are now becoming the mainstream
technology due to demands such as higher resolution, further reduction in process cost, and the need for
more environmentally friendly chemicals. These systems typically utilize polybenzoxazole (PBO) material
that has a thermal resistance equivalent to polyimides but requires cure temperatures >300◦ C to achieve
optimum end-use properties. However, in order to reduce damage and improve yields on certain devices
such as the next generation memory devices, a low temperature cure, less than 250◦ C, is now becoming an
important requirement. To meet this demand, a low temperature curable (<250◦ C) aqueous-developable,
positive-tone PBO has been developed by adopting an effective cure promoter without sacrificing any litho-
graphic (140 mJ/cm2 exposure energy at 5 µm cured film thickness) or end-use properties (70% elongation).
Keywords
Polybenzoxazole, low temperature cure
1. Introduction
Due to their excellent thermal, mechanical and electrical properties, polyimides
have been widely used as stress buffer coatings to improve the reliability of semi-
conductor devices [1] and, more recently, have also been used as cover coating films
for wafer level chip size packages having re-routing distribution layers of copper.
In recent years, the need for process cost reduction and the capability for produc-
*To whom all correspondence should be addressed. Tel.: 81-294-23-8990; Fax: 81-294-23-8992; e-mail:
m-ooe@hkg-net.co.jp
182 M. Ohe et al.
Figure 1. Mechanisms of positive-tone photo-definable PBO with NQD system (ICA: Indene Car-
boxylic Acid; TMAH: Tetramethylammonium hydroxide).
ing fine patterns have resulted in gradual replacement of non-PDPIs by PDPIs.

While the first PDPI products on the market were solvent-developable negative-
acting types, aqueous-developable positive-tone materials are now becoming the
mainstream technology due to additional demands such as higher resolution, further
reduction in process cost and the need for more environmentally friendly chemicals
and this has resulted in the commercialization of aqueous-developable positive-tone
products based on both polyimide and PBO polymers [2–4]. An example of the
photo-mechanism that utilizes poly(o-hydroxyamide) (PBO precursor) and naph-
thoquinonediazide (NQD) is shown in Fig. 1.
Although DRAM, SRAM and flash memory are widely used for conventional
semiconductor memory devices, it should be noted that these memory devices do
have advantages and disadvantages with regards to performance. For example, with
DRAM and SRAM, the writing and reading times are very fast but the memories
are volatile. On the other hand, flash memory is non-volatile but the writing time
is longer when compared to DRAM and SRAM and the rewriting number is also
limited. With this in mind, the next generation memory devices currently under
development (such as FeRAM and MRAM) will address these disadvantages and it
is expected that these devices will become the mainstream in the future. Regarding
the fabrication process of these future memory devices, temperatures above 250◦ C
will not be acceptable due to the low heat resistance of certain materials being
utilized in these devices and this means that a low temperature curable material
will be needed for use as a stress buffer coating on these new memory devices.
In addition, a low temperature curable material is also required as a cover coat
film for wafer level chip size packages in order to reduce yield loss related to high
temperature cure. To meet these demands, a low temperature curable (<250◦ C)
Low Temperature Curable, Photo-Definable Poly(benzoxazole) 183
Table 1.
Process conditions (cured thickness: 5 µm)
Process Process conditions Remarks
Coating 1000 rpm/5 s + 3100 rpm/30 s –

Pre-baking 120◦ C/180 s –
Exposure 100–810 mJ/cm2 i-line stepper
Development 40 s + 40 s (2 step) 2.38% TMAH∗
* 2.38%-Tetramethylammonium hydroxide aqueous solution.
aqueous-developable positive-tone photo-sensitive PBO has been developed and,

in this paper, the approaches for the material development and various process and
end-use properties of the developed material are described.
2. Experimental
2.1. Materials
Poly(o-hydroxyamide), the PBO precursor, was synthesized by the condensation
reaction of an aromatic dicarboxylic acid, an aromatic bis(aminophenol) deriva-
tive and an end-capping reagent in 1-methyl-2-pyrrolidone solvent. The synthesized
polymer was subsequently isolated by precipitation, washing and drying.
The NQD compound, cross-linker, cure promoters and solvents were used as
received.
2.2. Preparation of Photo-Sensitive Compositions
The photo-sensitive compositions were prepared by dissolving the PBO precursor,
NQD compound, cross-linker and cyclization promoter into γ -butyrolactone and
then filtered using a 3 µm pore-size filter.
2.3. Process Conditions for Lithographic Properties Evaluation
Lithographic properties were evaluated under the conditions shown in Table 1. Sil-
icon wafers were used for the evaluations.
2.4. Thermal Analyses
The cyclization temperature of the polymers was determined by differential scan-
ning calorimetry (DSC) using a Perkin Elmer DSC-7. The glass transition temper-
ature (Tg ) of the films was measured by thermo-mechanical analysis (TMA) using
a SEIKO TM/SS6000. 1% and 5% weight loss temperatures of the films were mea-
sured by thermo-gravimetric analysis (TGA) using a SEIKO TG/DTA6300.
2.5. Evaluation of Mechanical Properties
Mechanical properties were measured on 10 µm thick cured films using a Shimadzu
AGS-100NH system.
184 M. Ohe et al.
2.6. FT-IR Analysis

Fourier transfer infrared (FT-IR) analysis to determine the degree of cyclization of
PBO films was performed using a Shimazu FT-IR-8300, equipment.
2.7. Adhesion Strength Measurement
The adhesion strength between cured PBO film and the substrate (Si, SiN) was
measured using a stud-pull tester. An Al stud-pin was bonded to the substrate with
the PBO film with an epoxy adhesive and the adhesion strength was measured by
pulling the Al stud-pin. The tests were conducted before and after pressure cooker
test (121◦ C, 100% RH).
2.8. Viscosity Measurement
The viscosity was measured using a Brookfield Viscometer Model DV-2+ at 25◦ C.

3.1. Guidance for the Development
It is known that some bases (4-hydroxypyridine, benzimidazole) and acids (p-
toluenesulfonic acid) are effective in promoting cyclization reactions of both PI
and PBO precursors [5–7]. However, the addition of these materials into photo-
sensitive-PBO based formulations can reduce the storage stability (especially sta-
bility at room temperature). With this in mind, it was decided to evaluate materials
(for example, salt types or acid ester types) that were both stable at pre-bake tem-
peratures (ca. 120◦ C) and which could also release bases or acids through decom-
position during the cure cycle (ca. 200◦ C). For this selection procedure, TGA was
used as a guideline for decomposition temperature and DSC was used to confirm the
cyclization promotion effect. A DSC trace of a conventional PBO precursor (poly
(o-hydroxyamide)) is shown in Fig. 2 where an endothermic peak is observed at
around 300◦ C. In the case where a particular cyclization promoter has been added
(Fig. 3), the endothermic peak is observed at around 260◦ C, so resulting in a ca.
40◦ C shift towards a lower cure temperature. Using this method, the effect of se-
lected additives on cyclization temperature was determined.
Furthermore, FT-IR was employed to measure the degree of cyclization. IR spec-
tra of the pre-baked and cured films are shown in Fig. 4. The degree of cyclization
is calculated from the absorbance of the C–N stretch at 1539 cm−1 and of the
C=O stretch at 1651 cm−1 which are derived from the PBO precursor. These peaks
decrease as the cyclization reaction proceeds and the degree of cyclization is calcu-
lated from the following equation,
Cyclization degree (%) = (A0 − At )/(A0 − A100 ) × 100, (1)
where At is the absorbance of the film after curing under various conditions at either
1539 or 1651 cm−1 , A0 is the absorbance of the film at 0% cyclization degree that
Figure 2. DSC trace of conventional PBO precursor.
Figure 3. DSC trace of PBO precursor with cyclization promoter.
is obtained from the pre-baked film processed at 120◦ C for 3 min and A100 is the
absorbance of the film at 100% cyclization degree that is obtained from the film
cured at 320◦ C for 1 h.
3.2. Optimization of Formulation of Low Temperature Curable Material
An alkali-developable, positive-tone, photo-sensitive PBO comprising of a PBO

precursor, NQD compound and cross-linker as the main components has already
been developed by Hitachi Chemical DuPont MicroSystems. Using this technol-
ogy, a number of cyclization promoters were tested using the evaluation parameters
shown in Table 2 and, from these tests, a formulation containing a particular cy-
clization promoter was selected.
186 M. Ohe et al.
Figure 4. IR spectra of the pre-baked and cured films.
Table 2.
Evaluation parameters for screening cyclization promoter
Parameters Details of evaluation on addtive effect of promoter
Cyclization degree Promotion effect of cyclization degree

Lithographic properties Additive effect of promoter on lithoraphic properties
Development properties Checking of residue after development
Mechanical properties Impact on mechanical properties
Thermal propeties Impact on thermal properties
Moisture uptake Impact on moistue uptake
Adhesion Adhesion to various substrates after PCT∗
Storge stabiliy Checking precipitation and property change after storage
* Pressure cooker test (121◦ C/100% RH, 2 atm).
3.3. Various Properties of the Developed Material
(A) Cyclization degree

The FT-IR spectra of the PBO formulations with the selected cyclization pro-
moter cured at different temperatures are shown in Fig. 5 and the degree of cycliza-
tion both with and without the cyclization promoter at various cure temperatures is
given in Fig. 6. As can be seen from the FT-IR spectra, the peak at 1539 cm−1 due to
the C–N stretch and the peak at 1651 cm−1 due to the C=O stretch are not observed
above a cure temperature of 225◦ C. In this case, the degree of cyclization was cal-
culated at >90% at cure temperatures above 225◦ C as compared to the formulation
without any cyclization promoter where around 80% conversion was obtained when
cured at 250◦ C. These results indicate that the addition of the cyclization promoter
is very effective in reducing the cure temperature.
Figure 5. IR spectra of the cured films at various temperatures.
Figure 6. Effect of promoter addition on cyclization degree.
(B) Film properties

Cured film properties are shown in Table 3 where it can be seen that the devel-
oped material has a high % elongation after being cured at 225◦ C (as indicated by
the high degree of cyclization). It should be noted that as the cure temperature is
raised, the tensile strength, Tg and weight loss temperature tend to increase gradu-
ally while the modulus decreases gradually.
Table 4 shows a comparison of the film properties between the developed mate-
rial, a conventional PBO formulation and PIX-3400 (conventional non-PDPI). The
Tg and weight loss temperature for the developed material cured at 225◦ C were
slightly lower than those obtained with the conventional PBO cured at 320◦ C and
PIX-3400 cured at 350◦ C. However, there were no remarkable differences in the
other end-use properties. Based on these results, the cure temperature for the PBO
188 M. Ohe et al.
Table 3.
Film properties of the developed material at various cure temperatures
Properties Unit Cure temperature (X ◦ C)

200 225 235 250 320
Mechanical properties Tensile strength MPa 100 110 105 130 135
Modulus GPa 2.6 2.4 2.2 2.1 1.8
Elongation % 25 75 70 90 65
Thermal properties Tg ◦C 240 260 270 280 310
CTE ppm/◦ C 72 60 69 58 58
1% weight loss temp. ◦C 260 310 325 345 415
5% weight loss temp. ◦C 345 435 450 460 490
* Cure conditions: 150◦ C → X ◦ C/90 min + X ◦ C/60 min.
Table 4.
Comparison of film properties
Properties Unit Developed Conventional PIX-34001

material PBO material (Non-PDPI)
Cure conditions ◦ C/min 225/60 320/60 350/60

Mechanical properties Tensile strength MPa 110 155 112
Modulus GPa 2.4 2.1 2.6
Elongation % 75 83 100
Thermal properties Tg ◦C 260 300 298
CTE ppm/◦ C 60 57 43
1% weight loss temp. ◦C 310 389 450
5% weight loss temp. ◦C 435 465 515
Electrical properties Dielectric constant – 3.0 2.9 3.4
Breakdown voltage kV/mm 384 470 250
Moisture uptake % 1 <0.5 1.6
Adhesion retention time2 h >500 >500 >500
1 Trade name of Hitachi Chemical DuPont MicroSystems Ltd.

2 Test conditions.
Substrate: Si, PCT (Pressure cooker test): 121◦ C/100% RH, 2 atm, Tape test.
formulation with the selected cyclization promoter can only be lowered by ca. 70◦ C
without sacrificing any end-use properties as compared to conventional PBO sys-
tems.
(C) Lithographic properties
Figure 7 shows the film retention as a function of i-line exposure dose and shows
that the pattern can be opened with exposure energy greater than 140 mJ/cm2 for
a 5 µm thick cured film. In addition, good linearity between mask size and pattern
size was obtained as shown in Fig. 8.
Figure 7. Sensitivity curve for i-line exposure (cured film thickness: 5 µm).
Figure 8. Pattern size vs mask size (i-line exposure, cured film thickness: 5 µm).
Figure 9. SEM cross-sectional images of the film (thickness after cure: 5 µm, mask size: 7 µm).
(D) Pattern shapes of the film

Figure 9 shows SEM cross-sectional images of the film (both before and after
cure). No residues were observed after development and no significant change in
pattern shape was obtained after cure.
190 M. Ohe et al.
Figure 10. Adhesion strength between cured films and substrates (the data in parentheses show the
cured temperatures).
Figure 11. Viscosity of photo-sensitive formulation at RT storage.
(E) Adhesion strength between cured film and substrates

Figure 10 shows the adhesion strength between the cured film and the substrates
(Si, SiN) at various cure temperatures. The films cured at 225◦ C showed good ad-
hesion strength both before and after PCT treatment for both substrates.
(F) Storage stability
Figures 11 and 12 show the effect of storage time on viscosity of photo-sensitive
formulation at both room temperature (RT) and −18◦ C. It can be seen that no no-
ticeable change in viscosity was found after 30 days at RT or after 12 months at
−18◦ C. In addition, no precipitation of any components and no effects on litho-
graphic properties were observed over this time period.
Figure 12. Viscosity of photo-sensitive formulation at −18◦ C storage.
4. Conclusion
A low temperature curable (<250◦ C) positive-tone aqueous-developable PBO has
been developed by introducing an effective cure promoter without sacrificing any
lithographic or end-use properties. The cure temperature for this system is ca. 70◦ C
lower as compared to conventional PBO products, thus allowing its use as a pro-
tective layer on next-generation memory devices where low temperature cure is
a requirement.
References
1. K. Sato, S. Harada, A. Saiki, T. Kimura, T. Okubo and K. Mukai, IEEE Trans. Parts. Hybrid
Packaging PHP-9, 176 (1973).
2. H. Makabe, T. Banba and T. Hirano, J. Photopolym. Sci. Technol. 10, 307–312 (1997).
3. M. Tomikawa, M. Suwa, S. Yoshida, Y. Fujita, R. Okuda and G. Ohbayashi, J. Photopolym. Sci.
Technol. 13, 357–360 (2000).
4. M. Nunomura, M. Sasaki, M. Ohe, N. Yamazaki and T. Miwa, Hitachi Chemical Technical Report
34, 25–28 (2000).
5. F. Ghaffari and K. A. Hodd, Thermochimica Acta 41, 213–224 (1980).
6. T.-K. Kim, K.-Y. Choi, K.-S. Lee, D.-W. Park and M. Y. Jin, Polymer Bulletin 44, 55–62 (2000).
7. M. Oba and Y. Kawamonzen, J. Appl. Polym. Sci. 58, 1535–1542 (1995).
Multiple Relaxation Processes in Swift Heavy
Ion Irradiated Kapton-H Polyimide
J. K. Quamara ∗ , Maneesha Garg and Geetika Goyal

Department of Physics, National Institute of Technology, Kurukshetra-136119, India
Abstract
Swift heavy ion irradiation effects in Kapton-H polyimide have been investigated using thermally stimulated
depolarization current (TSDC) technique. The Kapton-H samples were irradiated with 50 MeV Si, 50 MeV
Li, 75 MeV O and 50 MeV Ni using Pelletron facility at the Nuclear Science Centre, New Delhi. The range
of fluence varied from 1011 to 1013 ions/cm2 . The two major relaxation processes in Kapton-H polyimide
viz. β-relaxation associated with dipolar nature of the carbonyl groups and α-relaxation associated with the
space charge relaxation process are both affected by ion irradiation. The TSDC spectra obtained for various
poling temperatures (80–250◦ C) and poling fields (40–600 kV/cm) of irradiated samples reveal, on the one
hand, a significant loss in the β-relaxation due to demerization of carbonyl groups and, on the other hand,
there is an increase in the space charge relaxation (α-relaxation) owing to the formation of new energy trap
centers. Though the general behaviour of the TSDC spectra is not much affected by the nature of ion, the
fluence and the energy of the ion play a significant role in modifying these relaxation processes. Some new
radiation induced relaxation processes due to increase in water absorption (γ -relaxation) and unsaturated
bond formation (δ-relaxation) have been recognized. The distribution of the relaxation times for the TSDC
spectra has been estimated using the Cole–Cole distribution function. The Cole–Cole distribution curves
show the formation of new sub-polar groups with different characteristic times. The different relaxation
parameters viz. activation energy and relaxation time have also been calculated.
Keywords
High-energy heavy ion irradiation, Kapton-H polyimide, FT-IR spectroscopy
1. Introduction
The effect of high-energy heavy ion irradiation on polymers has drawn considerable
attention during the last couple of decades. These studies have been mainly directed
in two directions. On the one hand, the modification in physico-chemical properties
due to irradiation in the development of new sensor devices [1, 2] or generating con-
ducting properties in polymers, and, on the other hand, the morphological changes
occurring which affect the performance of polymers in a radioactive environment
* To whom correspondence should be addressed. Tel.: +91 1744- 239636; Fax: +91 1744-238050; e-mail:
jkquamara@yahoo.com
194 J. K. Quamara et al.
have been of considerable interest. Though many groups [3–7] have been concerned
with working out the mechanism of energetic ion interaction with polymers which
results in a significant change in electrical properties, little has been reported on the
high-energy heavy ion irradiation effect on dielectric behaviour [8, 9]. The study
of dielectric relaxation behaviour after irradiation may play an important role in
determining the configurational changes in polymers due to irradiation. The dielec-
tric relaxation investigations, in particular, will not only reveal the morphological
changes such as formation of free radicals, demerization of carbonyl groups, cross-
linking, generation of new energy trap centers and appearance of new phases due to
high-energy irradiation but would also be helpful in determining the suitability of
using polymers in such environment.
In the present work the various relaxation processes in Kapton-H polyimide irra-
diated with oxygen, silicon, nickel and lithium ions at different fluences have been
reported. To accomplish this, the thermally stimulated discharge current (TSDC)
technique is the most suitable tool, as it provides a deep insight into the molecu-
lar dynamics associated with the different relaxation processes [10, 11]. The major
advantage of this technique lies in the wide combination of excitation conditions
that can be used. The various current peaks appearing in the TSD current spectrum
of the polarized sample (termed as electret) are capable of analyzing the various
operative relaxation mechanisms in a polymer.
The TSD current density i(T ) at a temperature T can be expressed as [11]:
τ
Nμ2 Ep U 1 −U
i(T ) = exp − − exp − dT , (1)
3kTp τ0 kT βτ0 τ0 kT
where U is the activation energy, Ep an average field, N the dipolar concentration,
μ the dipole moment, β the linear heating rate, τ0 the relaxation time when T → ∞,
k the Boltzmann’s constant and T0 the room temperature.
From equation (1) one obtains
∞
i(t) dt
τ (t) = , (2)
t (T ) i(t)
where i(t) is the current at time t.

The value of τ at any temperature (T ) can be estimated with fairly good accuracy
as
Area under the TSDC curve between T0 and T
τ (T ) = .
Value of TSD current at temperature T
The temperature variation of τ is given by the Arrhenius equation as
U
ln τ = ln τ0 + . (3)
kT
Values of τ at different temperatures can be obtained from equation (2). A linear
plot between ln(τ ) and 1/T yields the value of activation energy.
Multiple Relaxation Processes in Swift Heavy Ion Irradiated Kapton-H Polyimide 195
2. Experimental
The Kapton-H polyimide used in the present investigations was procured from
DuPont, USA. The samples were irradiated with high-energy heavy ion beams uti-
lizing the 15 UD Pelletron Accelertaor Facility at the Nuclear Science Centre, New
Delhi, India. The different ion beams and their corresponding energies are listed in
Table 1.
The samples (irradiated as well as pristine) were polarized [11] by subjecting
them to the desired dc field (Ep ) at a constant temperature (Tp ) for 1 h. The samples
were then cooled to room temperature within a fixed time (30 min) in the pres-
ence of Ep . The frictional stray surface charges accumulating during polarization
were eliminated by short circuiting the electrodes for about 10 min. In order to ob-
tain depolarization current the system was short-circuited through an electrometer
(Keithley 610 C) and the samples were depolarized at a constant heating rate of
2◦ C/min. The corresponding activation energy was calculated through the Bucci-
plot method [11].
3. Results
The TSDC spectra of Kapton-H samples irradiated with different ions consist,
in general, of maxima situated around 40–60◦ C, 110–130◦ C, 200–210◦ C and
230–245◦ C. These have been named as γ -, β-, α- and ρ-peaks, respectively. The
number of peaks actually appearing in a particular TSDC spectrum, their locations,
heights and sharpness are governed not only by the poling parameters, storage and
depolarization parameters, but also by the fluence at which the samples have been
irradiated. Some additional small peaks (kinks) may also appear in addition to these
maxima in certain TSDC characteristics.
Table 1.
Sample thickness and ion beams with corresponding energies and fluences
Ion beam Thickness Energy Fluence

(µm) (MeV) (ions/cm2 )
Si 25 50 2.3 × 1012
1.38 × 1013
Li 12.5 50 105
O 25 75 1.88 × 1011
1.88 × 1012
1.88 × 1013
5.63 × 1012
Ni 25 80 1.4 × 1012
3.1. TSDC Behaviour of Si Ion Irradiated Samples
The TSDC spectra of 50 MeV Si ion irradiated (fluence 2.3 × 1012 and 1.8 × 1013
ions/cm2 and thickness 25 µm) Kapton-H samples polarized at Ep = 400 kV/cm,
Tp = 90 and 200◦ C with different fluences are depicted in Fig. 1(a) and 1(b).
The low Tp polarized samples (Fig. 1(a)) show the presence of β-peak, and high
Tp polarized samples (Fig. 1(b)) show the presence of α-peak only and a constant
TSD current is obtained up to 180◦ C and then it declines.
3.2. TSDC Behaviour of Li Ion Irradiated Samples
The TSDC spectra of 50 MeV Li ion irradiated (fluence 1015 ions/cm2 and thick-
ness 12.5 µm) Kapton-H samples poled at different fields (Tp = 90◦ C) and different
(a)
(b)
Figure 1. TSDC spectra of silicon ion irradiated Kapton-H samples for different fluences: Q pristine,
F 2.3 × 1012 ions/cm2 and 2 1.38 × 1013 ions/cm2 ((a) Tp : 90◦ C and Ep : 400 kV/cm; (b) Tp : 200◦ C
and Ep : 400 kV/cm).
poling temperatures (Ep = 128 kV/cm) are illustrated in Fig. 2(a) and 2(b), respec-
tively.
The high Tp poled samples (Fig. 2(b)) show the presence of an α-peak around
180–200◦ C depending on the Tp . In certain cases a γ -peak also appears around
30◦ C. The magnitude of the α-peak increases with increasing Tp . The activa-
tion energy for the α-peak (0.32–0.99 eV) was found to be field dependent (not
shown).
(a)
(b)
Figure 2. TSDC spectra of lithium ion irradiated Kapton-H samples polarized at different (a) fields:
F 80 kV/cm, 2 128 kV/cm and Q 400 kV/cm (Tp : 90◦ C); (b) temperatures: F 90◦ C, 2 150◦ C and
Q 200◦ C (Ep : 128 kV/cm).
3.3. TSDC Behaviour of Oxygen Ion Irradiated Samples

The TSDC spectra of 75 MeV oxygen ion irradiated (fluence 1.8 × 1012 ions/cm2
and thickness 25 µm) Kapton-H samples poled at different fields (Tp = 80◦ C and
220◦ C) and different poling temperatures (Ep = 320 kV/cm) are illustrated in Figs
3(a), 3(b) and 4, respectively.
All the TSDC spectra for low Tp consist of one prominent peak around 120◦ C,
i.e., the β-peak. Though the peak is very broad for low Ep , it gets better defined
with increasing Ep . The peak height (Ip ) increases with Ep . In the low temperature
region, though not well defined, a peak (γ -peak) around 50◦ C appears (shown in the
(a)
(b)
Figure 3. TSDC spectra of 75 MeV oxygen ion irradiated Kapton-H samples (fluence:
1.875 × 1012 ions/cm2 ) for different polarizing fields (Ep ): Q 160 kV/cm, 2 320 kV/cm and
F 400 kV/cm (Tp : (a) 80◦ C, (b) 220◦ C). Inset: Enlarged view of the low temperature TSDC curves.
Figure 4. TSDC spectra of 75 MeV oxygen ion irradiated Kapton-H samples (fluence:
1.875 × 1012 ions/cm2 ) for different polarizing temperatures (Tp ): Q 80◦ C, 2 120◦ C, F 150◦ C,
" 220◦ C and × 250◦ C (Ep : 320 kV/cm). Inset: Enlarged view of the low temperature TSDC curves.
insets of Figs 3(b) and 4), the height of which increases with increasing Ep . In the
high temperature region, the TSDC spectra do not show any well-defined maxima
but a few kinks appear in the temperature region 190–220◦ C. In the high Tp TSDC
spectra (Fig. 3(b)), the β-peak appears only in low Ep samples, whereas the α-peak
is present in high Ep samples. Though the height (Ip ) of the α-peak is not much
affected by Ep , a shift in the peak temperature occurs with increasing poling field.
The ρ-peak, around 240◦ C, is also observed in most of the cases and appears to be
independent of Tp .
The effect of fluence on the TSDC spectra of Kapton-H samples polarized
at different Ep /Tp has been illustrated in Fig. 5(a) and 5(b). The β-peak mag-
nitude initially decreases with increase in fluence from 1.8 × 1011 ions/cm2 to
1.8 × 1012 ions/cm2 . But, a significant increase in the peak magnitude (Ip ) oc-
curs at higher fluence (1.8 × 1013 ions/cm2 ). However, in all the irradiated samples
the β-peak magnitude is less than that of pristine sample. The α-peak (around
200–210◦ C) magnitude decreases with increasing fluence. A marginal increase in
the peak temperature is also observed with decreasing fluence.
3.4. TSDC Behaviour of Nickel Ion Irradiated Samples
The TSDC spectra of 80 MeV Ni ion irradiated (fluence 1.4 × 1012 ions/cm2 and
thickness 25 µm) Kapton-H samples poled at different fields (Tp = 80 and 200◦ C)
and different poling temperatures (Ep = 320 kV/cm) are illustrated in Figs 6(a),
6(b) and 7, respectively. The low temperature TSDC spectra are almost similar to
those obtained for oxygen ion irradiated samples except that the β-peak is sharper
in the former case. However, the nature of TSD currents in the high temperature
region is different in the two cases. In case of low Tp polarized oxygen ion irradi-
(a)
(b)
Figure 5. TSDC spectra of pristine and 75 MeV oxygen ion irradiated Kapton-H samples
for different fluences Q pristine, 2 1.875 × 1011 ions/cm2 , F 1.875 × 1012 ions/cm2 and
(a) " 1.875 × 1013 ions/cm2 polarized at Tp : 80◦ C and Ep : 160 kV/cm; (b) " 3.93 × 1013 ions/cm2
polarized at Tp : 80◦ C and Ep : 320 kV/cm.
ated samples where we observe a continuous rise with temperature above 170◦ C
(Fig. 3(a)), in case of Ni ion irradiated samples the TSD current gives a small max-
imum around 220◦ C.
In Fig. 7 we observe a peak around 30◦ C. This peak is equivalent to the low
temperature TSDC γ -peak observed in oxygen ion irradiated samples (Fig. 4). The
β-peak appearing around 110–150◦ C is prominent in samples poled at low temper-
atures whereas the peak around 210◦ C appears in high Tp samples.
4. Discussion
The various current maxima appearing in the TSDC spectra of pristine and irradi-
ated samples represent the different relaxations or transitions. In pristine Kapton-H
polyimide two major transitions characterized as β and α occurring in the ascend-
ing order of temperature have been reported [12–14], assigned to the electric dipole
(a)
(b)
Figure 6. TSDC spectra of 80 MeV Ni ion irradiated Kapton-H samples with different polarizing fields
(Ep ): (a) Q 160 kV/cm, 2 320 kV/cm, F 400 kV/cm and × 600 kV/cm (Tp : 80◦ C); (b) Q 160 kV/cm
and 2 320 kV/cm (Tp : 220◦ C).
Figure 7. TSDC spectra of 80 MeV Ni ion irradiated Kapton-H samples (fluence:

1.44 × 1012 ions/cm2 ) for different polarizing temperatures (Tp ): Q 80◦ C, F 150◦ C, " 220◦ C and
× 250◦ C (Ep : 320 kV/cm).
orientation and the space charge polarization, respectively. The low Ep /Tp samples
are favorable for β-peak, whereas α-peak is dominant at high Ep /Tp [11]. In irra-
diated samples in addition to β- and α-transitions in the low temperature region of
the TSDC spectra a relaxation associated with absorbed water also occurs.
4.1. γ -Peak
In most of the TSDC spectra (Figs 2(a), 3(b), 4 and 7), we observe current max-
ima in the low temperature region 30–70◦ C. Although this current maximum is
not sharp, its presence cannot be ignored. The cause of this peak can be associated
with a relaxation process due to absorbed water. The water-dependent relaxation
behaviour studied by many groups in Kapton-H polyimide is quite complex and pre-
sumably results from several different competitive mechanisms involving physical
states of absorbed water. Shirnet et al. [15] have shown an increase in water absorp-
tion capacity in Kapton-H due to irradiation. Usually this γ -relaxation process lies
in the vicinity of β-relaxation as discussed later and sometimes it becomes difficult
to seperate these two relaxations. The presence of γ -relaxation and its enhancement
due to irradiation can be confirmed from the following: after irradiation the magni-
tude of β-peak should decrease due to the decomposition of carbonyl groups but,
on the contrary, we observe a significant increase in the peak current magnitude.
This shows the dominance of γ -relaxation process in the irradiated samples. This
is further confirmed from the following facts: (i) there is an increase in the β-peak
magnitude with increase in fluence, (ii) no linear relationship between Im (peak cur-
rent) and Ep is observed in irradiated samples which is an essential feature for the
pure dipolar relaxation process, (iii) the presence of low temperature peak around
30◦ C in high Tp irradiated samples.
4.2. Dipolar Relaxation (β-peak)
The low temperature peak appearing in the TSDC spectra of pristine as well as ir-
radiated Kapton-H samples in the temperature range 80–110◦ C has been assigned
to electric dipole orientational polarization resulting from the molecular motion of
the carbonyl groups (C=O in the polymer film). The dipolar origin of this peak has
been established by Bhardwaj et al. [12] owing to the facts: (i) peak current varies
linearly with the polarizing field which is an indication of uniformly induced bulk
polarization, (ii) in comparison to the other peaks, its magnitude has been found
to be less sensitive to the variation in forming parameters viz. Ep , Tp and ts (stor-
age time) as well as to the specimen thickness, (iii) during the electret storage it
is the low temperature tail region of this peak that gets affected the most. These
are the usual characteristics associated with β-relaxation [12, 13]. It is well doc-
umented [12] that in pristine Kapton-H polyimide, at a particular Tp , if the value
of Ep is systematically increased, more and more polar groups are in a position to
orient by surmounting the intermolecular forces as a result of the increasing torque
experienced by them. Consequently, the corresponding relaxation peaks heighten.
Similarly, when Tp is increased at a particular value of Ep , more and more dipoles
are in position to orient due to the increased mobility, resulting in an increase in

the peak height. The effect of energetic ion interaction on β-relaxation is evident
from the TSDC behaviour of irradiated Kapton-H samples. The smoothness of the
β-peak is lost in almost all cases. Investigations of Marletta et al. [16] show that
although ether bonds are weak in Kapton-H but the carbonyl group (C=O) is most
prone to high-energy ion induced radiation damage. Phenyl rings are most stable
to radiation induced damage. This can also be confirmed from the energies associ-
ated with various bonds viz. 3.03 (C–N), 3.59 (C–C), 3.63 (C–O), 4.29 (C–H) and
6.34 eV for (C=C) at 25◦ C [3]. So a release of carbonyl group from imidic rings re-
quires 6.63 eV to break one C–C and one C–N bond. Similarly a release of oxygen
from ether linkage requires 7.27 eV to break two C–O bonds and a release of car-
bon from a phenyl ring requires 9.93 eV to break σ and π bonds. This suggests that
with ion-irradiation, severe damages occur at carbonyl groups and ether linkages but
phenyl rings will be least affected. Thus it is most probable that the ion irradiation
causes the decomposition of both carbonyl groups and ether linkages. This is further
confirmed from the FT-IR analysis of pristine and irradiated Kapton-H polyimide
discussed later. The average activation energy for β-peak (0.60 ± 0.01 eV) is higher
than the theoretically predicted value of E (≈0.2 eV) for the side group orientation
in a polymer [17]. This shows that β-relaxation process in an irradiated sample is
accompanied by other relaxation processes. We term this as δ-relaxation. The ex-
istence of δ-relaxation in the vicinity of β-relaxation was also confirmed from the
TSDC behaviour of re-poled sample (not shown) where we observed significantly
broad maximum. Usually, this maximum does not appear in the case of pristine
re-poled samples, since the β-relaxation gets suppressed in high temperature an-
nealed samples. This is also supported by the fact that even in high Tp samples we
find the presence of current maxima which is not very significant. The major source
of δ-relaxation is the formation of shallow energy trap centers due to ion irradia-
tion. Investigations of Marletta et al. [16], Garg and Quamara [4] have shown that
the carbonyl groups are most prone to high-energy ion induced damage. A loss of
a sufficient number of carbonyl groups would decrease the β-peak magnitude in
irradiated samples, which should be proportional to fluence (Figs 1(a) and 5(a)).
However, in high fluence irradiated samples enhancement in the β-peak may be
associated with a significant presence of δ-relaxation.
A further confirmation of the presence of δ-relaxation along with β-relaxation
is provided by the aging effect. We observe a fast erosion in residual polarization
under the β-peak during aging (Fig. 8) which is not normal for a pure β-relaxation
process. Since the dipoles which are responsible for the β-relaxation process pos-
sess nominal mobility at room temperature, the decay in the residual polarization
should be very low for this peak [12]. It is not easy to make any certain comment
on the origin of δ-relaxation but the above-mentioned observations suggest that this
relaxation has the space charge relaxation character associated with shallow energy
trap centers. These shallow trapping centers in irradiated samples may be provided
Figure 8. Effect of aging on the TSDC spectra of 75 MeV oxygen ion irradiated Kapton-H samples
at Q 0 h, 2 15 h and F 50 h, polarized at Tp : 100◦ C and Ep : 320 kV/cm.
by the demerization of imide linkage which is less reactive in free radical formation
as compared to carbonyl groups [3].
4.3. Space Charge Relaxation (α-peak)
The current maximum corresponding to α-relaxation in the TSDC spectrum of pris-
tine Kapton-H has been reported to be around 180–200◦ C [12, 13]. This relaxation,
which is more prominent in high Ep /Tp samples, has been mainly attributed to the
space charge polarization relaxation process. The unsaturated phenyl and carbonyl
side groups present in the structure of Kapton-H can act, to a significant extent,
as effective charge trap centers contributing to the α-relaxation. The high-energy
ion irradiation of Kapton-H polyimide results in the cross-linking of imide groups
which also act as as deep/shallow energy traps [16]. The formation of conjugated
and triple bonds due to delocalization of electrons and the demerization of carbonyl
groups [3], which leads to the formation of free radicals due to high-energy ion ir-
radiation in Kapton-H polyimide, would also provide a new source of deep energy
traps. The α-peak in irradiated samples is certainly affected by these new processes.
All these processes give a very complex nature to the α-relaxation. It is not surpris-
ing that unlike in pristine Kapton-H samples we observe a shift in the peak position
with Ep (Figs 3(b) and 6(b)) and also we do not find an Ep dependence of Im . The
dependence of the α-peak magnitude on polarizing temperature (Fig. 2(b)) also in-
dicates the existence of distributed polarization in irradiated samples. At low values
of Tp only the fast sub-transitions could be operative but with the increasing Tp
more sub-transitions with longer relaxation times could be activated. This will shift
the peak towards higher temperature and will also increase the peak magnitude.
An increase in α-peak magnitude with increase in fluence has been reported by
Garg and Quamara [4]. The α-peak, due to the space charge relaxation process,
mainly depends on the deep energy trap center density. An increase in fluence
should not only increase the peak height due to the availability of more trapping
centers, but should also shift the peak towards higher temperature. However, in the
present study we find a decrease in the α-peak magnitude and a shift in the peak
position towards lower temperature side with increase in fluence (Fig. 5(a)). This
opposite trend can be attributed to the different orders of the fluence used for irra-
diation in the two studies. In case of Garg and Quamara [4] the order of the fluence
was very low (5 × 104 ions/cm2 and 5 × 105 ions/cm2 ), whereas in the present
case it ranged from 1.8 × 1011 to 1.8 × 1013 ions/cm2 . Hioki et al. [8] and some
other groups [18, 19] have shown a significant increase in the electrical conductiv-
ity of Kapton-H polyimide due to energetic heavy ion irradiation at high fluences.
The increase in conductivity, which is proportional to the fluence, has been associ-
ated to an increase in the delocalized π electrons from the unsaturated bonds and
an increase of charge mobility enhanced by chain cross-linking [3]. In irradiated
Kapton-H samples during the process of polarization when subjected to Ep , the
amount of charge carriers which may get depleted would be more than the charge
trapped in deep/shallow trapping centers. This would result in a decrease in the
α-peak magnitude with increase in fluence and also shift of the peak towards lower
temperature. The depletion of charge carriers from deep energy centers would also
result in lowering the activation energy for this relaxation, which is found to be
1.04 ± 0.01 eV in comparison to the activation energy for the same relaxation in
pristine Kapton-H sample, which is of the order of 1.10 ± 0.01 eV.
4.4. Space Charge Relaxation (ρ-peak)
In some of the TSDC spectra (Figs 3(a), 4, 5(a) and 5(b)), we observe a current
maximum in a high temperature region around 240–250◦ C, termed as the ρ-peak.
A similar peak has also been observed in case of pristine Kapton-H samples by
Bhardwaj et al. [12] at high Ep . This peak has been associated with the charges
injected into the polymer from the electrode directly. In the present case also the
same mechanism appears to be responsible for the ρ-peak. The value of activation
energy for this relaxation process (≈1.32 ± 0.01 eV) is comparable to the energy
generally associated with the migration of ions [12].
4.5. Effect of Nature of Ion on TSDC Behaviour
As far as the nature of the ion is concerned we have not observed any significant
change in the general nature of TSDC spectra. Both oxygen and Ni ion irradiated
Kapton-H samples give TSDC characteristics revealing a similar behaviour. How-
ever, the high temperature TSD currents for oxygen ion irradiated samples have
been found to be much higher as compared to Ni ion irradiated samples. Oxygen ion
irradiation has been known to increase the electrical conductivity [3] of Kapton-H
polyimide by several orders of magnitude. Therefore, it is not unlikely that we may
observe large TSD currents particularly in the high temperature region (dominance
of conduction current) in oxygen ion irradiated samples. This is also in accord with
our FT-IR investigations as discussed later.
4.6. Cole–Cole Distribution Curves

The first mathematical distribution function was derived by Wagner assuming that
an infinite number of independent causes could disturb the original relaxation time
τm (τm is the relaxation time). Since then many authors [20–22] have reported
several other distribution functions. For the TSDC analysis the most suitable dis-
tribution function is considered to be that given by Cole–Cole and is termed as the
Cole–Cole distribution function. We calculate the Cole–Cole distribution function
of β-peak for Kapton-H samples irradiated by Li ions, for different poling parame-
ters using the following equation [22]
sin(βπ)
f (u) = , (4)
2π[cosh(βu) + cos(βπ)]
where −∞ < τ < ∞ and 0 < β 1.
The values of τ and τm are evaluated using the Arrhenius equation given as

E
τ (T ) = τ0 exp , (5)
kT
where τ0 is the pre-exponential factor or relaxation time at infinite temperature and
E is the activation energy of dipolar orientation or disorientation.
The graph between ln(τ/τm ) and f (ln τ/τm ) (Fig. 9) is plotted to find the dis-
tribution in τ . The Cole–Cole distribution curves for O ion irradiated Kapton-H
samples polarized with different fields at Tp = 80◦ C are illustrated in Fig. 9. These
curves show an increase in distribution in τ with increasing polarizing field.
This is corroborated by the fact that with increasing polarizing field there should
be an enhancement in the spectrum of dipolar relaxation times. The spectrum of
relaxation times in the irradiated samples may be due to the variation in the shape
and size of the rotating dipolar groups and formation of new sub-polar groups with
different relaxation times. In addition, the anisotropy of the internal field in which
the dipoles reorient may also contribute to the relaxation spectrum.
4.7. FT-IR Investigations
4.7.1. Pristine Kapton-H Polyimide
The FT-IR spectra of pristine Kapton-H polyimide for three different thicknesses
are illustrated in Fig. 10. The characteristic bands (for 25 µm thickness) appear-
ing around 717, 1079, 1160, 1230, 1367, 1494, 1601, 1703 and 1770 cm−1 are
assigned, respectively, to C=O bending, C–O–C stretching, C–C bending, C–N
stretching (two peaks, i.e., 1230, 1367 cm−1 ), aromatic C=C ring stretching (two
peaks, i.e., 1494, 1601 cm−1 ), C=O symmetrical stretching and C=O asymmetri-
cal stretching. This is in conformity with the FT-IR analysis results on Kapton-H
polyimide obtained by other groups [23–25].
4.7.2. Irradiated Kapton-H Polyimide
Figure 11 illustrates the FT-IR spectra of 75 MeV O+ ion irradiated (fluences:
9.1×1011 , 5.6×1012 and 1.8×1014 ions/cm2 ) Kapton-H polyimide. The reduction
(a)
(b)
(c)
Figure 9. The Cole–Cole distribution graphs for oxygen ion irradiated Kapton-H samples at different
polarizing fields (Ep ): Q 160 kV/cm, 2 320 kV/cm and F 400 kV/cm.
in the intensity of typical bands in the FT-IR spectra indicates that certain mor-
phological changes have occurred in the polymer due to irradiation. A significant
decrease in the ether and carbonyl bands after irradiation has been reported by De-
vanas [25]. Interestingly, we observe a very broad band between 2500–3500 cm−1
which is absent in case of pristine samples. We associate this broad band to the
presence of absorbed water in irradiated samples. The fact that this band is very
broad and appears at a lower frequency than the band for the free O–H stretching
(3600 cm−1 ) which is very sharp in nature confirms the presence of absorbed water.
Further, the presence of H–O–H bending peak around 1640–1615 cm−1 (depend-
Figure 10. FT-IR spectra of pristine Kapton-H polyimide for different thicknesses: (a) 7.5 µm,
(b) 12.5 µm and (c) 25 µm.
Figure 11. FT-IR spectra of 75 MeV O+ ion irradiated Kapton-H polyimide (25 µm) for different
fluences: (a) 9.1 × 1011 , (b) 5.6 × 1012 and (c) 1.8 × 1014 ions/cm2 .
Figure 12. FT-IR spectra of 80 MeV Ni+ ion irradiated Kapton-H polyimide (7.5 µm) for different
ing on the fluence) confirms the association of band around 2500–3500 cm−1 with
the absorbed water. The presence of absorbed water in irradiated polymers has also
been discussed by Zhu et al. [26] In the pristine Kapton-H polyimide structure the
presence of carbonyl groups is revealed by the band appearing around 1703 cm−1 .
A decrease in the intensity of this band in irradiated samples shows the decrease in
the number of carbonyl groups.
The FT-IR spectra of Kapton-H polyimide irradiated with 80 MeV Ni (Fig. 12)
and 50 MeV Li (Fig. 13) ions are almost similar to that of oxygen ion irradiated
polyimide. A moderate shift in the frequency of the characteristic bands is due to
different Linear Energy Transfer (LET) of different ions. This is also in confirma-
tion with our earlier TSDC investigations of irradiated polyimide [27]. It has been
shown that the radiation induced effects in Kapton-H polyimide are irrespective of
the nature of ion.
5. Conclusions
The thermally stimulated depolarization current (TSDC) characteristics have been
investigated in 50 MeV Si, 50 MeV Li, 80 MeV Ni and 75 MeV oxygen ion ir-
radiated Kapton-H polyimide in the temperature range 30–250◦ C. The analysis of
these characteristics shows that high-energy ion irradiation affects all the prominent
relaxation processes in Kapton-H. The enhancement in water absorption due to ion
irradiation gives rise to a low temperature relaxation process (γ -relaxation). The de-
merization of carbonyl groups which not only affects the dipolar relaxation process
(β-relaxation) but also results in the formation of shallow energy trap centers giving
Figure 13. FT-IR spectra of 50 MeV Li+ ion irradiated Kapton-H polyimide (12.5 µm) for different
rise to a new relaxation process (δ-relaxation) in the vicinity of β-relaxation. The

radiation induced cross-linking of imide groups, formation of unsaturated bonds
and free radicals provide new deep energy centers affecting the α-relaxation.
Acknowledgements
Funds were provided by the Defense Research and Development Organization,
Government of India and the Nuclear Science Centre, New Delhi. Authors are
thankful to Dr. D. K. Avasthi (scientist, NSC, New Delhi) for his help in irradi-
ating of the samples.
References
1. H. Tahara, T. Kawabata, L. Zhang, T. Yasu and T. Yoshikawa, Nucl. Instrum. Meth. Phys. Res.
B121, 446 (1997).
2. L. Zhang, T. Yasu, H. Tahara and T. Yoshikawa, Jpn J. Appl. Phys. 36, 5268 (1997).
3. E. H. Lee, in: Polyimides: Fundamentals and Applications, M. K. Ghosh and K. L. Mittal (Eds),
p. 471. Marcel Dekker, New York, NY (1996).
4. M. Garg and J. K. Quamara, Nucl. Instrum. Meth. Phys. Res. B179, 389 (2001).
5. A. M. Ektessabi and S. Hakamata, Thin Solid Films 377–378, 621 (2000).
6. S. K. Rutledge and J. A. Mihelcic, in: Material Degradation in Low Earth Orbit, V. Sriniwasan
and B. A. Bank (Eds), p. 35. TMS, Warrendrce, PA (1990).
7. A. M. Ektessabi and Y. Watanabe, Rev. Sci. Instrum. 73, 849 (2002).

8. T. Hioki, S. Noda, V. Sugiura, M. Kakeno, K. Yamada and J. Kawamato, Appl. Phys. Lett. 43, 30
(1983).
9. J. K. Quamara, B. K. Kaushik, S. K. Mahna and K. Asokan, Vacuum 48, 995 (1997).
10. J. Van Turnout, in: Electrets, G. M. Sessler (Ed.), p. 81. Springer-Verlag, New York, NY (1980).
11. J. Vanderschueren and J. Gasiot, in: Thermally Stimulated Relaxation in Solids, P. Braunlinch
(Ed.), p. 135. Springer-Verlag, New York, NY (1979).
12. R. P. Bhardwaj, J. K. Quamara, B. L. Sharma and K. K. Nagpaul, Phys. Stat. Sol. (a) 77, 347
(1983).
13. T. Tanaka, S. Hirabayashi and K. Shibayama, J. Appl. Phys. 49, 784 (1978).
14. J. K. Quamara, P. K. C. Pillai and B. L. Sharma, Acta Polymerica 34, 265 (1983).
15. V. Shirnet, U. K. Chaturvedi and S. K. Agarwal, in: Polyimides: Synthesis, Characterization and
Applications, K. L. Mittal (Ed.), Vol. 1, pp. 555–571. Plenum Press, New York, NY (1984).
16. G. Marletta, C. Oliveri, G. Ferla and S. Pignataro, Surf. Interface Anal. 12, 447 (1988).
17. W. Reddish, Trans. Faraday Soc. 46, 459 (1950).
18. J. S. Abel, H. Mazurek and D. R. Dey, in: Metastable Materials Formation by Ion Implantation,
S. T. Picraux and W. J. Choyke (Eds), p. 173. Elsevier, New York, NY (1982).
19. S. R. Forest, M. L. Kalpan and P. H. Schmidt, Appl. Phys. Lett. 41, 708 (1982).
20. M. Gevers, Philips Res. Rep. 19, 447 (1946).
21. K. W. Wagner, Ann. Phys. 40, 817 (1913).
22. K. S. Cole and R. H. Cole, J. Chem. Phys. 9, 341 (1941).
23. T. Tohru, in: Polyimides: Fundamental and Applications, M. K. Ghosh and K. L. Mittal (Eds),
p. 7. Marcel Dekker, New York, NY (1996).
24. I. H. Loh, W. Oliver and P. Sioshansi, Nucl. Instrum. Meth. Phys. Res. B 343, 37 (1988).
25. J. Davenas, Appl. Surface Sci. 43, 218 (1989).
26. Z. Zhu, Y. Sun, C. Liu, J. Liu and Y. Jin, Nucl. Instrum. Meth. Phys. Res. B 193, 271 (2002).
27. M. Garg and J. K. Quamara, Nucl. Instrum. Meth. Phys. Res. B 246, 355 (2006).
TSDC Spectroscopy Investigation of Polyetherimide
B. K. Kaushik ∗ , Geetika Goyal and Randhir Singh ∗∗

Physics Department, National Institute of Technology, Kurukshetra-136119, India
Abstract
Thermally stimulated depolarization current (TSDC) technique has been recognized as a most suitable in-
vestigative tool in the realm of dielectric spectroscopy for polymeric materials. The technique provides
a deep insight into the molecular dynamics associated with the different relaxation processes. In the present
paper we discuss the various dielectric relaxation processes investigated in polyetherimide, a well-known
member of high performance polyimide family, using this technique. The samples were polarized at various
polarizing fields (Ep : 23–341 kV/cm) and polarizing temperatures Tp varying from 290 K to 573 K. The
TSD currents were recorded at a linear heating rate of 2 K/min. The TSDC spectra consist of four current
maxima namely γ , β, α and ρ appearing around 313 K, 383 K, 473 K and 503 K, respectively. The origin of
γ - and β-peaks has been attributed to the dipolar nature of ether linkages and carbonyl groups respectively,
whereas α- and ρ-maxima have been attributed to the space charge polarization process. However, in certain
cases, we observe merging of β- and α-relaxation processes to become the αβ-relaxation process appearing
around 413–453 K. The observed dependence of the peak temperature Tm on polarizing temperature Tp and
polarizing time tp indicates a continuous distribution of relaxations. The various relaxation parameters and
activation energy for different relaxations have been calculated.
Keywords
Polyetherimide, thermally stimulated depolarization current, pre-exponential factor, activation energy, dipo-
lar relaxation, dielectric loss
1. Introduction
Polyetherimide (PEI) with a high glass transition temperature is known to exhibit
outstanding mechanical and electrical properties [1, 2]. The PEI has found applica-
tions in the fields of microelectronics and aerospace industry, besides many other
industrial applications due to its heat and radiation resistance [2]. In view of its
wide-ranging applications, a deep understanding of its various dielectric relaxation
processes becomes highly imperative. To accomplish this, the thermally stimulated
discharge current technique is the most suitable investigative tool, as it provides
* To whom correspondence should be addressed. Tel.: +919416911871; Fax: 01744-238050; e-mail:
bkk_47@yahoo.co.in
** Physics Department, S. K. Institute of Engg. and Technology, Kurukshetra
214 B. K. Kaushik et al.
a deep insight into the molecular dynamics associated with the different relaxation
processes [3, 4]. The major advantage of this technique lies in the wide combination
of excitation conditions that can be used. The various current peaks appearing in the
TSD current spectrum of the polarized sample (termed as electret) are capable of
analyzing the various operative relaxation mechanisms through which an electret
stores its charge. The TSDC investigations on PEI have been reported by several
researchers [5–7]. However, a detailed TSDC analysis of PEI is yet to be done. The
present investigation was undertaken in an effort to augment the earlier studies by
extending the ranges of polarizing field and temperature as well as by introducing
some additional parameters, e.g., polarizing time, specimen thickness, storage time,
electrode material and heating rate.
2. Theory of TSDC
When a polarized dielectric (electret) is heated at a uniform rate, the decay of po-
larization P (t) is given as:
t
Nμ2 dt
P (t) = Ep exp − , (1)
3kTp 0 τ
where τ , N , μ, k and Ep are dipolar relaxation time, concentration of dipoles,
dipole moment, Boltzmann’s constant and applied field, respectively.
The temperature variation of relaxation time τ is given by the Arrhenius equation
[8] as:

U
τ = τ0 exp , (2)
kT
where τ0 is the pre-exponential factor or relaxation time at infinite temperature (the
reciprocal of which is known as the characteristic frequency) and U is the activation
energy.
The depolarization current density i is given as the rate of change of polarization
with respect to time, i.e.,
dP (t) P (t)
i(t) = − = . (3)
dt τ
Since, during the TSD experiment, the time and temperature both are varying si-
multaneously, a new variable T is introduced by assuming a simple temperature
program, most probably a linear one, by increasing the temperature starting from
an initial temperature T0 , so that:
T = T0 + ht, (4)
where h = dT /dt is the heating rate.
From (1), (2), (3) and (4) the TSD current density at a particular temperature T
is given as:
T
Nμ2 U 1 U
i(T ) = Ep exp − − exp − dT . (5)
3kTp τ0 kT hτ0 T0 kT
TSDC Spectroscopy Investigation of Polyetherimide 215
The relaxation time τ at any arbitrary temperature T can be calculated as:

Area under the TSDC curve between T0 and T
τ (T ) = . (6)
Value of TSD current at temperature T
Taking natural logarithm on both sides of (2) one obtains:
U
ln τ = ln τ0 + . (7)
kT
The slope of the straight line plot (ln τ versus 1/T ) obtained from (7) gives the
value of activation energy (U ).
3. Materials and Methods

Polyetherimide used in the present investigations was procured from Goodfellow
(U.K.) in a thin film form of nominal thickness d of 85 µm. The samples used
(0.45 cm in diameter) were metallized on both sides by vacuum evaporation of
aluminium to form good electrical contacts. The sample holder designed for this
purpose was suspended in a specially prepared temperature controlled furnace,
shielded against stray pickups and capable of heating the sample at some arbitrary
uniform rate from room temperature to nearly 573 K. The experimental setup has
been described in the literature [3].
The thermoelectrets were prepared by subjecting the samples to the desired dc
field (Ep ) at a constant temperature (Tp ) for a certain time (tp ) and subsequently
cooling them to room temperature within a fixed time in the presence of Ep . In order
to eliminate the frictional stray surface charges accumulated during the polarization,
the electrodes were short-circuited for about 10 min.
TSD currents were obtained by short-circuiting the sample through an electrom-
eter. Polarized samples were heated at a desired linear rate. For the study of storage
time effect (electret ageing), polarized samples were stored in a desiccator contain-
ing dry ambient air under non-shorted conditions for the predetermined time period
(ts ). Throughout the measurements the values of d, ts , tp and h were 85 µm, 0 h, 1 h
and 2 K/min respectively, unless mentioned otherwise. The Bucci plot method was
used for the evaluation of activation energy (U ).

Polyetherimide with the chemical structure as shown in Fig. 1 appears to have at
least two major relaxation processes (β- and α-relaxations) and a couple of weak re-
laxation processes. In addition we also encounter a relaxation process representing
the merging of α- and β-relaxations, termed as αβ-relaxations process. However,
a single TSDC characteristic is not capable to reveal all these relaxations simultane-
ously. The dominance of a particular relaxation process depends on the polarizing
parameters.
Figure 1. Chemical structure of polyetherimide.
TSDC characteristics in the temperature range 293–525 K for various polarizing

and storage conditions are illustrated in Figs 2–8. The number of peaks actually
appearing in these spectra and their location, height and sharpness are governed by
the forming, storage and depolarization parameters. The peaks appearing around
403 K and in the temperature range 453–463 K have been termed as β- and α-
peaks, respectively.
Figure 2(a) shows the TSDC characteristics of PEI samples polarized under iden-
tical conditions but different fields, ranging from 23 to 341 kV/cm. In each case, we
observe a well-defined peak around 423 ± 5 K. The peak temperature (Tm ) is almost
independent of Ep but the peak current (Im ) increases with increase in Ep . A linear
1/2
variation of Im with Ep is observed (inset in Fig. 2(a)). In certain cases, we also
observe a low temperature peak around 313 K (termed as β -peak). The sample
polarized at high Ep /Tp (Fig. 2(b)) shows a current maximum around 513–533 K.
The values of activation energy (U ) and pre-exponential factor (τ0 ) for these peaks
with Ep are given in Table 1.
The effect of Tp on the TSD current spectra of PEI samples polarized at 47 kV/cm
is shown in Fig. 3(a). All spectra show a well-defined current peak. However, the
location of the peak (Tm ) is significantly affected by the polarizing temperature. In
general, the peak shifts towards higher temperature on increasing the Tp . Interest-
ingly, for low values of Tp (333–373 K), the peak appearing around 403 K is almost
unaffected by the polarizing temperature. The peak magnitude increases with in-
creasing Tp . However, for higher values of Tp a decrease in peak current Im is also
noticeable (Fig. 3(b)).
Figure 4 exhibits the TSD current spectra of PEI samples corresponding to dif-
ferent values of tp , polarized at 47 kV/cm (Tp = 393 K). The increase in polarizing
time (tp ) results in slight enhancement in peak temperature (Tm ) and increase in
peak current magnitude (Im ).
Figure 5 illustrates the effect of heating rate (h) employed for depolarization on
the TSDC spectra of identically polarized samples of PEI. An increase in peak tem-
perature (Tm ) as well as peak current (Im ) is observed with increase in h. However,
the initial slope is almost unaffected.
Figure 6 illustrates the effect of storage time, ts , on the TSD current behaviour
of PEI samples polarized and depolarized under similar conditions. An increase in
ts has been observed to suppress the whole TSDC envelope. Though we do not
find any shift in the peak position, the peak broadens and its magnitude reduces
considerably with increasing ts . In case of long storage times, the discharge current
spectra exhibit a striking feature as directional reversal (anomalous TSDC).
(a)
(b)
Figure 2. (a) TSDC characteristics of PEI samples, at Tp = 393 K, polarized with different Ep . In-
1/2
set: Peak current ‘Im ’ versus square root of polarizing field ‘Ep ’. (b) TSDC characteristics of PEI
samples, at Tp = 473 K, polarized with different Ep .
The β-peak (the peak appearing around 403 K) in amorphous polymers mainly
arises from localized rotational fluctuations of the dipoles and, therefore, it is also
referred as dipolar relaxation process [9]. The molecular mechanism of β-relaxation
has been discussed by many groups [10–13]. In PEI the β-relaxation can be asso-
ciated with the electric dipole orientation of the large number of carbonyl groups
(>C=O) present in its structure. Though we do not observe a single isolated peak
(a)
(b)
Figure 3. (a) TSDC characteristics of PEI samples, at Ep = 47 kV/cm, polarized with different Tp .
(b) TSDC characteristics of PEI samples, at Ep = 176 kV/cm, polarized with different Tp .
fully devoted to this relaxation, the occurrence of this relaxation can be judged from
certain characteristics observed for this peak such as (i) the activation energy asso-
ciated with this peak (Table 2) is very close to the theoretically predicted value of U
Figure 4. TSDC characteristics of PEI samples, at Ep = 47 kV/cm and Tp = 393 K, polarized with
different tp .
Figure 5. TSDC characteristics of PEI samples, at Ep = 47 kV/cm and Tp = 393 K, depolarized with
different h.
(≈0.22 eV) for the side group orientation in polymers [11], (ii) a shift in peak tem-
perature (Tm ) towards higher temperature with increasing Tp for Tp < Tg (Fig. 3(a)).
Interestingly, the shift in the peak location with increasing Tp for Tp > 373 K is very
Figure 6. TSDC characteristics of PEI samples, at Ep = 47 kV/cm and Tp = 393 K, stored for differ-
ent ts .
Figure 7. Variation in dielectric loss with temperature at a frequency of 120 Hz.

Figure 8. TSPC characteristics of ULTEM samples under different polarizing fields (12, 35 and
47 kV/cm).
Table 1.
Activation energy and relaxation time for αβ-relaxation (Tp = 393 K,
tp = 3.6 × 103 s)
Polarizing field Activation energy Peak temp. Relaxation time

Ep (kV/cm) U (eV) (K) τ0 (s)
23 0.35 423 9 × 10−4

47 0.29 428 3.5 × 10−2
70 0.36 438 1.64 × 10−2
129 0.36 428 2.5 × 10−2
270 0.30 413 1.3 × 10−2
341 0.34 423 1.2 × 10−2
large. We observe a shift in Tm of the order of 363 K. Such a large shift in Tm with
increasing Tp cannot be attributed to the dipolar relaxation process alone. We may
ascribe this shift in the Tm to the onset of another relaxation process. We know from
the earlier studies that low Ep /Tp is usually favourable for β-relaxation, whereas
high Ep /Tp is favourable for α-relaxation [14]. It is reasonable to attribute peaks
appearing in the high temperature region (in the temperature range 453–463 K) to
space-charge relaxation (α-relaxation) process. The space-charge formation occurs
via interfacial polarization (the Maxwell–Wagner–Sillars effect) [15]. In PEI, the
material is almost amorphous in nature and, therefore, the possibility of interfa-
cial polarization is minimal. The glass transition temperature (Tg ) of PEI is around
Table 2.
Activation energy and relaxation time for β-relaxation (Ep = 47 kV/cm,
tp = 3.6 × 103 s)
Polarizing temp. Activation energy Peak temp. Relaxation time

(K) U (eV) (K) τ0 (s)
333 0.31 373 9.76 × 10−2

353 0.23 398 6.2 × 10−2
373 0.28 403 5.76 × 10−2
Table 3.
Activation energy and relaxation time for α-relaxation (Ep = 47 kV/cm,
tp = 3.6 × 103 s)

(K) U (eV) (K) τ0 (s)
433 0.34 453 2.17 × 10−1

443 0.36 453 1.14 × 10−1
453 0.35 463 1.98 × 10−1
488 K [2]. It will be very reasonable to associate the α-peak with the glass tran-
sition temperature of PEI. The space-charge origin of this peak is also confirmed
1/2
by the following facts: (i) its peak current (Im ) varies linearly with Ep (inset in
Fig. 2(a)), (ii) during the electret storage, the peak vanishes almost completely dur-
ing long storage time, (iii) its peak magnitude depends on the nature of electrode
material [16], the dependence is mainly due to the different degrees of blocking ow-
ing to the difference in their work functions and (iv) the average activation energy
U for this peak (0.34 eV; Table 3) does not differ much from the activation energy
value reported for α-relaxation in Kapton-H polyimide [17]. Further, the unsatu-
rated carbonyl side groups in PEI can act significantly as charge trapping centers
contributing to their transport and trapping [11].
Usually the TSDC characteristics are quite capable of resolving the β- and
α-relaxations in polymers. However, in some polymeric systems there exists a tem-
perature region where the characteristic relaxation time for the β-relaxation process
reaches a similar order of magnitude as that of the primary α-relaxation and thus we
may observe merging of α- and β-processes to become the αβ-relaxation process
[18]. The occurrence of αβ-relaxation is more significant and dominating if the
same molecular mechanism is responsible for both α- as well as β-relaxations. As
in case of PEI, the carbonyl groups, on the one hand, due to their rotational mo-
tion, are responsible for β-relaxation and, on the other hand, being charge trapping
centers become responsible for α-relaxation. The TSD current maximum appearing
Table 4.
Activation energy and relaxation time for αβ-relaxation (Ep =
47 kV/cm, tp = 3.6 × 103 s)

(K) U (eV) (K) τ0 (s)
393 0.29 418 3.7 × 10−1

403 0.30 433 4.5 × 10−1
around 413–433 K can be associated with the αβ-relaxation. These peaks appear
to have the hybrid character, i.e., they have certain characteristics of β- and certain
characteristics of α-relaxation. We observe a shift in the peak temperature (Tm ) with
Tp , which is a β-relaxation characteristic and at the same time the peak magnitude
Im varies non-linearly with Ep which is a α-relaxation characteristic. The activa-
tion energy associated with these peaks is almost equal to the average of activation
energies for α- and β-relaxations combined (Tables 1 and 4). Heating rate is an im-
portant parameter, which can be helpful in deciding the mode of relaxation. Increase
in peak intensity Im and rise in Tm are directly related to the increasing heating
rate. This is mainly owing to the fact that the polymer responds less promptly to
the higher heating rates and initial polarization has to be released in a shorter time
[19]. In the present case, a significant rise in Tm with increasing sharpness shows
the presence of a distributed αβ-relaxation process in PEI.
The high temperature TSD current maximum appearing around 530 K, the so
called ρ-peak (Figs 2(b) and 3(b)) is found to emerge only for high values of
Ep /Tp . The high Ep /Tp values are favourable for injection of charge carriers from
the electrodes into the polymer directly or through Townsend breakdown [20]. In
the present case since the vacuum deposited electrodes were used for electrical con-
tacts, the possibility of Townsend breakdown can be ruled out [11]. The value of
activation energy U estimated for this relaxation (≈1.38 eV) is comparable to the
energy generally associated with the migration of ions [14]. Thus, the charge injec-
tion phenomenon might be responsible for this peak as the charge carriers injected
into the polymer are ionic in nature [21]. The dependence of Tm on Tp indicates the
existence of a distributed polarization. In a dielectric relaxation process, the peak
temperature is given as [8]:
1/2
U hτ0 U
Tm = exp , (8)
k kTm
where τ0 , U , h and k are pre-exponential factor, activation energy, heating rate and
Boltzmann’s constant, respectively.
The above equation shows that in case of a single relaxation time, Tm should
be independent of the polarizing temperature Tp for a fixed heating rate. However,
in low temperature region only the fast sub-polarizations are operative but as Tp
increases, more sub-polarizations with longer relaxation times become activated.

This results in shifting of the current maximum towards higher temperature side
and also enhancing the peak magnitude. The shift of the peak temperature with
varying heating rate employed during depolarization also affirms the involvement of
a distribution in relaxation frequencies. The association of continuous distribution
of relaxation times with β-, α- or αβ-relaxations can also be observed from the fact
that both activation energy U and pre-exponential factor τ0 are distributed (Table 1).
In the low temperature region, the TSD currents show the presence of a weak
relaxation process termed as β -relaxation process. Since the onset of a dipolar
relaxation process occurs at low temperatures, the β -relaxation can be considered
as a satellite of the main dipolar peak (β-peak). The origin of β -relaxation can be
associated with the dipolar nature of the ether linkages in PEI. The weak nature of
this relaxation process can be ascribed to the fact that the ether linkage is present in
the main chain and its response towards the external field is not as strong as in case
of carbonyl groups. The occurrence of β -relaxation associated with ether linkages
has also been shown in poly-p-phenylene sulphide by Quamara et al. [12]. The tan δ
loss peak appearing around 323 K in dielectric loss (ε ) versus T characteristics
(Fig. 7) is in conformity with the presence of β -relaxation. The TSP current spectra
(Fig. 8) also confirm some of these relaxation processes, particularly the dipolar
relaxation due to ether linkages.
5. Conclusions
From the present investigation we can draw the following conclusions:
(1) Polyetherimide shows many relaxation processes associated with its molecular
dynamics.
(2) The various current maxima are due to dipolar orientation, space charge polar-
ization, and trapping of charges injected into electrodes.
(3) TSP current spectra are in conformity with the above findings (Fig. 8).
References
1. J. A. Brydson, Plastics Materials, 6th edn, p. 508. Butterworth-Heinemann (1995).
2. W. I. Serfaty, in: Polyimides: Synthesis, Characterization and Applications, K. L. Mittal (Ed.),
Vol. 1, pp. 149–161. Plenum Press, New York, NY (1984).
3. J. van Turnhout, in: Topics in Applied Physics: Electrets, G. M. Sessler (Ed.), Vol. 33, p. 81.
Springer-Verlag, Berlin (1980).
4. J. Vanderschuren and J. Gasiot, in: Topics in Applied Physics: Thermally Stimulated Relaxation in
Solids, P. Braunlich (Ed.), Vol. 37, p. 135. Springer-Verlag, Berlin (1979).
5. M. Mudarra, J. Belana and R. Diaz-Calleja, in: Proc. of 10th International Symposium on Elec-
trets, pp. 71–74 (1999).
6. J. Belana, J. C. Canadas and J. A. Diego, Polymer Intl. 46, 29–32 (1998).
7. B. L. Sharma and J. K. Quamara, in: Proc. 5th International Symposium on Electrets, pp. 241–246.
Heidelberg (1985).
8. J. Vanderschuren and J. Gasiot, in: Topics in Applied Physics: Thermally Stimulated Relaxation in
Solids, P. Braunlich (Ed.), Vol. 37, pp. 145–146. Springer-Verlag, Berlin (1979).
9. A. Schonhals, in: Broadband Dielectric Spectroscopy, F. Kremer and A. Schonhals (Eds), p. 242.
10. M. Garg and J. K. Quamara, Nucl. Instrum. Meth. Phys. Res. B 246, 355–363 (2006).
11. R. P. Bhardwaj, J. K. Quamara and K. K. Nagpaul, Phys. Stat. Sol. (a) 77, 347 (1983).
12. J. K. Quamara, N. Singh and A. Singh, Macromol. Chem. Phys. 202, 1955–1960 (2001).
13. J. P. Runt and J. J. Fitzgerald, Dielectric Spectroscopy of Polymeric Materials, p. 86. American
Chemical Society, Washington, DC (1997).
14. T. Prabha and J. K. Quamara, Radiation Effects & Defects in Solids 160, 187–195 (2005).
15. M. Garg, S. Kumar and J. K. Quamara, Indian J. Pure Appl. Phys. 39, 261 (2001).
16. J. van Turnhout, in: Topics in Applied Physics: Electrets, G. M. Sessler (Ed.), Vol. 33, p. 100.
17. M. Garg and J. K. Quamara, Nucl. Instrum. Meth. Phys. Res. B 179, 87 (2001).
18. A. Arbe, J. Colmenero and D. Richter, in: Broadband Dielectric Spectroscopy, F. Kremer and
A. Schonhals (Eds), p. 707. Springer-Verlag, Berlin (2003).
19. J. L. Gil-Zambrano and Juhasz, J. Phys. D 14, 1661 (1981).
20. B. Gross, J. Chem. Phys. 17, 866 (1949).
21. H. Geiger and K. Scheel, Handbook of Phyics, Vol. 23. Springer-Verlag, Berlin (1943).
Space-Survivability and Characterization of Main-Chain
and Side-Chain POSS–Kapton® Polyimides
S. J. Tomczak a,∗ , M. E. Wright b , A. J. Guenthner b , B. J. Petteys b , T. K. Minton c ,

A. Brunsvold c , V. Vij d , L. M. McGrath d and J. M. Mabry a
a
Air Force Research Laboratory, Propulsion Materials Application Branch,
Edwards AFB, CA 93524, USA
b
Research & Engineering Sciences Department, Chemistry Division, NAVAIR-US Navy,
China Lake, CA 93555-6100, USA
c
Department of Chemistry and Biochemistry, Montana State University, Bozeman, MT 59717, USA
d
ERC Incorporated, Propulsion Materials Application Branch, Edwards AFB, CA 93524, USA
Abstract
Kapton® polyimide is an extensively used space material in solar arrays, spacecraft thermal blankets and
space inflatable structures. Upon exposure to atomic oxygen (AO) in low Earth orbit (LEO), Kapton®
is severely degraded. An effective approach to prevent this erosion is chemically bonding polyhedral
oligomeric silsesquioxane (POSS) into the polyimide matrix by copolymerization of POSS–diamine with
the polyimide monomers. POSS is a silicon and oxygen cage-like structure surrounded by organic groups
which can be polymerizable. The copolymerization of POSS provides Si and O in the polyimide matrix on
the nano-level. X-ray photoelectron spectroscopy (XPS) studies on POSS–Kapton® polyimides exposed to
AO from a hyperthermal AO beam generated in ground-based facilities, and in LEO on a 3.9 year flight on
the Materials International Space Station Experiment-1 (MISSE-1), show that a silica surface layer forms
on POSS–Kapton® polyimide films in AO-rich environments. After exposure to a fluence of 8.5 × 1020 O
atoms/cm2 in simulated LEO facility, erosion yields of 3.5 and 7 wt% Si8 O11 main chain (MC) POSS–
Kapton® samples were 3.7 and 0.98 percent, respectively, of the erosion yield for commercial Kapton H® .
MC- and SC-POSS–Kaptons® were found to have similar physical properties and comparable AO-resistance
in a simulated LEO environment.
Keywords
Hyperthermal atomic oxygen, polyimide, Kapton, polyhedral oligomeric silsesquioxane, POSS, space envi-
ronment, polymer erosion, low Earth orbit
*
To whom correspondence should be addressed. Tel.: (661) 275-5171; Fax: (661) 275-5171; e-mail:
sandra.tomczak@edwards.af.mil
228 S. J. Tomczak et al.
1. Introduction
Over the past decade, there has been increasing interest in the creation of POSS–
polyimide nano-composites for a variety of applications [1] that include low di-
electric constant materials [2–4], thermally curable aryl–ethynyl end-capped hybrid
composite resins [5–7], polyimides with improved thermal and mechanical proper-
ties [8], and thermally stable hyperbranched-polyimides [9]. It has been well es-
tablished through space flight experiments and ground-based simulations of atomic
oxygen (AO) in low Earth orbit (LEO) that polymeric materials undergo severe
degradation as a result of the aggressive conditions in LEO [10–26] where materials
are subjected to solar radiation, thermal cycling from −50◦ C to 150◦ C, bombard-
ment by low- and high-energy charged particles, as well as high incident fluxes
of AO [10–15, 23, 24]. A moving spacecraft encounters AO that has a kinetic en-
ergy of about 4.5–6 eV [24–26], resulting in the continual degradation of organic
polymeric materials if not protected by an inert inorganic coating.
To explore alternative methods for protecting Kapton® , and other polymers, for
space applications, polyhedral oligomeric silsesquioxane (POSS) nanostructures
have been incorporated into polyimides to impart remarkable resistance to AO
erosion [27–35]. POSS represents a unique family of nanoscale inorganic/organic
hybrid cage structures that contain a silicon/oxygen framework that is intermediate
(RSiO1.5 ) between silica (SiO2 ) and silicone (R2 SiO) [33, 36–39]. The Si/O cage is
surrounded by, and chemically bonded to, organic groups which can be tailored for
compatibility with polymer matrices or for polymerization into polymer backbones.
Kapton® type polyimide (0% POSS–Kapton® ) and POSS–Kapton® samples
were synthesized and cured in the same conditions. 0% POSS–Kapton and MC-
POSS–Kapton® were exposed to all aspects of LEO on the International Space
Station for 3.9 years. 0% POSS–Kapton® completely eroded within 4 months,
while a thin film of 3.5 wt% Si8 O11 MC-POSS–Kapton® remained after the flight,
demonstrating the space-survivability of POSS–Kapton® .
Through simulated LEO environments utilizing a hyperthermal O-atom beam,
we have, thus far, found that copolymerized POSS imparts approximately the same
degree of AO-resistance to Kapton® and other polyimides, regardless of differences
in POSS structure and distance of the POSS cage from the polyimide backbone.
Our previous work with POSS-modified fluorinated polyimides included the syn-
thesis of a POSS graft copolymer, which can be stored as a powder, dissolved in
solvent, and successfully sprayed to form an AO-resistant film. A visibly transpar-
ent film was achieved with 36 wt% POSS (17 wt% Si8 O12 ). This film exhibited
no detectable erosion, no changes in coloration, and no development of opacity, af-
ter exposure to 2.44 × 1020 O atoms/cm2 , whereas the Kapton H® witness sample
eroded 7.32 µm [35]. It is reasonable to project that the POSS-containing material
would continue to resist erosion, while Kapton H® would continue to erode, in an
extended AO exposure.
Wright invented a new Side-Chain (SC) POSS–diamine [40], which was sub-
sequently copolymerized to form SC-POSS–Kapton® polyimide. As will be dis-
Space-Survivability and Characterization of MC- and SC-POSS–Kapton® Polyimides 229
cussed in the following, the SC-POSS polyimides, relative to 0% POSS–Kapton®

polyimide, display minor effects on the storage modulus, glass transition temper-
ature, coefficient of thermal expansion, and successfully impart remarkable AO
resistance to the polymer system. Our data indicate that upon AO exposure, the
organic material in the POSS–polyimide film surface erodes with concomitant re-
action of the nano-dispersed POSS with AO to form an effective passivation layer of
silica. The environmentally dependent generation of a protective layer in ‘response’
to material damage that can occur during flight makes applicable the terms ‘smart
material’, ‘self-healing’, or ‘auto-repsonsive’ in referring to POSS–Kapton® . In this
work, we present the properties of SC-POSS–Kapton® that include coefficient of
thermal expansion (CTE) before and after AO exposure, mechanical properties and
AO-erosion data. The SC-POSS monomer synthesis has been scaled up to a 100 g
synthesis, which was carried out in two days, by a highly efficient synthesis route.
The sudden temperature changes experienced by materials in the LEO envi-
ronment make the CTE an important material property. Mismatches in the CTE
between polymeric materials and their coatings lead to cracks, crazing and me-
chanical material failure. Kapton® is often coated with silica to resist AO, and may
be coated with metal oxides to prevent the effects of electrostatic charge differen-
tials on spacecraft surfaces. Multilayer insulation blankets for spacecraft include
Kapton® coated with aluminum on the underside and silica on the top surface.
These layered systems increase the importance of the knowledge of CTE values
of POSS–Kaptons® , and their similarity to values for Kapton® materials currently
used in spacecraft systems.
Mechanical properties, such as storage modulus, loss modulus, and glass tran-
sition temperature, indicate the effect of POSS as a nano-filler on the polymer
matrix. POSS dispersion, or aggregation, can be witnessed through transmission
electron micrographs and mechanical property investigations. POSS dispersion is
anticipated to affect the homogeneous formation of the silica layer on the surface
of POSS materials during AO exposure.
To directly compare the effect of AO on Kapton H® , SiO2 -coated Kapton HN® ,
and 8.75 wt% Si8 O11 cage MC-POSS–Kapton® , these materials were exposed to
AO, scratched, and exposed to a second equivalent AO fluence. The erosion of the
three materials, inside and outside of the scratched area, was monitored by stylus
surface profilometry. The results of this study indicate that the 8.75 wt% Si8 O11
MC-POSS–Kapton® reproducibly eroded about 200 nm before forming a silica
layer [28].
MC-POSS–diamine and SC-POSS–diamine were separately copolymerized with
the Kapton® monomers without changes to the Kapton® polymerization proce-
dure, resulting in the POSS–Kaptons® shown in Figs 1 and 2. The SC-POSS
was synthesized using the SC-POSS–diamine, N-{3-(iso-butyl7 Si8 O12 )propyl}-
3,5-diaminobenzamide. Approximately the same wt% POSS cage was used for
both MC-POSS and SC-POSS containing polymers. The amounts were inher-
ently approximate because MC-POSS is a Si8 O11 structure, while SC-POSS is a
Figure 1. MC-POSS–Kapton® structure demonstrating the PMDA-MC-POSS repeat unit and the
PMDA–ODA repeat unit, R — cyclopentyl.
Figure 2. SC-POSS–Kapton® structure demonstrating the PMDA-SC-POSS repeat unit and the
PMDA–ODA repeat unit, R — iso-butyl.
Si8 O12 structure. This means that less SC-POSS monomer can be used, relative to
MC-POSS, to deliver the equivalent amount of POSS cage into the polymer matrix.
In reference to previous work [28–34, 41, 42], 3.5 and 7 wt% POSS cage equals
about 10 and 20 wt% MC-POSS diamine monomer and 8.5 and 17 wt% SC-POSS
diamine monomer, respectively.
2. Experimental
2.1. Synthesis of Side-Chain POSS Monomers
N-{3-(iso-butyl7 Si8 O12 )propyl}-3,5-diaminobenzamide was synthesized from
3-aminopropyl-hepta(iso-propyl)POSS in excess of 95% yield. The synthesis
of SC-POSS and a second SC-POSS with cyclohexyl R-groups, N-{3-(cyclo-
hexyl7 Si8 O12 )propyl}-3,5-diaminobenzamide), has been described in detail [40].
2.2. Preparation of Kapton® Equivalent

Kapton® is a registered trademark of DuPont. The monomers used in this study are
the same as for Kapton® . Processing and additives cause variation from DuPont
produced Kapton® .
Polymerizations were carried out in a N2 filled dry box by the following general
procedure: A 50 ml RB-flask was charged with 4,4 -oxydianiline (ODA) (0.918 g,
4.59 mmol) and N,N-dimethylacetamide (DMAc) (2 ml). A separate 25 ml RB-flask
was charged with pyromellitic dianhydride (PMDA) (1.00 g, 4.59 mmol) and
DMAc (1 ml) to form a light yellow slurry. The latter slurry was added drop-wise to
the ODA solution with stirring. Additional DMAc (2 ml aliquots) was used to wash
the PMDA flask and complete the transfer to the ODA reaction flask. This was re-
peated until a total of 20 ml DMAc was added to the polymerization reaction. The
addition of PMDA to ODA was typically over a period of 30 min. The poly(amic
acid) solution became viscous upon final PMDA addition and was allowed to react
for 16 h with stirring at ambient temperature. The poly(amic acid) DMAc solution
was stored under nitrogen in a freezer and was found to produce high-quality films
even after 5 years of storage.
2.3. POSS/PMDA/ODA Copolymerization
In a N2 filled glovebox, a 25 ml RB-flask was charged with SC-POSS (N-{3-iso-
butyl7 Si8 O12 )propyl}-3,5-diaminobenzamide) (0.3749 g, 0.3719 mmol), a stir bar,
and DMAc (2 ml) and the mixture allowed to stir until homogeneous. A second
50 ml RB-flask was charged with ODA (0.8437 g, 4.21 mmol) and a stir bar. The
POSS/DMAc solution was transferred by syringe into the ODA and was stirred.
PMDA (1.00 g, 4.59 mmol) was weighed into a 25 ml RB-flask, and transferred
drop-wise by syringe, as a slurry in DMAc, into the diamine mixture over 30 min.
A total of 20 ml DMAc was added to the polymer solution. The POSS-poly(amic
acid) solution became moderately viscous, and was stirred overnight at room tem-
perature. The POSS-poly(amic acid) was stored under nitrogen in a freezer, and was
found to produce high-quality films even after 2 years of storage. This storage time
is due to the limited history of SC-POSS. Comparative MC-POSS poly(amic acid)
produced high-quality films even after 5 years of these storage conditions and it is
very possible that other POSS monomers will behave similarly.
2.4. Polyimide Film Fabrication
The poly(amic acid) solution, or POSS-poly(amic acid) solution, with approxi-
mately 2 wt% solids, was poured on a clean glass plate which was balanced in a
vacuum oven, equipped with a N2 flow. The DMAc solvent was evaporated at 80◦ C
for 4 h, followed by a cure schedule of 120◦ C for 1 h, 200◦ C for 1 h and 275◦ C for
2 h. The amber colored and visibly transparent films were removed from the slide
by placing them under running deioinized water, lifting the film corner with a razor
blade, and running water between the glass and the film. The films were stored in
air.
2.5. Low Earth Orbit Exposure of POSS–Polyimides

MC-POSS–Kaptons® containing 0, 1.75 and 3.5 wt% Si8 O11 , were flown in LEO
on the outside of the International Space Station as part of the MISSE-1 [43]. The
samples were flown in Passive Experiment Container 1 (PEC 1), in Tray 1, de-
veloped by NASA Langley Research Center (LaRC), and were assigned numbers
38, 39, 40. Prior to flight, the samples were held into place in the tray by an alu-
minum top plate with open circles to expose the samples. The top plate also protects
the outer area from exposure to the space environment. The flight experiment was
launched on August 10, 2001 and was retrieved after 3.9 years. The samples were
exposed to the RAM and therefore to all components of the LEO environment,
including AO and UV light. The step-height difference between the unexposed sam-
ple area and the neighboring exposed sample area was measured in twelve locations
using profilometry. The atomic composition of sample surfaces was determined by
XPS [25]. Free-standing films of MC-POSS–Kaptons® containing 0, 1.75, 3.5, 5.25
and 7.0 wt% Si8 O11 were sewn to a Kapton® blanket and flown in LEO on MISSE-5
for about one year. The AO fluence was estimated at 1.8 × 1020 atoms/cm2 , based
on the Kapton H® erosion rate during this flight.
2.6. Exposure of POSS–Kapton® Polyimides to Atomic Oxygen
Ground-based AO exposures of POSS–Kapton samples were performed with a
pulsed AO beam, operating at a repetition rate of 2 Hz and containing O atoms
that were generated by the laser-induced breakdown of O2 gas in a conical nozzle
with the use of a 7-Joule-per-pulse CO2 laser [8, 26, 31]. The hyperthermal beam
contains neutral O atoms and molecular oxygen, with an ionic species component
of <0.01%. The mole fraction of AO in the beam was above 70% and, for some
exposures, above 90%. Kinetic energies of the fast O atoms in the beam averaged
5.2 eV, with a full width at half maximum in the distribution of ∼2 eV. Prior to
exposure, samples were covered with a stainless steel mesh in order to mask ar-
eas and achieve both AO exposed and unexposed areas. The O-atom fluences of
the exposures varied depending on exposure duration, but a common fluence was
∼2.7 × 1020 O atoms/cm2 for all experiments. All samples were handled in ambient
air after exposure and prior to surface analysis, including etch depth determination
(by profilometry) and other surface topography (AFM, SEM and TEM) and surface
chemistry measurements.
Three samples each of Kapton H® , 8.75% Si8 O11 MC-POSS–Kapton® , and
silica-coated Kapton HN® (provided by Astral Technology Unlimited, Inc., USA,
Lot No. 00625-007, with a 130 nm SiO2 coating) were exposed to 100k pulses of the
beam, along with a Kapton H® standard. After the first exposure, the etch depths of
the screened samples were measured. A set of unscreened samples also underwent
surface measurements. The third set of samples was scratched. For each sample,
one approximately 40 µm wide and 1 µm deep scratch was made with a diamond-
tipped scribe, and two approximately 20 µm wide and 1 µm deep scratches were
made with a razor blade. All scratches were measured by profilometry (with a 5 µm
radius probe tip). Screens were placed over the scratched samples and these samples
were exposed to additional 100k shots of the hyperthermal AO beam. Profilometry
was used to measure step height differences between exposed and unexposed areas
and to profile each scratch in AO exposed and unexposed areas.
2.7. Physical Property Characterization of POSS–Kapton® Polyimides
Polyimide samples were analyzed using a DMTA V from TA Instruments with a
5◦ C/min temperature ramp from room temperature to 500◦ C in tensile mode. CTE
measurements were taken on a thermomechanical analyzer (TMA 2940) from TA
Instruments with a film fiber attachment in a nitrogen atmosphere. The cured films
were cut into 15 mm by 3 mm samples, the force applied was 0.05 N and 0.10 N,
and the sample was heated at 5◦ C/min. The CTE was calculated as α = (L ×
K)/(L × T ), where L — length, K — cell constant, T — temperature in ◦ C.
Test variability was ±2.306 ppm/◦ C based on five tests on Kapton H® .
2.8. Surface Characterization of POSS–Kapton® Polyimide Films
XPS data were obtained using a non-monochromatic Mg Kα radiation (1253.6 eV)
and a hemispherical CLAM 2 (VG Microtech) analyzer. Scanning electron mi-
croscopy (SEM) was performed using an ISI CL6 operating at 15 keV equipped
with a Kevex X-ray detector, at Edwards AFB. SEM images of MISSE-1-flown
samples were acquired similarly at Montana State University.
TEM images were obtained on either a Hitachi H7600 transmission electron mi-
croscope at 120 kV or a Hitachi HD2000 STEM at Clemson University Electron
Microscope Facility.

To compare the present polymers with previously synthesized MC-POSS–
Kaptons® , approximately the same POSS cage composition was used in SC-POSS–
Kaptons® . All aspects of film fabrication, including cure cycle, were consistent
with the protocol for Kapton® . In comparison to previous work [28–34] with
MC-POSS–Kaptons® , greater amounts of SC-POSS–Kapton® were copolymer-
ized before opacity was visible to the eye. 10.5 wt% Si8 O11 MC-POSS–Kapton®
was completely opaque with visible white aggregates and shattered during cure.
Conversely, 10.5 wt% Si8 O12 SC-POSS–Kapton® was transparent with areas con-
taining visible opaque streaks. 12.3 wt% and 14 wt% Si8 O12 SC-POSS–Kaptons®
have increasing brittleness and areas of visible opaque streaks, but also have trans-
parent flexible areas that mimic the color and appearance of Kapton® . This implies
that fine-tuning of film fabrication may lead to transparent and flexible films with
higher SC-POSS loadings than achieved with MC-POSS.
As shown in the following, the dynamic and mechanical thermal properties of
MC- and SC-POSS–Kaptons® have insignificant differences from Kapton® in the
temperature range experienced in LEO (−50 to 150◦ C). However, fast temperature
Table 1.
Coefficients of thermal expansion for Kapton® and POSS–Kapton® films
Sample a CTE (µm/m◦ C)
Kapton H® 30.2
0 wt% POSS–Kapton® 33.1
7 wt% Si8 O11 MC-POSS–Kapton® 33.5
8.75 wt% Si8 O11 MC-POSS–Kapton® 35.0
7 wt% Si8 O12 SC-POSS–Kapton® 35.9
7 wt% Si8 O12 SC-POSS–Kapton® exposed to AOb 33.6
a Test variability based on 5 Kapton H® tests = ±2.306 ppm/◦ C.

b 2.3 × 1020 oxygen atoms/cm2 .
changes which occur in the space environment give importance to the differences
in the CTE between polymer films and their coatings. The CTE for several samples
are shown in Table 1. The addition of POSS slightly increases the CTE of Kapton®
polyimide, and a slight decrease is seen after exposure of the SC-POSS–Kapton®
to AO. The CTE of fused silica is about 0.55 µm/m◦ C [44]. The comparison of this
value to the data in Table 1, approximates the CTE difference between the POSS–
Kapton® polymer matrix and the silica passivation layer formed in the presence of
AO. Whether the silica layer is deposited, as with state-of-the-art space materials,
or self-generated as with POSS–polyimides, this mismatch between the silica pas-
sivation layer and the underlying-Kapton® is likely to cause cracking and flaking
of the silica layer. If this occurs with POSS–polyimides, a fresh silica passivation
layer is formed in the areas where cracks or damage have caused the loss of the
previously formed silica. This silica regeneration process continually protects the
underlying POSS–Kapton® polymer matrix from reactions involving AO.
The physical properties of MC-POSS–Kaptons® were evaluated by DMTA and
have been discussed previously [27, 28, 30]. All Tg (glass transition temperature)
values, except for 8.8 wt% MC-POSS–Kapton® , were slightly lower (5–10%) than
those of the equivalent SC-POSS–Kaptons® . The tan δ plots were symmetrical,
and as the wt% of the Si8 O11 MC-POSS cage increased as: 0, 3.5, 7.0 and 8.75, the
Tg s decreased: 420, 394, 390 and 383◦ C, respectively. The 7% Si8 O11 MC-POSS–
Kapton® exhibited a tan δ peak very similar in intensity and breadth to the 0%
POSS–Kapton® peak, while the 3.5 and 8.75 wt% Si8 O11 MC-POSS–Kapton® tan
δ peaks were slightly greater than half the height, but of similar width, to the 0
and 7 wt% POSS samples. The behavior of the 7 wt% Si8 O11 MC-POSS–Kapton®
may be due to having chain packing similar to the 0% POSS–Kapton® sample. The
storage modulus of the MC-POSS–Kaptons® decreased only very slightly relative
to the 0% POSS–Kapton® , below the Tg .
The physical properties of SC-POSS–Kaptons® were evaluated by DMTA as
shown in Fig. 3. As the wt% of the Si8 O12 SC-POSS cage increased as: 0, 1.75,
Figure 3. Dynamic mechanical thermal analysis of 0, 1.75, 3.5, 7 and 8.8 wt% Si8 O12
SC-POSS–Kapton samples.
3.5, 7 and 8.8, the Tg s decreased as: 407◦ C, 403◦ C, 406◦ C, 392◦ C and 365◦ C,
respectively. It is likely that increasing the amount of the bulky SC-POSS increases
the free volume in the polymer matrix, and decreases the areas of the strong charge
complex formation between carbonyls and nitrogen groups of adjacent polyimide
backbone chains.
To understand better the effect of POSS on the polyimide matrix interactions,
a series of POSS–Kaptons® were subjected to an extended hold at 275◦ C for 16 h.
The storage modulus and tan δ curves versus temperature are shown in Fig. 4.
The Tg increased by about 4◦ C for 7.0 and 8.8 wt% SC-POSS–Kapton® after the
extended cure period, while the storage modulus and tan δ curves for the 0% POSS–
Kapton® remained unchanged.
In our study, the films were solvent-cast by pouring the POSS-poly(amic acid)
solution onto a glass plate with no blade casting or forced directionality. Atomic
Force Microscopy (AFM) of 0, 1.75, 3.5, 5.25 and 7 wt% Si8 O11 MC-POSS–
Kaptons® shows variations in surface morphology with increasing POSS loading.
3.5 and 5.25 wt% Si8 O11 MC-POSS–Kaptons® repeatedly demonstrate the unique
morphology with raised and depressed regions as seen in Fig. 5. Energy Disper-
sive X-Ray (EDX) analysis of three elevated and three depressed regions in each
of 3.5 and 5.25 wt% Si8 O11 MC-POSS–Kaptons® showed the same silicon content
throughout all areas of the film. A smoother surface morphology is present with the
7 wt% Si8 O11 MC-POSS–Kapton® films.
Scanning Electron Micrographs (SEMs) of AO exposed 7 wt% Si8 O11 MC-
POSS–Kapton® are seen in Fig. 6. Elevated square regions of roughly 3 by 3 µm
are present, and AO-erosion cavities are seen at the periphery of the square regions.
Figure 4. Dynamic mechanical thermal analysis of 0, 7 and 8.8 wt% Si8 O12 SC-POSS–Kapton®
samples with final cure at 275◦ C for 2 h (0 wt% POSS, 7.0 wt% SC-POSS and 8.8 wt% SC-POSS)
and at 275◦ C for an additional 16 h (0 wt% POSS 275◦ C, 7.0 wt% SC-POSS 275◦ C and 8.8 wt%
SC-POSS 275◦ C).
(a) (b) (c)
(d) (e)
Figure 5. AFM images of MC-POSS–Kaptons® . (a) 0 wt% POSS–Kapton® (image area:

20 µm2 ), (b) 1.75 wt% Si8 O11 MC-POSS–Kapton® (image area: 5 µm2 ), (c) 3.5 wt% Si8 O11
MC-POSS–Kapton® (image area: 10 µm2 ), (d) 5.25 wt% Si8 O11 MC-POSS–Kapton® (image area:
20 µm2 ) and (e) 7 wt% Si8 O11 MC-POSS–Kapton® (image area: 20 µm2 ).
(a) 5 µm: (b) 5 µm:
Figure 6. SEM images of 7 wt% Si8 O11 MC-POSS–Kapton® exposed to 2.3 × 1020 O atoms/cm2 ;
(a) gold-coated and measured at 5 kV, direct magnification: 2600×, (b) non-coated and measured at
30 kV, direct magnification: 5000×.
This indicates that these regions may be POSS aggregates, or silicon containing
aggregates, formed during exposure to AO.
Further insight into POSS dispersion and POSS–POSS aggregation can be
gained by TEM imaging of POSS–Kapton® films. Wei and coworkers [3, 4] carried
out TEM imaging on POSS-end-capped Kapton® films, which were directionally
blade-cast before cure. It was reported that POSS–POSS aggregates formed lamel-
lae about 2–3 nm wide, as seen by layers of dark lines perpendicular to the film
plane and casting direction, as well as by dark dots when the image was taken par-
allel to the film plane and casting direction. Interestingly, the Si–Si distance in a
POSS cage is 0.5 nm and the width of a POSS cage with cyclohexyl groups is
1.5 nm.
As seen in Figs 7 and 8, the 7.0 and 8.8 wt% Si8 O11 MC-POSS–Kapton® films
appear to have areas of POSS dispersion and areas with darkened POSS–POSS
aggregates, creating lamellae that are approximately 5 nm wide. There are also a
very small number of darkened dots approximately 5 nm wide. Fabrication of MC-
POSS–Kaptons® has been carried out with much iteration over a multi-year time
span, resulting in films with consistent physical properties and surface morpholo-
gies per MC-POSS wt percent. Figures 7 and 8 show initial TEM results of these
materials.
The SC-POSS–Kaptons® result in different TEM images than the MC-POSS
analogues. These newer materials are under analysis for characteristic morpholo-
gies. Despite the differences between MC- and SC-POSS–Kaptons® , the physical
properties described above, and AO resistance described in the following, are fairly
similar. In fact, the AO resistance data are better correlated with wt% POSS than
with the position of the POSS co-monomer, relative to the polymer chain.
Previously, the AO exposure of 0, 3.5, 7.0, 8.75 wt% Si8 O11 MC-POSS–
Kapton® films was carried out in a simulated LEO environment where these ma-
(a) Unexposed, 100 nm: (b) AO-Exposed, 100 nm:
Figure 7. TEM image of 7.0 wt% Si8 O11 MC-POSS–Kapton® measured at (a) 120 kV, direct mag-
nification: 150 000× and (b) 120 kV, direct magnification: 200 000×.
(a) 100 nm: (b) 100 nm:
Figure 8. TEM images of unexposed 8.8 wt% Si8 O11 MC-POSS–Kapton® measured at (a) 120 kV,
direct magnification: 150 000× and (b) 120 kV, direct magnification: 200 000×.
terials were etched by exposure to a hyperthermal O-atom beam. The difference

in etch depth between the eroded and stainless steel screen-protected areas of the
samples made it possible to calculate an AO reaction efficiency (Re ) or erosion rate
of the material for a given AO flux [18]. The AO reaction efficiency of the Kap-
ton H® reference sample was used to calculate the Kapton® equivalent fluence and
erosion yield of each sample after exposure. For various exposures, the step heights
(or etch depths) of POSS–Kapton® films were plotted as a function of the step
height of the Kapton H® film [18, 31]. The derivative functions indicated that the
3.5 and 7.0 wt% Si8 O11 POSS–polyimide films reached erosions rates of 3.7 and
0.98%, respectively, of the erosion rate for Kapton H® after 395 000 beam pulses
(8.47 × 1020 O atoms/cm2 ) [28, 33]. 8.75 wt% Si8 O11 MC-POSS–Kapton® sam-
ples had an erosion rate that was 0.3 percent of the erosion rate for Kapton H® , and
one-third of 7.0 wt% POSS–Kapton® at a fluence of 8.5 × 1020 O atoms/cm2 .
A ‘self-passivation test’ was carried out to better assess the protective silica layer
formation witnessed by XPS analysis of POSS–Kaptons® . Kapton H® , SiO2 -coated
Kapton HN® , and 8.75 wt% Si8 O11 MC-POSS–Kapton® were exposed to equiva-
lent AO fluences. The etch depth of the Kapton H® after the initial exposure was
7.0 ± 0.2 µm, indicating an O-atom fluence of 2.3 × 1020 O atoms/cm2 . This sample
appeared significantly roughened after exposure. The etch depth of 0.26 ± 0.15 µm
of the 8.75 wt% Si8 O11 MC-POSS–Kapton® film was difficult to measure since
the overall etch depth was not much greater than the slight roughness caused by
the exposure. For the SiO2 -coated Kapton HN® the etch depth was below the
practical measurement limit of the profilometer and the sample surface appeared
unaffected [29].
After the initial AO-exposure, Kapton H® , SiO2 -coated Kapton HN® , and
8.75 wt% Si8 O11 MC-POSS–Kapton® were scratched and underwent a second
exposure to 2.3 × 1020 O atoms/cm2 . SEM images are shown in Fig. 9. For the
SEM image of Kapton H® , the exposed region is on the left side of the image and
has a more roughened surface with a deepened scratched area. The etch depth of
the Kapton H® sample outside of the scratch was about 5.5 µm. The unexposed
scratch was 20 µm wide and 1 µm deep, and increased in depth to 1.4 µm after
the second exposure. The top area of the SiO2 -coated Kapton HN® image was
exposed to AO and only had erosion in the scratched area with unaffected neigh-
boring silica-coated Kapton HN® , demonstrating the effects of damage to silica
coatings on Kapton® . Here the unexposed scratch was 20 µm wide and 1 µm deep
and the exposed scratch was 8 µm deep relative to the neighboring exposed surface,
amounting to 7 µm of erosion in the scratch. 8.75 wt% Si8 O11 MC-POSS–Kapton®
was exposed to AO in the darkened upper right area in Fig. 9(c). A difference in
step height between exposed and unexposed unscratched areas of the 8.75 wt%
Si8 O11 MC-POSS–Kapton® was <200 nm and was not visible by microscopy. In
Fig. 9(c), the scratch was 35 µm wide and 1.4 µm deep before exposure and about
1.8 µm deep after AO exposure. This result indicates that 8.75 wt% Si8 O11 MC-
POSS–Kapton® experienced about 200 nm of erosion inside the scratch during AO
(a) (b) (c)
Figure 9. SEM images from a self-passivation experiment. (a) Kapton H® , (b) SiO2 -coated Kapton
HN® (Provided by Astral Industries Inc., USA) and (c) 8.75 wt% Si8 O11 cage MC-POSS–Kapton® .
Samples were exposed to 2.3 × 1020 O atoms/cm2 , scratched, covered by a wire screen and exposed
again to 2.3 × 1020 O atoms/cm2 . (The area covered by the wire screen was (a) the right side, (b) the
bottom portion and (c) the left side.)
Table 2.
Erosion depths for unscratched and scratched AO-exposed samples
Kapton H® SiO2 -coated Kapton HN® 8.75 wt% Si8 O11

with aluminum MC-POSS–Kapton®
under-coating
a Erosion depth (µm) 5.5 0 0.2
b Erosion depth (µm) measured 5 0 0
on the unscratched area
b Erosion depth (µm) measured 5 7 0.2
on the scratched area
a Sample history: (1) covered with a screen and (2) exposed to 2.7 × 1020 O atoms/cm2 .
b Sample history: (1) exposed to 2.7 × 1020 O atoms/cm2 , (2) scratched 1 µm deep, (3) covered
with a screen and (4) exposed to 2.7 × 1020 O atoms/cm2 .
Table 3.
AO erosion data for SC-POSS–Kapton® polyimides
Weight % Si8 O12 ∗ Kapton® -equivalent Erosion Percent erosion

in SC-POSS–Kapton® fluence/1020 O atoms/cm2 depth of Kapton H®
(µm) witness sample
1.75 2.71 1.99 ± 0.01 24.5

3.5 2.66 1.29 ± 0.05 16.15
7.0 2.68 0.390 ± 0.04 4.9
8.8 2.68 0.132 ± 0.02 1.64
10.5 2.71 0.249 ± 0.03 3.06
12.3 2.71 0.113 ± 0.03 1.39
14.0 2.71 Undetectable ∼0
∗ Calculated using the accepted erosion yield for Kapton H in LEO: 3.00 × 1024 cm3 /O atom.
exposure. In comparison, Kapton H® eroded about 5 µm both inside and outside

the scratch, plus Kapton H® eroded an additional 0.2 µm in the scratch during ex-
posure. In all instances, the scratched regions apparently eroded slightly more than
the unscratched exposed areas. This may be explained by the undercutting effect
described above where the O atoms are focused deeper into the polymer matrix
by scattering from the sidewalls of the scratches. These results are summarized in
Table 2.
The series of SC-POSS–Kaptons® was exposed to AO with a total fluence of
2.7 × 1020 O atoms/cm2 . The results in Table 3 show a dramatic decrease in surface
erosion with increasing POSS content, until undetectable erosion is reached with
14 wt% SC-POSS–Kapton® .
In an AO exposure with a total fluence of 3.53 × 1020 O atoms/cm2 , 7% Si8 O12

SC-POSS–Kapton® had an erosion yield that was 3.3% of Kapton H® . In a sep-
arate AO exposure with a total fluence of 4.10 × 1020 O atoms/cm2 , 7% Si8 O11
MC-POSS–Kapton® had an erosion yield that was 3.8% of Kapton H® . These ma-
terials were also exposed side by side to 2.68 × 1020 O atoms/cm2 . The % erosion of
the POSS–Kaptons® relative to Kapton H® was 4.25 ± 0.48 µm for 7 wt% Si8 O11
MC-POSS–Kapton® and 4.86 ± 0.47 µm for 7 wt% Si8 O12 SC-POSS–Kapton® .
The laboratory-scale synthesis of the polymer films is likely to produce slight dif-
ferences in molecular weights and cure cycles, thus potentially having minor effects
on the AO-erosion of the materials. The limited existing data indicate that the ef-
fect of MC-POSS and SC-POSS on POSS–Kapton® film erosion is comparable.
Thus, the level of erosion resistance afforded by POSS appears, with current data,
to depend on the wt% of the monomer and not on how it is bound to the polymer
chain.
MC-POSS–Kapton® films were flown on MISSE-1 as described above. Vari-
ous images of the samples were taken throughout the flight which show that the
0% POSS–Kapton® was completely eroded in less than four months [43]. The in-
nermost portion of the retrieved samples were thinner indicating they experienced
greater erosion than the portions closer to the unexposed (masked) area, and/or
they were thinner in this center area before flight. From the step heights from the
unexposed area and the neighboring exposed portion of the sample, measured in 12
places around the circumference of the exposed sample area, it was determined that
the thickness of the masked 0% POSS–Kapton® was 32.6 ± 0.9 µm. The erosion
data are shown in Table 4, and SEM images of the flown 1.75 wt% and 3.5 wt%
Si8 O11 MC-POSS–Kaptons® are shown in Figs 10 and 11. It was determined by
XPS that the atomic percentages of the top 10 nm of both POSS–Kapton® films
were 34% Si, 59% O and 7% C for both the 1.75 and 3.5 wt% Si8 O11 MC-POSS–
Kapton® samples.
0, 1.75, 3.5, 5.25 and 7.0 wt% Si8 O11 MC-POSS–Kaptons® were flown on the
top side of the International Space Station on MISSE-5 as described above. Sam-
Table 4.
Erosion data from samples flown on MISSE-1
Sample a Average step Standard Erosion in

height (µm) deviation (µm) 3.9 years (µm)
0 wt% Si8 O11 MC-POSS–Kapton® 32.6 0.9 b 240

1.75 wt% Si8 O11 MC-POSS–Kapton® 5.8 1.3 5.8
3.5 wt% Si8 O11 MC-POSS–Kapton® 2.1 0.3 2.1
a Data from 12 step-height measurements around the perimeter of exposed area of circular sample
with neighboring outer ring masked and not exposed to LEO.
b 0% POSS–Kapton total erosion estimated using the AO fluence approximated for the total flight:
8 × 1021 O atoms/cm2 .
(a) 10 µm: (b) 200 µm:
Figure 10. Scanning Electron Micrographs of (a) 1.75 wt% Si8 O11 MC-POSS–Kapton flown on
MISSE-1 and (b) 3.5 wt% Si8 O11 MC-POSS–Kapton flown on MISSE-1.
(a) 10 µm: (b) 200 µm:
Figure 11. Scanning Electron Micrographs of 3.5 wt% Si8 O11 MC-POSS–Kapton flown on MISSE-1.
Table 5.
Erosion data from MC-POSS–Kapton samples flown on MISSE-5
Sample Erosion depth (µm)
Kapton H 2.42 ± 0.47

0 wt% POSS–Kapton 2.89 ± 0.59
1.75 wt% Si8 O11 MC-POSS–Kapton 1.15 ± 0.21
ples were situated such that a part of the ISS obstructed the RAM and so a much
less harsh environment was experienced. All samples survived and erosion depths
decreased with increasing POSS content as shown in Table 5. As is the case with
AO exposures in simulated LEO experiments, the commercial Kapton H® eroded
less than the 0% POSS–Kapton® , which is synthesized by the same technique as

the POSS–Kaptons® .
MC-POSS–Kaptons® and SC-POSS–Kaptons® containing 3.5, 7.0 and 8.75 wt%
POSS cage are currently undergoing LEO exposure on the MISSE-6 which was
launched in March 2008.
4. Conclusions
The incorporation of POSS nanostructures into Kapton® polyimides has been

shown to significantly extend the lifetime of these materials in low Earth orbit.
Studies on the effect of a hyperthermal O-atom beam on POSS–Kaptons® have
shown improved oxidation resistance imparted to polyimides by the addition of
POSS. XPS data of both the AO-exposed and space-flown POSS–Kapton® materi-
als indicated that the improved oxidation resistance of these materials was due to
a rapidly formed silica layer upon exposure of POSS–Kaptons® to high incident
fluxes of atomic oxygen.
The effect of MC-POSS on Kapton® surface morphology and POSS-dispersion
in the polymer matrix was studied by Atomic Force Microscopy and Transmis-
sion Electron Microscopy (TEM), respectively. It was found that 3.5 and 5.25 wt%
Si8 O11 MC-POSS loadings resulted in a unique POSS–Kapton® surface morphol-
ogy with raised and depressed areas, by the film casting and curing techniques used
in this work. Dark lamellae were seen in the TEM images of 7 wt% and 8.8 wt%
Si8 O11 MC-POSS–Kaptons® .
The incorporation of POSS slightly reduced the glass transition temperature
(Tg ) of Kapton® polyimide by about 5 to 10%. However, the Tg values of POSS–
Kaptons® are well above the temperatures experienced by materials in many appli-
cations. The new SC-POSS monomer imparts comparable AO resistance to poly-
imides as the MC-POSS monomer, and transparent flexible films can be made at
higher SC-POSS loadings than MC-POSS loadings. The modulus, Tg , and CTE of
SC-POSS–Kaptons® were determined before and after exposure to atomic oxygen,
and were similar to the values of these properties for Kapton H® .
Acknowledgements
The authors would like to thank Dr. Darrell Marchant for carrying out dynamic
mechanical thermal analysis (DMTA) and coefficient of thermal expansion (CTE)
measurements, as well as directing 2nd Lt. Laura Moody in performing DMTA
and CTE measurements. We thank Mrs. Marietta Fernandez for SEM images and
EDX studies, Mrs. Rebecca Morello for AFM images, and Dr. Timothy Haddad for
MC-POSS synthesis guidance. We appreciate a meticulous film fabrication by Dr.
Gregory R. Yandek. Dr. Rene Gonzalez is gratefully acknowledged for his work
with MC-POSS–Kapton® . This work was financially supported by Defense Ad-
vanced Research Projects Agency (DARPA), the Air Force Office of Scientific
Research (Grant Nos. F49620-01-1-0276 & LRIR-92PL0COR).
References
1. L. Klein, M. De Guire, F. Lorraine and J. Mark, Mater. Res. Soc. Symp. Proc. 576, 465 (1999).
2. C. Yi, S. Wen, K. Ho, T.-H. Hsieh, F.-C. Chang and Y.-Z. Wand, Macromol. Rapid Commun. 27,
452 (2006).
3. C.-M. Leu, Y.-T. Chang and K.-H. Wei, Macromolecules 36, 9122 (2003).
4. C.-M. Leu, Y.-T. Chang and K.-H. Wei, Chem. Mater. 15, 3721 (2003).
5. D. A. Schorzman, M. E. Wright, F. J. Feher and R.-Z. Jin, Polymer Preprints (American Chemical
Society, Division of Polymer Chemistry) 43(2), 1104 (2002).
6. D. A. Schorzman and M. E. Wright, Polym. Mater. Sci. Eng. 85, 482 (2001).
7. M. E. Wright, D. A. Schorzman, F. J. Feher and R. Z. Jin, Chem. Mater. 15, 264 (2003).
8. J.-C. Huang, C.-B. He, Y. Xiao, K. Y. Mya, J. Dai and Y. Siow, Polymer 44, 4491 (2003).
9. Y. Yamada and J. Sakai, PCT Int. Appl., 46 (2006).
10. R. D. Karam and P. Zarchan (Eds), Satellite Thermal Control for Systems Engineers, Progress in
Astronautics and Aeronautics, Vol. 181. NASA (1998).
11. K. K. de Groh and B. A. Banks, in: Proceedings, Society for the Advancement of Materials and
Process Engineering, Long Beach, CA (2002).
12. S. L. Koontz, L. J. Leger, J. T. Visentine, D. E. Hunton, J. B. Cross and C. L. Hakes, J. Spacecraft
Rockets 32, 483 (1995).
13. K. K. de Groh and B. A. Banks, J. Spacecraft Rockets 31, 656 (1994).
14. B. A. Banks, S. K. Rutledge, K. K. de Groh, M. J. Mirtich, L. Gebauer, R. Olle and C. M. Hill, in:
Proc. 5th Int. Symp. on Materials in a Space Environment, Cannes Mandelieu, France (1991).
15. R. C. Tennyson, Surface Coatings Technol. 68/69, 519 (1994).
16. J. W. Gilman, D. S. Schlitzer and J. D. Lichtenhan, J. Appl. Polym. Sci. 60, 591 (1996).
17. R. I. Gonzalez, S. H. Phillips and G. B. Hoflund, J. Spacecraft Rockets 37, 463 (2000).
18. T. K. Minton and D. J. Garton, in: Chemical Dynamics in Extreme Environments, R. A. Dressler
(Ed.), p. 420. World Scientific, Singapore (2001).
19. D. Dooling and M. M. Finckenor, Material Selection Guidelines to Limit Atomic Oxygen Effects
on Spacecraft Surfaces. NASA/TP – 209 (1999).
20. X. Zhang, Y. Wu, S. He and D. Yang, Yuhang Cailiao Gongyi 37(4), 19–23 (2007).
21. S. Wagner, H. Dai, R. A. Stapleton, M. L. Illingsworth and E. J. Siochi, High Performance Poly-
mers 18, 399 (2006).
22. X. Wang, X. Zhao, M. Wang and Z. Shen, Nucl. Instrum. Methods Phys. Res. B 243, 320 (2006).
23. D. Devapal, S. Packirisamy, R. M. Korulla and K. N. Ninan, J. Appl. Polym. Sci. 94, 2368 (2004).
24. J. Grobner and J. B. Kerr, J. Geophysical Research — Atmospheres 106(D7), 7211 (2001).
25. K. S. W. Champion, A. E. Cole and A. J. Kantor, in: Handbook of Geophysics and the Space
Environment, A. S. Jursa (Ed.), pp. 14–43. National Technical Information Service, Springfield,
VA (1985).
26. A. E. Hedin, J. Geophysical Research — Space Physics 88(A12), 170 (1983).
27. S. L. Koontz, L. J. Leger, S. L. Rickman, C. L. Hakes, D. T. Bui, D. E. Hunton and J. B. Cross,
J. Spacecraft Rockets 32, 475 (1995).
28. S. J. Tomczak, D. Marchant, S. Svejda, T. K. Minton, A. L. Brunsvold, I. Gouzman, E. Grossman,
G. C. Schatz, D. Troya, L. Sun and R. I. Gonzalez, Mater. Res. Soc. Symp. Proc. 851, 395 (2005).
29. S. J. Tomczak, V. Vij, T. K. Minton, A. L. Brunsvold, D. Marchant, M. E. Wright, B. J. Petteys,

A. J. Guenthner, G. R. Yandek and J. M. Mabry, ACS Symposium Series 978, 140 (2008).
30. S. J. Tomczak, T. K. Minton, A. L. Brunsvold, V. Vij, M. E. Wright, B. J. Petteys, A. J. Guenthner,
S. A. Svejda and J. M. Mabry, Proc. SAMPE, 52 (2007).
31. A. L. Brunsvold, T. K. Minton, I. Gouzman, E. Grossman and R. Gonzalez, J. High Performance
Polymers, Special Edition 16(2), 303 (2004).
32. R. I. Gonzalez, Ph.D. Dissertation, Chem. Eng. Department, University of Florida,
http://purl.fcla.edu/fcla/etd/UFE1000127 (2002).
33. S. H. Phillips, T. S. Haddad and S. J. Tomczak, Current Opinion Solid State Mater. Sci. 8, 2 (2004).
34. S. J. Tomczak, V. Vij, D. Marchant, T. K. Minton, A. L. Brunsvold, M. E. Wright, B. J. Petteys
and A. J. Guenthner, Proc. SPIE 6308, 4 (2006).
35. M. E. Wright, B. J. Petteys, A. J. Guenthner, S. Fallis, G. R. Yandek, S. J. Tomczak, T. K. Minton
and A. L. Brunsvold, Macromolecules 39, 4710 (2006).
36. F. J. Feher, F. Nguyen, D. Soulivong and J. W. Ziller, Chem. Commun., 1705 (1999).
37. F. J. Feher, R. Terroba and J. W. Ziller, Chem. Commun., 2309 (1999).
38. F. J. Feher, D. Soulivong and A. G. Eklund, Chem. Commun., 399 (1998).
39. K. Pielichowski, J. Njuguna, B. Janowski and J. Pielichowski, Adv. Polym. Sci. 201, 225 (2006).
40. M. E. Wright, S. J. Tomczak, B. J. Petteys, A. J. Guenthner, V. Vij, T. K. Minton, A. Brunsvold,
L. M. McGrath and J. M. Mabry, in preparation.
41. S. J. Tomczak, M. E. Wright, A. J. Guenthner, B. J. Petteys, T. K. Minton, A. Brunsvold, V. Vij,
L. M. McGrath and J. M. Mabry, in: Proceedings High Temperature Resins and Coatings I,
SAMPE (2008).
42. S. J. Tomczak, M. E. Wright, A. J. Guenthner, B. J. Petteys, A. Brunsvold, C. Knight, T. K. Minton,
V. Vij, L. M. McGrath and J. M. Mabry, in: Proc. 9th Int. Conf. on the Protection of Materials in
a Space Environment, Toronto, Canada (publication in process 2008).
43. http://misse1.larc.nasa.gov
44. Optical Components.com
Part 2
Surface and Adhesion Aspects
Fluorination of PMDA-ODA Polyimide Using Hyperthermal
Atomic Fluorine Beams
Masahito Tagawa a,∗ , Kumiko Yokota a , Ken-ichi Maeda a , Akitaka Yoshigoe b ,

Yuden Teraoka b , Kensuke Akamatsu c and Hidemi Nawafune c
a
Department of Mechanical Engineering, Graduate School of Engineering, Kobe University,
Rokko-dai 1-1, Nada, Kobe 657-8501, Japan
b
Japan Atomic Energy Agency, Kouto 1-1-1, Sayo-cho, Sayo-gun, Hyogo 679-5148, Japan
c
Faculty of Science and Engineering, Konan University, Okamoto 8-9-1, Higashi-Nada,
Kobe 658-8501, Japan
Abstract
Hyperthermal fluorine atomic beam interaction with PMDA-ODA polyimide (PI) has been studied. It was
found that CF, CF2 and CF3 groups were formed by reaction of PI with atomic fluorine with an impingement
energy of 11.7 eV. Advancing contact angle of water increases rapidly to 100◦ , followed by a slow increase
due to the increase in surface roughness. A gasification reaction of PMDA-ODA was observed under hyper-
thermal atomic fluorine exposure, and the erosion yield was of the same order of magnitude as that by atomic
oxygen. Surface analysis results of atomic fluorine-exposed highly oriented pyrolytic graphite (HOPG) us-
ing synchrotron radiation photoelectron spectroscopy showed that the fluorine concentration was limited to
only topmost HOPG (0 0 0 1) layer. From the HOPG experiment, it is revealed that the physical sputtering
effect can be ignored in this hyperthermal atomic beam process and fluorination reaction was restricted only
to the topmost graphite layer.
Keywords
Atomic fluorine, polyimide, HOPG, contact angle, synchrotron radiation
1. Introduction
When atoms or molecules collide with surfaces, complicated gas/surface reactions

occur. They are categorized as chemical and physical processes. Physical sputtering
phenomenon, which is one of the physical processes, disturbs the atomic arrange-
ment of the surface and affects the surface physical/chemical/electrical properties.
It has been successfully applied to remove surface layers before surface analysis for
many years, however, sputtering phenomenon is not appropriate for the preparation
* To whom correspondence should be addressed. Tel./Fax: +81-78-803-6126; e-mail: tagawa@mech.kobe-

u.ac.jp
250 M. Tagawa et al.
of the atomically-controlled surfaces. In order to avoid the physical sputtering effect

at the sample surface, reducing the impingement energy of gas ion/atom is essen-
tial. Since the threshold energy of physical sputtering phenomenon is a few tens of
eV, it is necessary to limit the collision energy below this threshold. On the other
hand, conventional physical vapor deposition (PVD) or chemical vapor deposition
(CVD) processes use only thermal energy (<0.1 eV) thus no sputtering effect is
involved in these process. However, in order to promote chemical reaction, ther-
mal energy (sample heating) is usually applied for overcoming the potential energy
barrier of the chemical reaction. The necessity of heat treatment restricts the ap-
plication of these processes to low melting point materials. Impingement energy of
the atoms/ions in the range 5–15 eV could produce both effects simultaneously, i.e.,
avoidance of physical sputtering phenomenon and promotion of chemical reactions.
Technologically, it has been difficult to accelerate atoms/ions to 15 eV. It is too
high energy for conventional molecular beam technology and too low energy for
ion beams. This energy gap of atomic beam techniques was filled by the laser
detonation method. This method was originally investigated for the laser propul-
sion and later applied for the space environmental effect studies [1, 2]. The broad
atomic beam with a translational energy of 5–10 eV formed by a laser detonation
method also has the unique capability as a surface modification technique: no charg-
ing effect, large area treatment, high reactivity, high intensity, etc. These features
are suitable for modifying non-conductive surfaces with a high spatial resolution,
beyond micrometer-level [3–5]. It has been demonstrated that the hyperthermal
atomic oxygen and atomic fluorine beams formed by this method can be useful
for altering the polymer surface chemistry [6].
In this study, effects of hyperthermal atomic fluorine beam exposure on the poly-
imide surfaces were investigated based on the analysis results obtained by X-ray
photoelectron spectroscopy (XPS), atomic force microscopy (AFM) and contact
angle measurements. Detailed analysis on the atomic fluorine beam and surface car-
bon atom was carried out using the basal plane of highly oriented pyrolytic graphite
(HOPG) as a specimen. A high-resolution, ultra surface-sensitive technique using
synchrotron radiation photoelectron spectroscopy (SR-PES) was utilized for HOPG
analysis.

Samples used in this study were commercially available PMDA-ODA polyimide
(Kapton-H and SP-510, Toray-DuPont) and HOPG. Kapton-H surface was rinsed
with ethanol before the experiment. SP-510 was spin-coated onto a quartz crystal
microbalance (QCM) in order to measure the real-time mass change during the
beam exposure. Heat treatments of 150◦ C for 1 h and 300◦ C for 1 h were carried
out in a N2 environment. HOPG was cleaved just before the exposure.
Figure 1 shows the schematic drawing of the laser detonation fast atomic beam
facility used in this study. The atomic beam source is based on the laser break-
Fluorination of Polyimide Using Atomic Fluorine Beams 251
Figure 1. Schematic drawing of the laser detonation atomic fluorine beam source used in this study.
down phenomenon. This type of atomic beam source uses a pulsed CO2 laser
(5–6 J/pulse) and a pulsed supersonic valve (PSV). The laser light was focused
on the nozzle throat with a concave Au mirror located 50 cm away from the noz-
zle. The PSV introduced pure target gas into the nozzle and the laser light was
focused on the target gas in the nozzle. The energies for dissociation and accel-
eration of molecules/atoms were provided by the multiphoton absorption process.
SF6 was used as a target gas for forming atomic fluorine beam. The atomic beam
was characterized by a time-of-flight (TOF) distribution measured by a quadrupole
mass spectrometer installed in the beamline. Translational energies of the species
in the beam were calculated using TOF distributions with a flight length of 238 cm.
Atomic fluorine (m/z = 19) was the major component when SF6 was used as the
target gas. Figure 2 represents typical TOF distribution of m/z = 19, which cor-
responds to atomic fluorine. The zero time in Fig. 2 corresponds to laser firing.
Beside the main peak at 220 µs, sharp spikes are obvious in the 100–150 µs range.
These spikes are due to the residuals after subtraction of ion component. (Note that
the detection yield of fluorine ion in the beam is much higher than that of fluo-
rine atoms. The ion component, which was measured with the filament-off, was
subtracted from the TOF spectra with the filament-on.) The peak velocity of the hy-
perthermal atomic fluorine beam at 220 µs corresponds to 10.7 km/s (translational
energy: 11.3 eV). The atomic fluorine beam fluxes were measured with a Ag-coated
QCM. The resonance frequency of Ag-coated QCM decreased due to the forma-
tion of silver fluoride. With an assumption of accommodation coefficient of atomic
fluorine at the Ag surface being 1.0, the lower bound of the beam flux could be
estimated. Atomic fluorine flux at the sample position, 48 cm away from the nozzle
throat, was 3.6 × 1014 atoms/cm2 /s.
Water contact angles on polymer surfaces were measured by a video contact
angle measurement system using the sessile drop method. The advancing contact
Figure 2. Time-of-flight spectrum of the m/z = 19 used in the in situ mass change study.
angles were measured from the time dependence of the contact angle measured
by the sessile drop method. Details of the time-resolved sessile drop method are
described in the literature [7].
A conventional XPS system with a non-monochromatic Mg Kα X-ray source was
used for surface analysis of PMDA-ODA polyimide. In contrast, high-resolution,
ultra surface sensitive analysis of HOPG was carried out by synchrotron radiation
photoelectron spectroscopy (SR-PES). SR-PES measurement was carried out with
the SUREAC2000 end-station installed in the soft X-ray beamline, BL23SU, at
SPring-8 facility in Hyogo, Japan [8]. 840 eV soft X-rays, which generates F1s
photoelectron with kinetic energy of 150 eV, were used for HOPG analysis. This
kinetic energy provides the shortest inelastic mean free path (IMFP) in solids, i.e.,
the minimum escape depth of photoelectron signals (extremely surface sensitive)
[9].

3.1. PMDA-ODA
Figure 3 shows the real-time resonance frequency shift of the PMDA-ODA poly-
imide during 11.3 eV atomic fluorine beam exposure. The resonance frequency was
recorded every 10 s. Note that the resonance frequency of QCM (5 MHz) can be
directly converted to mass change by the following equation:
M = 2.002 × 10−8 F, (1)
where M is the mass change in gram and F the frequency change in Hz. At
700 s elapsed time, the fluorine beam exposure started. It is clearly indicated that
the resonance frequency of the PMDA-ODA polyimide decreased at the beginning
Figure 3. Frequency shift of the PMDA-ODA polyimide during atomic fluorine beam exposure. En-
ergy: 11.3 eV, flux: 3.6 × 1014 atoms/cm2 /s.
of the exposure and was followed by a linear frequency increase. This tendency is
similar to that observed for atomic oxygen exposure on the PMDA-ODA polyimide
[10]. The linear mass loss is due to the formation of volatile reaction products. In
this case, CFx would be major reaction products, however, the volatile products
cannot be detected in this system. The slope of linear frequency increase observed
in Fig. 3 is calculated to be 0.02 Hz/s. From this slope, the erosion yield of PMDA-
ODA polyimide by fluorine beam is calculated to be 2.9 × 10−24 cm3 /atom (in this
calculation, 1.4 g/cm3 is used as the density of polyimide). This erosion yield is
of the same order as that by atomic oxygen (3 × 10−24 cm3 /atom). The effect of
impingement energy on erosion rate was also studied. Figure 4 shows the frequency
shifts of the PMDA-ODA polyimide under exposures to 11.3 eV and 5.3 eV atomic
fluorine beams. The slopes of the frequency shifts, i.e., mass loss rates, are identical
for both beams. It is natural to consider that the mass loss rate, i.e., reaction proba-
bility of gasification reaction, is strongly related to the impinging energy, however,
no such effect was observed in Fig. 4. This surprising result could be related to the
byproducts from the SF6 plasma such as extreme ultraviolet (EUV) emission or to
the amount of hyperthermal fluorine component.
Figure 5 shows the advancing contact angles of water on the atomic fluorine
beam exposed polyimide (Kapton-H). Advancing contact angle increased with in-
creasing atomic fluorine fluence, and it reached as high as 120◦ which is greater
than that of Teflon FEP. The initial increase in advancing contact angle shown in
Fig. 5 is rather rapid. This is explained using the theory proposed by Johnson and
Dettre [11, 12]: advancing contact angle is sensitive to fractional coverage by low-
energy region on predominantly high-energy surfaces. The advancing contact angle
reaches 110◦ with atomic fluorine beam exposure of 1 × 1019 atoms/cm2 . This ad-
Figure 4. Comparison of frequency shifts of PMDA-ODA polyimide under atomic fluorine exposures
with different beam energies: Ek = 11.3 eV and 5.3 eV.
Figure 5. Advancing contact angle of water as a function of atomic fluorine fluence.
vancing contact angle is equivalent to that of Teflon FEP. It is thus considered that
the fluorination of the topmost layer would be completed at the beginning of ex-
posure (approximately 1 × 1018 atoms/cm2 ) and further exposure leads to etching
of PMDA-ODA. This etching process increases the surface roughness and thus the
advancing contact angles are further increased. This conclusion is supported also by
the surface fluorine concentration measured by XPS and the roughness measured
by AFM (Table 1).
Figure 6 shows the C1s core level spectra of PMDA-ODA polyimide before and
after exposure to atomic fluorine beam. The exposure conditions of the fluorine
Table 1.
Surface roughness of PMDA-ODA polyimide before and after ex-
posure to atomic fluorine beam
Atomic fluorine fluence (×1018 atoms/cm2 ) 0 6.5 52

Surface roughness Ra (nm) 0.05 1.3 5.7
Figure 6. C1s core level XPS spectra of PMDA-ODA polyimide before and after atomic fluorine
exposure. Energy: 6.2 eV, and fluence: 6.3 × 1019 atoms/cm2 .
beam were as follows: Energy, 6.2 eV and fluence, 6.3 × 1019 atoms/cm2 . The peak
synthesis protocol of the standard PMDA-ODA used in this study is described in
the literature [13]. In the C1s spectrum, C atoms in ODA are located at 284 eV,
those in PMDA at 285.0 eV (peak shift of +1.0 eV), COC at 285.6 eV (+1.6 eV),
and >C=O at 287.9 eV (+3.9 eV). Additional peaks by fluorination are obvious
after atomic fluorine beam exposure. Peak shift of fluorinated carbon is reported
to be CF (+4.7 eV), CF2 (+6.7 eV) and CF3 (+9.0 eV) [14]. The presence of
high-energy tail at 288–294 eV suggests that CF, CF2 and CF3 moieties are formed
by the atomic fluorine exposure. Stability of surface fluorination has been reported
elsewhere [15]. It has been shown that the fluorine concentration does not change
significantly even after storage in air for a few months. Hydrophobic property of
fluorinated PMDA-ODA was conserved. This stable fluorination was applied to the
patterning of electroless Cu plating. The capability of site-selective Cu metallization
was demonstrated elsewhere [15].
3.2. HOPG
Hyperthermal atomic fluorine beam reaction with carbon material is studied us-
ing HOPG as a reference surface. HOPG (0 0 0 1) surface was exposed to 10.4 eV
atomic fluorine beam with a fluence of 1.0 × 1018 atoms/cm2 . The atomic fluorine-
(a)
(b)
Figure 7. SR-PES survey spectra of the atomic fluorine exposed-HOPG. (a): before exposure and
(b): after exposure. Energy: 10.4 eV, and fluence: 1.0 × 1018 atoms/cm2 . Excitation energy: 840 eV.
exposed HOPG was removed from the laser detonation beam source and transferred
to the SUREAC2000 installed in the BL23SU at SPring-8 in Hyogo, Japan. 840 eV
soft X-rays were used for SR-PES measurements. This excitation energy was se-
lected to minimize the escape depth of F1s photoelectrons. SR-PES measurements
were carried out at a pressure lower than 1 × 10−8 Pa with a take-off angle of pho-
toelectrons 0 to 70◦ from the surface normal. Figure 7(a) shows the SR-PES survey
spectrum of atomic fluorine-exposed HOPG (0 0 0 1) with a take-off angle of 0◦
(surface normal). The abscissa reveals kinetic energy of photoelectrons. Beside the
intense C1s signal, O1s and F1s signals are obvious. Oxygen signal is observed due
to the sample exposure to the ambient air during transfer from the fluorine beam
Figure 8. Take-off angle dependence of the F1s intensity normalized by C1s intensity in the SR-PES
spectra.
source to SPring-8 facility. Figure 7(b) shows the SR-PES spectrum measured with
a take-off angle of 60◦ . It is clearly indicated that the F1s signal is enhanced with
this take-off angle. This result suggests that the fluorine is concentrated in a very
shallow region. Relative intensity of fluorine compared to the C1s intensity is cal-
culated as a function of the take-off angle of photoelectrons. Figure 8 shows the
results. It is clearly indicated that the relative fluorine intensity increased rapidly at
the take-off angle of 40◦ . From the kinetic energy of F1s photoelectron (150 eV),
IMFP is estimated to be 0.5 nm. In contrast, IMFP of C1s photoelectron (kinetic
energy of 550 eV) is estimated to be 1.0 nm. The escape depth of F1s is estimated
to be 0.38 nm at the take-off angle of 40◦ . It decreases to 0.32 nm at 50◦ . Those for
C1s are 0.76 nm and 0.64 nm for the take-off angles of 40◦ and 50◦ , respectively.
The increase in fluorine concentration is due to the difference in distributions of
carbon and fluorine in the escape depth of photoelectrons. The fact that the rela-
tive fluorine intensity rapidly increases near the escape depth of 0.3 nm suggests
that the fluorine atoms are concentrated at the topmost HOPG layer (the lattice
constant along with c-axis of HOPG is 0.34 nm). It is concluded that the fluorine
atoms with the impingement energy of 10.4 eV do not penetrate the topmost HOPG
(0 0 0 1) layer. Figure 9 shows the C1s high-resolution spectra of fluorine exposed
HOPG (0 0 0 1). Beside the main C1s signal, two additional peaks are observed at
548.2 and 550.6 eV. These peaks are labeled as CF2 and CF, respectively. These
peaks are also enhanced in the take-off angle of 60◦ relative to the main C1s peak
at 554.7 eV. It is confirmed that the fluorine atoms concentrated in the topmost
layer are chemically bonded with topmost carbon atoms. It was discovered from the
atomic fluorine beam experiment with HOPG that both fluorine concentration and
the carbon-fluorine bonding were distributed only in the topmost HOPG (0 0 0 1)
layer. These SR-PES results suggest that the hyperthermal atomic fluorine reaction
Figure 9. C1s core level SR-PES spectra of the atomic fluorine-exposed HOPG.
with carbon material occurs in the first layer at the surface. This analytical conclu-
sion proves that the physical sputtering effect can be ignored in the hyperthermal
atomic beam in the range of 10 eV. This phenomenon is also supported by the fact
that the impact energy is released at the first collision and low-temperature process
prevents further reaction of ‘thermal’ fluorine atoms.
4. Conclusions
Hyperthermal fluorine beam interaction with PMDA-ODA polyimide and HOPG

was studied. It was revealed that CF, CF2 and CF3 groups were formed at the atomic
fluorine-exposed PMDA-ODA surface. Advancing contact angle of water increases
rapidly to 100◦ , followed by a slow increase due to the increase in surface rough-
ness. It was confirmed that the gasification reaction of PMDA-ODA occurs under
the hyperthermal atomic fluorine exposure, and the erosion yield by atomic fluorine
is of the same order as that by atomic oxygen. Surface analysis results of atomic
fluorine-exposed HOPG using SR-PES revealed that the physical sputtering effect
can be ignored in the hyperthermal atomic beam process. This is due to the fact
that the impact energy is released at the first collision and low-temperature process
prevents further reaction of fluorine atoms.
Acknowledgement
Synchrotron radiation experiments were carried out through the nanotechnology

support program of JASRI.
References
1. G. E. Caledonia, R. H. Krech and D. B. Green, AIAA J. 25, 59–64 (1987); idem U.S. Patent No.
4894511 (1990).
2. A. Paillous, B. Cazaubon, A. Barassin and R. Thomas, in: Proceedings of the 7th International
Symposium on Materials in a Space Environment, Toulouse, France, pp. 185–190 (1997).
3. M. Tagawa, T. Ema, H. Kinoshita, N. Ohmae, M. Umeno and T. K. Minton, J. Appl. Phys. Jpn
37, L1455–L1457 (1998).
4. M. Tagawa, K. Yokota, N. Ohmae, H. Kinoshita and M. Umeno, J. Appl. Phys. Jpn 40, 6152–6156
(2001).
5. K. P. Giapis, T. A. Moore and T. K. Minton, J. Vac. Sci. Technol. A13, 959–963 (1995).
6. M. Tagawa, K. Yokota, Y. Ohki and N. Ohmae, J. Adhesion Sci. Technol. 18, 243 (2004).
7. K. Gotoh, M. Tagawa, N. Ohmae and M. Tagawa, in: Polymer Surface Modification: Relevance to
Adhesion, K. L. Mittal (Ed.), Vol. 2, pp. 445–457. VSP, Utrecht (2000).
8. Y. Saitoh, T. Nakatani, T. Matsushita, A. Agui, A. Yoshigoe, Y. Teraoka and A. Yokoya, Nucl.
Instrum. Meth. Phys. Res. A 474, 253 (2001).
9. S. Tanuma, C. J. Powell and D. R. Penn, Surf. Interface Anal. 21, 165 (1994).
10. M. Tagawa, K. Yokota, N. Ohmae and H. Kinoshita, J. Spacecraft Rockets 39, 447 (2002).
11. R. E. Johnson, Jr and R. H. Dettre, in: Surface and Colloid Science, E. Matijevic (Ed.), Vol. 2.
Wiley and Sons, New York, NY (1969).
12. R. E. Johnson, Jr and R. H. Dettre, J. Phys. Chem. 68, 1744–1749 (1969).
13. H. Kinoshita, M. Tagawa, M. Umeno and N. Ohmae, Trans. Jpn Soc. Aeronautics Space Sci. 41,
94 (1998).
14. G. Beamson and D. Briggs, High Resolution XPS of Organic Polymers: Scienta300 Database.
Wiley, Chichester (1992).
15. M. Tagawa, K. Maeda, T. Kajita, K. Yokota, K. Akamatsu and H. Nawafune, Langmuir 23, 11351–
11354 (2007).
Preparation of Highly Reflective and Conductive
PI/Ag Composite Films by an Ion-Exchange
Self-Metallization Technique
Shengli Qi a , Zhanpeng Wu a , Dezhen Wu a,∗ and Frank W. Harris b

a
College of Materials Science and Engineering, Beijing University of Chemical Technology,
Beijing 100029, China
b
Maurice Morton Institute and Department of Polymer Science, College of Polymer Science and
Polymer Engineering, The University of Akron, Akron, OH 44325-3909, USA
Abstract
A direct ion-exchange self-metallization technique is used to prepare a surface silver-metallized polyimide
(PI) film. The method involves performing an ion-exchange reaction of damp-dry poly(amic acid) films in
silver aqueous solution to form silver(I)-containing precursor films. Thermal treatment under tension con-
verts the poly(amic acid) into polyimide and simultaneously reduces the silver(I) to silver(0), yielding silver
(Ag) layers with excellent reflectivity and conductivity on both film sides. This work focuses on the influence
of polyimide structure, the kind of silver salt, ion-exchange time and the curing conditions on the morphol-
ogy and properties of films. Three polyimide matrices derived from 3,3 ,4,4 -benzophenonetetracarboxylic
dianhydride (BTDA)/4,4 -oxydianiline (ODA), 4,4 -oxydiphthalic anhydride (ODPA)/ODA and pyromel-
litic dianhydride (PMDA)/ODA and two silver salts, silver nitrate (AgNO3 ) and silver fluoride (AgF), were
investigated. XRD, TEM, SEM, XPS and AFM measurements were used to characterize the structure and
morphology of polyimide/silver composite films. The results show that films prepared from BTDA/ODA-
based polyimide and AgF are more efficient to obtain highly reflective and conductive films than the others.
Keywords
Polyimide, silver, ion exchange, metallization
1. Introduction
Polymeric matrices containing different metals have attracted great attention from
both fundamental and applied research because of their potential applications
in space, microelectronics and electrochemistry [1–7]. In this paper, we report
our work on the synthesis of surface silver-metallized polyimide films with high
reflectivity and conductivity via a newly-developed direct ion-exchange self-
*To whom correspondence should be addressed. Tel.: +86-10-6442-1693; Fax: +86-10-6442-1693; e-mail:
wdz@mail.buct.edu.cn
262 S. Qi et al.
metallization procedure. The method refers to developing metallized films by

thermal curing of the silver(I)-containing polyimide precursor films, in which the
metal species are loaded through the ion exchange of the damp-dry poly(amic acid)
(PAA) films in an aqueous silver solution. Thermal treatment of the silver(I)-doped
precursor films induces cycloimidization of the PAA to form the final polyimide
matrix and the concomitant reduction of the silver(I) species to silver(0). And dur-
ing the thermal cycle the reduced metal atoms and small clusters formed in the film
aggregate at the surface yielding the metallic silver layer. Due to the combination
of the excellent optical and electrical properties of the surface silver layer and the
outstanding thermal and mechanical properties of the polyimide matrix, such films
have been considered to be very promising for many uses including printed cir-
cuit boards, magnetic data storage, thin film reflectors and concentrators in space
environments for solar thermal propulsion, etc. [4, 8–15].
To construct such metallized films, several classic approaches have been em-
ployed including external deposition [13, 16, 17], supercritical fluid impregnation
[18, 19] and in situ single-stage self-metallization [12, 20–23]. Although film met-
allization could be achieved, these usual techniques have limitations. The major
problem with the external deposition procedure, which directly deposits a metal
layer onto a previously prepared substrate, is the poor adhesion at the metal–
polymer interface [24–26] and surface modification is usually needed to enhance
the interfacial bonding strength [3, 27–29]. As an alternative, Boggess et al. [18]
and Rosolovsky et al. [19] produced silver-metallized polyimide films with modest
reflectivity (max. 67.1%) and strong adhesion at the silver–polymer interface by su-
percritical fluid infusion of silver complex into the fully cured polyimide films using
supercritical CO2 as the carrier followed by annealing. However, besides only mod-
est reflectivity and non-conductivity, only one silver dimer could be used for this
technique and metallization could be realized only on a polyimide matrix with a ke-
tonic carbonyl group. The in situ single-stage self-metallization process [23], which
refers to producing metallized films by thermal curing of the films cast from the ho-
mogeneous solution composed of silver complexes and poly(amic acid), has been
considered to be the most prospective since it possesses advantages of processing
simplicity and outstanding adhesion at the metal–polymer interface. However, suc-
cess was only realized when complex silver compounds with bridging β-diketonate
ligands, which are usually very unstable, are required to be freshly prepared, and
are dramatically expensive, were utilized as the silver precursor. Attempts to fabri-
cate silvered polyimide films using simple silver salts never achieved real success
[16, 30]. In particular, films prepared using silver nitrate as the metal precursor were
visually degraded and brittle without mechanical usefulness [31].
Due to these limitations, the direct ion-exchange self-metallization method
was developed in our group with the purpose of fabricating surface-silvered
polyimide films by employing simple silver compounds, such as silver nitrate
(AgNO3 ) and silver fluoride (AgF), as the precursor. The synthesis protocol
is illustrated in Scheme 1. The uniqueness of this approach lies in the load-
Preparation of Highly Reflective and Conductive PI/Ag Composite Films 263
Scheme 1. Illustrative protocol for the synthesis of silver metallized BTDA/ODA based polyimide film via
direct ion-exchange self-metallization process.
ing of silver into the precursor film through its direct ion exchange in an
aqueous silver solution and the achievement of two-sided surface metallization.
Here, we present our efforts on achievement of silver metallization on 3,3 ,4,4 -
benzophenonetetracarboxylic dianhydride/4,4 -oxydianiline (BTDA/ODA), 4,4 -
oxydiphthalic anhydride (ODPA)/ODA and pyromellitic dianhydride (PMDA)/ODA
(Scheme 2) polyimide matrices via this direct ion-exchange self-metallization tech-
nique with AgNO3 and AgF as the silver precursors.
2. Experimental
2.1. Materials
All chemicals were obtained from commercial sources. The poly(amic acid) solu-
tions were prepared by first dissolving the diamine in dimethylacetamide (DMAc)
and then adding the dianhydride (1 mol% offset) gradually. The solid contents
were set at 20 wt%, 12 wt% and 10 wt% for the BTDA/ODA, ODPA/ODA and
PMDA/ODA precursor solutions, respectively.
264 S. Qi et al.
Scheme 2. Chemical structures of the BTDA/ODA, ODPA/ODA and PMDA/ODA poly(amic acid)s.
2.2. Preparation of Silvered Polyimide Films

Damp-dry poly(amic acid) films were prepared by spreading the prepared homoge-
neous precursor solution onto a clean glass plate followed by solvent evaporation in
an atmosphere of slowly flowing dry air or in a vacuum oven. These films were then
peeled from the glass substrate and treated in the aqueous silver solution to load sil-
ver ions into the matrix through ion exchange. After being rinsed thoroughly with
deionized water, the silver(I)-doped films were thermally treated under tension to
cycloimidize the poly(amic acid) and to reduce the silver(I) producing metallized
polyimide films, as depicted in Scheme 1. The thermal cycles were heating over 1 h
to 135◦ C and holding for 1 h, heating to 300◦ C over 2 h, and maintaining at 300◦ C.
For clarity, the surface of the damp-dry PAA film in contact with the glass substrate
is referred to as the upside, while that exposed to the atmosphere is referred to as
the underside.
2.3. Measurements
Surface reflectivity (relative to a BaSO4 mirror set at 100% reflectivity) was mea-
sured on a Shimadzu 2501PC UV/VIS spectrophotometer with an incidence angle
of 8◦ at a wavelength of 531 nm. Surface electrical resistances were determined with
an RTS-8 four-point probe meter. Scanning electron micrographs were recorded
on a HITACHI S-4300 system operating at 15 kV. The samples were coated
with ca. 5 nm platinum prior to measurements. Atomic force micrographs (AFM)
were collected with a Digital Instruments NanoScope IIIA multimode scanning
probe microscope. The cross-sectional structure of the films was observed using
a HITACHI H-800 transmission electron microscope. X-ray diffraction (XRD) pat-
terns were recorded on an X-ray diffractometer ((D/Max2500VB2+/PC, Rigaku,

Japan). X-ray photoelectron spectroscopy (XPS) measurements were performed on
an ESCALAB 250 spectrometer (Thermo Electron Corporation) with a take-off an-
gle of 45◦ .

3.1. Surface Reflectivity and Electrical Conductivity
Upon thermal treatment of the ion-exchanged precursor films under tension,
double-surface-silvered polyimide films were obtained. The development of the
metallic surfaces was followed by measuring the surface properties of the in-
dividual films as they were withdrawn from the oven at selected times and
temperatures. Figure 1 shows the variation of the surface reflectivity during the
thermal curing cycle for the polyimide/silver films derived from (a) BTDA/ODA-
AgNO3 , (b) BTDA/ODA-AgF, (c) ODPA/ODA-AgNO3 , (d) ODPA/ODA-AgF and
(e) PMDA/ODA-AgNO3 . The ion-exchange time and the concentration of the sil-
ver solution are given in each figure. The reflectivity curves indicate that all the
hybrid films were well-metallized with good reflectivity on both film surfaces after
curing at 300◦ C for several hours. The BTDA/ODA-AgNO3 film has achieved the
optimum reflectivity of 84%/71% on the upside/underside after at 300◦ C for 9 h
(Fig. 1(a)). The surface resistance data in Table 1 indicate that this film had been
metallized with surface resistance as low as 4.1/1.3 /sq on the upside/underside,
which is in the range of high electrical conductivity. What is more, when AgF is
used as the silver precursor, the BTDA/ODA polyimide films were metallized with
reflectivity over 80%/90% (Fig. 1(b)) and surface resistance of 0.6/0.2 /sq on the
upside/underside (Table 1). Thus, highly reflective and conductive films were ob-
tained via the present ion-exchange procedure.
However, in our work with ODPA/ODA as the matrix (Fig. 1(c) and 1(d)), the
films were prepared with only modest reflectivity (no higher than 72%), and con-
ductivity was never realized. Also, as shown in Table 1 and Fig. 1(e), although
excellent conductivity was obtained, the PMDA/ODA-AgNO3 films were never
metallized with highly reflective surfaces. It is clear that not all polyimide films
could be metallized with high quality. Metallized films with desirable performance
could only be achieved on certain polyimide matrices using appropriate silver pre-
cursors, as for the BTDA/ODA-AgF system.
The realization of surface silver metallization by employing simple silver com-
pounds is a great advantage over the in situ protocol, with which the films were often
prepared with only very poor reflectivity and/or were brittle and degraded without
mechanical usefulness. The present protocol differs from the in situ method in that
the silver was loaded into precursor films not by just physically blending the silver
compounds with the poly(amic acid) solution but via ion exchange to form silver
polyamate in the damp-dry poly(amic acid) films [32, 33]. It is suggested that the
formation of silver carboxylate groups during the ion-exchange process makes the
266 S. Qi et al.
(a) (b)
(c) (d)
(e)
Figure 1. Surface reflectivity as a function of the cure time/temperature for the metallized films
prepared using BTDA/ODA, ODPA/ODA and PMDA/ODA as the matrix and AgNO3 or AgF
as the silver precursor. (a) BTDA/ODA-AgNO3 , (b) BTDA/ODA-AgF, (c) ODPA/ODA-AgNO3 ,
(d) ODPA/ODA-AgF and (e) PMDA/ODA-AgNO3 . (Time zero is at 135◦ C for 1 h.)
silver reduction and film formation more favorable [9], resulting in metallized films
with good surface properties and mechanical properties.
The SEM images for the five polyimide/silver films are shown in Fig. 2. As dis-
played, well-defined silver layers were formed on both the upside and underside
Table 1.
Surface electrical resistance of metallized polyimide films
Film sample Thermal Surface resistance (/sq)

history
Upside Underside
BTDA/ODA-AgNO3 300◦ C–9 h 4.1 1.3

BTDA/ODA-AgF 300◦ C–4 h 0.6 0.2
ODPA/ODA-AgNO3 300◦ C–7 h >106 >106
ODPA/ODA-AgF 300◦ C–7 h >106 >106
PMDA/ODA-AgNO3 300◦ C–4 h 0.4 0.5
Figure 2. Scanning electron micrographs of the silvered polyimide films derived from
(a) BTDA/ODA-AgNO3 , (b) BTDA/ODA-AgF, (c) ODPA/ODA-AgNO3 , (d) ODPA/ODA-AgF and
(e) PMDA/ODA-AgNO3 . (U: upside of the composite film; D: underside of the composite film.)
of the metallized films. However, the surface topographies of the silvered films
are rather different from each other. SEM micrographs exhibit continuous net-like
silver surfaces for the films derived from BTDA/ODA- (Fig. 2(a) and 2(b)) and
PMDA/ODA- (Fig. 2(e)) based polyimides, which are consistent with their excel-
lent electrical conductivity. While for the ODPA/ODA-based polyimide, the films
(Fig. 2(c) and 2(d)) were metallized with only island-like silver aggregates sepa-
rated from one another on the surfaces, which accounts for the non-conductivity.
The ODPA/ODA polyimide differs from the BTDA/ODA only in replacing the car-
bonyl groups of BTDA with an ether linkage in ODPA. However, they produce
metallized films with completely different surface morphologies and surface prop-
erties. This implies that the polyimide structure has very significant influence on the
268 S. Qi et al.
film metallization process. Here, the nucleophilic and reactive carbonyl groups in
the BTDA, which are considered to be more favorable to silver reduction [34], are
suggested to be responsible for the superior surface properties and morphologies of
the BTDA/ODA-based films.
Also, SEM images in Fig. 2 indicate that the ligand/anion of the silver precur-
sor still plays a dominant role in the polyimide film metallization process. When
AgNO3 was employed, the ODPA/ODA polyimide was metallized with many uni-
form spherical silver particles distributing on the surfaces, as shown in Fig. 2(c).
Whereas, the ODPA/ODA-AgF film gives completely different metallic surfaces
on which the silver aggregates are present as lumps, as observed in Fig. 2(d). Such
huge effect of the ligand on film metallization is not expected since the ligand/anion
species is not supposed to be loaded into the precursor film in the ideal ion-exchange
process, as illustrated in Scheme 1. Further studies at this point are now being per-
formed.
The cross-sectional structures of the BTDA/ODA-AgF and the ODPA/ODA-
AgNO3 hybrid films are shown in Fig. 3. As can be clearly observed, a sandwich
(a)
(b)
Figure 3. Transmission electron micrographs of the polyimide/silver composite films prepared from
(a) BTDA/ODA-AgF cured to 300◦ C for 3 h and (b) ODPA/ODA-AgNO3 cured to 300◦ C for 7 h.
structure was formed with silver clusters mainly concentrated in the near-surface
silver layer and with only a few silver clusters in the inner bulk. For the conductive
BTDA/ODA-AgF films (Fig. 3(a)), TEM images indicate that compact silver layers
with thickness of 150–200 nm were formed on both sides after heating at 300◦ C
for 3 h. However, only thin (less than 50 nm) and non-uniform silver layers were
observed for the ODPA/ODA-AgNO3 films heated to 300◦ C for 7 h (Fig. 3(b)) and
the film never becomes conductive even after heating at 300◦ C for 14 h. The results
presented here clearly show that film properties and morphologies are sensitive to
both the chemical structure of the silver compound and the polyimide matrix.
3.2. Silver Aggregation
The plots in Fig. 1 indicate that the reflectivity of the silvered films increases
gradually with the thermal curing time and temperature but in a stepwise manner.
Actually, this is consistent with the silver reduction and the gradual silver aggrega-
tion to form silver layers on the film surfaces. Figure 4 shows the X-ray diffraction
Figure 4. X-ray diffraction patterns in silver 111, 200, 220, 311 and 222 regions for the
ODPA/ODA-AgNO3 composite films cured at different thermal stages.
270 S. Qi et al.
patterns for the ODPA/ODA-AgNO3 composite films during the whole thermal cy-
cle. After heating at 135◦ C for 1 h, XRD pattern exhibits no peaks for any silver
crystallites, and the hybrid films have the similar light yellow transparent appear-
ance of the pure ODPA/ODA precursor films. Whereas, the XRD patterns soon
show very distinct peaks characteristic of the crystalline state for the face-centered
cubic silver metal after the temperature was raised to 218◦ C. However, the strongest
characteristic reflection (111) at this point is very broad, indicating that the silver
crystallites generated at this early thermal stage were very small. Simultaneously,
the silver(I)-doped films present an appearance of dark blue–black and become
completely non-transparent. Further thermal treatment converts the non-transparent
films to blue, grass green and finally silvery. The frequent color changes at this ther-
mal stage are suggested to originate from the silver reduction and the continuous
silver aggregation to form silver surfaces. Meanwhile, XRD diffractograms in Fig. 4
show that the main peak in the (111) reflection region narrows rather rapidly and the
(200), (220), (311) and (222) peaks all become more and more distinct with increas-
ing cure time and temperature, indicating the formation of larger and more perfect
silver crystallites. Both color changes and XRD patterns suggest silver aggregation.
Further thermal treatment at 300◦ C induces further silver aggregation, as indi-
cated by the continuously narrowing peaks in the XRD patterns. Such continuous
silver aggregation at high temperature is very important, since in our present work
with all polyimides and silver precursors, to obtain more desirable surface reflectiv-
ity and conductivity, it was necessary that the metallized films were cured at 300◦ C
for certain times. Figure 5 displays the AFM images for the underside surfaces of
the BTDA/ODA-AgF hybrid films. For the samples cured at 300◦ C for 3 h, the
AFM image in Fig. 5(a) shows a continuous but rather irregular morphology and
the film was metallized with a surface resistance of 1.6 /sq. However, after further
thermal treatment at 300◦ C for only one more hour, the surface resistance of this
film was reduced to as low as 0.2 /sq. And as shown in Fig. 5(b), the film exhibits
rather different topography, i.e., a very flat surface without any distinct bulges. This
is supposed to be attributed to the silver aggregation that sinters the silver parti-
cles together on the surfaces to form the well-defined silver layers [32]. In addition,
it is suggested that thermal curing at high temperature would also induce metal-
catalyzed oxidative decomposition of the polyimide overlayer in the near-surface
layer making the silver surfaces more conductive [23, 33]. Figure 6 presents the
XPS Ag 3d core-level spectra for the BTDA/ODA-AgNO3 silvered films cured at
300◦ C for 1 h, 3 h and 9 h. As can be seen, the silver 3d peaks become more and
more intense with increasing cure time, indicating a gradual exposure of silver layer
on the surface, which is in accord with polyimide degradation.
The chemical nature of the silver surfaces on the final metallized polyimide films
is always of primary concern, since thermal treatment might convert the silver(I)
ions into many kinds of silver species. However, XRD diffractograms in our work
indicate that the silver is mainly present in the form of FCC silver crystallites. More-
over, XPS Ag 3d core-level spectra with a silver 3d5/2 peak at 368.2 eV and a silver
(a) (b)
Figure 5. AFM images of the underside surfaces of the silvered BTDA/ODA polyimide
films ion-exchanged in 0.1 M aqueous AgF solution for 20 min and cured to 300◦ C for
(a) 3 h (reflectivity = 92.4%, resistance = 1.6 /sq) and (b) 4 h (reflectivity = 89.77%,
resistance = 0.2 /sq).
Figure 6. XPS Ag 3d core-level spectra for the BTDA/ODA-AgNO3 silvered films cured at 300◦ C
for 1 h, 3 h and 9 h. (Measurements were performed on the underside.)
3d3/2 peak at 374.2 eV suggest that the silver on the final metallized film surface is
mainly present as the native metal [35].
3.3. Thermal and Mechanical Properties
In the present method, due to the existence of the weak and hydrolysable amide
groups in the poly(amic acid) molecules, the direct utilization of damp-dry PAA
film to perform ion exchange is adventurous, since the PAA chains might be seri-
ously hydrolyzed during ion exchange in the aqueous silver(I) solution. In addition,
the oxidative and catalytic effects of doped silver clusters might also decompose
the polyimide matrix and, consequently, produce metallized films with substan-
tially degraded mechanical properties. However, the characterization results suggest
272 S. Qi et al.
that although the percent elongations of the hybrid films were reduced after film
metallization, the tensile strength and modulus of the metallized films were never
seriously diminished, as reported previously [32, 33, 36, 37]. This is not unexpected
since the sandwich film morphology with only a few silver clusters in the inner film
bulk, as shown in the TEM images of Fig. 3, makes it possible for the metallized
films to retain the essential mechanical properties of the pure polyimide film.
Thermal characterization results [32, 33, 36, 37] also suggest that the thermal
stability of the silver-doped films, as characterized by the 10%-weight-loss temper-
ature, did not decrease in a nitrogen environment. However, in air, the degradation
temperatures are considerably decreased (120–160◦ C lower) as compared to the
host polyimide films, which is mainly attributed to strong catalytic and oxidative
degradation effects of silver on the polymeric matrix. Nevertheless, the remaining
thermal properties of the composite films still make it more than adequate for many
applications.
4. Conclusions
We have prepared reflective and/or conductive double-surface-silvered polyimide
films via the direct ion-exchange self-metallization process using simple silver
compounds (AgF and AgNO3 ) as the silver precursor. Both polyimide structure
and silver species play an important role in the film metallization process. One
cannot simply take any silver compound with a poly(amic acid) and obtain metal-
lized films of high quality. Metallized films with desirable performance could only
be achieved on certain polyimide matrices using proper silver precursors. In the
present work with BTDA/ODA, ODPA/ODA and PMDA/ODA as the matrices and
AgNO3 and AgF as the silver precursors, the BTDA/ODA-AgF system proved to be
the most efficient to obtain highly reflective and conductive polyimide/silver films.
Although the polyimide precursor might be hydrolyzed during ion-exchange and
the polyimide matrix might be decomposed under the catalytic effect of silver dur-
ing thermal treatment, the final metallized films maintain the major mechanical and
thermal properties of the pristine polyimide.
Acknowledgements
The authors thank the financial supports from the National High Technology Re-
search and Development Program (Project No. 2007AA03Z537), the National Nat-
ural Science Foundation of China (NSFC, Project No. 50573007) and the program
for the New Century Excellent Talents in University (NCET-040118).
References
1. K. Akamatsu, S. Ikeda, H. Nawafune and S. Deki, Chem. Mater. 15, 2488 (2003).
2. K. Akamatsu, H. Shinkai, S. Ikeda, S. Adachi, H. Nawafune and S. Tomita, J. Am. Chem. Soc.
127, 7980 (2005).
3. M. Charbonnier, Y. Goepfert and M. Romand, in: Polyimides and Other High Temperature Poly-
mers: Syntheis, Characterization and Applications, K. L. Mittal (Ed.), Vol. 2, pp. 289–314. VSP,
Utrecht (2003).
4. N. N. Kariuki, J. Luo, S. A. Hassan, I.-I. S. Lim, L. Wang and C. J. Zhong, Chem. Mater. 18, 123
(2006).
5. S. Matsuda, Y. Yasuda and S. Ando, Adv. Mater. 17, 2221 (2005).
6. R. Murdey and J. T. Stuckless, J. Am. Chem. Soc. 125, 3995 (2003).
7. C. C. Perry, S. R. Carlo, J. Torres, A. J. Wagner and D. H. Fairbrother, in: Polyimides and Other
High Temperature Polymers: Syntheis, Characterization and Applications, K. L. Mittal (Ed.),
Vol. 2, pp. 345–358. VSP, Utrecht (2003).
8. K. Akamatsu, S. Ikeda and H. Nawafune, Langmuir 19, 10366 (2003).
9. D. Andreescu, A. K. Wanekaya, O. A. Sadik and J. Wang, Langmuir 21, 6891 (2005).
10. L. L. Beecroft and C. K. Ober, Chem. Mater. 9, 1302 (1997).
11. S. Qi, W. Wang, D. Wu, Z. Wu and R. Jin, Eur. Polym. J. 42, 2023 (2006).
12. S. Qi, D. Wu, Z. Wu, W. Wang and R. Jin, Polymer 47, 3150 (2006).
13. S. Rifai, C. A. Breen, D. J. Solis and T. M. Swager, Chem. Mater. 18, 21 (2006).
14. Y. Tachibana, K. Kusunoki, T. Watanabe, K. Hashimoto and H. Ohsaki, Thin Solid Films 442, 212
(2003).
15. W. W. Zhao and F. J. Boerio, Surf. Interface Anal. 26, 316 (1998).
16. R. E. Southward and D. M. Stoakley, Prog. Org. Coat. 41, 99 (2001).
17. J. D. Warner, M. Pevzner, C. J. Dean, D. E. Kranbuehl, J. L. Scott, S. T. Broadwater, D. W. Thomp-
son and R. E. Southward, J. Mater. Chem. 13, 1847 (2003).
18. R. K. Boggess, L. T. Taylor, D. M. Stoakley and A. K. St. Clair, J. Appl. Polym. Sci. 64, 1309
(1997).
19. J. Rosolovsky, R. K. Boggess, A. F. Rubia, L. T. Taylor, D. M. Stoakley and A. K. St. Clair, J.
Mater. Res. 12, 3127 (1997).
20. A. F. Rubira, J. D. Rancourt and L. T. Taylor, in: Metal-Containing Polymeric Materials,
C. U. Pittman, C. E. Carraher, B. M. Culbertson, M. Zeldin and J. E. Sheets (Eds), pp. 357–368.
Plenum Press, New York, NY (1996).
21. A. F. Rubira, J. D. Rancourt, L. T. Taylor, D. M. Stoakley and A. K. St. Clair, J. Macromol. Sci. —
Pure Appl. Chem. A35, 621 (1998).
22. T. Sawada and S. Ando, Chem. Mater. 10, 3368 (1998).
23. R. E. Southward and D. W. Thompson, Adv. Mater. 11, 1043 (1999).
24. S. Kudaikulova, O. Prikhodko, G. Boike, B. Zhubanov, V. Y. Voytekunas and M. J. M. Abadie, in:
Polyimides and Other High Temperature Polymers: Syntheis, Characterization and Applications,
K. L. Mittal (Ed.), Vol. 2, pp. 389–406. VSP, Utrecht (2003).
25. R. E. Southward, D. S. Thompson, D. W. Thompson and A. K. St. Clair, Chem. Mater. 11, 501
(1999).
26. R. E. Southward and D. W. Thompson, Mater. Design 22, 565 (2001).
27. M. Danziger and W. Voitus, in: Polyimides and Other High Temperature Polymers: Syntheis, Char-
acterization and Applications, K. L. Mittal (Ed.), Vol. 2, pp. 315–329. VSP, Utrecht (2003).
28. J. Friedrich, G. Kuhn, R. Mix, I. Retzko, V. Gerstung, S. Weidner, R. D. Schulze and W. Unger, in:
Polyimides and Other High Temperature Polymers: Syntheis, Characterization and Applications,
29. E. Ranucci, A. Sandgren, N. Andronova and A.-C. Albertsson, J. Appl. Polym. Sci. 82, 1971
(2001).
30. R. E. Southward and D. W. Thompson, Chem. Mater. 16, 1277 (2004).
274 S. Qi et al.
31. R. E. Southward, D. S. Thompson, D. W. Thompson, M. L. Caplan and A. K. St. Clair, Chem.

Mater. 7, 2171 (1995).
32. S. Qi, Z. Wu, D. Wu, W. Wang and R. Jin, Langmuir 23, 4878 (2007).
33. S. Qi, Z. Wu, D. Wu, W. Wang and R. Jin, Chem. Mater. 19, 393 (2007).
34. C. Yang, J. S. Chen and S. L. C. Hsu, J. Vac. Sci. Technol. A 23, 862 (2005).
35. G. Beamson and D. Briggs, High Resolution XPS of Organic Polymers: The Scienta ESCA300
Database. Wiley, Chichester, UK (1992).
36. S. Qi, D. Wu, Z. Bai, Z. Wu, W. Yang and R. Jin, Macromol. Rapid Commun. 27, 372 (2006).
37. R. Jin, S. Qi, Z. Wu and D. Wu, PMSE Preprints 97, 926 (2007).
Adhesion and Tribological Properties of Sputtered Polymer
Thin Films from a Polyimide Target
Satoru Iwamori ∗ , Akihiro Uemura and Kazuya Kezuka

Graduate School of Natural Science & Technology, Kanazawa University Kakuma-machi,
Kanazawa City 920-1192, Japan
Abstract
Polymer thin films were sputtered from a polyimide, biphenyl dianhydride-p-phenylene diamine (BPDA-
PDA), target onto a copper substrate by conventional RF sputtering with argon (Ar) and nitrogen (N2 ).
These polymer thin films (sputtered PI thin films with Ar and N2 ) were characterized, and their adhesion
and tribological properties were evaluated.
Nitrogen content in the sputtered PI thin film with N2 increased compared to that of the sputtered PI thin
film with Ar and target material (pristine). Surface roughness of the sputtered polymer thin film with N2
showed higher values than that of the polymer thin film with Ar.
Friction coefficients of the sputtered PI thin films with Ar were almost the same as that of the pristine PI,
however, that of the sputtered PI thin film with N2 was higher than that of the pristine PI. The wear life of
the sputtered PI thin film with N2 was much higher than that of the sputtered PI thin film with Ar.
The adhesion strength between the sputtered PI thin film with N2 and copper substrate was higher than
that between the thin film sputtered with Ar and copper substrate. In addition, the sputtered PI thin film with
N2 was introduced between the copper substrate and the sputtered PI thin film with Ar (Ar/N2 laminate).
The adhesion strength of this laminate was higher than that between the sputtered PI thin film with Ar and
copper substrate.
Keywords
Polyimide, thin film, copper substrate, RF sputtering, adhesion strength, friction coefficient, wear life
1. Introduction
Sputtering is widely used in electrical and mechanical industries, because a sput-
tered thin film has a uniform structure and an excellent adhesion property to most
substrates. Polymer thin films such as polyethylene, polypropylene, polyimide (PI)
and poly(tetrafluoroethylene) (PTFE) have been sputtered since the 1970s [1–13]
and their tribological properties have been characterized [2, 5, 10]. Sputtering rate
of PI is known to be lower than those of the other polymers, namely, polyethylene,
*To whom correspondence should be addressed. Tel.: +81-76-234-4950; Fax: +81-76-234-4950; e-mail:
iwamori@t.kanazawa-u.ac.jp
276 S. Iwamori et al.
poly(methyl methacrylate) (PMMA), nylon and PTFE. Friction coefficient and wear
life of the sputtered PI thin film are known to be superior to those of the sputtered
inorganic thin films, e.g., boron nitride, titanium nitride and molybdenum disul-
fide thin films [5]. As the tribological properties of the PTFE are known to be only
slightly influenced by humidity, a PTFE thin film can be used as a solid lubricant.
The wear life of the PTFE thin film was found to be shorter than that of the PI,
because the PTFE thin film had poor adhesion property to the substrate [2].
We characterized the polymer thin films prepared by RF sputtering with two
kinds of PI targets, PMDA-ODA and BPDA-PDA, and evaluated their adhesion
strength to a copper substrate, as well as their tribological properties [14, 15]. In
this paper, we discuss relationships between adhesion, tribological properties, and
structures of the polymer thin films prepared by RF sputtering from the BPDA-PDA
target.
2. Experimantal
Upilex-S™ (BPDA-PDA: 125 µm thick, Ubekosan, Japan), PI was used as the sput-
tering target. Figure 1 shows the molecular structure of the PI target. Glass slides
and copper plates (60 mm square, 2.5 mm thick) were used as the substrates. Glass
slide substrates were used for surface structure and chemical bonding states analy-
ses of the polymer thin films prepared by RF sputtering. The copper substrates were
used for evaluation of the adhesion and tribological properties. The copper plates
were polished with different grades of sandpaper (#400, #800, #2000). The arith-
metical mean roughness (Ra ) of the copper surfaces was measured using a surface
roughness measurement instrument, Surfcom 1400A-6™ (Tokyo Seimitsu, Inc.,
Japan). The Ra value was 50 nm after the polishing. These polished substrates were
washed with distilled water and ultrasonically cleaned in acetone.
Polymer thin films were sputtered onto these substrates from the PI target
with argon gas. After the sputtering chamber was evacuated to a pressure of
1.0 × 10−5 Torr, the pressure in the chamber was maintained between 2 and
80 mTorr by adjusting the main valve installed in the sputtering system. The thick-
ness of these polymer thin films was determined by measuring the height between
the film and glass slide with a profilometer, Surfcom 1400A-6™ (Tokyo Seimitu,
Figure 1. Molecular structure of PI target.

Adhesion and Tribological Properties of Sputtered Polymer Thin Films 277
Inc., Japan). The thickness of the thin films used for evaluation of wear life and
adhesion strength was 0.5–1.0 µm.
Surface morphologies of these polymer thin films were analyzed by scanning
probe microscopy (SPM) (SPM-9600: Shimadzu Corporation, Japan). The elemen-
tal compositions and chemical bonding states of these polymer thin films were
determined by X-ray photoelectron spectroscopy (XPS) (Quantum 2000: Ulvac
— Phi Co. Ltd, Japan) using Al Kα radiation with an energy of 1486.6 eV. The
chemical bonding states were also analyzed with Fourier transform-infrared (FT-
IR) spectroscopy measurements. Polymer thin films were peeled off from the glass
slide substrate, and mixed with KBr. These mixtures were used for analyses with the
FT-IR spectroscopy (Model 610: JASCO International Co. Ltd, Japan). Absorption
in the range 2100–600 cm−1 was measured.
A pin-on-disk type friction and wear test apparatus with a bearing steel ball
(2 mm in diameter) as the slider was used for evaluation of friction coefficient and
wear life [10, 14]. The speed of the disk was 5 revolutions/min and revolution diam-
eter of the pin was 8 mm. The load on the bearing steel ball was 20 g for measuring
friction coefficient, and 50 g or 100 g for evaluation of wear life. When a polymer
thin film is worn or peels off from the copper substrate, an electric current flows
between the pin and the substrate [14, 15].
The adhesion strength between the polymer thin films and copper substrate was
measured with two methods. A stud (6 mm in diameter) was bonded to the polymer
thin films with an epoxy resin. The adhesion strength was determined by pulling
the stud using Tensilon™ (Touyou Sokuki, Inc., Japan) [14, 15]. The pull speed
was 0.01 mm/min. This is a conventional method to determine adhesion strength,
however it is impossible to determine it when the adhesion strength between the
thin film and substrate is higher than that between the stud and thin film. Another
method to measure the adhesion strength between the polymer thin film and cop-
per substrate was SAICAS (Surface and Interface Cutting Analysis System: Daipla
Wintes Co. Ltd, Japan). Figure 2 shows a schematic diagram of SAICAS. Vertical
and horizontal reaction forces at a single crystal diamond blade during cutting of the
thin film were monitored. After the tip of the blade reaches the interface between
the thin film and substrate, it moves in the horizontal direction. This force measures
Figure 2. Schematic diagram of the SAICAS.

the adhesion strength between the thin film and substrate. The adhesion strength, P,
can be represented as follows [16]:
P = FH /w (N/m), (1)
FH and w denote the horizontal reaction force and width of the blade, respectively.
3. Structure Analyses of Sputtered PI Thin Films

Figure 3 shows SPM images of the sputtered PI thin films with Ar and N2 . Surface
roughnesses of the PI thin films with Ar and N2 were 2.9 nm and 10.5 nm, respec-
tively. Table 1 shows elemental composition of these polymer thin films. Carbon
content of the PI thin films with Ar increased, but oxygen and nitrogen contents
decreased compared to those of the pristine PI. Nitrogen content of PI thin film sput-
tered with N2 dramatically increased. Figure 4(a) and 4(b) shows C1s XPS spectra
of the sputtered PI thin films with Ar and N2 , respectively [15]. Peak intensity of
peak 2, which contains C–N and C–O bonds, in Fig. 4(b) increased considerably
compared to that in Fig. 4(a). Figure 5 shows FT-IR spectra of the pristine PI, and
sputtered PI thin films with Ar and N2 [17]. Absorption peaks at 1780, 1720 and
1380 cm−1 represent the imide moiety, and the peak at 1500 cm−1 represents the
benzene ring in the PI structure. These peaks can be clearly recognized only in the
pristine PI spectrum. However, a large and broad peak around 1000–1800 cm−1
can be found in the spectrum of the sputtered PI thin film with N2 . Although peak
intensity of the sputtered PI thin film with Ar is lower than that of the sputtered thin
film with N2 , a broad peak can also be found. These peaks around 1000–1800 cm−1
mean that there are wide varieties of chemical bonding states in these thin films. C1s
XPS spectra of these polymer thin films, as shown in Fig. 4, were deconvoluted into
three peaks, peak 1–3. However, in fact, these peaks contain various chemical bond-
ings, in addition to CH2 (peak 1), C–N, C–O (peak 2) and C=O (peak 3) shown in
Fig. 4. Surface free energies of these PI thin films were calculated from the contact
angles of water and methylene iodide (Fig. 6) [15]. The surface free energy of the
sputtered PI thin film with Ar decreased compared to that of the pristine PI, how-
ever that of the sputtered thin film with N2 was almost of the same as that of the
pristine PI.
4. Adhesion and Tribological Properties of Sputtered PI Thin Films

The friction coefficients of the PI thin films sputtered onto the copper substrate were
measured (Table 2) [15]. The friction coefficient of the sputtered PI thin film with
Ar was almost the same as that of the pristine PI. However, that of the sputtered
thin film with N2 was higher than the pristine PI. Figure 7 shows the extent of wear
after 50 and 300 revolutions. The wear life of the sputtered PI thin film with N2 was
much higher than that of the sputtered thin film with Ar.
The pull strength between the sputtered PI thin films and copper substrate was
evaluated with a stud (6 mm in diameter) bonded to the thin films with an epoxy
(a)
(b)
Figure 3. SPM images of the PI thin films sputtered with Ar (a) and N2 (b).
Table 1.
Elemental composition (at%) of sputtered PI thin films
and pristine PI
C O N
Pristine 79 14 7
Ar-sputter 83 12 5
N2 -sputter 60 10 30
(a)
(b)
Figure 4. C1s XPS spectra of the PI thin films sputtered with Ar (a) and N2 (b).
resin (Fig. 8) [15]. The dotted line in Fig. 8 means the pull strength between the
adhesive (epoxy resin) and thin films. The pull strength between the sputtered PI
thin film with N2 and copper substrate was higher than that between the sputtered
PI thin film with Ar and copper substrate. Figure 8 also shows pull strength of
the sputtered PI thin film with Ar and N2 mixture (Ar + N2 ). It was also higher
than that of the sputtered thin film with Ar alone. In order to enhance the pull
Figure 5. FT-IR spectra of the PI thin films sputtered with Ar and N2 and pristine PI.
Figure 6. Surface free energy and its components for the PI thin films sputtered with Ar and N2 and
pristine PI (target).
Table 2.
Friction coefficient of pristine PI and sputtered PI thin
films
Friction coefficient
BPDA-PDA (pristine) 0.15

BPDA-PDA (Ar-pristine) 0.16
BPDA-PDA (N2 -pristine) 0.40
Figure 7. Extent of wear of polymer thin films after 50 and 300 revolutions. Hatched and black bars
show the extent after 50 revolution and 300 revolutions, respectively.
Figure 8. Pull strength between PI thin films sputtered with Ar and N2 and copper substrate. Dotted
line shows the adhesion strength between the stud and these thin films.
strength between the sputtered thin film with Ar and copper substrate, sputtered PI
thin film with N2 was introduced between the sputtered thin film with Ar and copper
substrate (Ar/N2 laminate). The pull strength of this laminate showed higher value
than that between the sputtered thin film with Ar and copper substrate [15].
In order to evaluate adhesion strength between the sputtered PI thin film with N2
and copper substrate, it was measured with the SAICAS as shown in Fig. 2. The
adhesion strength of the sputtered thin film with N2 was almost two times higher
than that sputtered with Ar (Fig. 9) [17]. The surface free energy of the sputtered PI
thin film with N2 was higher than that of the thin film with Ar as shown in Fig. 6.
In addition, the surface roughness of the sputtered PI thin film with N2 was higher
than that of the sputtered PI thin film with Ar as shown in Fig. 3. This high adhesion
strength could be related to the high surface free energy imparted by hydrophilic
groups and the anchor effect provided by the surface roughness.
Figure 9. Adhesion strength of the PI thin films sputtered with Ar and N2 to copper substrate measured
with the SAICAS.
5. Conclusions
Polymer thin films were prepared by r.f. sputtering from a polyimide target, and
adhesion and tribological properties were evaluated.
The following conclusions were drawn from this study.
• The adhesion strength between the sputtered PI thin film with N2 and copper
substrate was almost two times higher than that between the sputtered thin film
with Ar and copper substrate. The adhesion strength between the sputtered thin
film with Ar and copper substrate was enhanced due to the introduction of the
sputtered PI thin film with N2 between the sputtered thin film with Ar and
copper substrate.
• Friction coefficient of the PI thin film sputtered with Ar was almost the same as
that of the pristine PI, however, that of the PI thin film sputtered with N2 was
higher. The wear life of the sputtered PI thin film with N2 was much longer than
that of the sputtered thin film with Ar.
Acknowledgements
The authors would like to thank Mr. Itsuo Nishiyama of Daipla Wintes Co. Ltd
for adhesion strength measurements of sputtered thin films with SAICAS and
Ms. Nobuko Edo of Shimadzu Analytical & Measuring Center, Inc. for SPM analy-
ses of the sputtered thin films.
References
1. H. Biederman, S. M. Ojha and L. Holland, Thin Solid Films 41, 329 (1977).
2. Y. Yamada, K. Tanaka and K. Saitoh, Surface Coatings Technol. 43–44, 618 (1990).
3. N. Marechal and Y. Pauleau, J. Vac. Sci. Technol. A10, 477 (1992).
4. H. Biederman, Vacuum 31, 285 (1995).
5. M. Kitoh and Y. Honda, Thin Solid Film 271, 92 (1995).
6. G. A. Hishmeh, T. L. Barr, A. Sklyarov and S. Hardcastle, J. Vac. Sci. Technol. A14, 1330 (1996).
7. H. Biederman, Vacuum 59, 594 (2000).

8. H. Biederman, M. Zeuner, J. Zalman, P. Bilkova, V. Stelmasuk and A. Boldyreva, Thin Solid Films
392, 208 (2001).
9. I. Kholodkov, H. Biederman, D. Slavínská, A. Choukourov and M. Trchova, Vacuum 70, 505
(2003).
10. Y. Nagayama, S. Iwamori and Y. Yamada, J. Vac. Soc. Jpn 46, 827 (2003).
11. V. Stelmashuka, H. Biederman, D. Slavínskáa, M. Trchováa and P. Hlidek, Vacuum 75, 207 (2004).
12. S. Iwamori, Y. Yamada and K. Ikeda, Mater. Sci. Technol. 42, 54 (2005).
13. Y. Yamada, S. Iwamori and Y. Takahashi, J. Vac. Soc. Jpn 47, 574 (2004).
14. S. Iwamori, A. Uemura and Y. Yamada, J. Adhesion Sci. Technol. 18, 1771 (2004).
15. S. Iwamori and A. Uemura, J. Adhesion Soc. Jpn 42, 89 (2006).
16. I. Nishiyama and M. Saito, J. Printing Sci. Technol. Jpn 42, 285 (2005).
17. S. Iwamori, K. Kezuka and A. Uemura, Molecular Crystals Liquid Crystals 471, 99 (2007).
Adhesion and Mechanical Properties of
Poly(tetrafluoroethylene)–Poly(vinyl alcohol) Composite
Thin Films Prepared by Spin Coating Method
Satoru Iwamori a,∗ , Masaaki Mizoguchi a and Itsuo Nishiyama b

a
Graduate School of Natural Science & Technology, Kanazawa University Kakuma-machi,
Kanazawa City 920-1192, Japan
b
DAIPLA WINTES Co., Ltd, 1-3-26 Nagasu Nishidouri, Amagasaki City 660-0807, Japan
Abstract
Poly(tetrafluoroethylene) (PTFE) and poly(vinyl alcohol) (PVA) and their composite thin films coated onto
a nickel–titanium substrate by spin coating were characterized. Relationship between the sintering tem-
perature and mechanical properties, such as friction coefficient, adhesion strength and shearing stress, was
evaluated. The friction coefficient of PTFE thin films slightly decreased with increase of the sintering tem-
perature, however, that of the PVA thin films increased. Friction coefficient of the PTFE–PVA thin films was
almost the same as that of the PTFE thin film. The adhesion strength and shearing stress of the PTFE–PVA
composite and PVA thin films decreased with increasing sintering temperature. Higher sintering temper-
atures would cause decomposition of the PVA molecules and influence the friction coefficient, adhesion
strength and shearing stress of the PTFE–PVA composite and PVA thin films.
Keywords
Poly(tetrafluoroethylene) (PTFE), poly(vinyl alcohol) (PVA), composite thin film, adhesion strength, shear-
ing stress, friction coefficient, spin coating
1. Introduction
Fluorocarbon polymers have been used for various applications due to their excel-
lent thermal and chemical stabilities and tribological properties. Surfaces of metal,
inorganic and organic material substrates can be modified by coating with PTFE
thin films. However, PTFE thin films coated onto metal and inorganic material sub-
strates often show poor adhesion. Coating techniques for a stronger adhesion are
definitely needed.
*To whom correspondence should be addressed. Tel./Fax: +81-76-234-4950; e-mail:

iwamori@t.kanazawa-u.ac.jp
(a) (b)
Figure 1. Molecular structures of PTFE (a) and PVA (b).
Spin coating is one of sol–gel coating methods, and is widely used in the coat-
ing of inorganic and organic materials, e.g., tetraethylorthosilicate (TEOS) [1],
manganese oxide [2], tin oxide [3], semiconducting polymers [4], and PTFE [5].
In order to enhance the lubricating property of nickel–titanium (Ni–Ti) substrate,
a PTFE thin film was coated by a conventional spin coater. PTFE thin film coated
onto nickel–titanium (Ni–Ti) substrate by spin coating showed low friction coef-
ficient, however adhesion strength between the thin film and Ni–Ti substrate was
so low that it was easily peeled off from the substrate. In order to enhance the ad-
hesion strength, a hydrophilic polymer, poly(vinyl alcohol) (PVA) was mixed with
the PTFE dispersion, and introduced between the Ni–Ti substrate and PTFE thin
film by spin coating [6]. The introduction of PVA thin film between the substrate
and PTFE thin film was effective in enhancing adhesion. Moreover, introduction of
functionally graded thin films composed of PTFE and PVA was more effective for
the enhancement of adhesion [6].
In this paper, we report on the relationship between sintering temperatures of
these thin films and their mechanical properties i.e., friction coefficient, adhesion to
the Ni–Ti substrate and shearing stress of the films.
2. Experimental
Figure 1 shows molecular structures of PTFE and PVA used in this study. A PTFE
dispersion (POLYFLON® PTFE: Daikin Industries, Co., Ltd, Japan) and PVA resin
(POVAL® : Kuraray Co., Ltd, Japan) which was dissolved in distilled water were
used for spin coating. PTFE particles, whose diameter ranged from 0.2 to 0.4 µm,
were dispersed in distilled water, and the concentrations of the PTFE dispersion and
PVA solution were 60% and 8% (w/w), respectively. Figure 2 shows the schematic
diagram of the spin coater. The PTFE thin films were coated onto nickel (Ni)–
titanium (Ti) substrates by the spin coater. After coating, they were cured at tem-
peratures in the range from 100◦ C to 250◦ C for 30 min. The thickness of all the
thin films was in the range of 1.0 to 2.0 µm.
The elemental compositions and chemical bonding states of these thin films were
determined by X-ray photoelectron spectroscopy (XPS) (Quantum2000: Phi Co.,
Ltd) using Al Kα radiation with an energy of 1486.6 eV.
PTFE–PVA Composite Thin Films Prepared by Spin Coating Method 287
Figure 2. Schematic diagram of the spin coater.
Contact angles of water were determined by taking photographs of the droplets

on the PVA thin films. The volume of each droplet used was 20 µl. The measure-
ments were performed in the open air.
A pin-on-disk type friction test apparatus with a steel ball (2 mm in diameter) as
the slider was used for determination of friction coefficient [7].
Shearing stress is one of the important mechanical properties of thin films,
which indicates brittleness of the thin film. Adhesion strength between these thin
films and Ni–Ti substrate and shearing stress of these thin films were measured by
SAICAS (Surface and Interface Cutting Analysis System: DAIPLA WINTES Co.,
Ltd, Japan). Vertical and horizontal reaction forces at a single crystal diamond blade
during cutting the thin film were monitored. Figure 3(a) and 3(b) shows a schematic
diagram of the SAICAS and a typical profile of the reaction forces of the SAICAS.
The shearing stress can be calculated by measuring the horizontal reaction force
during cutting the thin films (cutting mode). After the tip of the blade reaches to
the interface between the thin film and substrate, it moves only in the horizontal
direction (peel mode). The thin films are peeled off from the substrates due to the
tip of the blade moving in the horizontal direction. This force means the adhesion
strength between the thin film and substrate (Fig. 3(b)). The adhesion strength, P ,
and shearing stress, τ , can be represented as follows [8]:
P = FH /w (N/m), (1)
τ = FH /(2wd cot φ) (MPa), (2)
FH , w, d and φ denote the horizontal reaction force, width of the blade, coating
thickness and shear angle, respectively. In this experiment, we assumed φ as 45◦
and calculated τ according to (2).

Table 1 shows elemental compositions of the PTFE, PTFE–PVA composite and
PVA thin films prepared by spin coating at sintering temperatures of 150◦ C and
250◦ C. Mixing ratio (wt/wt) of the PTFE and PVA in the PTFE–PVA composite
thin film was 3:7. The PTFE thin film sintered at 150◦ C contained oxygen. The
oxygen is considered to originate from water in the PTFE dispersion. The PTFE
thin film sintered at 250◦ C and PVA thin film sintered at 150◦ C had almost the same
(a)
(b)
Figure 3. (a) Schematic diagram of SAICAS and (b) a typical profile of the reaction forces of the
SAICAS.
elemental compositions as those of pristine PTFE and PVA, respectively. Elemental

ratio, oxygen to carbon, in the PVA thin film sintered at 250◦ C decreased compared
with that in the film sintered at 150◦ C. Figure 4 and Table 2 show C1s XPS spectra
of the PVA thin films sintered at 150◦ C and 250◦ C, and area (%) of peaks from the
C1s XPS spectra, respectively. Peak intensity of the ether group, C–O, decreased
and that of carbonyl group, C=O, slightly increased in the PVA thin film sintered at
250◦ C compared with the values in the PVA thin film sintered at 150◦ C. It is known
that PVA is decomposed on heating over 200◦ C [9]. Table 3 shows contact angles of
water droplets on the PVA thin films sintered at various temperatures. The contact
angle increased with increase of temperature. These results mean that hydrophilic
groups, i.e., hydroxyl groups, in the PVA molecules decrease due to the sintering.
The hydroxyl group in the PVA would be decomposed, and carbonyl group, C=O,
produced as shown in Fig. 5. Fluorine content increased and oxygen content de-
creased in the PTFE–PVA composite thin film sintered at 250◦ C compared with
those in the PTFE–PVA composite thin film sintered at 150◦ C.
Table 1.
Elemental composition (at%) of PTFE, PTFE–PVA composite and
PVA thin films sintered at 150◦ C and 250◦ C
Sample Temperature (◦ C) C O F
PTFE 150 41 9 50
250 34 0 66
PTFE–PVA 150 68 27 5
250 62 18 21
PVA 150 68 32 0
250 76 24 0
Figure 4. C1s XPS spectra of PVA thin films sintered at 150◦ C (a) and 250◦ C (b).
Table 2.
Areas (%) of peaks from C1s XPS spectra of the PVA
thin films sintered at 150◦ C and 250◦ C
–(CH2 )n – –CO– C=O O–C=0
150◦ C 31 32 1 4
250◦ C 46 20 3 7
Table 3.
Contact angle of water on the PVA thin films sintered at
various temperatures
Sintering temperature (◦ C) 150 180 200 250

Contact angle (◦ ) 47 48 55 60
Figure 5. Decomposition of PVA by heating over 200◦ C.
Figure 6. Friction coefficient of the PTFE, PTFE–PVA composite and PVA thin films sintered at
various temperatures.
Figure 6 shows friction coefficient of the PTFE, PTFE–PVA composite and PVA
thin films sintered at various temperatures. The friction coefficient of PTFE thin
films slightly decreased with increase of sintering temperature, however that of the
PVA thin films increased. It is noteworthy that friction coefficient of the PTFE–
PVA thin films was almost of the same level as that of the PTFE. It was difficult
to measure the friction coefficient of PTFE thin film sintered at 100◦ C, because the
thin film easily peeled off from the Ni–Ti substrate.
Figure 7 shows the adhesion strength between these thin films sintered at 100◦ C,
180◦ C and 250◦ C and the Ni–Ti substrate as measured by the SICAS. The high-
est adhesion strength was for the PVA thin film, and the second highest was for
the PTFE–PVA composite thin film at every sintering temperature. The adhesion
strength of PTFE–PVA composite and PVA thin films decreased with increase of
sintering temperature. The hydrophilic groups such as hydroxyl group in the PVA
molecules decreased and carbonyl groups increased due to the sintering. We think
the decrease of the adhesion strength is caused by the decomposition of PVA.
Figure 8 shows the shearing stress of these thin films sintered at 100◦ C, 180◦ C
and 250◦ C as measured by the SICAS. The highest shearing stress was for the PVA
Figure 7. Adhesion strength of the PTFE, PTFE–PVA composite and PVA thin films sintered at
100◦ C, 150◦ C and 250◦ C to Ni–Ti substrates.
Figure 8. Shearing stress of the PTFE, PTFE–PVA composite and PVA thin films sintered at 100◦ C,
150◦ C and 250◦ C to Ni–Ti substrates.
thin film, and the second highest was for the PTFE–PVA composite thin film at
every sintering temperature. The shearing stress of PTFE–PVA composite and PVA
thin films decreased with increase of sintering temperature as was the case with the
adhesion strength shown in Fig. 7. However, the shearing stress of the PTFE thin
films slightly increased with increase of the temperature. The shearing stress of the
PTFE thin films was so low that it may be beyond the detection limit in the SICAS.
4. Conclusion
Friction, adhesion and shearing stress of PTFE, PTFE–PVA composite and PVA
thin films coated onto the nickel–titanium substrate by spin coating were character-
ized by a pin-on-disk type friction test apparatus and SICAS.
The following conclusions were drawn from this study.
(1) The friction coefficient of PTFE thin films slightly decreased with increase of
sintering temperature, however that of the PVA thin films increased. Friction
coefficient of the PTFE–PVA thin films was almost of the same value as that of
the PTFE.
(2) The adhesion strength of PTFE–PVA composite and PVA thin films decreased
with increase of the sintering temperature.
(3) The shearing stress as well as the adhesion strength of PTFE–PVA composite
and PVA thin films decreased with increase of sintering temperature. Higher
sintering temperatures would cause decomposition of the PVA molecules and
influence the adhesion strength and shearing stress of the PTFE–PVA compos-
ite and PVA thin films.
References
1. A. Kumar, R. R. Pandey and B. Brantley, Talanta 69, 700 (2006).
2. S. Ching, S. M. Hughes, T. P. Gray and E. J. Welch, Microporous Mesoporous Mater. 76, 41
(2004).
3. S.-C. Lee, J.-H. Lee, T.-S. Oh and Y.-H. Kim, Solar Energy Mater. Solar Cells 75, 481 (2003).
4. T. Nyberg, Synthetic Metals 140, 288 (2004).
5. A. Yoshida, E. Matsumoto, H. Yamada, H. Okada and A. Wakahara, Nucl. Instrum. Methods Phys.
Res. B 199, 370 (2003).
6. S. Iwamori and M. Mizoguchi, Mater. Sci. Technol. 43, 99 (2006).
7. S. Iwamori, A. Uemura and Y. Yamada, J. Adhesion Sci. Technol. 18, 1771 (2004).
8. I. Nishiyama and M. Saito, J. Printing Sci. Technol. Jpn 42, 285 (2005).
9. Kuraray Poval® , Material Safety Data Sheet, Kuraray Co., Ltd (2007).
Part 3
Applications
Polyimide–Aerogel Hybrid Foam Composites for
Advanced Applications
Trent M. Smith a,∗ , Martha K. Williams a , James E. Fesmire a , Jared P. Sass a and
Erik S. Weiser b
a
NASA John F. Kennedy Space Center, Spaceport Technology Division, KSC, FL 32899, USA
b
NASA Langley Research Center, Advanced Materials and Processing Branch, Hampton,
VA 23681, USA
Abstract
NASA has had a growing need for high-performance polymer foams for cryogenic insulation, fireproofing,
energy absorption and other applications. Commercially available foams, however, do not meet all of the
requirements for surviving extreme environments. Most low-density foams crack at cryogenic temperatures,
and the foams that can withstand extreme temperatures do not have the required structural integrity or ther-
mal performance. Chemists and Engineers at NASA’s Kennedy Space Center and Langley Research Center
have collaborated in the development of a polyimide foam composite based on TEEK polyimide foam tech-
nology with enhanced thermal performance and vibration attenuation properties. The foam composites can
be fabricated to target densities: high-density foam composites for structural applications and low-density
foam composites where more flexibility might be advantageous. Heat transfer reduction is driven primar-
ily by inorganic aerogel filler, with aerogel blanket composites having the most significant reduction in
heat transfer and vibration attenuation. The TEEK composites are fire resistant and have excellent weather-
resistant properties. This composite technology also lends itself to other foam systems. These materials may
prove useful in process fluid piping, tanks for transporting and storing hot or cold fluids, ship and boat
building and especially aerospace applications.
Keywords
Polyimide, aerogel, foam, foam composite, aerospace insulation, high-performance foam, TEEK
1. Introduction
Aerogel materials have been known as excellent thermal [1–3] and acoustic insu-
lation materials for nearly eighty years [4–6], but their use has been limited due to
their extremely fragile nature. Present-day commercial aerogel materials are con-
sidered some of the best insulators on the planet at 101.3 kPa (760 torr) and are
more robust; however, many forms still lack sufficient structural integrity for use in
* To whom correspondence should be addressed. Tel.: 321-867-7492; Fax: 321-861-2975;

e-mail: Trent.M.Smith@nasa.gov
296 T. M. Smith et al.
many applications. Many forms of commercial aerogel also present a contamina-

tion concern due to shedding of aerogel powder. New cross-linked aerogel materials
(x-aerogel) are under development at the Glenn Research Center in Cleveland, OH
which show improved mechanical strength over state-of-the-art commercially avail-
able aerogels, however, current process requirements limit their production to small
sizes [7–9]. Other attempts to make aerogel composite materials have met with lim-
ited success [10–17].
For many years polyimide foam materials have been utilized for acoustic and
thermal insulation in applications requiring these properties in conjunction with
high use temperature, non-flammability, and excellent chemical resistance. In most
cases, these foams were limited to non-structural applications due to their inherent
very high flexibility, low density, and poor mechanical properties. Not until the late
1990s with the introduction of the TEEK polyimide foam technology did a poly-
imide foam become available which had good structural integrity at both cryogenic
and extremely high temperatures while retaining excellent chemical resistance and
good long-term weathering characteristics [18–30]. Although performance evalua-
tions indicated that the TEEK foams exhibited an excellent combination of prop-
erties, these foams still did not achieve the very low thermal conductivity achieved
by commercially available polyurethane and polyisocyanurate foams which are cur-
rently used to insulate the Space Shuttle External Fuel Tank.
In certain applications to meet future requirements set forth within NASA’s Con-
stellation program, further reduction in heat transfer in these inherently stable foams
is preferred. The introduction of aerogel materials into a robust polyimide foam
such as TEEK polyimide foam would be one mechanism by which this reduction
in thermal conductivity could be achieved without significant loss in mechanical
integrity.

2.1. Materials
Polyimide–aerogel hybrid foam composites were prepared from 4,4 -oxydiphthalic
anhydride/3,4 -oxydianiline (TEEK-H) friable balloons from Unitika Ltd, Osaka,
Japan, combinations of Spaceloft® and Cryogel® aerogel blankets from Aspen
Aerogels, Northborough, MA, and/or Nanogel® aerogel beads from Cabot Corpora-
tion, Billerica, MA. TEEK refers to the first names of the inventors of the polyimide
system.
2.2. Fabrication of Composites
TEEK–aerogel bead composites (TEEK XX% aero) were made by mixing the
appropriate amount of Nanogel® beads with TEEK friable balloons followed by
placing the mixture in a mold and curing in a convection oven at 200◦ C for 2 h.
The TEEK–aerogel blanket (TEEK-AB X-layer) composites were made in a simi-
lar manner: (1) half of the TEEK friable balloons were disbursed evenly across the
Polyimide–Aerogel Hybrid Foam Composites for Advanced Applications 297
mold surface, (2) the aerogel blanket was placed in the center of the mold, (3) the
remaining TEEK friable balloons were poured over the blanket and finally (4) the
foam composite was cured at 200◦ C for 2 h. The foam composite with the pocket
of aerogel beads was made in a similar manner as the aerogel blanket composites
by substituting a bed of beads in place of the blanket. The combined TEEK–aerogel
bead and blanket composites were made by substituting a mixture of balloons/beads
in place of the pure balloons. The diagonal strips designation refers to the aerogel
blanket being cut into strips and placed diagonally across the mold instead of a sin-
gle blanket.
2.3. Test Methods
Thermal conductivity was measured on a Netzsch Lambda 2300 heat flowmeter
per ASTM C518. Cryogenic fluid storage measurements were carried out by sub-
merging 0.152 m × 0.152 m × 0.0318 m foam specimens in liquid nitrogen for
2 min, draining the free liquid from the foam, and placing the foam on a balance
with a 0.0254 m thick piece of rigid polyurethane foam to insulate the balance pan.
The weight loss was monitored with respect to time. Vibration attenuation mea-
surements were accomplished by sandwiching foam specimens between aluminum
plates and striking with an instrumented modal impact hammer. Accelerometers
strategically placed on the opposite aluminum plate recorded the vibration that was
transmitted through the foam. Mechanical properties were measured in an Instron
4507 tensile test machine by compressing 0.0254 m × 0.0254 m × 0.0254 m foam
specimens to greater than 75% strain at a rate of 0.00635 m/min.
3. Results
The TEEK–aerogel bead composites had aerogel uniformly dispersed throughout
the foam composite as shown in Fig. 1. Aerogel beads retained their original spher-
ical shape. In the case of the aerogel blanket composites, dissection of an early
fabricated composite revealed that the blanket material was in good shape and was
surrounded by a foam shaped cavity close to the blanket. Incorporating aerogel into
the TEEK polyimide foam reduced heat transfer through the material as measured
by the heat flowmeter method. Increased loading of aerogel in the composite re-
sulted in reduced heat transfer through the composite. The thermal conductivity
(k-value) decreased with increasing density as shown in Fig. 2. Typically k-value
increases with increasing foam density due to increased solid conduction in the
denser material. The thermal conductivity of the 96 kg/m3 samples as shown in
Fig. 3 clearly demonstrate the beneficial effect of aerogel loading on heat transfer
in the resulting foam composites. Keeping density relatively constant the heat trans-
fer was reduced approximately 18% with aerogel bead loading of 25%. The drop
in k-value with respect to aerogel bead loading appears to have a linear relation-
ship (Fig. 3). The largest reduction in heat transfer was just over a 40% reduction
by the TEEK–aerogel 2-layer composite in comparison to the TEEK control at
Figure 1. Photograph of TEEK–aerogel hybrid foam composite (left) and TEEK foam (right). The
photographed foam samples were 96 kg/m3 and cut to 0.0254 m square, and 0.0318 m thick. The
Nanogel® beads are evident and are uniformly dispersed in the TEEK–aerogel hybrid foam.
96 kg/m3 as shown in Table 1 and Fig. 4. The aerogel loading effect and density
effect are clearly evident in Fig. 4. The composites which had aerogel strategically
placed within the foam performed the best. The TEEK–aerogel foam with a pocket
of beads which was 20% by weight aerogel placed in the center similar to a Pita
pocket and the TEEK–aerogel 1-layer composite had a 26% reduction in heat flow
in comparison to the 96 kg/m3 TEEK control. Cutting aerogel blanket and placing
in the diagonal orientation negatively impacted the k-value when compared to the
other aerogel blanket composite materials.
Foam–aerogel hybrid foam composites can be used as a cryogen storage medium
for increased safety and reduced cryogen boiloff. Aerogel due to its extremely high
surface area and open mesoporous and nanoporous structure lends itself as an ex-
cellent gas adsorption material and thus an excellent storage medium for cryogenic
fluids [31]. Liquid nitrogen storage capacity and retention followed aerogel load-
ing trends as shown in Fig. 5. Discrepancies in the trends are likely due to foam
morphology differences.
Rigid TEEK foam control and rigid TEEK–aerogel blankets composites were
tested for vibration attenuation characteristics. The rigid aerogel blanket containing
foams, 96 kg/m3 , were selected to obtain the highest differential vibration response
for comparison purposes. The aluminum plate as expected had the highest high g
and highest low g performance as is shown in Fig. 6 and Table 2. The lowercase
Figure 2. Density and thermal conductivity measurements for TEEK foam and TEEK–aerogel hybrid
foam composites. Note that the thermal conductivity decreases with increased loading of aerogel
beads.
Figure 3. Density and thermal conductivity measurements for rigid TEEK foam and TEEK–aerogel
hybrid foam composites. Note that the thermal conductivity clearly decreases with increased loading
of aerogel beads while foam density stays relatively constant.
g refers to the acceleration of gravity at 9.8 m/s2 and is the normal way in which
these data are analyzed (multiples of g). The best performer was the TEEK–aerogel
2-layer composite with a 4-fold decrease with respect to the TEEK control. The
Table 1.
Thermal conductivity and density measurements on TEEK control foam and TEEK–aerogel hybrid
foam composites
Sample ID Material Density Thermal conductivity

(kg/m3 ) (mW/m-K)
113 TEEK control 39.9 32.98

110 TEEK-10% aero w/w 41.7 32.22
124 TEEK-10% aero w/w 96.7 33.76
131 TEEK-10% aero w/w 37.4 31.97
111 TEEK-20% aero w/w 46.3 30.82
125 TEEK-20% aero w/w 99.0 31.32
132 TEEK-20% aero w/w 38.1 30.82
112 TEEK-25% aero w/w 47.6 30.17
126 TEEK-25% aero w/w 95.9 29.69
133 TEEK-25% aero w/w 39.0 30.63
115 TEEK-30% aero w/w 47.8 29.48
134 TEEK-30% aero w/w 39.3 29.85
116 TEEK-40% aero w/w 54.2 28.09
135 TEEK-40% aero w/w 40.3 29.20
122 Pocket of beads 46.6 26.48
121 TEEK–aeroblanket 1-layer 98.8 26.18
123 TEEK–aeroblanket 2-layer 98.0 21.04
139 TEEK–aeroblanket diag strips (9) 96.0 33.20
136 TEEK-10% aero + aeroblanket 1-layer 46.1 24.83
138 TEEK-30% aero + aeroblanket 1-layer 55.3 24.22
137 TEEK-30% aero + aeroblanket diag strips (9) 37.5 27.00
TEEK–aerogel 1-layer composite did have an initial larger response, 23.5g, than
the TEEK control, 22.3g, but Fig. 6 shows that TEEK–aerogel 1-layer composite
was just below that of the TEEK control from 25 ms to 60 ms.
4. Discussion
TEEK–aerogel hybrid foam composites had reduced heat transfer in comparison
to TEEK control foams. Normally heat transfer through foams occurs due to solid
conduction through cell struts and windows, gas conduction within cells, gas con-
vection through cells, and radiation. These modes of heat transfer are all dependent
on cell structure (density), foam composition or chemistry, and foam architecture
(open/closed cell content). Environmental factors which affect the modes of heat
transfer are temperature, temperature difference between the cold boundary and
warm boundary, pressure (internal and external), and gas composition (internal
Figure 4. Density and thermal conductivity measurements for TEEK foam and TEEK–aerogel hybrid
foam composites. Note that the thermal conductivity decreases with increased loading of aerogel
beads. The most significant reduction in heat transfer is clearly in the composites in which the aerogel
was strategically located rather than homogeneously distributed throughout the material. Examples
are the Pocket of Beads, TEEK–aeroblanket 1-layer, and the most significant reduction of thermal
conductivity was for TEEK–aeroblanket 2-layer.
and external). Increased foam density normally causes better heat flow through the
material due to better solid conduction as evident by the data collected on TEEK
control samples shown in Fig. 7. It is clear that incorporating aerogel into the foam
system diminishes the density effect and aerogel loading becomes the dominant
factor in the heat transfer as is shown in Fig. 7. Aerogel beads and blankets which
by themselves range between 80–128 kg/m3 are about 90% air and thus have min-
imal solid conduction. Commercial aerogel materials also have pore sizes ranging
from 10 to 20 nm thus impeding gas convection and gas conduction by making the
path for a gas molecule or collection of gas molecules tortuous (reduction of mean
free path). Aerogel materials are not effective radiation blockers unless treated with
an opacifying agent. Incorporating aerogel into a foam system therefore reduces all
three modes of heat transfer assuming that the aerogel materials chosen are better
thermal insulators than the parent foam system. Thus improving heat insulation by
incorporating aerogel into foam systems has diminishing returns when the parent
foam itself is an excellent thermal insulator. Additionally, mechanical properties
will likely suffer, but to what extent is dependent on how the foam is processed,
the type of aerogel incorporated, and the loading amount. The fabrication of foam–
aerogel hybrid composites can open up new applications for foam systems that were
Figure 5. Liquid nitrogen storage data for TEEK foam and TEEK–aerogel hybrid foam composites.
The number in parentheses denotes the number of different samples tested. All samples were run
twice. Higher aerogel loading resulted in increased capacity and retention of nitrogen over time. The
architecture of the foam composite though plays a critical role in allowing nitrogen access to the
aerogel and more open-cell foams should result in storing larger quantities of cryogen since the aerogel
is more accessible.
previously not possible due to inadequate thermal performance, but had other desir-
able properties. Further improvements in cryogen fluid storage applications require
an open-cell morphology with excellent cryogenic mechanical properties. Tuning
TEEK–aerogel hybrid foam fabrication and performance parameters for potential
utilization will be required to increase fuel and oxidizer densities, retention times
and transfer rates.
5. Summary
Incorporation of aerogel into a high-performance polyimide foam resulted in re-

duced heat transfer in the composite foam in comparison to the parent foam
material. TEEK–aerogel hybrid composites also had better vibration attenuation
properties and thus can be used as an integral part of a cryogenic fluid storage
system. New applications for these foam systems are being explored and new
foam systems are also being investigated. These foam systems as a part of inte-
gral thermal management plan can help NASA establish a Lunar Base and a Mars
Base. These foam hybrids also have several advanced terrestrial applications on
Earth.
Figure 6. Vibration attenuation data measurements on TEEK foam and TEEK–aerogel hybrid foam
composites at 5 kHz. The TEEK–aerogel foam composite with two layers of aerogel blanket clearly
performs the best with an acceleration decrease of 5 times that of the TEEK control and 16 times less
than that of the aluminum plate.
Table 2.
Vibration attenuation data measurements of rigid, 96 kg/m3 , TEEK foam and TEEK–aerogel hybrid
foam composites at 5 kHz. The lowercase g refers to acceleration of gravity at 9.8 m/s2 and is the
normal way in which vibration data are analyzed. High g is the highest acceleration for the sample
and low g is the lowest acceleration for the sample in the test time. TEEK–aerogel blanket 2-layer was
the most effective at attenuating the energy with a high g measurement of 5.45, which was the lowest of
the foam materials. TEEK–aerogel blanket 1-layer had an initial high g measurement slightly greater
than the TEEK control, but then was consistently lower showing it to be an improved attenuation
material over the TEEK control
Material Sample High g Low g
TEEK control N115 22.3 5.07

TEEK–aeroblanket 1-layer N117 23.5 2.02
TEEK–aeroblanket 2-layer N119 5.45 1.23
Aluminum plate Nxxx 78.3 57.9
Acknowledgements
The authors wish to thank Rudy Werlink and Dr. LaNetra Tate for vibration at-
tenuation measurements; Dr. Barry Meneghelli, Phil D’Andreamatteo, and Mark
Velasco for cryogen storage measurements; Wayne Heckle, Sr. for thermal conduc-
tivity measurements; and Courtney Hornsby for her work in fabricating samples.
This document was prepared under the sponsorship of the National Aeronautics
and Space Administration. Neither the United States Government nor any person
Figure 7. Thermal conductivity measurements on TEEK foam and TEEK–aerogel hybrid foam com-
posites. Typically heat transfer increases with increased foam density as shown by the TEEK control.
Incorporating aerogel into the foam determines heat transfer mechanisms and diminishes the density
effect with increased loading of aerogel.
acting on behalf of the United States Government assumes any liability resulting
from the use of the information contained in this document, nor warrants that such
use will be free from privately owned rights. The citation of manufacturers’ names,
trademarks, or other product identification in this document does not constitute an
endorsement or approval of the use of such commercial products.
References
1. J. E. Fesmire, Cryogenics 46, 111–117 (2006).
2. J. E. Fesmire, S. D. Augustynowicz and S. Rouanet, Adv. Cryogenic Eng. 47, 1541–1548 (2002).
3. J. E. Fesmire and J. P. Sass, Cryogenics 48, 223–231 (2008).
4. S. S. Kistler, Nature 127, 741 (1931).
5. S. S. Kistler, J. Phys. Chem. 36, 52–64 (1932).
6. S. S. Kistler and A. G. Caldwell, Ind. Eng. Chem. 26, 658–662 (1934).
7. A. Katti, N. Shimpi, S. Roy, H. Lu, E. F. Fabrizio, A. Dass, L. A. Capadona and N. Leventis,
Chem. Mater. 18, 285–296 (2006).
8. L. A. Capadona, M. A. B. Meador, A. Alunni, E. F. Fabrizio, P. Vassilaras and N. Leventis, Poly-
mer 47, 5754–5761 (2006).
9. M. A. B. Meador, L. A. Capadona, L. McCorkle, D. S. Papadopoulos and N. Leventis, Chem.
Mater. 19, 2247–2260 (2007).
10. C. L. Jones, Paint Industry Magazine 62, 390–391 (1947).
11. H. F. Church and J. R. Scott, J. Rubber Res. 18, 13–27 (1949).

12. G. Biesmans, D. Randall, E. Francais and M. Perrut, Cellular Polymers 17, 17–30 (1998).
13. J. K. Floess, R. Field and S. Rouanet, J. Non-Crystal. Solids 285, 101–108 (2001).
14. A. Rigacci, J. C. Marechal, M. Repoux, M. Moreno and P. Achard, J. Non-Crystal. Solids 350,
372–378 (2004).
15. A. A. Obaid, S. Andersen, J. W. Gillespie, Jr, R. Vaidyanathan and A. Studley, in: Proc. Interna-
tional SAMPE Symposium and Exhibition, Vol. 50, 2002–2013 (2005).
16. E. Bardy, J. Mollendorf and D. Pendergast, J. Phys. D: Appl. Phys. 39, 1908–1918 (2006).
17. E. Bardy, J. Mollendorf and D. Pendergast, J. Phys. D: Appl. Phys. 39, 4068–4076 (2006).
18. E. S. Weiser, F. F. Baillif, B. W. Grimsley and J. M. Marchello, in: Proc. International SAMPE
Symposium and Exhibition, Vol. 43, 730–744 (1998).
19. E. S. Weiser, T. F. Johnson, T. L. St. Clair, Y. Echigo, H. Kaneshiro and B. W. Grimsley, High
Performance Polymers 12, 1–12 (2000).
20. M. K. Williams, G. L. Nelson, J. R. Brenner, E. S. Weiser and T. L. St. Clair, in: Recent Advances
in Flame Retardancy of Polymeric Materials, M. Lewin (Ed.), Vol. 12, pp. 137–150. Business
Communications Co., Norwalk, CT (2001).
21. M. K. Williams, G. L. Nelson, J. R. Brenner, E. S. Weiser and T. L. St. Clair, ACS Symp. Ser. 797,
49–62 (2001).
22. R. J. Cano, T. H. Hou, E. S. Weiser and T. L. St. Clair, High Performance Polymers 13, 235–250
(2001).
23. E. S. Weiser, B. W. Grimsley, R. B. Pipes and M. K. Williams, in: Proc. International SAMPE
Symposium and Exhibition, Vol. 47, pp. 1151–1162 (2002).
24. C. I. Cano, R. B. Pipes and E. S. Weiser, 61st Annual Technical Conference — Society of Plastics
Engineers 2, 1835–1840 (2003).
25. C. I. Cano, E. S. Weiser and R. B. Pipes, Cellular Polymers 23, 299–309 (2004).
26. T.-H. Hou, E. S. Weiser, E. J. Siochi and T. L. St. Clair, High Performance Polymers 16, 487–504
(2004).
27. E. S. Weiser, Synthesis and Characterization of Polyimide Residuum, Friable Balloons, Mi-
crospheres and Foams, Ph.D. Dissertation, Department of Applied Science, College of William
and Mary, Williamsburg, VA (2004).
28. M. K. Williams, O. Melendez, J. Palou, D. Holland, T. M. Smith, E. S. Weiser and G. L. Nelson,
J. Adhesion Sci. Technol. 18, 561–573 (2004).
29. M. K. Williams, E. S. Weiser, J. E. Fesmire, B. W. Grimsley, T. M. Smith, J. R. Brenner and
G. L. Nelson, Polym. Adv. Technol. 16, 167–174 (2005).
30. M. K. Williams, D. B. Holland, O. Melendez, E. S. Weiser, J. R. Brenner and G. L. Nelson, Polym.
Degrad. Stab. 88, 20–27 (2005).
31. J. A. Starkovich, US Patent No. 2006/0218940 A1 (2006).
Polyimide Foam Nanocomposites from Powder Precursors
C. Rocio Misiego, Camilo I. Cano and R. Byron Pipes ∗

School of Chemical Engineering, Purdue University, 480 Stadium Mall Dr., West Lafayette,
IN 47907-2100, USA
Abstract
Polyimide foam nanocomposites have been developed through an in situ polymerization process followed
by solid state powder foaming. The dispersion of the acid functionalized multi-walled carbon nanotubes
in the reaction medium is achieved by means of surfactants, mechanical mixing and sonication. In the last
step of the dispersion technique, a solution of carbon nanotubes in tetrahydrofuran is sonicated with the
diamine monomer in order to improve carbon nanotube dispersibility by favoring the interaction of the acid
functionalities of carbon nanotubes and the basic amine groups of diamine monomers. The carbon nanotube
dispersion is used as the reaction medium for the polymerization of diamine and diester monomers, giving
rise to a poly(amic acid)–carbon nanotube solution. Partial solvent volatilization produces solid poly(amic
acid) with embedded carbon nanotubes that is processed to a powder. Polyimide foam is produced from
poly(amic acid) powder after a thermal treatment termed ‘solid state powder foaming’. Differential scanning
calorimetry and thermo-gravimetric analyses have been used to determine optimal processing conditions
and the influence of different surfactants. An analysis of the influence of carbon nanotubes on the polyimide
matrix has been carried out to fully understand the nature, properties and behavior of this type of material.
Keywords
Polyimide, foam, powder, carbon nanotubes, surfactant
1. Introduction
The development of high-performance polyimide foams has been spurred by the

demand of the aerospace industry for materials with low density and high thermal
resistance. These foams have applications in cryogenic, acoustic and structural in-
sulation [1–3]. The properties of polyimide foams can be modified by the addition
of carbon nanotubes (CNTs). The incorporation of CNTs into polymeric materials
improves their mechanical properties and modifies thermal and electrical conductiv-
ities [4, 5]. The performance of polyimide foams can be enhanced with the addition
of CNTs, providing improved mechanical behavior. Depending on CNT concen-
* To whom correspondence should be addressed. Tel.: (765) 418-5447; Fax: (765) 446-9992; e-mail:
bpipes@purdue.edu
308 C. R. Misiego et al.
tration, the electrical and thermal conductivities of polyimides can be significantly

modified [6–8]. Since the thermal properties of the foam are mostly determined by
the gas present in the foam [9], the addition of CNTs to polymeric foams will not
likely reduce significantly the thermal insulation properties of the foam.
Solid state foaming of poly(amic acid) (PAA) precursors with an embedded
blowing agent is one of the novel techniques for the production of polyimide foams.
By applying a thermal treatment to powder polymeric precursors containing a blow-
ing agent it is possible to produce cellular microstructures that give rise to polyimide
foam [10].
Due to the poor compatibility of CNTs with polyimide matrix, the dispersion of
CNTs is a fundamental barrier to improving the properties of the final nanocompos-
ite. In the past years, several techniques have been used for CNT dispersion, such
as high and low energy sonication and chemical modification of the CNT surface
[11–19]. Another approach for CNT dispersion is the use of surfactants [20–30]. For
most applications, the presence of surfactants in the polymer matrix causes process-
ing problems; therefore, the surfactant has to be eliminated after the production of
the polymer nanocomposite. However, in the case of polyimide foams, the use of
surfactants is required in order to reduce the surface tension of the growing cellular
microstructure.
Therefore, the use of surfactants to aid CNT dispersion is an attractive alternative
for the development of polyimide–carbon nanotube foam nanocomposites because,
in this case, there is no need for surfactant elimination.
The first stage of the polyimide foam nanocomposite production process is the
synthesis of the PAA–CNT precursor. In this step, a diester and a diamine react to
produce PAA in a medium containing CNTs, surfactants to help the dispersion of
the CNTs and solvents that will act as blowing agents in the later inflation stage.
Once the PAA–CNT composite is produced, a thermal treatment allows the inflation
of PAA cellular microstructures. The blowing agent forms hydrogen bonds with the
carboxylic and amide groups in the PAA. As temperature increases, the hydrogen
bonds can be broken, releasing blowing agent to drive the inflation process [9].
Polyimide foam is finally obtained after thermal imidization at higher temperatures.
2. Experimental
The carbon nanotubes used were multi-walled carbon nanotubes with carboxylic
functionality (MWCNT–COOH). The monomers used for the poly(amic acid)
(PAA) synthesis were 3,3 ,4,4 -benzophenonetetracarboxylic dianhydride (BTDA)
and 4,4 -oxydianiline (4,4 ODA). Tetrahydrofuran (THF) and methanol (MeOH)
were used as solvents. The surfactants employed were sodium dodecyl benzene sul-
fonate (SDBS), hexadecyltrimethyl ammonium bromide (CTAB), sodium dodecyl
sulfate (SDS), Tergitol NP9® (TNP9), DC 193® , Tergitol XJ® and Triton X200® .
A Branson 1510 ultrasonic cleaner and a Hielscher UP200S Ultrasonic Processor
were used for CNT dispersion. The viscosity of the dispersions was measured in
PI Foam Nanocomposites from Powder Precursors 309
Figure 1. Scheme of the reaction of MWCNT–COOH with 4,4 ODA.
a Cannon L153 glass capillary viscometer. PAA precursors were dried in a VWR
1415M vacuum oven. Optical micrographs were obtained with an Olympus ZX-12
stereomicroscope and a Digital Instruments SPOT digital camera. Micrographs of
cellular microstructures at different stages of inflation were recorded with an IN-
STEC heating stage connected to an Olympus ZX-12 stereomicroscope.
Modulated differential scanning calorimetry (MDSC) and differential scanning
calorimetry (DSC) analyses were carried out in a TA Instruments Q100 DSC. Ther-
mogravimetric analysis (TGA) experiments were conducted in a TA Instruments
Q500 TGA with a constant nitrogen flow.
2.1. Dispersion of Carbon Nanotubes
The as-received MWCNT–COOH nanotubes were first ground in THF with a ce-
ramic pestle and mortar to separate most of the macroscopic agglomerates. 0.025 g
of MWCNT–COOH were bath sonicated for 2 h in 31.64 g of THF under mechan-
ical stirring, then they were horn sonicated at 20 W for 10 min. In order to favor a
chemical interaction of the acid functionalities of the CNTs and the amine groups
of 4,4 ODA, 0.40 g of SDBS surfactant and 1.30 g of 4,4 ODA were added to the
mixture and bath sonicated under mechanical stirring for 2 h. Figure 1 represents
a scheme of the reaction path towards MWCNT–CO–ODA. Other surfactants used
instead of SDBS were CTAB, SDS, Tergitol NP9® , DC 193® , Tergitol XJ® and
Triton X200® . The viscosity of the dispersions was measured in a glass capillary
viscometer submerged in a constant temperature oil bath at 25◦ C. The viscosity of
dispersions prepared with the same procedure but without CNTs was also measured
to determine the increase in viscosity due to the CNTs.
2.2. Synthesis of Poly(amic acid)–Carbon Nanotube (PAA–CNT) Precursors
The PAA–CNT precursors were prepared via the diester diamine route. The diester
was prepared under nitrogen in a three-neck reactor submerged in a water bath at
80◦ C, with magnetic stirring and under solvent reflux. 12.30 g of THF were mixed
with 11.66 g of MeOH and 14.62 g of BTDA were added and reacted for approx-
imately 2 h, when the solution became clear due to the end of the esterification
reaction, as summarized in Fig. 2. The CNT dispersion prepared as described in
Section 2.1 was mixed with 7.75 g of 4,4 ODA. After 30 min of bath sonication,
Figure 2. Scheme of the reaction of BTDA with methanol to produce BTDE.
Figure 3. Scheme of the reaction of BTDE, 4,4 ODA and MWCNT–CO–ODA to produce
PAA–MWCNT precursor.
the diester solution was added and bath sonicated for 2 h under nitrogen, with me-
chanical mixing, in a three-neck reactor with solvent reflux at approximately 65◦ C.
A scheme of the amidation reaction can be seen in Fig. 3. The same protocol was
used for preparing PAA–MWCNT composites with the surfactants SDBS, CTAB,
SDS and Tergitol NP9® , all with 0.1% MWCNT–COOH (0.1 g of MWCNT per
100 g of solvent-free nanocomposite). PAA without MWCNTs and with the same
surfactants was also synthesized. Figure 4 represents the chemical structures of the
surfactants.
The PAA was then partially devolatilized in a vacuum oven at 70◦ C for 5 h. After
partial solvent volatilization, the solid PAA was ground with a ceramic pestle and
mortar to produce small particles and sieved to separate particle sizes. Micrographs
of the reaction mixture at different stages of the reaction process for the production
of PAA–MWCNT with SDBS are shown in Fig. 5. The sample preparation of the
PAA precursors for calorimetric and thermo-gravimetric analyses consisted of the
selection of precursors with particle size <100 µm which were devolatilized for
5 days in vacuum oven at 70◦ C.
Figure 4. Chemical structures of the surfactants used for the dispersion of MWCNT–COOH.
Figure 5. Micrographs taken at different stages of the PAA–MWCNT amidation reaction with SDBS
(a) MWCNT dispersion before starting the reaction, (b) after adding the diester solution, (c) after 2 h
of bath sonication and (d) after partial solvent volatilization and size reduction.
2.3. Thermo-Gravimetric Analysis and Polyimide (PI) Foam Production
A TA Instruments Q500 TGA in a nitrogen atmosphere was used for the thermo-
gravimetric analysis of all the PAA solid precursors with a temperature increase
of 5◦ C/min. Polyimide foam samples, produced from solid powder precursors with
particle size in the range of 180–300 µm, were prepared in the TGA pan using
a heating rate of 50◦ C/min and two consecutive isothermal steps at 150◦ C for 1 h
and 325◦ C for another hour. A scheme of the imidization reaction is shown in Fig. 6.
A sample of the PI foam obtained was also analyzed using a temperature increase
of 5◦ C/min.
2.4. Calorimetric Analysis
The calorimetric analysis of the devolatilized PAA solid precursors and the PI foams
was carried out with modulated differential scanning calorimetry (MDSC) and con-
secutive differential scanning calorimetry (DSC) analyses. The MDSC was run at
a 5◦ C/min heating rate, followed by two DSC scans at 20◦ C/min, an isothermal step
at 325◦ C for 1 h and a DSC scan at 20◦ C/min.
Figure 6. Scheme of the reaction of PAA–MWCNT to produce PI–MWCNT.
Table 1.
Increase in viscosity due to MWCNTs in dispersions with 3.9%
4,4 ODA in THF using different surfactants
Surfactant Surfactant MWCNTs Viscosity increase

(%) (%) due to MWCNTs (%)
SDBS 1.2 0.07 28

DC-193® 1.2 0.07 22
CTAB 1.2 0.07 18
Tergitol XJ® 1.2 0.07 16
SDS 1.2 0.07 10
Triton X-200® 1.2 0.07 8
Tergitol NP9® 1.2 0.07 5

3.1. Dispersion of MWCNTs Using Different Surfactants
The viscosity of a dispersion containing fillers is a function of the aspect ratio of
the filler, defined as the length divided by the diameter [31, 32]. The aspect ratio of
dispersed CNTs is much greater than that of CNT agglomerates, since the diameter
of the agglomerates is greater than the diameter of isolated CNTs. Therefore, the
viscosity of dispersions with better dispersed CNTs should be greater [33, 34]. The
increase in viscosity due to MWCNTs of the dispersions prepared as described in
Section 2.1 has been used as the screening tool for the selection of surfactants for
Figure 7. Image sequence of PAA–MWCNT with SDBS cellular microstructure formation as temper-
ature increases (a) precursor particle at room temperature, (b) bubble nucleation, (c)–(e) initial bubble
growth, (e)–(h) mature bubble growth, (h)–(i) final bubble growth and (i) PAA cellular microstructure
at 150◦ C.
the synthesis of PAA precursors. Table 1 summarizes the increase in viscosity in

the dispersions caused by addition of MWCNTs. The surfactants SDBS and Ter-
gitol NP9® provide 28% and 5% respective increases in viscosity, being the best
and worst MWCNT dispersants according to the viscosity increase criterion. In
order to study the effect of different surfactants in the polyimide nanocomposite,
SDBS, CTAB, SDS and Tergitol NP9® were selected for the synthesis of the PAA–
MWCNT composites. SDBS and SDS are anionic, CTAB is cationic and Tergitol
NP9® is non-ionic.
3.2. Cellular Microstructures of PAA–MWCNT
When the PAA particles have appropriate size and blowing agent content, a sim-
ple thermal treatment is sufficient to produce cellular microstructures, as showed
in Fig. 7. Starting from room temperature, as temperature increases the blowing
agent contained in the material diffuses to nucleation sites. When the glass transi-
tion temperature of the PAA is reached, the viscosity of the material is reduced and
the diffusion of blowing agents into the nucleus becomes easier. The blowing agent
accumulated in the nucleus creates a pressure difference which is the driving force
for the initial bubble growth. The diffusion of blowing agent becomes important
for mature bubble growth. In the terminal stage of the inflation process, the growth
is determined by both diffusion and pressure difference. At temperatures close to
150◦ C, an amidation reaction is produced in the bubble walls, increasing the poly-
mer molecular weight and, therefore, its viscosity. When the viscous forces become
dominant, the bubble growth ceases.
The factors that determine the nature of the inflation process are heating rate,
temperature, blowing agent content of the precursor, initial particle size and geom-
etry, surfactant used, MWCNT content and conversion of the initial PAA precursor.
The variation of these parameters leads to different nucleation temperatures, rates
of bubble growth, bubble sizes, wall thicknesses and bubble morphologies (i.e.,
production of single bubbles or multi-bubbles). Figure 8 shows the wall surface of
bubbles produced with and without MWCNTs with all the surfactants used.
3.3. Kinetic Processes and Glass Transitions of PAA and PI Nanocomposites
Differential scanning calorimetry (DSC) can be used to study the amount of heat
that is being consumed or released in the polymer as temperature increases. The
analysis of this heat variation allows the determination of glass transitions, solvent
evaporation and kinetic processes. In a modulated DSC test, a combination of two
heating profiles, one linear and one oscillatory, allows the separation of reversible
and nonreversible heat flows. A typical DSC ramp test consists of a linear heating
rate profile that produces a heat flow signal combining reversible and non-reversible
phenomena. In order to avoid cellular microstructure inflation inside the DSC sam-
ple holders, small precursor particles with low blowing agent content were used.
Figure 9 shows a typical MDSC thermogram of PAA precursors devolatilized
for 5 days with particle size <100 µm. The first drop in non-reversible heat flow
is related to free solvent evaporation (a), the next heat flow consumption drop
corresponds to the overlap of blowing agent decomplexation and amidation en-
dotherms (b). The last drop that can be observed is the imidization endotherm (c).
The reversible heat flow signal can be used to determine the glass transition tem-
peratures (Tg ) of PAA (d) and PI (f). After the glass transition of PAA there is an
Figure 8. Micrographs of cellular microstructure surfaces: (a) PAA without MWCNTs,

(b) PAA–MWCNT with SDBS, (c) PAA–MWCNT with CTAB, (d) PAA–MWCNT with SDS and
(e) PAA–MWCNT with Tergitol NP9® .
Figure 9. Modulated DSC thermograms of PAA precursor with SDBS (particle size <100 µm)
devolatilized for 5 days showing (a) solvent evaporation, (b) blowing agent decomplexation and ami-
dation, (c) imidization, (d) PAA glass transition, (e) enthalpic relaxation and (f) PI glass transition.
energy recovery due to enthalpic relaxation (e). The MDSC results are summarized
in Table 2.
For all the surfactants studied, the temperature for free solvent evaporation is
slightly reduced with MWCNTs. The amidation-blowing agent decomplexation and
imidization endotherms do not appear to be correlated with MWCNTs. Glass tran-
sition temperatures of both PAA and PI increase with the presence of MWCNTs,
the latter being more strongly affected by MWCNTs.
The influence of PI conversion on the glass transition can be studied by means of
consecutive DSC analysis, since the PI conversion increases in each DSC run. Ta-
ble 3 shows the glass transition of the PI for different conversions, different surfac-
tants and with and without MWCNTs. The glass transition of PI due to MWCNTs
tends to increase as the conversion is advanced. The larger increase in the glass
transition is assumed to be related to better dispersion of the MWCNTs. According
to this criterion, SDBS provides the best dispersion among the surfactants tested,
followed by CTAB and SDS. The glass transition temperature determines the maxi-
mum temperature at which the polymer can be used. Therefore, the combination of
MWCNTs and surfactants that enhance their dispersibility increases the maximum
temperature at which the polymer can be used.
The glass transition of PI foams was also studied. As with PAA precursors, the
glass transition of PI increases with MWCNTs, but the behavior is not the same as in
the case of PAA powder precursors that do not produce cellular microstructures. As
summarized in Table 4, the glass transition increase due to MWCNTs for the CTAB
surfactant is much lower in PI foam than in powder precursors. The causes of this
behavior are not yet clearly understood. A possible explanation is the occurrence
of interactions of the surfactant with the polymer matrix in PAA powders that are
different from the interactions of the surfactant with PI foams.
Table 2.
Results obtained from MDSC of PAA precursors (particle size <100 µm) devolatilized for 5 days (Tg :
glass transition temperature, T : temperature)
Sample Tg Tg T free Specific heat T amidation– T Specific

PAA PI solvent heat of decomplexa- imidization of amidation
(◦ C) (◦ C) evaporation free solvent tion peak peak and imidization
peak (◦ C) (J/g) (◦ C) (◦ C) (J/g)
SDBS 121 262 55 14.8 147 188 215

SDBS–MWCNT 123 273 51 13.3 150 188 212
CTAB 119 257 52 16.4 147 188 207
CTAB–MWCNT 120 266 50 16.1 146 189 213
SDS 118 261 60 19.8 145 191 208
SDS–MWCNT 121 266 56 9.7 146 191 200
TNP9 118 264 56 16.8 143 189 218
TNP9–MWCNT 122 267 55 11.8 143 190 203
3.4. Desorption Processes and Decomposition of PAA and PI Nanocomposites

A thermo-gravimetric analyzer (TGA) measures the weight fraction of material that
is left in the TGA sample holder as temperature increases. This analysis has been
used to determine diffusivity changes and temperatures for blowing agent volatiliza-
tion and decomposition. Figure 10 shows the TGA signals of a PAA precursor with
particle size <100 µm devolatilized for 5 days. Changes in weight loss are easily
interpreted in the derivative of the weight with respect to the temperature. The first
phenomenon that can be observed is the change in diffusivity produced when the
glass transition of the PAA is reached (a). At this point, the blowing agent can easily
leave the material and there is an increase in weight loss. At greater temperatures,
two peaks in the weight derivative can be observed. The blowing agent decomplex-
ation and the volatiles produced in the amidation process diffuse out of the powder
precursors, producing the first peak in the weight derivative (b). The second peak
in the derivative curve accounts for the blowing agent produced in the imidization
reaction (c). The blowing agent volatilization ends when the amidation process is
completed (d). At greater temperatures, close to 550◦ C, the decomposition of the
polymer produces the last peak in the weight derivative (e).
The TGA analysis of PAA precursors reveals different responses for the different
surfactants, with no major changes related to the MWCNTs. As shown in Fig. 11,
the imidization peak in the weight derivative is slightly sharper for the samples
without MWCNTs. The observation of a sharper peak in the derivative is related
Table 3.
Polyimide glass transition temperatures measured using MDSC on PAA precursors (particle size
<100 µm) devolatilized for 5 days (DSC1 after MDSC, DSC2 after DSC1, DSC3 after DSC2 and
1 h at 325◦ C)
Sample Tg PI Tg PI Tg PI Tg PI Tg PI Tg PI Tg PI Tg PI
(◦ C) increase (◦ C) increase (◦ C) increase (◦ C) increase
(MDSC) due to (DSC 1) due to (DSC 2) due to (DSC 3) due to
MWCNTs MWCNTs MWCNTs MWCNTs
SDBS 262 275 278 282

SDBS– 273 11 294 19 295 17 303 21
MWCNT
CTAB 257 280 280 294
CTAB– 266 9 287 7 289 9 303 9
MWCNT
SDS 261 278 279 285
SDS– 266 5 284 6 285 6 293 8
MWCNT
TNP9 264 273 276 285
TNP9– 267 3 281 8 282 6 288 3
MWCNT
Table 4.
Polyimide foam glass transition temperatures measured using MDSC on the PI foams (DSC1 after
MDSC, DSC2 after DSC1, DSC3 after DSC2 and 1 h at 325◦ C)
Sample Tg PI Tg PI Tg PI Tg PI Tg PI Tg PI Tg PI Tg PI
(◦ C) increase (◦ C) increase (◦ C) increase (◦ C) increase
(MDSC) due to (DSC 1) due to (DSC 2) due to (DSC 3) due to
MWCNTs MWCNTs MWCNTs MWCNTs
SDBS 282 286 286 289

SDBS– 293 11 296 10 296 10 299 10
MWCNT
CTAB 287 292 293 297
CTAB– 288 1 294 2 294 1 300 3
MWCNT
SDS 282 288 288 290
SDS– 286 4 292 4 292 4 295 5
MWCNT
TNP9 281 286 286 291
TNP9– 286 5 293 7 293 7 297 6
MWCNT
Figure 10. TGA of PAA precursor (particle size <100 µm) with CTAB devolatilized for 5 days show-
ing (a) blowing agent volatilization after diffusivity increase, (b) blowing agent decomplexation and
amidation, (c) imidization, (d) end of imidization and (e) decomposition.
to a greater diffusivity. Therefore, the diffusivity of the material in the imidization

step is reduced with the MWCNTs. The difference in the imidization peaks in the
samples with and without MWCNTs is more remarkable for the samples prepared
with the surfactants SDBS and CTAB, the better MWCNT dispersants according to
the PI glass transition.
Figure 11. TGA of PAA precursors (particle size <100 µm) devolatilized for 5 days with different
surfactants and with and without MWCNTs.
Table 5 contains the information obtained from TGA analysis of PAA precur-
sors. The results displayed show that the temperature at which the initial weight
loss starts is greater in the samples with MWCNTs. The onset temperature for
blowing agent volatilization is related to the increase in diffusivity produced when
the glass transition of the PAA is reached. Since the samples with MWCNTs
have greater PAA glass transition, the increase in the onset temperature for blow-
ing agent volatilization is consistent with the glass transition results. Amidation–
decomplexation and imidization peak temperatures are close to those obtained with
DSC. The temperature at which blowing agent volatilization stops after the end of
the imidization reaction is greater in the samples with MWCNTs, this result be-
ing consistent with the lower diffusivity related to the MWCNTs. The temperature
for 5% weight loss and the decomposition peak temperature do not appear to be
correlated with the presence of MWCNTs.
The TGA signals for PI foam are shown in Fig. 12. In the case of PI foam, the
weight loss in the material is negligible until the decomposition starts. The only
peak that can be observed in the weight derivative curve is due to the material
decomposition.
The TGA results for PI foams are shown in Table 6. The temperature for 5%
weight loss and the peak temperature for decomposition of PI foams are lower for
the samples that contain MWCNTs. The magnitude of this difference is greater for
samples with surfactants that produce better MWCNT dispersion according to the
PI glass transition.
4. Conclusions
The study of the development of polyimide foam nanocomposites from powder
precursors leads to the following conclusions:
Table 5.
TGA results of PAA precursor (particle size <100 µm) devolatilized for 5 days (T : temperature)
Sample T initial T amidation– T T 5% T end of T

weight decomplexa- imidization weight imidization decomposi-
loss tion peak peak loss (◦ C) tion peak
(◦ C) (◦ C) (◦ C) (◦ C) (◦ C)
SDBS 113 145 192 167 278 558

SDBS–MWCNT 114 152 190 171 295 555
CTAB 112 146 189 167 244 565
CTAB–MWCNT 114 144 191 165 249 566
SDS 113 149 191 163 258 554
SDS–MWCNT 114 152 193 171 262 555
TNP9 111 143 191 163 277 563
TNP9–MWCNT 112 146 197 168 280 562
Figure 12. TGA results of PI foam with CTAB.
– Solid state foaming from powder precursors is a feasible technique for the de-
velopment of polyimide–carbon nanotube foams.
– In order to produce homogeneous nanocomposites, the dispersion of carbon
nanotubes is a fundamental step. This dispersion can be improved with surfac-
tants.
– The increase in viscosity of the dispersion containing carbon nanotubes com-
pared to the same dispersion without carbon nanotubes is related to the degree
of dispersion of the carbon nanotubes.
– Differential Scanning Calorimetry can be used to assess the transformation of
poly(amic acid) into polyimide for samples prepared with and without car-
Table 6.
TGA results of PI foams (T : temperature)
Sample T 5% weight Increase in 5% mass T decomposition Increase in decompo-

loss (◦ C) loss T due to peak (◦ C) sition T due to
MWCNTs (◦ C) MWCNTs (◦ C)
SDBS 454 556

SDBS–MWCNT 484 30 563 7
CTAB 508 563
CTAB–MWCNT 525 17 571 8
SDS 496 555
SDS–MWCNT 516 20 557 2
TNP9 516 559
TNP9–MWCNT 507 −9 560 1
bon nanotubes. This technique was employed and it was found that the glass
transition temperature of the polyimide increased with the presence of carbon
nanotubes. The glass transition temperature increase due to carbon nanotubes
can be used as a tool to characterize dispersion analytically.
– Among the surfactants SDBS, CTAB, SDS and Tergitol NP9® , SDBS allows
the best dispersion of carbon nanotubes in the polyimide.
– Thermo-gravimetric analysis can be used to study the effect of carbon na-
notubes on the diffusivity. There is evidence of a slight decrease in diffusiv-
ity during the imidization process due to the addition of carbon nanotubes.
– The decomposition temperature of polyimide foams is slightly increased with
carbon nanotubes.
Acknowledgements
The authors wish to acknowledge SABIC IP and Purdue University for the financial
support for this project.
References
1. E. S. Weiser, T. F. Johnson, T. L. St. Clair, Y. Echigo, H. Kaneshiro and B. W. Grimsley, High
Performance Polymers 12, 1 (2000).
2. E. S. Weiser, T. L. St. Clair, Y. Echigo and H. Kaneshiro, US Patent 6,180,746 B1 (2001).
3. C. I. Cano, E. S. Weiser, T. Kyu and R. B. Pipes, Polymer 46, 9296 (2005).
4. A. I. Nakatami, R. P. Hjelm, M. Gerspacher and R. Krishnamoorti, Mater. Res. Soc Symp. Proc.
661, 1 (2001).
5. J. H. Koo, Polymer Nanocomposite: Processing, Characterization and Applications. McGraw-Hill
(2006).
6. Z. Ounaies, C. Park, K. E. Wise, E. J. Siochi and J. S. Harrison, Composites Sci. Technol. 63, 1637
(2003).
7. X. Jiang, Y. Bin and M. Matsuo, Polymer 46, 7418 (2005).
8. L. Gao, X. Zhou and Y. Ding, Chem. Phys. Lett. 434, 297 (2007).
9. N. Mills, Polymer Foams Handbook: Engineering and Biomechanics Applications and Design
Guide. Elsevier (2007).
10. C. I. Cano, Polyimide Microstructures from Powdered Precursors: Phenomenological and Para-
metric Studies on Particle Inflation, PhD Thesis, Graduate Faculty, University of Akron, Akron,
OH (2005).
11. D. Hill, Y. Lin, L. Qu, A. Kitaygorodskiy, J. W. Connell, L. F. Allard and Y.-P. Sun, Macromole-
cules 38, 7670 (2005).
12. C. Park, Z. Ounaies, K. A. Watson, R. E. Crooks, J. Smith, S. E. Lowther, J. W. Connell,
E. J. Siochi, J. S. Harrison and T. L. St. Clair, Chem. Phys. Lett. 364, 303 (2002).
13. K. Wise, C. Park, E. J. Siochi and J. S. Harrison, Chem. Phys. Lett. 391, 207 (2004).
14. A. Yu, H. Hu, E. Bekyarova, M. E. Itkis, J. Gao, B. Zhao and R. C. Haddon, Composites Sci.
Technol. 66, 1190 (2006).
15. K. A. Watson, S. Ghoseb, D. M. Delozier, J. G. Smith and J. W. Connell, Polymer 46, 2076 (2005).
16. D. M. Delozier, K. A. Watson, J. G. Smith and J. W. Connell, Composites Sci. Technol. 65, 749
(2005).
17. J. G. Smith, J. W. Connell, D. M. Delozier, P. T. Lillehei, K. A. Watson, Y. Lin, B. Zhou and
Y. P. Sun, Polymer 45, 825 (2004).
18. J. G. Smith, D. M. Delozier, J. W. Connell and K. A. Watson, Polymer 45, 6133 (2004).
19. B. Zhou, Y. Lin, D. E. Hill, W. Wang, L. M. Veca, L. Qu, P. Pathak, M. J. Meziani, J. Diaz,
J. W. Connell, K. A. Watson, L. F. Allard and Y. P. Sun, Polymer 47, 5323 (2006).
20. L. Vaisman, G. Marom and D. Wagner, Adv. Functional Mater. 16, 357 (2006).
21. L. Vaisman, D. Wagner and G. Marom. Adv. Colloid Interface Sci. 128–130, 37 (2006).
22. O. Matarredona, H. Rhoads, Z. Li, J. H. Harwell and L. Balzano, J. Phys. Chem. B 107, 13357
(2003).
23. M. J. O’Connell, P. Boul, L. M. Ericson, C. Huffman, Y. Wang, E. Haroz, C. Kuper, J. Tour,
K. D. Ausman and R. E. Smalley, Chem. Phys. Lett. 342, 265 (2001).
24. A. I. Gopalan, K. P. Lee, P. Santhosh, K. S. Kim and Y. C. Nho, Composites Sci. Technol. 67, 900
(2007).
25. Y. Tomonari, H. Murakami and N. Nakashima, Chemistry — A European Journal 12, 4027 (2006).
26. H. Wang, W. Zhou, D. L. Ho, K. I. Winey, J. E. Fischer, C. J. Glinka and E. K. Hobbie, Nano
Letters 4, 1789 (2004).
27. X. Gong, J. Liu, S. Baskaran, R. D. Voise and J. S. Young, Chem. Mater. 12, 1049 (2000).
28. Y. Lin, S. Taylor, H. Li, K. A. S. Fernando, L. Qu, W. Wang, L. Gu, B. Zhou and Y.-P. Sun,
J. Mater. Chem. 14, 527 (2004).
29. N. Nakashima, Intl J. Nanosci. 4, 119 (2005).
30. J. Hilding, E. A. Grulke, Z. G. Zhang and F. Lockwood, J. Dispersion Sci. Technol. 24, 1 (2003).
31. C. D. Han, Rheology and Processing of Polymeric Materials. Oxford University Press (2007).
32. S. G. Advani, Flow and Rheology in Polymer Composites Manufacturing. Elsevier (1994).
33. H. Cai, F. Yan and Q. Xue, Mater. Sci. Eng. A 364, 94 (2004).
34. P. Potschke, T. D. Fornes and D. R. Paul, Polymer 43, 3247 (2002).
High-Performance Composite Membranes Based on
Sulfonated Polyarylenethioethersulfone and Sulfonated
Polybenzimidazole for Fuel Cell Applications
Zongwu Bai a , Shane B. Juhl b , Narayanan Venkat a and Thuy D. Dang b,∗
a
University of Dayton Research Institute, University of Dayton, Dayton, OH 45469, USA
b
Air Force Research Laboratory/RXBP, Wright-Patterson Air Force Base, OH 45433, USA
Abstract
Fabrication of novel composite proton exchange membranes (PEMs) comprising a fully sulfonated pol-
yarylenethioethersulfone (SPTES) and sulfonated poly(p-phenylene benzobisimidazole) (SPBI) and the
evaluation of the membrane properties are described. The composite membrane was obtained via a solvent-
cast process in a mixture of DMAc and methanol as solvents. The composite membrane proton conductivity,
as measured by four-probe impedance spectroscopy, was found to increase with increase in the SPTES con-
tent in the composite. The highest proton conductivity obtained was ∼80 mS/cm at 65◦ C and 85% relative
humidity for the SPTES/SPBI 70/30 composite membrane which was considerably less than the proton
conductivity of 300 mS/cm for the pure SPTES membrane under the same conditions, but it was found that
the swelling of the composite membranes was dramatically reduced due to the decreased water uptake of
the composite membranes relative to that of SPTES. The morphology of the SPTES/SPBI composite mem-
branes was also examined by a combination of techniques such as Wide Angle X-ray Diffraction (WAXD),
Scanning Electron Microscopy (SEM) and tapping mode Atomic Force Microscopy (AFM) to confirm the
dispersion of SPBI and study the composite micro-structure. These studies also involved the preliminary
evaluation of the MEA (Membrane Electrode Assembly) performance of the composite membranes for
H2 /O2 fuel cells applications.
Keywords
Sulfonated polyarylenethioethersulfone (SPTES), sulfonated polybenzimidazole (SPBI), polymer compos-
ite, proton exchange membrane (PEM), proton conductivity
1. Introduction
Proton exchange membrane (PEM) fuel cells have received considerable attention
over the last ten years due to their use as a portable power source and as a poten-
tial replacement for batteries. It has resulted in the development of various classes
of polymer membranes with diverse mechanical properties and proton conductiv-
* To whom correspondence should be addressed. E-mail: thuy.dang@wpafb.af.mil
324 Z. Bai et al.
ities. Perfluorosulfonate ionomer membranes such as Nafion (DuPont), Flemion

(Asahi Glass Co.) and Acipex (Asahi Chemical Co.) have been used as PEMs in fuel
cells [1]. However, these perfluorosulfonic acid membranes are not suitable in large
scale commercial applications due to their high price, decrease in proton conduc-
tivity at temperatures higher than 80◦ C and at low relative humidity, and methanol
crossover, especially in direct methanol fuel cell (DMFC) applications [2, 3].
Optimized proton and water transport properties of the membrane and proper
water management are crucial for efficient fuel cell operation. While dehydration
of the membrane reduces proton conductivity, an excess of water can lead to flood-
ing of the electrodes. Both conditions result in poor fuel cell performance. The
proton conductivity of the ionomer membranes relies on the presence of water. At
temperatures above 100◦ C there is a drastic decrease in conductivity of the mem-
brane because of high evaporation rate of water. Thus, a modification of the polymer
membranes for operation at higher temperatures seems to be an alternative [4]. The
composite approach represents one of the ways to improve the performance of the
PEMs as the desired properties of the two components can be combined in one
composite. The preparation of composite membranes based on proton conducting
materials such as sulfonated polyaryleneetheretherketone membranes showed ex-
cessive swelling, which would render the membranes brittle upon drying [5]. The
membrane swelling can be potentially reduced by blending proton conducting poly-
mers with basic polymers such as polybenzimidazole (PBI), which are capable of
forming hydrogen bonds, leading to the compatibilization of the blend polymers.
The acid–base blend membranes showed reduced methanol permeability and were
thermally and chemically stable, in addition to showing moderate proton conduc-
tivities at higher temperatures [6].
Sulfonated polyarylenethioethersulfone (SPTES) has been synthesized by direct
polymerization employing sulfolane as the polar aprotic solvent. Membranes fabri-
cated from the fully sulfonated polyarylenethioethersulfone homopolymer SPTES
(or SPTES-100, to distinguish this composition from its copolymer composition)
[7] showed outstanding proton conductivity (∼300 mS/cm) at 65◦ C and 85% rel-
ative humidity. However, the potential usefulness of the membranes in H2 /O2 fuel
cells was limited by excessive swelling and the consequent loss of dimensional sta-
bility of the hydrophilic polymer due to its high water uptake.
Polybenzimidazole-based PEMs have been extensively studied, in the form of
their complexes with phosphoric acid [8, 9]. The unique hydration capability of the
benzimidazole heterocycles, along with their mechanical integrity and high ther-
mal and thermo-oxidative stabilities, make them eminently suitable for evaluation
as high temperature PEMs. For example, a conductivity of 40 mS/cm was observed
at 200◦ C for a cast PBI membrane imbibing 5.1 mol of H3 PO4 per polymer re-
peat unit at a water partial pressure of 1.5 atm [8]. But since the rigid-rod PBI is
soluble only in strong acids, it cannot be compatibilized with the organo-soluble
SPTES for the formation of a miscible blend. To overcome the miscibility issue
between SPTES polymer and rigid-rod PBI polymer in solution-based processing,
High-Performance Composite Membranes Based on SPTES and SPBI 325
a sulfonic acid-pendant rigid-rod polybenzimidazole (SPBI) [10, 11] was utilized as

the blend component instead since SPBI can be solubilized in organic solvents such
as methanol and DMAc as its triethylammonium salt [12]. SPBI was synthesized
by grafting pendant acid functionality directly onto the polymer backbone [10, 11].
Sulfonic acid-functionalized PBI is known to have exceptional mechanical prop-
erties due to the rod-like conformation of the polymer [12]; the incorporation of
SPBI in the SPTES matrix is expected to provide compatible blends resulting in
robust solution-cast micro-composite membranes for potential evaluation as high
temperature PEMs.
The rationale behind this composite approach for Proton Exchange Membranes
(PEMs) is two-fold: one is to mitigate the water susceptibility of the fully sulfonated
SPTES without substantially reducing the proton conductivity of the membrane,
thus improving its performance; the other is the potential utilization of the compos-
ite membrane for high temperature (150–200◦ C), low relative humidity (<25%)
fuel cell applications, by exploiting the unique hydration/water retention capability
of the imidazole heterocycle [13] in SPBI.
The present work describes the preparation of SPTES/SPBI composite mem-
branes by solution casting in methanol/N,N-dimethylacetamide (DMAc) and the
measurement of their proton conductivity by impedance spectroscopy, as a func-
tion of SPTES content in the composite. The membranes were characterized by
X-ray diffraction (XRD), Scanning Electron Microscopy (SEM) and Atomic Force
Microscopy (AFM).
2. Experimental
2.1. Materials
N,N -dimethylacetamide (DMAc), methanol (MeOH) and triethylamine (TEA)
were purchased from Aldrich and used as received. Sulfonated polyarylenethio-
ethersulfone homopolymer (SPTES or SPTES-100) and sulfonated polybenzimida-
zole (SPBI) were synthesized in our laboratories, as described elsewhere [7, 10].
Both polymers were of high molecular weight and could be solvent-cast into films
of outstanding mechanical integrity. The chemical structures of SPBI polymer
and SPTES polymer are shown in Fig. 1. The synthesis and characterization of
SPTES-50 copolymer membrane [14], which is described in this paper for the com-
parison of MEA performance, have been reported in our previous work.
2.2. Preparation of SPTES/SPBI Composite Membranes
The polymer composite membranes were fabricated via a solvent-cast process in
which the rigid-rod SPBI was first solubilized as its triethylammonium salt in a mix-
ture of DMAc and MeOH as solvents followed by the addition of SPTES and the
subsequent volatilization of TEA and the solvents. The preparation of SPTES/SPBI
polymer composite membranes (SPTES/SPBI = 70/30 as an example) was as fol-
lows: DMAc (40 ml), MeOH (10 ml) and TEA (10 ml) were charged into a three-
326 Z. Bai et al.
Figure 1. Chemical structures of SPBI and SPTES polymers.
necked reaction flask equipped with a mechanical stirrer and a nitrogen inlet/outlet
and were stirred for 30 min at room temperature. SPBI polymer (1.2 g) was added
into the flask. The mixture was stirred at 140◦ C, in a silicone oil bath, until the so-
lution became clear. Then, SPTES polymer (2.8 g) was added to the SPBI solution.
The resulting mixture was stirred under a nitrogen atmosphere at 80◦ C for 2–4 h.
The SPTES/SPBI polymer composite solution was then filtered. The SPTES/SPBI
polymer composite solution was directly transferred to a casting dish, and was dried
at 80◦ C for 4–8 h in a vacuum oven, then at 100◦ C for 24 h. Once completed, the
membrane was stored in an acid solution bath (2M sulfuric acid) for 24 h at room
temperature. A free-standing SPTES/SPBI (70/30) polymer composite membrane
was thus fabricated.
2.3. Characterization of SPTES/SPBI Composite Membranes
2.3.1. Water Uptake
Water uptake of SPTES/SPBI polymer composite membranes was determined from
the difference in weight (wt) between the dry and the swollen membranes. The
membrane, cast from the polymer composite solution after vacuum drying at 100◦ C
for 24 h, was weighed and then soaked in water for 24 h, and wiped with a blot-
ting paper (soaking paper) and re-weighed. The percentage of water uptake was
calculated with reference to the weight of the dry membrane specimen from the
following equation:
Wwet − Wdry
Water uptake (wt%) = × 100.
Wdry
2.3.2. Proton Conductivity
Proton conductivity was measured using AC Impedance Spectroscopy and utilized
a standard 4-electrode measurement setup to eliminate electrode and interfacial
effects. A Teflon sample fixture was placed inside a temperature and humidity con-
trolled oven and was situated in such a way to allow the sample to be exposed to
the humidified air within the chamber. The two outer electrodes were made of plat-
inum foil and these acted as a source for current in the sample. Two inner platinum
wire electrodes (spaced 1 cm apart) were used to measure the voltage drop across
a known distance. By measuring the impedance of the material as a function of
frequency at a set temperature and humidity, the conductivity of the membranes is
obtained using the magnitude of the impedance in a region where the phase angle
is effectively zero. The conductivities were calculated by using the formula:
L
σ= ,
R×A
where σ is the proton conductivity, L, the distance between the two electrodes, R,
the resistance, and A, the test area, given by T × W , T being the thickness of test
membrane sample and W , the width of the test membrane.
2.3.3. Thermogravimetric Analysis (TGA)
The thermal stability of SPTES/SPBI polymer composite membranes was deter-
mined on an Auto TGA 2950HR V5.4A instrument. All the samples were dried in
vacuum at 120◦ C for at least 12 h for the removal of absorbed water. Approximately
5–10 mg samples of SPTES/SPBI polymer composites were preheated to 150◦ C at
10◦ C/min and kept at this temperature for 30 min to remove any residual water or
solvent remaining in the polymer composite membranes under N2 ; then the sam-
ples were cooled to 30◦ C and reheated to 900◦ C at 10◦ C/min. The analyses were
performed in the range of 30–900◦ C at a heating rate of 10◦ C/min in air.
2.4. Wide Angle X-Ray Diffraction (WAXD)
A Bruker AXS D8 Discover wide angle X-ray diffractometer in transmission mode
was used. X-rays with λ = 1.5418 Å were generated by a Cu Kα source. The sample
to detector distance was 20 cm. The wide angle X-ray scattering, WAXS, intensity
was collected using an evacuated Statton camera (Warhus, DE, USA) with point-
collimated X-ray beam via two pinholes (0.05 mm), separated by ∼15 cm at Cu Kα
wavelength of 1.5418 Å.
2.5. Scanning Electron Microscopy (SEM)
Surface morphology of the cryo-fractured dried and swollen membranes was exam-
ined by a Serion field emission SEM (FEI, USA). The instrument was operated at
5 kV in the secondary electron imaging mode.
2.6. Atomic Force Microscopy (AFM)
Tapping mode AFM was used to obtain height and phase imaging data simultane-
ously on a Dimension 3100 Nanoscope III Controller (Veeco, CA). Etched silicon
probes (Nanoprobes, Veeco) were used at their fundamental resonance frequen-
cies which typically varied from 270 to 350 Hz. The lateral scan frequency was
0.5–1 Hz. The AFM was operated with a double vibration isolation system at var-
ious humidities by using dry nitrogen to dehumidify and nitrogen bubbled through
328 Z. Bai et al.
water to humidify the chamber. The images were obtained in the tapping mode.
Tapping mode allows the probing of more delicate samples due to the lower lat-
eral forces applied to the surface. In the tapping mode, cantilever oscillates near its
resonance frequency, contacting the surface for very short time during it oscillation
cycle. Obtaining both phase images and topographical information is possible in
the tapping mode.
2.7. Membrane Electrode Assembly (MEA)
A painting technique was used to form electrodes on both sides of the composite
membranes, using ink slurry. This slurry consisted of platinum black, water and
Nafion solution (5 wt%) as binder. Catalyst loading of 5 mg/cm2 was used. The
ink slurry was sonicated to break up the catalyst powder in order to obtain a ho-
mogeneous distribution. Several layers of the ink were painted on both sides of
the composite membranes, and dried at 80◦ C under vacuum for 10 h to remove all
solvents. MEAs with effective areas of 5 cm2 were fabricated.

3.1. Properties and Composition of SPTES/SPBI Composite Membranes
The polymer composite membranes were fabricated with varying weight ratios of
SPTES and SPBI, such as 30/70, 50/50 and 70/30. The SPTES polymer was the
acidic component with a high proton conductivity due to its fully disulfonated back-
bone. SPBI, with its benzobisimidazole structure, served as the basic component,
which was incorporated into the SPTES/SPBI polymer composite to enhance its
mechanical integrity and durability as a fuel cell membrane by maintaining high
water resistance while retaining relatively high proton conductivity.
Factors such as water uptake, proton conductivity and membrane swelling need
to be delicately balanced in order to obtain a composite PEM which is suitable
for durable operation in a H2 /O2 fuel cell. Higher contents of the SPTES polymer
component in the composite would result in higher proton conductivities due to the
increased water uptake of the membrane. On the other hand, the SPBI polymer, be-
sides being compatible [12] with SPTES for the generation of homogeneous blends,
also serves to reduce the water uptake and thus the excessive swelling of the SPTES
polymers. This means there would be a reduction in the overall proton conductivity
of the composite membrane as a function of SPBI incorporation in the composite
membrane; it is to be stated, in this context, that sulfonic acid-containing rigid-
chain polybenzimidazole membranes themselves are known to have relatively low
to modest proton conductivities [10] (∼1 mS/cm) except at temperatures exceeding
150◦ C.
The properties, i.e. solubility, water uptake and proton conductivity, of
SPTES/SPBI polymer composites depend on the content of sulfonated groups in
the polymer backbone. Solubility of SPTES/SPBI polymer composites was deter-
mined at a concentration of 10% (w/v) in a number of solvents, including NMP,
Table 1.
Properties of SPTES/SPBI polymer composite membranes
Polymer Solubility Solubility Solubility in Proton

in water in methanol DMAc and NMP conductivity∗
(mS/cm)
SPTES/SPBI 30/70∗∗ N N N 15.3

SPTES/SPBI 50/50∗∗ N N N 32.5
SPTES/SPBI 70/30∗∗ N N SW 81.3
SPTES-100 polymer∗∗∗ S S S 300
SPBI polymer∗∗ N N N 1.0
S — soluble, N — insoluble, SW — swollen.

∗ The membrane samples was tested at 65◦ C, 85% relative humidity.
∗∗ Could be solubilized in MeOH/DMAc only in the presence of triethylamine.
∗ ∗ ∗ For comparison with composite samples.
DMAc, methanol and water at room temperature. The results are listed in Table 1.
From Table 1, it is evident that while the SPTES homopolymer is soluble in a range
of solvents such as water, methanol and DMAc, SPBI is insoluble in all the sol-
vents. As stated earlier, both SPBI and SPTES polymer blends with SPBI can be
solubilized in methanol/DMAc only in the presence of triethylamine.
The proton conductivity of the SPTES/SPBI polymer composite membranes was
measured using AC Impedance Spectroscopy and utilized a standard 4-electrode
measurement setup to eliminate electrode and interfacial effects. At a set tempera-
ture and humidity, the proton conductivity of the SPTES/SPBI polymer composite
membranes is listed in Table 1.
From Table 1, it is observed that the composite membrane proton conductivity
increased from 15.3 mS/cm to 81.3 mS/cm as the amount of the SPTES polymer
in the composite was varied from 30 to 70 wt%. Clearly, the increase in proton
conductivity is associated with the presence of the conductive SPTES polymer in
the composite membrane. In the acid–base blends, strong interactions due to hy-
drogen bridges and electrostatic forces by proton transfer from the acidic to the
basic group are present [15]. While the acid–base specific interaction is responsible
for the SPTES–SPBI blend compatibility, it also causes a significant reduction in
the proton conductivity of the composite membrane relative to that of SPTES-100.
Presumably, at the molecular level, this can be attributed to the strong ionic associ-
ation of some of the SO3 H groups in the SPTES backbone with the basic imidazole
structure of SPBI, thus lowering proton mobility. Also, the lowering of the proton
conductivity can be correlated with the reduction in the water uptake of the com-
posite relative to that of SPTES-100.
The proton conductivity is temperature dependent at a given relative humidity
[16, 17] for most of the polymer electrolyte materials. The results of the proton
conductivities of the SPTES/SPBI polymer composite membranes at various tem-
330 Z. Bai et al.
Figure 2. Proton conductivity of SPTES/SPBI composite membranes as a function of temperature.
peratures are given in Fig. 2. Increased SPBI content in the composite decreased the
membrane proton conductivity owing to the formation of ionic crosslinks by proton
transfer from the free –SO3 H groups to the imidazole heterocycle. With increase in
temperature at constant (85%) relative humidity, a gradual increase in proton con-
ductivity was observed in the case of the SPTES/SPBI 30/70 and 50/50 composite
membranes while the corresponding increase was much sharper in the case of the
70/30 SPTES/SPBI composite membrane.
In general, it is known that proton conductivity of composite membranes does not
directly correlate with either water uptake or ion exchange capacity across differ-
ent families of polymers [17]. The water uptake results of SPTES/SPBI composite
membranes with varying compositions are displayed in Fig. 3. As can be seen, the
water uptake increased with increasing SPTES polymer content in the composites;
this mirrors the proton conductivity trend observed for these composite membranes.
The water uptake decreased with the addition of SPBI to SPTES, resulting in the
reduction in the swelling of the polymer composite membranes.
3.2. Thermal Properties of SPTES/SPBI Composite Membranes
The thermal properties of SPTES/SPBI polymer composites were examined
by TGA in an air atmosphere at a heating rate of 10◦ C/min. The results for
SPTES/SPBI polymer composites as well as for SPBI and SPTES for compari-
son, are shown in Fig. 4. It is evident that there is considerable difference in the
thermo-oxidative degradation profiles of SPTES and SPBI polymers. In general,
Figure 3. Water uptake of SPTES/SPBI composite membranes.
Figure 4. TGA traces of SPTES/SPBI polymer composites.
a two-stage weight loss profile is observed for the polymers and the polymer com-
posites. The onset of first weight loss corresponding to the pendant sulfonic acid
decomposition occurs at 260◦ C for SPTES while it is considerably higher (350◦ C)
for SPBI. This is due to the higher thermal stability associated with the imidazolium
sulfonate group in SPBI. The second weight loss results from polymer backbone
332 Z. Bai et al.
degradation and the onset for the degradation occurs around 400◦ C for SPTES
and 500◦ C for SPBI. In the case of the SPTES/SPBI composites (70/30, 50/50 and
30/70 w/w), the degradation of the sulfonic acid pendant groups occurs over a much
broader temperature range and is increasingly shifted towards higher temperatures
as a function of increasing SPBI content. Increased thermal stability for the second
stage decomposition is also observed in the case of all the composites, relative to
SPTES polymer. These results also corroborate the incorporation of SPBI in the
SPTES matrix.
The ordering and morphology of the SPTES/SPBI composite membranes were
also examined by a combination of techniques such as WAXD, SEM and tapping
mode AFM to confirm the dispersion of SPBI and study the composite micro-
structure.
3.3. X-Ray Diffraction of SPTES/SPBI Composite Membranes
The SPTES/SPBI polymer composite membranes were also investigated using
X-ray diffraction (XRD). The X-ray diffraction images for the composite mem-
branes containing 30%, 50% and 70% (wt) of SPTES in SPTES/SPBI polymer
composite membranes showed qualitatively similar trends. The XRD patterns
(Fig. 5) were found to increase in intensity with increasing content of SPBI. A com-
parison of these images also reveals that the patterns become sharper as the SPBI
content in the composite membrane is increased.
The integrated equatorial two-theta (2θ ) plots of the XRD patterns of the three
SPTES/SPBI composites are shown in Fig. 6. Two major diffraction peaks, centered
at 2θ = 25.8◦ and 18.3◦ , are seen. These correspond to d-spacings of 3.45 Å and
4.84 Å, characteristic of the planes formed by face-to-face and side-to-side lateral
packings, respectively, of SPBI chains [12]. These peaks increase in intensity with
increase of the SPBI content in the composite.
Figure 5. XRD images of SPTES/SPBI polymer composite membranes.

Figure 6. WAXD plots for SPTES/SPBI polymer composite membranes.
3.4. AFM Studies of SPTES/SPBI Composite Membranes

Tapping mode images of the SPTES polymer membrane and SPTES/SPBI 50/50
polymer composite membrane were taken under various relative humidity con-
ditions and at various size scales in order to monitor the membrane morphology
during hydration. The SPTES polymer and SPTES/SPBI 50/50 polymer composite
membranes had to be purged overnight under dry nitrogen in the AFM chamber
to remove the surface water adhering to the hydrophilic sulfonic acid groups of
the membrane. The humidity was raised slowly over a period of several hours
while imaging of the film continued. Water droplets can be seen to develop on the
SPTES polymer (Fig. 7) surface as the relative humidity was raised from (a) 24%,
(b) 32%, (c) 42%. No apparent phase separation or ‘clusters’ could be seen on
the SPTES polymer membrane when imaged at higher magnification (down to
500 nm × 500 nm).
The SPTES/SPBI 50/50 composite membrane was shown to have a noticeable
phase separation in its morphology. The micro-scale dispersion appears fairly uni-
form with only a few locations of contrasting phase difference as shown in Fig. 8.
Closer observation (Fig. 9) of the different regions revealed nano-scale dispersion
of the SPTES and the SPBI within each region. The apparent phase difference
on the micro-scale, seen in Fig. 8, is due to a local concentration variation of the
SPTES/SPBI ratio as seen in the two phase images in Fig. 9.
Figure 9(a) has noticeably less SPBI than in Fig. 9(b) which is from the local
region of higher SPBI concentration. The lighter regions are associated with the
334 Z. Bai et al.
Figure 7. AFM images of SPTES polymer under increasing relative humidity. (a) 24% RH,
(b) 32% RH and (c) 42% RH.
Figure 8. SPTES/SPBI 50/50 composite height (200 nm) image (a) and the corresponding phase (90◦ )
image (b) showing local regions of concentration variation.
harder SPBI polymer and the darker regions with the softer SPTES polymer. The
amplitude set point was varied to confirm phase data validity.
3.5. Structure and Morphology of SPTES/SPBI Composite Membranes
Cross-sectional images of SPTES/SPBI 70/30, 50/50 and 30/70 composites were
obtained on the Sirion (low magnification) SEM and the Hitachi S-5200 SEM.
The images were made of freeze-fractured cross sections of the polymer composite
membranes. The thickness of the membranes was approximately 35 µm. Figure 10
shows that all the three membranes have a micro-structure with no apparent phase
separation. A layered micro-structure, typical of the rigid-rod polymer, is evident
with increased SPBI incorporation and is particularly revealing in the case of the
SPTES/SPBI 30/70 composite. Overall, these results indicate good dispersion of
Figure 9. Phase (120◦ ) images of the SPTES/SPBI 50/50 composite showing morphology differences
resulting from local concentration variations. (a) 1.63 µm and (b) 1.00 µm height.
Figure 10. SEM images of SPTES/SPBI polymer composite membranes. (a) SPTES/SPBI 70/30,
(b) SPTES/SPBI 50/50 and (c) SPTES/SPBI 30/70.
SPBI in the SPTES matrix, providing a homogeneous morphology for all the com-
posite films.
3.6. MEA Fabrication and Performance of SPTES/SPBI Composite Membranes
Membranes with painted electrodes along with a gas diffusion layer (carbon cloth)
were positioned in a single cell fixture with graphite blocks as current collec-
tors, having a serpentine flow-pattern. The gases used (H2 , air) were humidified,
with the gas humidity bottles set at 80◦ C; 80% fuel utilization, along with a sto-
ichiometry of 2 was used for the fuel gas and oxidant. Humidity was determined
with an in-line high temperature humidity probe (55% relative humidity). Cell per-
formance was determined at 1 atmosphere with cell temperature at 80◦ C. MEAs
using SPTES/SPBI 50/50 composite, SPTES-50 copolymer and Nafion-117 mem-
branes (all solvent-cast films) with effective areas of 5 cm2 were fabricated. Polar-
ization plots for the Nafion-117, SPTES-50 copolymer, and SPTES/SPBI 50/50
336 Z. Bai et al.
Figure 11. Polarization curves of MEAs from SPTES/SPBI 50/50 composite, SPTES-50 copolymer
and cast Nafion-117 membrane.
composite-based MEAs were obtained and a comparison of their performances

under the conditions described above is shown in Fig. 11. SPTES/SPBI 50/50 com-
posite MEA, compared with Nafion-117 MEA and SPTES-50 copolymer MEA,
showed an overall lower performance. A lower current density of 180 mA/cm2 for
SPTES/SPBI 50/50 composite membrane was obtained compared to 520 mA/cm2
and 260 mA/cm2 for the SPTES-50 copolymer membrane and the Nafion-117
membrane at the same potential (0.6 V) under the same experimental conditions.
Competing processes which include catalyst activity, interfacial resistance, and dif-
fusion processes occur to different extents and contribute to losses observed in the
polarization curve.
Table 2 shows the overall electrochemical performances of MEAs from
SPTES/SPBI 50/50 composite membrane, SPTES-50 copolymer membrane and
Nafion-117 membrane. SPTES/SPBI 50/50 composite membrane MEA exhibited
a lower OCV (open circuit voltage) compared with MEAs from SPTES-50 copoly-
mer and Nafion-117 membranes. Presumably, this can be attributed to micro-porous
defects in the SPTES/SPBI 50/50 composite membrane that often occur during the
membrane fabrication process. Calculated area specific resistance (ASR) is higher
(0.95 ± 0.1 ohm cm2 ) for SPTES/SPBI 50/50 composite membrane compared to
an area specific resistance of 0.13 ± 0.02 ohm cm2 for the SPTES-50 copolymer
membrane MEA and 0.30 ± 0.03 ohm cm2 for the Nafion-117 membrane MEA.
Estimates of hydrogen fuel permeability based on measured open circuit voltage
Table 2.
Comparison of cell performance of composite membrane and copolymer membrane vs Nafion
Sample ASR∗ (ohm cm2 ) Rct∗∗ (ohm cm2 ) Open circuit

all voltage ranges at 0.7 V voltage (OCV)
SPTES/SPBI 50/50 composite (100 µm) 0.95 ± 0.1 1.7 0.87

SPTES-50 copolymer (100 µm) 0.13 ± 0.02 0.8 0.95
Cast Nafion-117 (250 µm) 0.30 ± 0.03 1.1 0.99
∗ ASR — area specific resistance (ohm cm2 ).

∗∗ Rct — charge-transfer resistance (ohm cm2 ).
indicate that SPTES-50 copolymer membrane MEA exhibits similar rates of fuel
crossover.
4. Conclusions
In this paper, preparation, proton conductivity and morphology of SPTES/SPBI
composite membranes have been reported. The membranes showed good proton
conductivities, up to 80 mS/cm, at 65◦ C and 85% relative humidity; the proton
conductivity was found to increase with increasing SPTES content. While the com-
posite membranes had much lower proton conductivities compared to SPTES, they
exhibited much better dimensional stability than the SPTES homopolymer under
high relative humidity conditions. The decreased proton conductivity of the mem-
branes relative to that of SPTES can be attributed to the strong association of the
SO3 H groups in the SPTES backbone with the basic imidazole heterocycle of SPBI.
Viable MEAs were fabricated from these composite membranes and their electro-
chemical performance was evaluated. Morphological examination of the composite
membranes confirmed the presence of SPBI in these composite membranes and
showed uniform dispersion of the SPBI phase. The inherent water retention ca-
pability associated with the SPBI component makes these composite membranes
prospective candidates for potential utilization in high temperature fuel cells oper-
ation.
Acknowledgements
The authors would like to thank AFRL/PR for funding this research, Dr. Joseph P.
Fellner (AFRL/PR) for the management of the Fuel Cells program and Marlene D.
Houtz (University of Dayton Research Institute) for TGA measurements.
References
1. M. Rikukawa and K. Sanui, Prog. Polym. Sci. 25, 1463 (2000).
2. A. Kuver and K. Potje-Kamloth, Electrochim. Acta 43, 2527 (1998).
338 Z. Bai et al.
3. A. S. Arico, S. Srinivasan and V. Antonucci, Fuel Cells 1, 133 (2001).

4. S. M. Javaid Zaidi, Electrochimica Acta 50, 4771 (2005).
5. J. A. Kerres, A. Ullrich, F. Meier and Th. Haring, Solid State Ionics 125, 243 (1999).
6. Y. S. Kim, F. Wang, M. Hickner, T. A. Zawodzinski and J. E. McGrath, J. Membrane Sci. 212,
263 (2003).
7. Z. Bai and T. D. Dang, Macromol. Rapid. Commun. 27, 1271 (2006).
8. J. Wainright, J. T. Wang, D. Weng, R. F. Savinell and M. Litt, J. Electrochem. Soc. 142, L121
(1995).
9. A. Schechter and R. F. Savinell, Solid State Ionics 147, 181 (2002).
10. T. D. Dang, M. J. Dalton, M. F. Durstock, N. Venkatasubramanian and F. E. Arnold, ACS PMSE
Preprint 89, 508 (2003).
11. H. Pu and Q. Liu, Polymer International 53, 1512 (2004).
12. N. Venkatasubramanian, D. R. Dean, T. D. Dang, G. E. Price and F. E. Arnold, Polymer 41, 3213
(2000).
13. D. W. Tomlin, A. V. Fratini, M. Hunsaker and W. W. Adams, Polymer 41, 9003 (2000).
14. Z. Bai, M. F. Durstock and T. D. Dang, J. Membrane Sci. 281, 508 (2006).
15. J. A. Kerres, J. Membrane Sci. 185, 3 (2001).
16. X. Li, C. Zhao, H. Lu, Z. Wang and H. Na, Polymer 46, 5820 (2005).
17. R. W. Kopitzke, C. A. Linkous, H. R. Anderson and G. L. Nelson, J. Electrochem. Soc. 147, 1677
(2000).
New Ion-Containing Polyimides for the Purification of
Bio-fuels by a Membrane Separation Process
Anne Jonquières ∗ , Michel Awkal, Robert Clément and Pierre Lochon

Laboratoire de Chimie Physique Macromoléculaire, UMR CNRS-INPL 7568, Nancy Université,
ENSIC, 1 rue Grandville, BP 20451, 54 001 NANCY cedex, France
Abstract
Two international patents and also recent results have shown that polymeric membranes containing cationic
groups are highly efficient for the removal of protic species (e.g., alcohols) from organic mixtures, with
an important potential application for the purification of ethyl-tert-butyl ether (ETBE). By an appropriate
tax policy for the past ten years, the European Union has been strongly encouraging to produce this alkyl
ether from bio-ethanol. Due to its specific advantages and its much better biodegradability than methyl-tert-
butyl ether (MTBE), ETBE is currently considered as one of the most promising bio-fuels. Nevertheless, its
industrial synthesis process leads to an azeotropic mixture containing 20 wt% of ethanol which has to be
removed for ETBE purification. Taking advantage of our former experience with polyimide block copoly-
mers for this separation, we have recently synthesized and characterized 3 families of new ion-containing
polyimide copolymers, with a control of the number of their cationic ammonium groups, the length of their
alkyl side chain and the type of their counter-ions. On the basis of sorption and permeability results obtained
for the purification of ETBE, the membrane features are discussed in terms of structure–property relation-
ships. In particular, it is shown that simply changing the chemical structure of the cationic groups enables
to increase sharply the permeability with a very low impact on selectivity, therefore overcoming the usual
permeability/selectivity trade-off.
Keywords
Polyimide membranes, cationomers, structure–property relationships
1. Introduction
Ethyl-tert-butyl ether (ETBE) is an environment-friendly fuel octane number en-
hancer with a much better biodegradability than methyl-tert-butyl ether (MTBE)
owing to the presence of H atoms in the β-position to the ether group [1, 2]. It is
also considered as one of the most promising bio-fuels in the European Union where
it is widely produced from bio-ethanol using the following reaction with isobutene
(Fig. 1).
*To whom correspondence should be addressed. Tel.: 33 3 83 17 50 29; Fax: 33 3 83 37 99 77; e-mail:
Anne.Jonquieres@ensic.inpl-nancy.fr
340 A. Jonquières et al.
Figure 1. Industrial synthesis of ETBE.
Due to an appropriate tax policy, the French capacity for production of ETBE
has almost doubled over the past ten years, up to a current capacity exceeding
400 000 tons/year according to the French Association for the Development of Agri-
cultural Fuels (ADECA). Nevertheless, the presence of an ethanol (20 wt%)/ETBE
azeotropic mixture in the industrial synthesis process is a recurrent problem. The
conventional separation process involves a ternary or ‘azeotropic’ distillation which
combines three distillation towers, with related high cost in terms of time and
energy. An economical alternative has been proposed by the Institut Français du
Pétrole (IFP — French Petroleum Institute) [3]. This hybrid process combines a
single distillation tower and a membrane separation process. It is both simple and
economical, with yearly energy savings estimated between 2 and 4.7 millions eu-
ros (depending on the membrane properties) for the current French production
capacity [4]. Significant energy savings were also reported by several authors in
closely related works on pervaporation, alone or in hybrid processes [5], for the
purification of MTBE [6, 7]. Coming back to the ETBE process, two international
patents [8, 9] and also recent results [10] have shown that polymeric membranes
containing cationic groups are highly efficient for the removal of ethanol from the
EtOH/ETBE azeotropic mixture. Nevertheless, even if the former polymeric mem-
branes were well performing for this separation, the rather poor control of their
chemical structure did not allow any detailed analysis of the influence of the cationic
groups on their permeability. Our former experience with polyimide block copoly-
mers demonstrated that several of these polymers also display interesting properties
for this particular separation [11–13]. Another advantage of polyimides is the great
variety of chemical structures offered by step-growth polymerization. We have re-
cently developed a new synthesis scheme for cationic polyurethaneimides (PUIs)
based on a careful control of the chemical reactivity in complex polycondensa-
tion systems [14, 15]. Three families of new ion-containing polyurethaneimides
were synthesized from N -methyldiethanolamine (MDEA) with a good control in
the number of cationic groups x, the length of the alkyl side chain R and the type
of counter-ions X − (Fig. 2).
These ion-containing polymers were obtained by quantitatively quaternizing PUI
precursors (PUI/x means it contains a controlled number x of MDEA tertiary amine
groups) (Fig. 2). The first family corresponded to the quaternization of the PUI
precursors with dimethyl sulfate (DMS), leading to PUI cationomers with a con-
trolled number of cationic groups x in the range 0.2–0.7. For the sake of simplicity,
New Ion-Containing Polyimides for the Purification of Bio-fuels 341
Figure 2. Cationic polyurethaneimides investigated in this work and their precursors.
the proposed nomenclature for these polymers is PUI/x/DMS. The second fam-
ily (PUI/0.5/CH3 X) was then obtained by quaternizing a PUI precursor containing
0.5 equivalent of MDEA (PUI/0.5) with various methylating agents CH3 X, thus
giving polymers with different counter-ions X − (i.e., methyl sulfate, triflate, tosy-
late, chloride, bromide and iodide). The third family (PUI/0.5/n-alkyl triflate) was
eventually synthesized by quaternizing the same PUI precursor (PUI/0.5) with var-
ious n-alkyl triflates, leading to polymers varying in the length of their n-alkyl side
chain R. The detailed synthesis, characterization and nomenclature for these PUI
cationomers and their precursors have been described elsewhere [14, 15].
In this work, the permeability features of these new polymers were systemati-
cally explored for the separation of the targeted azeotropic mixture by a membrane
separation process called pervaporation. New structure–property relationships are
proposed for the fundamental analysis of the influence of cationic groups on the
permeability features of cationic polymers for the separation of purely organic mix-
tures by pervaporation.

2.1. Introduction to Pervaporation
Pervaporation (PV) is both a permeability measurement technique with particularly
well-defined boundary conditions for mass transfer, as well as a membrane sepa-
ration process which has already found several industrial applications [16, 17]. As
a membrane separation process, it is usually considered as one of the most cost-
effective and environment-friendly alternatives to conventional processes for the
separation of azeotropes or closely boiling liquids [16, 18]. Pervaporation involves
partial vaporization of a liquid feed mixture through a dense membrane, whose
downstream side is usually kept under low pressure to ensure a strong driving force
for the diffusion process (Fig. 3) [19].
A membrane is usually characterized by its flux and its separation factor, αPV ,
commonly calculated from the feed and permeate compositions according to [20]:
C C
αPV = × (1)
1 − C 1 − C
with C: weight fraction of the preferentially permeated species in the feed mixture,
C : weight fraction of the preferentially permeated species in the pervaporate. The
pervaporate enrichment factor, β, is also defined as:
C
β=
. (2)
C
In this work, all the experiments were carried out with the same feed mixture, i.e.,
the EtOH/ETBE azeotropic mixture at 30◦ C and all the membranes were selective
towards ethanol. Therefore, for the sake of simplicity, the membrane selectivity will
be simply regarded as the ethanol weight fraction of the permeate.
Figure 3. Schematic principle of pervaporation according to Néel [19]. C: weight fraction of the pref-
erentially permeated species in the feed mixture, C : weight fraction of the preferentially permeated
species in the permeate (or pervaporate), J (kg · µm/h · m2 ): permeation flux given for a normalized
membrane thickness of 1 µm.
In the following, the membrane properties will be presented as a function of the

polyimide structure. In particular, the influence of the number of cationic groups x,
the type of their counter-ions X − and the length of their n-alkyl side chain R will
be discussed.
2.2. Influence of the Number of Cationic Groups
Figure 4 shows the variation of the pervaporation flux, J , as a function of the
MDEA content, and thus the number of cationic groups, for polyimide cationomers
obtained by quantitative quaternization of the polyimide precursors PUI/x with
dimethyl sulfate DMS (PUI/x/DMS — Family 1). It appears readily that the PV
flux decreases strongly with the number of cationic groups for this first family of
polyimide cationomers. Moreover, this trend appears much more complex than the
linear decrease observed with the corresponding polyimide precursors.
On the contrary, the pervaporation selectivity C , expressed here as the pervapo-
rate ethanol content for the sake of simplicity, sharply increased with the number
of cationic groups in the polyimide cationomers (Fig. 5) and the less permeable
polyimide cationomer (x = 0.7 equiv.) enabled a pervaporate enrichment in ethanol
by a factor β of more than 3. Nevertheless, a comparison between the properties of
the polyimide cationomers and their precursors revealed that the pervaporation se-
lectivity of the polyimide cationomers did not differ significantly from that of their
precursors. Therefore, it appeared clearly that the quaternization of the polyimide
precursors had a much greater impact on permeability than on selectivity.
It thus appears that the quaternization of the polyimide precursors with dimethyl
sulfate led to polymers with usually decreased permeability and only a slight in-
crease in selectivity. Moreover, this first family of polyimide cationomers followed
the usual permeability/selectivity trade-off in pervaporation, meaning that an in-
Figure 4. Influence of the number of cationic groups x (or polymer MDEA content) on the PV flux.
A comparison between the polyimide cationomers PUI/x/DMS (Family 1) and their polyimide pre-
cursors PUI/x.
Figure 5. Influence of the number of cationic groups x (or polymer MDEA content) on the PV
selectivity C . A comparison between the polyimide cationomers PUI/x/DMS (Family 1) and their
polyimide precursors PUI/x.
Figure 6. Influence of the polymer soft block content on the PV flux J . A comparison between the
polyimide cationomers PUI/x/DMS (Family 1) and their polyimide precursors PUI/x.
crease in selectivity was obtained to the detriment of the flux which decreased
sharply in the same conditions. These first results could be considered as fairly
discouraging for the targeted separation.
Former analyses of the permeability of block copolymers have already shown
that the polymer soft block content is a key parameter for mass transfer through
block polymers [13]. Further analysis of the permeability data with respect to the
polymer soft block content then offered a new insight which gave new prospects for
this work. Figure 6 shows the variation of the PV flux as a function of the polymer
soft block content for the polyimide cationomers and their precursors and reveals
very different trends for their permeability. The PV flux of the ion-containing block
polyimides increased exponentially with their soft block content. This exponential
trend is typical of block copolymers with polyether soft blocks [13]. Nevertheless,
what was really surprising was the behavior of the polyimide precursors whose
PV flux increased linearly (rather than exponentially) with the polymer soft block
content. This particular trend corresponded in fact to what is usually observed with
block copolymers containing polyester (rather than polyether) soft blocks. In this
case, the phase separation is much less than for the related polyether-based block
copolymers owing to much better interactions between the soft and hard blocks.
For the polyimide precursors used in this work, specific interactions involving the
tertiary amine groups, which are H-bonding acceptors, could promote phase mixing
and be the reason for their particular behavior with respect to permeability.
In addition to clearly showing for the first time that the chemical nature of the soft
block is not the sole parameter governing the general permeability behavior of block
copolymers, the comparison between the properties of the polyimide cationomers
and their precursors showed that the chemical structure of the hard blocks also
played a critical role in determining the type of observed behavior. It prompted us
to analyze the influence of the hard block chemical structure on the permeability
features of the polyimide cationomers. The hard block chemical structure was then
varied in two different ways, by varying the type of counter-ion X − and the length
of the n-alkyl side chain R.
2.3. Influence of the Type of Counter-Ions X −
The type of counter-ion was varied by using different methylating agents CH3 X
for the quantitative quaternization of the polyimide precursor containing 0.5 equiv-
alent of MDEA (i.e., PUI/0.5 [15]), eventually leading to six different counter-ions
(methyl sulfate, chloride, bromide, iodide, triflate and tosylate — Fig. 2) for the
polyimides of the second family (PUI/0.5/CH3 X). Nevertheless, the iodide counter-
ion was not stable during pervaporation due its slow oxidation by traces of perox-
ides appearing over time in the EtOH/ETBE azeotropic mixture. Therefore, only
the data obtained for the counter-ions methyl sulfate, chloride, bromide, triflate and
tosylate will be discussed hereafter.
Figure 7 shows that the counter-ion X − had a very strong influence on the PV
flux. Whereas the counter-ion methyl sulfate corresponding to the first family of
polyimide cationomers led to a slight permeability decrease (−22%) compared to
that of the corresponding polyimide precursor (PUI/0.5), all the other counter-ions
were responsible for a strong increase in permeability which was even more than
doubled for the counter-ion tosylate CH3 PhSO3 − .
However, the counter-ion X − had almost no influence on the PV selectivity ex-
cept for the fairly bulky counter-ion tosylate which was responsible for both a very
sharp increase in permeability (×2.5) and a rather strong decrease in selectivity
(−33%) (Fig. 8).
Figure 7. Influence of the counter-ion X − on the pervaporation flux J of the polyimide cationomers
PUI/0.5/CH3 X (Family 2). The flux of the corresponding polyimide precursor (PUI/0.5) is also given
for the sake of comparison.
Figure 8. Influence of the counter-ion X − on the pervaporation selectivity C for the polyimide
cationomers PUI/0.5/CH3 X (Family 2). The selectivity of the corresponding polyimide precursor
(PUI/0.5) is also given for the sake of comparison. The stars correspond to the counter-ions enabling
to overcome the common permeability/selectivity trade-off.
For the counter-ions chloride, bromide and triflate, permeability increased

strongly with a very low impact on selectivity as compared to the properties of
the corresponding polyimide precursor. Therefore, an appropriate choice of the
counter-ion X − enabled to overcome the usual permeability/selectivity trade-off
and to increase the membrane productivity (equal to J (β − 1) [19]) by a factor of
ca. 1.5.
According to the solution-diffusion model [21], the overall permeability of dense
membranes is usually considered as the combined contribution of two complemen-
Figure 9. Influence of the counter-ion X − on the relative total flux J and swelling S for the polyimide
cationomers PUI/0.5/CH3 X (Family 2).
tary steps. The first sorption step consists of polymer swelling at the upstream side
of the membrane. In this first step, a part of the liquid mixture in contact with the
membrane is absorbed by the polymer, corresponding to a thermodynamic equilib-
rium between the liquid phase and the swollen polymer phase. The second step is
the diffusion of the absorbed species through the polymer membrane owing to the
strong chemical potential gradient imposed by the low downstream pressure.
Therefore, a complementary sorption analysis was carried out in the same condi-
tions as for PV to help analyze the reason for the particular permeability behavior.
The total swelling S was determined by measuring the weights, m0 and m, of poly-
mer samples before and after sorption equilibrium in the EtOH/ETBE azeotropic
mixture at 30◦ C:
m − m0
S= × 100. (3)
m0
Figure 9 shows a comparison between the PV flux and the total swelling for the
polyimide cationomers with different counter-ions X − . In this figure, all the data
were normalized relative to the properties of the non-ionic corresponding polyimide
precursor to allow a better comparison of the properties of the different polymers.
A particularly striking fact in Fig. 9 is the quasi-constancy of the total swelling S
for all the polyimide cationomers, whereas the total flux clearly differed from one
cationomer to another. Figure 9 indeed shows that the total swelling did not depend
on the type of counter-ion and was always slightly less than that of the non-ionic
polyimide precursor.
According to the solution-diffusion model, the PV separation factor αPV is usu-
ally considered as the product of the sorption separation factor αS and the diffusion
Figure 10. Influence of the counter-ion X − on the sorption, diffusion and pervaporation separation
factors, αS , αD and αPV , respectively, for the polyimide cationomers PUI/0.5/CH3 X (Family 2). The
properties of the corresponding polyimide precursor (PUI/0.5) is also given for the sake of comparison.
separation factor αD . Therefore, the experimental data on both αPV and αS enable
to estimate the contribution of diffusion to the overall separation factor. Further ex-
periments were thus carried out to determine the sorption separation factor αS by in
situ desorption of the absorbed mixture in carbon tetrachloride in a 1 H NMR tube
according to a new technique we reported recently for determining sorption selec-
tivity in a very simple and efficient way [22]. Figure 10 shows the results for the
three separation factors αS , αD and αPV determined for the EtOH/ETBE azeotropic
mixture at 30◦ C. It appears that the sorption separation factor αS is not significantly
influenced by the type of counter-ion X − and that the slight differences in the PV
separation factor αPV have to be ascribed to the same differences as in the diffusion
separation factor αD . In good agreement with Fig. 9, Fig. 10 again confirms the key
role played by the diffusion step for these polyimide cationomers.
2.4. Influence of the Length of the Alkyl Side Chain R
Another way of varying the chemical structure of the hard blocks is to quat-
ernize the polyimide precursor containing 0.5 equiv. of MDEA with alkylating
agents RX with n-alkyl chain R of increasing length. First attempts were made
with n-alkyl bromides [14] but it turned out that they led to a concurrent elim-
ination reaction responsible for the loss of the structural control of the resulting
polyimide cationomers [15]. Using n-alkyl triflates rather than n-alkyl bromides re-
cently avoided the former drawback by leading to quantitative quaternization and a
good control of the cationomer chemical structure. The corresponding third family
Figure 11. Influence of the length of the n-alkyl side chain R on the PV flux J for the polyimide
cationomers PUI/0.5/n-alkyl triflate (Family 3). The flux of the corresponding polyimide precursor
(PUI/0.5) is also given for the sake of comparison.
Figure 12. Influence of the length of the n-alkyl side chain R on the PV selectivity C for the poly-
imide cationomers PUI/0.5/n-alkyl triflate (Family 3). The selectivity of the corresponding polyimide
precursor (PUI/0.5) is also given for the sake of comparison.
of polyimide cationomers (PUI/0.5/n-alkyl triflate) was thus obtained with increas-

ing length of the alkyl side chain from C1 to C6 [15].
All the polyimide cationomers of this third family displayed pervaporation fluxes
which were higher (×1.5) than that of the corresponding polyimide precursor
(Fig. 11). Quite unexpectedly, the length of the alkyl side chain had almost no
influence on the pervaporation flux (Fig. 11) as well as selectivity (Fig. 12). Never-
theless, the quaternization of the polyimide precursor with n-alkyl triflates provided
another way of overcoming the common permeability/selectivity trade-off in per-

vaporation.
3. Conclusion
A comparison between the polyimide cationomers of the first family (PUI/x/DMS)
and their polyimide precursors showed that very different permeability behaviors
could be obtained with block copolymers containing the same polyether soft block
but different hard blocks. While the former analysis of permeability through block
copolymers had pointed out the importance of the soft block chemical nature [13],
these new results revealed the key role played by the polyimide hard blocks and
prompted further investigations based on new polyimide cationomers with different
hard blocks.
The chemical structure of the hard blocks was then easily varied with different
types of counter-ions (Family 2). An appropriate choice of the counter-ion (i.e.,
chloride, bromide and triflate) led to a sharp increase in permeability with almost
no negative impact on selectivity. A complementary sorption analysis showed that
the sorption step was not significantly influenced by the type of counter-ion and that
the differences in permeability had to be ascribed to the diffusion process.
Another way of varying the chemical structure of the hard blocks was to increase
the length of the n-alkyl side chain of the ammonium cationic groups (Family 3).
Although the resulting polyimide cationomers were much more permeable than
the corresponding polyimide precursor which was very desirable for the targeted
separation, it appeared that the length of the n-alkyl side chain had no influence on
the permeability features over the investigated range (C1–C6).
In conclusion, it appeared that an appropriate choice of the hard block chemical
structure enables to overcome the usual permeability/selectivity trade-off in perva-
poration, with interesting prospects for the design of new block polyimides with
improved permeability properties.
Acknowledgements
The authors wish to acknowledge the French Petroleum Institute and also the
French Membrane Club for having awarded the best PhD Prize 2006 to one of
the authors of this work (Michel Awkal).
References
1. H. Noureddini, ACS Symposium Series 799, 107 (2002).
2. R. Koenen and W. Puettmann, Grundwasser 10, 227 (2005).
3. C. Streicher, US Patent 5,607,557 (1997).
4. E. Maucci and I. Prevost, Gains énergetiques induits par l’introduction d’une opération de perva-
poration dans le procédé de fabrication de l’ETBE (Energy savings by introducing pervaporation
in the ETBE production process), Public report of the French Petroleum Institute (2000).
5. F. Lipnizki, R. W. Field and P. K. Ten, J. Membrane Sci. 153, 183 (1999).

6. M. S. Chen, R. M. Eng, J. L. Glazer and C. G. Wensley, US Patent 4,774,365 (1988).
7. U. Hömmerich and R. Rautenbach, J. Membrane Sci. 146, 53 (1998).
8. H. Steinhauser and H. E. A. Brüschke, Eur. Pat. Appl. 674 940 B1 (1995).
9. H. Steinhauser and H. E. A. Brüschke, US Patent 5,700,374 (1997).
10. S. Touchal, D. Roizard and L. Perrin, J. Appl. Polym. Sci. 99, 3622 (2006).
11. A. Jonquières, D. Roizard and P. Lochon, J. Membrane Sci. 118, 73 (1996).
12. A. Jonquières, R. Clément, C. Dole and P. Lochon, J. Polym. Sci. Part A: Polym. Chem. 38, 614
(2000).
13. A. Jonquières, R. Clément and P. Lochon, Prog. Polym. Sci. 27, 1803 (2002).
14. M. Awkal, A. Jonquières, G. Creffier, R. Clément and P. Lochon, Macromolecules 37, 684 (2004).
15. M. Awkal, A. Jonquières, R. Clément and P. Lochon, Polymer 47, 5724 (2006).
16. A. Jonquières, R. Clément, P. Lochon, J. Néel, M. Dresch and B. Chrétien, J. Membrane Sci. 206,
87 (2002).
17. P. Shao and R. Y. M. Huang, J. Membrane Sci. 287, 162 (2007).
18. P. Aptel, N. Challard, J. Cuny and J. Néel, J. Membrane Sci. 1, 271 (1976).
19. J. Néel, Pervaporation. Lavoisier Tec&Doc Publisher, Paris (1997).
20. K. W. Böddeker, J. Membrane Sci. 51, 259 (1990).
21. J. Wijmans and R. Baker, J. Membrane Sci. 107, 1 (1995).
22. M. Awkal, A. Jonquières, O. Fabre, R. Clément and P. Lochon, J. Membrane Sci. 286, 3 (2006).
Zeolite-Modified Polyimides as Absorption Materials
for Molecular Contamination
Shabnam Virji ∗ , Sara Halper, Judy Ying, Rika Anderson and Randy Villahermosa
Physical Sciences Laboratory, The Aerospace Corporation, P.O. Box 92957/M2-242,
Los Angeles, CA 90009, USA
Abstract
Polyimide is a unique spacecraft material because of its high strength, heat and chemical resistance. Mod-
ifying polyimide with chemical specificity while retaining its material properties is important to develop
multi-functional materials to solve challenging contamination, thermal regulation, and space environment
problems. A particular contaminant of interest is water. Moisture contamination leads to deterioration of
both absorbance and transmittance properties of optical surfaces. Zeolites have been widely known to ab-
sorb water and other molecular contaminants. These moisture absorption properties are further enhanced
when synthesized on the nanoscale level. Nanozeolite particles were surface modified to promote uniform
inclusion into the polyimide matrix. The polyimide–nanozeolite composite material shows only little devia-
tion from the parent material when characterized for thermal and optical properties. However, the composite
material exhibits outstanding moisture absorption properties when compared to the unmodified parent ma-
terial. The amount of nanozeolite was varied in the composite material and there appears to be a threshold
value of nanozeolite that produces the best water absorption characteristics as well as comparable thermal
and optical characteristics to the unmodified polyimide.
Keywords
Polymer, polyimide, nanozeolite, hybrid material, spacecraft contamination, moisture
1. Introduction
The accumulation of contamination on orbiting satellites is detrimental to the
performance of their optical systems [1]. A contaminant of interest is moisture.
Polymer composites used in the construction of spacecraft structures and compo-
nents outgas rapidly in vacuum, yielding moisture as the primary contaminant. This
outgassing can lead to ice formation on hardware that operates at cryogenic tem-
peratures. In order to minimize the ice formation, active thermal systems can be
used to ‘bake out’ the water. This approach adds to spacecraft weight, power, and
volume which leads to higher costs. Therefore, there is a need to develop multi-
* To whom correspondence should be addressed. Tel.: 310-336-7555; Fax: 310-563-2551; e-mail:

shabnam.virji@aero.org
354 S. Virji et al.
functional materials that can absorb moisture contamination with minimial impact
to spacecraft design.
Polyimides possess a broad array of superior properties over other polymers,
such as excellent thermal and chemical stability [2, 3]. Polyimides are currently
being used as insulators on spacecraft hardware. In order to make polyimides more
multi-functional, zeolites are incorporated into the polymer for water absorption ca-
pability. Zeolites are inorganic materials that have a porous molecular structure with
different diameter pore sizes [4] that can absorb contaminants. Nanoscale zeolites
have recently received much attention due to their small size that offers a greater
absorption surface area [5]. This paper focuses on the synthesis and characterization
of polyimide–nanozeolite multi-functional materials and their moisture absorption
characteristics.
2. Experimental
2.1. Polyimide Synthesis
Polyimide is synthesized by curing poly(amic acid) at high temperatures. Poly(amic
acid) is synthesized through a well-known process that begins with the condensation
of a diamine with a dianhydride [2, 3]. The polyimide synthesized here is similar to
Kapton® H (Scheme 1). Other types of polyimides can easily be prepared by mak-
ing substitutions on the diamine and dianhydride building blocks. Substitution of
the polymer needs to be performed before the polymer cure process, because once
the polyimide is formed, it becomes very difficult to modify the polymer further.
All glassware was pre-dried in an oven and hooked up to a nitrogen line
as it cooled to eliminated moisture from contaminating the glassware. N,N-
Dimethylacetamide (Aldrich, Sure-Seal Bottle® ) was used as the solvent. 4,4 -
Oxydianiline (Aldrich, 0.902 g, 4.50 mmol) was dissolved in dry dimethylac-
etamide while stirring under nitrogen for 5 min. Pyromellitic dianhydride (Aldrich,
1.001 g, 4.59 mmol) was added as a solid to the oxydianiline and dissolved slowly
over time. The reaction mixture was allowed to stir overnight under nitrogen to
Scheme 1. Synthesis of poly(4,4 -oxydiphenylene-pyromellitimide) (Kapton® H).

Zeolite-Modified Polyimides as Absorption Materials for Molecular Contamination 355
yield a viscous pale yellow solution of poly(amic acid). The viscosity of the solu-
tion could be altered by adding different amounts of dimethylacetamide.
Curing the poly(amic acid) leads to the formation of the polyimide. The cure
cycle consists of ramping the temperature to 80◦ C over 30 min, holding at 80◦ C
for 3 h, ramping the temperature to 150◦ C over 30 min, holding at 150◦ C for 1 h,
ramping the temperature to 300◦ C over 1 h, and holding the temperature at 300◦ C
for 2 h.
2.2. Nanozeolite Synthesis
The nanozeolite particles were synthesized using a modified literature procedure [6]
as outlined in Scheme 2. The nanozeolites are a silicate variety, TPA-silicalite-1,
synthesized by the hydrolysis and subsequent polymerization of tetraethyl orthosil-
cate (TEOS). The nanoscale particle size is achieved by controlling the alkalinity
of the solution. A high pH induces a high initial nucleation rate and creates a larger
number of relatively smaller particles [7–9].
The nanozeolites were synthesized by first adding tetraethyl orthosilicate (TEOS,
Aldrich, 14 ml) to a polypropylene bottle. While this solution was stirring
tetrapropylammonium hydroxide (TPAOH, Aldrich, 22.6 ml), doubly distilled wa-
ter (3.4 ml) and sodium hydroxide (Aldrich, 0.02 g) were added slowly. After
everything was added, the solution was stirred for another 10 min. The solution was
then shaken for 12 h on a gyratory shaker to allow for hydrolysis of the tetraethyl
orthosilicate. A gyratory shaker allows the solution to be gently shaken in a circular
motion. After shaking the solution, the solution was heated without stirring in an
oil bath at 98◦ C for 24 h. After heating, the solution was cooled to room tempera-
ture and centrifuged for 3 h at 36 000 g/4◦ C. The supernatant was removed and the
zeolite pellet was resuspended in doubly distilled water. The centrifugation and su-
pernatant removal steps were repeated. The rinsed nanozeolites were resuspended
in 2 ml doubly distilled water. The nanozeolites were transferred to a small glass
vial and freeze suspended in a liquid nitrogen bath. Freeze suspension occurred by
rotating the vial in a liquid nitrogen bath so that the nanozeolites coated the inside
surface of the vial. The vial with the suspension was incorporated into an apparatus
that was attached to a Schlenk line. The apparatus was submerged into a liquid ni-
trogen bath while under vacuum overnight in order to remove excess solvent. The
nanozeolites adhered well to the glass surface of the vial because of the chemical
Scheme 2. Nanozeolite synthesis.

356 S. Virji et al.
Scheme 3. Functionalization of nanozeolite particle surface.
compatibility of the nanozeolites and the silica glass. The resulting dried nanozeo-
lites were then removed from the liquid nitrogen bath and vacuum and modified as
described below.
The functionalization of the nanozeolite particles was achieved by modifying
a literature synthesis method [10] that was developed to chemically modify glass
surfaces with organic molecules. Chemically, the surface of the nanozeolites is very
similar to glass, thus making the literature method applicable. The synthesis of the
functionalized nanozeolites is outlined in Scheme 3.
Toluene (2.5 ml) was added to the dry nanozeolites and the suspension was son-
icated to fully suspend the nanozeolites (5 pulses for 10 s/pulse). A small amount
of water (10 µl) was added to the toluene suspension and the suspension was
stirred for 10 min. Different amounts of 3-aminopropyltrimethoxysilane (APTMS)
were added to the nanozeolite–toluene suspension. According to the literature [10]
19 mmol of APTMS per gram of zeolite is the ideal quantity to modify the ze-
olite. Four different samples of functionalized nanozeolite made with 19 mmol,
28.5 mmol, 38 mmol and 76 mmol APTMS were added to the nanozeolite–toluene
suspension. These suspensions were then stirred at room temperature overnight.
Each suspension was centrifuged for 2 h at 36 000 g/4◦ C. The supernatant was re-
moved and 2.5 ml of fresh toluene was added. The solution was refluxed at 90◦ C
for 1 h, cooled, and centrifuged again for 2 h at 36 000 g/4◦ C. The supernatant was
removed and 15 ml of isopropanol were added to the nanozeolites. This new iso-
propanol dispersed nanozeolite suspension was centrifuged for 2 h at 36 000 g/4◦ C.
The supernatant was removed and the purification process was repeated. The sam-
ples were then transfered to glass vials and dried in a vacuum oven at 100◦ C.
2.3. Polyimide–Nanozeolite Hybrid Material Synthesis
2.3.1. Mixing Technique
The method for blending the functionalized nanozeolites into polyimide materi-
als involves combining the two materials at a precursor stage of the polymer cure
process. Poly(amic acid), the viscous precursor of polyimide, was mixed with the
functionalized nanozeolites. The mixture was then cured through a standard proce-
dure for converting the acid into polyimide. During the curing process, the polymer
covalently attaches to the ‘tails’ that decorate the surface of the nanozeolites, pro-
moting an even distribution of nanozeolite throughout the polymer.
The APTMS–nanozeolite was suspended in 3 ml of N,N-dimethylacetamide
(0.008 g/ml). The nanozeolite suspension was sonicated for 10 s to disperse the
nanozeolite in the solvent. In a glass vial with 4 ml of poly(amic acid), while
stirring, 2.379 ml of the nanozeolite suspension was added. The dispersion was
stirred for 4 h. Films were made by drop-casting the suspension on a Teflon
substrate and curing the poly(amic acid)/nanozeolite suspension to form the poly-
imide/nanozeolite hybrid film.
2.3.2. Airbrushing Technique

Airbrushing the nanoparticles consists of using a fixed concentration of nanozeo-
lites in the solvent and spraying the nanozeolites onto a partially cured poly(amic
acid) film using an artist’s airbrush. The resulting film is then cured completely
resulting in a polyimide–nanozeolite hybrid film.
Nanozeolites (0.02387 g) were added to isopropanol (7.6 ml) and stirred
overnight at room temperature to fully suspend the nanozeolites in the solvent.
Poly(amic acid) (3 µl) was drop-cast onto a quartz crystal and placed in an oven to
partially cure it at 60◦ C for 30 min. This allowed solvent to be evaporated and pro-
duced a film that was less viscous. The nanozeolite suspension (500 µl) was sprayed
with nitrogen as the background gas on the surface of the film with a mask and the
final hybrid material was placed back in the oven to complete the cure process.
2.3.3. Stamping Technique

Stamping results from placing a partially cured poly(amic acid) film flat onto a sur-
face that has been airbrushed with nanoparticles. The resulting film is then cured
completely resulting in a polyimide–nanozeolite hybrid film. This stamping tech-
nique results in adhesion of a random number of nanoparticles to the surface of the
polyimide film, thus there is no control over how much nanozeolite attaches to the
surface.
The nanozeolite–isopropanol suspension mentioned above was sprayed onto a
Teflon block. The partially cured poly(amic acid) film on a quartz crystal surface
mentioned above was then placed face-down on the Teflon block to ‘stamp’ the
poly(amic acid) film with the nanozeolites. This Teflon block with the crystal sitting
on it face down was then placed back in the oven in this position to complete the
cure process.
2.4. Polyimide–Nanozeolite Hybrid Material Characterization
Thermogravimetric Analysis (TA Instruments) was performed with a heating rate of

10◦ C/min in a nitrogen atmosphere. To perform thermogravimetric measurements
a free-standing film of polyimide or the polyimide–nanozeolite hybrid material
was used. The free-standing film was made by curing either the poly(amic acid)
or poly(amic acid)–nanozeolite mixed solution cast in a thin film mold.
358 S. Virji et al.
Moisture uptake experiments were performed using a quartz crystal microbal-

ance. A 6 MHz quartz crystal obtained from Sycon Instruments (Syracuse, NY)
was used. Moisture uptake was calculated using the Sauerbrey equation [11] and
calculating the mass change of the crystal from the frequency change. Equations
(1)–(3) show the Sauerbrey equations and how the percent moisture uptake was
calculated. In these equations fq is the uncoated crystal frequency, fc is the coated
crystal frequency, ρq is the density of quartz (2.649 g/cm3 ), νq is the shear wave
velocity of quartz (332 200 cm/s), M is the mass of the film, Mw is the mass of the
film under a humid (wet) environment, and Md is the mass of the film under a dry
environment (no humidity). Films were made on the quartz crystal microbalance by
drop-casting 3 µl of the poly(amic acid) precursor solution. This film was then cured
under the heating conditions stated above to form a well adhered film on the sensor.
All experiments were performed using gas flows generated from calibrated mass
flow controllers (MKS). The humidity was measured using a humidity meter (Vi-
asala) and the humidity was generated by flowing nitrogen gas into a bubbler that
contained water.
2fq2
Cf = , (1)
ρq × νq
fq − fc
M= , (2)
Cf
Mw − Md
uptake (%) = × 100. (3)
Md

3.1. Functionalized Nanozeolite Particles
Zeolites are inorganic silicate or aluminosilicate materials that have a porous mole-
cular structure. The structured pores within the material, which are typically on the
order of 0.1 to 10 nm in diameter, are ideally sized to adsorb and retain common
spacecraft contaminants, including moisture. Nanoscale zeolite particles afford two
advantages: (1) the small size, which is comparable to the molecular dimensions
of polymer chains, is expected to promote homogeneous blending with the host
polymer, and (2) the smaller diameter would afford a greater adsorption surface
area, which becomes an important factor when they are incorporated into pseudo
two-dimensional structures, such as polymeric thin film materials.
The nanozeolites were characterized using SEM analysis (Fig. 1), which shows
that particle sizes ranged from 65 nm to 80 nm. The SEM image also shows that
the particles are not aggregating, which is a common problem for larger zeolite
particles.
In order to successfully incorporate the nanozeolite particles into polymeric ma-
terials without any of the side-effects common to blended materials — such as
extrusion of the additive, loss of tensile strength or ductility, or inhomogeneous
Figure 1. SEM image of nanozeolite particles.
blending — an important technical advance was needed. Inspired by biological

systems, in particular the action of viruses on living cells, a novel scheme for mod-
ifying the surface of the nanozeolites was conceived (Scheme 3). In this scheme, an
inorganic–organic linker molecule, 3-aminopropyltrimethoxysilane (APTMS), was
covalently attached to the zeolite to coat the surface of the nanoparticle through a
silane moiety. The other end of the linker molecule was then able to bind to the
polymer itself, thus incorporating the zeolite nanoparticles into the polymer matrix.
3.2. Polyimide–Nanozeolite Hybrid Materials
A major driver behind poor blending of additives into polymers is the chemical
mismatch between the two starting materials. Surface energies, differences in polar-
ities, and incompatibilities in chemical functional groups all contribute to rejection
of additives. These challenges were overcome by ‘tricking’ the polymer into think-
ing the nanozeolites were a native or friendly material by modifying the surface of
the particles with a chemical ‘tail’ consisting of the inorganic–organic linker mole-
cule, 3-aminopropyltrimethoxysilane (APTMS). One end of the linker molecule has
a silane group that covalently bonds to the zeolite surface, while the other end of
the linker molecule is able to bind to the polymer itself, thus homogeneously incor-
porating the zeolite nanoparticles into the polymer matrix. By making the surface
appear to be more ‘polymer-like’, the nanozeolite tricks the polymer into accepting
the particle in much the same way that a virus is able to gain access to the interior
of a cell.
The result of a good blend would be the creation of a multi-functional version of
polyimide that has all of the favorable thermal and strength properties of the unmod-
ified material, but with the added benefit of water absorption to protect cryogenic-
and contamination-sensitive spacecraft surfaces.
Combining the poly(amic acid), the polyimide precursor, and the nanozeolites
should result in the covalent attachment of the polymer to the APTMS functional-
360 S. Virji et al.
Figure 2. Proposed distribution of nanozeolite particles in polyimide matrix.
(a) (b)
Figure 3. SEM images of an unmodified polyimide film (a) and of nanozeolite–polyimide film (b).
ity of the nanozeolites after the cure process. Figure 2 shows a schematic of this
new hybrid material. As depicted in Fig. 2, the addition of the polymer chain to the
nanozeolite particles should promote their even distribution throughout the poly-
mer. Ideally, the zeolites would only be present at the top of the film but as will be
shown later this method was not designed to do that.
The cured, nanozeolite–polyimide multi-functional material and unmodified
polyimide films were characterized with SEM (Fig. 3). The SEM image of the
multi-functional material has particle-like features, with a typical diameter of 3 µm
or less. These features are likely to be clusters of nanoparticles, as examined from
energy dispersive X-ray analysis measurements. The energy dispersive X-ray analy-
sis measurement shows the appearance of silicon and aluminum when examining
the particle-like features of the multi-functional material, which is indicative of
nanozeolite properties.
(a) (b)
Figure 4. SEM images of airbrushed nanozeolite–polyimide film at the center of the film (a) and at
the edge of the film (b).
Figure 5. SEM image of stamped nanozeolite–polyimide film.
As seen from Fig. 3 the nanozeolite particles are dispersed both within the poly-
imide film as well as on the surface of the film. In order to achieve higher absorbing
properties it is ideal to have the nanozeolites situated mostly on the surface of
the film. Two techniques were utilized to achieve this orientation: airbrushing and
stamping the nanoparticles onto the surface of a partially cured polyimide film.
SEM images of the airbrushed samples are shown in Fig. 4. Figure 5 shows the
SEM image of the stamped nanoparticle polyimide sample.
As seen from Figs 4 and 5 the different techniques of incorporating the nanoze-
olites into the polyimide film yield different amounts of zeolite on the film sur-
face. The stamping technique yields zeolites only on the polyimide surface but the
amount of zeolite is very small. The amount of zeolite in the stamping technique
can vary randomly depending on the amount of zeolite sprayed on the surface. The
airbrushing technique yields more zeolite on the surface but most of it is concen-
trated on the edges of the film. This edge effect is a result of the airbrush pushing
the zeolites with air to the edge of the film.
362 S. Virji et al.
3.2.1. Thermogravimetric Analysis

Thermogravimetric analysis (TGA) and the TGA first derivative for the nano-
zeolite–polyimide hybrid materials are shown in Fig. 6. The TGA measures the
weight change of the sample as the temperature is increased. The TGA first deriva-
tive shows the temperature at which there is a change in the behavior of the material.
Figure 6 shows the temperature characteristics of unmodified polyimide, nanozeo-
lites and different nanozeolite–polyimide hybrid materials that were modified with
different amounts of the APTMS linker added to the nanozeolite, with sample A
(a)
(b)
Figure 6. TGA results for nanozeolite–polyimide films (a) and TGA first derivatives of nanozeo-
lite–polyimide films (b). The four different samples were made with 19 mmol (A), 28.5 mmol (B),
38 mmol (C) and 76 mmol (D) APTMS added to the nanozeolite suspension that was then mixed with
the polyimide to make the composite material.
having the smallest amount of linker and sample D having the largest amount. This
data set includes nanozeolites that were mixed into the polyimide. As seen from
the figures unmodified polyimide is stable to about 600◦ C after which it begins to
degrade. Nanozeolite, not incorporated into polyimide, degrades at around 300◦ C.
Polyimide–nanozeolite hybrid sample A most closely matches the unmodified poly-
imide trace and is the most stable of the different polyimide–nanozeolite samples.
Polyimide–nanozeolite samples B, C and D have first derivative traces with 2 peaks.
The small peak at the lower temperature is assumed to be the degradation of the
nanozeolite. When compared to the pure nanozeolite, the nanozeolites trapped in
the polyimide matrix are more stable because they degrade at higher temperatures.
3.2.2. FT-IR Spectra
Figures 7 and 8 show the FT-IR traces of the unmodified and various modi-
fied polyimide–nanozeolite hybrid films. Table 1 shows the FT-IR peak assign-
Figure 7. FT-IR spectra of unmodified polyimide (top) and modified nanozeolite–polyimide film A
(bottom). The composite material (A) was made by adding 19 mmol APTMS to the nanozeolite sus-
pension which was then mixed with the polyimide.
Figure 8. FT-IR spectra of modified nanozeolite–polyimide films B, C and D. The three different sam-
ples were made with 28.5 mmol (B), 38 mmol (C) and 76 mmol (D) APTMS added to the nanozeolite
suspension which was then mixed with the polyimide to make the composite material.
364 S. Virji et al.
Table 1.
Peak assignments for the polyimide film
Peak position (cm−1 ) Peak assignment
1780 C=O (in-phase)

1720 C=O (out-of-phase)
1513 C=C (benzene)
1370 C=O (imide)
1100 (OC)2 NC (imide)
720 Imide ring deformation
Figure 9. QCM measurements on unmodified polyimide while varying the relative humidity (right
axis) and cycling the valve on and off (left axis, 1 is valve on and 0 is valve off).
ments [12]. As seen from the FT-IR traces the unmodified polyimide and modified
polyimide–nanozeolite, sample A has the most closely matched spectrum, while
samples B, C, D have different FT-IR traces. As more APTMS–nanozeolite is added
to the polyimide films the C=O in- and out-of-phase peak, C=C benzene, and
C=O imide peaks are shifted to the right. These peaks also become sharper as more
APTMS–nanozeolite is added to the polyimide. The (OC)2 NC imide peak becomes
broader as more APTMS–nanozeolite is added in the film as seen from Fig. 8.
3.2.3. Contamination Absorption Properties
A quartz crystal microbalance was used to measure the mass uptake upon exposure
to the contaminant, water. Figure 9 shows a typical experiment with unmodified
polyimide. Valve ‘on’ (moisture uptake 1.0%) means exposure to contaminant and
valve ‘off’ (moisture uptake 0.0%) means no exposure to contaminant. The humid-
ity variation from 10% to 100% is shown on the right axis. The solid black trace
is the moisture uptake in percent of the unmodified polyimide film, shown on the
(a)
(b)
Figure 10. QCM measurements on nanozeolite–polyimide films (a) and zoom-in of first trace of
QCM measurements of nanozeolite–polyimide films (b). The three different samples were made with
19 mmol (A), 38 mmol (C) and 76 mmol (D) APTMS added to the nanozeolite suspension which was
then mixed with the polyimide to make the composite material.
left axis. The moisture uptake percentage is calculated using the Sauerbrey equation
where the mass is calculated from the change in frequency between the dry and wet
polyimide film. At 100% humidity this film takes up 3.5% water. The moisture up-
take is completely reversible, when the valve is turned off the water is immediately
desorbed from the film.
Figure 10 shows the response of different polyimide–nanozeolite films. Again
the different films are indicative of the different amounts of APTMS linker added
to the nanozeolites in the polyimide film, with sample A having the smallest amount
366 S. Virji et al.
of linker and sample D having the largest amount. As seen from the figure, it
appears that the nanozeolite-modified polyimide outperforms the unmodified poly-
imide when examining the moisture uptake experiment. Also, there is a difference in
the moisture uptake between samples A, C and D, with sample A having the largest
response compared to samples C and D which have similar responses; sample B
was a defective sensor. This figure also shows that the response is fully reversible,
with the water desorbing from the film when the valve is turned off. Ideally, the
zeolites would absorb the water and retain it. However, upon closer examination of
a cycle on the graph, it shows that the nanozeolite-modified polyimide does appear
to have a longer desorbing time than the unmodified polyimide (Fig. 10). The un-
modified polyimide film returns immediately back to its initial weight upon valve
closure and purging of the film in dry nitrogen, readily releasing any water that
it absorbed upon exposure to moisture. The nanozeolite-modified polyimide, how-
ever, does not readily release its absorbed water, taking longer to reach back to its
initial weight. This shows that there is promise for incorporation of nanozeolites
into polyimide films to increase their moisture absorption. The mixing technique
was used to make the films resulting in small amounts of nanozeolite particles on
the surface of the film, most of the nanozeolite particles are embedded within the
polyimide film. In order to increase the amount of nanozeolites on the surface of
the film the airbrushing and stamping techniques were used. If more nanozeolites
resided on the surface of the film there is a chance to further increase the moisture
uptake of the film. However, both the stamping and airbrushing techniques result in
irreproducible films because the amount of nanozeolite added to the film surface is
highly dependent on the amount of nanozeolite airbrushed.
4. Conclusions
Polyimide is an excellent spacecraft material due to its high chemical and heat
resistance and can be used as a precursor to create multi-functional materials
with wider property ranges to solve contamination and space environment prob-
lems. Here, polyimide was modified with nanozeolites to absorb the contami-
nant, moisture. Nanozeolites were used because they have porous structures that
have been known to absorb different molecules depending on the pore size.
In order to fully incorporate the nanozeolites, an inorganic material, into the
polyimide, an organic material, the nanozeolites were modified with an organic
linker, 3-aminopropyltrimethoxysilane (APTMS). This linker helped the nanozeo-
lites disperse well throughout the polyimide film creating a new multi-functional
polyimide–nanozeolite hybrid material. The properties of this new material were
examined with thermogravimetric analysis (TGA), FT-IR, and quartz crystal mi-
crobalance moisture uptake experiments. Different amounts of APTMS were added
to the nanozeolites to make different hybrid films, of which the best film was
produced by adding 19 mmol APTMS to the nanozeolite. The film properties of
the material with 19 mmol APTMS most closely resembled the parent unmod-
ified polyimide through examination with both TGA and FT-IR. This film also
exhibited outstanding moisture absorbing capabilities compared to the unmodified
polyimide film. In order to further enhance the composite film’s moisture absorb-
ing characteristics more surface nanozeolites need to be present on the film allowing
more nanozeolite active sites for the moisture to interact with. In order to produce
films with more surface nanozeolite structures that should promote greater moisture
absorbing characteristics more reproducible airbrushing and stamping techniques
need to be examined. The kinetics of absorption and desorption will be examined
to determine how much better the modified film is compared to the unmodified film.
Acknowledgement
We would like to thank The Aerospace Corporation’s Independent Research and
Development Program for funding.
References
1. C. Stein, T. R. King, W. G. Wilson and R. Robertson, Proc. SPIE 3427, 56–64 (1998).
2. D. L. Dunson, Ph.D. Thesis, Department of Chemistry, Virginia Polytechnic Institute and State
University, Blacksburg, VA (2000).
4. B. Lindlar, M. Lüchinger, A. Röthlisberger, M. Haouas, G. Pirngruber, A. Kogelbauer and
R. Prins, J. Mater. Chem. 12, 528–533 (2002).
5. B.-Z. Zhan, M. A. White, M. Lumsden, J. Mueller-Neuhaus, K. N. Robertson, T. S. Cameron and
M. Gharghouri, Chem. Mater. 14, 3636–3642 (2002).
6. A. E. Persson, B. J. Schoeman, J. Sterte and J.-E. Otterstedt, Zeolites 14, 557 (1994).
7. B. J. Schoeman, J. Sterte and J.-E. Otterstedt, Zeolites 14, 568 (1994).
8. S. M. Auerbach, K. A. Carrado and P. K. Dutta, in: Handbook of Zeolite Science and Technology.
Marcel Dekker, New York, NY (2003).
9. J. N. Watson, L. E. Iton, R. I. Keir, J. C. Thomas, T. L. Dowling and J. W. White, J. Phys. Chem. B
101, 10094 (1997).
10. M. Luechinger, R. Prins and G. D. Pirngruber, Microporous Mesoporous Mater. 85, 111 (2005).
11. G. Sauerbrey, Zeit. Physik 155, 206–222 (1959).
12. S.-J. Park, E.-J. Lee and S.-H. Kwon, Bull. Korean Chem. Soc. 28, 188–192 (2007).
Cobalt-Containing Polyimides for
Moisture Absorption and Sensing
Sara R. Halper ∗ , Shabnam Virji and Randy M. Villahermosa

Space Materials Laboratory, The Aerospace Corporation, P.O. Box 92957/M2-242,
Los Angeles, CA 90009, USA
Abstract
Contamination is a constant concern for many space-grade materials. Therefore, the ability for materials to
sense and absorb molecular contamination is greatly needed. A polyimide has been created that both senses
and absorbs moisture, a contaminant detrimental to optical space systems. The material was prepared by
incorporating cobalt(II) chloride, a well-known moisture absorber and sensor, into polyimide, with the intent
of retaining the original properties of both the metal and the polymer. The final material was found to be
fully cured, retain the metal without any leeching, and was thermally stable. Further, the material underwent
a visible color change when exposed to moisture. The moisture absorption could also be observed through
infrared spectroscopy. Finally, the material could be easily regenerated through drying, such as exposing it
to nitrogen.
Keywords
Polyimide, cobalt, moisture, contamination, infrared spectroscopy
1. Introduction
Molecular contamination has been a general concern for many aerospace materi-
als and systems including optical systems. Moisture as a contaminant is a common
problem for space hardware, especially sensors that operate in the cryogenic tem-
perature regime where ice formation is a possibility. Typical mitigation strategies
involve the use of active thermal systems to ‘bake out’ the water. This approach
assumes porous materials, such as composite structures, thermal blankets, and
other non-metallic components, will absorb a certain amount of moisture. While
removing water on-orbit has proven effective, it is time consuming and requires
intervention from a ground operator. The ability of a space-grade material to sense
and absorb water is of great use, possibly eliminating or reducing the need for bake-
outs.
* To whom correspondence should be addressed. Tel.: 310-336-1397 (work); Fax: 310-563-5708; e-mail:
Sara.R.Halper@aero.org
370 S. R. Halper et al.
Polyimides are polymers used for various aerospace applications such as part
of multilayer thermal blankets or as electrical insulators. Our goal was to obtain a
water absorbing polyimide, which could serve to protect underlying materials from
moisture damage. Polyimides are known to absorb only very small amounts of wa-
ter at high levels of humidity [1, 2]. The addition of a second moisture absorbing
material would greatly increase the total amount of moisture absorbed by the poly-
imide. Previous polymer modifications have been performed to decrease the amount
of water absorption for certain applications, such as making dielectric materials [3],
but no example could be found that increased the amount of moisture absorption.
Certain metal complexes such as cobalt(II) chloride are known to readily bind
to water. It is also widely used in moisture absorption applications because of the
physical color change it undergoes during binding, thus also creating a moisture
sensor/indicator [4]. It can be coated onto materials such as calcium sulfate to create
a desiccant with a visible moisture sensor. It has also been embedded into materials
such as polymers and composites [5, 6].
2. Experimental
2.1. Materials
4,4 -Oxydianiline (Aldrich), pyromellitic dianhydride (Aldrich), cobalt(II) chlo-
ride hexahydrate (Aldrich), and anhydrous N,N-dimethylacetamide (Aldrich,
Sure/Seal™ bottle) were used without further purification and drying.
2.2. Synthesis of Unmodified Polyimide (PI)
4,4 -Oxydianiline (0.901 g, 4.50 mmol) was dissolved in 22 ml of N,N-dimethyl-
acetamide (DMAc) while stirring under N2 . Pyromellitic dianhydride (1.001 g,
4.59 mmol) was added and the reaction mixture was stirred for 5 h. The viscous
solution was drop-cast on Teflon® (DuPont™) blocks and cured under nitrogen in
an oven using the following cure schedule: 80◦ C for 3 h, 150◦ C for 1 h and 300◦ C
for 2 h, yielding a thin yellow film of PI.
2.3. Synthesis of Cobalt Containing Polyimide (Co-PI)
The poly(amic acid) was prepared as above. Cobalt(II) chloride hexahydrate
(0.054 g, 0.23 mmol) was added to the viscous solution and stirred for 1 h. The
blue solution was drop-cast on Teflon® blocks and then cured following the same
cure schedule as above, yielding a thin green film of Co-PI.
2.4. Analysis and Characterization
Thermogravimetric analysis (TGA) was performed using a TA Instruments 2950
HR Thermogravimetric Analyzer under nitrogen. Infrared spectra were obtained
using a Nicolet Magna 550 FT-IR spectrometer. Scanning Electron Microscopy
(SEM) and Energy-Dispersive X-ray Spectroscopy (EDX) was performed using
a JSM-6460LV (JEOL) Field Emission Scanning Electron Microscope with an
Oxford INCA x-sight 7573 Detector.
Cobalt-Containing Polyimides 371

3.1. Preparation
The incorporation of metal ions and metal complexes into polyimides is a well-
established concept and has been executed by many research groups [7–19]. While
cobalt(II) chloride-containing polyimides have previously been prepared [20–22],
a new procedure was adapted in this study to make the synthesis more facile. The
preparation began with the standard formation of the poly(amic acid) (Scheme 1).
The addition of the cobalt(II) chloride before curing insured a better incorporation
of the metal in the polymer. The cobalt(II) chloride was not dried prior to incor-
poration, therefore it did not have to be handled under nitrogen. The small amount
of water bound to the cobalt(II) chloride was quickly extracted by the addition of
the DMAc solvent, resulting in the immediate color change from a purple solid to a
blue solution. The use of Teflon® blocks made the material easy to delaminate after
curing.
3.2. Characterization
The final unmodified polyimide was a thin flexible yellow film. While the cobalt
poly(amic acid) solution was a blue color, curing of the material formed a turquoise
solid at elevated temperatures to finally a green solid at full curing (Fig. 1).
While the film appeared to be uniform in color, suggesting that the metal was
well dispersed, SEM–EDX was used to look for any large scale clustering of the
cobalt within the polymer. No regions of high and low metal concentrations were
Scheme 1. Synthesis of Co-PI.
Figure 1. Images of PI (left) and Co-PI (right).

Figure 2. SEM–EDX spectra of PI (top) and Co-PI (bottom).
observed; therefore, it is assumed that the metal was evenly distributed throughout
the polyimide. In addition to a peak indicative of cobalt, a peak for chlorine was
also observed (Fig. 2). This leads us to believe that the chlorine was still bound to
the cobalt, and was not liberated from the metal at elevated temperatures.
3.3. Thermal Behavior
The incorporation of a new component into a polyimide can alter the thermal be-
havior of the material. For example, if a metal compound prevents the cyclization
of poly(amic acid) to form polyimide, it may reduce the thermal stability of the
polymer. TGA was performed on the modified polyimide to determine its thermal
stability. As can be seen in Fig. 3, Co-PI is less thermally stable compared to the
unmodified PI, yet maintains its integrity up to at least 300◦ C, which will make the
material suitable for a variety of applications. As is common with most polyimides,
Figure 3. TGA results of PI (—–) and Co-PI (- - - -) under nitrogen.
we were unable to accurately determine the glass transition temperature, Tg , of the

modified material.
3.4. Degree of Imidization
Fourier Transform-Infrared Spectroscopy (FT-IR) was used to determine the de-
gree of imidization of the polyimides. The degree of imidization was calculated
using the ratio of the C–C stretching of para-substituted benzene (1500 cm−1 ) and
C–N stretching of the imide group (1381 cm−1 ) of the freshly prepared materials
and compared to the commercially available polyimide Kapton® H (trade name of
DuPont). The unmodified polyimide had a degree of imidization of 1.0, showing
that our synthetic preparations were well suited to form a highly imidized material
(Fig. 4). Interestingly, the cobalt-modified polyimide also had a degree of imidiza-
tion of 1.0, comparable to Kapton® H and the unmodified PI (Fig. 5). This leads
us to believe that the cobalt component does not interfere with the formation of the
imide ring. The exact binding of the cobalt to the poly(amic acid) and the polyimide
is unknown. Some reports have suggested a metal can bind to the polymer backbone
through the carboxylic acid of the poly(amic acid) before imidization or even break
the polyimide ring to form the carboxylic acid [11, 17, 23, 24]; however, the FT-IR
spectrum of Co-PI shows full imidization.
3.5. Moisture Sensing and Absorption
Co-PI is a green film in low humidity environments at room temperature; however,
it was found that when the material was submerged in water, it slowly converted to
a yellow film. As the material was removed from water and allowed to dry, it con-
verted back to the original green color. This leads us to believe that the metal is well
retained in the polymer. In addition, when the water solution that the film was sub-
merged in was concentrated, no cobalt salt that could have potentially leeched out
Figure 4. FT-IR spectra of Kapton® H (top) and unmodified PI (bottom).
Figure 5. FT-IR spectra of PI (top) and Co-PI (bottom).
of the material was observed. This is a big advantage over other moisture absorb-
ing/indicating materials that can have the metal indicator wash off while submerged
in water.
Figure 6. FT-IR spectra of Co-PI at low to high (top to bottom) humidity levels.
Upon examining the FT-IR spectra further, an interesting peak in the region
3200–3600 cm−1 was detected in Co-PI (Fig. 6). As the humidity of the environ-
ment was increased, the intensity of the peak increased as well. This broad peak
is attributed to the binding of water in the polymer. No peak was observed in the
unmodified PI at any humidity level. The increase in peak intensity at a specific
wavenumber with respect to an increase in humidity could also be exploited as a
moisture indicator.
As mentioned above, when the material was removed from a water solution,
the original green color returned, therefore the polyimide reversibly binds to water.
Only drying the material in a less humid environment is needed to remove the ab-
sorbed water. This is in contrast to other drying agents, and specifically cobalt(II)
chloride itself, which binds water but requires vacuum and/or elevated temperatures
to release the water again.
4. Conclusions
A polyimide that both absorbs and senses water has been created. The metal is
well retained within the polyimide and the overall polymer is thermally stable. The
exact binding nature of the cobalt in the polymer as well as the binding nature of the
water molecules within the polymer is unknown. However, due to the color change
observed in the cobalt, it is assumed the water molecules bind to the metal.
The material will be characterized further to determine its usefulness in space
systems. For example, the effects of atomic oxygen and radiation, seen in a low
earth orbit environment, on the cobalt-modified polyimides will be investigated.
The aging of the polymer [25] and the long-term effects of moisture exposure will
be studied as well. While it has previously been determined that moisture uptake
of unmodified polyimide is independent of temperature [26], further studies to de-
termine the effect of extreme temperature conditions on moisture uptake of the
cobalt-modified polyimide will be carried out.
Acknowledgement
This work was supported by The Aerospace Corporation’s Independent Research
and Development Program.
References
1. C. R. Moylan, M. E. Best and M. Ree, J. Polym. Sci. B: Polym. Phys. 29, 87–92 (1991).
2. N. C. B. Tan, W. L. Wu, W. E. Wallace and G. T. Davis, J. Polym. Sci. B: Polym. Phys. 36, 155–162
(1998).
3. L.-Y. Jiang, C.-M. Leu and K.-H. Wei, Adv. Mater. 14, 426–429 (2002).
4. T. Asami, Anal. Chem. 40, 648–650 (1968).
5. A. Tsigara, G. Mountrichas, K. Gatsouli, A. Nichelatti, S. Pispas, N. Madamopoulos,
N. A. Vainos, H. L. Du and F. Roubani-Kalantzopoulou, Sensors Actuators B 120, 481–486
(2007).
6. G. Manasis, A. Tsigara, A. Giannoudakos, G. Anyfantis, K. Gatsouli, G. Mousdis, S. Pispas,
N. Madamopoulos and N. Vainos, Glass Technol. 46, 171–174 (2005).
7. J. J. Bergmeister and L. T. Taylor, Chem. Mater. 4, 729–737 (1992).
8. S. H. Chan, W. T. Wong and W. K. Chan, Chem. Mater. 13, 4635–4641 (2001).
9. S. A. Ezzell and L. T. Taylor, Macromolecules 17, 1627–1632 (1984).
10. E. Khor and L. T. Taylor, Macromolecules 15, 379–382 (1983).
11. M. Nandi, J. A. Conklin, L. Salvati and A. Sen, Chem. Mater. 2, 772–776 (1990).
12. W. Y. Ng, X. Gong and W. K. Chan, Chem. Mater. 11, 1165–1170 (1999).
13. G. M. Porta, J. D. Rancourt and L. T. Taylor, Chem. Mater. 1, 269–276 (1989).
14. T. Sawada and S. Ando, Chem. Mater. 10, 3368–3378 (1998).
15. R. E. Southward and D. W. Thompson, Chem. Mater. 16, 1277–1284 (2004).
16. T. Verdianz, H. Simburger and R. Liska, Eur. Polym. J. 42, 638–654 (2006).
17. Z. Wu, D. Wu, S. Qi, T. Zhang and R. Jin, Thin Solid Films 493, 179–184 (2005).
18. S. Yoda, A. Hasegawa, H. Suda, Y. Uchimaru, K. Haraya, T. Tsuji and K. Otake, Chem. Mater.
16, 2363–2368 (2004).
19. W.-X. Yu, L. Hong, B.-H. Chen and T.-M. Ko, J. Mater. Chem. 13, 818–824 (2003).
20. E. Khor and L. T. Taylor, Polym. Mater. Sci. Eng. 49, 198–202 (1983).
21. J. D. Rancourt and L. T. Taylor, Polym. Mater. Sci. Eng. 55, 328–333 (1986).
22. J. D. Rancourt and L. T. Taylor, Macromolecules 20, 790–795 (1987).
23. M. M. Plechaty and R. R. Thomas, J. Electrochem. Soc. 139, 810–821 (1992).
24. K. Akamatsu, S. Ikeda, H. Nawafune and H. Yanagimoto, J. Am. Chem. Soc. 126, 10822–10823
(2004).
25. M. C. Buncick and D. D. Denton, J. Appl. Polym. Sci. 46, 271–280 (1992).
26. W. H. Hubbell, Jr, H. Brandt and Z. A. Munir, J. Polym. Sci. B: Polym. Phys. 13, 493–507 (1975).
Part 4
General Papers
Durability of Si-Containing Polyimide in a Simulated Atomic
Oxygen Environment for Low Earth Orbit Applications
Kumiko Yokota a,∗ , Masahito Tagawa a , Minoru Iwata b , Eiji Miyazaki c ,

Jun-ichiro Ishizawa c , Mineo Suzuki c and Rikio Yokota d
a
Department of Mechanical Engineering, Graduate School of Engineering, Kobe University,
Rokko-dai 1-1, Nada, Kobe 657-8501, Japan
b
Laboratory of Spacecraft Environment Interaction Engineering, Kyushu Institute of Technology,
Sensui-cho 1-1, Tobata, Kita-Kyushu 804-8550, Japan
c
Japan Aerospace Exploration Agency, Sengen 2-1-1, Tsukuba 305-8505, Japan
d
Japan Aerospace Exploration Agency, Yoshino-dai 3-1-1, Sagamihara 229-8510, Japan
Abstract
Real-time measurement of the erosion rate of a commercially available Si-containing polyimide (BSF30#1)
on hyperthermal atomic oxygen (AO) beam exposure, which simulates the AO environment in low Earth
orbit (LEO), was performed. It was found that the erosion rate of BSF30#1 decreased with increasing AO
fluence and it reached as low as 4% of the conventional PMDA-ODA polyimide. X-ray photoelectron spec-
troscopic analysis confirmed that the surface of AO-exposed BSF30#1 was covered by SiO2 layer which
functioned as a protective coating. The self-healing capability of this material is desirable for a space-
applicable polyimide.
Keywords
Atomic oxygen, polyimide, protective coating, silicon
1. Introduction
Atomic oxygen (AO)-induced material degradation in low Earth orbit (LEO) space
environment is a hazardous phenomenon for the materials used for the exterior
surfaces of a spacecraft. In many cases, exterior surfaces of spacecraft are covered
with a multilayer thermal insulation (MLI) [1]. Since the outermost surface of MLI
consists of bare polymeric materials such as Kapton® or FEP Teflon® , AO-induced
erosion of these materials leads to a loss of MLI performance.
*To whom all correspondence should be addressed. Tel./Fax: +81-78-803-6126; e-mail:

yokota@mech.kobe-u.ac.jp
380 K. Yokota et al.
In protecting polymeric materials used from the AO bombardment, optically

transparent inorganic thin films have been considered as overlay protective coatings.
Many types of protective coatings such as indium tin oxide (ITO), PTFE Teflon
film, tin dioxide (SnO2 ), Al2 O3 , or silicon dioxide (SiO2 ) have been proposed [2–
4]. However, there are three major problems in the current hard coating technology.
One is pinholes in the protective coatings which are introduced during the produc-
tion process. Unlike microelectronics applications, anti-AO protective coatings are
required to be pinhole-free over large areas. This is because even a small pinhole
leads to a large undercutting of the polymer [5]. This is a technical challenge for
thin film processes. Another problem are cracks which are introduced during han-
dling. These cracks are larger than pinholes, and thus can become a major source
for damage to hard coatings. The last problem is the collision with space debris in
the orbit. Small pinholes may still be created during the flight due to the collision
with space debris or micrometeoroids. A hard coating with micrometer-thickness
could be easily penetrated by these high-energy particles. Some of these protective
coatings are used for actual space systems such as International Space Station, how-
ever, severe degradation of these coated materials has been reported [6]. From the
considerations listed above, development of materials with self-healing capability
is required. Such new materials are under development, for example, polyhedral
oligomeric silsesquioxane (POSS) is incorporated in the main chain of the poly-
imide [7]. Robustness against simulated AO environment has been reported for this
material. POSS polyimide is a newly synthesized material and has not been used
industrially. If the existing material, which has been widely used industrially, has
similar anti-AO capability, it reduces the cost and time for research and develop-
ment.
In this study, we evaluated the AO-induced erosion properties of a commercially
available Si-containing polyimide which is widely used in the microelectronics
industry in Japan. Real-time mass loss of the material and chemical states after
hyperthermal AO beam exposure were analyzed with a quartz crystal microbalance
(QCM) and by X-ray photoelectron spectroscopy (XPS), respectively.
2. Experimental
Samples used in this study were commercially available PMDA-ODA polyimide

(SP-510, Toray-DuPont, equivalent to Kapton-H) and Si-containing polyimide
(BSF30#1) supplied by the Nippon Steel Chemical Co. Ltd, Kisarazu, Japan. Chem-
ical structures of these materials are shown in Fig. 1. Certain physical and chemical
properties of these materials are also listed in Table 1. Precursors of these materials
were spin-coated onto the Au-electrode of QCM sensor crystal at 15 000 rpm, and
heated at 150◦ C for 1 h followed by heating at 300◦ C for 1 h in N2 . Photographs
of the polyimide-coated QCM sensor crystals are shown in Fig. 2. The thickness of
the polyimide film is approximately 100–200 nm.
Durability of Si-Containing Polyimide in Atomic Oxygen Environment 381
Figure 1. Molecular structures of PMDA-ODA polyimide (SP-510) and Si-containing polyimide

(BSF30#1).
Table 1.
Physical and chemical properties of PMDA-ODA and BSF30#1 polyimide
PMDA-ODA BSF30#1
Tensile strength (MPa) 340 61

Young’s modulus (GPa) 3.4 1.0
Density (g/cm3 ) 1.42 –
Thermal expansion coefficient (×10−6 /◦ C) 27 136
Glass transition temp. (◦ C) – 175
Dielectric constant 3.3 –
Water absorption (%) 2.9 –
Figure 2. Photograph of the polymer-coated QCM sensor crystals. BSF30#1 (left) and SP-510 (right).
The hyperthermal AO beam source used in this study was a laser detonation
source at Kobe University. Details of the system are described in a companion paper
[8]. In this study, two QCMs were installed at 5 cm from the center axis of the sys-
tem and the frequency shifts were recorded simultaneously (Fig. 3). A QCM with
(a)
(b)
Figure 3. (a) Photograph of the laser detonation atomic oxygen beam source and (b) the schematic
drawing of the QCM locations.
Ag electrodes was used to calibrate AO beam fluxes at both positions. This tech-
nique measures the mass of oxygen forming silver oxides at the silver surface. Since
silver oxides are not formed by molecular oxygen, it is suitable to measure atomic
oxygen flux [9]. After the AO flux calibration, SP-510 and BSF30#1 polyimide-
coated sensor crystals were installed in each QCM, and exposed to an AO beam
Figure 4. Time-of-flight spectrum of AO beam used in this experiment.
simultaneously. The temperature of the QCMs was kept at 38◦ C during the expo-
sure. Time-of-flight (TOF) spectrum of AO beam (m/e = 16) used in this study
is shown in Fig. 4. Ion component in the AO beam was subtracted from this TOF
spectrum. The sharp spike at 160 µs is due to the residual after ion subtraction.
A flight time of 340 µs which is the summit of TOF spectrum in Fig. 4 corresponds
to a velocity of 7.0 km/s (4.1 eV) which is close to an orbital velocity of spacecraft
traveling in LEO.

3.1. Flux Calibration with Ag-Coated QCM
Figure 5 shows the frequency shifts of Ag-coated QCMs at two sample positions
during the AO beam exposure. The frequency shift of QCM (AT-cut, 5 MHz) can
be directly converted to the mass change with the equation below:
M = 2.002 × 10−8 F, (1)
where M is the mass change in gram and F the frequency change in Hz. Open
circles show the frequency shift of QCM located at the right position, and filled
circles do that at the left position. The contamination level at the initial Ag surface
is different in every QCM crystal, and an initial mass decrease (positive frequency
shift) is obvious in the QCM at the right position. However, the slopes of the mass
increases are identical for both QCMs. Since the oxidation of silver by AO forms
a non-volatile reaction product (Ag2 O3 ), slope of the mass change decreases with
increasing AO fluence. This is due to the fact that the reaction of Ag with AO is
the rate-limiting process at the beginning of exposure. However, it is followed by
the diffusion-controlled process of Ag ions or AO in the Ag2 O3 layer. This reaction
Figure 5. Frequency shifts of the Ag-coated QCMs. Open circle: right position, and filled circle: left
position.
scheme is similar to that of AO–Si system [10]. The maximum slope of mass in-
crease is −0.17 Hz/s, which is converted to an AO flux of 1.6 × 1014 atoms/cm2 /s.
This AO flux was calculated with the accommodation coefficient of AO at silver
surface to be 1.0 so that the AO flux shown here is the lower bound of the actual
flux. From the experimental results shown in Fig. 5, it is confirmed that the AO
fluxes at the two QCM positions are identical. Thus, the erosion of polymeric mate-
rials measured at the two QCM positions simultaneously can be compared directly.
3.2. Real-Time Mass Loss Measurement of Polyimides
Figure 6(a) shows the frequency shifts of polyimide-coated QCMs under 7.0 km/s
AO beam exposure. The abscissa shows the elapsed time, and the frequency mea-
surement was carried out at every 10 s with a frequency resolution of 0.1 Hz. The
AO beam exposure started at 1800 s. Black line indicates the frequency shift of
SP-510 and the gray line does that of BSF30#1. It is clearly indicated that the fre-
quency of SP-510 increases linearly with time (constant erosion rate). The slope
of the linear frequency shift of SP-510 is 0.13 Hz/s. On the other hand, BSF30#1
shows much lower erosion under the same AO exposure condition. Figure 6(b) is
the close-up of Fig. 6(a). It is clearly seen that the frequency shift of BSF30#1 is
not linear with elapsed time; the mass loss rate gradually decreases with time (or
AO fluence). By differentiating these experimental data with time, the erosion yield
of BSF30#1 is compared to that of SP-510. The mass loss rates of both materials
shown in Fig. 6 were approximated by linear and power functions, respectively.
The best-fit curves were expressed by the following functions: y = 0.13x − 171
for SP-510 and y = x 0.037 − 1203 for BSF30#1. At the elapsed time of 14 000 s,
which corresponds to AO fluence of 2.2 × 1018 atoms/cm2 , it was calculated that the
relative erosion yield of BSF30#1 was less than 4% of that of SP-510. From the ex-
(a)
(b)
Figure 6. (a) Frequency shifts of QCMs coated with SP-510 (black) and BSF30#1 (gray). (b): close-up
view of (a).
perimental data presented here, it is evident that the BSF30#1 has high survivability
in the simulated AO environment.
3.3. XPS Measurements
In order to investigate the protection mechanism of BSF30#1 polyimide against
the hyperthermal AO bombardment, the effect of the AO exposure on the surface
chemistry of BSF30#1 was analyzed by XPS. In this study, an XPS spectrometer
with a non-monochromatic Mg Kα X-ray source was utilized. Figure 7 shows the
XPS survey spectra of BSF30#1 before and after AO exposure. The AO exposure
was performed with 5.0 eV and 2.2 × 1019 atoms/cm2 . Beside C1s and O1s peaks,
small Si2p peak near 100 eV is observed in Fig. 7(a). In contrast, significant in-
crease in O1s intensity and decrease in C1s signals are observed after the AO beam
exposure (Fig. 7(b)). Figure 8 shows the O1s spectra before and after AO beam ex-
(a)
(b)
Figure 7. XPS survey spectra of BSF30#1 before (a) and after AO exposure (b). Energy: 5.0 eV,
fluence: 2.2 × 1019 atoms/cm2 .
posure. A clear peak shift due to AO exposure is observed for the O1s peak. O1s
peak at 532.0 eV is due to O=CN< and O–Si–O structures originally included in
the BSF30#1 polyimide [11]. In contrast, the peak at 532.5 eV is due to the forma-
tion of SiO2 [11]. Si2p spectra before and after AO beam exposures (Fig. 9) also
show peak shift corresponding to a structural change from O–Si–O (101.7 eV) to
SiO2 (103.3 eV) [12]. From the XPS results shown above, it was confirmed that
the surface of BSF30#1 was covered by the SiO2 layer which was formed by the
oxidation reaction with hyperthermal AO.
From the XPS measurements shown in Figs 7–9, it was confirmed that the surface
of AO-exposed BSF30#1 is covered by SiO2 layer. The major reaction path of mass
loss for polyimide with hyperthermal AO beam is considered to be the oxidation
of carbon atoms which creates the volatile reaction products such as CO or CO2 .
(a)
(b)
Figure 8. O1s core level spectra of BSF30#1 before (a) and after AO exposure (b). Energy: 5.0 eV,
The SiO2 layer at the BSF30#1 surface prevents a direct collision of hyperthermal
AO with carbon atoms. Since the SiO2 layer was formed under AO bombardment,
the self-healing capability, in which the cracks and pinholes introduced by debris
impact during the satellite operation can be recovered by further exposure to AO,
can be expected in BSF30#1 polyimide.
4. Conclusions
The erosion rate of Si-containing polyimide (BSF30#1) on exposure to hyperther-
mal AO beam, which simulates the AO environment in LEO, was evaluated. It was
(a)
(b)
Figure 9. Si2p core level spectra of BSF30#1 before (a) and after AO exposure (b). Energy: 5.0 eV,
found that the erosion yield of BSF30#1 decreased with increasing AO fluence and
reached as low as 4% of the conventional PMDA-ODA polyimide. XPS Si2p and
O1s core level spectra both confirmed that the surface of BSF30#1 was covered by
SiO2 layer after AO exposure, which functioned as a protective coating. The self-
healing capability of this material is desirable for a space-applicable polyimide.
Since this polyimide is currently mass-produced, application of this material to real
space systems can be easily realized.
Acknowledgements
The authors express their gratitude to Dr. K. Tokuhisa, M. Yuasa, A. Mori and
Y. Wada of Nippon Steel Chemical Co. Ltd, for supplying the BSF30#1 material.
References
1. D. G. Gilmore, in: Satellite Thermal Control Handbook. The Aerospace Corporation Press, El
Segundo, CA (1994).
2. B. A. Banks, M. J. Mirtich, S. K. Rutledge and D. M. Swec, Thin Solid Films 127, 107 (1985).
3. J. H. Sanders and B. J. Tatarchuk, Thin Solid Films 192, 79 (1990).
4. M. Tagawa and T. K. Minton, in: Proceedings of the 20th International Conference on Space
Technology and Science, Morioka, Japan, Paper No. 96-b-07 (1996).
5. S. K. Rutledge and J. A. Mihelcic, Surf. Coating Technol. 39–40, 607 (1989).
6. B. Banks, A. Shyder, S. K. Miller and R. Demko, NASA/TM-2002-211577 (2002).
7. R. I. Gonzalez, S. J. Tomczak, T. K. Minton, A. L. Brunsvold and G. B. Hoflund, in: Proceed-
ings of the 9th International Symposium on Materials in a Space Environment, Noordwijk, The
Netherlands, ESA-540, 113 (2003).
8. M. Tagawa, K. Yokota, K. Maeda, A. Yoshigoe, Y. Teraoka, K. Akamatsu and H. Nawafune,
Companion paper.
9. V. Matijasevic, E. L. Garwin and R. H. Hammond, Rev. Sci. Instrum. 61, 1747 (1990).
10. M. Tagawa, K. Yokota, N. Ohnae and K. Kinishita, High Perform. Polym. 12, 53 (2000).
11. G. Beamson and D. Briggs, High Resolution XPS of Organic Polymers: Scienta 300 Database.
Wiley, Chichester (1992).
12. J. F. Moulder, W. F. Stickle, P. E. Sobol and K. D. Bomben, Handbook of X-ray Photoelectron
Spectroscopy. Perkin-Elmer Corporation, Eden Prairie, MN (1992).
Thermo-Chemical Behaviour of Organosilicon Films
Deposited by a Cold Plasma Polymerization Process
Brigitte Mutel ∗ and Philippe Supiot

Laboratoire de Génie des Procédés d’Interactions Fluides Réactifs — Matériaux (GéPIFRéM)
UPRES-EA 3571, Cité Scientifique, Université des Sciences et Technologies de Lille,
59655 Villeneuve d’Ascq cedex, France
Abstract
While organic compounds with polymerizable structure are required for conventional way of polymer-
ization, any organic compound can be used for plasma-assisted polymerization process. In this study,
polysiloxane thin films were obtained from a Cold Remote Nitrogen Plasma polymerization of 1,1,3,3-
tetramethyldisiloxane monomer with and without premixing with oxygen. This reactive medium, far from
the discharge, allows to obtain high deposition rate. The films deposited in the present work are characterized
by Fourier Transform Infra-red Spectroscopy, X-ray Photoelectron Spectroscopy, Atomic Force Microscopy,
Scanning Electron Microscopy and thermo-gravimetric analysis. Their thermo-chemical behaviour is also
studied after an in situ plasma assisted post-treatment and after an ex situ thermal post-treatment. A total
mineralization of the coating occurs only after a thermal treatment at 600◦ C and in this case the film struc-
ture looks like silica. After the plasma post-treatment, the uppermost layer tends towards a silica structure
while the core of the coating remains unchanged. The efficiency of the polysiloxane film as a fire retardant
coating for Polyamide-6 nano-composite substrates is evidenced. Fire retardant properties are studied from
Limiting Oxygen Index (LOI) tests and cone calorimetry measurements.
Keywords
Cold remote plasma, plasma polymerization, polysiloxane, fire retardancy, tetramethyldisiloxane, poly-
amide-6, thermal behaviour
1. Introduction
While organic compounds with polymerizable structure are required for conven-
tional way of polymerization, any organic or inorganic compound available in
vapour form can be used in the plasma-assisted plasma polymerization process. In
fact, in this latter case, polymerization occurs because electrons or reactive species
of the plasma have energy enough to create free radicals whose recombination in-
duces the growth of a polymer film which can be deposited on various substrates.
* To whom correspondence should be addressed. Tel.: +33 3 20 43 65 40; Fax: +33 3 20 43 41 58; e-mail:
brigitte.mutel@univ-lille1.fr
392 B. Mutel and P. Supiot
Generally, plasma-assisted polymerization processes have a low deposition rate

as they involve a discharge plasma. These plasmas are characterized by high con-
centrations of charged particles and there is a competition between the deposition
step and etching induced by electron and ion bombardment phenomena.
Our laboratory has developed another approach for plasma-assisted polymeriza-
tion process involving a Cold Remote Nitrogen Plasma (denoted by CRNP). This
reactive medium, located in the afterglow, far from the discharge, is free of charged
particles. Hence, etching effect is reduced or eliminated and it is possible to obtain
deposition rate reaching 100 times that obtained with processes involving discharge
plasmas [1–4].
Depending on the chosen monomer, the plasma deposited films can be used for a
great number of applications. In this study, a CRNP process was employed to induce
the polymerization of the 1,1,3,3-tetramethyldisiloxane (TMDSO). The deposited
film shows an interesting thermo-chemical behaviour which is shown to improve
fire retardant properties of polyamide-6 (PA-6) substrate.
The deposited films are characterized by Fourier Transform Infra-red Spec-
troscopy (FT-IR), X-ray Photoelectron Spectroscopy (XPS), Atomic Force Mi-
croscopy (AFM) and Scanning Electron Microscopy (SEM). Their thermal stability
is studied by thermo-gravimetric analysis (TGA) and their fire retardant properties
are studied from Limiting Oxygen Index (LOI) tests and cone calorimetry measure-
ments.
2. Experimental
2.1. CRNP and Plasma-Assisted Deposition
The experimental set-up of the CRNP-assisted deposition reactor is shown in Fig. 1.

The nitrogen flow (purity 99.995%) was excited by means of a microwave discharge
(2450 MHz) produced in a fused silica tube. By a continuous pumping (primary
pump — Alcatel), the reactive species flow from the discharge to the deposition
chamber where the CRNP appears like a yellow afterglow due to the transition
between two electronically excited states of nitrogen molecules (N2 (B3 g )v →
N2 (A3 + u )v ). Indeed, this phenomenon originates from recombination of nitrogen
atoms. The main reactive species in the CRNP are atomic nitrogen in the ground
electronic state N(4 S) and vibrationally excited nitrogen molecules in the ground
state (N2 (X1 + g )v ). The long lifetime of atomic nitrogen which allows to obtain
a CRNP far from the discharge and with volume reaching several m3 could be
explained by a very low surface induced recombination [5] and under some condi-
tions by a re-dissociation mechanism [6]. The CRNP is free of charged particles.
It is characterized by a strong deviation from thermodynamic equilibrium: the gas
temperature is around the ambient while vibrational temperature of N2 (X1 + g )v is
around 3000 K [7, 8].
Polysiloxane Film Deposited by Plasma Process 393
Figure 1. Experimental set-up of the CRNP-assisted deposition reactor.
The TMDSO monomer (Sigma Aldrich — purity: 97%) which can be premixed
with O2 (purity 99.5%) was injected in the CRNP through a coaxial injector.
Gaseous flows were regulated by means of MKS mass flow controllers.
CRNP-assisted polymerization process always includes two steps. First, the sub-
strate to be coated is pre-treated by the CRNP for 15 min in order to clean it, to
desorb contaminants present or, in case of polymer, to increase its adhesion quality.
Second, the deposition step is performed, without air exposure, by the addition of
the monomer.
2.2. Characterization Techniques
Deposited films were analysed by FT-IR using a Perkin-Elmer instrument (Spec-
trum One) and by XPS using a VG ESCALAB 220XL. Their morphologies were
studied by Scanning Electron Spectroscopy (LEO 982 ZEISS Microscope-electron
beam operating at 2 kV) and by Atomic Force Microscopy using Nanoscope III —
(scan size: 1 µm × 1 µm).
Thermo-gravimetric analyses (TGA) were performed using a Setaran appara-
tus from 20 to 800◦ C at a heating rate of 7.5◦ C/min under air flow (Air Liquide
grade, 5.10−7 m3 /s measured under standard conditions). They were carried out on
coatings in the powder form (∼10 mg) obtained by scraping the films deposited
on aluminium plates. The powder was placed in an open silica pan. Precision of
temperature measurements was ±1.5◦ C. Thickness of the deposited film was mea-
sured by surface profilometry using an Alphastep piezoelectric stylus (accuracy
±0.01 µm).
Limiting Oxygen Index (LOI) tests were performed using a Stranton Redcroft
instrument (Fire Testing Technology Ltd, UK) according to the ASTM D 2863/77
standard [9]. This test allows to determine the minimal oxygen percentage in an
oxygen–nitrogen mixture which is required to sustain the combustion of the sample
(standard size: 100 mm × 10 mm × 3 mm) placed vertically. The larger the LOI
value is, the better is the fire retardant property of the material.
Cone calorimetry measurements were carried out with a Stranton Redcroft
calorimeter (Fire Testing Technology Ltd, UK) according to the ASTM E 1354-90a
standard [10]. Samples with standard size (100 mm × 100 mm × 3 mm) were ex-
posed to a 35 kW/m2 external heat flux which represents the heat flux found in
the vicinity of solid-fuel ignition source. Conventional data such as Rate of Heat
Release (RHR), ignition time, Total Heat Evolved (THE), Volume of Smoke Pro-
duction (VSP), CO and CO2 flow rates as combustion products and residual weights
were obtained.
3.1. Mechanism Occurring in the System CRNP/TMDSO-O2
Spectroscopic studies of both gas phase and deposited film have allowed to propose
the mechanism occurring in the system CRNP/TMDSO-O2 [11, 12].
The initiation step consists in hydrogen capture by N(4 S) according to:
(a)
By sequential reactions, N(4 S) induces mainly these three radicals:
(b) (c) (d)

and this mechanism occurs both with or without addition of O2 to TDMSO. The
two smaller fragments (c and d) may originate from N(4 S) + N(4 S) + M reaction
where M is presently the siloxane radicals a or b.
Without O2 admixture, the TDMSO shows an efficient methyl abstraction
through reactions with N(4 S) and with excited nitrogen molecules resulting from
N(4 S) recombination reaction. The following small silicated multiradical fragments
are formed:
These fragments recombine onto the substrate and lead to a highly cross-linked
silica-like structure but at a very low deposition rate.
The addition of O2 to the TMDSO limits the methyl abstraction step and leads
to an increase of (≡Si–O• )solid or gaseous radical density and to an increase of the
average length of the growing polymeric fragment according to
So, using TMDSO premixed with oxygen, it is possible to obtain films showing a
polysiloxane-like structure with high deposition rates
3.2. Characterization of the Deposited Films

3.2.1. Influence of Oxygen Flow Rate Added to TMDSO
The film depositions were performed using nitrogen and TMDSO flow rates of
1800 and 5 sccm, respectively. The oxygen flow rate ((O2 )) was set to values
ranging from 10 to 200 sccm. The microwave power transmitted to the plasma was
200 W. The total pressure in the reactor ranged from 540 to 600 Pa depending on
the oxygen flow rate.
Figure 2 shows FT-IR spectra of a 1 µm thick film deposited on Si, for oxygen
flow ranging from 10 to 150 sccm. In this O2 flow range, the chemical nature of the
films is rather similar.
The band located at 2145 cm−1 and attributed to Si–H stretching band [13] in the
TMDSO spectrum (Fig. 3) completely disappears in the film spectra with increasing
(O2 ). This spectrum feature confirms hydrogen abstraction by N(4 S) atoms as a
Figure 2. FT-IR spectra of the film deposited on Si (thickness = 1 µm) for (O2 ) = (a) 10, (b) 25,
(c) 50 and (d) 150 sccm.
preferential initiation step. The bands located at 2910 and 2960 cm−1 correspond
to the symmetric and asymmetric stretching modes of methyl group [14].
In the range 600–1500 cm−1 the band located at 1260 cm−1 is assigned to the
CH3 symmetric bending mode and is characteristic of Si(CH3 )2 in a polysiloxane
structure [13, 15]. Correlatively, the band observed at 800 cm−1 is characteristic of
Si–C bending motion in the same structure [13, 15].
Figure 3. FT-IR spectrum of TMDSO.
Figure 4. Evolution of deposition rate (rd ) and density (ρ) of the deposited films versus oxygen flow
(O2 ) added to TMDSO.
The polymeric structure of the film is evidenced by the broad band located be-
tween 1000 and 1150 cm−1 which is assigned to the Si–O–Si stretching mode in
this group [13, 16]. The position of this band indicates a very weak cross-linking.
High cross-linking would induce a shift of this band towards 1200 cm−1 . When
oxygen flow increases, a large band between 3200–3400 cm−1 appears, due to the
stretching band in Si–OH [13, 17].
Figure 4 shows the evolution of the deposition rate versus (O2 ). Coatings were
deposited on an aluminium substrate (diameter 54 mm). The maximum deposition
(a) (b)
(c) (d)
Figure 5. AFM images (1 µm × 1 µm) of the film ((O2 ) = 50 sccm). Evolution of the morphol-
ogy versus the thickness. (a) Thickness = 0.04 µm, roughness = 0.3 nm; (b) thickness = 0.2 µm,
roughness = 0.5 nm; (c) thickness = 0.4 µm, roughness = 0.8 nm; (d) thickness = 1.3 µm,
roughness = 1.8 nm.
rate equal to 43.5 nm/s is obtained for (O2 ) = 50 sccm. This oxygen flow value
was chosen in the following study.
3.2.2. Morphology of the Films

Figure 5 shows the results obtained by AFM. The film roughness increases with
the thickness. For a 1.3 µm thick film, it is equal to 1.8 nm. Thicker films ex-
amined by SEM (Fig. 6) give evidence for a granular structure showing a linear
arrangement. For a 10 µm thick film, the coalescence phenomenon appears. These
thick films show a very low surface energy (contact angle with water equal to
160◦ ) which can be explained by preferential orientation of apolar methyl groups
(a) (b)
(c)
Figure 6. SEM images of the film ((O2 ) = 50 sccm). (a) Thickness = 5.2 µm;
(b) thickness = 7.8 µm; (c) thickness = 13.0 µm.
towards the surface and probably also by a large surface roughness (not yet mea-
sured).
3.2.3. Thermo-chemical Stability
The thermo-gravimetric trace and its derivative (Fig. 7) show that the degradation
process occurs in two steps:
• The first step occurs between 300 and 390◦ C with a very low degradation rate.
At the end of this first step, the residual weight is around 89%. FT-IR spectrum
of the film at 300◦ C is the same as the one shown at the ambient (spectra a and
b in Fig. 8). Only a very weak band at 470 cm−1 attributed to ρ (Si–O–Si) can
be noticed [16].
• The second degradation step, which is the main one, occurs between 390 and
530◦ C with a maximum at 415◦ C. At 400◦ C, a modification of the FT-IR
spectrum (spectrum c in Fig. 8) shows methyl removal leading to some more
inorganic structure. Several new bands characteristic of Si–O–Si appear and are
summarized in Table 1.
Figure 7. Thermo-gravimetric trace and its derivative for coating (in powder form) deposited with
(O2 ) = 50 sccm.
Even at 800◦ C, the degradation of the film is not complete and the residual
weight is around 75%. The methyl contribution at 2950 cm−1 has completely dis-
appeared, due to a complete carbon removal of the coating which has a silica-like
structure. The FT-IR spectrum only shows Si–O–Si and Si–OH bands [18].
3.2.4. Influence of Thermal Treatment on Deposited Films
In order to study the thermal behavior of this material as a coating, a 5 µm thick
film was deposited on Al SI316L stainless steel substrate (40 mm × 20 mm × 2mm)
and submitted to thermal treatment performed in a furnace operating at atmospheric
pressure. The temperature ramp was controlled by a thermal regulator (Nabertherm
GmbH, Germany). Before their introduction in the CRNP reactor the substrate sam-
ples were degreased in acetone and dried at 110◦ C for 12 h. The deposition step
occurred after a pre-treatment by the CRNP for 15 min.
FT-IR analysis of the deposited film shows that it is similar to the one deposited
on Si.
The first thermal treatment was applied up to 450◦ C (heating rate = 5◦ C/min) for
durations denoted by t and ranged from 1 to 5 h. Intensity of bands related to organic
contribution (CH3 ) decreases as soon as t = 1 h but still remains even after t = 5 h.
The relative methyl content of the film during the treatment was evaluated from
the ratio of peak intensities at 2950 cm−1 and 1000 cm−1 , corresponding to organic
(CH3 ) and mineral (Si–O–Si) contributions, respectively. Irrespective of t, this ratio
Figure 8. Evolution of FT-IR spectra of the film versus sample temperature. (a) 20◦ C; (b) 300◦ C;
(c) 400◦ C; (d) 800◦ C.
Table 1.
FT-IR bands of the film at different temperatures. Heating under air (7.5◦ C/min). (O2 ) = 50 sccm
Wavenumber Ambient Heated at Vibrational mode Ref.(s)

(cm−1 ) 300◦ C 400◦ C 800◦ C
3800–3000 X X X X ν (O–H) in Si–OH or H2 O [11, 15]

2960 X X X ν (CH3 )as [12]
2910 X X ν (CH3 )s [12]
1410 X X δ (CH3 )as in Si–(CH3 )x [11, 13]
1260 X X X δ (CH3 )s in Si–(CH3 )2 [11, 13]
1200–1000 X X X X ν (Si–O–Si)as [11, 14]
900 X X ρ (CH3 ) ν (Si–C) in Si–(CH3 )2 [11, 13]
805 X X δ (Si–O–Si) [16]
760 Weak Weak ρ (CH3 ) ν (Si–C) in Si–(CH3 )3 [11]
600 X ν (Si–O–Si)s [11]
470 X X δ (Si–O–Si) [14]
Table 2.
Surface composition of films calculated from XPS analyses. Influence of the thermal post-treatment
and of the plasma post-treatment
Sample Surface composition
“fresh” deposited film Si C1.08 O1.15 N0.04

After thermal post-treatment 450◦ C–5 h Si C0.12 O1.67
After plasma post-treatment 5 min Si C0.42 O1.78 N0.04
remains constant, giving evidence that the combustion of the organic part is limited.
It can be due to a strong decrease of oxygen diffusion in the core of the material.
After the second thermal treatment at 650◦ C for 1 h, the methyl band at
2950 cm−1 is no longer observed. The same result is obtained at 800◦ C. It seems
that beyond 650◦ C a full mineralization (disappearance of CH3 bands) of the coat-
ing is reached. The FT-IR spectrum is close to that of silica. This result gives
evidence that it is possible, using a selected thermal post-treatment, to obtain a coat-
ing with a controlled mineralization rate. XPS analyses [19] allow to calculate the
atom ratio normalized with respect to Si of the surface material. Results are shown
in Table 2 and confirm the FT-IR results. Up to 450◦ C, the surface contains an or-
ganic contribution, but from 650◦ C it is mainly mineralised and the composition is
near that of silica.
3.2.5. Influence of Plasma Post-treatment
Because of the presence of methyl group, the deposited film is very hydrophobic
and it can be a disadvantage for further applications. So, the influence of plasma
post-treatment on the film deposited on steel was studied. After the deposition step,
the TMDSO flow was switched off and the film was exposed to the CRNP+O2 mix-
ture for durations denoted by τ and ranged from 5 to 35 min. XPS results (Table 2)
show that the surface composition was strongly modified as soon as τ = 5 min:
the carbon to silicon content ratio decreases from 1.08 to 0.42 while the oxygen
to silicon content ratio increases from 1.15 to 1.78. For longer τ values, C content
smoothly decreases and becomes stable at 0.3 for τ = 35 min. FT-IR analysis of the
plasma post-treated films shows that irrespective of τ value, the spectrum remains
unchanged with respect to that of the as-deposited coating.
This result gives evidence that after the plasma post-treatment, only the upper-
most layer is modified. The core of the coating is still polysiloxane-like.
3.2.6. Efficiency of Polysiloxane-Like Film as a Fire Retardant Coating [19–22]
Taking into account the interesting thermal behaviour shown in the previous section,
deposited polysiloxane films were tested as a fire retardant coating.
Polymers are often very flammable and require modification before use in many
applications. Flammability can be reduced in several ways. The most classical one
is the incorporation of flame retardant additives to the polymer [23, 24]. This tech-
nique is quite simple and cheap, and such additives are now commercially available.
However, they can reduce the thermal and mechanical properties of the polymer
as the loadings are rather high (50–70 wt%) [25–27]. Moreover, halogenated and
phosphorous additives, commonly used [28–30], can lead to environmental prob-
lems [31, 32]. The chemical modification of the macromolecules is another way
[33–37]. As loadings are much lower (<10 wt%), thermal and mechanical prop-
erties of the polymer are preserved, but this technique, not easy to conduct on the
industrial scale, is rather expensive and not much used.
Literature data on the methods involving fire retardant coatings are very scarce
[38–41]. But this approch seems very attractive as it allows to impact fireproof prop-
erties at the polymer surface only where the flammability occurs without altering
the polymer bulk properties.
As polysiloxane-based films are known to have good thermal stability [42, 43],
fire retardancy efficiency of our polysiloxane film was tested on polyamide-6. Fig-
ure 9 shows that irrespective of the film thickness, the LOI value of polyamide-6
sample is not changed by the coating. Results from cone calorimetry measurements
are shown in Table 3. Results obtained for the PA-6 agree with literature data. The
coating does not allow to reduce significantly either the ignition time, or RHR and
CO2 peaks. The apparent decreases of THE, VSP peak and CO2 peak do not corre-
spond to an improvement of the system. As can be seen in Fig. 10, the combustion
Figure 9. Evolution of LOI value versus the film thickness.
Table 3.
Results of LOI and cone calorimetry measurements for PA-6, coated PA-6, PA-6 nano and coated
PA-6 nano1
PA-6 Coated-PA-6 PA-6 nano Coated-PA-6 nano
LOI 21 — — ↑130%
Ignition time 66 s — ↑ 49% ↑ 46%
RHR peak 1053 kW/m2 — ↓ 34% ↓ 41%
VSP peak 4.4.10−4 m3 /s ↓ 43% ↓ 31% ↓ 37%
CO peak 253 ppm ↓ 50% ↓ 63% ↓ 68%
CO2 peak 2.46 vol% — ↓ 39% ↓ 52%
THE 1345 kJ ↓ 38% ↓ 29% ↓ 33%
1 LOI — Limiting Oxygen Index, RHR — Rate of Heat Release, VSP — Volume of Smoke Pro-
duction, THE — Total Heat Evolved.
is not delayed rather it is even accelerated from 100 s onward. In addition, a de-
lamination of the coating is observed under heating. The presence of new chemical
functions grafted on the PA-6 surface after the pre-treatment could both promote a
faster degradation and lead to the formation of non-flammable products, explaining
the decrease of the THE as well as of VSP and CO2 peaks.
Literature data [43–46] report an improvement of PA-6 fire retardancy properties
by the addition of clay nano-composite even at low loadings (2–5 wt%). A compar-
ative study of results obtained with so-called PA-6 nano and coated PA-6 nano was
performed. PA-6 nano was obtained by incorporating 2 wt% clay at the nanoscale
into the polymer. The LOI value of PA-6 nano is not improved in comparison to
the PA-6 (Fig. 9). PA-6-nano combustion was delayed up to 50 s in comparison
to PA-6, slightly slowed down and led to a residual weight equal to 4% instead of
1% for PA-6 (Fig. 10). Table 3 shows that the clay incorporation into PA-6 leads
Figure 10. Evolution of residual weight. (Film thickness = 1.5 µm.)
to a decrease of conventional indexes obtained by cone calorimetry measurements.

Because of clay incorporation, a protective layer is formed at the polymer surface
during the combustion, reducing mass and heat transfer between the flame and the
polymer.
Once coated, the PA-6 nano showed better performances with regard to LOI and
cone calorimetry tests. Indeed, LOI value is strongly increased (Fig. 9): it sharply
increases as soon as the film is 0.6 µm thick, reaches a maximum value (48%) for
1.5 µm thickness, then decreases and becomes stable around 45%. No delamina-
tion occurs during heating due to a very good adhesion between the film and the
PA-6 nano, as also evidenced by the cross-cut-tape test. The combustion is delayed
and slowed down (Fig. 10) in comparison to PA-6 nano. Cone calorimetry results
(Table 3) show that all peaks are decreased, as well as the THE value. The barrier
effect caused by the protecting layer which is formed during the combustion due to
the clay addition is reinforced by the presence of the polysiloxane coating.
4. Conclusion
The CRNP-assisted polymerization of TMDSO premixed with O2 allows to deposit
a polysiloxane film with a deposition rate as high as 2.5 µm/min.
Thermo-gravimetric analysis shows that a depolymerization of the coating oc-
curs by heating but its degradation is not complete as even at 800◦ C the residual
weight is around 75%. The residue is mainly inorganic and shows a silica-like
structure. Thermal treatment of this film deposited on stainless steel shows that
the mineralization rate of the film is strongly dependent on the temperature. At
450◦ C, an organic contribution remains while the film is mainly free of carbon at
800◦ C. This behaviour can be very interesting for processes occurring at high tem-
peratures. Steel is protected by a silica layer which can act as an efficient diffusion
barrier against oxidation. After a post-treatment by the CRNP mixed with oxygen,
it is possible to obtain a coating with a core polysiloxane structure while the up-
permost surface, which is silica-like, acts as a diffusion barrier against oxidation at
room temperature.
These coatings allow to improve fire retardant properties of PA-6 nano-
composite substrates. In comparison to the classical method involving commonly
used additives, the method described in this study led to a significant and simul-
taneous improvement of performances evaluated by LOI, RHR and THE indexes
determination. A synergistic effect due to simultaneous action of the clay nano-
composite and of the thin film is shown. During the combustion, the formation of
a carbonaceous and silica-like layer acts as a barrier thus limiting mass and heat
transfer between the flame and the polymer and, consequently, slows down emis-
sion of toxic gases and smoke production.
References
1. J. D. Quensierre, C. Jama, E. Radeva, O. Dessaux and P. Goudmand, in: Proc. of 11th Intl. Sym-
posium on Plasma Chemistry, Loughborough, UK, p. 1047 (1993).
2. F. Callebert, P. Supiot, P. Goudmand and O. Dessaux, in: Proc. of 11th Intl. Symposium on Plasma
Chemistry, Loughborough, UK, p. 1493 (1993).
3. F. Callebert, B. Revel, O. Dessaux and P. Goudmand, High Temp. Chem. Process. 4, 59 (1995).
4. F. Callebert, R. Stephan, O. Dessaux and P. Goudmand, in: Proc. of 12th Intl Symposium on
Plasma Chemistry, Minneapolis, MN, p. 1105 (1995).
5. A. Talsky and V. Zvonieck, in: Proc. of 13th European Sectional Conf. on the Atomic and Molec-
ular Physics of Ionised Gases, Poprad, Slovakia, p. 325 (1996).
6. O. Dessaux, P. Goudmand and B. Mutel, J. Quant. Spectros. Radiat. Transfer 30, 137 (1983).
7. B. Mutel, M. Bridoux, M. Crunelle-Cras, O. Dessaux, F. Grase and P. Goudmand, Chem. Phys.
Lett. 104, 290–295 (1984).
8. D. Blois, P. Supiot, M. Barj, A. Chapput, C. Foissac, O. Dessaux and P. Goudmand, J. Phys. D:
Appl. Phys. 31, 2521 (1998).
9. ASTM D 2863/77. Standard test method for measuring the minimum oxygen concentration to
support candle like combustion of plastics (Oxygen Index).
10. ASTM E 1354-90a. Standard test method for heat and visible smoke release for materials and
products using an oxygen depletion calorimeter.
11. F. Callebert, P. Supiot, K. Asfardjani, O. Dessaux, P. Goudmand, P. Dhamelincourt and J. Lau-
reyns, J. Appl. Polym. Sci. 52, 1595 (1994).
12. P. Supiot, F. Callebert, O. Dessaux and P. Goudmand, Plasma Chem. Plasma Process. 13, 539
(1993).
13. D. R. Anderson, in: Infrared, Raman and Ultraviolet Spectroscopy: Analysis of Silicones,
A. L. Smith (Ed.), Chap. 10, p. 247. Wiley, London (1974).
14. L. L. Tedder, G. Lu and J. E. Crowel, J. Appl. Phys. 69, 7073 (1991).
15. C. Rau and W. Kulisch, Thin Solid Films 249, 28 (1997).
16. P. G. Pai, S. S. Chao, Y. Takagi and G. Lucovsky, J. Vac. Sci. Technol. A4, 689 (1986).
17. J. A. Theil, J. G. Brace and R. W. Knoll, J. Vac. Sci. Technol. A12, 1365 (1994).
18. T. M. Parrill, J. Mater. Res. 7, 2230 (1992).
19. L. Guillou, P. Supiot and V. Le Courtois, Surf. Coat. Technol. 202, 233 (2008, avaible on
www.sciencedirect.com).
20. A. Quédé, C. Jama, P. Supiot, M. Le Bras, R. Delobel, O. Dessaux and P. Goudmand, Surf. Coat.
Technol. 151–152, 424 (2002).
21. A. Quédé, B. Mutel, P. Supiot, C. Jama, O. Dessaux and R. Delobel, Surf. Coat. Technol. 180–181,
265 (2004).
22. A. Quédé, B. Mutel, P. Supiot, C. Jama, O. Dessaux, M. Le Bras and R. Delobel, in: Fire Retar-
dancy and Polymers, M. Lebras (Ed.), Chap. 20, p. 276. Royal Society of Chemistry, Cambridge,
UK (2005).
23. P. R. Hornsby and C. L. Watson, Plastic Rubber Process. Appl. 6, 169 (1986).
24. P. R. Hornsby, J. Wang, R. Rorthon, G. Jackson, G. Wilkinson and K. Cossick, Polym. Degrad.
Stab. 51, 235 (1996).
25. M. D. Schubert and R. Leeuwendal, in: Proc. of the Flame Retardant 98 Conf., Elsevier Applied
Science, London, p. 163 (1998).
26. D. Sallet, V. Mailhos-Lefievre and B. Martel, Polym. Degrad. Stab. 30, 29 (1990).
27. S. V. Levchik and E. D. Weil, Polym. Int. 49, 1033 (2000).
28. www.specialchem.com:site internet officiel de specialchem.
29. E. D. Weil, in: Flame Retardancy of Polymer Material, W. C. Kuryla and A. J. Papa (Eds), Vol. 13,
Chap. 3. Marcel Dekker, New York, NY (1978).
30. Y. B. Yaakov and R. Minke, US Patent 4,605,708 (1986).
31. L. W. D. Weber and H. Greim, J. Toxicology Environmental Health 50, 195 (1997).
32. J. H. Mennear and C. C. Lee, Environmental Health Perspectives 102 (Suppl. 1), 265 (1994).
33. P. Potin and R. De Jaeger, Eur. Polym. J. 27, 341 (1991).
34. R. De Jaeger and M. Gleria, Prog. Polym. Sci. 23, 179 (1998).
35. H. R. Allcock, Adv. Mater. 6, 106 (1994).
36. H. Horacek and W. Grabner, Makromol. Chem., Macromol. Symp. 74, 271 (1993).
37. E. Weil and B. McSwigan, J. Coating Technol. 66 (839), 75 (1994).
38. T. Jana, B. C. Roy and S. Maiti, Polym. Degrad. Stab. 69, 79 (2000).
39. G. S. Learmonth and D. G. Thwaite, Polym. J. 1, 154 (1969).
40. I. Ravadis, A. Toth, G. Marosi, A. Marton and A. Szep, Polym. Degrad. Stab. 74, 419 (2001).
41. L. Shi, Eur. Polym. J. 36, 2611 (2000).
42. D. Das, J. Farjas, P. Roura and G. Viera, Diamond and Related Materials 10, 1295 (2001).
43. G. Camino, S. M. Lomakin and M. Lazzari, Polymer 42, 2395 (2001).
44. J. W. Gilman, Appl. Clay Sci. 15, 31 (1999).
45. J. W. Gilman, C. L. Jackson, A. B. Morgan, R. Harris, E. Manias, E. P. Giannelis, M. Wuthenow,
D. Hilton and S. H. Phillips, Chem. Mater. 12, 1866 (2000).
46. F. Dabrowski, S. Bourbigot, R. Delobel and M. Le Bras, Eur. Polym. J. 36, 273 (2000).
47. J. W. Gilman, T. Kashiwagi, E. P. Gnelis, E. Manias, S. Lomakin, J. D. Lichtenhan and P. Jones,
in: Fire Retardancy of Polymers: The Use of Intumescence, M. LeBras, G. Camino, S. Bourbigot
and R. Delobel (Eds). Royal Chem. Soc., Cambridge, UK (1998).
Aspects of Lifetime Assessment of High-Temperature Fibers
and Lifetime Enhancement
T. Bahners ∗ , U. Schloßer, T. Textor and E. Schollmeyer

Deutsches Textilforschungszentrum Nord-West e.V., Adlerstr. 1, 47798 Krefeld, Germany
Abstract
The aging behavior of textiles made of polyimide (PI) and other high-performance polymers was studied by
accelerated aging simulation. The samples were exposed to basic or acidic vapors and the tensile strength
of the samples measured as a function of exposure time, i.e., duration of storage. The data show that m-
aramide is quite resistant to the basic vapor and shows only a slight decrease in tensile strength after 300 h
in acidic vapors. However, a significant increase in degradation due to acidic vapor is found at higher tem-
peratures under ‘closed-system’ conditions. PI, on the other hand, is highly sensitive to basic conditions
and is significantly damaged in these atmospheres already after 2 h at 60◦ C. The second part of this study
was to investigate the potential of inorganic–organic hybrid polymers deposited as thin layers to increase
the resistance of exemplary technical textiles against aging-promoting influences. GPTMS-based sols were
prepared and applied to the fabric samples. The data showed that the sol–gel coating did not improve the
aging behavior of m-aramide as well as glass fiber fabrics in acidic vapor, whereas the inorganic–organic
hybrid polymer formed a highly efficient barrier against the alkaline attack from basic vapor in the case of
the glass fiber samples.
Keywords
High-temperature fibers, technical textiles, aging, lifetime assessment
1. Introduction
In contrast to common clothing and home textiles, technical textiles as applied in
fields as diverse as automotive industry, modern architecture and building construc-
tion, filtration, or general engineering, where textiles are used as machine elements,
are subject to influences that have detrimental effects on usable lifetime. Depend-
ing on the field of application, not only the durability of the textile as a whole,
but also the main performance features of the textile are critical factors with re-
gard to function and safety. Accordingly, high-performance polymers such as para-
and meta-aramides, polyimide (PI), poly(etheretherketone) (PEEK), and also fibers
* To whom correspondence should be addressed. Tel.: +49 (0)2151 843-156; Fax: +49 (0)2151 843-143;
e-mail: bahners@dtnw.de
410 T. Bahners et al.
Figure 1. Scanning electron micrograph of damaged polyimide (PI) fibers after 2 h exposure to am-
monia vapor.
made of carbon and glass are often employed to meet the conditions of a specific
application.
The aging of a fiber material may be caused by factors as diverse as mechanical
stress, e.g., abrasion, high tensile forces, climate, aggressive chemical environments
as exemplified in Fig. 1, or high temperatures, to name a few [1–3]. Given this
background, the reliable prediction of the lifetime of these products, mainly based
on appropriate aging simulation, is of increasing importance to manufacturers and
users alike. Aging simulation is designed (a) along the conditions of application
and (b) to accelerate aging (days/weeks instead of years!). The simulation program
might include:
– exposure to a fast changing climate, i.e., temperature and relative humidity as
a function of time,
– exposure to aggressive gases or vapors (with/without variation of temperature
and humidity),
– storage in aggressive liquids, and
– exposure to UV radiation.
Depending on application, additional factors might be taken into account, and,
sometimes, mechanical stress is also applied simultaneously. An intrinsic problem
of accelerated aging simulation is, however, that there are NO RULES for the design
of the simulation program. Also, the correlation between simulated and real-time
aging is not easily assessed.
Several efforts have been made in the past years to improve the aging behavior of
high-performance fibers with the required properties by surface modification tech-
niques. An example was given by a study of the alkaline hydrolysis of poly(ethylene
terephthalate) (PET) fibers (textile fabrics) by Opwis et al. [4]. A significant in-
Aspects of Lifetime Assessment of High-Temperature Fibers and Lifetime Enhancement 411
Figure 2. Basic concept for the preparation of an inorganic–organic hybrid polymer from modified
alkoxysilanes.
crease in the alkali-resistance was found after UV irradiation in presence of reactive

media such as 1,7-octadiene or diallylphthalate (DAP), which produced highly
cross-linked thin layers on the fiber surfaces.
Another promising approach is the use of the sol–gel technique to deposit
inorganic–organic hybrid polymers based on organically modified alkoxysilanes on
the fiber surfaces. Besides the 3-glycidoxypropyltrimethoxysilane (GPTMS) inves-
tigated in this study, further typical examples are 3-chloropropyltrimethoxysilane
and methacryloxypropyltrimethoxysilane (MPTS). In the presence of certain
amounts of water, under basic or acidic conditions, the alkoxy groups undergo
hydrolysis and condensation to form a sol, which is then applied to the textile sub-
strate. In a following curing step, the condensation of the hydrolyzed alkoxysilanes
and enhanced cross-linking of the functional groups take place [5] (Fig. 2). The sol
can be applied by simple dipping or padding processes yielding very thin layers
with only a slight increase in textile weight.
These hybrid polymers combine qualities of ceramics and synthetic polymers
and offer an immense potential for creative modifications of surface properties with
a comparatively low technical effort at moderate temperatures. It is important to
note that functionalities can be controlled by the choice and/or modification of the
inorganic and/or organic part of the precursors [6–10]. The silane chemistry allows
also stabilization of nano-sized metal oxides, introducing an additional ‘degree of
freedom’ with respect to functionality. Here, the particles can be integrated in the
three-dimensional network of the hybrid polymers derived by the sol–gel process,
where they can be employed to achieve certain optical effects or to enhance me-
chanical strength.
In view of the technical application of textile products used in filtration and other
industrial installations — including outdoor application in polluted industrial areas
— the objective of the first part of this work was to investigate the behavior of tex-
tiles made of polyimide (PI) and other high-performance polymers in accelerated
aging simulations of various designs. Experimental measurements in aggressive
atmospheres were aimed at a general characterization of the materials as well as
an illustration of the relevance of parameters of simulation design. The scope of
the second part of this study was to investigate the potential of inorganic–organic
hybrid polymers deposited as thin layers to increase the resistance of exemplary

technical textiles against aging-promoting influences. The textiles under investiga-
tion were made of high-performance polymers such as polyimide (PI), m-aramide,
and p-aramide. In addition, a glass fiber fabric was included in the experimental
study because of its importance for various industrial applications, e.g., as a rein-
forcement fabric in construction. In closing, remarks on lifetime prediction methods
are made.
2. Experimental
2.1. Textile Samples
The textile samples under investigation were commercial products made of high-
performance materials, namely a polyimide (PI) felt, a felt and a (woven) fabric
made of m-aramide (Nomex® ), and a p-aramide (Twaron® ) fabric. These were stan-
dard products used, e.g., for filtration or as ballistic material (bullet-proof vests) and
were provided by VS-Indutex GmbH, Krefeld, Germany, BWF Tec GmbH & Co.
KG, Offingen, Germany, and Heimbach GmbH & Co. KG, Düren, Germany, re-
spectively. Additionally, a reinforcement fabric made of glass fibers was included
in the study.
2.2. Sol–Gel Coating
The sols applied in this study were based on 3-glycidoxypropyltrimethoxysilane
(GPTMS). For preparation of the sol, 8 ml of GPTMS were pre-hydrolyzed with
0.01 N HCl. The resulting sol was diluted with about 16 ml ethanol before adding
1.875 g bisphenol A (BPA). Immediately before application 160 µl Zonyl FSA
(Aldrich) and 128 µl methylimidazole as a catalyst were added. UV-absorbing coat-
ings were prepared by pre-hydrolyzing GPTMS and addition of BPA. This basic
sol was modified either with ZnO dispersed in water or a freshly prepared titanium
dioxide sol. The proportion of the particles in the resulting coatings was approxi-
mately 10 wt%.
The sols were applied to the textile samples using a laboratory padder (Werner
Mathis AG, Switzerland), with a coating speed of 0.5 m/min. The maximum width
of the textile samples was about 0.30 m. After padding, the samples were dried in a
Labcoater LTE-S (Werner Mathis AG) at 130◦ C for at least 40 min.
2.3. Aging Simulation and Characterization of Textile Samples
The aging behavior was studied in accelerated aging programs simulating (a) ag-
gressive vapor environments, (b) exposure to aggressive liquids and (c) exposure to
intense UV radiation.
To simulate vapor environments, the samples — fabrics of size 20 × 150 mm2
— were suspended over either ammonia (NH3 ) or sulfurous acid (H2 SO3 /SO2 ) at
constant temperatures with varying storage times. The samples were not in physical
contact to the acid or base. For temperatures below 100◦ C the samples were stored
in an Erlenmeyer flask (open atmosphere); for temperatures of 100◦ C and higher,

the aging experiment took place in an autoclave (closed atmosphere).
The effect of aggressive liquids was investigated only in the case of the glass
fiber fabric, which is used as a reinforcement fabric, e.g., for concrete. The aging
was simulated by storing the samples in a cement slurry at 60◦ C for 6 days.
Aging by exposure to UV radiation was investigated only in the case of the
p-aramide (Twaron® ) fabric. The samples were placed in a UV-chamber for about
6 h (Rayonet Photochemical Chamber Reactor, type RPR-100). It should be noted
that ozone was not excluded during the irradiation.
Qualitative characterization of textile samples was done by scanning electron
microscopy (SEM). Quantitatively, the aging was characterized by measurement of
the tensile strength of the fabrics.

3.1. Aging Behavior
The aging behavior of the various fiber materials was studied by accelerated aging
simulation. To evaluate the behavior in aggressive environments, the samples were
exposed to basic (ammonia) or acidic (sulfurous acid) vapors at a constant tem-
perature of 60◦ C and the tensile strength of the samples measured as a function of
exposure time, i.e., duration of storage. The elevated temperature served to accel-
erate the aging process in comparison to real-time aging at ambient temperature. It
should be noted that the temperature in the simulation program has to be chosen
in reference to the actual application of the textile. It is quite obvious that materi-
als used, e.g., in hot gas filtration will require far higher temperatures in the aging
program.
Figure 3(a) and 3(b) summarizes the results obtained from m-aramide and PI
fabrics and felts. Additionally, data from an earlier study by the authors [2] on the
behavior of common fiber polymers such as PET and polyamide (PA) 6.6 are given
for comparison. The data in Fig. 3(a) show that at 60◦ C m-aramide is quite resistant
to the basic vapor and does not show any loss of mechanical strength after 300 h
storage. Similar to PET, PI, however, is highly sensitive to basic conditions and is
significantly damaged under these conditions, losing about 20% of its strength al-
ready after 2 h, and nearly 70% after 3 h storage (cf. also Fig. 1). The degradation of
m-aramide is higher in the case of acidic vapors (Fig. 3(b)) leading to a decrease in
tensile strength of approximately 20% following exposure for 300 h. Nevertheless,
m-aramide shows a rather good chemical resistance in vapor environments.
It is known from practical experience, however, that high-temperature polymers
such as m-aramide are damaged or even destroyed in quasi-closed systems at tem-
peratures far below the specific working temperature of the polymer, but with high
humidity. An example of such conditions is the application of textile transport belts
in high-temperature continuous presses as, e.g., a chipboard press, where the tex-
tile is exposed to high pressure and high humidity at temperatures of the order of
(a)
(b)
Figure 3. Tensile strength of textiles made of PI and m-aramide as a function of exposure to (a) basic
and (b) acidic vapors at 60◦ C. The fabric samples were suspended over base or acid in an Erlenmeyer
flask. For comparison, data of similar experiments on PET and PA 6.6 by the authors [2] are shown.
All values are normalized to the tensile strength of the ‘as-received’ sample.
200◦ C. It was found that m-aramide is destroyed at this rather moderate tempera-
ture due to hydrolysis. Obviously, the process leading to the aging, or destruction,
of the polymer is accelerated by the conditions of the closed system.
Figure 4. Tensile strength of m-aramide felt as a function of exposure time to acidic vapor at different
temperatures. The samples were suspended over a pool of the acid in an autoclave. All values are
normalized to the tensile strength of the ‘as-received’ fabric.
Given this background, it is sensible to perform accelerated aging experiments

under similar, ‘closed-system’ conditions. In order to study the effect of high vapor
temperatures and pressures, experiments were carried out in an autoclave. Again,
the samples were suspended over either NH3 or H2 SO3 at constant temperatures
with varying storage times. Typical results are shown in Fig. 4 for the example of
the aging of m-aramide in an acidic atmosphere (H2 SO3 /SO2 ). While the polymer
is quite resistant to the acidic vapor in an open atmosphere, a significant increase in
degradation is found at higher temperatures in the ‘closed-system’ conditions.
An often overlooked fact in the consideration of long-term application of tech-
nical textiles is the deposition of particulate matter on the textiles, either directly
due to the application — e.g., in filtration — or from airborne dust, fumes or other
emissions. Particles on surfaces have a certain relevance to aging as the chemi-
cal conditions in the contact area between the particle and the fiber can be more
extreme than in the general atmosphere/environment. Chemical processes might
originate from, e.g., organic substances adsorbed on particle surfaces or may be
due to the hydration shell on particles. As a consequence, a strong chemical attack
occurs, often leading to the formation of holes in the textile.
In order to demonstrate this phenomenon, two felts made of m-aramide were
stored for 7 days in acidic atmosphere (H2 SO3 /SO2 ) at 60◦ C. Again, the samples
were suspended over the acid in an Erlenmeyer flask. In this experiment, however,
one sample was artificially soiled with graphite particles. As the data summarized
in Fig. 5 show the aging is significantly accelerated by the presence of graphite
particles.
Figure 5. Tensile strength of m-aramide aged for 7 days in acidic vapor at 60◦ C. The samples were
suspended over a pool of the acid in an Erlenmeyer flask. One sample was artificially soiled with
graphite particles. All values are normalized to the tensile strength of the ‘as-received’ fabric.
Inorganic fibers play an increasingly important role in the manufacture of high-

performance textiles. Today, fabrics made of glass fiber are used, among other
applications, as reinforcement fabrics for concrete. Here, the extreme alkaline na-
ture of the concrete during its setting — pH up to 14 — provokes a rapid aging of
the fiber material. Based on this, aging studies were conducted with the inorganic
material in this work, also.
Besides the already described aging scenario of storage in alkaline and acidic va-
pors, the glass fiber fabric was also stored in a cement slurry keeping the application
mentioned in mind. Results of these experiments are given in Fig. 6. The samples
were stored in a saturated ammonia atmosphere for 24 h, over sulfurous acid and
in cement slurry for 6 days. The samples were kept at 60◦ C in all cases. As could
be expected, the glass fiber fabric was unaffected by the acidic vapor, while alka-
line conditions led to rapid aging with a decrease of the tensile strength to 61% in
ammonia vapor and to 88% in the cement slurry.
3.2. Improvement of Aging Behavior by Nano-composite Thin Layers
The second part of the experiments was concerned with the potential of inorganic–
organic hybrid polymers deposited as thin layers in a sol–gel process on the fibers
to increase their resistance to aggressive environments, vapor as well as liquid.
The inorganic–organic hybrid polymers (often termed as nano-composites) were
of interest because of (a) their partly inorganic character and (b) their cross-linked
structure. Principal concepts for the preparation of appropriate sols have been pub-
lished by Textor et al. [6]. Based on this work, GPTMS-based sols were prepared
and applied to the fabric samples.
Figure 6. Tensile strength of glass fiber fabrics artificially aged in ammonia vapor (24 h at 60◦ C), in
an acidic atmosphere (H2 SO3 /SO2 ) (6 days at 60◦ C) and cement slurry (6 days at 60◦ C). All values
are normalized to the tensile strength of the ‘as-received’ fabric.
Figure 7. Tensile strength of as-received (open symbols) and sol–gel coated (closed symbols) fabrics
made of m-aramide artificially aged in an acidic atmosphere (H2 SO3 /SO2 ). All values are normalized
to the tensile strength of the ‘as-received’ fabric.
Exemplary results are given for m-aramide and glass fiber fabrics in Figs 7 and 8,
respectively. The data show that the sol–gel coating could not improve the aging be-
havior of m-aramide in acidic atmosphere (H2 SO3 /SO2 ) at 60◦ C as well as at 130◦ C
(closed system). On the contrary, the measured tensile strengths were slightly be-
low the values of the uncoated samples (Fig. 7). The same observation was found
Figure 8. Tensile strength of as-received (open symbols) and sol–gel coated (closed symbols) glass
fiber fabrics artificially aged in an acidic atmosphere (H2 SO3 /SO2 ) and in ammonia vapor at 60◦ C.
All values are normalized to the tensile strength of the ‘as-received’ fabric.
in the case of aging of glass fiber fabrics in acidic vapor (60◦ C only), whereas the
inorganic–organic hybrid polymer formed a highly efficient barrier against the al-
kaline attack from ammonia vapor. As shown in Fig. 8, the strength of the fabric is
totally unaffected by the aggressive environment. Again, the glass fiber fabric was
also subjected to the alkaline attack from the cement slurry. As was found for the
ammonia vapor, no loss of tensile strength could be measured following storage of
the sample in the cement slurry for 6 days (Fig. 9).
As stated before, the silane chemistry allows also stabilization of nano-sized
metal oxides, providing an additional ‘degree of freedom’ to introduce functional-
ity. Given the favorable UV absorption properties of nano-sized ZnO and TiO2 par-
ticles (see, e.g., [11]), specific sols were prepared for UV protection of p-aramide.
Like other aromatic polymers, this fiber polymer has a low UV-resistance, i.e., it
can be degraded following intensive exposure to UV light with a resulting loss in
tensile strength.
When using nano-sized particles, it is extremely important to keep the particles
separated for effective modification. Theoretical considerations on this topic can
be found in the literature, see, e.g., [10]. Here, silane chemistry offers appropriate
means to stabilize the particles, i.e., avoid aggregation and agglomeration, as well
as provides possible ways to immobilize the particles on the fiber surface by em-
bedding the particles in the silane-based hybrid polymers. In these processes, the
OH-groups present on the particle surface are usually modified with alkoxysilanes
in a condensation step. A network is built following the reaction scheme sketched
in Fig. 10.
Figure 9. Tensile strength of as-received and sol–gel coated glass fiber fabrics, artificially aged in
ammonia vapor (24 h at 60◦ C) and cement slurry (6 days at 60◦ C). Values are normalized to the
tensile strength of the ‘as-received’ fabric.
Figure 10. Chemical concept for embedding nano-sized metal oxide particles in silane-based hybrid
polymers.
The sols were again based on 3-glycidoxypropyltrimethoxysilane (GPTMS) pre-

hydrolyzed with hydrochloric acid. Bisphenol A was added as the cross-linking
agent. ZnO-particles with an average size between 5–10 nm were dispersed into the
sol with a bead mill, while TiO2 particles were added as a freshly prepared sol.
As can be taken from the data shown in Fig. 11, the tensile strength of a com-
mercial ‘as-received’ fabric decreases by about 80% following a standardized test in
Figure 11. Tensile strength of as-received and sol–gel coated p-aramide fabric before (light grey
columns) and after (dark grey columns) UV exposure. The sol was modified either with ZnO dispersed
in water or a freshly prepared titanium dioxide sol. All values are normalized to the tensile strength of
the ‘as-received’ fabric.
a UV reactor. It should be mentioned that the decomposition might be partly due to

ozone produced in the chamber. The basic sol already improves the UV-resistance
impressively due to the absorption of UV-light by the aromatic BPA used as cross-
linking agent but best results are obtained with the coatings filled with about 10 wt%
of the nano-particles. The data show that the coated samples show only a slight de-
crease in tensile strength, if any, after UV exposure. The protective performance of
the TiO2 -filled sol was slightly higher than that achieved using ZnO particles.
4. Summary
In view of the technical application of textile products used in filtration and other in-
dustrial installations — including outdoor-application in polluted industrial areas —
the aging behavior of textiles made of polyimide (PI) and other high-performance
polymers was studied by accelerated aging simulation. The samples were exposed
to basic (ammonia) or acidic (sulfurous acid) vapors at a constant temperature of
60◦ C and the tensile strength of the samples measured as a function of exposure
time, i.e., duration of storage. The data showed that at 60◦ C m-aramide is quite re-
sistant to the basic vapor and does not show any loss of mechanical strength even
after 300 h storage. PI, however, is highly sensitive to basic conditions and is sig-
nificantly damaged under these conditions, losing about 20% of its strength already
after 2 h, and nearly 70% after 3 h storage. The degradation of m-aramide is higher
in the case of acidic vapors leading to a decrease in tensile strength of approxi-
mately 20% following an exposure for 300 h. While this polymer is quite resistant
to the acidic vapor in an open atmosphere, a significant increase in degradation is
found at higher temperatures in the ‘closed-system’ conditions.
The second part of this study was to investigate the potential of inorganic–organic
hybrid polymers deposited as thin layers to increase the resistance of exemplary
technical textiles against aging-promoting influences. GPTMS-based sols were pre-
pared and applied to the fabric samples. The data showed that the sol–gel coating
did not improve the aging behavior of either m-aramide or glass fiber fabrics in
acidic vapor, whereas the inorganic–organic hybrid polymer formed a highly effi-
cient barrier against the alkaline attack from ammonia vapor in the case of the glass
fiber fabrics.
It should be remarked that the experimental measurements illustrated the rele-
vance of the various parameters of simulation design. Aging simulation is designed
along the given conditions of application and to accelerate aging in the interest of
lifetime prediction. An intrinsic problem of accelerated aging simulation is, how-
ever, that there are NO RULES for the design of the simulation program. Also, the
correlation between simulated and real-time aging is not easily assessed.
5. Outlook — Aspects of Lifetime Predicton

A fundamental problem is the assessment of lifetime from the aging simulation.
A standard procedure to predict the lifetime of a product (fiber, yarn, fabric) is
to perform an accelerated aging simulation following some design and plot ten-
sile strength vs storage duration on a logarithmic timescale. The data are then
extrapolated linearly to, e.g., a pre-determined residual tensile strength. The actual
viscoelastic behavior of a polymer, however, is more complex than a simple log-
linear decrease of the modulus. It is properly described by the ‘Relaxation Master
Curve’ (RMC), which describes the development of the elastic modulus with time
and shows different regimes of relaxation as the example of the RMC of a PET
multifilament yarn published by Beier and Schollmeyer [12, 13] shows (Fig. 12).
The RMC was proposed for lifetime-prediction of polymers already several years
ago.
Since the RMC covers many time decades, it is not readily accessible by experi-
ment. Numerical solutions to construct the RMC on the basis of data from relaxation
curves measured in short time interval of, e.g., several minutes have been published
in the last years. The relaxation measurements are performed by varying an experi-
mental parameter such as temperature, initial strain etc. By means of appropriate
mathematical algorithms based on the various viscoelastic models found in the
literature (referred to as ‘time-analogies’ in the following), the difference of the
varied experimental parameter from a reference value is related to a time interval
by which the measured relaxation curves are shifted along the time axis. Already
in 1955, Williams, Landel and Ferry [14] proposed a time-shift of an individual
short-time relaxation curve measured at a temperature T on the log(t)-axis, i.e.,
Figure 12. Relaxation Master Curve (RMC) of a PET yarn according to Beier and Schollmeyer [13].
The graph shows the data of 4 short-time relaxation measurements from which the RMC was con-
structed by superposition. The short-time curves covered only several minutes and were recorded with
the initial elongation ε0 , i.e., ε (t = 0), ranging from 5 to 30%.
a ‘time–temperature-analogy’, according to:

a · (T − Tg )
log10 (aT ) = ,
b + (T − Tg )
for amorphous polymers, where a, b are constants, T the measurement temperature
and Tg the glass transition point of the polymer. Analogous mathematical algo-
rithms have been published recently, e.g., by Barbero and Ford [15] and models for
crystalline polymers can be found in the literature (see, e.g., [16]).
Alternative numerical models based on a ‘time–strain-analogy’ have been pub-
lished by, e.g., Schulz et al. [17], who superposed isothermal relaxation measure-
ments from different initial elongations ε (t = 0) as shown in Fig. 13. Beier and
Schollmeyer [12, 13] made use of a ‘time–strain-rate-analogy’, i.e., they superposed
isothermal measurements of the elastic modulus using different rates of elongation.
With regard to aging simulation and lifetime prediction, the scope of future work
will be to determine the RMC following accelerated aging simulation in order to
predict lifetime more accurately.
Acknowledgements
The experiments presented here were conducted under national projects AiF-no.
12431N, 20Z, and 12882N. The authors wish to acknowledge financial support by
the Forschungskuratorium Textil e.V. for these projects. This support was granted
within the program Industrielle Gemeinschaftsforschung (IGF) from resources of
Figure 13. RMC-construction from a spectrum of short-time measurements taken at different initial
elongations ε (t = 0), which are denoted εn to εn−i . By relating the difference of the initial strain
from a reference value to a shift on the log-time axis (‘time–strain-analogy’), the individual curves
are superposed to form the RMC (dashed line). E0 and E∞ denote the initial modulus of the new
material and the final modulus after long-time relaxation (from [16]).
the Bundesministerium für Wirtschaft und Technologie (BMWi) via a supplementary

contribution by the Arbeitsgemeinschaft Industrieller Forschungsvereinigungen e.V.
(AiF).
References
1. J. N. Hay, Pure Appl. Chem. 67, 1855–1858 (1995).
2. Th. Bahners, U. Schloßer and E. Schollmeyer, Technische Textilien 46, 78–80 (2003).
3. Th. Bahners and E. Schollmeyer, Chemical Fibers Intl. 53, 420–421 (2003).
4. K. Opwis, T. Bahners and E. Schollmeyer, Chemical Fibers Intl. 54, 116 (2004).
5. C. J. Brinker and G. Scherer, Sol–Gel Science: the Physics and Chemistry of Sol–Gel-Processing.
Academic Press, Boston (1990).
6. T. Textor, D. Knittel, Th. Bahners and E. Schollmeyer, in: Silanes and Other Coupling Agents,
7. T. Textor, T. Bahners and E. Schollmeyer, Prog. Colloid Polym. Sci. 117, 76–79 (2001).
8. H. Schmidt, J. Non-Cryst. Solids 178, 302–312 (1994).
9. C. Roscher and M. Popall, Mater. Res. Soc. Symp. Proc. 435, 547–552 (1996).
10. Th. Bahners, T. Textor and E. Schollmeyer, in: Silanes and Other Coupling Agents, K. L. Mittal
(Ed.), Vol. 4, pp. 141–152. VSP/Brill, Leiden (2007).
11. J. P. Hewitt, Drug and Cosmetic Industry 157(2), 28–36 (1995).
12. M. Beier and E. Schollmeyer, Angew. Makromol. Chem. 169, 1–15 (1989).
13. M. Beier and E. Schollmeyer, Angew. Makromol. Chem. 168, 53–79 (1989).
14. M. L. Williams, R. F. Landel and J. D. Ferry, J. Am. Chem. Soc. 77, 3701 (1955).
15. E. J. Barbero and K. J. Ford, J. Eng. Mater. Technol. 126, 413–419 (2004).
16. R. Kosfeld and W. Lohmann, Colloid Polym. Sci. 258, 209 (1980).
17. J. Schulz, A. Stalevitsch, A. Eickmeier, A. Bossmann and E. Schollmeyer, Angew. Makromol.
Chem. 234, 159–175 (1996).
Because of their many desirable attributes, polyimides
constitute an important class of materials and they have
found applications in many and varied arenas ranging from
microelectronics to aerospace to membrane technology to
adhesives.
The book is divided into four parts as follows: Part 1:

Polyimides and
Synthesis, Properties and Bulk Characterization; Part 2:
Surface and Adhesion Aspects; Part 3: Applications; Part
4: General Papers. The topics covered include: synthesis of Other High
various kinds of polyimides with different characteristics;
structure-property relationships; photo-definable polyimi-
des and other materials; space-survivability of polyimides; Temperature
Polyimides and Other High Temperature Polymers:

Synthesis, Characterization and Applications,
heavy ion irradiation of polyimides; surface modification
and adhesion aspects of polyimides; metallization of po-
lyimides; polyimide foam nanocomposites; ion-containing
polyimides; zeolite modified polyimides; cobalt-containing
Polymers:
polyimides; application of polyimides in aerospace, micro-
electronics, purification of bio-fuels, membrane techno-
logy, moisture absorption and sensing, low Earth orbit, Synthesis,
and others; organosilicon films and lifetime assessment of
high-temperature fibers.
This book represents the cumulative wisdom of many

Characterization
world-renowned scientists and technologists and reflects
the current state-of-the-knowledge in the domain of po-
lyimides and other high temperature polymers. All signals
and Applications,
indicate that interest in these materials will continue
unabated and as their new and improved versions are deve-
loped, more new applications will emerge. Volume 5
In essence, this book provides a repository of current
information on this topic and should be of interest and
Edited by
value to anyone interested in polyimides and other high
temperature polymers.
K.L. Mittal
K.L. Mittal
Volume 5
(Ed.)
9 789004 170803
VSP
brill.nl A.S.
POHTPS5.indd 1 18-03-2009 14:12:57

Polyimides and Other High Temperature Polymers - Synthesis Characterization and Applications

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Polyimides and Other High Temperature Polymers - Synthesis Characterization and Applications

Uploaded by

Copyright:

Available Formats

Because of their many desirable attributes, polyimides

constitute an important class of materials and they have

The book is divided into four parts as follows: Part 1:

Polyimides and Other High Temperature Polymers:

This book represents the cumulative wisdom of many

POHTPS5.indd 1 18-03-2009 14:12:57

No claim to original U.S. Government works

International Standard Book Number-13: 978-9-04-742411-6 (eBook - PDF)

TSDC Spectroscopy Investigation of Polyetherimide

Part 2: Surface and Adhesion Aspects

Cobalt-Containing Polyimides for Moisture Absorption and Sensing

This volume chronicles the proceedings of the Fifth International Symposium

tionships; photo-definable polyimides and other materials; space-survivability of

Ying-Ling Liu ∗ and Shi-Yi Chen

* To whom correspondence should be addressed. Tel.: +886-3-2654130; Fax: +886-3-2654199; e-mail:

prove their heat-distortion temperature and solvent resistance by cross-linking. The

4,4 -(Hexafluoroisopropylidene)diphthalic anhydride (6FDA), 4,4 -methylene di-

2.2. Characterization and Measurements

2.5. Preparation of α, ω-Furan-Terminated Polyimide (PI-F)

3. Results and Discussion

Figure 1. FT-IR spectra of α, ω-functionalized polyimides.

Figure 2. 1 H NMR spectra of α, ω-functionalized polyimides.

Figure 3. TGA thermograms of α, ω-functionalized polyimides in nitrogen under a heating rate of

monitored by measurements of the inherent viscosities of the resulting polyimides.

0.46 dl/g in 350 h of reaction. The formation of maleimide-furan adducts linking

polyimides are coded as PI-DA-I-R and PI-DA-II-R, respectively. Table 2 repre-

Rate constants of retro-DA reaction at various temperatures Activation energy

Temperature (◦ C) 110 120 130 140 150 160 170 180

* Tel.: (+81-942) 35-9408; Fax: (+81-942) 35-9329; e-mail: tsuda@kurume-nct.ac.jp

Figure 1. Synthesis of conventional polyimides and soluble polyimides.

Figure 2. Representative methods for the enhancement of solubility of polyimides.

Figure 3. Aromatic diamines bearing long-chain alkyl groups.

and used as received. 3,3 ,4,4 -Benzophenone tetracarboxylic dianhydride (BTDA),

JEOL JNM-AL400 FT NMR in CDCl3 or dimethyl sulfoxide-d6 with tetramethyl-

3. Results and Discussion

Figure 4. Synthesis of aromatic diamines bearing long-chain alkyl groups.

catalyzed by triethyl amine in THF gave alkyl 3,5-dinitrobenzoate in satisfactory

Figure 5. Synthesis of diaminophenylalkoxybenzamide (DPABA).

Figure 6. Synthesis of dendritic terphenyl diamines (12G1 and G2-AG-terphenyldiamines).

shown in Fig. 6 using the first- and second-generation Percec-type monodendrons.

3.2. Polymer Synthesis

3.3. Effect of Linkage Group

Soluble Polyimides Bearing Long-Chain Alkyl Groups

tive AODB unit for soluble polyimides, and BTDA/AODB-12/DDE copolyimides

effect on solubility of polymers with polymethylene units. It can be assumed that

3.6. Effect of Branched Alkyl Group

To improve the solubility of rigid polyimides based on DBAE, branched alkyl

3.7. Dendronized Soluble Polyimides

Diaminea Poly(amic acid) Polyimide

Dianhydridea, b Diaminec Comonomerd Solvent Poly(amic acid) Polyimide Tg j Td (◦ C)k

BTDA 12G1 NMPe 0.31 insoluble

Soluble Polyimides Bearing Long-Chain Alkyl Groups

a BTDA: 3.3 ,4,4 -benzophenonetetracarboxylic dianhydride.

Figure 12. 1 H-NMR spectrum of a copolyimide based on BTDA/ADBA-12/DDE (100/50/50). 1 H

3.9. Alignment Layer Application for VAN-LCDs

M. Sarojadevi ∗ , R. Hariharan and N. Amutha

procedure for the preparation of bis(4-amino-3,5-dimethyl phenyl)4 -chloro phenyl

Scheme 1. Synthesis of diamines.

2.5. Synthesis of Polyaspartimides

Fourier Transform Infrared (FT-IR) spectra were recorded on a Nicolet spectrom-

3. Results and Discussion

Figure 1. 13 C-NMR spectrum of BABOM.

3.2. Synthesis of Polyimides