Ring-Opening Polymerization of Lactones Catalyzed by Silicon-Based Lewis Acid

. Many catalysts containing various elements at their active sites have been reported for the ring-opening polymerization (ROP) of cyclic esters. However, to our knowledge, silicon-based catalysts for ROP have never been reported. Here we report the ROP of cyclic esters and cyclic carbonates catalyzed by the derivatives of bis(perchlorocatecholato)silane (Si(cat Cl ) 2 ), which is a neutral silicon-based Lewis acid recently reported by Greb et al. The catalyst systems show high activity for the ROP of seven-and six-membered ring monomers such as ε -caprolactone, δ -valerolactone, and trimethylene carbonate to produce the polymers with molecular weights up to 32kg/mol. The matrix-assisted laser desorption/ionization time-of-ﬂ ight mass spectrometry (MALDI-TOF MS) and nuclear magnetic resonance analysis of the obtained polymers indicates the predominant formation of cyclic polymers.


Introduction
Aliphatic polyesters such as poly(ε-caprolactone) (PCL) and poly(L-lactide) are mainly synthesized by the ring-opening polymerization (ROP) of the corresponding cyclic ester monomers. One of the most popular catalysts is tin(II) 2-ethylhexanoate (Sn(Oct) 2 ), which has been widely used in both industrial and academic applications of the ROP of cyclic esters [1][2][3][4][5]. Sn(Oct) 2 is relatively easy to handle because it is bench stable and moderately active. The molecular weights of the resulting polymers can be controlled by using protic compounds such as alcohols as initiators and adjusting the feed ratio of the monomer and the initiator. However, the catalyst residue of Sn(Oct) 2 could potentially be harmful, especially in biomedical applications of biodegradable aliphatic polyesters [6][7][8]. Therefore, alternative catalysts to Sn(Oct) 2 are required. Various catalyst systems have been developed for the ROP of cyclic esters [9][10][11][12][13][14][15][16][17]. Several metal-free organocatalysts and enzyme catalysts have also been developed [10,18,19]. Among group 14 elements, carbon [10,18,19], germanium [20][21][22][23], and lead [24,25] have also been applied to the active site of the ROP catalysts as well as tin. However, to the best of our knowledge, silicon has never been applied to the reaction site for the catalyst for the ROP of cyclic esters [14]. Silicon is abundant and low in elemental toxicity; therefore, a silicon-based catalyst for the ROP of cyclic esters can be interesting, if it has high catalytic activity. Some Lewis acids have been known to catalyze the ROP of cyclic esters. Methylaluminum bis(phenolate), MeAl(OC 6 H 3 Ph 2 -2,6) 2 , promoted the ROP of ε-caprolactone (ε-CL) to produce PCL with narrow molecular weight distribution [26]. The ROP of ε-CL by scandium trifluoromethanesulfonate (Sc(OTf) 3 ) has been proposed to proceed by a monomer-activation mechanism [27,28]. A borane Lewis acid, B(C 6 F 5 ) 3 , promoted the controlled ROP of cyclic esters such as ε-CL [29]. We have reported that Al(C 6 F 5 ) 3 (THF) can promote the ROP of ε-CL, while that of lactide required the combination of Al(C 6 F 5 ) 3 (THF) with bulky phosphines [30].

Results and Discussion
2.1. ROP of ε-CL. In this work, acetonitrile or diethyl ether adducts of bis(perhalocatecholato)silane derivatives [31,32] were used as catalysts rather than donor-free ones for the ease of preparation and handling. The results of the ROP of ε-CL by Si(cat Cl ) 2 (MeCN) 2 (1a), Si(cat Cl ) 2 (Et 2 O) 2 (1b), Si(cat Br ) 2 (MeCN) 2 (2a), and Si(cat F ) 2 (MeCN) 2 (3a) are summarized in Table 1. As a result, 1a was found to catalyze the ROP of ε-CL in toluene at 100°C to produce PCL with M n of 12 kg/mol (run 1). To the best of our knowledge, this is the first example of the ROP of cyclic esters using Si-based catalysts. Because the reported Lewis acid-catalyzed ROPs have often been combined with some alcohols as initiators [26][27][28][29], the ROP of ε-CL using 1a was examined in the presence of different amounts of benzyl alcohol (BnOH) (runs 1-3). With increasing feed amount of BnOH, the polymer yields decreased, indicating that BnOH was not necessary for the catalyst system of 1a, but rather hampered the reaction. This is in sharp contrast to that the polymerization of ε-CL catalyzed by MeAl(OC 6 H 3 Ph 2 -2,6) 2 [26], Sc(OTf) 3 [27,28], and B(C 6 F 5 ) 3 [29] can be co-initiated with alcohol. We speculate that the addition of BnOH should cause the decomposition of Si(cat Cl ) 2 by protonolysis. The increased 1a feed amount enhanced polymer yield and did not affect the molecular weight of the resulting polymer (run 4). The diethyl ether adduct 1b showed similar activity to that of 1a to give PCL with lower molecular weight (run 5). The activities of 2a and 3a were compared with that of 1a (runs 6 and 7 vs. run 1). In comparison with 1a, the polymer yield by 2a was lower despite the higher Lewis acidity of 2 than that of 1 [31,32]. We speculate that the lower stability of 2 could result in faster deactivation to cause a lower polymer yield. The gel permeation chromatography (GPC) curve obtained with 2a showed a peak having a shoulder ( Figure S11), while that obtained with 1a showed an unimodal peak ( Figure S10), suggesting a multiplicity of active species in the 2a system possibly due to the partial decomposition of 2a. The fluoro-derivative 3a did not produce PCL under the same conditions. When the ε-CL polymerization by 1a was performed at 60°C (run 8), the polymer yield decreased to 35%, while M n increased to 17 kg/mol compared with those at 100°C (run 1). The attempted polymerization of ε-CL at room temperature (r.t.) did not give PCL (run 9). The increase and decrease of the feed monomer to catalyst ratio (runs 10 and 11 vs. run 1) resulted in lower and higher polymer yield, respectively, and the molecular weight of the resulting PCL was somewhat decreased in run 11. When the initial monomer concentrations were increased from 0.41 to 3.0 mol/L (runs 12, 1, and 13), quantitative polymer yield was achieved, and the M n of the resulting PCL increased from 8 kg/mol to 15 kg/mol. Thus, bulk polymerization of ε-CL using 1a was examined (runs [14][15][16][17]. The bulk ROP of ε-CL by 1a was almost completed within 10 minutes to give PCL with M n of 29 kg/mol (run 15), which demonstrated high catalytic activity of 1a. The increased feed monomer to catalyst ratio from 30 : 1 to 100 : 1 increased the molecular weight of the resulting polymer (run 15 and 16), while a further increase to 300 : 1 did not increase the molecular weight (run 17). Perchlorocatechol (cat Cl H 2 ) did not polymerize ε-CL under the same conditions, suggesting that the active site in this polymerization catalyst should be the Si-center of 1a.
The PCL with rather low molecular weight obtained with 1b-BnOH in run 5 was analyzed by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) ( Figure 1). The spectrum of the PCL exhibited a series of repeating peaks separated by the molecular weight of ε-CL, which were in good agreement with the formation of cyclic PCLs. This result also suggests that BnOH was not acting as an initiator in this system despite the presence of BnOH in run 5. The MALDI-TOF MS spectrum of the PCL with relatively low molecular weight obtained with 1a without BnOH (run 12) is shown in Figure S5. The spectrum also exhibited a series of repeating peaks separated by the molecular weight of ε-CL. The peaks around m/ z = 1400-1900 are in good agreement with cyclic polymers ( Figure S5(b)). On the other hand, the peaks of H 2 Oterminated species were also observed in the higher m/z region ( Figure S5(c)). The MALDI-TOF MS spectrum of the PCL obtained by bulk polymerization with 1a in run 16 showed the peaks for H 2 O-terminated linear PCLs with comparable intensity to those for cyclic PCLs ( Figure S6). Thus, the polymerization under lower concentration enhanced the selectivity for cyclic polymers as expected.
The 1 H nuclear magnetic resonance (NMR) spectrum of the PCL (run 1, Figure S1) showed the signals of the methylene protons next to the terminal OH group at 3.65 ppm (e ′ ) as well as those of the methylene signals in repeating units at 4.0 ppm (e). The molecular weights of PCLs determined by GPC (polystyrene standards, THF)  International Journal of Polymer Science  3 International Journal of Polymer Science were reported to be close to those determined by NMR [34]. From the M n value in run 1 (12 kg/mol) determined by GPC analysis, the intensity ratio of internal and terminal methylene protons of 105 : 1 can be expected. However, the observed intensity ratio from the spectrum is approximately 492 : 1, indicating a much smaller amount of the terminal group than that expected from the molecular weight. This also supports the predominant formation of cyclic polymers. Although the reaction mechanism is not clear yet, the speculated mechanism for the formation of cyclic polymers is shown in Scheme S1.
The ROP of δ-VL by 1a (runs 19-22) proceeded similarly to those of ε-CL. Bulk polymerization of δ-VL by 1a rapidly proceeded to afford poly(δ-VL) with M n of 20-25 kg/ mol, while solution polymerization gave those with lower M n . The 1 H NMR and MALDI-TOF MS spectra of the resulting poly(δ-VL) ( Figures S2 and S7, respectively) indicated their predominant cyclic structure as observed in the case of PCL. The polymerization of β-PL by 1a gave low molecular weight poly(β-PL) in low yield (run 23). The electrospray ionization mass spectrometry (ESI MS) ( Figure S8) and 1 H NMR spectrum ( Figure S3) of the obtained poly(β-PL) suggested multiple peaks indicating its complicated chain ends. Bulk polymerization of TMC by 1a proceeded to produce poly(TMC) with M n of 21 kg/mol in a quantitative yield within 10 minutes (run 28). The solution polymerization of TMC by 1a produced poly(TMC) with lower molecular weight (run 29). The bulk polymerization of TMC by 1a was accompanied by decarboxylation to form some ether linkage (ca. 1.5 mol% in run 28) as indicated by the 1 H NMR and ESI MS analysis of the produced poly(TMC) (Figures S4 and  S9, respectively). In comparison with bulk polymerization of TMC, the decarboxylation was hampered in solution polymerization (ca. 0.1 mol% in run 29), although the polymerization rate was much lower. The polymerization of DTC by 1a (run 30) was slower than that of TMC (run 28) possibly due to steric hindrance and also accompanied by decarboxylation (ether bond content 1.0 mol%).
In general, glass-transition temperatures (T g s) of linear polymers tend to decrease with decreasing their molecular weights due to the increasing effect of their end groups. In contrast, cyclic polymers generally show little dependence of T g s on their molecular weights due to their constrained topology and the absence of chain ends, resulting in higher T g of cyclic polymers than those of linear polymers with relatively low molecular weights [35,36]. Thus, the T g values of the polymers synthesized by 1a were compared with those synthesized by typical catalysts to produce linear polymers    Table 3). The PCL synthesized by 1a showed T g at −46.2°C, which is substantially higher than that of linear PCL (−60.8°C) synthesized by Sn(Oct) 2 -BnOH [37], supporting its predominant cyclic structure. Similarly, the polymers of δ-VL and TMC produced by 1a also showed higher T g than those obtained by Sn(Oct) 2 -BnOH or 1,5,7-triazabicyclo [4.4.0]dec-5-ene (TBD)-BnOH [38,39]. These results also support the formation of cyclic polymers by 1a. On the other hand, the poly(β-PL) synthesized by 1a showed lower T g than that synthesized with Al(O i Pr) 3 [40], suggesting its linear structure.

Conclusions
The Si-based Lewis acid, 1a, was found to exhibit high catalytic activity for the ROP of seven-or six-membered cyclic esters and cyclic carbonates without substituents. 1a promoted the ROP without an alcoholic initiator and afforded cyclic polymers predominantly. Such Si-based compounds can be promising catalysts for the ROP of cyclic esters and cyclic carbonates due to the high natural abundance and low elemental toxicity of silicon. In comparison with the reported catalyst systems based on other group 14 elements, the present silicon-based catalyst is unique in that cationic polymerization proceeds without proton mediation.

Data Availability
The experimental section and the data used to support the findings of this study are included within the article and the supplementary information file.

Conflicts of Interest
The authors declare that they have no conflicts of interest.

Acknowledgments
This work was supported by the Japan Society for the Promotion of Science KAKENHI (grant no. 21H02002).