Вы находитесь на странице: 1из 23

359

Macromol. Rapid Commun. 2003, 24, 359381

Feature Article: The theory of step-growth polymerizations including the cascade theory is discussed in the light of
new results focussing on the role of cyclization reactions. The
identification of cyclic oligomers and polymers in reaction
products of step-growth polymerizations has been eased
considerably by means of MALDI-TOF mass spectrometry.
Experimental examples concern syntheses of polyesters,
polycarbonates, polyamides, polyimides, poly(ether sulfone)s, poly(ether ketone)s and polyurethanes. It was found
in all cases that the percentage and molecular weight of the
cycles increases when the reaction conditions favor high
molecular weights. In the absence of side reactions all reaction products will be cycles when conversion approaches
100%. Cyclization may even take place in the nematic phase
but even-numbered cycles are favored over odd-numbered
ones due to electronic interactions between mesogens
aligned in parallel. In contrast to Florys cascade theory,
cyclization also plays a decisive role in polycondensations of
abn-type monomers, and at 100% conversion all hyperbranched polymers have a cyclic core. Furthermore, it is
demonstrated that in a2b3 polycondensations intensive
cyclization in the early stages of the process has the consequence that either no gelation occurs or the resulting net-

works consist of cyclic and bicyclic oligomers as building


blocks. Finally, a comparison between cyclization of synthetic polymers and biopolymers is discussed.

Schematic representation of a network structure mainly


consisting of cyclic oligomers and multicyclic building
blocks as derived from a2 b3 polycondensation.

Cyclic Polymers by Kinetically Controlled


Step-Growth Polymerization
Hans R. Kricheldorf,* Gert Schwarz
Institut fur Technische und Makromolekulare Chemie, Bundesstr. 45, 20146 Hamburg, Germany
Fax: 49-40-42838 6008; E-mail: kricheld@chemie.uni-hamburg.de

Keywords: cyclic polymers; MALDI-TOF; polycondensation; polyaddition

1 Introduction
The classical theory of step-growth polymerizations as it is
present in all textbooks is mainly based on the work of
Carothers[1,2] and Flory.[3,4] The work of both authors is
one of the greatest milestones in the history of polymer
science for three reasons. Firstly, they demonstrated that the
reactivity of end groups does not depend on the chain length
of oligomers and polymers (dimers and trimers may deviate
from this rule in individual cases). This fundamental aspect
needs to be emphasized because at the time when Carothers
started his experimental studies, around 1927, Staudinger[5]
and other polymer chemists[6] believed that the reactivity of
end groups decreases with higher chain length. Secondly,
the inauguration of polycondensation chemistry opened a
field of polymer syntheses which may be considered as an
equivalent to in-vivo syntheses of biopolymers. All syn-

theses of biopolymers in living organisms involve


condensation steps, although the kinetic course is different from that of typical step-growth polymerizations.
Thirdly, the first man-made fibers were produced
(on the basis of Nylon 6,6) and this event was the starting
point for the invention and production of all kinds of textile
fibers.
Despite its merits, the classical theory of step-growth
polymerizations has two shortcomings. Firstly, it does not
differentiate between kinetically controlled (KCP) and
thermodynamically controlled (TCP) step-growth polymerizations, and secondly, it does not consider a significant
contribution of cyclization reactions. Kinetic control means
that equilibration reactions are absent and the reaction
mixtures do not represent the thermodynamic optimum of
the system under investigation. Thermodynamic control
means that equilibration reactions including ring-chain and

Macromol. Rapid Commun. 2003, 24, No. 5/6 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim, 2003

1022-1336/2003/05-0604359$17.50.50/0

360

H. R. Kricheldorf, G. Schwarz

ring-ring equilibria are present throughout the polymerization process, so that reaction mixtures represent the
thermodynamic optimum at any stage of polycondensation.
As explained by Carothers[1] and Flory[3] in detail, a
significant contribution of cyclization reactions was not
taken into account, because syntheses of macrocycles containing more than 20 ring atoms had proven extremely
difficult before 1940. The formation of cycles in TCPs has
been studied intensively by Stockmayer[2,8] and other
authors[9] for the past 50 years, comparable studies of KCPs
however were not conducted. The development of MALDITOF mass spectrometry (MS) during the past decade has
provided a new powerful tool for qualitative analyses of
complex reaction mixtures. In this context, the present work
focuses on the role of cyclization in kinetically controlled
step-growth polymerization.

2 Theory of Kinetically Controlled


Polycondensations
The classical theory of polycondensation is based on the
chain-growth of difunctional linear monomers, oligomers
and polymers, whereby all reactive species can react with
each other at any time. In contrast to polyaddition which
obeys the same kinetic scheme, polycondensation is also
characterized by the elimination of a small molecule, such
as water, HCl or NaCl, in any growing step. This difference
has a strong influence on the thermodynamic properties of
these step-growth polymerizations but not on the kinetic
course. In both cases conversion is defined as the molar ratio
of reacted functional groups versus the sum of initially
existing functional groups, and it is not defined by consumption of monomers in contrast to chain-growth polymerizations. Chain growth expressed by the average degree
of polymerization (DP) is connected to conversion ( p) by

Equation (1), the so-called Carothers equation


DP

1
1p

where p is the conversion of functional groups. The


graphical version of this equation is presented as curve A
in Figure 1. Frequency distribution of kinetically controlled
step-growth polymerizations obeys Equation (2a). The
corresponding mass distribution is obtained by multiplication of the frequencies by the molar mass of chains with
individual DP (Equation (2b))
ff pDP1 1  p

2a

Wf mDP  pDP1 1  p

2b

where ff is the frequency of oligomers or polymers having


identical degrees of polymerization (DP), mf is the mass
fraction of polymers having identical degrees of polymerization, and mDP is the mass of a polymer chain with
individual DP.
The graphic illustration of both distribution functions is
presented in Figure 2 and 3.
The influence of cyclization reactions on the course of
kinetically controlled step-growth polymerizations was
calculated by Stepto and coworkers[10,11] and by Gordon
and Temple.[12] They assumed that cyclization can compete
with propagation even at high conversion and in concentrated solution (Figure 4). Although not clearly indicated in
Figure 4, it may be concluded that all reaction products of
an ideal KCP (no side reactions, perfect stoichiometry)
should be cycles at 100% conversion. Another mathematical approach was published more recently by Mandolini
and co-workers.[13] However, this approach is limited to
cyclic oligomers up to DP of 12 and to dilute solutions. All
three mathematical treatments of KCPs indicated that

Hans R. Kricheldorf was born in Freiburg i.Br. (Germany) in 1942. He studied chemistry at the
University of Freiburg i.Br., obtained his Ph.D. in 1969 and finished his habilitation in 1975. He was
appointed assistant professor in 1976 and associate professor in 1980 at the University of Freiburg i.Br.
In 1981 he was appointed full professor at the Institute of Technical and Macromolecular Chemistry at
the University of Hamburg. His research fields are ring-opening polymerization, focusing on the
synthesis of biodegradable polymers, and polycondensation with a focus on new synthetic methods,
cyclic polymers and liquid crystalline materials.
Gert Schwarz was born in Herbolzheim (Germany) in 1947. He studied chemistry at the University of
Freiburg i.Br. and obtained his Ph.D. under the supervision of Prof. Kricheldorf in 1981. In 1982,
together with Prof. Kricheldorf, he joined the faculty of chemistry of the University of Hamburg, where
he holds the position of an assistant professor. His research interests concern the synthesis and
characterization of polycondensates with focus on cyclic polymers and liquid crystalline polymers.

Cyclic Polymers by Kinetically Controlled Step-Growth Polymerization

Figure 3. Mass distribution in the classical theory of step growth


polymerizations calculated for three different conversions (Equation (2b)).

Figure 1. Chain growth/conversion curves as calculated from


Equation (4) with X 1.2 and various ratios Vp/Vc: (A) 1, (B) 20,
(C) 15, (D) 5, (E) 0.1, and (F) 0.

dilution of the monomers favors cyclization at low and


medium conversions (Figure 4) in agreement with the
Ruggli-Ziegler dilution principle.[14]
Our recent results presented below confirm the main
conclusions presented by Stepto et al.[10,11] and Gordon

Figure 2. Frequency distribution in the classical theory of stepgrowth polymerization[3] calculated for three different conversions (Equation (2a)).

et al.[12] without supporting all aspects of their calculations.


For instance, even KCPs conducted in bulk must yield
cyclic products at 100% conversion (not calculated and
presented in Figure 4). A major reason for this hypothesis is
the fact that all step-growth polymerizations are selfdiluting reactions. This means the molar concentration of

Figure 4. The mass fractions of cycles in kinetically controlled


step-growth polymerization as calculated in ref.[11,12] (with
increasing concentration from A to C).

361

362

H. R. Kricheldorf, G. Schwarz

all active linear species decreases with conversion according to Equation (3)
Lap Lao  1  p

where La is the reactive linear species (incl. monomers).


Therefore, all KCPs proceed under extreme dilution when
conversion approaches 100%, regardless of the initial
monomer concentration. An important consequence of this
fact is a high probability of cyclization, even for long
chains. Another important consequence is that cyclization
limits chain growth. As expressed by Equation (4)[15]
DP

1p 1

1
a

with a Vp =Vc

where Vp and Vc are the rates of propagation and


cyclization, X is a constant > 1.0, allowing the adaptation
of this equation to different concentrations, and as illustrated in Figure 1, the rate of propagation versus the rate of
cyclization is decisive for the maximum DP, KCP can
achieve under ideal conditions (curve BE in Figure 1). The
unlimited chain growth predicted by Equation (1) and curve
A in Figure 4 requires the absence of any cyclization (Vc 0
in Equation (4)).[15,16] Such an extreme case requires in turn
an extremely stiff chain with good solubility in inert solvents and in the absence of any side reactions. A polycondensation having this combination of properties is
difficult to realize. More realistic is the second extreme
included in Equation (4), namely rapid cyclization with no
propagation (Vp 0, curve F in Figure 1). Condensations
representing this scenario have been known for a long time
in organic chemistry. A typical example is the basecatalyzed dehydrohalogenation of b-chloro- or bromoalcohols yielding epoxides (Equation (5)). Although epoxides
are thermodynamically unstable, cyclization is favored
kinetically so much that almost no linear oligomers or
polymers are formed (dioxanes may result as byproducts).
5

The Vp/Vc ratio which, according to Equation (4), is


decisive for the extent of chain growth or cyclization
depends on the flexibility of oligomer and polymer
backbones and on the stereochemistry of the building
blocks. When growing chains can adopt a large number of
conformations unfavorable for cyclization, Vp/Vc will be
high and this situation is typical for flexible aliphatic chains.
However, aliphatic polyethers are more inclined to
cyclization than aliphatic polyesters or polyalkanes. In
alkane chains all-trans conformation is the energetically
most favorable situation, but this conformation is unfavorable for cyclization. In the case of polyethers, repulsive

electronic interactions between the O-electrons and sbonds in b-position favor gauche conformations, so that the
tendency to form loops and cycles is much higher. Incorporation of stiff aromatic monomers reduces the number of
conformations per unit chain length, and may thus reduce
the Vp/Vc ratio compared to alkane chains. The influence of
aromatic moieties depends very much on stereochemistry.
ortho- and meta-functional monomers favor cyclization
more than para-functional ones. However, the influence
of para-functional monomers depends very much on the
linking groups. When the bond angle is relatively small as in
the case of aromatic sulfides and sulfones, Vp/Vc is much
lower than in the case of aromatic polyamides or polyesters.
For para-linked aromatic polyamides and polyesters linear
chain conformations are energetically and stereochemically favored, and thus, cyclization tendency is extremely
low.[15,16]
In this connection, it should be mentioned that the Vp/Vc
ratio also depends on a statistical factor varying between 1
and 2. When monomers that contain two different functional groups (a-b monomers) are polymerized, all
oligomers and polymers will have one a and one b end
group, and thus, are in principle capable of cyclization.
However, a combination of a-a and b-b monomers generates three types of linear species (having a-b, a-a, or b-b
end-group combinations) two of which cannot cyclize.
Since the molar concentration of a-b chains equals the sum
of a-a and b-b concentrations, the Vp/Vc ratio is
reduced to 50% of the a-b case. A further increase of the
Vp/Vc ratio for statistical reasons may occur in interfacial
phosgenations of diphenols as will be discussed below.
The permanent competition of cyclization and propagation has, of course, consequences for the molecular weight
distribution (MWD) of the reaction products. Equation (2)
of the classical theory is not valid anymore. In the classical
theory the concentration of monomers and oligomers
continuously decreases with higher conversions and finally
tends toward zero. In contrast, the cyclic oligomers formed
in KCP are stable, approach a limiting value when all linear
oligomers are consumed, but their concentration never decreases. Furthermore, three different pairs of MWDs have
to be distinguished: (i) frequency and mass distribution of
cycles, (ii) frequency and mass distribution of linear
species, and (iii) distribution curves of the sum of cyclic
and linear chains
All these distribution curves systematically vary with
increasing conversion. A characteristic consequence of a
high extent of cyclization is the tendency towards bimodal
mass distribution combined with monomodal frequency
distribution. The first maximum in the mass distribution
results from cyclic oligomers. The second maximum may
contain cycles and/or linear chains. Figure 5 illustrates
the more or less developed bimodal character of the
mass distribution of three different classes of polymers
and three synthetic methods. Unfortunately, an adequate

363

Cyclic Polymers by Kinetically Controlled Step-Growth Polymerization

where r Na/Nb, with Na/Nb being the total numbers of


functional groups present initially, including those of ChT.
This equation is defined for either a polycondensation of
a-b or a-a b-b monomers with addition of a ChT
having two b groups. In the a-a b-b case, excess
b-b monomer has the same effect. In our new theory of
KCPs, Equation (6) has to be replaced by Equation (7).
Cyclization reactions keep DP systematically below the
[M]/[ChT] ratio, even at 100% conversion. When the
cyclization tendency is high (e.g., Vp/Vc  2), small
amounts of ChT have no influence on DP. When the Vp/
Vc ratio is high (e.g. > 100), the resulting DPs mainly
depend on the [M]/[ChT] ratio. Addition of a ChT reduces
the extent of cyclization, but it does not suppress cyclization
completely, unless its concentration approaches the monomer concentration. Yet, in this case the ChT also suppresses
chain growth.
DP

Figure 5. SEC elution curves: (A) polycarbonate prepared by


triethylamine-catalyzed interfacial polycondensation of bisphenol-A with diphosgene (M n  65 000 Da, M w  19 000 Da); (B)
poly(ether sulfone) prepared from silylated 4-tert-butylcatechol,
4,40 -difluoro-diphenylsulfone and K2CO3 in NMP at 120 8C (48 h;
M n  19 000 Da, M w  180 000 Da); (C) poly(ether ketone)
prepared from 4-tert-butylcatechol, 4,40 -difluoro-benzophenone
and K2CO3 in a refluxing sulfolane/xylene mixture (M n 
3 000 Da, M w  17 000 Da).

mathematical treatment of these distribution functions is


still missing.

3 The Role of Chain Terminators and Initiators


Chain terminators, ChTs, are frequently used in technical
polycondensation processes to control molecular weights.
In the classical theory of polycondensation DP approaches
the [M]/[ChT] ratio at 100% conversion (Equation (6))[3,4]
DP

1r
1
2r1  p 1  r

1r



1
1
1r
2r 1  p 1  a
X

This line of argumentation is, of course, only valid for


polycondensations obeying step-growth kinetics. However,
a different scenario results from addition of a highly
reactive ChT to an a-b monomer, when the reactive chain
end of the initiated oligomers is significantly more
reactive than the a or b group of the monomer. In this
case, the a-b monomers will preferentially or exclusively
react with the ChT-initiated growing chain and not with
each other. In this case no cyclization can take place,
because the ChT end group is a dead chain end. However,
such polymerization obeys chain-growth kinetics, and is
thus not included in the classical definition of a polycondensation. This point needs to be emphasized, because
such polymerizations have recently been called chaingrowth polycondensations. A better terminology would
be chain-growth condensation polymerization. Several
interesting examples of such chain-growth condensation
polymerizations have recently been published by Yokozawa and co-workers.[1720]

4 Polyesters
Technical syntheses of polyesters, such as poly(ethylene
terephthalate) and poly(butylene terephthalate), involve
equilibration reactions, and thus do not obey the definition
of a KCP. However, several polyester syntheses used for
research purposes or for the preparation of biodegradable
materials and the syntheses of polycarbonates by phosgenation or via bisphenol bischloroformates (see below) have
a kinetically controlled course. First extensive studies focussing on the formation of cyclic polymers were performed
with pyridine-promoted polycondensations of diphenols or
diols with aliphatic dicarboxylic acid dichlorides (ADADs)

364

H. R. Kricheldorf, G. Schwarz

(Equation (8)) or aromatic dicarboxylic acid dichlorides.[16] This mild method excludes any interference of
transesterification reactions. The synthesis of simple esters
from monofunctional alcohols or phenols and carboxylic
acid chlorides with pyridine as catalyst and HCl acceptor
has been known for more than a hundred years in organic
chemistry. However, syntheses of polyesters via this method
have rarely been used, mainly because of the prejudice that
high molar masses cannot be obtained. However, with
careful optimization of the reaction conditions numberaverage molecular weights (M ns) around 30 00035 000
can be achieved.[16] This optimization process requires low
temperatures (around 0 8C) during the addition of pyridine
to the monomer mixture, a large excess of (e.g. 100%) of
pyridine, long reaction times (studied up to 7 d) and ideal
stoichiometry. Interestingly, ideal stoichiometry is not
necessarily identical with a 1:1 feed ratio of both monomers. A slight excess of the dicarboxylic acid dichlorides
(in most cases around 1 mol-%) proved to be necessary for
the optimization of molecular weights.

prepared in bulk is, in principle, the same as that in concentrated solutions. For the following reasons this conclusion is rather trivial, although it has been questioned by
numerous discussion partners of the author. Firstly, the
molar concentration of neat monomers, which is decisive
for the Vp/Vc ratio, may vary over a broad range. For
instance, the molar concentration of glycolic acid is around
13 mol/L and that of lactic acid around 11 mol/L. A mixture
of bisphenol-A and 4,40 -dichlorodiphenylsulfone (as is
used for the technical production of poly(ether sulfone)s
(PES)) has a molar concentration of around 2.5 mol/L.
Hence, dilution by a factor of 5 is necessary to bring the
concentration of glycolic acid down to the level of neat
sulfone monomers. Secondly, the molar concentration of
all active species rapidly decreases with higher conversion
(Equation (3)), so that slight differences of the initial
monomer concentrations do not play any role for the main
course of the KCPs. Several more polycondensation
methods are known allowing the syntheses of polyesters
in a kinetically controlled way, but those methods have not
been studied yet with respect to the formation of cyclic
polymers.

8
If small amounts of dicarboxylic acid dichlorides were
lost by reactions with OH groups on the glass walls of the
equipment, by hydrolysis with traces of moisture, or by
b-elimination of HCl (Equation (9)) has not been
elucidated. Yet, because of the b-elimination of HCl
(Equation (9)), tertiary amines that are more basic than
pyridine are unfavorable as catalyst and HCl acceptors. In
the case of polyesters derived from ADADs, the high
sensitivity to hydrolytic cleavage also needs to be considered in as much as the cleavage of a single ester bond
suffices to transform a cycle into a linear chain.
9
In summary, all the numerous polyester syntheses
performed by the author via the pyridine method had
in common that the molar fraction of cycles and their molar
masses increased when the reaction conditions favored
higher molar masses of the entire sample. In other words,
any optimization favoring chain growth favored the extent
of cyclization in agreement with the new theory of KCPs. In
the case of catechol, the pyridine method was compared
with polycondensations of the bis(trimethylsilyl) derivative
which were conducted in bulk (Equation (10)).[21] Nearly
identical molar masses and MALDI-TOF mass spectra
were obtained. The MALDI-TOF mass spectra proved the
cyclic polyesters to be the most abundant reaction products
up to the limit of these measurements at 6 000 Da. These
results demonstrated that cyclization tendency of polyesters

10

5 Polycarbonates
In 1980 Horbach et al.[22,23] reported that polycarbonates
prepared by interfacial hydrolytic polycondensation of
bisphenol-A bischloroformate may contain high fractions
of cyclic polycarbonates (Equation (11)). This was the first
report on the formation of cyclic polymers in KCP.
However, the postulation of cyclic polycarbonates was
exclusively based on end-group analyses, because at that
time no analytical method existed allowing direct identification. Therefore, neither concrete information on ring
size and quantity of the cyclic polycarbonates nor a clearcut relationship between cyclization and reaction conditions was disclosed, only triethylamine was mentioned as
useful catalyst. In a recent reinvestigation[24] all important
parameters, such as temperature, time, NaOH feed ratio and
catalyst concentration, were varied to optimize the
molecular weight. MALDI-TOF mass spectra revealed
higher fractions of cyclic polycarbonates when average
molecular weights of the products increased. In the
optimum case M n of 65 000 Da and a M w of 300 000 Da
was obtained (SEC, triple detection). The MALDI-TOF
mass spectrum of this sample displayed peaks of cyclic
oligo- and polycarbonates up to 18 000 Da and no peaks for

365

Cyclic Polymers by Kinetically Controlled Step-Growth Polymerization

Figure 6. MALDI-TOF mass spectrum of a fraction of a polycarbonate prepared by


triethylamine-catalyzed hydrolytic polycondensation of bisphenol-A bischloroformate.

linear chains. Fractionation by SEC gave 20 fractions, 14 of


which could be analyzed by means of MALDI-TOF MS.
The optimum mass spectrum obtained from these mass
spectrometric studies is presented in Figure 6. No linear
species was detected, whereas the peaks of cycles were
observable up to 55 000 Da corresponding to DP around
210. This result is important for three reasons. Firstly, it
demonstrates the conclusions of Horbach et al.[22,23] to be
correct in principle. Secondly, it proves that even long
polymer chains easily cyclize in a clean KCP. Thereby,
these results definitely prove our new theory of KCPs,
which postulates that cyclization competes with propaga-

tion at any stage (i.e. at any chain length). Thirdly, the


identification of cycles having more than 2000 atoms in the
ring allows for an interesting comparison with cyclic
biopolymers (see last section).

11

366

H. R. Kricheldorf, G. Schwarz

The formation of high-molar-mass polycarbonates


including large cycles in the hydrolytic polycondensation
of bisphenol-A bischloroformate is not a priori trivial
because it demonstrates the almost total absence of side
reactions. It also means that propagation and cyclization
are faster by a factor of at least 104 than the hydrolysis of the
chloroformate groups (Equation (12)), otherwise linear
chains having two OH end groups should be detectable.
Almost all reports on syntheses and properties of cyclic
oligo- and polycarbonates are based on polycondensations
of bisphenol bischloroformates suggesting a special mechanism favoring ring-closing reactions. However, the new
theory of KCPs suggests that cyclic oligo- and polycarbonates should also be formed in the direct phosgenation of
bisphenols. A recent study[25] of this problem revealed that
triethylamine and quarternary ammonium salts are particularly suited to catalyze chain growth and cyclization in
the interfacial polycondensation of bisphenol-A with
diphosgene. It was found that small amounts of the catalyst
(25 mol-% relative to bisphenol-A) are optimal in the case
of triethylamine, but large amounts ( 50 mol-%) are
requested in the case of phase-transfer catalysts. Under
optimum conditions, M ws up to 106 Da were obtained.
Regardless of the catalyst, the fraction of cycles increased
with increasing average molecular weights.
12

In order to reduce molar masses and keep the content of


cycles on a high level (> 90 mol-%), the interfacial
polycondensation of bisphenol-A with both diphosgene
and triphosgene was performed by using the pseudo-high
dilution method.[26,27] It was found again that low feed
ratios of triethylamine and high feed ratios of phase-transfer
catalysts favored a high extent of cyclization. Furthermore,
triphosgene yielded slightly higher fractions of cycles
than diphosgene with triethylamine as catalyst, probably
due to a lower sensitivity to hydrolysis. Figure 7 displays
the MALDI-TOF mass spectrum of polycarbonate with
M n 4 500 Da and M w 36 000 Da prepared by means
of triphosgene. No linear species are detectable up to
24 000 Da. This spectrum and the one obtained after
fractionation by means of SEC indicates that this sample
contains more than 95 mol-% of cycles. Another interesting
result concerns the use of phase-transfer catalysts. When
diphosgene was used, TEBA chloride gave much higher
yields of cycles than Ph4PCl, whereas with triphosgene the
reverse trend was observed.[26,27] Finally, a tremendous
influence of the temperature on molecular weights was
found. Variation of the initial temperature from 5 to
40 8C may change the molecular weights by a factor of 10.
In summary, the direct phosgenation of bisphenol-A may
give high yields of cyclic polycarbonates and allows a
systematic variation of the average molecular weights
in perfect agreement with our recent theory of KCPs.
A typical bimodal mass distribution resulting from

Figure 7. MALDI-TOF mass spectrum of a polycarbonate prepared by triethylaminecatalyzed phosgenation of bisphenol-A with triphosgene via the pseudo-high-dilution
method.

Cyclic Polymers by Kinetically Controlled Step-Growth Polymerization

triethylamine-catalyzed polycondensation of bisphenol-A


with diphosgene is presented in Figure 5A.

6 Poly(ether sulfone)s and Poly(ether ketone)s


PESs are like polycarbonates a group of engineering plastics, technically produced via KCPs. The standard procedure consists of the polycondensation of a bisphenol with
4,40 -dichlorodiphenylsulfone catalyzed by K2CO3 in
DMSO (analogous to Scheme 1). Commercial PES have
M ns in the range of 15 000 to 25 000 Da. MALDI-TOF MS
revealed a moderate content of cyclic PES in commercial
PES, increasing parallel to the molecular weights.[28] Furthermore, the mass peaks of OH- and Cl-terminated linear
chains were detectable (Figure 8 and Scheme 1) indicating
that conversion was far from completion. When PESs were
prepared under standard conditions from bisphenol-A and
the more reactive 4,40 -difluorodiphenylsulfone (DFDPS)
(Scheme 1), optimum stoichiometry was found to require
an excess of DFDPS of around 1 mol-%.[29] The sample with
the highest molecular weight had an M n around 45 000 Da
and an M w around 140 000 Da. The MALDI-TOF mass
spectrum displayed mass peaks of cycles up to 13 000 Da,
and no linear species was detectable. Polycondensation of
silylated bisphenol-A with DFDPS gave almost identical
results[28] after optimization of time and stoichiometry.
Furthermore, polycondensations of silylated 4-tertbutylcatechol with DFDPS were studied in NMP at 120
or 140 8C[30] (Scheme 2). Variation of the stoichiometry
proved again that an excess of 1 mol-% of DFDPS is advan-

Scheme 1.

Figure 8. MALDI-TOF mass spectrum of a commercial poly(ether sulfone) (Aldrich


Catalog No. 37,420-6). La, Lb and Lc represent the linear chains formulated in Scheme 1.

367

368

H. R. Kricheldorf, G. Schwarz

Scheme 2.

tageous for maximum molecular weight. Regardless of the


variation of stoichiometry or time (conversion), the fraction
of cycles increased with higher molecular weights. When
the optimized sample was fractionated by means of SEC,
cycles (and nothing else) were detected up to masses around
27 000 Da.[30] At incomplete conversion, all linear species
corresponding to the La, Lb and Lc chains in Scheme 1
were detectable by MALDI-TOF MS. Cyclic PESs of low
average molecular weight were also obtained by polycondensation of free 4-tert-butylcatechol with DFDPS in
sulfolane[31] (Scheme 2). A typical bimodal mass distribution resulting from polycondensations of silylated 4-tertbutylcatechol is illustrated in Figure 5B.
In the case of poly(ether sulfone)s derived from 4-tertbutylcatechol, cyclic and predominantly linear samples of
similar molecular weight were mixed and the MALDI-TOF
mass spectra were recorded with variation of matrix, dopant
and solvent. These 1:1 mixtures contained more cycles than
linear species below 3 000 Da because the linear samples
contained small fractions of cycles (Figure 9A). As
exemplarily illustrated in Figure 9B, the linear chains
flew better than the cycles, regardless of the conditions
used for the preparation of the irradiation targets. Similar
results were obtained from poly(ether ketone)s. In other
words, our MALDI-TOF mass spectra of poly(ether
sulfone)s and poly(ether ketone)s have the tendency to
underestimate the content of cycles.

Figure 9. MALDI-TOF mass spectra of: (A) poly(ether sulfone)


prepared with a 10 mol-% excess of silylated 4-tert-butylcatechol
in NMP at 145 8C/24 h; (B) equimolar mixture of the linear
poly(ether sulfone) and a cyclic poly(ether sulfone) having almost
identical average molecular weights.

When 4-tert-butylcatechol was polycondensed with 4,40 difluorobenzophenone in DMSO, sulfolane or NMP (containing toluene or xylene for the azeotropic removal of
water) the poly(ether ketone)s (PEKs; 13) with relatively
low molecular weights were obtained (Table 1 and Experimental Part). MALDI-TOF MS revealed that intensive
cyclization was the limiting factor for chain growth. Particularly clean polycondensation and cyclization reactions
were found in sulfolane. Similar molecular weights together with a high extent of cyclization were observed, when
the silylated diphenol was polycondensed in NMP (Table 1).
Somewhat higher molecular weights but analogous trends
were found when the PEK (14) was synthesized from
bisphenol-A or silylated bisphenol-A (these results will be
published elsewhere). Also these polycondensations
yielded bimodal mass distributions (Figure 5C).

13

369

Cyclic Polymers by Kinetically Controlled Step-Growth Polymerization

Table 1. Syntheses of poly(ether ketone)s from 4,40 -difluorobenzophenone and 4-tert-butylcatechol (TBC) or its bis(trimethylsilyl)
derivatives.
Exp. no.

1
2
3
4
5
6b)
7
8b)
a)
b)

Diphenol

4-TBC
4-TBC
4-TBC
4-TBC
4-TBC
4-TBC
silylated 4-TBC
silylated 4-TBCb)

Reaction medium

DMSO toluene
DMSO xylene
sulfolane toluene
sulfolane toluene
sulfolane xylene
NMP toluene
NMP
NMP

Temp.

Time

Yield

Zinha)

8C

dL/g

130
150
150
150
170
140
160
160

6
6
6
6
6
6
48
48

96
97
98
97
98
87
90
98

0.08
0.11
0.08
0.08
0.13
0.08
0.15
0.12

Measured at 20 8C with c 2 g/L in CH2Cl2.


1 mol-% excess of 4,40 -difluorobenzophenone was used.

14

15

7 Polyamides
Technical syntheses of aliphatic polyamides, such as Nylon
6, Nylon 12 or Nylon 6.6 are usually conducted in the melt
at temperatures above 220 8C and involve equilibration
reactions. However, syntheses of aromatic polyamides
from free diamines and dicarboxylic acid dichlorides are
usually performed in NMP at temperatures  0 8C and obey
the definition of KCP. Unfortunately, several attempts to
measure MALDI-TOF mass spectra of poly(p-phenylene
terephthalamide) failed. In the case of more soluble analogs
derived from substituted terephthalic acids (15a,b),[32]
acceptable mass spectra were obtained after extraction with
chloroform/trifluoroethanol mixtures. As expected for
these rigid-rod-type polymers, no cycles were detected in
the extracts.[15] In other words, syntheses of polyamides
from para-functional monomers at low temperature obey
largely the classical theory of polycondensation (Equation
(1)). Furthermore, silylated aromatic diamines (e.g., mphenylene diamine, Equation (16)) were polycondensed
with ADADs. ADADs are significantly more reactive than
aromatic acid chlorides, so that higher conversions can be
achieved. The low basicity of the aromatic diamines (when
compared to aliphatic ones) and silylation which makes
basic HCl acceptors obsolete, prevented side reactions such
as b-elimination of HCl (Equation (9)). Therefore, high
molecular weights were obtained (M n up to 35 000 Da) and
MALDI-TOF mass spectra showed peaks of cycles as the
predominant reaction products up to 13 000 Da (technical
limit of these measurements).[15]

16
A large number of experiments was studied dealing
with syntheses of aliphatic polyamides at temperatures
 10 8C.[15] When silylated 1,12-diaminododecane was
polycondensed with ADADs (Equation (17)), moderate
molecular weights were obtained, and mass spectra displayed significant amounts of linear chains resulting from side
reactions such as b-elimination of HCl (Equation (9)).
Interfacial polycondensations of 1,12-diaminododecane
(Equation (18)) or of the diamine (19) gave again moderate
molecular weights. With a 1.0:1.0 feed ratio the resulting
polyamides only contained a small fraction of cycles and
the prevailing linear chains were terminated by two NH2
groups. The predominance of these chains is a normal
consequence of the hydrolysis of acid chloride groups,
which is an unavoidable side reaction of the interfacial
method. When ADADs are used in excess (26 mol-%),
the number of NH2-terminated chains tremendously
decreases, whereas the fraction of cyclic polyamides
strongly increases, so that the cycles were prevailing in
the MALDI-TOF mass spectra up to 5 000 Da. However, it
is clear that side reactions of acid chlorides are unavoidable
in interfacial polycondensations, so that polyamides
containing > 50 mol-% of cycles cannot be prepared in
this way.

370

H. R. Kricheldorf, G. Schwarz

17

densations of the bis(4-chlorophenyl)ester (21). Obviously,


the neutral or weakly basic character of the liberated
imidazole is an advantage at high temperature. In summary,
polycondensations of aliphatic diamines with activated
esters of dicarboxylic acid may yield high-molar-mass
polyamides containing high fractions of cycles, but careful
optimization of the reaction conditions is necessary.[34]

18
22
19
The third approach studied in detail was based on
polycondensations of free diamines (19) and (20) with the
bis(4-chlorophenyl)ester of sebacic acid (21).[33] Solvent,
temperature, time and the concentration of water were
varied and optimized. The comparison of N,N-dimethylformamide, NMP, sulfolane and DMSO revealed that
DMSO was the best reaction medium. Small quantities of
water ( 1 vol.-%) had no negative influence. Surprisingly,
the temperature played a key role for success. At 20 8C
polycondensation was too slow for high conversions within
one week. At 100 8C no polyamides were obtained due to
unknown side reactions. Yet, high-molar-mass polyamides
were isolated, when tertiary amines were added at 100 8C.
This finding suggests that the acidity of the liberated
4-chlorophenol is decisive for the side reactions. The highest
molecular weights and the highest extent of cyclization was
achieved in DMSO at 60 8C. The polyamide of diamine (20)
prepared under these optimum conditions did not show any
linear chain in the MALDI-TOF mass spectra, whereas the
mass peaks of cycles were observable up to 13 000 Da.

23

8 Polyimides
Six polyimides soluble in chloroform (or dichloromethane)
and in some cases soluble in tetrahydrofuran were prepared
from dianhydrides (24a,b) in combination with diamines
(25) and (26a,b).[35] Four different synthetic methods were
compared: (i) polycondensation in boiling m-cresol with
azeotropic removal of water, (ii) polycondensation in 1,2dichlorobenzene (with azeotropic removal of water) followed by polycondensation in the melt, (iii) polycondensation in NMP with acetic anhydride as condensing agent, and
(iv) polycondensation in NMP with triphenylphosphite and
pyridine as condensing agent (Higashi method).

20

24

21
25
Despite higher reactivity, polycondensations of the
bis(N-hydroxysuccinimide)ester (22)[34] gave considerably
poorer results than those of the bis(4-chlorophenyl)ester.[33]
The side reactions responsible for this failure were not
elucidated. Better results were obtained with the sebacic
acid bisimidazolide (23).[34] Polycondensation with the
diamine (20) in DMSO at 60 8C proved again to be close to
the optimum but both molecular weight and extent of
cyclization were slightly lower than the best results obtained with the bis(4-chlorophenyl)ester (21). Two characteristic differences are worth to be mentioned. Firstly, the
bisimidazolide is more sensitive to moisture and, thus,
azeotropic drying of the reaction medium is advantageous.
Secondly, polycondensations at 100 8C were successful
(although inferior to those at 60 8C) in contrast to polycon-

26
MALDI-TOF MS proved the formation of cycles in all
experiments. Highest molecular weights were obtained in
m-cresol, and in these cases only cyclic polyimides were
detectable by means of MS (up to 10 000 Da).

9 Polyurethanes
More than thirty years ago, Stepto and co-workers studied
polyadditions of 1,6-diisocyanatohexane and mixtures of

371

Cyclic Polymers by Kinetically Controlled Step-Growth Polymerization

tetra- and penta(ethylene glycol) in benzene.[10] They


concluded from a comparison of end group analyses and
measurements of absolute molecular weights that large
amounts of cycles were formed. Absolute molar masses
were determined by means of cryoscopy and ebullioscopy
in benzene. A recent investigation by Kricheldorf and
Schwarz revealed that such polyurethanes are insoluble in
benzene (due to intermolecular H-bonds). Therefore, the
molecular weight measurements reported by Stepto and coworkers are at least questionable.
Furthermore, polyadditions of tetra(ethylene glycol)
(TEG) and 1,6-diisocyanatohexane were performed in bulk
at 60 8C. Dibutyltin diacetate was used as catalyst and the
conversion of the isocyanate groups was monitored by IR
spectroscopy. The molar ratio of diol and diisocyanate was
varied and its influence on the solution viscosity and on the
structure of the reaction products was studied (Table 2 and
Experimental Part). The polyurethanes were isolated and
characterized after precipitation into methanol to allow
complete reaction of isocyanate groups with methanol.
MALDI-TOF MS proved the presence of cycles in all
samples (Equation (27)). However, samples prepared with
a 1.0:1.0 feed ratio or with an excess of the diol
mainly contained OH-terminated linear chains ((28) and
Figure 10A). The content of cycles increased with the
feed ratio of diisocyanate (Figure 10B). With an excess of 4
7 mol-% of diisocyanate, the OH-terminated linear chains
vanished, but the fraction of cycles having an allophanate
group with a pendent hexamethylenediamine residue (29)
increased. Furthermore, the Mei data in Table 2 indicate that
the fraction of cycles passes through a maximum when the
diisocyanate is used in an excess of around 4 mol-% (Mei:
molar mass at which the peaks of cycles and linear species
exhibit equal intensities in MALDI-TOF mass spectra).

Table 2. Syntheses of polyurethanes from tetra(ethylene glycol)


(TEG), and 1,6-hexamethylene diisocyanate (HMDI) in bulk at
60 8C (24 h, Bu2Sn(OAc)2 as catalyst).
Exp. no.

1
2
3
4
5
6
7
8
9
10
a)
b)
c)

TEG/HMDIa)

24:20.0
12:20.0
21:20.0
20:20.0
20:20.4
20:20.8
20:21.4
20:22.0
20:23.0
20:24.0

Yield

Zinhb)

Meic)

dL/g

Da

86
88
90
94
96
97
98
94
91
88

0.33
0.34
0.37
0.41
0.52

0.44
0.58

< 1 000
< 1 000
< 1 000
1 700
2 500
5 000
3 500
3 000
2 500
2 000

Ratios in mmol.
Measured at 20 8C with c 2 g/L in DMF.
Mass where peaks of cycles and linear chains showed equal
intensities in MALDI-TOF MS.

27

28

29

10 Ring-Expansion Polycondensation
The classical theory of polycondensation is based on a
chain growth of linear monomers via linear oligomers to
linear polymers. In contrast, ring-expansion polycondensation may be defined as a polymerization process starting
from cyclic monomers and yielding cyclic polymers. For a
proper understanding of its characteristics, kinetically
controlled ring-expansion polycondensation should
be compared with kinetically controlled ring-expansion
polymerization. As schematically illustrated for polymerization of lactones (Scheme 3), ring-expansion polymerization free of side reaction involves reaction mixtures that,
at any time, exclusively consist of cyclic species.[36,37]
Furthermore, the kinetic course is typically that of a chaingrowth polymerization. In the case of ring-expansion
polycondensation, one monomer may be linear and all
active growing species are linear. Since 100% formation of
cycles requires 100% conversion as for all KCPs, the
products of real ring-expansion polymerization may consist
of cycles by 99% but never by 100%. The third difference
relative to ring-expansion polymerization is the kinetic
course which obeys, of course, the pattern of step-growth
polymerization.
The first example of ring-expansion polycondensation
was based on the reaction between 2,2-dibutyl-2-stanna1,3-dioxepane and various ADADs (Equation (30)).[38]
This highly exothermic reaction yielded polyesters with
M ns in the range of 10 000 to 20 000 Da. Unfortunately, the
MALDI-TOF mass spectra recorded previously were limited to the detection of mass peaks around 5 000 Da, but in
this mass range only cyclic polyesters were detected as
reaction products. For the present work, several syntheses
were repeated by a somewhat modified procedure and
MALDI-TOF MS proves the formation of cyclic polyesters

372

H. R. Kricheldorf, G. Schwarz

Figure 10. MALDI-TOF mass spectra of polyurethanes prepared from tetra(ethylene


glycol) and 1,6-diisocyanatohexane in bulk at 60 8C: (A) feed ratio 1.0:1.0 (No. 4, Table 2);
(B) 4 mol-% excess of 1,6-hexamethylene diisocyanate (No. 6, Table 2).

Scheme 3.

as predominant reaction products up to masses around


7 000 Da. Analogous polycondensations were conducted with sulfur-containing cyclic monomers (Equation
(31)).[39] A further extension of this synthetic strategy has
been studied quite recently for polyesters of oligo- and
poly(ethylene glycol)s (PEGs).[40] PEGs react under thermodynamic control with dibutyltin dimethoxide to yield
tin-containing oligo- and polyethers (Scheme 4).[41] Characteristic of the MALDI-TOF mass spectra in theses cases
is the coexistence of two independent polydispersities
(Figure 11), namely the polydispersity resulting from the
polycondensation process and the polydispersity of the
commercial PEGs. Their polycondensations with aliphatic
or aromatic dicarboxylic acid dichlorides may yield a

373

Cyclic Polymers by Kinetically Controlled Step-Growth Polymerization

31
Scheme 4.

11 Odd-Even Effects

broad variety of cyclic poly(ether esters).[39] All these ringexpansion polycondensations share the same problems of
normal KCPs. The success, what means a high degree of
cyclization, requires the absence of side reactions and
careful optimization of the reaction conditions with nearly
quantitative conversion.

30

According to Jacobson and Stockmayer,[7] the probability


of kinetically controlled cyclizations should decrease with
the degree of polymerization byDP3/2 and in ring-chain
equilibria byDP5/2. Regardless if these frequency distributions or Equation (2) are taken into account, they do not
include any preference of odd- or even-numbered cycles.
However, in real KCPs a deviation from this rule of equal
probability may occur for various reasons, and three scenarios favoring even- over odd-numbered rings have been
reported so far.
When polycarbonates are prepared by polycondensation
of bisphenols with phosgene or diphosgene, odd- and evennumbered cycles are formed with equal probability as
discussed above. However, when bisphenol-A is polycondensed with bisphenol-A bischloroformate (BABC) in dry
inert solvents, even-numbered cycles should be formed
exclusively (Equation (32)). Such polycondensations were
reported for the first time by Schnell and Bottenbruch,[42]
who isolated cyclic tetramers from the reaction mixture.
However, at that time, characterization of the entire reaction
mixture by chromatographic or mass spectrometric methods was not feasible. The exclusive formation of evennumbered cycles in a polycondensation as described in
Equation (32) is, in principle, trivial. However, it was found
that with higher reaction temperatures more and more oddnumbered cycles are formed. How their formation can be
explained is a problem currently under investigation.

32
Figure 11. MALDI-TOF mass spectrum of a polyester prepared
by polycondensation of a stannylated PEG-1000 with phthaloyl
chloride.

Another scenario supporting a predominant formation of


even-numbered cycles is based on polycondensations of the

374

H. R. Kricheldorf, G. Schwarz

imide dicarboxyl chloride with silylated catechols (Equation (33)).[43,44] Below 250 8C this condensation method
does not involve transesterification reactions. The discrimination of odd-numbered cycles is obviously the result of
a hair-pin conformation with strong electronic interactions
(p-p, donor/acceptor (DA), and dipoly interactions) among
imide units aligned in parallel (Equation (33)). The catechols play the role of loops or hinges for this special
conformation. This interpretation is supported by the
following observations. Firstly, when the catechols
are replaced by resorcinols, odd- and even-numbered
cycles are formed with equal probability. Due to meta
functionalization, the resorcinols cannot act as hinges
for the hair-pin conformation. Secondly, no preference of
the even-numbered rings was found when terephthaloyl
chloride was used as comonomer of the catechols. Obviously, the terephthaloyl unit is too small and less suited for
DA interactions to stabilize the hair-pin conformation.

prepared either from the silylated or from the free diphenol


(Equation (35)) in isotropic phase, odd- and even-numbered
rings were formed with equal probability. The results suggest
that in the nematic phase hair-pin conformations were formed
stabilized by electronic interactions between the mesogens,
and ring closures occurred on the basis of these hair-pin
conformations(Scheme5).Suchascenarioresemblingthatof
the poly(ester imide)s in Equation (33) is, of course,
unfavorable for the generation of odd-numbered cycles.
Further intensive studies of correlations between monomer
structure and odd-even effects in the cyclization of oligomers
and polymers are certainly needed.

34

35

33
The third system favoring even-numbered cycles is based
on polycondensations in the nematic melt. It is well
established that oligomer chains and chain segments adopt
a nearly parallel alignment in nematic (and smectic) phases
with the consequence of highly anisotropic physical properties of the liquid or solid materials. Therefore, the
formation of entanglements or cycles in liquid crystalline
phases seemed to be rather unlikely. However, polyesters
prepared by polycondensation of silylated hydroquinone
4-hydroxybenzoate with ADADs (Equation (34)) in the
nematic phase contained even-numbered cycles as revealed
by MALDI-TOF MS.[45] When the same polyesters were

Scheme 5.

12 Hyperbranched Polymers
from abn Monomers
Hyperbranched polymers are polymers with a tree-shapetype of branching. This kind of branching (like all other
important polymer architectures) is an invention of evolution realized by polysaccharides like glycogen or
amylopectin. Polycondensations of a-bn monomers and
their copolycondensations with a-b monomers may
yield hyperbranched polymers when crosslinking side
reactions are absent. A first mathematical treatment of
a-bn homo- and copolycondensations, called cascade
theory, was elaborated by Flory around 1950. A systematic
contribution of cyclization reactions was not taken
into account, although Flory acknowledged that a few intramolecular reactions might occur as minor side reactions.

Cyclic Polymers by Kinetically Controlled Step-Growth Polymerization

Since neither NMR spectroscopy nor powerful MS were


available at the time of Florys work, clear-cut identification
and quantification of intramolecular reactions was not
feasible.
It is obvious from our theory of KCPs that the cascade
theory needs modification quite analogously to modification of the classical theory of linear polycondensations
(Florys terminology). When in a hypothetical experiment a
clean polycondensation of an a-bn monomer is taken to
100% conversion, either all monomers have merged into
one giant hyperbranched polymer (classical theory) or a
large number of cyclic polymers having hyperbranched side
chainsis formed (new theory). According to our new theory of
KCPs, cyclization competes with propagation in a-bn
polycondensationatany stage and concentration, and both the
number and size of branched cycles depend on the Vp/Vc ratio
quite analogously to linear polycondensation. In other
words, Equation (4) is, in principle, also valid for kinetically
controlled polycondensations of a-bn monomers.
The first experimental approach designed to yield
hyperbranched homo- and copolymers via KCP was
elaborated by Kricheldorf et al.[46] based on polycondensations of 3,5-bistrimethylsiloxybenzoyl chloride and
3-trimethylsiloxybenzoyl chloride. Unfortunately, several
attempts to analyze the structure of these aromatic (co)polyesters by means of MALDI-TOF MS failed. Meanwhile, two
research groups[47,48] have reported on ring closures in
kinetically controlled polycondensations of a-bn monomers. Either PESs were prepared from monomer (36a)[47] or
fluorinated polyethers from monomer (36b).[48] Mass peaks
of cycles were detected up to 3 000 or 5 000 Da and discussed
as byproducts resulting from the cyclization of oligomers.

36
Quite recently polycondensations of 3,5-bis(4-fluorobenzoyl)phenol, first described by Miller et al.,[49] in three different reaction media have been reinvestigated (Scheme 6).[50]
The lowest molecular weights (M n  3 500 Da, M w 
8 000 Da) were obtained in mixtures of sulfolane and
toluene or xylene. However, MALDI-TOF MS proved a
clean polycondensation with intensive cyclization. Only
mass peaks of cycles were detectable up to 14 000 Da,
indicating a nearly quantitative cyclization. In other words,
cyclization is not a minor side reaction in polycondensations
of a-bn monomers, it is a main reaction limiting the chain
growth quite analogously to linear polycondensations.
Furthermore, cyclization limits the growth of polydispersity,
which according to Florys cascade theory should reach

Scheme 6.

infinity, when conversion approaches 100%. This also means


that the so-called cascade theory is an unproven largely
misleading hypothesis, but not a theory.

13 Cyclization and Gelation


This section deals with polycondensations of a2 b3
monomer mixtures. The first experimental and theoretical
studies in this direction were reported by Kienle et al.,[51,52]
Flory[3,4] and Stockmayer.[53] All the experimental work
was based on the acid-catalyzed esterification of aliphatic
diols, triols (e.g., glycerol), or tetrols (e.g., pentaerythritol)
with diacid or tribasic carboxylic acid. Analytical tools
allowing the detection of side reactions, such as the formation of ether groups, vinyl groups or ring closure reactions,
were not available at that time.
The theoretical treatment of Flory was focused on a
branching factor, a, which was defined as the probability
that a chain segment reaching out from a branching unit (an)
is connected to another branching unit. Factor a is
correlated to the functionality of monomer an
in Equation (37) and with conversion in Equation (38). In
the case of an a2 b3 polycondensation with
equivalent feed ratio of a and b functional groups, (r 1)
a p2 1/2 is valid. Hence, gelation should occur at 76%
conversion. The experimental gel points were found at
somewhat higher conversion, and this difference relative to
the theoretical value was attributed to a few intramolecular
reactions. However, when other undetected side reactions
may occur, even this conclusion is mere speculation.
Furthermore, it should be mentioned that neither in the

375

376

H. R. Kricheldorf, G. Schwarz

experimental nor in the theoretical studies of the aforementioned authors, differentiation between KCPs and
TCPs was made.
a

1
n1

by Flory, which is based on constant, equivalent (not


equimolar!) feed ratios and variation of conversion.

37

r
1
a 2  pb2
pa
r

39
38

In the present work two syntheses of PESs from triphenols (Equation (39) and (40)) should be discussed[54,55]
and an analogous synthesis of PEKs (Equation (41)).[56]
The silylated triphenols were used to avoid side reactions
observed for analogous polycondensations of free triphenols.[55] Reaction conditions were optimized towards
almost quantitative conversion in connection with syntheses of linear PESs.[27] When exactly equimolar feed ratios
were used, no gelation was observed. In a hypothetical
experiment with 100% conversion and without any side
reaction or cyclization, a slight excess (e.g., 1 mol-%) of aa monomers should suffice to cause crosslinking and
gelation. The three polycondensations discussed here showed a decreasing tendency of gelation in the following order:
Equation (41) < Equation (40) < Equation (39). In the case
of PEK synthesis (Equation (41)), no crosslinking was
observed with a 5 mol-% excess of the difluoro monomer,
but an excess of 10 mol-% caused gelation. In the case of the
analogous PES (Equation (40)) even an excess of 10 mol-%
of the 4,40 -difluorodiphenylsulfone (DFDPS) did not
suffice for gelation which required an excess of 20 mol-%
(or little less). The polycondensations of silylated phloroglucinol even required an excess of 30 mol-% of DFDPS to
reach the gel point. MALDI-TOF mass spectra (discussed
below) did not reveal any significant side reactions but
confirmed a high extent of cyclization reactions such as
back-biting shown in Scheme 7. With these results in mind,
a plot of gel points versus molar excess of b-b monomers
allows a crude comparison of cyclization tendencies in
a2 bn polycondensations. This is an alternative
approach to the classical evaluation of gel points described

Scheme 7.

40

41

The discussion of a MALDI-TOF mass spectrum of a


PES, prepared according to Equation (40) with a 10 mol-%
excess of DFDPS, should illustrate what kind of information may be extracted from such mass spectra. As illustrated
in Figure 12, the mass range from 1 000 to 15 000 Da was
evaluated and all mass peaks were assigned. Almost all
peaks represent products containing at least one ring
(labeled CYor BxCY). Here Y indicates the total number of
repeat units in the cycles or their isomers. Increment x
counts the number of so-called bridge units B meaning

Cyclic Polymers by Kinetically Controlled Step-Growth Polymerization

Figure 12. MALDI-TOF mass spectrum of a poly(ether sulfone)


prepared from silylated 1,1,1-tris(4-hydroxyphenyl)ethane and
4,40 -difluorodiphenylsulfone in NMP (K2CO3) at 140 8C
(Equation (40)).

diphenylsulfone units bridging two-dimensional cycles


or connecting two rings. Scheme 8 illustrates this
nomenclature and provides a crude idea of the numerous
isomers which can be formed in an a2 b3 polycondensation (the analytical methods available in 2002 do not
allow the identification of individual isomers without
chromatographic separation). These mass spectrometric
results allow the following conclusions. Firstly, intensive

Scheme 8.

cyclization reactions not only produce two-dimensional


rings (series C2C9), but also bicycles and multicycles,
such as BC3, BC4, BC40 or BC5 and BC50 . Part of these

377

378

H. R. Kricheldorf, G. Schwarz

bicyclic and multicyclic compounds is monofunctional


(BC3, BC5, BC50 ), what means that a normal a2 b3
polycondensation produces chain terminators even in the
absence of any side reaction. Formation and incorporation
of these chain terminators explain why the molecular
weights of the linear and branched polymers formed before
the gel point are rather low. The same observation was
reported by Kienle et al.[51,52] and Flory[4] for their polyesters, but without any explanation. The systems studied by
these authors need reinvestigation by means of MALDITOF MS to elucidate, if those polyesters also form
monofunctional bi- and multicyclic reaction products. An
increasing feed ratio of monomer a2 has two consequences. On the one hand, it intensifies the formation of
bicycles and multicyclic components. On the other hand, it
promotes the connection of cyclic building blocks (intermolecular bridging) which finally results in a crosslinking
process. Therefore, the excess of monomer a2 needed for
crosslinking at nearly 100% conversion depends on the
overall rate of propagation versus the rate of all cyclization
steps. These results also suggest that the overall rate of
cyclization steps passes through a maximum. With increasing concentration of bicyclic and multicyclic species,
stereochemical and conformational situations favorable of
rapid cyclization and bridging will finally decrease, and the
ratio of propagation versus cyclization rates will increase,
resulting in gelation. A consequence of this scenario is the
formation of networks which mainly consist of cyclic
oligomers and multicyclic building blocks as illustrated in
Scheme 9. Polycondensations of a2 bn monomers
under kinetically controlled reaction conditions have been
studied by several research groups. Yet, to the best of our
knowledge, a detailed analysis by means of MALDI-TOF
MS has not been published so far.

Scheme 9.

14 Comparison with Cyclic Biopolymers


All biopolymers are synthesized in vivo by a polymerization process which involves condensation steps like any
normal polycondensation, but the kinetic course is that of
chain-growth polymerization. Two more similarities are a
kinetically controlled course (no equilibration) and the
possibility that cyclic oligomers and polymers are formed.
Mainly in the groups of oligopeptides and polypeptides or
in the case of DNA, living organisms produce numerous
cyclic species. Conspicuous is the fact that all living
organisms contain a smaller or larger fraction of cyclic
DNA, although its double-stranded helix seems, at a first
glance, particularly unsuited for cyclization. The molecular
weights and ring sizes of these cyclic DNAs vary over a
broad range. In protozoae relatively small rings containing
around 400 base pairs (equivalent to 2 400 atoms in the ring)
are known which have molar masses around 120 kDa.
Cyclic DNAs having molar masses of several million Da are
also known. The cyclization of such large chains seems to
require a special environment in the living cell, favoring
high coil density along with a cyclic overall conformation. However, it has been found recently that circular DNA
may be prepared by in-vitro experiments using the enzyme
ligase which does nothing else but closing phosphoric ester
bonds.[57] Such experiments led Stasiak to conclude:[58]
Ligase, which is frequently used for DNA circularization
experiments waits until thermal motion brings together
DNA ends with the correct polarity (30 with 50 ) and then
connects these ends covalently. The ligase works in two
steps. The first step closes linear molecules into so-called
nicked circles where only one strand is covalently closed
into a circle and there is a nick or interruption in the other
strand. However, in the second step of the reaction, the

Cyclic Polymers by Kinetically Controlled Step-Growth Polymerization

ligase closes the nick as soon as the ends of the other strand
come into close contact.
Certainly, the cyclization tendency of biopolymers in a
living cell may be favored by a restricted volume limiting
the space available for conformational changes. Furthermore, it may be favored by thermodynamically favored
loop-like conformations resulting from a certain sequence
of the monomers. Moreover, the intermolecular chain
extension reactions may be sterically hindered by the
spacial arrangement of ligase molecules. Nonetheless, the
fact remains that cyclization of biopolymers finally occurs
as a consequence of statistical conformational motions. If
this were to be true, similar cyclization tendencies should
exist in synthetic polymers and they should be observable
when side reactions can be avoided. In this context, it is
of great interest that cyclic polycarbonates were identified
up to molecular weights around 5560 kDa[2426]
(see Figure 6). This molecular weight corresponds to a
ring size of 2 6002 700 atoms which can rival with the ring
sizes of circular DNA in microorganisms. The result proves
that synthetic polymers may spontaneously form large
cycles which can rival with biogenic DNA, despite the
permanent competition of cyclization and chain growth, as
it is typical for normal KCP. Finally, it should be emphasized that the detection of large cycles in synthetic polymers
is currently restricted by technical limits of MALDI-TOF
mass spectroscopy rather than by the chemistry.

16 Experimental Part
Poly(ether ketone)s (Table 1)
Polycondensation of 4-tert-Butylcatechol (4-TBC)
4-TBC (10 mmol), 4,40 -difluorobenzophenone (10 mmol) and
K2CO3 (10.1 mmol) were weighed into a 150 mL three-necked
flask equipped with mechanical stirrer, dropping funnel and
distillation head. Sulfolane (50 mL) and xylene (20 mL) were
added and the reaction mixture was stirred at 170 8C for 6 h.
The xylene lost by distillation was gradually replaced from the
dropping funnel. Finally, the cold reaction mixture was poured
into methanol.
Polycondensation of Silylated 4-TBC
O,O0 -Bistrimethylsilyl 4-TBC (10 mmol), 4,40 -difluorobenzophenone (10 mmol) and dry K2CO3 (10.1 mmol) were weighed
into a 50 mL Erlenmeyer flask equipped with a magnetic bar
and dry NMP (65 mL) was added. The reaction vessel was
thermostated at 145 8C for 48 h. Afterwards, the reaction
mixture was precipitated into methanol. When such polycondensations were performed in 15 mL NMP, higher
molecular weights were obtained. Analogous polycondensations were performed with bisphenol-A or its bistrimethylsilyl
derivative.
4,40 -Difluorobenzophenone and 4-TBC were purchased
from Aldrich Co. (Milwaukee, USA) and used as received.
NMP (donated by BASF AG, Ludwigshafen, Germany) was
distilled twice over P4O10, whereas all other solvents were used
as received.

15 Conclusions

Polyurethanes (Table 2)

The results of numerous polycondensations and polyadditions have unambiguously demonstrated that kinetically
controlled step-growth polymerizations obey Equation (4).
Cyclization competes with propagation at any concentration and any stage of the polymerization until 100% cycles
are formed at 100% conversion. Cyclization of long chains
at high conversions is favored by self-dilution of the active
linear chains (Equation (3)). The average molecular weights
(M n and M w) of the cycles obtained at high conversions can
be shifted to lower values by dilution of the reaction mixture
and via the pseudo-high dilution method according to the
Ruggli-Ziegler dilution principle.[14]
Particularly interesting and useful is the comparison with
thermodynamically controlled polycondensations (TCPs).
As we have demonstrated recently,[59] thermodynamically
controlled reaction mixtures are dominated by ring-ring
equilibria at high conversion and the molar concentrations
of cycles reaches again 100% at 100% conversion (in the
absence of side reactions). Therefore, both KCPs and TCPs
may be summarized as follows: It is the fundamental
tendency of all step-growth polymerizations to yield cycles
as (kinetically or thermodynamically) stable end products.
The linear (or hyperbranched) chains are nothing but
reactive intermediates or end products of side reactions.

TEG (20 mmol) and 1,6-hexamethylene diisocyanate (HMDI;


20 mmol) were mixed at room temperature in an Erlenmeyer
flask (50 mL, with ground glass joint). A drop of a 0.5 M
solution of dibutyltin bisacetate in toluene was added, and the
reaction mixture was thermostated at 60 8C for 24 h. The
product was diluted with dry DMF (2050 mL) depending on
the viscosity and precipitated into water. The polyurethane was
isolated by filtration, intensively washed with water and dried
at 65 8C in vacuo.
MALDI-TOF Mass Spectrometry
All MALDI-TOF mass spectra presented in this work and
in most previous publications were recorded on a Bruker
Biflex III mass spectrometer equipped with a nitrogen laser
(l 337 nm). An acceleration voltage of 20 kV (rarely 25 kV)
was used and 150300 scans were accumulated. All mass
spectra were measured with a cut-off range of 1 000 Da, but in
selected cases, an additional measurement was performed with
a cut-off range of 3 000 Da. Irradiation targets were prepared
from tetrahydrofuran or chloroform solutions. In selected cases
(e.g., polyamides), trifluoroacetic acid or hexafluoroisopropylalcohol were added to the chloroform. Dithranol was used as
matrix, but for several polyamides better results were obtained
when 2-(4-hydroxyphenylazo)benzoic acid was used as
matrix. The polymer matrix ratio was varied between 1:1 and

379

380

H. R. Kricheldorf, G. Schwarz

1:200, depending on the success of the measurement. Although


all polymers prepared in standard glass ware contain enough
Na ions for doping, K trifluoroacetate was used as dopant for
two reasons. Firstly, an additional dopant allows control of
its quantity. Secondly, in the case of Na doping, a coincidental
K-doped peak of a cycle may be confused with a Na-doped
peak H2O.
From the PESs presented in Scheme 1 and 2, samples mainly
containing OH-terminated Lb chains were prepared. Furthermore, samples mainly consisting of D-F terminated Lc chains
were synthesized. Samples almost exclusively consisting of
cycles with molecular weights similar to those of the linear
samples were prepared. Finally, MALDI-TOF mass spectra
of 1:1 mixtures of cyclic and linear samples were recorded and
matrix dopant and solvent were varied. Regardless of all
variations, OH-terminated Lb chains flew best, the cyclic
polymers worst (Figure 9).
TEG, HMDI and dibutyltin bisacetate were purchased from
Aldrich Co. TEG was azeotropically dried with toluene and
distilled in a high vacuum over a short-path apparatus.
Received: February 19, 2003
Accepted: March 17, 2003

[1] W. H. Carothers, J. Am. Chem. Soc. 1929, 31, 2548.


[2] Collected Papers of W.H. Carothers on Polymerization,
H. Merk, G.S. Whitby, Eds., Interscience Publ., New York
1940.
[3] P. J. Flory, Chem. Rev. 1946, 39, 137.
[4] P. J. Flory, Principles in Polymer Chemistry, Cornell
University Press, Ithaca, New York 1953, chapter VIII.
[5] H. Staudinger, Die hochmolekularen organischen Verbindungen, Kautschuk und Cellulose, Springer Publ., Berlin
1932, p. 149.
[6] H. Mark, R. Raff, High Polymeric Reactions. Their Theory
and Practice, Interscience Publ., New York 1941, pp. 139
140, 151155, 176177.
[7] H. Jacobson, W. H. Stockmayer, J. Chem. Phys. 1950, 18,
1600.
[8] H. Jacobsen, C. O. Beckmann, W. H. Stockmayer, J. Chem.
Phys. 1950, 18, 1607.
[9] J. A. Semlyen, in: Large Ring Molecules, John Wiley &
Sons, New York 1996, chapter 1.
[10] R. F. T. Stepto, D. R. Waywell, Makromol. Chem. 1972, 152,
263.
[11] J. L. Stanford, R. F. T. Stepto, D. R. Waywell, J. Chem. Soc.
Faraday Trans. 1975, 71, 1308.
[12] [12a] M. Gordon, W. B. Temple, Makromol. Chem. 1972,
152, 277; [12b] M. Gordon, W. B. Temple, Makromol. Chem.
1972, 160, 263.
[13] G. Ercolani, L. Mandolini, P. Mencarelli, Macromolecules
1988, 21, 1241.
[14] K. Ziegler, Ber. Dtsch. Chem. Ges. 1934, 67A, 139.
[15] H. R. Kricheldorf, S. Bohme, G. Schwarz, Macromolecules
2001, 34, 8879.
[16] H. R. Kricheldorf, M. Rabenstein, M. Maskos, M. Schmidt,
Macromolecules 2001, 34, 713.
[17] T. Yokozawa, S. Horio, Polymer J. 1996, 28, 633.

[18] T. Yokozawa, T. Asai, R. Sugi, S. Ishigooka, S. Hiraoka, J.


Am. Chem. Soc. 2000, 122, 8313.
[19] T. Yokozawa, H. Suzuki, J. Am. Chem. Soc. 1999, 121,
11573.
[20] T. Yokozawa, Y. Suzuki, S. Hiraoka, J. Am. Chem. Soc. 2001,
123, 9902.
[21] H. R. Kricheldorf, A. Lorenc, J. Spickermann, M. Maskos, J.
Polym. Sci., Part A: Polym. Chem. 1999, 37, 3861.
[22] A. Horbach, H. Vernaleken, K. Weirauch, Makromol. Chem.
1980, 181, 111.
[23] A. Horbach, Polym. Prepr. (Am. Chem. Soc., Div. Polym.
Chem.) 1980, 21, 185.
[24] H. R. Kricheldorf, S. Bohme, G. Schwarz, C.-L. Schultz,
Macromol. Rapid Commun. 2002, 23, 803.
[25] H. R. Kricheldorf, S. Bohme, G. Schwarz, C.-L.
Schultz, J. Polym. Sci. Part A: Polym. Chem. 2003, 41,
890.
[26] H. R. Kricheldorf, S. Bohme, G. Schwarz, C.-L. Schultz,
R. Wehrmann, Macromol. Chem. Phys., submitted.
[27] H. R. Kricheldorf, S. Bohme, G. Schwarz, C.-L. Schultz,
Macromolecules, submitted.
[28] H. R. Kricheldorf, S. Bohme, G. Schwarz, C.-L. Schultz,
Macromolecules 2003, in press.
[29] H. R. Kricheldorf, S. Bohme, G. Schwarz, R.-P. Kruger, G.
Schulz, Macromolecules 2001, 34, 8886.
[30] H. R. Kricheldorf, L. Vakhtangishvili, G. Schwarz, R.-P.
Kruger, G. Schulz, Polymer 2003, in press.
[31] H. R. Kricheldorf, L. Vakhtangishvili, G. Schwarz, unpublished results.
[32] H. R. Kricheldorf, B. Schmidt, R. Burger, Macromolecules
1992, 25, 5465.
[33] H. R. Kricheldorf, M. Richter, G. Schwarz, Macromol.
Chem. Phys. 2003, 204, 646.
[34] H. R. Kricheldorf, M. Richter, G. Schwarz, J. Macromol. Sci.
- Pure Appl. Chem., submitted.
[35] H. R. Kricheldorf, S.-C. Fan, G. Schwarz, High Perform.
Polym., submitted.
[36] H. R. Kricheldorf, S. R. Lee, N. Schittenhelm, Macromol.
Chem. Phys. 1998, 199, 273.
[37] H. R. Kricheldorf, S. Eggerstedt, Macromol. Chem. Phys.
1998, 199, 283.
[38] H. R. Kricheldorf, D. Langanke, J. Spickermann, M.
Schmidt, Macromolecules 1999, 32, 3559.
[39] H. R. Kricheldorf, N. Probst, G. Schwarz, G. Schulz, R.-P.
Kruger, J. Polym. Sci., Part A: Polym. Chem. 2000, 38,
3656.
[40] H. R. Kricheldorf, M. Al Masri, G. Schwarz, unpublished
results.
[41] H. R. Kricheldorf, D. Langanke, Macromol. Chem. Phys.
1999, 200, 1174.
[42] H. Schnell, L. Bottenbruch, Makromol. Chem. 1962, 37,
1.
[43] H. R. Kricheldorf, G. Schwarz, A. Domschke, V. Linzer,
Macromolecules 1993, 26, 5161.
[44] A. A. Shaik, H. R. Kricheldorf, G. Schwarz, Polymer 2003,
44, 2221.
[45] H. R. Kricheldorf, M. Richter, G. Schwarz, Macromolecules
2002, 35, 5449.
[46] H. R. Kricheldorf, Q.-Z. Zang, G. Schwarz, Polymer 1982,
23, 1821.
[47] J. K. Gooden, M. C. Gross, A. Mueller, A. D. Stefanescu, K.
L. Wooley, J. Am. Chem. Soc. 1998, 120, 10180.
[48] C. A. Martinez, A. S. Hay, J. Polym. Sci., Part A: Polym.
Chem. 1997, 35, 2015.
[49] T. M. Miller, T. Y. Neenan, E. W. Kwock, S. M. Stein, J. Am.
Chem. Soc. 1993, 115, 356.

Cyclic Polymers by Kinetically Controlled Step-Growth Polymerization

[50] H. R. Kricheldorf, L. Vakhtangishvili, G. Schwarz, R.-P.


Kruger, Macromolecules, submitted.
[51] R. H. Kienle, F. A. van der Meulen, F. E. Fetke, J. Am. Chem.
Soc. 1939, 61, 2258.
[52] [52a] R. H. Kienle, F. E. Petke, J. Am. Chem. Soc. 1940, 62,
1053; [52b] R. H. Kienle, F. E. Petke, J. Am. Chem. Soc.
1941, 63, 481.
[53] W. H. Stockmayer, J. Chem. Phys. 1944, 12, 125.
[54] L. Vakhtangishvili, D. Fritsch, H. R. Kricheldorf, J. Polym.
Sci., Part A: Polym. Chem. 2002, 40, 2967.

[55] H. R. Kricheldorf, L. Vakhtangishvili, G. Schwarz, D.


Fritsch, Macromolecules, submitted.
[56] D. Fritsch, L. Vakhtangishvili, H. R. Kricheldorf, J.
Macromol. Sci. - Pure Appl. Chem. 2002, A39, 1335.
[57] J. Bednas, P. Fierrer, J. Dubochet, E. H. Egelmann, A. D.
Bates, J. Mol. Biol. 1994, 235, 825.
[58] A. Stasiak, in: Large Ring Molecules, J. A. Semlyen, Ed.,
John Wiley & Sons, Chichester, New York 1996, chapter 2,
p. 55.
[59] H. R. Kricheldorf, Macromolecules 2003, 36, 2302.

381

Вам также может понравиться