Handbook of Polymer Synthesis Second Edition Episode 9 ppsx

9
Polyimides
Javier de Abajo and Jose
´
G. de la Campa
Institute of Polymer Science and Technology, Madrid, Spain
I. INTRODUCTION
Polyimides are polymers incorporating the imide group in their repeating unit, either
as an open chain or as closed rings. However, only cyclic imides are actually of interest
concerning polymer chemistry. Thus, under the generic name polyimides, we will
exclusively refer to cyclic polyimides in this chapter.
The first reference to a polyimide was dated at the beginning of the 20th century [1],
but the actual emergence of polyimides as a polymer class took place in 1955 with a patent
of Edwards and Robinson on polymers from pyromel litic acid (1,2,4,5-tetracarboxy-
benzene) and aliphatic diamines [2]. Since then, growing interest in polyimides has brought
about a big expansion of the science and technology of this family of special polymers,
which are characterised by excellent mechanical and electrical properties along with
outstanding thermal stability. Among the wide list of reported heat-resistant condensati on
polymers [3–5], polyimides have gained a prominent position due to their good properties–
price–processability balance. And from the production figures, it can be inferred that
polyimides stand virtually alone with respect to providing useful, available, technological
materials.
Furthermore, while at the beginning polyimides found application in a rather
restricted variety of technologies, mainly on the form of films and varnis hes for the
aerospace and electrical industries, the discovering of addition polyimides, and, more
recently, of thermoplastic, processable aromatic polyimides has widened the range of
properties and application possibilities to a great extent. Presently, they should be
considered as versatile polymers with an almost unlimited spectrum of applications as
specialty polymers for advanced technologies [6–12].
In a list of applications of polyimides, the following should be included:
 Insulating films, coatings and laminates

 Molded parts
 Structural adhesives
 Insulating foams
 High-modulus fibers
 High-temperature composites
 Permselective membranes
Copyright 2005 by Marcel Dekker. All Rights Reserved.
From the beginning, the major proportion of research effort on polyimides was
directed to the development of wholly aromatic species, seeking for high thermal stability.
In this respect, wholly aromatic polyimides are materials that can retain their properties
almost unchanged for long periods at 250–300

C. But it was soon realized that the
application of aromatic polyimides, and in general aromatic polyheterocycles, was not
possible from the melt and, furthermore, their extreme structural rigidity and high density
of cohesive energy made them insoluble in any organic media. Given the excellent
properties of the aromatic polyimides, structural modifications were soon outlin ed in
order to overcome these limitations, and as a consequence of the many research efforts
made in this direction, the chemistry of polyimides has greatly enriched thanks to the
many improvements achieved in the last thirty years [9–11,13–15].
II. CONDENSATION POLYIMIDES
A. Polyimides via Poly(amic acid) from Dianhydrides and Diamines.
Reaction Conditions and Monomers Reactivity
The polycondensation of an organic dianhydride and a diamine is the traditional method
employed in the synthesis of polyimides (Scheme 1).
ð1Þ
This general scheme is valid for both aliphatic and aromatic polyimides. Since this is
the route preferably used for aromatic, aliphatic and cycloaliphatic polyimides of technical
importance, it has been the subject of numerous studies, and the main aspects of
the mechanisms and kinetics are fairly well known [16]. It is a two-step reaction. In the

first step the nucleophilic attack of the amine groups to the carbonyl groups of the
dianhydride gives rise to the opening of the rings yielding an intermediate poly(amic acid)
(Scheme 2).
ð2Þ
The symmetrical and unsymmetrical poly(amic acid)s are intended, since both are
possible.
Copyright 2005 by Marcel Dekker. All Rights Reserved.
The poly(amic acid) is converted, in the second step, to the corresponding polyimide
through a cyclodehydration reaction (Scheme 3).
ð3Þ
This simplified scheme may be envisioned in a more complete form by using
monofunctional species (Scheme 4).
ð4Þ
The first step is crucial to attain high molecular weight, and the second has a great
influence in the final nature of the polyimide since a quantitative conversion in the
cyclodehydration process is needed to have a pure, fully cyclized polyimide. Highly polar
solvents are suitable med ia to dissolve monomers and poly(amic acid)s. N,N-dimethyl-
acetamide (DMA), N,N-dimethylformamide (DMF), dimethylsulfoxide (DMSO), and
N-methyl-2-pyrrolidinone (NMP) are the most adequate. Purity of solvents and reactants,
and strict stoichiometric balance are requirements of polycondensation reactions that
fully fit polyimides synthesis, where a careful control of the reaction variables is essential
to achieve high molecular weight [17–19]. For instance, rigorous exclusion of water is
a key condition, as well as a moderate polymerization temperature (about 0

C or less)
in poly(amic acid) formation in order to limit the competition of side reactions and a
premature release of imidation water.
A comparative study of the influence of side reactions has been made by Kolegov
et al. [20], who have considered the following sequence of possible reactions (Scheme 5).
The concurrence of these reactions can obviously alter the progress of the main

reactions 1 and 2 and may prevent a high molecular weight. Experimental data of
polycondensations of diamines and dianhydrides can generally be treated as second order
reversible reactions, but the comparatively great magnitude of K
1
allows the calculation
of rate constants according to an irrversible reaction. In fact K
1
is greater than K
2
, K
4
and
Copyright 2005 by Marcel Dekker. All Rights Reserved.
K
5
and K
5
by approximately seven orders of magnitude and over fifteen times greater
than K
3
[21].
ð5Þ
The reactants concentration also plays a determinant role. It has been stated that
on plotting the inherent viscosity of poly(amic acid) against the initial concentration of
monomers, a curve with a maximum can be attained. This maxi mum is presumably
different for each monomers combination and solvent, but from the available data
it is accepted that for high molecular weight to be obtained 0.4 to 0.8 mol/L
monomer concentration is to be used [22–24]. The figures correlate well with data
reported for the synthesis of aromatic polyamides from aromatic diamines and aromatic
diacid chlorides [25].

In order to carry out a successful polymerization, a fixed mode of monomers
addition has been suggested. Traditionally, the addition of the dianhydride (preferably as
a solid) on the diamine solution has been considered as the right mode of addition, and
that because the anhydride is sensitive to solvent impurities (water, amines), and even
to solvent reaction, in much greater degree than the diamine, so that with the diamine in
large excess the main reaction will be favoured [22,26,27]. Furthermore, unlike aromatic
diamines, aromatic dianhydrides are not easily dissolved at low temperature.
Nevertheless, the same results can be obtained regardless the order of monomers
addition in the synthesis of poly(amic acid)s from pyromellitic dianhydride and
oxydianiline if the reaction conditions are stretched in terms of dryness, stoichiometry,
and solvent and monomers purity [28]. This indicates that the classical order of mono mers
Copyright 2005 by Marcel Dekker. All Rights Reserved.
addition has been imposed by the sensitivity of dianhydrides to water and solvent
impurities more than by reactivity or solubility concerns.
The progress of the polycondensation reaction largely depends also on the nature of
the monomers, and particularly on the monomers reactivity. As a rule, electron deficient
diamines will react more slow ly than electron rich diamines. At this respect, some studies
have been made on the reactivity of diamines by conventional methods. A reliable
approach to quantify the reactivity of diamines and dianhydrides, is the calculation of
molecular parameters by means of the modern methods of Computational Chemistry. The
reactivity of diamines against acylating monomers like acid chlorides have been reported
[29,30]. Likewise, theoretical calculations can be made to estimate the relative reactivity of
diamine and dianhydride monomers.
Quantum semiempirical methods are reliable tools for the determination of
parameters involved in the reactivity of organic reactants [31]. In fact, some partial
studies were performed by Russian researchers more than twenty years ago to relate
electronic parameters with reactivity of polyimide monomers [16] . However, the methods
they used to calculate these parameters have been nowadays ov ercome, and consequently,
it seems interesting to obtain new theoretical data that could be correlated with
experimental results. Thus, the method AM1 [32] included in the MOPAC package,

version 6.0 [33] has been used for the calculations that follow.
In spite of the commercial importance of polyimides and of the huge number of
new monomers synthesized in the last twenty years, the amount of kinetic data for the
acylation reaction of diamines and dianhydrides is very scarce, and we have only been able
to find data for a few diami nes and an even shorter number of dianhydrides [34].
As commented before, the acylation reaction between a diamine and a dianhydride takes
place by the attack of the lone pair of the nitrogen of the amine to the centre of low
electronic density located in the carbonylic carbon of the anhydride. Therefore, the
reaction will be controlled by the interaction between the occupied orbitals of the diamine
and the unoccupied orbitals of the dianhydride. The reactivity of the amines will be
affected by both the electronic density on the nitrogen and by the energy of the Highest
Occupied Molecular Orbital (HOMO) [29,30]. In dianhydrides, the reactivity will be
determined by the electronic deficiency on the carbonylic carbon and by the energy of the
Lowest Unoccupied Molecular Orbital (LUMO).
As the reactivity will be higher when the difference between both orbitals will be
lower, higher values of E
HOMO
and lower values of E
LUMO
will indicate the more reactive
diamines and dianhydrides respectively. Tables 1 and 2 show the main parameters
calculated for several diamines and dianhydrides, from which kinetic data could be found
in the literature. The calculated values correspond, in all cases, to the more stable
conformation. In both cases, diamines and dianhydrides, the differences of charge, either
on the amino nitrogen or on the carbonylic carbon, are very scarce and, furthermore, in
the case of diamines, because of the fact that the C
Ar
–N bond is out of the plane of
the aromatic ring, the charge transfer from the amine to the ring is difficult. Therefore, the
presence of electronwithdrawing groups does not cause a decreas e of the charge on the

nitrogen but an increase on the polarizability of the N–H bonds.
The values of E
HOMO
in the diamines are controlled by the character of the groups
present in the structure, being higher (higher reactivity) in the case of electron donating
groups. In that way, the higher reactivity should correspond to p-phenylene diamine,
where the second amino group acts as activating of the first one. The lowest reactivity
corresponds to the sulfon yldianiline (DDS O), because of the strong electron withdrawing
character of the sulfone group. These values of E
HOMO
can be related with the
Copyright 2005 by Marcel Dekker. All Rights Reserved.
experimental values of acylation constants shown in Table 1 as it can be seen in Figure 1.
A very good linear relationship can be observed, thus confirming the influence of the
electronic parameters of the diamines in the determination of reactivity.
The reaction of the first amino group, that is converted to amide, causes a decrease
of the reactivity of the second amino group, as it could be expected, which is reflected by a
decrease of E
HOMO
(Table 1). However, contrarily to the expected, a smal l increase of the
electronic density in the amine nitrogen is observed. This effect is probably related with the
out of plane situation of the C
Ar
–N bond, that has been commented above. The decrease
in E
HOMO
is very small in all cases, even for p-phenylene diamine and practically no
influence of the structure of the diamine can be observed.
In Table 2 are shown the electronic characteristics of the dianhydrides (E
LUMO

and
charge on the carbonylic carbon) and their acylation constants. In this case, the presence
of electronwithdrawing groups causes a decrease of E
LUMO
. Thus, the most reactive
compound is the pyromellitic dianhydride, because of the strong activation produced by
the presence of the second anhydride group. Next in reactivity is the dianhydride with the
sulfonyl group, and the lower reactivity corresponds to monomers with a long separation
between both anhydrides, and with electron donating ether groups. However, in this case,
the correlation between theoretical and experimental data is not as good as in the case of
diamines, mainly because of the strong deviation of the linear behaviour observed in the
case of the pyromellitic dianhydride.
Table 1 Electronic parameters and kinetic data for several diamines and their corresponding
monobenzamides.
Diamine Q
N
a
E
HOMO
Q
N
b
amide
E
HOMO
amide
log K
acylation
À0.314 À7.92 À0.319 À8.06 2.48
À0.329 À8.26 À0.330 À8.40 0.00

À0.327 À7.94 À0.328 À8.06 0.37
À0.323 À8.11 À0.323 À8.25 0.79
À0.337 À8.65 À0.338 À8.73 À2.17
À0.326 À8.29 À0.326 À8.39 0.56
À0.354 À8.89 À0.356 À8.99 À2.66
À0.330 À8.32 À0.330 À8.36 0.15
a
Charge on the nitrogen of any of the amino groups in the diamine.
b
Charge on the remaining amino group after the formation of the benzamide on the other side.
Copyright 2005 by Marcel Dekker. All Rights Reserved.
This must be attributed to the effect produced on the reactivity of the second
anhydride group by the form ation of the amide in the first one. Also in this case, the
occurrence of the first reaction causes a decrease in the reactivity of the second anhydride
(an increase of E
LUMO
), but a very small change of the charge on the carbonylic carbon.
However, in this case, the change in the orbitalic energy is significantly higher than for
diamines and it depends very much on the structure of the dianhydride (as most of the
Table 2 Electronic parameters and kinetic data for several dianhydrides and their corresponding
monoamides.
Dianhydrides Q
CðC¼OÞ
a
E
LUMO
Q
CðC¼OÞ
b
amide E

LUMO
amide log K acylation
0.344 À2.86 0.349 À2.18 0.79
0.349(m)
c
À2.20 0.350(m) À1.85 0.13
0.348( p) 0.349( p)
0.349(m) À2.03 0.349(m) À1.68 À0.006
0.351( p) 0.353( p)
0.350(m) À2.30 0.350(m) À2.02 À0.66
0.346( p) 0.346( p)
0.351(m) À2.45 0.352(m) À2.15 1.04
0.341( p) 0.342(p)
0.349(m) À1.67 0.349(m) À1.59 À0.32
0.354( p) 0.354( p)
0.349(m) À1.53 0.349(m) À1.46 À0.80
0.355( p) 0.354( p)
0.351(m) À2.09 0.349(m) À2.01 0.326
0.345( p) 0.355( p)
a
Charge on any of the carbonyl groups in the dianhydride.
b
Charge on the remaining carbonyl groups after the formation of amide on the other side.
c
m and p refer to the carbonyl in meta or para position to the substituent.
Copyright 2005 by Marcel Dekker. All Rights Reserved.
dianhydrides are not symmetrical there are two possibilities of ring opening, one with the
amide group in meta to the substituent and one with the amide in para). Although there
are small differences between both, they are not significant and consequently the values of
E

LUMO
shown in Table 2 are the mean of both possibilities). The reaction of one group in
pyromellitic dianhydride increases E
LUMO
in 0.68 eV (maximum change for diamines was
0.14 eV), but in the case of the dianhydride with the aliphatic chain and the ether groups
between both rings, only an increase of 0.07 eV is observed.
This means that the reactivity for the global acylation does not depend on the
reactivity of the dianhydride but on the reactivity of the less reactive molecule, that is,
the monoreacted anhydride. Consequently, it can be confirmed that the reactivity of
these species is controlled by the energy of the LUMO. A representation of E
LUMO
(monoamide) versus log K is shown in Figure 2. The correlation in this case is very good,
thus confirming the useful ness of the electronic parameters to predict the reactivity, even
in a semiquantitative way.
Thus, the value of E
LUMO
can be used to predict the reactivity of dianhydrides, when
no kinetic data are available. In Tabl e 3 are shown the E
LUMO
values of several important
dianhydrides, for which kinetic data are not available.
All these dianhydrides should have a very high reactivity, because of the lower values
of E
LUMO
for both the dianhydride and the monoamide. In fact, hexafluoroisopropyliden
4,4
0
-diphthalic anhydride should be only slightly less reactive than benzophenone
tetracarboxylic dianhydride, and 2,3,6,7-naphthalene tetracarboxylic dianhydride should

be very similar to biphenyl dianhydride. But if the reaction is controlled by the
monoamide, as we have postulated, the most reactive dianhydride should be 1,4,5,8-
naphthalene tetracarboxylic dianhydride, because E
LUMO
is almost the same than for
pyromellitic dianhydride, but E
LUMO
monoamide is lower than E
LUMO
monoamide of the
pyromellitic (À 2.33 versus À 2.18 eV).
To conclude, it can be said that the reactivity of diamines and dianhydrides to
give polyamic acids, and consequently polyimides, is co ntrolled by the energy of the
frontier orbitals of both types of molecules. Although the charges could also play a role in
Figure 1 Correlation between E
HOMO
of the diamines and log K.
Copyright 2005 by Marcel Dekker. All Rights Reserved.
the control of reactivity, the differences between them are very small and, in addition, in
the case of diamines it is very difficult to determine the real value of charge on the nitrogen
because the amino group is not in the same plane that the aromatic ring.
For the theoretical study of reactivities, selected diamines and dianhydrides have
been chosen along those more frequently used in the preparation of aromatic polyimides.
Most of them are commercially available, but some of them have been produced only at
laboratory scale.
Table 3 Electronic parameters for dianhydrides from which there are no kinetic data available.
Dianhydride E
LUMO
E
LUMO

monoamide
À2.237 À2.03
À2.15 À1.92
À2.85 À2.33
À2.20 À1.905
Figure 2 Correlation between E
LUMO
of the monoreacted dianhydrides and log K.
Copyright 2005 by Marcel Dekker. All Rights Reserved.
For some specific applications, particularly for microelectronics, the purification of
these monomers is sometimes so critical that the isolation of suitable reactants requires
sophisticated purification methods. For instance, miniaturization and tougher processing
requirements for advanced microelectronics have forced researchers to attain ultrapure
poly(amic acid)s from monomers purified by zone refining, and dianhydrides isolated in
solid ingot form [35].
As to the molecular weights of poly(amic acid)s and polyimides, they had been only
very seldom measured and reported. Thus, the usual criterion for molecular size in
poly(amic acid)s and soluble polyimides had traditionally been the inherent viscosity (
inh
)
until the size exclusion chromatography techniques (GPC) were refined and implemented in
last years. The development of many new soluble thermoplastic polyimides has moved also
for a growing interest in knowing the molecular weights, and for an improvement of the
analytical technique for the determination of M
n
’s and M
w
’s. GPC columns, that can work
with aggressive solvents like DMF, DMA or m-cresol at temperatures up to 70–80


C, are
available nowadays and can be used for the analysis of many soluble polyimides [36,37].
Of greatest importance is the cyclodehydration reaction leading from poly(amic acid)s
to polyimides. The general approach in the application of insoluble, wholly aromatic
polyimides as materials involves the elimination of solvent and water at high temperature.
When a poly(amic acid) solution is heated over 200

C, or at a lower temperature in the
presence of a dehydrating agent, such as acetic anhydride/base, the polyimide is attained in
few hours. Logically, the first approach received much more attention in the early years,
because the research effort was mainly focussed to insoluble polyim ides based on
pyromellitic dianhydride, although the chemical imidization of poly(amic acid) films in
the solid state has been the subject of several studies [38–40]. Thermal imidization associated
to classical aromatic polyimides actual ly needs temperatures of about 300

C to ensure total
rings closing, and that is far from being an optimal approach in many instances because
elimination of solvent and water at high temperatures can approach about surface
irregularities, microvoids and even polymer degradation. Furthermore, high temperatures
help for cross-linking side reactions, for example (Scheme 6):
1. ÀNH
2
end groups with the imide rings of the chains:
2. Thermal imidization by means of non-cyclized ortho-carboxyamides:
ð6Þ
Copyright 2005 by Marcel Dekker. All Rights Reserved.
3. Amidation of ÀNH
2
free groups and ortho-carboxyamides:
These reactions do help for a faster immobilization of the chain and, consequently,

for additional difficulties to get 100% cyclodehydration. The strong interactions between
the poly(amic acid) and the solvent also greatly interferes with the intramolecular
cyclization, and its presence does not certainly aid a quantitative conversion. It has
been demonstrated that solvents and poly(amic acid)s readily give rise to complexes [41],
and that solvent rests can remain joined to the polymer even through covalent bonds
[42,43].
A novel preparative method of poly(amic acid)s from aromatic diamines and
dianhydrides consists of carrying out the polycondensation reaction in a precise mixture
of tetrahydrofurane/methanol (9/1 to 6/4 by weight), at room temperature [44]. Average
molecular weights (M
w
) exceeding 150,000 g/mol have been reported for poly(amic acid)s
attained by this method from oxydianiline and pyromellitic anhydride [45]. Moreover,
thermal imidization seems to be more easily achievable on replacing classical high boiling
amide solvents such as DMA or NMP by the easy to evaporate THF and methanol
mixtures [46].
Chemical imidization is normally promoted by acetic anhydride, in combination
with organic bases, for instance pyridine or triethylamine, but other dehydrating agents
can be used, such as propionic anhydride, trifluoroacetic anhydride, N,N-dicylo-
hexylcarbodiimide and the like. Although it can be performed on polyimide films,
chemical imidization is mostly carried out in solution, with the final polyimide being
collected as a precipitate, but most conveniently remaining dissol ved all over the process.
A premature precipitation of the polymer does not ensure total imidization at all,
as partially imidized species can be insoluble in the organic medium. Temperatures and
reaction times amply vary depending on the polymer and the cyclization system. Thus, if
the reaction is conducted at room temperature 24 to 48 h are needed for total imidization,
while some few hours are eno ugh if the chemical cyclodehydration reaction proceeds at
100

C.

The imidization process, either thermally or chemically induced, may be followed
by a variety of means. It has been traditionally studied on poly(a mic acid)s, as well
as with molecular models, by IR and NMR spectroscopy [47,48]. But many other
analytical methods have been used, for instance: TGA [41,49 ,50], DSC [42,51], polarizing
microscopy [41], gas chromatography [52,53], microdielectrometry [54], or torsional
braid analysis [55]. From the numerous contributions on this topic some conclusions
can be drawn. Among other features, we remark that a rate reduction of the imidation
and the rate constant occurs as the conversion increases, so that it can not be considered
as a classical first order reaction. This phenomenon has been explained by consider ing
entropic factors [56]. Since the kinetic data could not be unequivocally assimilated
to a determined reaction order, they were interpreted as if the imidization reaction
could be divided into rapid and slow first order cyclization steps. The retardation in
Copyright 2005 by Marcel Dekker. All Rights Reserved.
the ring closur e reaction has also been explained by the existence of various amide-acid
groups with different reactivities and by the mobility reduction when the linear polymer is
converted into a cyclic chain [38,57].
The formation of isoimide as an intermediate step to imide has been confirmed
also in many instances (Scheme 7). Isoim ides are less stable than imides, so that
isoimide formation does not seem to play any significant role when conversion into
imide is forced by thermal treatment, but it can affect the imid ization process when
rings closure is performed by chemical treatment at moderate tempe ratures.
Furthermore, it has been proved that some solvent/anhydride/base combinations
clearly favou r the formation of isoimide [58], what can in turn, offer some advantage
from a practical view point as polyisoimides are much more soluble than polyimides
[14,59,60].
Chemical imidization is less attractive for commercial and experimental polyimides
that are tested and used in the form of films, but chemical imidation has been the preferred
method concerning experimental polyimides that are soluble in organic solvents in the
state of full imidation. At this respect it is worthy to remark that 100% conversion in the
ring closure step is virtually impossible to achieve, particularly for thermal imidation

at high temperature (about 300

C) in the solid state, due to the complexity of the process
and to the inherent molecular regidity of insol uble polyimides. However, for soluble
polyimides, solution imidization is possible at mild reaction temperatures, for instance
150–200

C, with 100% conversion, and avoiding undesirab le side reactions which lead to
insolubility and infusibility [61].
ð7Þ
By using monomers other than dianhydrides and diamines, a number of methods has
been outlined to synthesize polyimides, for instance from tetracarboxylic acids and their
half diesters. This method can be successfully applied to the preparation of aliphatic–
aromatic polyimides by melt polycondensation of the salt from the diamine and an
aromatic tetracarboxylic acid or half-diester (Scheme 8). The reaction achieved some
importance when polyimides appeared in the 1950s as an alternative for uses such as
Copyright 2005 by Marcel Dekker. All Rights Reserved.
fiber-forming and injection-molding polymers [2,62].
ð8Þ
The method was believed to be valid only for polyimides with melting points
low enough to remain molten during the polymerization, and solution methods were
considered as not suitable since the ‘polyamic salts’ are not soluble in aprotic organic
solvents, so that only low molecular weight polymeric salts were obtained. The method
was not used for many years, mainly because aliphatic polyimides did not show
much higher Tm than conventional nylons, and their main advantage, their high T
g
value, did not mean any useful improvement as their performances under service
conditions were comparable to the semicrystalline aliphatic polyamides, which are in turn
much cheaper.
However, polyalkylenimides can be prepared from pyromellitic anhydride and

a,o-diaminoalkanes in solution of NMP. NMP seems to provide a much more convenient
medium for these reaction than other organic solvents, and in this way, high molecular
weight poly(amic acid)s and polyimides have been attained [63,64].
A revision of this appro ach has been made in last years, and aliphatic and aromatic
salt monomers have been studied as precursors of high molecular weight polyimides.
Salt monomers have shown to be actually highly reactive, as they can produce directly
polyimides in a very short reaction time, and this feature has recently been observed
not only for aliphatic precursors but for aromatic salts as well [65]. Moreover, high
molecular weight polyimides can be achieved by combining the salt monomer method
with high pressure polycondensation, or with microwave induced polycondensation
[65,66]. Other advantages of the salt monomer method is that polycondensations can
progress at high conversion in solid state, at temperatures substantially lower than the
melting point (T
m
) of the polymers, and at a lower tempe rature than the salt monomer
melting point. Thus, imide ring closure takes place simultaneously to water or alcohol
elimination, rendering polyimide in an one-step direct reaction without passing
through poly(amic acid). In general, better results (higher inherent viscosities) are
achieved from half esters than from tetracarboxylic acids, and another feature of this
recently revised method is that highly crystalline polyimides, both aliphatic and aromatic,
can be attained [66].
Half diesters and their derivatives have been extensively used also in the preparation
of aromatic polyimides by the two-step method. The initial step involves the preparation
of the modified monomers, which consist of the half esters themselves or of activated
derivatives. The activation of the half esters is normally directed to the enhancement of the
carboxylic acids reactivity, by converting them into other highly electrophilic groups such
as acyl chlorides. The global synthetic route is depicted in Scheme (9). The high reactivity
of acid chlorides against diamines, makes the solution method at low temperature not
only recommendable but virtually the only possible one if high molecular weights are to be
accomplished.

Copyright 2005 by Marcel Dekker. All Rights Reserved.
Low to medium molecular sizes can be also obtained from the half esters directly
with specific amidation catalysts [67,68].
ð9Þ
Imidization is achieved by thermal treatment of the poly(amic ester) precursor in
the usual way, with elimination of alcohol. This method, because of its relative complexity,
has not got practical significance for conventional polyimides. However it has been of
great importance in the development of photocurable condensati on polyimides [8], and
to study the behaviour of different isomers as starting materials for model polyimides [67].
B. One-step Polycondensation. Thermoplastic Polyimides
As mentioned before, the first generation of fully aromatic homopolyimides, could be used
only in a few application because they had to be applied in the form of soluble polyamic
acids, what limited the materials to be transformed almost exclusively into films or
coatings. They all had to be synthesized by a two-step method.
Further improvements in the chemistry of polyimides during the last years have been
directed towards novel, linear species that are soluble in workable organic solvents or
melt-processable while fully imidized. Thus, changes had to be introduced in the chemical
structure to adapt the behaviour and performance of these specialty polymers to the
demands of the new technologies. As a consequence, a new generation of condensation
polyimides has appeared, the so-called thermoplastic polyimides.
The difficulties to process conventional aromatic polyimides are due to their
inherent molecular features, what is particularly true for the most popular of them:
polypyromellitimides. Molecular stiffness, high polarity and high intermolecular associa -
tion forces (high density of cohesive energy) make these polymers virtually insoluble in any
organic medium, and shift up the transition temperatures well over the decomposition
temperatures. Thus, the strategies to novel processable aromatic polyimides have focussed
on chemical modifications, mainly by preparing new monomers, that provide less
molecular order, torsional mobility and lower intermolecular bonding.
From the various alternatives to design novel processable polyimides some general
approaches have been universally adopted:

 Introduction of flexible linkages, which reduces chain stiffness.
 Introduction of side substituents, which helps for separation of polymer chains
and hinder molecular packing and crystallization.
 Use of 1,3-substituted instead of 1,4-substituted monomers, and/or asymmetric
monomers, which lower regularity and molecular ordering.
 Preparation of co-polyimides from two or more dianhydrides or diamines.
Polyimides with flexible linkages have been known from the advent of high
temperature aromatic polyimides. In fact, most of the commercial, fully aromatic
polyimides contain ketone or ether linkages in their repeating units [69], and early works
Copyright 2005 by Marcel Dekker. All Rights Reserved.
in the field soon demonstrated that dianhydrides having two phthalic anhydride moieties
joined by bonding groups, gave more tractable polyimides [26,70–73].
Many different linkages have been introduced with these purposes, but the most
promising are: –O–, C¼O, –S–, –SO
2
–, –C(CH
3
)
2
–, –CH
2
–, –CHOH–, and –C(CF
3
)
2
–.
These bonding groups may be located on the dianhydride, on the diamine or on both
monomers, or they can even be formed during the polycondensation reaction, when
some functional monomers containing preformed phthalimide groups are used [74,75].
The presence of flexible linkages has a dramatic effect on the properties of the final

polymers. First, ‘kink’ linkages between aromatic rings or between phthalic anhydride
functions cause a breakdown of the planarity and an increase of the torsional mobility.
Furthermore, the additional bonds mean an enlargement of the repeating unit and,
consequently, a separation of the imide rings, whose relative density is actually responsible
of the polymer tractability. The suppression of the coplanar structure is maximal when
bulky groups are introduced in the main chain, for instance sulfonyl or hexafluoro-
isopropylydene groups, or when the monomers are enlarged by more than one flexible
linkage. Some diamines and dianhydrides with a flexible linkage in their structure have
been listed in Tables 4 and 5. The combination of those dianhydrides and diamines,
and also the combination of some of them with conventional rigid monomers like
benzenediamines, benzidine, pyromellitic dianhydride or biphenyldianhydride, offer a
major possibility of different structures with a wide spectrum of properties, particularly
concerning solubility and meltability [76–80].
However, very few of the polymers that can be synthesized combining monomers of
Tables 3 and 4 have been reported as melt-processable, although many of them are soluble
in highly polar organic solvents. All of them show high glass transition temperatures,
commonly over 250

C, and, theoretically, they can develop crystallinity upon a suitable
thermal treatment, mainly those containing polar connecting groups. Thus, depending on
the nature of X and Y in the general formula (Scheme 10), polyimides can be prepared that
show an acceptable degree of solubility in organic solvents.
ð10Þ
Table 6 shows the T
g
’s and solubility of some selected polyimides among those
prepared from monomers of Tables 4 and 5. The combination of non-planar dianhydrides
and aromatic diamines containing flexible linkages, provides the structural elements
needed for solubility and melt processability. Some aromatic polyimides marketed as
thermoplastic materials are based on these statements [69,81–83].

Structural modifications to attain soluble aromatic polyimides have been also carried
out by introducing side substituents, alkyl, aryl or heterocyclic rings. One of the first
references of this approach described the synthesis of soluble aromatic polyimides
containing side phthalimide groups [84,85]. Since then, many attempts have been made
to prepare new monomers, diamines and dianh ydrides, with pendent groups for novel
processable polyimides. Table 7 shows some of these monomers.
Probably, the most promising species are those containing phenyl pendent groups.
The phenyl rest does not introduce any relevant weakness regarding thermal stability, and
provides a factor of molecular irregularity and separation of chains very beneficial in terms
of free volume increasing and lowering of the cohesive energy density [80–91]. Fluorene
Copyright 2005 by Marcel Dekker. All Rights Reserved.
Table 4 Diamines for polyimides containing flexible linkages.
Copyright 2005 by Marcel Dekker. All Rights Reserved.
Table 5 Dianhydrides for polyimides containing flexible linkages.
Copyright 2005 by Marcel Dekker. All Rights Reserved.
diamines and the so-called ‘cardo’ monomers, can be considered in this section, and they
can be seen also as valuable alternatives for the preparation of processable polyimides
[92–94].
For the preparation of this new generation of aromatic polyimides, synthetic
methods have been outlined which allow to achieve the polymers in their state of
full imidization in only one-step. However, the classical sequence poly(amic
acid) ! polyimide is generally followed somehow, although imidization occurs virtually
at the same time that propagation. Amide solvents of high boiling point, as NMP of
N-cyclohexylpyrrolidinone (CHP), nitrobenzene, chloroaromatics, phenols, cresols
[36,37,61,95–100], and even carboxylic acids as benzoic acid [101], are solvents
successfully used for the preparation of processable polyim ides. Moreover, the
beneficial effect of some basic (isoquinoline, tri ethylamine, pyridine) and acid (benzoic,
hydroxybenzoic, salicylic) catalysts have been observed [80,95]. The reaction proceeds
usually at low or moderate temperature in the first stage to favour the formation
of a high molecular weight poly(amic acid), while the second part is led at high

temperature to promote the cyclodehydr ation reaction and to force water separation.
For the splitting off of water, azeotropic solvents are frequently used too. From
a mechanistic point of view, it is to presume that bases help for the nucleophilic
attack of the diamine to the anhydride to form amic acid, and acids catalyse the
closing of the ring with evolution of water. Nevertheless, the role of acids in
the formation of six-membered ring imides, like naphthalimides, needs still an
explanation [102].
Table 6 Properties of selected polyimides from monomers containing flexible bridging groups.
Solubility
Polymer T
g
(

C) NMP m-cresol
259 Æ
267 þþ þþ
299 þ
305 þþ þþ
223 þþ
Copyright 2005 by Marcel Dekker. All Rights Reserved.
Table 7 Diamines and dianhydrides used in the preparation of polyimides with bulky
side groups.
Diamines Dianhydrides
(continued )
Copyright 2005 by Marcel Dekker. All Rights Reserved.
C. Polyimides from Dianhydrides and Diisocyanates
Although the reaction of anhydrides with isocyanates to give imides was very early
reported [103], it was not until about 1970 that the reaction found application in the
synthesis of polyimides and copolyimides [104–106] (Scheme 11).
ð11Þ

Table 7 Continued.
Diamines Dianhydrides
Copyright 2005 by Marcel Dekker. All Rights Reserved.
The reaction takes different pathways depending on the conditions. In the absence of
catalyst the reaction has been claimed to proceed through a seven-membered polycycle
intermediate (Scheme 12) that finally gives rise to polyimide with separation of carbon
dioxide.
ð12Þ
Spectroscopic evidence of the seven-membered rings has been found in the
preparation of polyimides from pyromellitic dianhydride and methylenediphenyl-
diisocyanate (MDI) [105]. The reaction is conducted in solution of aprotic solvents,
with reagents addition at low temperature and a maximum reaction temperature of about
130

C. On the other hand, polyimides of very high molecular weight have not been
reported by this method. The mechanism is different when the reaction is accele rated by
the action of catalysts. Catalytic quantities of water or alcohols facilitate imide formation,
and intermediate ureas and carbam ates seem to be formed, which then react with
anhydrides to yield polyimides [106]. Water as catalyst has been used to exemplify the
mechanism of reaction of phthalic anhydride and phenyl isocyanates, with the conclusion
that the addition of water, until a molecular equivalent, markedly increases the forma-
tion of phthalimide [107] (Scheme 13). The first step is actually the hydrolysis of the
isocyanates, and it has been claimed that ureas are present in high concentration during
the intermediate steps of the reaction [107]. Other conventional catalysts have been widely
used to accelerate this reaction. Thus, tertiary amines, alkali metal alcoholates, metal
lactames, and even mercury organic salts have been attempted [108].
ð13Þ
For the preparation of polyimides, conventional dianhydrides have been combined
with aliphatic and aromatic diisocyanates which are well known in the chemistry of
polyurethanes. Other more modern diisocyanates have been also studied [109,110].

Copyright 2005 by Marcel Dekker. All Rights Reserved.
Diisocyanates containing an aliphatic sequence with phenylisocyanate end groups [111],
and diisocyanates containing preformed imide rings [112], have been recently synthesized
and used as monomers against aromatic dianhydrides.
Another approach to polyimides from diisocyanates is based on the reaction of
isocyanates with half esters. The isocyanate group readily reacts with the carboxy group
in solution, without catalyst under mild conditions, to yield amic ester with splitting off of
carbon dioxide (Scheme 14).
ð14Þ
In this way, polyimides have been reported from diisocyanates and half diesters via
soluble poly(amic ester)s, which were converted into the final cyclized polyimides with loss
of alcohol by the classical imidization method at high temperature [113,114].
A related reaction is the condensation of anhydrides with cyanates to imide
carbamates (Scheme 15).
ð15Þ
This reaction has been used to synthesize polyimides (more properly polyimide
carbamates) from dianhydrides and dicyanates [23]. The reaction proceeds in nitrobenzene
at high temperature, catalysed by triethylamine. The thermal resistance of these polymers
is much lower than that of pure aromatic polyimides, and, therefore, the reaction has not
found practical application.
D. Other Methods to Condensation Polyimides
1. From Diimides
Diimides of tetracarboxylic acids can be used for the synthesis of polyimides. Several
reactions have been used:
(a) Polycondensation of Diimides with Dihalides (Scheme 16).
ð16Þ
Copyright 2005 by Marcel Dekker. All Rights Reserved.
This reaction is accompanied by the separation of hydrogen halide, so that an acid
acceptor is needed to catalize the reaction, which is carried out in polar solvents at high
temperature [115]. Aromatic dihalides are not suitable reactants for this reaction, unless

activated dihaloaromatic monomers are used [116].
(b) Aminolysis of Diimides by Diamines. Ammonia and amines can readily react
with cyclic imides to yield ortho-diamides by nucleophilic attack and subsequent ring
opening. On the basis of this old reaction, polyimides have been synthesized from aromatic
diimides and diamines. The reaction has been classified as a migrational polymerization
[117]. It proceeds in solution through a lineal poly(ortho-diam ide), and this intermediate is
converted to the polyimide by heating, with evolution of ammonia, in a similar fashion to
the conversion of poly(amic acid)s into polyimides (Scheme 17).
ð17Þ
(c) Transimidization. Another possibility is the reaction of diamines with N,N
0
-
disubstituted diimides, by a synthetic route that can be considered as a transimidization,
with evolution of monoamine from the intermediate poly(amic amide). In this exchange
reaction, the nature of R plays an important role as the residue R–NH
2
has to be
eliminated to accomplish ring closing, so that short-length R substituents are, in princip le,
desirable for this approach (Scheme 18). No netheless, the chemistry involved in these
reactions has been studied also for the case when R is rather long and constituted by
aliphatic aminoacid moieties [118,119].
The global reaction is an equilibrium that moves to the right at relatively high
temperature, and it is necessary to have diamine monomers which are more nucleophilic
than the monoamine, unless specific catalysts as transition metal salts are employed
[120,121]. An alternative method that uses N,N
0
-bis-pyridyl- or N,N
0
-bis-pyrimidyl
bisphthalimides as monomers has been developed. By this transimidation approach,

2-aminopyridine or 2-aminopyrimidine are readily displaced by diamines to yield high
molecular weight polyimides [122].
ð18Þ
Copyright 2005 by Marcel Dekker. All Rights Reserved.
Polypyromellitimides have been prepared by condensation in solution of NMP
from aliphatic diamines and N, N
0
-dialkyloxycarbonyl pyromellitimides. The reaction
can be carried out by interfacial polycondensation, as illustrated in Scheme (19)
[123,124].
ð19Þ
The same reaction has been studied wi th aromatic and aliphatic diamines, and the
conclusion has been drawn that the procedure is only valid for aliphatic diamines, because
the low basicity of aromatic diamines does not allow for polymer formation in mild
conditions.
(d) From Diimides it is also Possible to Attain Polyimides by Reaction with
Diisocyanates (Scheme 20).
ð20Þ
However, the molecular weights reported for polyimides prepared by this procedure
are comparatively low (
inh
0.2–0.3 dL/g) [125,126]. Moreover, the presence of functional
grouping consisting of imide plus ureyl linkages makes these polymers thermally unstable.
This is also the case of polyimides synthesized from dihydroxyimides and diacyl chlorides
[127] or diisocyanates [128] (Scheme 21). The combination in a functional grouping
of imide and carbamates also makes these polyimides unstable and easily attackable
Copyright 2005 by Marcel Dekker. All Rights Reserved.
by nucleophiles.
ð21Þ
Another approach is the reaction of diimides with divinyl monomers. Examples of

this route, starting from divilylsulfone and pyromellitic, benzophenonetetracarboxylic
and cyclopentanetetracarboxylic diimides have been reported (Scheme 22). The polymer-
izations are carried out in solution in the presence of inhibitors of radical polymerization,
and the molecular weights achieved are not very high [129]. By a similar mechanism,
polyimides have been prepared from diallylesters and cycloaliphatic [130] and aromatic
diimides [131].
ð22Þ
2. From Silylated Diamines
The silylation method has received particular attention during last years for the
preparation of a variety of condensation monomers and polymers [132–134]. It has
been successfully applied to the synthesis of aromatic polyimides, and can be considered as
a recommendable method in some instances, particularly for less reactive diamines because
it has been proved that silylated aromatic diamines are more nucleophilic than free
diamines [135,136]. By this method, a poly(amic trimethyl silyl ester) is produced in the
first step (Scheme 23), which can be converted into polyimide by chemical means
[91,93,137].
ð23Þ
Copyright 2005 by Marcel Dekker. All Rights Reserved.