Living polymers from cyclodimethylsiloxanes through non-protonic initiation

Makromol. Chem.. Suppl. 15, I5 - 30 (1989) 15 Living polymers from cyclodimethylsiloxanes through non-protonic initiation Pierre Sigwalt 7 Pascal NicoI, MichPle Masure Laboratoire de Chimie Macromoleculaire associe au CNRS, Universite Pierre et Marie Curie, tour 44, 4, place Jussieu, 75252 Paris Cedex 05, France (Date of receipt: September 30, 1988) SUMMARY: Complexes of antimony pentachloride with acyl chlorides such as acetyl chloride or l-naph- thoyl chloride polymerize various completely methylated cyclosiloxanes such as hexamethylcyclo- trisiloxane (D3) or dodecamethylcyclohexasiloxane (D,) with high rates of initiation and propaga- tion, their activity being unchanged in the presence of the proton-trap 2,6di-ferf-butyl-4-methyl- pyridine. For polymerization of D, conducted at - 10°C in methylene dichloride, there is a simultaneous and proportional formation of high-molecular-weight polymer (HP), of small cycles of the D,, type (concentrations ID,] > ID9] > IDl2] > . . .) and of macrocycles. A linear increase of the molecular weight (M) of H P is observed up to at least 50% conversion, in agreement with the presence of a constant number of macromolecules of high molecular weight, mainly consisting of a living polymer population. For higher conversions, M may grow more rapidly which may be explained by an increasing copolymerization of D, and D, with D 3 . Homopolymerization of D, occurs under similar conditions with a rate smaller than that for D, , but also with linearly increasing M of H P with conversion. The simultaneous formation of the various products is attributed to the presence of two populations of macromolecules. One of these would bear either two active centres or one active centre and one non-reacting end-group, and the other population would bear one active centre and one reacting end-group, giving by end-to-end ring closure, macrocycles and eventually smaller cycles. Introduction Until now, the only cationic polymerization of cyclosiloxanes studied in some detail I - 3 ) is that initiated by trifluoromethanesulfonic acid (CF3S0,H) particularly for polymerizations in methylene dichloride near room temperature. Together with high-molecular-weight polymer, various cyclic compounds are formed, the mechanism of formation of which is still not fully understood I ) . With octamethylcyclotetrasil- oxane (D4) ‘), decamethylcyclopentiloxane (D,) and dodecamethylcyclohexasiloxane (D,),), the general agreement is that small cycles are formed by backbiting reactions which give all possible cycles with decreasing concentration for increasing ring sizes ([D,] > [D,) > [D J > [D,]. . .). For hexamethylcyclotrisiloxane (D,)6-8). the rate of polymerization is much higher, and there is a preferential formation of D, and of other cycles D3x, with [D,] > [D9] > [DJ . . . . This may be explained by the occurr- ence, competitively to polymer formation, of a reaction which might involve either an end-to-end ring closure7) or a ring-expansion*) mechanism. There is also, in all cases, a formation of polymer fractions with lower molecular weights Mthan that of the high- molecular-weight polymer (= los to 10,). For D,, a fraction with M = 2 * lo4 has been assumed to be linear, but fractions with 2 . lo’ Q M Q lo4 for both D, and D3 are generally considered to consist of macrocycles that might also be formed by end-to-end ring closure (or by back-biting). 0025-1 16X/89/$03.00 16 P. Sigwalt, P. Nicol, M. Masure The possibility to polymerize cyclodimethylsiloxanes by non-protonic initiators is still controversial. Friedel-Crafts acids such as SnC149), FeCI, lo), TiCl,,) and SbCI, d o not polymerize D, in the absence of water or of hydrogen chloride which play the part of cocatalysts. However, SbCI, which is inactive for both D, and D, in bulk, gives high-molecular-weight polymers when incorporated as an intercalation compound in graphite"). But the purity of such heterogeneous systems is difficult to establish, and these reactions are extremely slow, both with D, and D,. Triethyldiboronsesquitrifluoromethanesulfonate (Et,B,((TTf),) polymerized D, rapidly in CH,CI,I2) but the effect of residual water is important and the reaction did not occur when the reagents were carefully purified I,). But polymerization did occur in these last conditions when CH,Cl, contained 15 vo1.-Yo of acetonitrile, and this even so in the presence of the sterically hindered base 2,6-di-tert-butyl-4-methylpyridine (DTBMP). However, the existence of a competitive reaction of some impurity (e. g. CF,SO,H) between D, and the base cannot yet be excluded. The reactions are very slow and the molecular weights very high, which shows a very low efficiency of the initiator. The aim of the research described in the present article was to find non-protonic initiators that would polymerize rapidly and with high efficiency the various cyclosil- oxanes. It was hoped that a kinetic control might lead to a reduction of the proportion of small cycles. By introduction of detectable initiator fragments in macromolecules. some pieces of information about the reaction mechanism might also be obtained more easily than for the initiation by protonic acids, for which the nature of the active centres is still unknown. Experimental part Reagents were generally purified in devices sealed under vacuum: solvents (CH,Cl,, heptane) and monomers (D, , Dd on several sodium films, SbCI, on a P,O, film. The complexes SbCI, with CH,COCI or 1-naphthoyl chloride (naphCOC1) were prepared by the method of Olah et al. at temperatures of - 10°C. or lower, at which they were kept until their use as initiators at - 10°C in CH,CI,. 2.6-Di-rerr-butyl-4-methylpyridine (DTBMP) w a s sublimed for several times under vacuum. Polymerizations were made in vacuum in an apparatus permitting sampling through "Rota- flow@" stopcocks. A bulb containing the initiator solution in CH,Cl, was broken to start the reaction and the kinetics were followed by analysis of samples deactivated by a pyridine solution (first sample after 40 s to 1 min). Samples were analyzed using gel permeation chromatography (GPC) with an instrument from Waters Ass., using the 7 following columns with nominal pore sizes lo5, lo4, lo3, 500.2 x 100and 50 A; toluene was generally used as solvent, tetrahydrofuran (THF) being used for some experiments with naphCOC1. Gas chromatography was carried out with a Carlo Erba GC 6OOO equipped with a capillary column and temperature programming between 35 and 320 "C. 'H NMR spectra of CH,COCI. SbCI, were obtained with a Varian (90 MHz) and those of naphCoCl . SbCI, with a Bruker (250 MHz) apparatus. Living polymers from cyclodimethylsiloxanes through non-protonic initiation 17 Results and discussion Association complexes of SbCI, with acyl chlorides Complexes between antimony pentachloride and acyl chlorides may exist non- ionized or ionized according to the solvent We examined their nature in CH,Cl, solution. For CH,COCI.SbCI, (l), the 'H NMR spectrum exhibits two singlets at 2,61 and 2,74 ppm which we attribute, respectively, to free acetyl chloride and to the complex 1 in equilibrium. CH,COCI + SbCI, * CH,COCI. SbCI, 1 According to the ratio of integrations, the amount of complex 1 is 33% at 30 "C and 50% at - 10°C (total complex concentration: 5,3 mol L-I). The formation of an ionic form occurs only for more polar solvents. For example, with 1 in CH,N02/CH,C1, (vol. ratio 33/66), there are three peaks at 2,61 ppm, 2,85 ppm (non-ionic complex) and 3.60 ppm (ionic complex). For the complex with 1-naphthoyl chloride naphCOC1. SbCI, (2) in CH,Cl, , the spectrum is similar to that of 1-naphthoyl chloride alone but again with a shift of 0.12 ppm. It seems that in this case the equilibrium is completely displaced towards complex formation (for a concentration of The UV spectrum of 2 has a maximum at the wavelength 330 nm. The Beer-Lambert law is verified (molar decadic absorption coefficient E , , ~ = 8000 L . mol-' * cm-I). mole L-'). Kinetics of polymerization reaction 'Ikro experiments were conducted in CH2CI, at - 10 "C with SbCI, alone ([SbCl,] = 4,2 - lo-' mole L-I) for initial concentration of D,, [DJ0 = 1,17 mol - L- ' , either in the absence or in the presence of [DTBMP] = 0,21 * mole L-'. In the absence of DTBMP, polymerization occurred rather rapidly, even if it was 2 to 4 times slower than in the presence of CH,COCl. But in the presence of [DTBMP] = 0,21 * lo - , mole L - I the reaction was slowed down considerably and no high-molecular-weight polymer was formed. This shows that the SbCI, solution contained probably traces of HCl which cocatalysed the reaction in the absence of IYI'BMP. Acetyl chloride alone does not initiate the polymerizations of D, . We also observed through 'H NMR that n o complex is formed between DTBMP and SbCl, . With complexes 1 and 2, the polymerizations were very rapid: e. g. 96% yield in 3 min with 3,4 * lo-, mole L-' of 1, giving simultaneously high-molecular-weight polymer (50 wt.-Vo), oligomers (12 wt.-To), D, (25 wt.-Vo), D, (5 wt.-Vo) and smaller quantities of other cycles. With 1, the kinetics of D, consumption follows a first-order law (Fig. l) , and the results obtained in the absence or in the presence of DTBMP are practically identical. This shows that initiation is rapid, that IYI'BMP does not influence the initiation and the propagation rates, and that the initiation is aprotic The yields of the various products are also the same ("hb. 1). 18 P. Sigwalt, P. Nicol, M. Masure 0 1 2 3 L 0 10 20 30 t/min f/min Fig. 1. Fig. 2. Fig. 1. First-order plot for hexamethylcyclotrisiloxane (D,) consumption. Polymerization by [CH,COCI . SbCI,] = 3,4 mole L-' at - 10°C in CH,CI,; 0 : no 2,6-di-terf-butyl-4- methylpyridine (DTBMP) added, ID,], = 1.22 mol L - I ; 0 : with added DTBMP (total concen- tration2,1.10-4mol.L-l) , [D,], = 1,14mol.L-l Fig. 2. First-order plot for hexamethylcyclotrisiloxane (D,) consumption. Initiation by l-naph- thoyl chloride/SbCI, complex [naphCOCI . SbCI,] = 2. mol . L-' in CH,CI, at - 10OC; 0 : [D,], = 0,72mol .L- ' ; r: [D,], = 1,42mol.L-' With 2 as initiator, D, consumption is also of first-order (Fig. 2). It was found that the number of active centres formed is the same for two different monomer concentra- tions (ID,], = 0,72 and 1,42 mol L - I ) and depends only on the initiator concentra- tion. This is quite different from what was observed with CF,SO,H, for which a strong inhibiting effect of the monomer has been o b ~ e r v e d ' ~ ~ ) . The efficiency of the initiator for the formation of high-molecular-weight polymer (deduced from molecular weights, see below) was found to be about 60% with 2 (initial concentration [2], = 2 . mol . L- ' ) and to be lower with 1 (26% for [l], = 3,7. mo1.L-I; 61% for [l], = 2 . lo-, mol . L-I). Unexpectedly with 1, the apparent rate constant of D, consumption was found to be approximately proportio- nal to 111, (for [l], > 2 * mole L-I) which seems to indicate that the dissociation of the complex into its components is not very rapid compared to the initiation rate, when the more concentrated initiator solution is dispersed into the monomer solution. Some information about the initiation rate was obtained with 2 by following the decrease with time of the optical density of the UV maximum at 330 nm (Fig. 3) and taking into account the lower absorption of SbCI, (or SbCI, (?)) at the same wavelength. A first-order decrease was observed up to 50% conversion of 1 giving a n initiation rate constant of ki = 4,4 * L . mol-' . s at - 10 "C. This shows that 90% of the initiator was consumed in 1 min, while monomer conversion was only 9% ([2], = 2 * mol * L-', [DJ, = 0.72 mole L-'). U Y 0, a 3 0 E TB b. 1. Po ly m er iz at io n o f h ex am et hy lc yc lo tr is il ox an e ( D ,) at - 1 0° C in C H 2C 12 in iti at ed by C H ,C O C I. Sb C I, . E ff ec t o f a dd iti on o f a p ro to n- tra p (2 .6 -d i- fe rf -b ut yl -4 -m et hy lp yr id in e (M B M P )) on th e yi el ds o f th e pr od uc ts f ul ly m et hy la te d cy cl ot et ra - ( D 4) , - pe nt a- (D ,), - he m - ( D a) . - he pt a- ( D, ). -o ct a- (D & ?I z Y -n on as ilo xa ne (D 9) , m ac ro cy cl es (M C ), hi gh -m ol ec ul ar -w ei gh t p ol ym er ( H P) an d so m e of t he ir ra tio s % !. [C H ,C O C I. Sb C I,] t C on ve rs io n Y ie ld s in w t.- To H P/ D , H P/ M C M C /D , D 9/ D 6 f of D , in T o r- -. -- -- > a a m o I * L - ’ 2 D , D, D , D7 D , D 9 M C H P - 3. 4. l o- ’ 1 71 0, 4 1,s 22 .4 0.2 1.5 2, 2 8.3 34 .6 1.5 4. 2 0, 37 0. 10 Ua = 1 22 m ol /L ) 3.2 %,5 1. 0 4.5 24 .8 1.2 0.8 1.4 11. 8 51 .2 2.0 4.3 0, 47 0. 06 ¶ 3, 4. lo -’ 0, 8 65 .3 0.4 1, s 20 .3 0, 3 1, 3 1.9 7, l 30 .5 1, 5 4. 3 0. 35 0. 09 ~ ? (I D, ], , = 1 .14 m ol /L ; a [D TB M P] , = 3 96 .2 1, 3 4.3 24 .3 1.2 0.9 1.5 11 .7 50 ,9 2. 1 4.3 0. 48 0. 06 2.1 . lo -‘ ~ O I/ L ) 3 ’ 3 - 0 F 3. 20 a C u 0 2 $0.50. n Q 0.25. 0 P. Sigwalt, P. Nicol, M. Masure 11, , I I , I , I I ' I , ! , ' I , I , i ; ; ,s I 1 L . 8 5 I , I ' . t I , I Fig. 3. ([naphCOCI.SbCI,] = 2,l . lo-, mol . L- ' in CH,CI, at - 10 "C); initial spectrum (- - -); polymerizing solution after 6.5 min; initial hexamethylcyclotri- siloxane concentration [D,], = 0,72 UV spectrum of 1-naphthoyl , \ chloride/SbCI, complex ' 3 '. I ' L 5 I \ .. ,--.. . .. ..... - _ _ ) mol. L- ' (. . . . . . - - _ _ Nature and concentration of the polymerization products The two main products of the reaction are D, and the high-molecular-weight polymer HP, the relative proportion of HP increasing with initiator concentration. From the beginning of the reaction, smaller proportions of octadecamethylcyclonona- siloxane (D9) and tetracosamethylcyclododecasiloxane (D,,) are also formed, together with higher-molecular-weight oligomers (see Fig. 4 for a reaction with a yield of 78% stopped after 95 s). Small quantities of D,, D, , D, and D, are also detected from the beginning of the reaction, the D, and D, concentrations increasing with conversion but remaining low for conversions lower than 90% (Thbs. 1 and 2). A significant result, similar to that observed for D, polymerization initiated with CF,SO,H is that [D,] remains higher than [D J up to 100% conversion of D, and even for longer times (with [D,]/[D,] = 3 during the reaction). A difference with CF,SO,H initiation is, however, Fig. 4. Hexamethylcyclotrisilo- xane (D,) polymerization by CH,COCI. SbCl, at - 10°C. GPC curve (solvent toluene, refractometric detection; sample quenched after 1 min 35 s). [D,], = 1,14 mol . L-I; [CH,COCI. SbCIJp = 3 , 4 . 1 0 - ~ m 0 1 . ~ - ;2,6-di- rerf-butyl4methylpyridine con- centration [DTBMP] = 0,21-10-3mol .L- ' (D6, D,, D,,: completely methylated cyclohexa-, cyclonona-, cyclo- dodecasiloxanes, resp.; HP: high- molecular-weight polymer; olig. sup. stands for macrocycles (MC); dotted curve corresponds to a Gaussian distribution of H P HP lo5 loL lo3 Molecular weight in polystyrene equivalents c. 5 09 U 0 3 5 2 3 lh b. 2 . Ef fe ct o f in iti at or a nd m on om er c on ce nt ra tio ns o n th e yi el ds o f th e pr od uc ts f or h ex am et hy lc yc lo tr is il ox an e (D ,) po ly m er iz at io n in C H ,C I,. C om pa ris on o f Sb C I, /a cy l ch lo rid e co m pl ex es (a t - 10 "C ) a nd C F, SO ,H (a t 2 0 "C ) a s in iti at or sa ); (n ap hC O C 1 = 1 -n ap ht ho yl c hl or id e) ; D 31 0: in iti al co nc en tra tio n of D , 0 R Fi H P H P M C D 9 D6 M C D6 D, - - - - [I ni tia to r] [D ,], f C on v. Y ie ld s i n w t.- % m oI .L -' m ol /L m in of D3 - - __ - - ln % D, D, D6 D, D, D, M C H P ?I 3, 7. 1, lS 1 17 0. 10 0, 08 8, 5 0. 08 0, 11 0. 8 3, l 3, 2 0.3 7 1, 0 0, 36 0, 09 -. 7, 5 51 .7 0, 2 0, 4 30 ,6 0. 15 0, 23 3, 7 8, 5 7, 9 0, 26 0, 93 0, 27 0, 12 ' 1, 3 57 .7 0, 3 1.5 21 .2 0, 5 0, s 1, 9 8.3 23 .2 1, l 2, 8 0 3 0, 09 $ [C H ,C O C I * Sb C l,] : 2 . lo -, 1, 18 0, 6 29 0, 15 0. 45 13 ,5 0, 17 0, 37 1.3 6, 3 6.5 0, 48 1, 0 0, 46 0, lO 3, 4. l o- ' 1, 22 1 71 0, 4 1, 5 22 .4 0.2 1, 5 2, 2 8, 3 34 ,6 1. 5 4.1 0, 37 0, lO 3 [n ap hC O C I Sb C I,] : 5 2.1 1 0 -~ 0, 72 6 38 ,9 0. 17 0, 30 19 0, 12 0. 10 5, O 4.0 6,O 0, 31 1.5 0, 21 0, 27 2 a 4, 6. 1 0 -~ l, o 45 43 0, 2 0. 35 14 .5 b, b, 3, 9 3.0 20 ,7 1, 42 6, 9 0, 20 0. 27 E. 1. 25 lo -, 1 ,o 4, 3 60 0.1 0, 3 10 ,5 b, b, 4, l 20 .1 25 .6 2. 44 1.3 b, 4, 4 7, 1 21 ,3 0. 90 3.0 0, 30 0, 18 c. 7. 1 lo -, 0, 42 4.5 59 0, 4 1, 0 23 ,5 b, 7, i . lo -' 1, 63 9 15 0, 04 0, lO 2, 8 b, b, 1.3 3.1 7, 2 2.6 2, 3 a) Fo r e xp la na tio n of a bb re vi at io ns D , -D ,, M C a nd H P, s ee le ge nd lh b. 1 . b, N ot m ea su re d. C wa 1, 9. 10 -~ 1, 40 6 35 ,8 0, lO 0, 20 19 ,6 0, lO 0, 17 4, 2 2, 2 6, 4 0, 32 2, 9 0. 11 0, 21 $ [C F, SO ,H ]: 6 1, 9 0. 39 s: 1.1 0. 46 2. a 22 P. Sigwalt, P. Nicol, M. Masure Fig. 5 . Hexamethylcyclotrisiloxane (D,) polymerization by 1-naphthoyl chloride/SbCI, complex [naphCOCl . SbCI,] = 2.6. m o l . L - . ' a t - 1 0 0 ~ ; [ D ~ I ~ = i,04 mol . L - I . Sample quenched after 4 min. GPC with double detection (refractive index (R.I.) and U.V.) in THF after elimination of the small cycles (molecular weight of peak Mpeak = lo5 molecular-weight polymer; dashed curve shows that the distribution of HP is not symmetrical (see text); dotted curve: U.V. absorption of naphthoyl end-groups , - U.V. Macrocycles 65 La 50 55 in polystyrene equivalents). HP: high- Elution volume in mL that the relative proportion of D, is smaller: [D9]/[D6] = 0,l for 1 and 0,3 to 0,s for CF,SO,H, for comparable yields and initiator concentrations. The higher- molecular-weight oligomers have an approximately constant mass concentration (for GPC with double detection of a polymer initiated by 2 (Fig. 5 ) shows that UV absorption under the high-molecular-weight polymer peak increases relatively for lower M, as may be expected for end-groups, but disappears for the lower-molecu- lar-weight oligomers. This confirms that they very probably consist of macrocycles (MC). 2.103 G M G 2.104). 0 50 100 50 100 Conversion o f D, in % Fig. 6. Fig. 7 Conversion of D, in % Fig. 6. Polymerization of hexamethylcyclotrisiloxane (D,) ([D,], = 1,18 mol . L- ' ) initiated by [CH3COCI * SbCI,] = 3.7 * mol * L- ' in CH,CI, at - 10°C. Variation of weight fraction of the various species with conversion. o : HP (high-molecular-weight polymer); +: MC (macro- cycles); 0 : dodecamethylcycloheiloxane (D6); A: octadecamethylcyclononasiloxane (D,) Fig. 7. Polymerization of hexamethylcyclotrisiloxane (D ) ([DJ0 = 0,72 mol . L-I) with 1-naphthoyl chloride [naphCOCl] = 2.1 . mol . L - at - 10°C. Formation of high- molecular-weight polymer (HP) and cyclic species. .: HP; C: MC (macrocycles); 0 : dodeca- methylcyclohexasiloxane (D6) ; : octadecamethylcyclononasiloxane (D9) 1 Living polymers from cyclodimethylsiloxanes through non-protonic initiation 23 The variation of the weight fraction of the main products with conversion is shown in Fig. 6 for 1 and in Fig. 7 for 2. It may be seen that there is a proportional increase of the concentrations of D,, HP, MC and D, up to about 50% conversion with 1, and 40% conversion with 2. For higher conversions, [D,] increases less rapidly, [D,] is nearly constant, and HP and MC concentrations increase more rapidly. With D, as monomer (for experimental conditions as in Fig. 9), the reaction is slower than for D, by a factor of about 10, but the internal first-order in monomer is also observed, in agreement with a constant concentration of active centres. The type and amount of cyclic products is, however, quite different to the D, case. For example, a t a conversion of 30% (in 1,4 h) the yield of high-molecular-weight polymer was 19% and that of macrocycles 7,5%. The small cycles consisted mainly of D, and D,,, and of much smaller quantities of others (D,, D,, D,, D,, D,). D,, might eventually be formed by end-biting, but the mode of formation of D, is not clear. Variation of the molecular weight of the high-molecular-weight polymer with conversion The molecular weight distributions of the high-molecular-weight polymer fraction obtained with both initiators 1 and 2 remain approximately constant. It is not, however, always symmetrical, as may be seen by comparing Fig. 4 (initiation by [l] = 3,4 * lo-’ mol * L-I) for which it is, with Fig. 5 (initiation by [2] = 3,7 * lo-, mol . L-I) for which it is not. In various experiments conducted with 1, we found values of the ratio weight- to number-average molecular weights 1,3 Q Mw/@,, Q 1,4, without significant variation with conversion up to 80% conversion. The relative symmetry of the distribution led us to plot the apparent molecular weight a t the maximum (in polystyrene equivalents from GPC calibrated with poly- styrene standards I,)) against conversion. With polymers initiated by 1, the variation is linear starting from the origin up to D, conversions higher than 80% (Fig. 8). For polymers initiated by 2, the linearity is observed up to 50-60Vo conversion but the molecular weight then increases more rapidly with conversion. However, when the molecular weight is plotted against the weight fraction of polymer formed, the varia- tion is again linear (Fig. 9). All this shows the presence during the various experiments of a constant number of macromolecules with high molecular weight, i. c, the presence of a living polymer. It is, however, possible that a cyclic fraction of high molecular weight also contributes to the main peak. The more rapid increase of the molecular weight for high D, conversion may be explained by the copolymerization of D, and D, with D, , in agreement with the concentration variation of the former discussed above. With D, as monomer, the molecular weight of the high-molecular-weight polymer formed (Fig. 9) increases also approximately proportionally to its weight fraction (with some deviation for the low molecular weights). This again indicates the presence of a relatively constant concentration of growing macromolecules. 24 P. Sigwalt, P. Nicol, M. Masure 1 7 HP in w t - % I 17 30 52 86 Converslon ri % 1 "0 10 20 30 HP in wt.-% Fig. 8. Fig. 9. Fig. 8. Polymerization of hexamethylcyclotrisiloxane (D,) by CH,COCI. SbCI, at - 10 "C. Variation of molecular weight MWak (at GPC peak, in polystyrene equivalents of high-molecu- lar-weight polymer (HP)) with conversion. Initial concentration [D,], = 1.18 mol . L-I; [CH,COCI. SbCI,] = 3,7. Fig. 9. Polymerization of hexamethylcyclotrisiloxane (D,) and dodecamethylcyclohexasiloxane (Dd with 1-naphthoyl chloride/SbCI, complex [naphCOCI. SbCI,] at - 10°C. Increase of Mwk (see legend Fig. 8) with the amount of high-molecular-weight polymer (HP) formed. A: ID,], = 1,42 mol.L- ' , [naphCOCI.SbCI,] 1,9.10-4 mo1.L-I; A: [D,], = 0,72 [naphCOCI . SbCI,], = 2,4. mole L- ' mole L-I, [naphCOCI. SbCI,], = 2,l . 10- s = mole L-I; 0 : [D6], = 0,70 mol . L- ' , mol . L- ' Mechanism of formation of the cyclic species As for the initiation with CF,S0,H6*8), D,, D, and macrocycles are formed proportionally to high-molecular-weight polymer up to at least 50% conversion of D,. Chojnowski et al. explained this for the case of the acid by the occurrence of end- to-end ring closure reactions competitive with the formation of the high-molecu- lar-weight polymer by chain propagation. If ring opening of D, with an initiator AB gives two different end-groups A and B, for propagation occurring on active centres A the following general reaction scheme (2) was proposed6) (symbol k denoting the various rate constants): BD,A BD6A BD& BD,,A HP I k6 I kg D6 + AB D, + AB D,, + AB The preference for ring closure (or "end-biting") rather than for back-biting was attributed to the strong reactivity of the terminal group B. The probability of ring Living polymers from cyclodimethylsiloxanes through non-protonic initiation 25 closure is highest for short chains and decreases rapidly, high-molecular-weight polymer being formed when the polymer chain is long enough. Such a scheme would, however, lead to continuous reformation of initiator or initiating species AB when the polymerization goes on, and to a reinitiation with formation of new macromolecules during the whole polymerization. This is obviously not the case in cationic polymeri- zation of D, for which the number of growing macromolecules remains constant with conversion, as well for the non-protonic initiators described in the present paper as for polymerizations initiated with CF,S0,H8). How then may we explain the simultaneous formation of a living high-molecu- lar-weight polymer, of macrocycles and of large quantities of small cycles D,,? A first possibility might be the presence of two types of active centres, one type giving the high- molecular-weight polymer and the other one the cyclic compounds (eventually by end- to-end ring closure). However, since the relative proportion of HP to D, (and that of H P to MC) increases strongly (by a factor of up to 4) with the concentration of 1 (see Tab. 2), one would not expect the approximately first order of the global rate observed according to initiator concentration, because the relative proportion of the two types of active species should vary. The explanation we propose is the presence of two populations of macromolecules with similar active centres but differing by the nature of the end-groups. The popula- tion giving the living high-molecular-weight polymer would bear either two active centres or one active centre and one non-reactive end-group, while the population giving the macrocycles (and eventually also small cyclics) would bear one active centre and one reactive end-group permitting end-biting, with continuous reinitiation as in scheme (2). The possibility to form small cycles by a ring-expansion mechanism is not completely excluded, but seems to us unlikely for the formation of macrocycles. And since the relative proportion of D, and macrocycles does not change with initiator concentra- tion (contrary to the case of CF,SO,H initiation (see Xab. 2)), while the proportion of high-molecular-weight polymer changes considerably, this is in favour of their formation by a similar mechanism. The reactive end-group permitting the end-biting is probably a carboxylic ester resulting from initiation and containing a nucleophilic oxygen leading to end-biting by reaction with an activated Sib+ or with Si@. We verified that such a reaction may occur easily, as well as the reverse reaction, by examining, using 'H NMR, the model reaction between acetoxytrimethylsilane and chlorotrimethylsilane in the presence of SbCI,, as well as the reverse reaction between hexamethyldisiloxane and 1, which both lead to the equilibrium (K, : equilibrium constant): 26 P. Sigwalt, P. Nicol, M. Masure CH,CO,Si(CH,), + CISi(CH,), (CH,),SiOSi(CH,), + CH3COC1 (4) SbCI, The forward reaction (with [CH,CO,Si(CH,),] = [CISi(CH,),] = [SbCI,] = 4,l * lo-, mole L-I) was rapidly equilibrated a t - 10 “C in CH,CI,. i.e. in less than 1 min at this concentration (with K , = 4). It is difficult to compare quantitatively this bimolecular reaction with ring-closure which is unimolecular, but the latter might occur even more rapidly, particularly in dilute solutions. The end-groups other than active centres are then the acyloxy and chlorosilane groups. The unreactive end-groups are probably the chlorosilane ones, since the presence of C1 deactivates the oxygen of the last siloxane group towards electrophilic attack: I I I I - Si--(kSi+CI In order to test the possibility of activation of the >Si-Cl group for initiation and to verify that chlorosilane end-groups are little or non-reactive for back-biting, we have initiated the polymerization of D, by mixtures of SbCI, with dichlorodimethyl- silane”). With an excess of 100% of SbCI, over “active” >Si-CI ([CH,),SiCI,] = 7,2 * mole L-’; [SbCI,] = 2,9. lo-, mol. L--’) at - 10°C in CH,Cl,, it was found that no macrocycles were formed at the beginning of the reaction (but appeared above 80% conversion) and that the cycles, in low concentration only, occurred in quite different proportions (with [D,] > [DJ > [D,] > [DJ) to those observed when initiating with the complexes of acyl chlorides. This seems to confirm that end-biting does not occur easily with >Si-Cl end-groups. The relative yield of polymer is much higher (= 84% of the total yield at 77% conversion) and its molecular weight grows with conversion, but not linearly. It should be noted that the global polymerization rate is much lower than with similar concentra- tions of 1 or 2 (by a factor of 10 to 20), and since DTBMP was not added in these experiments, a possible initiation involving SbCI, and HCI cannot be excluded. In this case one would, however, have expected the formation of macrocycles by end-biting at terminal hydroxyl groups. It then seems that the active centres in this case might be different from those operating with 1 or 2. We have examined the possible formation of ionic species in CH,CI, by measuring the conductivities of solutions of (CH,),SiCI,, of SbCI, and of their equimolar mixture. The conductivity of the mixture was a little smaller than the sum of those of the individual components, which excludes the occurrence of ioniza- tion. And during a polymerization initiated with this mixture, no significant conductiv- ity change was observed. This shows that covalent species - or a minute amount of ions in equilibrium (?) - might be responsible for the formation of both polymer and cyclics in this case. Living polymers from cyclodimethylsiloxanes through non-protonic initiation 27 Mechanism of polymer formation The nature of the active centres in polymerizations initiated by 1 and 2 is still unknown. They might be chlorosilane or acyloxysilane groups activated by SbCI, (respectively, 3 and 4) I I Si-OCOR . SbCI, --- Si-CI . SbCI, --- I 3 ‘ 4 or siloxonium species 5 or 6, eventually in equilibrium with 3 and 4: 5 6 In all cases, they might behave as transitory silylium ions, as written in Eq. (3).High- molecular-weight polymer growth might occur either on one active end (3 to 6), or on two active ends. This last possibility would exist if both end-groups would be activated by SbCI, . It is still not known whether non-reacting end-groups consist only of chlorosilane groups or also of acyloxy groups stabilized by complexation with SbCI, (4, if this is not an active centre for polymer growth). Some of the RC0,Si groups d o not give end- biting in the naphthoyl chloride case, since they are found in the high-molecular-weight polymer (see Fig. 5 ) but this is perhaps specific of these more bulky groups which might react less readily than the acetoxy groups. Another possibility might also be that these end-groups stem from active centres of type 6 and are reincorporated into the polymer during the termination by pyridine addition. What is the reason of the relative increase in high-molecular-weight polymer yield when the initiator concentration increases (see B b . 2)? This might be linked to an equilibrium established during the initiation period according to Eq. (1). Its displace- ment towards the left might decrease the initiator efficiency for the formation of high- molecular-weight polymer if only complexes 1 or 2 give the active centres involved. But the complexation of RCO,Si< by SbCI, would also be favoured at higher initiator concentration. According to the “two populations” hypothesis (see above), the same types of centres may be active for the formation of high-molecular-weight polymer and macrocycles. But are the same active centres responsible for D, formation? The very large yield of D, compared with that of high-molecular-weight polymer is not in favour of their formation by end-biting Is), and seem to imply that their formation either occurs on active centres a t the end of the polymer chain or involves other reactive groups. The first possibility may be either ring expansion or a very selective type of back-biting. Some preliminary results are not in favour of ring expansion involving stable siloxonium ions. If they existed, their formation would be expected to be disfavoured 28 P. Sigwalt, P. Nicol, M. Masure in a less polar solvent. For a polymerization initiated by 2, by using a mixture (heptane/CH,Cl, (volume ratio 60140)) with a dielectric constant of about 4,8, the global rate decreases by a factor of about 3 only. But the formation of high- molecular-weight polymer decreases strongly relative to the formation of D, and D, (lhb. 3). This result shows that while we cannot exclude the possibility for siloxonium ions to be species active for polymer formation, they are not responsible for the forma- tion of D, by a simple ring-expansion mechanism. On the other hand, D, (and D,. . .) might be formed by end-biting between end-groups >SiOCOR and >Sic1 in equilibrium, in the presence of SbCl, . Xib. 3. Polymerization of hexamethylcyclotrisiloxane (D,) with the complex 1-naphthoyl chloride/SbCI, (naphCOC1. SbCI,) at - 10°C. Influence of the solvent on the relative propor- tion of the various products dodecamethylcycloheiloxane (D,), octadecamethylcyclonona- siloxane (D9), tetracosamethylcyclododemsiloxane (D,,), macrocyclics (MC) and high-molecu- lar-weight polymers (HP) (ID,]: (a) 0,72 mol * L-I, (b) 0,73 mol . L-'; [naphCOCI * SbC15]: (a) 2.1. I O - ~ ~ O I . L - ' , (b)2. I O - ~ ~ O I - L - ' ) Solvent t/min Conversion Yields in wt.-'To of D, in To \ D,j D9 Dl, MC HP (a) CH2C12 15 61,4 27.6 5,2 2,9 8,7 15,2 24 79.2 34,4 4,9 1,4 11,l 24,2 (vol. ratio 40:60) 150 78,2 38,6 18.5 5.2 8,O 7,9 (b) CH,CI,/heptane 100 64.6 29,2 18,6 3,6 8,4 4,8 Another possibility might be that D, formation results from a particular type of back-biting reaction involving an eight-membered cyclic transition state as shown in the following scheme: D, might be formed in a similar way but its probability of formation would be low on account of the lower stability of the siloxonium involving D,. This type of transition state would be strongly favoured by comparison with that which would give a smaller ring (like D,). But one would have also to assume, and this is less obvious, that larger cyclic transition states (which would give D, and Ds) are strongly disfavoured for conformational reasons, or that these monomers are consumed much more rapidly than D,. Living polymers from cyclodimethylsiloxanes through non-protonic initiation 29 Conclusion It was shown that rapid and efficient non-protonic initiation of the cationic poly- merization of D, is possible. There are many similarities, as well as some differences, in the polymerization of D, (and D,) initiated by these non-protonic initiators or by trifluoromethanesulfonic acid. Among the similarities, the use of initiators from complexes of SbCI, with acyl chlorides did not lead to a reduction in the proportion of macrocycles and of small cycles which could be even formed in larger proportion than with trifluoromethanesulfonic acid. In both cases, the small cycles are mainly of the D,, type and are formed proportionally to the high-molecular-weight polymer, and there is also a slow formation of D, in a larger amount than that of D,. The most striking similarity is the linear growth (starting from the origin) of the molecular weight of the high-molecular-weight polymer with the amount formed. The main differences between the behaviour of the two types of initiators were found for the kinetics of the polymerization of D,. With CF,SO,H, the external apparent first-order rate constant was of third order in CF,SO,H, and of a negative order (- 1,8) in [D,], and the explanations offered are still conjectural8). Contrarily with initiation by complexes of acyl chlorides, the more usual and intelligible first orders (external) in monomer and initiator concentrations were observed. With both types of initiators, the same general explanation may be offered for the simultaneous and proportional formation of the various products (and particularly that of the macrocycles (MC)), viz. the presence of two populations of macromolecules. One of those would consist of a living polymer, and the other one of polymer chains having reactive end-groups permitting the formation of cyclics. This hypothesis may explain satisfactorily the formation of high-molecular-weight polymer and macrocycles with the two families of initiators, but the interpretation of the formation of small cycles (particularly D,) is not so obvious. With CF,SO,H, the ratio [MC]/[D,] increases (while the ratio [HP]/[MC] decreases) when the initiator concentration increases; this seems to indicate that macrocycles are formed by end- biting (with external assistance by S O H groups) but that D, is formed by another mechanism. With the acyl chloride complexes the ratio [MC]/[D,] does not vary with initiator concentration, which points out to a formation of D, and [MC] depending only on the concentration of active centres. But it is not possible to say whether macrocycles and D, are formed or not by the same mechanism, because the kinetics would be similar for D, formation by back-biting, ring-expansion or end-to-end ring closure (when the latter is unimolecular). More data about the variation of [MC]/[Dd according to experimental conditions should be obtained, and eg. about the effect of adding an excess of SbCI, to the complexes. The preliminary results obtained with (CH,),SiCI, and SbCI, seem to show that it is possible to reduce more efficiently the quantity of macrocycles than that of D,, and even to suppress macrocycles completely with an excess of SbCI,. More systematic work with related initiators is under way, with the aim to identify the active centres, responsible for the formation of the various products. 30 P. Sigwalt, P. Nicol, M. Masure I ) S. Penczek, P. Kubisa, K. Matyjaszewski, Adv. mlym. Sci. 68-69, 216 (1985) ’) P. Sigwalt, Actual. Chim. 1986 (no. 3), 45 3, J. Chojnowski, R. Rubinsztajn, K. Wilczek, Acrual. Chim. 1986 (no. 3). 86 4, G. Sauvet, J. J. Lebrun, P. Sigwalt, in “CationicPolymerization andRelated Processes’: edited by E. J. Goethals, Academic Press, New York, N.Y. 1984, p. 237 C. Gobin, M. Masure, G. Sauvet, P. Sigwalt, Makromol. Chem.. Macromol, Symp. 6, 237 (1 986) ‘) J. Chojnowski, M. Scibiorek, J. Kowalski, Makromol. Chem. 178, 1351 (1977) T, J. Chojnowski, L. Wilczek, Makromol. Chem. 180, 117 (1979) P. Sigwalt, Polym. J . 19, 567 (1987) A. I. Chernyshev, U. V. Yastrebov, Vysokomol. Soedin., Ser. A: 11, 525 (1969) lo) T. C. Kendrick, J. Chem. SOC. 1965, 2027 ‘ I ) I. Rashkov, I. Gitsov, I. Panayotov, Polym. Bull. (Berlin) 10, 481 (1983) 12) J. Collomb, P. Arlaud, A. Gandini, H. Chkradame, in “Cationic Polymerization and Related Processes’: edited by E. J. Goethals, Academic Press, New York, N.Y. 1984, p. 49 S. Boileau, H. Chkradame, A. Gandini, E. Jordan, L. Lestel, 5th International Symposium on Ring-opening Polymerization, Blois, France, 22 - 26 June 1986, Preprints, Communication C13, p. 79 14) G. A. Olah, S. J. Kuhn, S. Tolgyesi, E. B. Baker, J. Am. Chem. SOC. 84, 2733 (1962) lJ) G. A. Olah, H. E. Moffatt, S. J. Kuhn, B. A. Hardie, J. Am. Chem. SOC. 86, 2198 (1964) L. Mandik, A. Foksova, J. Foltyn, J. Appl. mlym. Sci. 24, 395 (1979) G. Toskas, to be published ’*) S. Slomkowski, Makromol. Chem. 186, 2581 (1985)