Cationic Ring-Opening Polymerization (CROP) Major Mechanistic Phenomena

HIGHLIGHT

STANISLAW PENCZEK Polish Academy of Sciences, Department of Polymer Chemistry, Centre of Molecular and Macromolecular Studies, Sienkiewicza 112, 90-363, Lodz, Poland

Received 25 February 2000; accepted 6 March 2000

ABSTRACT: A number of to- which way we learned that polyac- “temperature jump” techniques in day’s accepted basic viewpoints etals are not exclusively cyclic (as determining rate constants of ac- related to cationic ring-opening it was assumed), why CROP ions tive species interconversions are polymerizations (CROP) were a and ion pairs have similar reactiv- discussed. © 2000 John Wiley & Sons, matter of vivid disagreements be- ities, and why it was necessary to Inc. J Polym Sci A: Polym Chem 38: 1919– tween various research groups in propose that CROP proceeds at 1933, 2000 the past. These controversies are certain conditions by Activated described in this article and rea- Monomer Mechanism. Among Keywords: polyacetals; poly- sons of some differencies in opin- other subtle kinetic problems, ap- ethers; living polymerization; cations are explained. It is shown in plication of the dynamic NMR and ionic polymerization

Professor Stanislaw Penczek has been professor and head of the Depart- ment of Polymer Chemistry at the Polish Academy of Sciences (Centre of Molecular and Macromolecular Studies in Lodz) since 1974. He received his Ph.D. in 1963 at the Institute of Technology in Leningrad and the D.Sci. (habilitation) at the Lodz University of Technology in 1970. He spent 1996-1997 with Professor Michael Szwarc in Syracuse, New York, working on synchronous cationic vinyl and ring-opening polymerization. At the Polish Academy of Sciences, he has been mostly involved in studies of kinetics, thermodynamics, and mechanisms of anionic and cationic polymerization of cyclic compounds, including extensive work on models of biomacromolecules prepared by ROP. He is a member of editorial boards of several international journals (including J. Polym. Sci., Macromol.Chem. & Phys., Polymer Interna- S. PENCZEK tional, Progress in Polymer Science, Biomacromolecules, Current Org. Chem., Coll. Czech. Chem. Comm.) and was visiting professor in CRM

in Strasbourg and Universities in Cleveland (CWRU), Ghent, Mainz, Kyoto, Paris (M. P. Curie), and Stockholm (KTH). Professor Penczek has been elected to the Polish Academy of Sciences and was given a title of Titular Professor of the French Academy of Sciences (1993). He received several scientiﬁc awards, including M. Sklodowska-Curie Award (Poland) and Palmes Academiques (France). From 1997 to 1999 he was a President of European Polymer Federation. He is a Titular Member of IUPAC Macromolecular Division and is to chair the World Polymer Congress-IUPAC Macro 2000 to be held in Warsaw.

INTRODUCTION On the other hand, basic research in academia became enhanced by the fact that both polymerization of 1,3- General Importance of the Ring-Opening dioxolane (DXL) and of THF provided excellent models Polymerization to study chemistry of elementary reactions in cationic polymerization. Moreover, just at that time, studies of Ring-opening polymerization contributed to the mac- living anionic polymerization provided intellectual and romolecular science and polymer technology in a num- technical tools and had shown how to attack similar ber of ways. The most important was the work of problems arising in CROP. It was in this atmosphere that Staudinger establishing the existance of the covalently several research groups had undertaken basic research bonded macromolecules, and using as models poly(eth- that has eventually led to the description of CROP in the ylene oxide) and polyformaldehyde.1 way having no equivalent in the cationic vinyl polymer- Preparation of thermally stable polyformaldehyde ization and approaching the level of understanding of failed however at that time. A few decades later the these few vinyl anionic systems, which established the discovery of “self stabilization” in the cationc copoly- background of the modern polymerization kinetics, in- merization of 1,3,5-trioxane (TR) with ethylene oxide or cluding new wave of radical polymerizations. 1,3-dioxolane solved this problem. Polyformaldehyde (polyacetal) has become since then one of the largest scale industrial polymers made by cationic polymeriza- Major Controversies of the Past tion comparable only (in terms of the value of produc- tion) with polyisobutylene.2 Our present understanding of the CROP is based on Another important polymer prepared by cationic ring several fundamental pillars coming from studies of opening polymerization, namely polytetrahydrofuran model monomers, mostly DXL and THF. These phe- (polyTHF), has also a rather long history. Polymerization nomena were often established and finally accepted as of THF was studied first by Meerwein.3 If Meerwein results of resolving controversies arising between vari- would have studied polymerization after 1956, that is, ous research groups. Our laboratory was involved in after the famous article by Szwarc appeared on living almost all of these discussions. Thus, a certain part of polymerization,4 he could have used the expression “liv- this article is devoted to the description of these different ing” for the polymerization of THF, since he and his viewpoints. group had all of the kinetic elements at hands. Later, all One of the basic early controversy was related to the of the features of livingness of CROP of THF became mechanism of polymerization of cyclic acetals and the apparent in works of four research groups that came to structure of related polymers. It was proposed that poly- this conclusion independently of each other and working acetals (e.g., polyDXL) are exclusively cyclic and are in Russia, Great Britain, and the United States.5 formed by “ring expansion”. The Keele and Mainz PolyTHF has became an important technical product, groups were involved in discussing this theory.7 We shall ␣ ␻ ៮ Ϭ since the , -dihydroxy oligomers with Mn 1000 3000 briefly repeat here our article, published after a decade of constitute the soft-elastic segments in thermoplastic elas- painstaking studies in this area, and showing how the tomers: polyurethanes (Lycra), polyesters (Hytrel), and, ion-trapping method allowed determination of the simul- more recently, polyamides. taneous presence of linear and cyclic macromolecules Besides, CROP has been applied in technical poly- fitted with ions and coexisting during propagation.7 merization of cyclic siloxanes, imines, oxazolines, and Studies of the living THF polymerization, particularly phosphazenes.6 polymerization with interconverting covalent and ionic HIGHLIGHT 1921

Scheme 1 active species, allowed measuring of all of the involved Meanwhile, in Mainz, Volker Jaacks, working on a rate constants of elementary reactions. Our contribution similar system but at higher monomer concentrations did will also be covered in this short review. The most observe end groups,11 and there was no doubt that both striking at the time of those studies (late sixties) was the laboratories were careful enough with their analytical result showing the identity of reactivities of ions and ion techniques and in formulating their final conclusions. pairs, since it was in disagreement with a dominating After the untimely passing away of our friend Jaacks paradigm of higher reactivity of ions, and based on the during his journey to Himalaya, we decided to enter into studies of anionic polymerization of dienes and styrenes. this not finished at that time of discussion. We assumed The established in our works kinetic features of the that not only the back-biting should take place in such a ionic and covalent propagations will also be reviewed, as system, but also the end-to-end biting, and that the extent well as the kinetics and mechanism of interconversion of of end-biting should depend on the chain length (given propagating ions and dormant (practically not contribut- below by n) (Scheme 2). ing to propagation) macroesters. Today, it became ap- Thus, in order to show that active linear and cyclic parent that in both vinyl cationic and radical polymer- macromolecules do coexist during the chain growth we izations, particularly nitroxy mediated and ATRP, simi- used our method of ion-trapping with phosphines,12–15 8 lar kinetic phenomena exist. assuming that the secondary oxonium ions (cyclic active Finally, this article will describe another general con- species that could be formed by end-biting) would give a tribution of our group, namely the Activated Monomer protonated phosphine by fast proton transfer (tertiary Mechanism (AMM) in CROP. This mechanism resem- phosphonium salt), whereas the oxocarbenium (and ox- bles the better known AMM in anionic polymerization of onium) ions should give quaternary phosphonium ions. NCA and lactams, where the propagation step is based Thus, if the rate of exchange between these two species on the addition of the ionized monomer molecule to the 9 is not higher than the rate of trapping, then the propor- nonionic growing chain end. tions of linear and cyclic active species should be related to the proportions of two phosphonium ions: quaternary Basic Mechanistic Principles of the Cationic Ring- and tertiary, respectively. Their chemical shifts are Opening Polymerization known and several ppm apart (Scheme 3). Below, in Figure 1, two 31P{1H} NMR spectra are I have chosen only a few examples of our contributions given, for a system with [M] /[I] ϭ 20 and [M] /[I] that on my judgement are related to the most basic o o o o ϭ 7⅐103. These ratios are not far from the determined phenomena of CROP. 16 experimentally DPn (slightly higher). There are two major conclusions following from these Structure of Polyacetal Chains and the Mechanism experiments: of Cyclic Acetals Polymerization In the late 1960s the idea of the ring-expansion polymerization was formulated in Keele when attempts failed to find the end groups in the poly-1,3-dioxolane, prepared 10 with HClO4 initiator. Thus, it was proposed that polymerization proceeds by subsequent additions of the monomer molecules to the protonated (activated) rings, enlarging in every next step of propagation (Scheme 1). A given proton was, of course, assumed to travel to any nucleophilic center in the system—oxygen atom in monomer or macrocyclic molecule. Scheme 2 1922 J. POLYM. SCI. PART A: POLYM. CHEM.: VOL. 38 (2000)

Scheme 3 Figure 1. 31P{1H}NMR spectra of the trapped active species in the polymerization of poly(1,3-dioxolane) initiated with HOSO CF . Trapping agent: P(n-C H ) in 1. Both oxocarbenium (and/or tertiary oxo- 2 3 4 9 3 CH Cl solvent. The [M] /[I] ratios (Ϸ DP ) equal 20 nium) ions, and secondary oxonium ions 2 2 0 0 n and 7⅐103 in (A) and (B), respectively. (i.e., “protons”) are present in the system, 1: P(n-C H ) therefore propagation can take place on the ᮍ 4 9 3 2: H-P(n-C H ) linear active macromolecules (this is not ex- 4 9 3 ᮍ 3: . . .-CH OCH P(n-C H ) . (Ref. 16) cluding propagation on the cyclic ones). 2 2 4 9 3 2. The proportion of the linear macromolecules ﬁtted with oxocarbenium (tertiary oxonium) ions is higher for longer chains, as it could growth by ring expansion could not be dismissed but be expected. could be considered as redundant. Further analysis of the work of the Keele and Mainz

Actually, if the rate of exchange were much higher groups revealed, that in Keele much lower [M]o/[I]o ratio than the rate of trapping, then only one kind of species was used than in Mainz. Therefore, in agreement with would have been observed. results of Figure 1, mostly cyclics could be observed in In the next series of experiments with triethyl oxo- Keele and high proportion of linear molecules could be nium ion as initiator and 1,3-dioxolane-d6 as monomer, it observed in Mainz. Thus, both groups were correct in was possible to show by 1H NMR that after initiation one describing their own results and not quite right in criti- ethyl group is transferred from the oxonium ion to the cizing of the work of the other group. noncharged chain end. 1H NMR allowed observation of Nevertheless, the “ring-expansion” has been an inge- these end groups in the growing macromolecules with nious enlargement of our vision of the way polymeriza- ៮ 4 17 Mn up to 10 . (Scheme 4). tion could have proceeded, and it was shown more re- Comparison of the chemical shifts of the correspond- cently that, for example, ␤-propiolactone, initiated with ing models clearly indicated that also in this instance cyclic tin dialkyldialkoxide grows by ring-expansion.18 there are linear active macromolecules, since the chem- More recent work conﬁrmed this general possibility in O ical shift of CH3CH2OCD2CD2 . . . differed substan- the polymerization of other lactones, although ﬁnal prep- ᮍ Ͻ tially from the chemical shift of CH3CH2 O .17 These aration of the entirely cyclic population of macrocyclic experiments did show that indeed growth proceeds on macromolecules by a direct polymerization is not yet the linear macromolecules. Thus the assumption of the achieved.19,20 Thus, still the most general way of preparing macrocyclics in polymerization is by thermodynamic control. Remembering that the concentration of macrocyclics of a given size at equilibrium is equal to ϭ A ⅐ nϪ5/2,21,22 the total concentration of all of the cyclic macromolecules (of all sizes) is given by:

Scheme 4 HIGHLIGHT 1923

Scheme 6

highly nucleophilic environment of the oxygen atoms from monomers and the polymer units. Therefore it became of interest to establish whether indeed oxocarbenium ions do exist and how large is their proportion. Several groups were involved in this discussion, but the arguments produced were based mostly of the indirect Scheme 5 evidences. We decided to study the model system, being as close as possible to the actual polymerizing system. Thus, we used the following systems (Scheme 5). Both Therefore, if the starting concentration of a given systems were studied at equilibrium by dynamic 1H monomer (e.g., DXL) is equal to this sum, thus, is NMR. From the line broadening (shown in Fig. 2) the then practically all of the macro- rate constants were determined.24–26 molecules should be cyclic.23 This relationship, based on The rate constants of reactions of carbenium ions with the Jacobson–Stockmayer equation, is valid for thermo- nucleophiles were found to be close to the values typical dynamically ideal systems (“no interactions”), in practice for the diffusion controlled processes (above 30 °C). some shorter linear oligomers can also be present. Further analysis of this system allowed establishing Attempts to prepare exclusively cyclic polymers is contributions of the carbenium and oxonium ions to the important, since there is a difference of various proper- chain growth, as well as the structure of active species ties including the higher stability of the cyclic macro- bearing monomer molecules at their ends. 1H- and 13C molecules devoid of the chain ends. Besides, breaking NMR studies of the model system25 (Scheme 6), indi- once the cyclic macromolecule does not change its po- cated, that after the ﬁrst “cationation” of monomer (very lymerization degree. fast) and subsequent ring opening, almost exclusively the The next important problem in the ROP of cyclic seven-membered ring is formed, being in equilibrium acetals is related to the structure of active centers. with its open form (Scheme 7). In the polymerization of cyclic acetals, because of the Similar, equilibrium with another monomer —seven- stabilization of the carbenium ion by the oxygen atom in membered 1,3-dioxepane—looks differently: the en- ␣-position, there in an enhanced possibility of the exis- largement to the nine-membered ring does not take place. tence of carbenium ions (oxocarbenium) in spite of the This is because the seven-membered ring is less strained than ﬁve-membered ring but also less strained than the nine-membered one.

Structure of Active Species in the Polymerization of Cyclic Ethers and Their Reactivities Carbenium ions are very strong hydride ions (Hᮎ)ac- ceptors.27,28 This reaction was observed in the polymer- O ization of DXL; and respectively the. . . CH2OCH3 end groups were detected, resulting from the chain transfer to O ᮍ the. . . CH2OCH2 active species. The absence of the Hᮎ transfer products in the polymerization of cyclic ethers was used as one of the arguments for the unique-

Figure 2. Fragments of 1H NMR spectra of2:1 ᮍ ᮎ reaction mixtures of CH3OCH3 and CH3OCH2 , SbF6 at the temperature range related to fast and slow exchange (ref. 25). Scheme 7 1924 J. POLYM. SCI. PART A: POLYM. CHEM.: VOL. 38 (2000)

ϩ Ϯ Figure 3. Determination of kp and kp in the poly- app merization of oxepane (OXP); kp plotted as function of ␣ ϭ the degree of dissociation of macroion pairs: [OXP]0 ⅐ Ϫ1 5.0 mol L , 25 °C. (1) CH2Cl2; (2) C6H5NO2 as solvent (ref. 33). Figure 4. Oxiranium-tertiary oxonium ion with ᮎ SbF6 anion. ness of the chemical structures of active species, namely the tertiary oxonium ions that by themselves are not abstracting Hᮎ anions. shown in Figure 4; anions are large and Coulombic These ions were first observed by 1H NMR in poly- attractions with cations are relatively weak (distances are merization of THF in our group29,30 (later an improved given in Å).The second explanation is based on the spectrum was published). Protons in the endo- and oxo- assumed mechanism of the propagation step. The picture methylene groups were separately observed as two su- in Figure 4 shows a counterion as a “large ball under the perimposed triplets.30 There is only one kind of chemical relatively steep roof” and it can be further visualized that species (in contrast to cyclic acetals), because of the this ball (counterion) flows to the new position when the much higher nucleophilicity of ethers when compared monomer adds to the oxonium ion, in which over 90% of with acetals of the similar ring sizes. the positive charge is located on the CH2 groups. This is In the first articles published on the determination of an additional argument that the counterion does not have ϩ Ϯ ϩ kp (on ions) and kp (on ion pairs) the kp in CH2Cl2 to be removed far away from its position in the ground Ϯ 31,32 solvent was found to be several times larger than kp . state during the propagation step, and there is no reason We reexamined this system and then studied polymer- for the activation energies for the propagation step on ization of THF and oxepane in a number of other con- ions or ion pairs to differ substantially. ditions, changing counterions and solvents. In all of these Thus, kϩ Ϸ kϮ, as found experimentally, is not sur- ϩ Х Ϯ p p systems we observed kp kp , within an experimental app prising for these systems. error. The dependence of the observed rate constant kp In the anionic ring-opening propagation the negative (proportional to the rate) on the degree of dissociation ␣ charge is concentrated on the anion (e.g., alkoxide an- is for oxepane shown in Figure 3. The independence of ion), and the stereochemistry of the monomer addition the rate on the proportion of a given kind of ions clearly requires an extensive charge separation in the ion pairs. indicated that both ions and ion pairs propagate with Therefore, in the propagation on the ion pairs, the path identical rate constants.33 Ϯ from the ground state to transition state requires addi- Further work has shown, that kp does not depend on the anion structure; this result complements the earlier tional energy to counteract the Coulombic attraction ϩ Х Ϯ when monomer is added to the carbanion. observation of kp kp . Thus, the similar reactivity of ions and ion pairs as Ϯ well as indepence of kp on the structure of anion is the most general feature of the CROP.6 Two interlinked explanations have been given to these facts. The first is related to the size of active species; tertiary oxiranium ion has structure and dimensions as Scheme 8 HIGHLIGHT 1925

Propagation on Active Covalent Bonds and Macroion Pairs and Their Interconversions

There are two groups of anions (counterions) used in Scheme 9 ᮎ ᮎ ᮎ CROP. The complex anions, like AsF6 ,BF4 , SbF6 that are not able to form covalent bonds, since their coordi- nation sphere cannot be further enlarged, and the non- ᮎ ᮎ ᮎ tion arises: is the external or internal ion- complex ones, like CH3SO3 ,CF3SO3 , ClO4 , that form covalent bonds, using two electrons of the central oxygen ization faster? atom to form a sigma bond. 4. How fast is the dominating ionization? If it Therefore, whenever polymerization proceeds with mostly proceeds (as it was shown and will be these noncomplex anions reversible ester formation takes discussed below) by an intramolecular pro- ᮎ cess, is it fast enough to assume an equilib- place (e.g., for polyTHF and CF3SO3 ) (Scheme 8).This particular process was studied in our group extensively, rium taking place between the ionic and reactivities of covalent and ionic species were deter- covalent species throughout the complete mined and the rate constants of interconversion were polymerization? measured. 5. How to determine contribution of covalent Soon after the ﬁrst article by Smith and Hubin on the and ionic species to the building of the mac- ᮎ romolecules? polymerization of THF with HSO3CF3 (thus, CF3SO3 counterion) was published34 we proposed a general scheme of this kind of polymerization.35 Finally all of these questions were answered, and Surprisingly, as it will be shown later in this para- because some of these answers are of general impor- graph, although this scheme was solved in details, it has tance, the way these answers were found is discussed not fully been comprehended by research groups work- below. ing with other cyclic monomers. Not to mention vinyl cationic polymerization, like polymerization of styrene Direct Observation of Macroesters and Macroions and (particularly) vinyl ethers where reversible formation of macroester was also postulated.36 One of the important early observations was related to The following questions were asked, and ﬁnally an- the NMR spectra of the covalent and ionic species. Some swered: early articles disregarded the presence of esters and/or gave incorrectly assigned chemical shifts. Finally, cor- 1. How to determine concentrations of esters rectly understood 1H NMR and 13C NMR spectra al- and ions? What does it mean “covalent prop- lowed estimation of concentrations of both species. This agation”? Could the monomer addition pro- is illustrated in Figure 5, where 1H NMR spectra are ceed by a direct inclusion across the cova- given for both ions and esters in three different solvents. lent bond, omitting the ion formation? (e.g., The relative concentrations of macroesters and mac-

for CF3SO3-ester, the simultaneous rupture roion pairs in CCl4,CH2Cl2, and CH3NO2 vary dramat- of two bonds and formation of two new ically. There are almost no ions in CCl4, and almost no bonds in the four center concerted mecha- covalent species in CH3NO2 (the sum of concentrations nism). Or, does it proceed, on the contrary, of both species was kept constant). The controversy that exclusively by nucleophilic attack of the arose already in 1976 was rather of quantitative charac-

monomer molecule on the CH2 group adja- ter, since various groups were not observing the same cent to the ester bond, followed by oxonium proportions of covalent and ionic species in the same

ion formation (external ionization: kei) solvents.Then, we have shown that the proportion of ions 2. Whatever is the covalent propagation, is it increases with a total concentration of active species. faster or slower than ionic propagation and Apparently ion pairs form higher ionic aggregates, in- why? creasing this way the concentration of ionic species ob- 3. If covalent propagation proceeds exclusively served at 1H NMR. Thus, in order to properly determine by “external ionization”, that is, by the bio- the thermodynamic equilibrium constant covalent spe- molecular attack of monomer molecules, cies ª ion-pair, measurements have to be done at suf- then another ionization process has also to ficiently low total concentration of active species. Oth- be taken into account, namely the “internal erwise there is more than this one equilibrium involved. ionization” (Scheme 9) where, due to the Concerning the second part of the first question, and anchimeric assistance mostly the five-mem- related to the covalent propagation (the four-center ad- bered oxonium ion is formed. Then the ques- dition), we proposed a direct experimental proof. 1926 J. POLYM. SCI. PART A: POLYM. CHEM.: VOL. 38 (2000)

Scheme 10

roion pairs (independently of the anion structure) and Ϯ macroions. Therefore, it was justiﬁed to assume, that kp ᮎ for macroion pairs with noncomplex anions (like ClO4 ᮎ or CF3SO3 ) are the same as for the complex anions ᮎ ᮎ cov (BF4 , SbF6 etc.). Then, determination of kp is straightforward and simple, since the rate of monomer consumption is given by the following relationship:

Ϫ ͓ ͔ ϭ Ϯ͓ Ϯ͔͓ ͔ ϩ cov͓ cov͔͓ ͔ dt M /dt kp Pi M kp Pi M

Since the rate is determined experimentally, concentra- 1 ␦ cov Ϯ Figure 5. 300-MHz H NMR spectra (the 4.5–5.0 tions of both kinds of active species, [Pi ] and [Pi ], are region only) of living polyTHF with CF SO Oᮎ anion: 1 Ϯ ϭ ϩ 3 2 known from H NMR and kp ( kp ) is also known, then (a) in CD NO , (b) in CH Cl , (c) in CCl solvent (ref. cov 3 2 2 2 4 from any kinetic run kp could directly be calculated. 29). Moreover, thus determined rate constant of reaction of cov the macroester with monomer, namely kp (in another cov notation kei was used for kp ) could directly be compared Let us imagine reaction of methyl triflate with THF with the rate constant of reaction of a low molar mass proceeding in two ways. Either by direct inclusion or by ester with THF. Indeed, within the accuracy of kinetic cationation–ion-collapse. In both ways eventually the measurements kin was found to be practically equal to cov ester-ion equilibrium will be reached. However, if equil- kp , at the otherwise similar conditions. Of course, kin ibration is established slower than the actual reaction had to be measured at the early stages of reaction, before takes place, then overshooting over the equilibrium con- the first ion pair starts to participate in the reverse and centration should be observed of these species that are forward reactions. formed first, at least at the very beginning of reaction. Thus, at these conditions the concentration of the species formed first should be higher (overshooting) than their The Rate of Exchange Between Macroesters and concentration at equilibrium. the Macroion Pairs The preliminary results have shown that indeed, at the Two ways of interconversions between the macroesters early stage of reaction concentration of ions is higher and the macroion pairs should be envisaged, namely the than their concentration at the established equilibrium. This overshooting indicates that reaction proceeds rather by “cationation” of the monomer by ester and not by a direct insertion, as shown below is Scheme 10.

The Relative Rates of Propagation: Covalent Versus Ionic As it has already been shown, the most basic feature of the cationic ring-opening polymerization is the equality Ϯ ϩ of kp and kp , that is, the equal reactivities of the mac- Scheme 11 HIGHLIGHT 1927 intramolecular (unimolecular) and the intermolecular (bimolecular, involving monomer) (Scheme 11). One may note that in the intramolecular reaction the back process involves attack on the endocyclic methylene group whereas in the intermolecular reaction the exocyclic group is attacked. The intramolecular process was found to be faster than the intermolecular one. If it is sufficiently faster, then the contribution of the bimolecular reaction is low enough and can be neglected. Then the relative proportions of macroesters and macroion pairs should be invariant during polymerization, since an equilibrium is practically established at the early stage of polymerization and is preserved throughout polymerization. This was indeed experimentally observed in the polymeriza- 1 tion of THF: within the accuracy of the H NMR mea- Figure 6. Determination of kii and ktt by the “tem- surements the relative concentrations of macroesters and perature jump”method. Dependence of 1H NMR (300 macroions pairs are constant during polymerization. MHz) spectra of living ends on time in the polymeriza- ᮎ Ϫ Moreover, the internal kinetic order on monomer is equal tion of THF with CF3SO3 anion at 18 °C after dis- ϩ to one, a feature only possible if the bimolecular inter- rupting an equilibrium established at 18 °C (after ϫ 2 time s 10 ): (a) 1.5, (b) 5.7, (c) 13.2, (d) 22.8. CCl4 conversion is not important and therefore the proportions ϭ ⅐ Ϫ1 ϭ solvent, [THF]0 8.0 mol L [CF3SO3CH3]0 of both kinds of coexisting species do not depend on the 8 ⅐ 10Ϫ2 mol ⅐ LϪ1 (ref. 38). monomer concentration. Therefore with monomer conversion the proportion of macroion pairs does not de- crease and the internal order on monomer does not de- viate from the first order. the external ionization would proceed with a rate comparable to the internal ionization provided, that THF concentration equals to 100 mol ⅐ LϪ1 (so-called effec- Rate of the Macroester–Macroion Interconversion tive concentration). This is almost 10 times higher then In order to measure these rates the “temperature jump” the bulk concentration of THF (i.e., “THF in THF”). It technique was used. The macroester–macroions inter- follows, that in the polymerization of THF in solution, if ⅐ Ϫ1 conversion was brought to equilibrium at a certain tem- one starts from 5.0 mol L of THF the contribution of perature in a reaction mixture contained in the NMR the external ionization to the total ionization processes is tube. Then, the temperature was abruptly changed and approximately 5%, falling down to approximately 2–3% kinetics of interconversion, leading at this new temper- when the living polymer–monomer equilibrium is ature to the new equilibrium was measured. The rate of reached. Therefore, indeed, at least in the polymerization interconversion was relatively low in comparison with of THF the contribution of the external ionization to the the rate of establishing the thermal equilibrium, that is, macroester–macroion pair interconversion can be ne- the new temperature in the NMR tube.37,38 glected. In Figure 6 the change of the 1H NMR spectrum in the An interesting article has recently appeared, treating a similar problem in the polymerization of 2-methyl ox- region in which macroesters and macroions are absorb- 39 ing is given.38 Since in the previous paragraph we al- azoline. For the first time, in spite of a long history of ready established that the unimolecular ionization is studies of oxazolines the effective concentration was determined as be equal to 980 mol ⅐ LϪ1, at 25°, in faster, we know that this is the kinetics of the unimo- ᮎ lecular opposed reaction that was measured (cf. also nitrobenzene solvent and for Br counterion. Scheme 11). An expression “temporary termination” was used in Contribution of Covalent and Ionic Species to the order to stress that the macroesters add the monomer Chain Growth molecules much slower than the macroions do; today I would rather use an expression “temporary deactivation” As it has already been discussed, it is interesting to or “reversible deactivation”. establish contributions of each species to the building of ϭ Comparison of kii and kei ( kp), that is, the rate the macromolecule in polymerizations involving simul- constant of internal and external ionizations, led to the taneously growing covalent and ionic species. The equa- conclusion that in, for example, CH2Cl2 solvent, at 25 °C tion relating the contribution of polymer units introduced 1928 J. POLYM. SCI. PART A: POLYM. CHEM.: VOL. 38 (2000)

Scheme 13

of THF, conducted in two solvents, namely in CCl4 and ϭ ⅐ Ϫ1 in CH3NO2 ([THF]0 8.0 mol L , 25 °C) (Scheme Scheme 12 14). The illustrated life records give the following infor- mation: ␶Ϯ and ␶E are the corresponding lifetimes of the ion pair and an ester, ⌾Ϯ and ⌾E are equal to the number by covalent species (␥) with the propagation rate con- p p 40–43 of monomer molecules added to the ionic and ester stants is given by the following expression: ⌾Ϯ active species, d is a number of monomer molecules that depropagate from ionic active species. The lifetimes ␣k ␥ ϭ ei are given by intramolecular exchange processes. ͑ Ϫ ␣͒ ϩ ␣ 1 kpi kei This picture is a general one and applies equally well to any other polymerization with a temporary deactiva- where ␣ is the proportion of covalent active species. In tion. It would differ, however, for vinyl polymerizations, deriving of this equation it was assumed, that there is no where the intramolecular interconversion cannot pro- direct insertion of a monomer into the covalent bond. ceed. Thus, the complete scheme of polymerization, involving The real life records of polymer chains are more ionic and covalent species reads (the meaning of the complex; lengths of segments corresponding to times of super- and subscripts are self-explanatory in this scheme) being in the given state, ionic or covalent, are not iden- (Scheme 12). tical, not only for different chains in the given polymer- The equation for ␥ was first derived neglecting the ization, but for the same chain as well. Distribution of 40,41 depropagation. The complete treatment, including lengths is approximately the most probable one. depropagation, is given in our article, published a few years ago.42 Introduction of reversibility did not change the final expression. Apparently, insensitivity of ␥ to Activated Monomer Mechanism depropagation is related to the uniformity of the polymer During the early 1980s a striking phenomenon was ob- chains, that is, the equal probabilities of finding at any served: addition of alcohols to the polymerization of chain position a polymer unit introduced by one or some oxiranes reduced proportion of cyclics, known to another species. be formed by back-biting. Moreover, the lower the instantaneous monomer concentration (i.e., the higher the Reversible Deactivation: Its Influence on the Overall Kinetics In a process with reversible deactivation, like in the polymerization of THF with triflates or perchlorates, a given macromolecule propagates on its ionic species for a certain time, adding a certain number of monomer molecules, then the ion pair collapses into the covalent species. If these are much less reactive than ionic species cov Ӷ i (kp kp) then this reaction is equivalent to the temporary deactivation. A macromolecule spends some time as an inactive one, until, as a result of internal ionization (kii ӷ kei[M]) it reconverts into the active ionic form. This record of life can be visualized as segments of the line. The lengths of these segments, shown below, are equivalent to the time spent in a given state, and not to the length of the macromolecule (Scheme 13). Below it is shown how these periods oscillate in the polymerization Scheme 14 HIGHLIGHT 1929

Table I. Polymerization of Propylene Oxide (PrOx): Inﬂuence of Instantaneous Monomer Concentration on the Proportion of Cyclic Oligomers Formed

Conversion of Tetramer ⅐ Ϫ1 No. Conditions [PrOX]inst mol L PrOx (wt)% Formed (wt)%

1 PrOx introduced at once 2.0 (30.0.) Complete ϳ50 2 PrOx continuously charged faster 0.3 82 2.35 3 Slower ϳ0.05 Completea 0.95

ϭ ⅐ Ϫ3 ⅐ Ϫ1 ϭ ⅐ Ϫ1 O ϭ ⅐ Ϫ1 ⅐ Ϫ1 [HBF4-Et2O] 2.5 10 mol L ; [PrOx]total 2.0 mol L ;[ OH] 4 10 mol L ,CH2Cl2,25°C. a Complete polymerization in approximately 24 h.

[alcohol]/[monomer] ratio)—the lower proportion of cy- our studies of polymerization of glycidol. In this poly- clics. Some typical results are given below in Table merization the polymer formed contains both primary I.43–45 and secondary hydroxyl groups. In the ACE polymeriza- Back-biting is the unimolecular reaction, therefore tion ring opening should lead to the repeating units with O any external factor, like addition of an alcohol, is not pendant CH2OH groups (Scheme 17). Thus, for both able to inﬂuence its rate, and consequently the proportion ␣- and ␤-ring openings (presumably, the ␣ opening of cyclics formed. Therefore, we proposed that in the should prevail) polymer units contain three atoms O O O O presence of alcohols there is a mechanism of propagation ( O C C ) with CH2OH pendant group. Chain in which formation of cyclics is either hampered or transfer involving hydroxyl groups would not change eliminated. The best candidate for such a mechanism is this structure, because any opening of the tertiary oxo- polymerization in which chain ends in the growing mac- nium ion (attack on CH2 or CH) would generate primary romolecules are devoid of ions. Then, ions should be hydroxyl group. Secondary hydroxyl groups can, how- located in the monomer molecules, and this kind of ever, be observed only if the AMM operates. Indeed, as polymerization is called “Activated Monomer Polymer- it is shown below, in one mode of attack, namely in ␣ ization” (AM). What we proposed for the cationic pro- opening, the corresponding repeating units are formed cess has already been introduced, for a different reason in (Scheme 18). Thus, concentration of the secondary hy- anionic polymerization. droxyls in the polymer units provides the lowest contri- Thus, the AM mechanism, in its simplest form, can be bution of the AMM in the chain growth.29 Si-NMR was described by an equation, in which the nucleophilic applied after silylation of the polymer and both primary attack of the hydroxyl end group from the macromole- and secondary OOH were detected. cule on the carbon atom, adjacent to the oxonium ion, provides the propagation step (Scheme 15). (“Hϩ” de- Some Aspects of the Kinetics of AM Polymerization notes a proton, located on one of the nucleophilic sites in the system, that is, OOH, monomer molecule, or repeat- When unsymmetrically substituted oxirane is polymer- ing unit.) ized, there are four propagation reactions possible, in- The AM mechanism may operate effectively, provid- volving either primary or secondary hydroxyls at the ing that the concentration of the nucleophilic end groups polymer chain end and either ␣-or␤-ring opening in the (reproduced in each propagation step) is high enough to oxiranium cation in the actual propagation step. These make this route competitive with respect to the ACE four reactions are shown below (Scheme 19). The indi- mechanism. This may be illustrated by the scheme given vidual rate constants k1,1, k1,2, k2,1, and k2,2 are denoted below (Scheme 16). in such a way that the ﬁrst digit describes the attacking Both ACE and AM mechanisms can contribute in building of the macromolecules, since both mechanisms can operate simultaneously. The proof of the AM polymerization to proceed was obtained in several ways. The most direct one came from

Scheme 15 Scheme 16 1930 J. POLYM. SCI. PART A: POLYM. CHEM.: VOL. 38 (2000)

Scheme 17 hydroxyl and the second one the formed hydroxyl (1 stands for the primary and 2 for the secondary ones). Studies of the polymer end groups, at various stages of polymerization, revealed that after the first few addi- Scheme 19 O tions to the initiator, the ratio of CH2OH/CHOH end groups becomes invariant. Thus, for XAH (propylene O ϭ A app oxide) [ CH2OH]/[CHOH] 45 : 55 and for X CL of propagation kp is related, as discussed earlier, to k1 ϭ appϭ ␣ ϩ ␣ (epichlorohydrin) [CH2OH]/[CHOH] 5 : 95. Thus, in and k2 in the following way: kp k1 (1- )k2, where the kinetic experiments initiator was composed from a ␣ denotes the proportion of the primary hydroxyls. Thus, mixture of alcohols modeling the growing chain ends in for the four unknowns there are four equations relating the proportion similar to that observed in polymerization. them, allowing k1,1, k1,2 and k2,2 to be independently app ϭ The apparent (observed) rate constants of propaga- determined once kp is known. Then, e.g., k1/k2 a; k1,1/ ϭ app ϭ ␣ ϩ ␣ tion, encompassing these four constants could be mea- k1,2 b; kp k1 (1- )k2. sured directly from the polymerization kinetics, but to Solutions of these relationships gives, for instance, ϭ app ␣ ␣ ϩ ϩ determine the individual rate constants model studies had k1,1 kp b/[( (1-1/ ) 1/a)](1 b). to be performed, allowing one to find the relationship The ratios of rate constants are listed in Table II app ϭ ␣ ϩ between these constants; actually kp (k1,1 k1,2) below. ϩ ␣ ϩ ␣ (1- )(k2,1 k2,2) where denotes molar fractions of The required apparent rate constant of propagation app the primary hydroxyl groups. Therefore, additional rela- kp , needed for determination of the rate constants of tionship between the constants had to be determined. elementary reactions, is related to the monomer con- This has been done in the following way for both mono- sumption by the usual dependence: mers described in this section: first the corresponding Ϫ ͓ ͔ ϭ app͓ ϩ ͔͓O ͔ primary and secondary alcohols were reacted separately d M /dt kp H M OH with a monomer under study, and the rates of formation of products, due to ␣-or␤-ring openings, were measured where [-OH] is the sum of concentrations of the second- by GLC in two separate sets of experiments: for the ary and primary hydroxyl groups. The total concentration primary and for the secondary alcohol, both modeling the of the hydroxyl groups and their proportion are invariant growing chain ends. These measurements gave the ratios (cf. preceding section). Thus, the knowledge of concen- k1,1/ k1,2, and k2,1/ k2,2. In the next step the relative tration of the activated monomer is required to determine app reactivities of primary and secondary hydroxyl groups kp . toward protonated monomer, that is, the ratio k1/k2 At the very early stage of polymerization, the only ϭ ϩ ϩ (k1,1 k1,2)/(k2,1 k2,2) were determined by reacting two nucleophilic sites in the reaction mixture are the a mixture of two alcohols (primary and secondary) with monomer molecule and hydroxyl chain ends. Then, we a monomer and measuring by GLC the relative rates of have Scheme 20. Combining the two equations given ϩ ϩ ϩ ϭ disappearance of the used alcohols. The ratio of these above, and remembering that [H M] [- OH2] ϩ ϩ ϭ relative rates is equal to k1/k2. The apparent rate constant [H ]0 (where [H ]0 [catalyst]0), we finally have, after eliminating the unknown [HϩM], Scheme 21. -d[M]/dt denotes the monomer consumption at the very early

Scheme 18 Scheme 20 HIGHLIGHT 1931

Table II. Ratios of Rate Constants of Elementary Reactions in Addition of Propylene Oxide and Epichlorohydrin to the Model Alcohols

Rate Scheme 21 Constantsa Propylene Oxide Epichlorohydrin

Ϯ Ϯ k1/k2 1.5 ( 0.3) 3.0 ( 0.5) k /k 0.90 0.013 hindered secondary group on the substituted carbon atom 1,1 1,2 in the AM ring (Scheme 22). k2,2/k2,1 0.77 120 In the polymerization of epichlorohydrin both primary a Ϫ1 ⅐ ⅐ Ϫ1 CH2Cl2, 25 °C (in mol L s ). and secondary hydroxyls react predominantly with the least substituted carbon atom. Thus, the electronic factor

(the -CH2Cl group has strong electron-withdrawing stage of polymerization (first additions), when the pres- properties) is overshadowed by the steric factor. Indeed, ence of polymer units can be neglected. At this stage this from a purely charge distribution viewpoint the rate is approximately equal to the rate of initiator (alco- -CH(CH2Cl) carbon atom should be (but is not) a pre- hol) consumption. On this bases the individual rate con- ferred site of attack. The importance of the steric factor stants have been determined. These constants are tabu- is much more pronounced for the polymerization of lated below for propylene oxide and epichlorohydrin epichlorohydrin than for propylene oxide. The primary polymerizations (Table III). hydroxyl groups are a few times more reactive than the In the polymerization of propylene oxide the individ- secondary ones, however, the effect of position of sub- ual rate constants do not differ considerably from each stituent is less important than the position of ring attack. other, indicating that there is no significant preference for On the basis of the thus determined rate constants it any of the four growth reactions involved. The relatively was possible to calculate the polymers microstructures. It small differences in the rate constants may be explained is remarkable, how well thus calculated microstructures by the interplay of two factors, namely steric and elec- agree with microstructures determined directly from the tronic factors. Secondary hydroxyls in poly(propylene NMR data. This comparison is given below (Table IV). oxide) oligomers are more basic than the primary ones, Cationic ring-opening polymerization involving acti- but, on the other hand, there is considerably more steric vated monomer is the latest complex and comprehensive hindrance for the reaction of the secondary groups. Thus, endeavor of our laboratory in the cationic field. For this both effects are partly counterbalanced. The steric factors reason our work in area is described in more detail. seem to be more important, because primary HO- groups During the last decade several other research groups used are slightly more reactive than the secondary ones. Thus, this approach either for preparation of polymers devoid of cyclics or in order to explain the results of their own. the highest rate constant (k1,2) is observed for an attack of the primary hydroxyl group on the unsubstituted carbon atom in the protonated monomer molecule, while the CONCLUSION lowest (k2,1) is for the attack of the more sterically In the preceeding sections some contributions of our group to the understanding of the cationic ring-opening Table III. Rate Constants of Elementary polymerization are reviewed. However, this is not a Propagation Steps in the Polymerization of typical review, and Professor Percec, when asked me to Propylene Oxide and Epichlorohydrin prepare this article has also mentioned, that it may be Proceeding by AM Mechanism more personal than usually. Therefore, I have taken a liberty in this last section to describe the contributions of Monomer my coworkers, that took part in this adventure, and quite a often provided solutions by themselves. ki,j Propylene (molϪ1 ⅐ L ⅐ sϪ1) Oxide Epichlorohydrin

k1,1 0.170 0.055 k1,2 0.190 0.41 k2,2 0.135 0.135 k2,1 0.105 0.0011

a CH2Cl2,25°C. Scheme 22 1932 J. POLYM. SCI. PART A: POLYM. CHEM.: VOL. 38 (2000)

Table IV. Fractions of the h-t, h-h, and t-t Dyads in Polymerization of Propylene Oxide and Epichlorohydrin According to the AM Mechanism, Calculated from

Kinetic Results (i.e., from the Values of k1,1, k1,2, k2,1, and k2,2) and Found From 13C NMR

Monomer

Propylene Oxide Epichlorohydrin Method of Determination h-t h-h t-t h-t h-h t-t

Kinetics 0.52 0.23 0.25 0.93 0.007 0.063 NMR 0.50 0.25 0.25 0.93 0.008 0.062

We started our work on the CROP in the early sev- REFERENCES AND NOTES enties, the peak of our activity in this area was ten years 1. Staudinger, H. From Organic Chemistry to Macromole- later, but even now, after 25 years we are still involved cules; Wiley Interscience: New York, London, 1970; in some related problems (activated monomer), although Chapters 9, 10. our major activity shifted to other areas of macromolec- 2. Vairon, J. P.; Spassky, N. In Cationic Polymerizations; ular science. Matyjaszewski, K., Ed.; Marcel Dekker: New York, 1996; Structure of polyacetals, and the mechanism of Chapter 8, p 683–741. polymerization of cyclic acetals were mostly estab- 3. Meerwein, H. In Houben-Weyl Methoden der Organischen lished by Przemyslaw Kubisa in his Ph.D. (1970) and Chemie, 4th ed.; Muller, E., Ed.; Georg Thieme Verlag: habilitation (1982). Studies of the reactivity of ions Stuttgart, 1965; Vol.VI/3, p 325. 4. Szwarc, M. Nature 1956, 178, 1168. and ion pairs in CROP were started in Ph.D. of Stanis- _ 5. Dreyfuss, P.; Dreyfuss, M. Poly(tetrahydrofuran); Gordon law Slomkowski (H transfer, carbenium–oxonium & Breach: New York, 1982. equilibria) and in his subsequent work on THF ions 6. Penczek, S.; Kubisa, P.; Matyjaszewski, K. Cationic Ring- and ion pairs. Then we were joined by Krzysztof Opening Polymerization; Springer-Verlag: Berlin, Heidel- Matyjaszewski, who in his Ph.D. (1976) and then in berg, New York, 1985; Vol. I, II. habilitation (1983) elaborated a more complete 7. Szymanski, R.; Kubisa, P.; Penczek, S. Macromolecules scheme of THF polymerization, including the ester- 1983, 16, 1000. ion pair equilibrium and kinetics of interexchange. 8. Penczek, S.; Szymanski, R.; Duda, A. Macromol Symp 1 1995, 98 193. The dynamic H NMR allowed Ryszard Szymanski 9. Penczek, S.; Sekiguchi, H.; Kubisa, P. Chapter Activated establishing in his Ph.D. work (1981) the reactivities Monomer Polymerization of Cyclic Monomers. In Macro- of carbenium and oxonium ions and the kinetics of molecular Design of Polymeric Materials; Hatada, K.; their interchange. Then, in his habilitation (1989) a Kitayama, T.; Vogl, O.; Eds.; Marcel Dekker: New York, general theory of equilibrium copolymerization was 1997; p 199. developed. All those scientists are already for some 10. Plesch, P. H.; Westerman, P. H. J Polym Sci Part C1968, time-independent researchers with their own research 16, 3837. groups. Sometimes, however, we join our efforts in 11. Jaacks, V.; Boehlke, K.; Eberius, E. Makromol Chem 1968,118, 354. solving new problems. Like the kinetic theory of mac- 12. Brzezinska, K.; Chwialkowska, W.; Kubisa, P.; Matyjas- rocyclization (with J. Chojnowski), the activated zewski, K.; Penczek, S. Ion Trapping in Cationic Polymer- monomer polymerization (with P. Kubisa) or the ki- izations. Makromol Chem 1977, 178, 2491. netic theory of the chain transfer to polymer (with S. 13. Kubisa, P.; Penczek, S. Makromol Chem 1979, 180, 1821. Slomkowski and R. Szymanski). K. Matyjaszewski 14. Matyjaszewski, K.; Penczek, S. Makromol Chem 1981, has recently chosen a world of radical polymerization, 182, 1735. and some of us are considering of joining him in this 15. Penczek, S. Makromol Chem Suppl 1979, 3, 17. 16. Kubisa, P.; Penczek, S. Makromol Chem 1976, 180, 1821. new endeavor where kinetics has some similarities 17. Kubisa, P.; Penczek, S. Makromol Chem 1978, 179, 445. with CROP, since it also involves reversibly formed 18. Davies, A. G. In Organotin Chemistry; VCH: Weinheim, dormant species. 1997; p 173. There was also a larger number of gifted Ph.D. stu- 19. Kricheldorf, H. R.; Eggerstedt, S. J Polym Sci Part A: dents and their names appear in the list of references. Polym Chem 1998, 38, 137. HIGHLIGHT 1933

20. Stridsberg, K.; Ryner, M.; Albertsson, A. Macromolecules, 34. Smith, S.; Hubin, A. J. J Macromol Sci 1973, A7, 1399. in press. 35. Matyjaszewski, K.; Kubisa, P.; Penczek, S. J Polym Sci 21. Jacobson, H.; Stockmayer, W. H. J Chem Phys 1950, 18, 1974, 12, 1333. 1600. 36. Matyjaszewski, K.; Sawamoto, M. In Cationic Polymer- 22. Semlyen, J. A. Adv Polym Sci 1976, 21, 41. izations; Matyjaszewski, K., Ed.; Marcel Dekker: New 23. Matyjaszewski, K.; Zielinski, M.; Kubisa, P.; Slomkowski, York, 1996; Chapter 4, p 300. S.; Chojnowski, J.; Penczek, S. Makromol Chem 1980, 37. Buyle, A. M.; Matyjaszewski, K.; Penczek, S. Macromol- 181,1469. ecules 1977, 10, 269. 24. Penczek, S.; Szymanski, R. Polym J 1980, 12, 617. 38. Buyle, A. M.; Matyjaszewski, K.; Penczek, S. J Polym Sci 25. Szymanski, R.; Penczek, S. Makromol Chem 1982, 183, 1976, 14, 125. 1587. 39. Dworak, A. Macromol Chem Phys 1998, 194, 1843. 26. Penczek, S.; Kubisa, P. Progress in Polymerization of 40. Matyjaszewski, K.; Szymanski, R.; Kubisa, P.; Penczek, S. Cyclic Acetals. In Ring-Opening Polymerization; Saegusa, Acta Polymerica 1984, 35, 14. T.; Goethals, E., Eds.; ACS Symp Ser 1977, 59, 58. 41. Penczek, S.; Szymanski, R. Makromol Chem Macromol 27. Penczek, S. Macromol Chem 1974, 175, 1217. Symp 1992, 60, 65. 28. Slomkowski, S.; Penczek, S. J Chem Soc Perkin Trans 2 1974, 1718. 42. Penczek, S.; Szymanski, R.; Duda, A. Macromol Symp 29. Matyjaszewski, K.; Penczek, S. J Polym Sci Polym Chem 1995, 98, 193–216. Ed 1974, 12, 1905. 43. Penczek, S.; Kubisa, P.; Matyjaszewski, K.; Szymanski, R. 30. Penczek, S.; Matyjaszewski, K. J Polym Sci 1976, 56, In Cationic Polymerizations and Related Processes; Goeth- 255. als, E. J., Ed.; Academic Press: New York, 1984; p 184. 31. Sangster, J. M.; Worsfold, D. J. Macromolecules 1972, 5, 44. Penczek, S.; Kubisa, P. In Ring-Opening Polymerization; 229. Brunelle, D. J., Ed.; Hauser Publishing: Munich, Vienna, 32. Bourdauducq, P.; Worsfold, D. J. Macromolecules 1975, 8, 1993; Chapter 2, p 14–86. 562. 45. Penczek, S.; Kubisa, P. In Comprehensive Polymer Sci- 33. Brzezinska, K.; Matyjaszewski, K.; Penczek, S. Makromol ence; Allen, G.; Bevington, J. C., Eds.; Pergamon Press: Chem 1978, 179, 2387. Oxford, 1989; p 751.