Perspectives on the contributions of Michael Szwarc to living polymerization

ARTICLE IN PRESS

Prog. Polym. Sci. 31 (2006) 1041–1067 www.elsevier.com/locate/ppolysci

Perspectives on the contributions of Michael Szwarc to living polymerization Johannes Smida, Marcel Van Beylenb, Thieo E. Hogen-Eschc, a

Chemistry Department, State University of New York College of Environmental Science and Forestry, Forestry Drive, Syracuse, NY 13210, USA b Katholieke Universiteit Leuven, Department of Chemistry, Laboratory of Macromolecular & Physical Organic Chemistry, Celestijnenlaan 200F, Louvain, B-3001 Belgium c Loker Hydrocarbon Research Institute and Department of Chemistry, University of Southern California, University Park, Los Angeles, CA 90089-1661, USA Received 11 September 2006; received in revised form 14 September 2006; accepted 15 September 2006

Abstract The review starts with a short historical introduction (Sections 1 and 2), followed by the famous work of Michael Szwarc on electron transfer to styrene and related monomers in THF and similar solvents (Section 3), forming the basis of most of his subsequent work. Section 4 describes his now classical work on the effects of ion pairing, ion pair solvation and triple ion formation on styrene anionic polymerization kinetics, as well as exploratory work on several related monomers and a brief description of work by others on related systems, such as the effects of LiCl and lithium alkoxides on the polymerization of styrene, and the mediation of styrene polymerizations by divalent Ba2. Section 5 starts with a summary of solvation studies of fluorenyl ion pairs that allowed a better understanding of the role of ion pairs in styrene polymerizations. The effects of solvent and carbanion structure on ion pair solvation and dissociation and their role in initiation equilibria are discussed. Finally a study of the dynamics of ion pair dissociation and triple ion formation based on the second Wien effect is discussed. Section 6 reviews anionic copolymerization studies by Szwarc and collaborators. This is followed by subsequent work on the Hammet relationships involving the addition of 1,1-diphenylethylenes to polystyrenelithium in hydrocarbon/THF. Studies by others on the role of Li–pi donors coordination in butadiene/styrene and isoprene/styrene copolymerizations in hydrocarbon media are also reviewed in this section, as is the role of this coordination in the LiOH mediated isotactic polymerization of styrene. r 2006 Published by Elsevier Ltd. Keywords: Anionic polymerization; Living polymerization; Vinyl anionic copolymerization; Ion pairs; Solvent-separated ion pairs; Radical anions

Contents 1. 2. 3.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1042 Discovery of living polymers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1043 Electron transfer studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1043

Corresponding author.

E-mail address: [email protected] (T.E. Hogen-Esch). 0079-6700/$ - see front matter r 2006 Published by Elsevier Ltd. doi:10.1016/j.progpolymsci.2006.09.001

ARTICLE IN PRESS J. Smid et al. / Prog. Polym. Sci. 31 (2006) 1041–1067

1042

4.

5.

6.

7.

Anionic polymerization of styrene and other vinyl monomers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1045 4.1. Polymerization of styrene . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1045 4.2. Other vinyl monomers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1048 4.3. Triple ion formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1048 Ion pair solvation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1050 5.1. Fluorenyl anions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1050 5.2. Solvent effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1052 5.3. Ion pairing effects in other carbanions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1052 5.4. Ion pairing effects on equilibria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1054 5.5. Dynamics of interconversion of ionic species . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1055 Copolymerization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1057 6.1. Polar media . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1057 6.2. Copolymerizations in apolar media. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1059 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1063 Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1063 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1064

1. Introduction The contribution by Michael Szwarc to chemistry and particularly to polymer chemistry has been farreaching and is still evident today. This is especially the case for his discovery of living anionic polymerization of styrene and related vinyl monomers and his extensive studies on the corresponding mechanisms of initiation (electron transfer) and polymerization. These subjects were covered by Szwarc in numerous reviews [1–3] and in a number of textbooks [4–7]. Michael Szwarc was born in Poland in 1909 and received his early science education at the Warsaw Polytechnic Institute. He immigrated with his family to Israel in 1935 and earned a PhD degree in organic chemistry from Jerusalem’s Hebrew University in 1942. His main interest however was physical chemistry, and his career really took off when he joined the group of Michael Polanyi, an outstanding physical chemist at the University of Manchester, England. He was awarded the prestigious Doctor of Science degree in physical chemistry in 1949 for his elegant research on bond dissociation energies. It was at Manchester University that Michael developed an interest in polymers when his pyrolysis work with pxylene led him to the discovery of poly(p-xylylene) [8,9], a polymer commercialized in later years by Union Carbide under the name of ‘‘Parylene’’. In 1952 he accepted a professorship in the Chemistry Department of the State University of New York College of Forestry (presently Environmental Science and Forestry, or SUNY ESF) in Syracuse, NY. He became the first director of its newly created Polymer Research Center in 1966.

After his retirement in 1979 Michael moved to La Jolla, CA and was appointed a Senior Fellow in the Hydrocarbon Research Institute of the University of Southern California in Los Angeles, a position he held until his death in 2000. Of his more than 500 publications, over 100 were published during his 21 years of ‘‘retirement’’, most of them in collaboration with other researchers, including one of us (MVB). His last publication in 2002 dealt with the effect of LiCl on the anionic polymerization of styrene [10]. The scope of Michael Szwarc’s research illustrates the versatility and breadth of his scientific interests. The topics on which he published include the following: bond dissociation energies; gas permeation through polymer films; radical affinities of unsaturated compounds; cage reaction of radicals; living anionic and cationic polymerization; block copolymers; electron transfer reactions in aromatic hydrocarbon; electron photon-ejection from radical anions and dianions; structure and properties of ion pairs; photolysis of carbanions; protonation of carbanions by alcohols; flexibility of polymer chains; and several other topics. Below we will focus on the mechanistic aspects of living anionic polymerization, stemming from Michael’s interests in the intricate mechanisms of these reactions. Our article is not intended as a comprehensive review of living anionic-polymerization. Many of the examples were taken from his work; others describe research, much of it our own, that was inspired by his monumental work. Thus, much excellent work on acrylate-type monomers aimed at improving the living character of their anionic polymerizations is not included. Nor does it cover the numerous synthetic applications inspired by his

ARTICLE IN PRESS J. Smid et al. / Prog. Polym. Sci. 31 (2006) 1041–1067

findings. A forthcoming article in this journal will address the current state of the art in this field that continues to be an exciting area of research. 2. Discovery of living polymers The first years of Michael’s research at SUNYSyracuse were devoted to studies of gas permeation through polymeric films (with Vivian Stannett), and addition of radicals (methyl, ethyl) to a variety of aromatic and olefinic compounds (‘‘methyl’’ affinities) [11,12]. He also had a postdoctoral fellow working at Brookhaven National Laboratory on the C–C bond dissociation energies of ethylene using radioactive tracers. It was there, on one of his monthly visits, that he met Prof. Samuel Weissman of Washington University, St. Louis, MO, who mentioned that his methyl affinity data on aromatic hydrocarbons correlated well with their electron affinities. This latter research involved electron transfer between the radical anion of, say, naphthalene and another aromatic hydrocarbon such as phenanthrene. It occurred to Michael that electron transfer to styrene might yield a styrene radical anion capable of radical polymerization from one end and, simultaneously, anionic polymerization from the other end. When asked if he had tried electron transfer to styrene, Weissman replied: ‘‘It’s no use, it polymerizes’’. With Weissman’s permission and assisted by his coworkers Moshe Levy and Ralph Milkovich, Michael added styrene to sodium naphthalemde in tetrahydrofuran (THF) and concluded that the unstable styrene radical anion formed by electron transfer almost instantaneously converted to a red-colored dimeric dicarbanion capable of initiating the anionic polymerization of styrene. He coined the product a two-ended ‘‘living polymer’’. No termination or chain transfer occurred as long as carbanion-terminating impurities such as moisture, oxygen or carbon dioxide were rigorously excluded. Michael quickly realized the far-reaching ramifications of his discovery which he outlined in his first paper on living polymers published fifty years ago in 1956 in Nature [13]: the feasibility to obtain narrow MW polymers, and macromolecules with controlled architecture. His publications generated great interest and inspired researchers around the world to utilize this concept in the synthesis of well-defined polymers. These include block and graft copolymers, star-and macrocyclic polymers and telechelic polymers. It

1043

eventually earned Michael in 1991 the prestigious Kyoto Award for advanced technology, in addition to many other honors. A more detailed account of the discovery of living polymers can be found in an article entitled ‘‘Living Polymers: Their Discovery, Characterization, and Properties,’’ published by Michael Szwarc in 1998 [14]. 3. Electron transfer studies As mentioned in the preceding section, the discovery of living anionic polymerization by Michael Szwarc had its origin in the observation by Weissman that transfer of an electron from sodium naphthalenide to styrene in THF resulted in a polymeric product. Not surprisingly, a considerable body of Michael’s work in subsequent years was devoted to the unraveling of electron transfer processes involved in vinyl initiation [15–18]. Biphenyl and naphthalene radical anions rapidly transfer an electron to styrene and other vinyl monomers. This is followed by rapid dimerization of the monomer radical anion into the more stable dianion, the driving force being the formation of a new C–C bond (Scheme 1). For styrene and a-methylstyrene this leads to polymer formation, although for the latter monomer the propagating poly(a-methylstyrene) carbanion remains in equilibrium with monomer, at least above 40 1C. However, the dianion is formed exclusively when this monomer in THF is stirred on Na/K alloy [16]. A mixture of dimer- and tetramer dianion (stirring of a-methylstyrene on a sodium mirror in THF) was for many years the preferred initiator for the polymerization of styrene and its derivatives. For 1,1-diphenylethylene (DPE) the end product in the electron transfer reaction is the dimeric dianion (Ph2)CCH2CH2C(Ph2) (DD) (Scheme 1, 11) This dianion effectively initiates methyl methacrylate or vinyl pyridines but is slow with styrene due to the different stabilities of Ph2CH and PhCH 2 . In this respect the dimeric dianion of a-isopropylstyrene (Scheme 1, 12) is a very effective initiator for styrene [16]. Like DPE, a-isopropylstyrene easily forms a dianion when stirred with alkali metal in THF but, unlike a-methylstyrene, steric hindrance prevents polymer formation above 100 1C [19]. The dimerization reaction of the vinyl radical anion is a reversible process (Scheme 1), although the equilibrium is displaced far to the dicarbanion. The initiation is difficult to study for the case of

ARTICLE IN PRESS J. Smid et al. / Prog. Polym. Sci. 31 (2006) 1041–1067

1044

._

R

+ D

.-

,

M+

1-4

M+

M+ k2

+ Ar

k-1

k-2

5-8

(2)

E+

1-4

R

.(3)

(4)

9-12 Ph

E

E+ = electrophile (H

D =naphthalene, biphenyl, e etc. R=CH3 ; 3,7,11

E

R

M =Li, Na. K, Cs

1,5 , 9 R = H; 2,6,10

M+ _ R

R _

k1

(1)

9-12

R

R =Phenyl ; 4,8,12

Ph

+

, alkyl, CO2 etc.)

R = CH(CH3 ) 2 .

Scheme 1. Initiation of vinyl monomers by electron transfer from electron donors, D  .

Table 1 Kinetic data on the formation and subsequent dimerization of radical anions 6 and 7 in THF at 25 1C Rate constants:(Scheme 1) +

7/Li 7/Na+ 7/K+ 7/Cs+ 6/K+

k1 108 (M1s1) 0.009 0.19 40 4100 —

k1 (s1) 15 33 p20 — —

styrene and similar monomers because rapid polymerization occurs along with initiation. However, for vinyl monomers with lower ceiling temperatures, such as a-methylstyrene (AMS) and especially DPE this process may be studied in detail [17,18]. Thus, careful studies by the Syracuse group, involving isolation of the dianion carboxylation and protonation products, showed that the radical anion dimerization is exclusively ‘‘tail to tail’’. Furthermore, ESR and other studies showed that the fraction of the radical anions is very low relative to that of the dimer dianions. Flash photolysis was used to generate the transient radical anion potassium salt of AMS from the corresponding dicarbanion [17]. In the dark period following the flash, the rate of coupling of the radical anion was measured from the increase of the 340 nm absorption maximum of 10 and by the decay at 400 and 600 nm of 6. The occurrence of an isosbestic point at about 375 nm implied that only two species were involved. Consistent with Scheme 1, second-order kinetics for the dimerization of 6 was observed. Similar flash photolysis measurements on the DPE dianion 11, provided the rate constants (k2) of

k2108 (M1 s1)

k2 (s1)

Reference

1.0 3.5 10 30 0.1

o10 — — — —

[18] [18] [18] [18] [17]

7

dimerization of radical anion, 7, into dianion, 11, in the presence of Li, Na, K, and Cs counter ions [18]. The rates of dissociation of radical anion, 7, into the hydrocarbon and the solvated electrons and the electron transfer from the solvated electron to monomer were also determined. Significant cation effects are seen in these reactions (Table 1). The rate constants of radical anion dimerization, k2, increase about an order of magnitude in going from Li to Cs cation. This was interpreted as indicating the absence of a role of solvent separated ion pairs in the radical anions or the dianions. However, we believe that this may also be due to the presumably tighter ion pairs in the dianion compared with the radical anion. The presence of the second anion is expected to increase the contact ion pair stability. Thus, the reaction may be slowed down by desolvation of the radical ion pairs in being converted to the presumably tighter dianions. For the rate of formation of the radical anions (k1) from the corresponding solvated electrons the differences between the cations are truly enormous being at least on the order of 104, much larger than seen for the rates of radical anion dimerization (k2) and dissociation (k1). It is interesting to speculate

ARTICLE IN PRESS J. Smid et al. / Prog. Polym. Sci. 31 (2006) 1041–1067

that this may be due to especially tight ion pairing in the solvated electron ion pair (for instance Li+, e or Li+, Li ion pairs) compared to that in the radical anions, 7. The rates of dimer dissociation were too slow to allow determination of the rate constant k2, consistent with expectations. These studies also indicated that the formation of the DPE dimer radical anion is very slow compared with that of the dianion as a result of the tight ion pair character of the dianion. An analogous approach using 14C labeling [21] of the dicarbanion allowed a rough estimate of the dissociation rates of 11 (Scheme 1) into its radical anion, 7. This along with other studies confirmed that radical anions do not play a role in the polymerization process. Isotope labeling was used in determining the dissociation of 10 into 6 [20]. The a-methylstyrene dianion potassium salt with deuterated CH3 and CH2 groups was allowed to exchange with nonlabeled 10 (Scheme 2). Since the dissociation is rate determining relative to the coupling reaction, the rate of formation of the mixed dianion can be obtained. The above at least in principle is applicable to the initiation of styrene and its derivatives by electron transfer in THF and similar solvents and anticipates the profound ion pairing effects illustrated in Sections 4 and 5. 4. Anionic polymerization of styrene and other vinyl monomers 4.1. Polymerization of styrene The stability of the propagating carbanionic chain end greatly facilitates mechanistic studies of living anionic polymerization. The active chain-end

K H3C _

+

+

K _ CH k-2 3

C6H5 C6H5

concentration can be determined spectrophotometrically or by other means. It permits the determination of absolute propagation rate constants, and, for copolymerization reactions, crossover rate constants [4]. Szwarc initially chose two-ended polystyrylsodium (PSNa) in ethereal solvents for his kinetic studies. The sodium counterion and styrene monomer turned out to be a fortunate choice, since in retrospect it gives the least complications. For example, the two polystyryl sodium chain ends do not interfere with one another, unlike that for twoended polystyryl cesium (vide infra). Also, the rate of polymerization of styrene can be followed by the UV absorption of styrene. Monomers other than styrene often depict a more complex behavior. The most straightforward system is that of polystyryl salts in dioxane where the reaction is strictly first order in both carbanion (‘‘active ends’’) and monomer, and slow enough to follow by a conventional batch method. Only tight ion pairs appear to exist in this low dielectric constant solvent. Propagation rate constants increase with counterion size: kp (M1 s1, 25 1C) equals 0.9 for Li+, 3.4 for Na+, 20 for K+ and 24.5 for Cs+ [22]. The carbanionic end is more nucleophilic and, therefore, more reactive in the presence of a larger alkali counter cation. This is plausibly due to the weak solvation of the counterions by dioxane that has both a low dielectric constant (D ¼ 2.2) and a dipole moment that is zero in the chair conformation. The pattern in higher dielectric constant ether solvents such as THF or DME is more complex. Reactions are very fast, often taking less than a second to complete, and require the use of fast flow techniques [23]. Apparent propagation rate

+ + K . . K _ _ CD3 k CH2 CD2 H3C -2 + C6H5 C6H5

6

10

1045

6-d5 k2

+

+ K K D3C _ CD2 CD2_ CD3

C6H5

C6H5

10 -d10

K+ H2 D2 K+ H3C _ C C _ CD3 10-d5 C6H5

C6H5

Scheme 2. Determination of rate constants of dissociation of dianion 10 into radical anion 6 in the presence of potassium ion at 25 1C obtained by deuterium exchange.

ARTICLE IN PRESS J. Smid et al. / Prog. Polym. Sci. 31 (2006) 1041–1067

1046

constants (kap) increase at lower active end concentration due to the presence of a small fraction of very reactive free ions. Assuming a free ion fraction less than 0.1, kap is given by [23] 1=2

kap ¼ ki þ ðkf  ki ÞK d C 1=2 , o

(1)

where Co is the total active end concentration, Kd the ion pair dissociation constant and ki and kf the respective propagation rate constants for ion pair and free ion. Fig. 1 shows that plots of kap versus C1/2 for polystyrylalkali salts are straight o lines, yielding values for kf K1/2 d (kf bki). Conductivity measurements in THF at 25 1C gave the following values for Kd  107 M: 1.86 (Li+), 1.52 (Na+), 0.71 (K+) and 0.021 (Cs+). The free ion rate constant, kf, in THF at 25 1C was found to be 65,000 M1 s1. Reliable ki values can not be obtained from the Fig. 1 plots. However, adding excess counter cations in the form of the alkali tetraphenylborate salt retards the propagation reaction by decreasing the concentration of free polystyryl ions. The common ion effect leads to the expression: þ 1

kap ¼ ki þ ðkf  ki ÞK d ½M  .

(2) +

K1/2 b

The free cation concentration [M ] ¼ 1/2 [M+BPh 4 ] , Kb being the dissociation constant of the borate salt. Plots of kap versus [M+]1 are linear with the intercepts yielding reliable ki values

[23]. Combining the respective slopes kf K1/2 d and kf Kd of plots in the absence and presence of borate salt gives kf and Kd. The Kd values agree with those obtained from conductance data. The order of ion pair propagation rate constants for polystyryl salts in THF at 25 1C is opposite to that found in dioxane: ki (M1 s1) ¼ 80 (Na) and 22 (Cs). Furthermore, as the temperature decreased from 25 to 60 1C, the reactivity of the Cs ion pair decreased whereas that of the Na ion pair increased from 80 to 270 M1 s1 [24]. Based on the discovery of solvent-separated (ssip) or ‘‘loose’’ carbanion pairs in ether solutions [25–27], it was suggested that the anomalous reactivity of the PSNa at lower temperature was caused by an increase in the fraction of the more reactive ssip (see below). As explained in the next section, the formation of loose carbanion pairs from tight ion pairs typically is an exothermic process. This may cause a peculiar temperature dependence of the apparent ion pair rate constant ki,app. It can be readily shown that [24,26]: ki;app ¼ ks K i ð1 þ K i Þ1 ,

where ks is the rate constant for the loose ion pair and Ki the equilibrium constant for the conversion of tight into loose ion pairs. This assumes that kc5ks Ki, where kc is the rate constant for the tight ion pair. The apparent activation energy is then given by DE app ¼ DE s þ DH oi =ð1 þ K i Þ,

Fig. 1. Plots of the apparent propagation rate constants of polystyryl salt in THF at 25 1C as a function of [living ends]0.5 [23].

(3)

(4)

where DHi1 is the enthalpy change on forming the loose polystyryl alkali ion pair from the tight ion pair, and DEs denotes the activation energy of the loose ion pairs. This prediction was verified in ethers by Szwarc [24,28], Schulz [29–31], and coworkers. For example, in several ether solvents the Arrhenius plots for the apparent ion pair rate constants of PSNa exhibit a sigmoidal type of curve (Fig. 2) [29,31]. Apparent negative activation energies are observed in the temperature range for which the ratio DHi/DEs41+Ki. This occurs for PSNa in THF between 20 and 60 1C, and in DME between 48 and 0 1C. The higher rates in THF and DME relative to those in MeTHF or oxepane are in agreement with the greater tendency of the former two solvents to form reactive PSNa loose ion pairs. The nearly identical activation energies for the tight ion pairs (dominant at high temperatures) and loose ion pairs (dominant at low temperatures) imply that

ARTICLE IN PRESS J. Smid et al. / Prog. Polym. Sci. 31 (2006) 1041–1067

1047

Fig. 3. Polymerization rate constants of living PSLi and PSNa ion pairs in THF as a function of temperature [3].

Fig. 2. Arrhenius plots of the rate constants of living PS Na ion pairs as a function of temperature in different media [29].

the reactivity of these polymerization systems are largely determined by specific solvation effects, and that differences in dielectric constants are not an important factor. Noteworthy is also the remarkable agreement between data published by Szwarc et al. and those from the group of G.V. Schulz in Mainz, in spite of different experimental approaches [28–31]. In contrast to PSNa, the ion-pair rate constant for PSLi in THF, interestingly, does not show a ‘‘negative’’ activation energy, but only a small decrease in activation energy at lower temperatures, i.e. a slightly curved Arrhenius plot, as shown in Fig. 3 [3]. The ion-pair dissociation constant of PSNa also increases much more with decreasing temperature (exothermic dissociation), while that of the Li salt is almost thermo-neutral, leading to a much lower dissociation constant for the Li+ salt than for the Na+ salt in the lower temperature domain [3]. This behavior is also seen with fluorenyllithium (FlLi) and FlNa in THF [27]. This could be due to the small size of the Li cation as well as to the

relatively localized benzylic anion (Section 5) forming a tight but externally solvated and hence less reactive ion pair. At higher temperatures the PSNa ion pair tends to loose some of its solvation. At low temperature, the larger Na ion favors the formation of loose pairs, making the PSNa much more reactive than the PSLi. In relatively weakly cation-coordinating solvents, such as tetrahydropyran, loose ion pairs can be generated by adding multi-dentate cation binding ligands such as tetraglyme (TG), i.e., CH3(OCH2 CH2)4OCH3. Plots of kapp vs.[LE]0.5 for PSNa at constant [TG] are linear (Fig. 4) [32,33]. Reliable ki intercept values can again be obtained by adding NaBPh4 (Eq. (2)) and these increase with increasing [TG] due to the formation of increasing fractions of TG-separated ion pairs for which the ki value is found to be 3900 M1 s1, considerably below that of 20,000 M1 s1 for the THF or DME-separated ion pairs (Fig. 4). This may be caused by a shorter inter-ionic ion pair distance in the TG-separated ion pairs. Alternatively, TG addition may form a mixture of tight and loose ion pairs (see below). The former are favored in less polar solvents (e.g., THP), or when dealing with less delocalized carbanions (polystyryl versus fluorenyl, see also Section 5). It could also be ascertained that the ‘‘loose’’ PSNa ion pair and the free Na+ ion are

ARTICLE IN PRESS 1048

J. Smid et al. / Prog. Polym. Sci. 31 (2006) 1041–1067 Table 2 Anionic homopolymerization rate constants of vinyl monomers in THF at 25 1C counterion ¼ Na+, [Anion]E3  103 M, [Monomer]E102 M [92] Monomer

kp

Monomer

M1 s1 Ref. Vinyl mesitylene a-Methylstyrene p-Methoxystyrene o-Methylstyrene p-(t-Butyl)styrene

Fig. 4. Plots of the apparent propagation rate constants of sodium polystyryl in THP at 25 1C as a function of [living ends]0.5 in the absence and presence of triglyme. Lines with half open circles are experiments with added NaBPh4 [32].

complexed to one tetraglyme molecule but to two triglyme molecules, consistent with results obtained with fluorenyl salts (Section 5). 4.2. Other vinyl monomers Exploratory studies on the propagation rate constants for the anionic polymerization of other vinylaromatic monomers in THF at 25 1C with Na as counterion are listed in Table 2 [34–38]. Because kp is concentration dependent, they are compared at about the same anion concentration (3  103 M). As expected, the polymerization is slower with electron-donating para-substituents in the phenyl group while the reaction speeds up with electronwithdrawing groups (e.g., vinylpyridines). However, crossover rate constants provide a better measure for monomer reactivities (see below) since the values of Table 2 do not account for differences in the ionpair dissociation constants of the propagating active ends. Polymerization of vinylmesitylene is slow due to steric strain in the monomer. Conjugation between double bond and phenyl group is adversely affected and reduces the reactivity of the double bond. The slow rate for a-methylstyrene is a result of increased electron density on the double bond as well as steric hindrance. This system was studied

0.9 2.5 52 170 220

[36] [34] [35] [35] [35]

kp M1 s1 Ref.

p-Methylstyrene Styrene 1-Vinylnaphthalene 2-Vinylpyridine 4-Vinylpyridine

210 950 850 7300 3500

[35] [24] [37] [38] [38]

extensively by Worsfold and Bywater [34] using dilatometry. Strain in the polymer chain, leading to a low ceiling temperature of the polymer, results in high equilibrium monomer concentrations at room temperature. The ion pair dissociation constant for poly(2vinylpyridyl) sodium (P2-VPNa) in THF is substantially lower than that for PSNa [39] in spite of increased charge delocalization into the aromatic ring (see also Section 5). The degree of ion pair dissociation is much lower as a result of strong interactions between the cation and the pyridine rings of the terminal and penultimate units [38]. It is interesting to note that when the penultimate unit in P2-VPNa is replaced by a styrene unit, kp increases from 7300 to 30,000 M1 s1, probably because of a higher concentration of reactive free (P2-VP) anions [38]. Another interesting observation is that the ion pair rate constant, ki, for P2-VPNa is essentially the same in dioxane, THP and THF contrary to that found for PSNa. This is most likely due to a negligibly low fraction of loose ion pairs, and may in part result from the fact that ssip formation is not known in 2-pyridyl-anions (Section 5). Furthermore, intra-molecular coordination of the Na ion by the pyridine groups is presumably strong so that competitive Na ion solvation by THF should not be favorable. 4.3. Triple ion formation In the concentration range used in polymerization studies, especially when the dissociation of ion pairs is small, triple-ions may form as shown by conductance measurements. Positive and negative triple-ions may exist: 2Bþ A Ð ðBABÞþ þ A

(5)

ARTICLE IN PRESS J. Smid et al. / Prog. Polym. Sci. 31 (2006) 1041–1067

2Bþ A Ð ðABAÞ þ Bþ

(6)

Kþ

Intramolecular formation of negative triple ions of living polymer anions was first observed in the cesium salt of polystyrene possessing two active chain ends [40]. The propagation of such polymers is slower than that of one-ended PSCs, whereas for the same concentration of active end-groups the conductance of their solutions is higher. This phenomenon was accounted for by ion-pair dissociation into free ions and by intramolecular cyclization to give triple anions (Eqs. (7) and (8)): + -

Cs , S

+ -

Cs , S

-

S , Cs

S

+

-

+ -

-

Cs , S

-

S + Cs

+

S, Cs ,S

-

+

Liþ þ ðLiClÞ2 Ð ðLiClÞ2 Liþ

Dissociation (Eq. (9)) increases the concentration of Li ions, which, in turn, suppresses the dissociation of PSLi into Li and PS ions by a common ion effect, thus resulting in a rate decrease since the PS anions are the main contributors to the rate. Reaction (10), on the other hand, reduces the Li ion concentration, thereby increasing the dissociation of PSLi and therefore the polymerization rate. The propagation rate of PSLi in THP is retarded by the addition of tertiary lithium butoxide (tBuOLi) but increased by adding n-BuOLi [44].

(7)

ðBuOLiÞ4 Ð Liþ þ ðBuOLiÞ3 BuO

(11)

(8)

Liþ þ ðBuOLiÞ4 Ð ðBuOLiÞ4 Liþ

(12)

The latter reaction consumes free PS anions, thereby decreasing the polymerization rate. The replenishment of the free anion by dissociation (Eq. (7)) results in a higher concentration of the Cs ions, which, in view of their greater mobility, now increases the conductance. Positively charged triple ions (Eq. (5)) were proposed by Sigwalt et al. for P2VPNa in THF, for which the dissociation, as mentioned above, is very low [41]. On the other hand, Kahn and HogenEsch showed that, for the case of purified 1-lithio-1(2-pyridyl)ethane (LiEP) model anion in THF, 7Li NMR in the 70 to 90 1C range showed two small but equal intensity Li peaks, slightly upfield from the main ion pair band with coalescence occurring at about 60 1C [42]. The relative intensities of the three bands were shown to be consistent with conductance measurements, indicating that the peaks are due to triple anions (Eq (6)). The formation of the (Li2, 2EP)+ triple ion would predict equivalent Li peaks. More recently it was shown that the addition of LiCl to a solution of PSLi in THP reduces the rate of propagation at low PSLi concentration but accelerates it at higher [PSLi] [43]. Moreover, the addition of LiCl, shown to be dimeric, increased the conductance. These observations were fully accounted for by the formation of (ClLiCl) triple anions (Eq. (7)) and by the ability of the dimers to scavenge Li+ ions forming quintuple ions (Eq. (8)): ðLiClÞ2 Ð Liþ þ ClLiCl

1049

(9) (10)

This was accounted for by the predominant effect of the dissociation (Eq. (11)) of t-BuOLi tetramers suppressing the ionization of PSLi by a common ion effect, thus diminishing the concentration of the reactive PS free anions. However, for n-BuOLi tetramers reaction 12 is more important than that of t-BuOLi tetramers (see below). This reduces the Li ion concentration and enhances the PSLi dissociation, thus yielding more of the reactive free PS anions. The equilibrium constant of Eq. (12) was shown to be almost 30 times greater than that of tBuOLi, probably because of steric hindrance in the t-BuOLi tetramers. These explanations were further substantiated by density functional theory calculations [45]. Although the influence of butoxides and R2Mg or R3Al, resulting in bimetallic complexes in non-polar media will not be discussed in this section, it was recently shown by Deffieux et al. that the rate of the Li ion mediated polymerization of styrene can be drastically reduced by the addition of R2Mg or R3Al allowing the controlled polymerization in bulk and at high temperature [46]. The nature of this effect will be discussed in a future paper in this series. Finally, the anionic styrene polymerization mediated by alkaline earth cations is of interest. For instance, surprisingly, the first-order propagation rate constant of the Ba salt of one-ended living polystyrene is PS2Ba concentration independent [47]. This was explained by assuming two modes of ion formation: the usual monomolecular dissociation (Eq. (13)): PS ; Ba2þ ; PS Ð PS ; Ba2þ þ PS K diss

(13)

ARTICLE IN PRESS J. Smid et al. / Prog. Polym. Sci. 31 (2006) 1041–1067

1050

yielding a free PS anion and a PS, Ba2+ cation and a bimolecular reaction: 2PS ; Ba2þ ; PS Ð PS ; Ba2þ þ ðPS Þ3 Ba2þ K þBa

1=2

½PS  ¼ K diss =K t;Ba 

(14) (15)

2+

species apparently acts as a The (PS )3Ba buffering agent, keeping the free PS anion at a very low concentration and independent of [PS2Ba] over a wide concentration range thus making the observed propagation rate constant [PS2Ba] independent. This is consistent with the free PS anion being the only contributor to the propagation. Its concentration for a low degree of dissociation of the salt and for Kt,Ba [(PS)2 Ba2+]bKdiss is given by Eq. (15). Addition of barium tetraphenylborate stops polymerization completely because it results in the formation of the mixed salt Ba(PS,BPh4). Upon dissociation this salt yields Ba2+, PS and a BPh 4 ion thus eliminating free PS ions from the solution [47]. Apparently, the (PS)2Ba species has a negligible reactivity compared to the PS alkali salts and the negative triple ion also should have a very low reactivity. The kinetics of polymerization of the strontium salt of monofunctional polystyrene can be explained by the same mechanism [48]. Living polymers of this type with two active end groups per chain (M2+, PS2) tend to form rings for which dissociation results in ring opening giving a non-conducting ion pair [49]. In this case the polymerization kinetics is more complicated. The conversion vs. time plots are sigmoidal indicating that the polymerization accelerates as the reaction progresses. This is expected since the concentration of active open chains increases as the rings become larger.

anion decreases and makes the anion more nucleophilic. However, the magnitudes of ion pair rate constants in THF and DME seemed to defy simple rationalization. It was realized at the time that in these strongly coordinating solvents this is due to cation solvation. At the time, exploratory work was being carried out on the UV-visible spectroscopy of fluorenyl salts for use as initiators for the polymerization of MMA. Initial studies indicated that these spectra were indeed well worth studying. Thus, the Na salt of fluorene (FlNa), prepared by reacting fluorene with living PSNa in THF, shows a maximum at 356 nm at ambient temperature. Below room temperature a new band at 373 nm emerges that increases with decreasing temperature while the band at 356 decreases [25,26]. At 30 1C the peaks at 356 and 373 nm are about equal (Fig. 5). Further cooling to 78 1C transforms almost all of the original absorption at 356 nm into the 373 nm peak. These changes were fully reversible upon heating and re-cooling. The equilibrium between the two species is concentration independent, at least above 104 M, and unaffected by addition of excess Na+ ions in the form of sodium tetraphenylborate. Moreover, conductance studies of Li, Na and Cs

5. Ion pair solvation 5.1. Fluorenyl anions The ion pair propagation kinetics for the alkali ion mediated anionic polymerization of styrene in THF and similar media obtained in the early sixties by Szwarc and coworkers were intuitively reasonable. Thus, as indicated in the previous section, the rate constants for the PS ion pairs in a low polarity solvent, dioxane, showed rate constants that increased with the ionic radius of the counter cation. As cation size increases, its perturbation of the

Fig. 5. Absorption spectrum of fluorenyl salts in THF at 30 1C as a function of counterion [26].

ARTICLE IN PRESS J. Smid et al. / Prog. Polym. Sci. 31 (2006) 1041–1067

fluorenyl salts at concentrations around 103 M preclude the presence of appreciable fractions (o 0.10%) of free anions [27]. At that time, so-called ssip’s had been invoked by Winstein [50] in the solvolysis of arene sulfonates and by Cram [51] to account for the electrophilic substitution at saturated hydrocarbon but had not been observed directly. It was soon realized that the species seen at 373 nm corresponds to ssip and not to ion pair aggregates or free anions and that the 356 nm maxima corresponds to that of a contact ion pair (cip). In contrast to the ssip’s, the maximum of the cip’s increases with increasing cation size from 349 nm for Li+ to 364 and 368 nm for Cs+ and tetrabutylammonium ion, respectively. The tendency to form ssip’s is directly related to cation size with the fraction of these species decreasing in the order Li4Na4K4Rb4Cs. Thus, for FlLi in THF at room temperature their fraction is about 0.80 while that of FlNa is only about 0.05 (Table 3). The fluorenyl salts of K, Rb and Cs in THF did not show evidence of the 373 nm species at ambient temperatures although there is evidence for their formation in more strongly solvating media at lower temperatures. As solvation is involved, a presumably rapid equilibrium exists between the two types of ion pairs (Scheme 3). We soon realized that these data could be used to establish semi-quantitative tendencies of various aprotic solvents to solvate specific cations as illustrated in Table 3.

1051

Conductance studies of Li, Na and Cs fluorenyl salts confirm the formation of loose ion pairs. Scheme 3 leads to the relationship [27]: K d ¼ K i K s ð1 þ K i Þ1 .

(16)

The ion pair dissociation constant Kd, is defined by Kd ¼ [Fl] [M+]/{[Fl M+]+[FlJM+]}. The results show that the dissociation of FlJM+ can be approximated by a ‘‘sphere in continuum’’ model, the two ions being separated by 6.3 A˚ for FlJLi+ and by 7 A˚ for FlJNa+ [27,52]. Differentiation of Eq. (16) with respect to 1/T yields: DH od ¼ DH oS þ DH oi ð1 þ K i Þ1 .

(17)

The enthalpy of the cip to ssip conversion (DHi) of Fl Na+ in THF equals 32 KJ/mol. Furthermore, the apparent enthalpy of dissociation, DHd, changes from about 40 KJ/mol at 25 1C, where Ki51 (predominantly cip’s), to around 5 KJ/mol below 78 1C, where Kib1 and DHdEDHs (predominantly ssip’s). These changes are not related to Ki -

Fl M

Fl // M +

+

K d,s

Kd,c Fl

+

M+

Scheme 3. Dissociation equilibria of contact and solventseparated ion pairs.

Table 3 Effect of cation and solvent or complexing agent on the formation of solvent-separated ion pairs (ssip) of fluorenyl salts at 25 1C M+ Li

a

Solvent/complexing agent

Fraction of SSIP

M+

Solvent/complexing agent

Fraction of SSIP

Dioxane 2-MeTHFa 3-MeTHF THP THF 2-MeOMeTHFb Oxetane DME Dioxane/DG [0.31 M]c Dioxane/TrG [0.0089 M]c Dioxane/TG [0.0041 M]c

o 0.02 0.25 0.47 0.30 0.80 40.98 40.98 40.98 0.5 0.5 0.5

Na

Dioxane 2-MeTHF THF DME DMSO

o 0.02d o 0.02d 0.05 0.95 1.0

K

THF DME

o 0.02 0.10

Cs

THF DME

o 0.02d o 0.02d

d

2-methyl THF. 2-methoxymethyl THF. c DG ¼ diglyme (glyme-3); TrG ¼ triglyme (glyme-4); TG ¼ tetraglyme (glyme-5) where the numbers denote the number of oxygen atoms. The brackets denote concentration of complexing agent for which [cip] ¼ [ssip]. d Approximate fraction of ssip that may be detected. b

ARTICLE IN PRESS 1052

J. Smid et al. / Prog. Polym. Sci. 31 (2006) 1041–1067

the increase in dielectric constant of THF at lower temperature. Thus, changes in the shape and width of the spectra, again, were consistent with equilibrium between distinct cip and ssip (Scheme 3). The presence of cip’s and ssip’s distinguishable by UV was also seen in naphthalene radical anions by UV/ visible spectroscopy and by ESR studies of sodium naphthalenide THP solutions containing TG [53]. Fluorenyl salts of divalent counter cations exhibit an interesting stepwise solvation [54,55]. For example, two equilibria are found between 20 1C and 100 1C for bis (9-fluorenyl)strontium [55]: Fl Sr2þ Fl Ð Fl jjSr2þ Fl Ð Fl jjSr2þ jjFl

(18)

The second step requires more energy because, following the first solvation step, the electrostatic interaction of the Fl anion and the divalent Sr2+ ion is increased. Its enthalpy (DH ¼ 12 kJ/mol) is much less exothermic than the DH ¼ 50 kJ/mol for the first step. Even at 100 1C the predominant species is the mixed tight/loose ion pair. Also, addition of dibenzo-18 crown-6 to bis(9-fluorenyl)barium gives an ion pair (actually an ‘‘ion triple’’) in which the barium forms a contact ion pair with one fluorenyl unit but is separated from the other Fl anion by the crown ether [54]. Pressure is another factor affecting the cip/ssip equilibria [56,57] Thus, the fraction of ssip increases when pressure is applied on a solution of Fl M+ in THF. The effect is caused by electrostriction, i.e., tight packing of solvent molecules around the cation of the loose ion pair. The respective DV values were found to be 24 ml/mol for Fl Na and 16 ml/mol for FlLi. 5.2. Solvent effects Formation of the ssip is driven chiefly by the coordination of the cation with the solvent or complexing agent as well as the electrostatic interaction with the anion. Since cation–solvent interactions, generally, are stronger with the smaller cations, Ki decreases with increasing cation size in spite of decreased Coulombic interactions in the ion pair. For unsubstituted cyclic ethers the increase in the fraction of loose ion pairs closely relates to the basicity of the solvent and its dipole moment rather than to dielectric constant. Multidentate ligands such as the dimethyl poly(ethyleneglycol) ethers of the general formula CH3O(CH2CH2O)nOCH3 (glymes) and even more so the macrocyclic polyethers (crown ethers) are powerful complexing

agents for alkali and alkaline earth cations because of the lower entropy of cation coordination. Not surprisingly, small quantities of such powerful coordinating agents added to fluorenylalkali salts (e.g., Fl Li in dioxane, Fl Na or Fl K in THF) convert tight into ‘‘loose’’ ion pairs [58–60]. The spectral changes observed in these solutions are shown in Fig. 6. In addition to the loose ion pairs (either a 1:1 complex, Fl G M+, or a 2:1 complex, Fl G M+ G) the glyme can form a glymated tight ion pair, Fl M+ G. In most instances a mixture of two or all three species are present. For example, glyme-5 (the number refers to the number of oxygen atoms in the chain) and glyme-6 form 1:1 glyme-separated ion pairs with Fl Na+ in THF or Fl Li+ in dioxane, while glyme-4 forms a 2:1 loose ion pair complex with Fl Na+ in THF. Glyme-6 and glyme-7 form a mixture of 1:1 glyme-complexed tight and loose ion pair with Fl K+ in THF, while glyme-5 form a 1:1 tight ion pair complex and a 2:1 loose ion pair complex. Similar complexes are found with crown ethers [61–64]. 5.3. Ion pairing effects in other carbanions The formation of ssip involves the (endothermic) separation of anion and cation to allow an increased degree of solvation (exothermic) by the cation. Evidently, for the case of relatively soft or sterically hindered anions, the inter-ionic Coulombic interaction is sufficiently weak to allow this solvation, thus giving two potential energy minima corresponding to the two types of ion pairs. Charge delocalization and bulky substituents close to the anionic site favor ssip formation. The substituent effect is apparent when comparing 9-alkylfluorenyl salts with fluorenyl itself. For example, Ki ¼ 0.33 for Fl Li+ in 2-MeTHF at 25 1C, while in this solvent its value is 1.5 for 9-(2-hexyl)-Fl Li+ [58]. The fraction of ssip for the tetraphenylborate salts is shown to be somewhat larger than those for the fluorenyl salts. Given the localized negative charge on boron this also appears to be a steric effect. Additional insight into the effects of charge delocalization in hydrocarbon aromatic anions is provided by comparisons of ion pairing in indenyl (In), fluorenyl (Fl), 2,3-benzofluorenyl (BFl), fluoradenyl (FD) and other carbanions (Scheme 4) [65]. For instance, while FlLi in DME is a ssip at room temperature, that of the corresponding InLi in DME is largely a cip. This latter system forms 50%

ARTICLE IN PRESS J. Smid et al. / Prog. Polym. Sci. 31 (2006) 1041–1067

1053

Fig. 6. Spectral changes on addition of glymes to fluorenyl salts at 25 1C [60].

-

-

-

-

Fl

BFl

In

P

-

-

P

H

H

FD

-

P

3

2

-

-

Ph

2

-

CH3 OCH3

O N

7

6

5 2 . P =PS

3

P =P2VP

-

P

2

2

Ph

4 P

Ph

N

1 P=PS

-

N

N

1

H

7 P =PMMA;

H

CN

8

8 P = PBD

PS = polystyrene; P2VP =poly(2-vinylpyridine); PBD =polybutadiene PMMA = polymethylmethacrylate Scheme 4. Carbanion structures used in ion pairing studies.

ssip only at 30 1C. Whereas FlNa in THF has only about 5 percent cip’s, the FDNa in the same solvent forms about 50 percent ssip’s. Another example is ssip formation in THF solutions of 1,4-dilithio1,1,4,4-tetraphenylbutane. It indirectly shows the formation of ssip by a gradual and reversible red shift from about 450–500 nm as the temperature is decreased from 50 to 45 1C [26]. The bandwidth goes through a maximum and there is no further red

shift below 45 1C, consistent with the predominance of ssip’s below that temperature. As shown in Table 4 the ion pair dissociation constant, Kd, decreases 25 and 170-fold upon introduction of nitrogen at the 2- and 4- positions, respectively, of the phenyl ring, consistent with appreciable localization of negative charge onto these nitrogens (compare anions 1–3 and 4–6) [65]. Following Eq. (16) this indicates the virtual absence

ARTICLE IN PRESS J. Smid et al. / Prog. Polym. Sci. 31 (2006) 1041–1067

1054

Table 4 Effects of carbanion structure on ion pair dissociation constants Kd, of two ended living polymers in THF at 25 1C (Scheme 4) Anion

M+

109Kd

Ref.

Anion

M+

109Kd

Ref.

1 2 3 4

Na Na Na Na

150 6.2 0.87 830

[27] [39] [39] [73]

5 6 7 8

Na Na K Na

3.7 1.9 0.2 4.0

[73] [73] [75] [76]

of ssip on these ‘‘hard’’ anions incapable of forming appreciable fractions of ssip’s. The low Kd values of anions 5–8 indicate a similar charge localization of negative charge onto the nitrogens of anions 5, 6 and 8 and on the carbonyl oxygen of 7. The Coulombic interaction of the highly charged heteroatoms with the counterions appears to be too strong to allow extensive solvent separation that is common for the relatively delocalized hydrocarbon anions 1 and 4. An interesting consequence of a change in anion delocalization is provided by the UV/visible emission of FlNa and other fluorenyl salts in THF and other ethers [66,67]. The FlNa contact ion pair, upon excitation, gives emission from the lowest excited (S1) state that is much more delocalized compared to the ground state. As a result the carbanion emission corresponds to that of a ssip as the fluorenyl anion life time (E108 s) is orders of magnitude longer than the time required for solvent reorganization. Given the comparatively localized polystyryl anion it is then not surprising that PSNa in THF at 40 1C is still predominantly a tight ion pair, although a significant red shift in its optical spectrum from 340 to 347 nm can be detected on cooling the solution to 80 1C [3,6]. Apart from the polymerization kinetics discussed in Section 4, spectroscopic evidence for ssip’s for a-methylstyrene (AMS) type polymer anions in THF was given by Comyn and Ivin [68] and for PSNa by Van Beylen and coworkers [69,70]. In contrast to the fluorenyl salts, in these cases no separate maxima for cips and the ssips are observed. Upon lowering the temperature, PSNa in THF at 70 1C shows a red shifted broad shoulder (Fig. 7). These changes are reversible with temperature as observed for other carbanions, consistent with ssip formation. This was corroborated by the polymerisation kinetics (Section 4) but also by the fact that the DHi and DSi values for the formation of ssip’s from cip’s (28.9 kJ/mol and 150 J/mol K, respectively) [70] were nearly identical to those derived

Fig. 7. Near UV spectrum of polystyryl-lithium (left) and sodium at 60 and 70 1C, respectively.

from kinetic and conductometric measurements by Schulz et al. [71]. PSLi was not found to form an appreciable amount of ssip’s by spectroscopy or by polymerization kinetics (Section 4). This is probably due to the fact that in the presence of the small Li ion much of the charge of the styryl anion resides on the benzylic C atom. As the size of the counter ion increases more delocalization in the ring occurs thus facilitating ssip formation. For PAMSLi in THF there appears to be more ssip formation due to greater charge delocalization caused by the presence of the methyl group as pointed out by Bywater [72] on the basis of 13C measurements of PSLi and PAMSLi under similar conditions. 5.4. Ion pairing effects on equilibria The effects of suitably placed heteroatoms in benzyl and similar anions can be even more dramatic. For instance whereas the DD2Li salts

ARTICLE IN PRESS J. Smid et al. / Prog. Polym. Sci. 31 (2006) 1041–1067

in THF contain large fractions of ssip’s, that of the corresponding 1-phenyl-1-(2-pyridyl) or 1-phenyl-1(4-pyridyl)anions (Scheme 4) are cip’s only, with the 4-pyridyl anions giving the tighter ion pairs as indicated by conductance and other studies [73,74]. Eq. (16) allows the assessment of the fraction of ssip, assuming the absence of ion pair aggregation and the ssip dissociation constant, Kd,s, being anion independent. Thus when ssip formation is complete, a large value of Kd,s (order of magnitude of 105 M) is expected [27]. For values of Ki51, Kd,a will be on the order of Kd,s/Ki so that a rough estimate of Ki may be made. These conclusions are applicable to anions involved in the polymerization of 2-vinylpyridine [74], MMA [75,76], acrylonitrile [77] and similar anions where there is strong delocalization onto the heteroatom (‘‘hard’’ anions). As shown in Table 5, the corresponding dissociation constants are in the order of 109 M for typical ‘‘hard’’ pyridyl or enolate anions (Scheme 4), indicating the virtual absence of ssip. At least in typical ether solvents, the predominant species should all be cip’s, regardless of temperature. Evidently, the enthalpy of ion pair separation is either positive or insufficiently negative to allow significant ssip formation. In accordance with this no negative activation energies seem to have been reported for any anionic polymerization involving such anions. Consistent with this, straight line Arrhenius plots have been obtained for the anionic polymerization of MMA for all cations studied [77,78]. Suitably placed heteroatoms can drastically affect ion pair equilibria, for instance for anionic initiation reactions and the rates of copolymerization (Section 6). As illustrated in Table 5 and Scheme 5 the equilibrium of addition of 9-methylfluorenyl Li, Na and Cs salts to a non polymerizable monomer 1-phenyl-1(4-pyridyl)ethylene (4-PPE) is highly cation/solvent dependent. [79] It can readily be

1055

shown that K ip ¼ K i :K d1 =K d2

(19)

where Kd1 and Kd2 are the dissociation constants of the fluorenyl (‘‘initiator’’) and 4-PPE ion pair, respectively, and Kip and Ki are the equilibrium constants for the ion pairs and the free anions. The ionization into free anions is negligible and the experimental equilibrium constant value is close to the ion pair equilibrium constant Kip. As seen from Table 5 there are very large differences in Kip for the various cation/solvent systems that correlate well with the Kd1/Kd2 ratios. The equilibrium is clearly dominated by the relative stabilities of the respective ion pairs. Thus, in the presence of the Li ion the equilibrium favors addition, presumably due to the very strong interactions of the two ‘‘hard’’ anions, whereas in the presence of the much larger Cs ion the equilibrium favors the fluorenyl initiator. In the presence of Li ion and THP, a poorer solvent, the initiation is favored even more whereas in the presence of Na ion and a crown ether (dibenzo -18 crown -6) the equilibrium is shifted to the left. This should have effects on the kinetics also. A similar rationale would appear to explain the much more favorable initiation of MMA by FlLi compared with the salts of other alkali cations [76]. 5.5. Dynamics of interconversion of ionic species In order to investigate dissociation–association kinetics of electrolytes in non-aqueous media, Persoons [80] developed a relaxation technique based on the second Wien effect [81]. An electric current flowing through a conductance cell is perturbed by a variable, high strength electric field. This increases the conductance by increasing the conversion rate of ion pairs into free ions, the ion association rate being independent of the field K ip

Table 5 Effects of cation on initiation equilibrium of 4-PPE with 9methylfluorenyl anions in THF and THP at 25 1C (Scheme 5) M+

Solvent

Kip

Li Li Na Na/CE Cs+

THP THF THF THF THF

14,200 680 3.0 0.40 0.05

108Kd1

108Kd2

MFl, M + 4-PPE Kd1

0.60 2.0 30 19

Kd2

Kd1/Kd2 -

390 96 370 1.6

MFl-4-PPE, M

650 48 12 0.08

MFl + M + +

Ki

-

MFl-4-PPE + M +

4-PPE Scheme 5. Effect of counterion on the equilibrium constants of addition of MFL to 4-PPE.

ARTICLE IN PRESS J. Smid et al. / Prog. Polym. Sci. 31 (2006) 1041–1067

1056

strength. The increase in the dissociation equilibrium constant Kd with increasing field strength can be measured by the change in conductance. The field-dissociation effect will be most pronounced in low polarity media. The relaxation time, t, of a simple dissociation association ion pair equilibrium (Eq. 20) kd

Aþ B Ð Aþ þ B

(20)

t1 ¼ kd þ 2kr C ion

(21)

kr

is given by Eq. (21) where Cion is the equilibrium concentration of the free ions. For low degrees of dissociation, kd52 kr  Cion and Cion ¼ (KdC)1/2, C being the total concentration of electrolyte. Eq. (21) can now be written as t1 ¼ 2kr ðK d CÞ1=2 þ 2ðkd kr Þ1=2 C 1=2 1

(22)

1/2

so that plots of t vs. C should yield straight lines the slopes of which permits the determination of kd and kr. Relaxation studies for fluorenyllithium (FlLi) in diethylether (DEE) and its mixtures with THF as a function of temperature in all cases give linear plots of t1 vs. C1/2 going through the origin [82–84]. This is expected when tight ion pairs directly dissociate into free ions, and vice versa. Values of kd in DEE were found to be (in s1): 0.73 (20 1C), 0.93 (0 1C), 1.15(20 1C) and 1.45(40 1C). Hence, kd increases at lower temperatures, corresponding to an apparent negative activation energy. The results imply that the dissociation step is not an elementary reaction. Apparently, the solutions contain minute quantities of loose ion pairs (not detectable spectroscopically) in rapid equilibrium with tight ion pairs. Only the loose ion pairs dissociate into free ions, and are the primary products of free ion association. The dissociation rate constant, kd, is given by (see also Eq. (3)). kd ¼ kds K i ð1 þ K i Þ1

(23)

kds being the dissociation rate constant of the loose ion pair. The activation energy of dissociation, Ead, is equal to (see also (Eq. 4)) E ad ¼ E ads þ DH i ð1 þ K i Þ1

(24)

Since in our case Ki51 so that: kd ¼ kds K i and E ad ¼ E ads þ DH i

(25)

The tight–loose ion pair equilibrium is virtually unperturbed by the electric field since the inter-

conversion rate is much faster than the dissociation of the loose ion pairs. The slow relaxation rate, therefore, pertains to the formation of free ions from loose ion pairs. This result, i.e., free ions being the dissociation product of loose ion pairs and not tight ion pairs agrees with a recent theory of ionic dissociation as outlined by Fuoss [85]. The observation that the tight–loose ion pair equilibrium is fast relative to the loose ion pair dissociation into free ions is consistent with the results from the fluorescence data of fluorenyl ion pairs [66]. Fuoss–Kraus plots (LC1/2 vs. C where L is equivalent conductance) give evidence of a small fraction of triple ions in the above systems [86]. However, from the linear dependence of t1 as a function of C1/2 it is clear that, in the 1  103–5  105 M concentration range, the contribution of triple ions to the relaxation signal is negligible compared to that of free ions. At higher FlLi concentrations (up to 1.3  102 M) deviations from the linear relationship occur, with t1 now being a function of C. This suggests that equilibria such as dimerization, or more extensive triple ion formation, may start to interfere. In principle one would expect, in this case, several relaxation processes but at each concentration only a single relaxation time is observed. Apparently, application of a high electric field chiefly perturbs the overall equilibrium between conducting and non-conducting species rather than the specific equilibrium between ion pairs and free ions. In the latter case charge separation takes place whereas, for triple ion formation and ion pair aggregation, only dipole–dipole or ion–dipole interactions occur. Relaxation studies were also carried out with FlLi in DEE/toluene mixtures, in tetrahydropyran (THP), and in oxepane [88]. In the low dielectric constant mixtures non-negligible intercepts and dependencies of t1 higher than C1/2 are found due to the presence of dimers and triple ions. In THP and in oxepane (except for the highest values of C) plots of t1 vs. C1/2 are straight lines going through the origin. Plots of log kd vs. 1/T in the various media are shown in Fig. 8. Negative apparent activation energies are found in the DEE/toluene mixtures, indicative of free ion formation from loose ion pairs although the latter again can not be detected spectroscopically. In THP and oxepane loose FlLi ion pairs can be detected by spectroscopy, and, therefore, Ki is much higher. No negative activation

ARTICLE IN PRESS J. Smid et al. / Prog. Polym. Sci. 31 (2006) 1041–1067

1057

Electric field perturbation produces an excess of triple ions and a relaxation results from their recombination into non-conducting ion pairs and their dimers. When Kdim is small and [triple ion] 5C, t1 can be shown to depend on C3/2. For larger quantities of THF the conducting species turn out to be Cs+ and the Fl2Cs triple anion. In this case t1 is proportional to C, and a non-negligible intercept is found. More detailed studies show that the mechanism may even be more complex [89]. However, the Fl, Cs+ system clearly demonstrates that relaxation techniques, combined with conductance measurements, can reveal the presence of otherwise undetectable species 6. Copolymerization 6.1. Polar media

Fig. 8. Temperature dependence of dissociation rate constants of fluorenyllithium ion pairs as a function of solvent media [88].

energies are found in these solvents. Apparently DHi/(1+Ki) is small and Ead approaches Eads (see Eq. (25)). For systems composed of free ions, triple ions, ion pairs and their dimer Everaert and Persoons derived a general equation for the reciprocal relaxation time, consisting of a dissociation and an association term [87]. An example of such a system is FlCs in dimethoxymethane (D ¼ 2.66) at, and its mixtures with THF where t1 is found to be proportional to C3/2. Conductance studies indicate that the triple ions Fl, Cs+, Fl and Cs+, Fl, Cs+ are virtually the only conducting species present [88,89]. The most probable mechanism involves rapid formation of dimeric ion pairs (Eq. (26)) 2Fl Csþ Ð ðFl Csþ Þ2

K dim

(26)

followed by triple anion formation (Eq. (27)). Fl Csþ þ ðFl Csþ Þ2 Ð Csþ Fl ; Csþ þ Fl ; Csþ ; Fl

(27)

The presence of multiple propagating ionic species increases the complexity of mechanistic vinyl copolymerization studies. For instance, the dissociation of one carbanion pair into free ions will be affected by the presence of the second ion pair as the reaction progresses. While not a cross-over vinyl addition, an example of such an effect is found in the reaction of PSNa with triphenylmethane in THF [90]. In this proton abstraction the PSNa ion pair (Kd ¼ 1.5  107 M) is replaced by the less basic Ph3C Na+ (Kd ¼ 7.6  106 M). Since the free PS anion is much more reactive than the PSNa ion pair (about a 1000 times), the reaction is self-retarding through a common ion effect, i.e., it slows down as more trityl carbanions are formed. Therefore, the copolymer composition should be sensitive to factors that can modify the solvation/dissociation of carbanion salts, e.g., solvent, counter cation, and temperature. Under carefully chosen conditions, Szwarc et al. were able to measure absolute cross-over rate constants (k12) for the reaction of PSNa in THF with several monomers (Table 6) [91–96]. Since they were all measured at the same PSNa concentration (about 3  103 M), ion pair dissociation should not affect a comparison between the rate constants of the various monomers. For the psubstituted styrenes the k12 values obey the Hammett relationship [35] as shown in Fig. 9. The r value of 5.0 is much larger than that in the corresponding radical polymerization (r ¼ 0:5), implying that the rate determining step is the attack of the carbanion on the vinyl double bond. As in the

ARTICLE IN PRESS J. Smid et al. / Prog. Polym. Sci. 31 (2006) 1041–1067

1058

homopolymerization, the bulk of the reaction proceeds via the free polymeric carbanion. Values for k21 for the reaction of styrene with different polymeric carbanions are collected in Table 7 [92]. Note that the k21 cross-over rate constants are much less sensitive to the changes in the substituent than the corresponding k12 values. For closely related monomers the k21 values do not differ much from the homopolymerization rate constants. For these monomer pairs, e.g. styrene

Table 6 Rate constants of addition of various monomers to poly(styryl sodium) in THF at 25 1C; [active ends ] 3  103 M [92] Monomers

k12 Monomers (M1 s1)

Butadiene Isoprene 2,3-Dimethylbutadiene Vinylmesitylene b-Methylstyrene a-Methylstyrene p-Methoxystyrene p-t-Butylstyrene 2,4-Dimethylstyrene p-Methylstyrene

33 17 0.48 0.9 18 27 50 110 160 180

o-Methylstyrene Styrene p-Vinylbiphenyl p-Fluorostyrene 1,1-Diphenylethylene 1-Vinylnaphthalene 2-Vinylnaphthalene p-Chlorostyrene 2-Vinylpyridine 4-Vinylpyridine

k12 (M1 s1) 320 950 1600 1800 2500 8000 8600 23,000 430,000 430,000

Fig. 9. Hammett’s relation for the addition of substituted styrenes to polystyrylsodium in THF at 25 1C [35–38].

(1)/p-methoxystyrene (2) the ratio’s k11/k12(r1) and k22/k21 (r2) will approach unity. This case corresponds to a truly random or Bernoullian polymerization. Furthermore, as r1 r2E1, it is also an ‘‘ideal’’ copolymerization. When the polarities of the monomer pair differ substantially, or where penultimate unit effects [38] exist (styrene/2-vinylpyridine), the product r1.r2 most likely will deviate from unity. In some cases steric hindrance plays an important role. This is the case for the monomer pair styrene/DPE. [96] In this system the addition of styrene to the diphenyl carbanion is reversible, the equilibrium constant being K ¼ 5  102 M1. An alternating copolymer is obtained in the presence of a large excess DPE. While in a number of anionic copolymerizations the classical copolymer composition equation developed for radical polymerization has been used, it can often lead to erroneous conclusions. Two examples may be cited to illustrate the danger in deriving reactivity ratios from copolymer composition data. One of the more interesting cases involves the addition of styrene to the poly(1-vinylnaphtalene) carbanion (PVN) [37]. This addition has a rate constant of 30 M1 s1. Combined with the three other rate constants a reasonable r1.r2 product of close to unity was obtained. However, when styrene is added, the optical spectrum reveals that the 556 nm peak of the purple PVN carbanion does not initially produce the 340 nm peak of the PS carbanion but a new maximum at 440 nm. Only when a large excess of styrene is used is the latter maximum replaced by the expected 340 nm peak of the PS anion. Scheme 6 illustrates that a peculiar penultimate unit effect may occur. Although the newly formed PSM+ does not react with free naphthalene, addition to the penultimate 1-naphthyl group appears to occur readily, presumably due to formation of a six–membered ring. This minimizes

Table 7 Rate constants of addition of styrene to various polymeric anions in THF at 25 1C ([active ends]E3  103; counterion, Na+) [92] Polymeric anion

k21, (M1 s1)

Polymeric anion

k21, (M1 s1)

a-Methylstyrenea Styrene p-Methylstyrene p-Methoxystyrene

780 950 1150 1100

Vinylmesitylene 1-Vinylnapthalene 1,1-Diphenylethyleneb 2-Vinylpyridine-

77 30 0.7 1

a

Tetramer dianion. Dimer dianion.

b

ARTICLE IN PRESS J. Smid et al. / Prog. Polym. Sci. 31 (2006) 1041–1067

-

1059

Ph

+

k1 = 30 M

-1

.sec

Ph

K

-1

rapid

440 nm

556 nm kp = 600 M

-1

.sec

-

-

340 nm Scheme 6. Mechanism of styrene addition to the polymeric carbanion of 1-vinylnaphthalene [37].

entropy of activation and gives a highly stable allylbenzyl carbanion that is incapable of adding styrene. However, this anion remains in equilibrium with a small fraction of the styryl carbanion that now adds styrene. Once the new styryl anion is more than one unit removed from the last VN unit it will not react with a naphthalene moiety. As a result, the 440 nm carbanion eventually disappears and the spectrum then only shows the 340 nm PS peak. Obviously, analysis of the copolymer composition would not have detected any of these details nor would its composition necessarily reflect accurate reactivity ratios. A second example is the monomer pair styrene/ vinylpyridine [38]. While the poly(vinylpyridyl) anion (PVPy) reacts with styrene, and all excess styrene is consumed, the optical spectrum does not show the presence of a styryl carbanion. The deep red color of the PVPy carbanion changes into a pale yellow. In this case the styryl carbanion that is formed initially reacts rapidly with a pyridine ring. The same reaction occurs when polystyryl sodium is mixed with a THF solution of terminated polyvinylpyridine. When two-ended PSNa+ is used extensive cross-linking occurs. Again, copolymerization composition studies of the anionic polymerization of this monomer pair would not have revealed these side reactions. 6.2. Copolymerizations in apolar media Anionic propagation of styrene in solvents such as benzene is known to proceed through monomeric PSLi ion pairs which are in rapid equilibrium with their unreactive (referred to as ‘‘dormant’’ by M

Szwarc) dimers (PSLi)2 [97–99]: K dim

ðPS Liþ Þ2 $ 2ðPS Liþ Þ

(28)

PSn Li þ S ! PSnþ1 Li

(29)

kp

The propagation rate is first order in monomer but one-half order in living ends, the propagation rate constant being kapp ¼ kp(Kdim/2)1/2. Viscosity measurements at high living end concentrations confirm the presence of dimers [99]. For the sodium salt the order in PSNa remains 0.5, but changes from 0.5 at 103 M to 1.0 at 2–5  105 M for PSK. No dimers are found for the Rb and Cs salts, and the reaction order in benzene remains unity. The respective values for Kdim (M) and kp (M1 s1) at 20 1C were found to be 107 and 40 for Li and 6  104 and 38 for PSNa. The kp values for the Rb and Cs salts are 22 and 13 M1 s1, respectively. The free PSLi ion pair is more reactive in benzene than in ethereal solvent [6,100]. Adding THF to PSLi in benzene breaks down the unreactive dimers, causing a higher rate. Both 1:1 and 1:2 complexes are formed with THF. The propagation rate goes through a maximum when more THF is added as the solvated PSLi ion pairs are less reactive than uncomplexed monomeric PSLi. Adding the Licomplexing agent tetramethylethylene diamine to PSLi in cyclohexane causes a similar rate pattern [101]. With o-methoxystyrene as monomer intramolecular solvation of the Li counterion by the OCH3 substituent interferes with the dimerization of the living PSLi ends [102]. This causes Kdim to increase, and the reaction order with respect to PSLi is now variable.

ARTICLE IN PRESS 1060

J. Smid et al. / Prog. Polym. Sci. 31 (2006) 1041–1067

In order to better understand the role of cation and living end in anionic copolymerization reactions in benzene, the Hammett relationship for a series of disubstituted DPE’s reacting with polystyryl salts (PSM) [103] was examined in benzene and THF at 20 1C [104]. Since the DPE’s are non-polymerizable, optical spectroscopy permits an easy way to determine the addition rate constants. All reactions were carried out with a large excess of PSM to avoid formation of mixed dimers. Results show that under the same conditions addition of the DPE’s to PSM is faster than the styrene addition, reflecting the higher resonance stabilization of the diphenyl carbanion.

The r values derived from plots shown in Fig. 10 reflect the ratios of rate constants of substituted and the unsubstituted DPE, irrespective of whether PSM is dimeric or monomeric. In benzene r ¼ +(1.7–1.9) for PSLi and +(2.2–2.4) for the K and Cs salts. In THF r ¼ +(2.8 – 3.5) for PSK, considerably lower than the value r ¼ +5.0 reported in the same solvent for the reaction of PSNa with substituted styrenes (Fig. 9). In this solvent free PS carbanions are the reactive species. The higher reactivity, and resulting lower selectivity of the DPE monomers may be the reason for the lower r values of the DPE additions. The larger r value for PSK in THF relative to that in benzene

Fig. 10. Hammett relations for the addition of living PSM (103–104 M) to substituted 1,1-diphenylethylenes (4.105 M) at 24 1C in benzene and THF: (A) PSLi in benzene (circles) and in cyclohexane (squares) (B) PSK in benzene; (C) PSCs in benzene; (D) PSK in benzene (o) and THF(x) [104].

ARTICLE IN PRESS J. Smid et al. / Prog. Polym. Sci. 31 (2006) 1041–1067

1061

may be caused by the greater sensitivity of the free PS anion to changes in the electron density on the vinyl double bond. In benzene, r decreases in the order Cs4K4Li, probably as a result of the increase in field strength of a smaller cation, which reduces the nucleophilicity of the carbanion pair. Reversible formation of a Li-monomer complex followed by monomer insertion may be an alternative explanation for the low r value of PSLi in benzene (Eq. (28)): Pn Li þ M Ð Pn Li; M ! Pnþ1 Li

(30)

This leads to the relationship: ðE 0  E S Þ=RT þ ðDH 0  DH S Þ=RT ¼ rs

(31)

where E0 and ES are the respective activation energies for the insertion of unsubstituted and substituted monomer, and DH0 and DHS the corresponding enthalpies of complex formation. Electron donating substituents (negative s) favor complex formation, making the sign of the second term of Eq. (31) opposite that of the first term. This leads to a lower r value for PSLi relative to that of PSCs where no complex formation occurs. In this context the anionic copolymerization of styrene and butadiene initiated in hydrocarbon solvents by lithium alkyls is very interesting and instructive. This copolymerization was first investigated by Korotkov [105] who reported that, although styrene homopolymerizes rapidly and the homopolymerization of butadiene is slow, surprisingly, the copolymerization of a mixture of these monomers starts slowly and initially only butadiene is consumed. After the depletion of this monomer, the reaction speeds up and styrene then polymerizes rapidly (Fig. 11). Korotov offered an ingenious explanation for this phenomenon, although this was later shown to be incorrect. Treating the monomers as solvents of the Li ion,butadiene was assumed to be a stronger solvating agent and was expected to be present almost exclusively at the growing chain-end, thereby excluding the more poorly solvating styrene from the reaction center. Thus, butadiene polymerizes preferentially, albeit slowly, followed by rapid polymerization of the more reactive styrene. This unusual behavior was correctly explained by O’Driscoll and Kuntz [106] who pointed out that although styrene homopolymerizes faster than butadiene, the cross-over rate constant kSB PSLi with butadiene is much larger than the homopropagation rate constant kSS for PSLi, whereas the rate constant for the addition of styrene to

Fig. 11. Copolymerization of butadiene and styrene by n-BuLi in cyclohexane. Note the slow initial polymerization of butadiene followed by the much faster styrene polymerization.

polybutadienyllithium (PBLi) is very small. Moreover, as the homopolymerization of styrene is faster than that of butadiene (kSS4kBB) virtually all the active centers in the early stages of the reaction are in the form of PBLi since addition of styrene will be rapidly followed by the addition of butadiene present and the PBLi will then slowly homopolymerize since its cross-over into PSLi is very slow. Only at the depletion of butadiene will styrene react and homopolymerize causing the increased rate. This idea was confirmed by the copolymerization kinetics of butadiene to PSLi and of styrene to PBDLi [107,108a]. Since for the above system only qualitative results could be obtained due to experimental difficulties, Worsfold [108b] investigated the closely related but experimentally more accessible copolymerization of isoprene–styrene (1/1, w/w) in cyclohexane. In this case the reaction rates of PSLi both with styrene and isoprene are half order in the salt, in accordance with the predominance of (unreactive) PSLi dimers, whereas the rates of addition of styrene and isoprene to isoprenyllithium are both proportional to the one-fourth power of polyisoprenyllithium (PILi) indicating formation of tetramers. The authors pointed out that there is no detectable PSLi in the early stages of the reaction and that virtually all the chains exist as PILi, due to the slow rate of reaction of styrene with PILi and the very rapid reaction of isoprene with PSLi, when the latter does form PILi. Should the low styrene content of the initial copolymer be attributed to preferential solvation of the chain end by isoprene, as suggested by Korotkov, then the rate of reaction of PILi with styrene would be lower in the presence of isoprene

ARTICLE IN PRESS 1062

J. Smid et al. / Prog. Polym. Sci. 31 (2006) 1041–1067

than in its absence, whereas such retarding effect was not observed. Thus, the copolymerization can be described in terms of the four individual rate constants without resorting to preferential solvation of the chain ends by dienes. The above raises the question why butadiene or isoprene add so much faster to PSLi than styrene [6]. It was already pointed out that the monomeric PSLi ion pair is more reactive in benzene than in ethereal solvents [6,100]. In order to account for the Hammett r values in the series Cs4K4Li in benzene, one of the explanations was the pcomplexation of the monomer with the counterion of the living PSLi followed by monomer insertion. Indeed, Szwarc et al. [108,109] concluded from the experiments of Worsfold, that Li ion coordination by monomers takes place prior to its insertion. Whether as a discrete intermediate complex or in the transition state, this coordination appears to make the Li ion pairs exceptionally reactive. In ether solvents Li–diene coordination would be prevented by Li ion solvation, causing a reduction of the reactivity of the Li ion pair. The occurrence of Li–diene coordination was also corroborated by Van Beylen et al. who reported detailed studies on the effect of p-complexing agents on alkyllithium initiated polymerizations of styrene [110–113]. As originally suggested by O’Driscoll and Patsiga [114], durene (tetramethylbenzene) forms a p-complex with the Li ion of PSLi. This results in an acceleration of the styrene polymerization, but at higher concentrations of durene the propagation decreases again. The effect resembles that found on adding THF [100] but higher durene concentrations are required to obtain the same effect (Fig. 12). The data are consistent with the presence of unreactive PSLi dimers in equilibrium with the reactive monomeric PSLi, with the latter species forming 1:1 and 1:2 p-complexes with durene (PSLiD and PSLiD2). The dissociation constant, Kdim, for the (PSLi)2 dimer in cyclohexane is approximately 2  107 M, while the respective formation constants of the 1:1 and 1:2 durene complexes were found to be 140 and 1 M1. The propagation rate constant for PSLi, PSLiD, and PSLiD2 are kp ¼ 480, 6.7 and 3.0 M1 min1, respectively. In the presence of 0.56 M durene the Kdim of (PSLi)2 in cyclohexane is increased from 2  107 to 6.2  103 M. Moreover, the reactivity of the non-aggregated uncomplexed ion pair is not only higher than that of the PSLi complexed with one or two durenes but also exceeds that of the tight

Fig. 12. Effect of durene on the propagation of polystyryllithium (PSLi) in cyclohexane at 20 1C. Concentrations: [PSLi] ¼ 5.2.104 M (triangles); [PSLi] ¼ 3.0.103 M (solid circles) [111].

ion pair in ether solvents. This is attributed to the formation of the Li ion p- or s-complexes impeding the Li ion–diene interactions in the ground or transition states. The conclusions of this investigation were confirmed by density functional calculations [112,113]. Studies with the p-complexing agents benzene, hexamethylbenzene, tetramethyl ethylene and tetraphenylethylene (TPE) show that only TPE is more strongly coordinated with PSLi in cyclohexane than is durene [110,111]. These experiments suggest that Li–ion-diene complexes are involved in the polymerization of dienes in hydrocarbon media [115], the Li–diene complex leading to a 1,4 microstructure as found in cyclohexane and similar solvents. In order to obtain SBS triblock copolymers with a high 1,4 micro-structure in the PBD block, bifunctional organolithium initiators in apolar solvents are used to initiate first butadiene followed by styrene. Small amounts of THF, triethylamine and similar Lewis bases have often been added in order to prevent intermolecular aggregation of the two-ended initiators or living polymers which can lead to gelation or to insoluble tridimensionally-associated dilithium organolithium initiators. However, the polar additives also cause an undesirable increase in the content of 1, 2microstructure. To avoid this, Hofmans and Van Beylen [116] reacted 1,3-diisopropenyl benzene in cyclohexane with two equivalents of t-BuLi in the presence of at least 0.5 M durene. On adding butadiene a narrow

ARTICLE IN PRESS J. Smid et al. / Prog. Polym. Sci. 31 (2006) 1041–1067

0.8M durene, 45 0 C + t-BuLi

Li

t -Bu

Li

t -Bu

t -Bu

C6 H12 , 30 hrs.

Li

t -Bu

Li

1063

+ BD

0.8M durene, 45 0 C Li-PBD-Li C6 H12 , 12 hrs.

3% THF in Styrene Li-PBD-Li

0

20 C, overnight

Li-PS-PBD-PS-Li

Ch3 OH

PS-PBD-PS

Scheme 7. Synthesis of SBS block copolymers having high 1,4-PBD content.

MW distribution bifunctional PBD dianion was obtained with the same content of 1,4-microstructure as in the absence of durene (Scheme 7). Apparently, p-complexation with the durene leaves the bifunctional Li initiator in a soluble form but the complex is not strong enough to prevent the crucial Li-diene interaction. The subsequent reaction of the PBD dicarbanion with styrene does not occur when durene is present. However, it is well known that styrene does add to PBDLi in non-polar media. As shown above, durene prevents the reaction of PBDLi sufficiently with styrene, which was shown to be already very slow in cyclohexane without the presence of durene. However, the reaction of PBDLi with BD in cyclohexane and in the presence of durene, still occurs because, apparently, BD forms a much stronger complex with PBDLi than durene so that this reaction can still occur (with conservation of 1,4 content). With the less strongly complexing styrene this addition is not possible. Even the reaction of styrene with PSLi is much slower than in the absence of durene. To form the SBS triblock copolymer, THF or TMEDA must be added. These s-complexing agents break down the durene p-complex with PBDLi, and now styrene is rapidly added. The coordination of hydrocarbon monomers with Li ion in non polar media such as hexane was also demonstrated by Makino and Hogen-Esch who found dramatic changes in stereochemistry in the LiOH-mediated polymerization of styrene [117]. Thus, in hexane and in the presence of 0.5 equivalent of H2O this t-BuLi initiated polymerization yielded PS with a highly isotactic content (95% isotactic mm triads as determined by 13C NMR).

Although the details of the stereocontrol are not clear, the stereochemistry of these polymerizations in the presence of p donors such as toluene, a moderately active p donor, becomes close to atactic (mm ¼ 14%). As indicated in the preceding, this result also supports the idea that Li ion does coordinate styrene and similar monomers in the absence of strong p or s-donors. 7. Conclusions As shown in this review, the research initiated by Michael Szwarc has lead to multiple lines of research. We have illustrated how his work on living polymers has lead to increased focus on the role of ion pair solvation and -ionization in anionic polymerization kinetics and chemical equilibria. Thus, the effects of solvent and carbanion structure on ion pair solvation and the dynamics of ion pair ionization have been discussed. Finally, the polymerization in non-polar solvents in the presence of Li ion and the issue of Li cation p-complexes in vinyl polymerizations has been illustrated. Other aspects of Michael’s research on living polymerization—mechanistic, synthetic or otherwise—will be the subject of a future paper in this series. Acknowledgments We wish to state our indebtedness to Michael Szwarc for his life-long guidance and inspiration. Support of some of the work by the present authors was provided by the Petroleum Research Foundation (JS), by NSF-DMR and the Loker Hydrocarbon Research Institute (THE) and by the University of Leuven, the Belgian National Science

ARTICLE IN PRESS 1064

J. Smid et al. / Prog. Polym. Sci. 31 (2006) 1041–1067

Foundation, GOA, and Asahi Chemical Industries (MVB). We also wish to thank our many coworkers who have contributed much with their energy and insight. Help with the references and figures by Scott Taylor and Dr. Fred Boschet at USC is greatly appreciated.

References [1] Szwarc M. Electron transfer reactions and living polymers. Proc Roy Soc 1964;A279:260–90. [2] Szwarc M. Living polymers and mechanisms of anionicpolymerization. Adv Polym Sci 1983;49:1–177. [3] Van Beylen M, Bywater S, Smets G, Szwarc M, Worsfold DJ. Developments in anionic polymerization—a critical review. Adv Polym Sci 1988;86:87–143 and references therein. [4] Szwarc M. Carbanions, living polymers and electron transfer processes. New York: Interscience; 1968. [5] Szwarc M, editor. Ions and ion pairs in organic reactions, vol. 1. 1972 vol. 2 (1974). [6] Szwarc M, Van Beylen M. Ionic polymerization and living polymers. New York: Chapman & Hall; 1993. [7] Szwarc M. Ionic polymerization: fundamentals. New York: Hanser; 1996. [8] Errede LA, Szwarc M. Chemistry of p-xylylene its analogues and polymers. Quart Rev 1958;12:301. [9] Gorham WF. Recent advances in the vapor deposition polymerization of p-xylylenes. Adv Chem Ser 1969(91): 643–59. [10] Verheyden H, Van Lierde P, Szwarc M, Litvinenko G, Van Beylen M. Dual influence of lithium chloride on the anionic propagation of polystyryllithium in ethereal solvents. J Polym Sci Polym Chem 2002;40:2148. [11] Szwarc M. Reaction of methyl radicals and their application to polymer chemistry. J Polym Sci 1955;16:367–82. [12] Smid J, Szwarc M. Ethyl affinities of aromatic and olefinic compounds. J Am Chem Soc 1956;78:3322–5. [13] Szwarc M. ‘Living’ polymers. Nature 1956;178:1168. [14] Szwarc M. Living polymers. Their discovery, characterization, and properties. J Polym Sci -Polym Chem 1998; 36(1):ix–xv. [15] Szwarc M. Electron-affinities of aromatic-hydrocarbons, their radical-anions and dianions. Adv Chem Ser 1988;217: 13–37. [16] Lee CL, Smid J, Szwarc M. The mechanism of formation of living a-methylstyrene dimer and tetramer. J Phys Chem 1962;66:904–7. [17] Wang HC, Levin G, Szwarc M. Flash-photolysis of salts of dimeric dianions of a-methylstyrene and dimerization of its radical-anions. J Phys Chem 1979;83:785–8. [18] Wang HC, Levin G, Szwarc M. The dimerization of Li, Na, K, and Cs salts of diphenylethylene radical-anions and equilibria and kinetics of their formation by attachment of electron–cation pairs. The nature of electron-cation aggregates. J Am Chem Soc 1978;100:6137–41. [19] Alberty K, Chen R, Hogen-Esch TE. A new bifunctional initiator for use in anionic polymerizations. J Polym Sci Polym Chem 2002;40:2108–25.

[20] Asami R, Szwarc M. Kinetics of dissociation of organic dimeric dianions into monomeric radical anions. J Am Chem Soc 1962;84:2269–70. [21] Spach G, Montero H, Levy M, Szwarc M. Chemistry of radical anions. The exchange between diphenylethylene and its dimeric radical anion. Trans Faraday Soc 1962;58: 1809–20. [22] Bhattacharyya DN, Smid J, Szwarc M. Alternative paths in anionic propagation of ion pairs. Effect of solvent and counterion. J Phys Chem 1965;69:624–7. [23] Bhattacharyya DN, Lee CL, Smid J, Szwarc M. Reactivities and conductivities of ions and ion pairs in polymerization processes. J Phys Chem 1965;69:612–23. [24] Shimomura T, To¨lle KJ, Smid J, Szwarc M. Energy and entropy of activation of propagation by the free polystyryl anions and their ion pairs. The phenomenon of ‘‘negative’’ activation energy. J Am Chem Soc 1967;89:796–803. [25] Hogen-Esch TE, Smid J. Solvent-separated ion pairs of carbanions. J Am Chem Soc 1965;87:669–70. [26] Hogen-Esch TE, Smid J. Studies of contact and solventseparated ion pairs of carbanions. I. Effect of temperature, counterion and solvent. J Am Chem Soc 1966;88:307–18. [27] Hogen-Esch TE, Smid J. Studies of contact and solventseparated ion pairs. II. Conductivities and thermodynamics of dissociation of fluorenyllithium, sodium and cesium. J Am Chem Soc 1966;88:318–24. [28] Shimomura T, Smid J, Szwarc M. Reactivities of contact and solvent separated ion pairs. Anionic polymerization of styrene in dimethoxyethane. J Am Chem Soc 1967;89: 5743–9. [29] Bo¨hm LL, Chmelir M, Lo¨hr G, Schmitt BJ, Schulz GV. States and reactions of carbanions in the anionic polymerization of styrene. Adv Polym Sci 1972;9:1–45. [30] Alvarino JM, Chmelir M, Schmitt BJ, Schulz GV. The kinetics of anionic polymerization of styrene in 1,2dimethoxyethane. J Polym Sci Polym Symp 1973;42: 155–62. [31] Bo¨hm LL, Lo¨hr G, Schulz GV. Transition rates between the different ionic states of polystyrylsodium in polar organic solvents. Ber Bunsenges 1974;78:1064–78. [32] Shinohara M, Smid J, Szwarc M. Effect of solvation on the reactivity of ion pairs. J Am Chem Soc 1968;90:2175–7. [33] Shinohara M, Smid J, Szwarc M. Studies of ion pair solvation: absolute rate constants of propagation of the tetraglyme-coordinated living sodium polystyrene. Chem Commun London 1969:1232–4. [34] Worsfold DJ, Bywater S. Anionic polymerization of amethylstyrene II. Kinetics. Can J Chem 1958;36:1141–3. [35] Shima M, Bhattacharyya DN, Smid J, Szwarc M. Hammett’s relations in anionic copolymerization. J Am Chem Soc 1963;85:1306–10. [36] Bhattacharyya DN, Smid J, Szwarc M. Anionic polymerization of vinylmesitylene. J Polym. Sci 1965;A3:3099–106. [37] Bahsteter F, Smid J, Szwarc M. Kinetics of anionic polymerization and copolymerization of the vinylnaphthalenes and styrene. The intramolecular charge transfer complexes. J Am Chem Soc 1963;85:3909–18. [38] Lee CL, Smid J, Szwarc M. Kinetics of anionic polymerization and copolymerization of vinylpyridines. The penultimate effects. Trans Faraday Soc 1963;59:1192. [39] Fisher M, Szwarc M. Kinetics of propagation of living poly-2-vinylpyridine. Macromolecules 1970;3:23–7.

ARTICLE IN PRESS J. Smid et al. / Prog. Polym. Sci. 31 (2006) 1041–1067 [40] Bhattacharyya DN, Smid J, Szwarc M. A novel approach to triple ion formation. J Am Chem Soc 1964;86:5024–5. [41] Tardi M, Rouge D, Sigwalt P. Etude de la conductivite des solutions carbanioniques de polymers vivants de poly(2vinylpyridine) dans le tetrahydrofuranne. Eur Polym J 1972;8:151–62. [42] Khan IM, Hogen-Esch TE. Characterization of carbanions in THF by conductance and 7 Li NMR. Evidence for the occurrence and nature of triple ion formation. J Polym Sci 1988;26:2553–9. [43] Verheyden H, Van Lierde P, Szwarc M, Litvinenko G, Van Beylen M. Dual influence of lithium chloride on the anionic propagation of polystyryllithium in ethereal solvents. J Polym Sci Part A: Polym Chem 2002;40:2148–57. [44] Van Lierde P, Litvinenko G, Van Beylen M. Effect of lithium butoxide additives on the anionic propagation of polystyryllithium in ethereal solvents. Polymer 2006;47: 49–55. [45] [a] Yakimansky AV, Mu¨ller AHE, Van Beylen M. Density functional theory study on the aggregation and dissociation behavior of lithium chloride in THF and its interaction with the active centers of the anionic polymerization of methyl methacrylate and styrene. Macromolecules 2000;33:5686–92; [b] Yakimansky AV, Van Beylen M. Quantum-chemical DFT calculations to interpret the effect of lithium butoxide additives on the anionic polymerization of styrene in polar solvents. Polymer 2002;43:5797–805. [46] [a] Desbois P, Fontanille M, Deffieux A, Warzelhan V, La¨tsch S, Schade C. Towards the control of the reactivity in high temperature bulk anionic polymerization of styrene. 1. Influence of n,s-dibutylmagnesium on the reactivity of polystyryllithium species. Macromol Chem Phys 1999;200:621–8; [b] Desbois P, Fontanille M, Deffieux A, Warzelhan V, Schade C. Towards the control of the reactivity in high temperature anionic polymerization of styrene: retarded anionic polymerization. 3. Influence of triisobutylaluminum on the reactivity of polystyryllithium species. Macromol Symp 2000;157:151–60. [47] De Groof B, Mortier W, Van Beylen M, Szwarc M. Anionic-polymerization of barium salt of one-ended living polystyrene. 1. Conductance and kinetic studies. Macromolecules 1975;8:396–400. [48] De Smedt C, Van Beylen M. Anionic-polymerization of the strontium salt of one-ended living polystyrene in tetrahydrofuran and tetrahydropyran. ACS Symp Ser 1980;179:97. [49] Mathis C, Christmann-Lamande L, Franc- ois B. Anionicpolymerization with alkaline-earth counterion. 1. Preparation and characterization of dicarbanionic oligostyrylbarium and oligostyrylstrontium in tetrahydrofuran. J Polym Sci 1978;16:1285–96. [50] Winstein S, Clippinger E, Fainberg AH, Robinson GC. Salt effects and ion pairs in solvolysis. J Am Chem Soc 1954;76:2597–8. [51] Cram DJ, Mateos JL, Hauck F, Langemann A, Kopecki KR, Nielsen WD, et al. Electrophilic substitution at saturated carbon. VI. Stereochemical capabilities of carbanions. J Am Chem Soc 1959;81:5774–84. [52] Hogen-Esch TE. Specific solvation phenomena of some alkali carbanions in aprotic solvents. PhD thesis; University of Leiden, 1967.

1065

[53] Hofelmann K, Jagur-Grodzinski J, Szwarc M. ESR studies of sodium naphthalide THP solutions containing tetraglyme (TG). J Am Chem Soc 1969;91:4645–51. [54] Hogen-Esch TE, Smid J. Ion pair structures of divalent carbanion salts. J Am Chem Soc 1969;91:4580–1. [55] Hogen-Esch TE, Smid J. Evidence for stepwise solvation of strontium carbanion salts in ethereal solvents. J Am Chem Soc 1972;94:9240–1. [56] Claesson S, Lundgren B, Szwarc M. The effects of pressure on the equilibrium tight ion pairs-loose ion pairs. Trans Faraday Soc 1970;66:3053–60. [57] LeNoble WJ, Das AR. Volume change during the solvent separation of a tight ion pair in a solvent of low dielectric constant. J Phys Chem 1970;74:3429. [58] Chan LL, Smid J. Contact and solvent separated ion pairs of carbanions. V. The role of solvent structure in alkali ion solvation. J Am Chem Soc 1968;90:4654–61. [59] Ellingsen T, Smid J. Studies of contact and solvent separated ion pairs of carbanions. VI. Conductivities and thermodynamics of dissociation of fluorenyl alkali salts in tetrahydrofuran and dimethoxyethane. J Phys Chem 1969; 73:2712–9. [60] Chan LL, Wong KH, Smid J. Complexation of lithium, sodium and potassium carbanion pairs with polyglycol dimethyl ethers (glymes). Effect of chain length and temperature. J Am Chem Soc 1970;92:1955–63. [61] Takaki U, Hogen-Esch TE, Smid J. Ion pair aggregation in solutions of complexes of carbanion salts and macrocyclic polyethers. J Phys Chem 1972;76:2152–5. [62] Wong KH, Konizer G, Smid J. Binding of cyclic polyethers to ion pairs of carbanion alkali salts. J Am Chem Soc 1970;92:666–70. [63] Takaki U, Hogen-Esch TE, Smid J. Complexes of macrocyclic polyethers and ion pairs. J Am Chem Soc 1971; 93:6760–6. [64] Hogen-Esch TE, Smid J. Conductivities and thermodynamics of dissociation of fluorenyl salts and their complexes with dimethylbenzo-18-crown-6. J Phys Chem 1975; 79:233–8. [65] Hogen-Esch TE. Ion pairing effects in carbanion reactions. In: Gold V, Bethell D, editors. Advances in physical organic chemistry, vol. 15. London: Acdemic Press; 1977. p. 153–266 and references therein. [66] Plodinec MJ, Hogen-Esch TE. Ion pairing in excited states of carbanions. I. Cation and solvent effects. J Am Chem Soc 1974;96:5262–3. [67] Plodinec MJ, Hogen-Esch TE. Ion pairing in excited states of carbanions. 2. Emission of fluorenyl and fluoradenyl alkali salts in aprotic media. J Am Chem Soc 1978;96: 7633–40. [68] Comyn J, Ivin KJ. The absorption spectra of living oligomers of a-methylstyrene. Eur Polym J 1969;5: 587. [69] Peeters S, Gils F, Van Beylen M. Novel findings on solvent separation of polystyryl salts in anionic polymerization. Proceedings of the 28th International Symposium on Macromolecules, Amherst, MA 1982:104. [70] Van Kerckhoven C, Kopmanis R, De Hertog D, Van de Sande W, van Beylen M. Unpublished results. [71] [a] Chmelir M, Schulz GV. Conductivity measurements on polystyryl sodium in dimethoxyethane and tetrahydrofuran. Ber Bunsenges 1971;75:830–3;

ARTICLE IN PRESS 1066

J. Smid et al. / Prog. Polym. Sci. 31 (2006) 1041–1067

[b] Schmitt BJ, Schulz GV. Complete kinetics of the anionic polymerization of polystyryl sodium in THF. Makromol Chem 1971;142:325–30. [72] Bywater S. Structure and mechanism in anionic polymerization. Prog Polym Sci 1994;19:287–316. [73] Chang CJ, Kiesel RF, Hogen-Esch TE. UV/Vis and conductometric studies of vinyl-pyridine type carbanions in aprotic media. J Am Chem Soc 1975;97:2805–10. [74] Hogen-Esch TE, Jenkins WL. NMR and conductometric studies of 2-pyridyl substituted carbanions. 2. Effects of cation size and coordination. J Am Chem Soc 1981;103:3666–72. [75] Figuerelo JE. Conductivity studies on living poly(methyl methacrylate) in tetrahydrofuran. Makromol Chem 1970;131:63–74. [76] Zagala AP. Anionic synthesis of narrow molecular weight distribution polymethylmethacrylate, poly(n-butylacrylate) and polymethylmethacrylate-poly(n-butylacrylate) block copolymers at elevated temperatures. PhD thesis; University of Southern California. 1995,p. 31–6. [77] Tsvetanov C, Sgonnic V, Panayotov I, Erussalimsky B. Langkettige alkalimetallorganische Verbindungen von Typ RCH2CH(CN)M: Herstellung, spektroskopische und elektrochemische Charakterisierung. Liebigs Ann Chem 1972;763:53–60. [78] Jeuck H, Mu¨ller AHE. Kinetics of the anionic polymerization of methyl methacrylate in tetrahydrofuran using lithium and potassium as counterions. Makromol Chem Rapid Commun 1982;3:121. [79] Chang CJ, Hogen-Esch TE. Effect of ion pair structure on the equilibrium addition of carbanions to substituted vinylpyridines. Tetrahedron Lett 1976;5:323–6. [80] Persoons A. Field dissociation effect and chemical relaxation in electrolyte solutions of low polarity. J Phys Chem 1974;78:1210. [81] [a] Wien AM. The voltage effect of the conductivity in strong and weak acids. Physik Z 1931;32:545–7; [b] Eckstrom HC, Schmeltzer C. The Wien effect: deviations of electrolytic solutions from Ohm’s law under high field strengths. Chem Rev 1939;24:367. [82] Persoons A, Van Beylen M. Dynamics of electric-field effects in ion-pairing processes. Pure Appl Chem 1979;51: 887–900. [83] Van Beylen M, Persoons A, Uytterhoeven H. Study of the dissociation and solvation of fluorenyllithium by electric field relaxation. International ISI symposium, We´pionLeuven, Preprint. 1974. p. 190. [84] Uytterhoeven H, Vanzegbroeck J, Persoons A, Van Beylen M. The mechanism of the ionic dissociation of fluorenyllithium in diethylether tetrahydrofuran mixtures as studied by electric field relaxation, Abstract: XXVI International congress of pure and applied chemistry; Tokyo, Japan. 1977. p. 173. [85] Fuoss RM. Non-Coulomb variation of ion pairing in polar solvents. J Am Chem Soc 1978;100:5576–7. [86] Fuoss RM, Kraus CA. Properties of electrolytic solutions. IV. The conductance minimum and the formation of triple ions due to the action of Coulombic forces. J Am Chem Soc 1933;55:2387–99. [87] Everaert J, Persoons A. The dynamics of ionic processes. In: Mc Grath JE, editor. Kinetics and Mechanisms and Synthesis. ACS symposium series no. 166; American Chemical Society. 1981. p. 153.

[88] Verbist E. Comparative study of the dissociation mechanisms of fluorenyllithium and polystyryllithium in low polarity solvents. A conductance and relaxation study. PhD thesis; University of Leuven, 1982. [89] Vanzegbroeck J, Van Beylen M. Dynamics of association, dissociation and solvation of fluorenylcesium ion pairs in low polar media. Abstracts: IUPAC IV; international symposium on solute–solvent Interactions, Vienna. 1978. p. 106. [90] Chan LL, Smid J. Reactivities of ion pairs and free ions in proton abstraction reactions. The reaction between polystyryl carbanion salts and triphenylmethane. J Phys Chem 1972;76:695–701. [91] Szwarc M, Smid J. The kinetics of propagation of anionic polymerization and copolymerization. In: Porter G, editor. Progress in reaction kinetics, vol. II. New York: Pergamon Press; 1964. p. 217–84. [92] Smid J. Elementary steps in anionic vinyl polymerization. In: O’Driscoll KF, Tsuruta T, editors. Structure and mechanism of vinyl polymerization. New York: Marcel Dekker; 1969. p. 345–448. [93] Bhattacharyya DN, Lee CL, Smid J, Szwarc M. The absolute rate constants of anionic copolymerization in the system styrene-a-methylstyrene. The effect of polymer structure on the rate of polymerization. J Am Chem Soc 1963;85:533–9. [94] Lee CL, Smid J, Szwarc M. The direct determination of the absolute copolymerization rate constant in anionic polymerization. J Am Chem Soc 1961;83:2961–2. [95] Smid J, Szwarc M. Kinetics of anionic copolymerization. Determination of absolute rate constants. J Polym Sci 1962;61:31–7. [96] Ureta E, Smid J, Szwarc M. Anionic copolymerization of styrene and 1,l-diphenylethylene. J Polym Sci 1966; A1(4):2219–28. [97] Worsfold DJ, Bywater S. Anionic polymerization of styrene. Can J Chem 1960;38:1891–900. [98] Roovers J, Bywater S. Kinetics of the anionic polymerization of styrene in benzene. Trans Faraday Soc 1966;62: 701–6. [99] [a] Hsieh HL, Quirk RP. Anionic polymerization: principles and practical applications. New York: Marcel Dekker, Inc; 1996 Chapter III; [b] Morton M, Bostic EE, Livigny RA. Advances in anionic polymerization. Rubber Plast Age 1961;42:397–401. [100] Bywater S, Worsfold DJ. Anionic polymerization of styrene. Effect of tetrahydrofuran. Can J Chem 1962;40: 1564–70. [101] He´lary G, Fontanille M. Etude de la polymerisation anionique du styrene en milieu non polaire, et en pre´sence de N, N, N0 , N0 , tetramethyl ethylene diamine. Eur Polym J 1978;14:345–8. [102] Geerts J, Van Beylen M, Smets G. Anionic polymerization of o- and p-methoxy styrene. J Polym Sci A-I 1969;7: 2859–73. [103] Laita Z, Szwarc M. Mixed dimerization of living polymers having Li counterions. Kinetics of addition of 1,1diphenylethylene to living lithium polystyrene. Macromolecules 1969;2:412–3. [104] Busson R, Van Beylen M. Hammett relations in anionic polymerization. Reaction of polystyrylalkali salts with disubstituted 1.1-diphenylethylenes. Macromolecules 1977; 10:1320–6.

ARTICLE IN PRESS J. Smid et al. / Prog. Polym. Sci. 31 (2006) 1041–1067 [105] Korotkov AA, Mitzengendler SR, Dornutzig LL. On the mechanism of copolymerization of styrene and methyl methacrylate with metallic lithium. J Polym Sci B 1966;4(11):809–14. [106] O’Driscoll KF, Kuntz J. Kinetics of anionic copolymerization of monomers of similar polarities. J Polym Sci 1962;61:19–24. [107] Morton M, Ells FR. Absolute rates in anionic copolymerization. J Polym Sci 1962;61:25–9. [108] [a] Johnson AF, Worsfold DJ. Anionic copolymerization of styrene and butadiene. Makromol Chem 1965;85:273–9; [b] Worsfold DJ. Anionic copolymerization of styrene and isoprene in cyclohexane. J Polym Sci A-1 1967;5: 2783–9. [109] Van Beylen M, Bywater S, Smets G, Szwarc M, Worsfold DJ. Developments in anionic polymerization. Adv Polym Sci 1988;86:87–143. [110] Hofmans J, Maeseele L, Wang G, Janssens K, Van Beylen M. The influence of p-complexing additives on the dissociation of organolithium compounds in non- polar medium: a long time unexplored area in anionic polymerization. Polymer 2003;44:4109–15. [111] Wang G, Van Beylen M. Influence of p-complexing agents on the anionic polymerization of styrene with lithium as

[112]

[113]

[114] [115]

[116]

[117]

1067

counterion in cyclohexane. 1. Effect of durene. Polymer 2003;44:6205–10. Yakimansky A, Wang G, Janssens K, Van Beylen M. Influence of p-complexing agents on the anionic polymerization of styrene with lithium as counterion in cyclohexane. 2. Quantum-chemical density functional theory calculations. Polymer 2003;44:6457–63. Wang G, Janssens K, Van Oosterwijck CH, Yakimansky A, Van Beylen M. Influence of p-complexing agents on the anionic polymerization of styrene with lithium as counterion in cyclohexane. 3. Effect of tetraphenylethylene. Polymer 2005;46:295–302. O’Driscoll K, Patsiga R. Solvent effects in anionic copolymerization. J Polym Sci Part A 1965;3:1037–44. Spirin YL, Arest Yakubovich AA, Polyakov DK, Gantmakher AA, Medvedev SS. Polymerizations catalyzed by lithium and alkyllithium. J Polym Sci 1962;58:1181–9. Hofmans J, Van Beylen M. Synthesis and development of a dilithium initiator and its use for the preparation of ABAblock copolymers in non-polar medium: the use of p-complexing additives. Polymer 2005;46:303–18. Makino T, Hogen-Esch TE. Anionic synthesis of highly isotactic polystyrene in hexane in the presence of lithium hydroxides. Macromolecules 1999;32:5712–4.