Rare Earth Metal‐Containing Ionic Liquid Catalysts for Synthesis of Epoxide/Cyclic Anhydride Copolymers

The perfectly alternating ring‐opening copolymerization of epoxides and cyclic anhydrides has been an emerging route to diverse polyesters. Simple catalysts have been targeted for this polymerization to decrease the cost and air‐sensitivity of the catalysis. This report improves upon the recently reported YCl3 ⋅ 6H2O/[PPN]Cl (bis(triphenylphosphine)iminium chloride) binary catalyst system, which shows fast rates of polymerization but suffers from low molecular weights due to the presence of many water chain transfer agents. In this study, use of a phosphonium chloride ionic liquid as both the cocatalyst and solubilizing agent for the metal salt to form a metal‐containing ionic liquid (MIL) leads to polymerization of high molecular weight polyesters at significantly faster rates than the original system. The identity of the ionic liquid, dryness of MIL, and synthetic prep of the MIL were all found to have a greater impact on the polymerization rate than the metal choice.


Introduction
Development of inexpensive and efficient methods for the synthesis of diverse degradable polymers is a critical challenge to commercialize degradable polymer alternatives to combat growing plastic pollution. The perfectly alternating ring-opening copolymerization (ROCOP) of epoxides and cyclic anhydrides has been a promising method for the production of a wide array of aliphatic polyesters with a range of thermal properties ( Figure 1). [1] However, future directions need to consider cost and synthesis of catalysts while working to enhance polymerization activity and control (i. e. dispersity and selectivity). [1,2] Recently, there has been an acceleration of studies with simple, inexpensive metal salts as active and controlled catalysts for the target polymerization.
In particular, metal carboxylate salts have been identified as active catalysts for the targeted polymerization without the need of a cocatalyst. Of these salts, cesium pivalate has been shown to be an active catalyst and has been used to make a wide range of polyesters through the target ROCOP (Figure 2A). [3][4][5][6][7][8] While Cs-pivalate has been shown to exhibit the highest activity of a series of alkali metal carboxylates tested for copolymerization of various monomer pairs, these polymerizations were met with some challenges such as long reaction times, the need for an initiator along with the catalyst, and airfree conditions. Additionally, extremely high monomer loadings are needed to acquire high molecular weight polymers (> 30 kDa). We have previously identified yttrium based catalysts are active for ROCOP of epoxides and cyclic anhydrides, in which the environment around the metal center can be tuned to impact the rate and control of polymerization. [9,10] Simple yttrium-based chloride salts, such as YCl 3 (THF) 3.5 and YCl 3 · 6H 2 O, along with a cocatalyst, were highly active for the perfectly alternating copolymerization of ten epoxide/cyclic anhydride monomer pairs, while maintaining good molecular weight control and suppressed side reactions ( Figure 2B). [10] While both yttrium catalysts displayed record turnover frequencies (TOFs) for various monomer pairs, there were some challenges for each catalyst. The YCl 3 · 6H 2 O catalyst displayed high conversions, however, the molecular weight of the samples fell below the expected theoretical molecular weight, likely from the water acting as a chain transfer agent. While YCl 3 (THF) 3.5 was synthesized in an inert atmosphere, preventing water content in the reaction, the rates of polymerization for most monomer pairs suffered in comparison to the hydrate catalyst. Additionally, the synthesis of the air-free YCl 3 (THF) 3.5 adduct is lowyielding, as the starting YCl 3 salt is only slightly soluble in THF. Alongside these challenges, both catalysts required the use of an expensive cocatalyst, bis(triphenylphosphoranylidene)ammonium chloride ([PPN]Cl), for the optimized conditions for the targeted polymerization. Despite these challenges, these results show promise for design of future catalysts with rare earth metals (which includes scandium, yttrium and the lanthanide series).
To improve the use of these simple yttrium salts (and other lanthanide salts) as catalysts for the target ROCOP, it is important to identify easy synthetic methods for highly soluble metal salt alternatives that could maintain fast and controlled rates of polymerization, while avoiding coordination of water to prevent lowering the polymer molecular weights. The introduction of an ionic liquid (IL) into the catalyst system could be expected to minimize the presence of water in the catalyst owing to the fact that the coordination sphere of the metal would be saturated with the counter anions from the ionic liquid leading to a visual loss of water, in some cases. [12] Sesto and coworkers demonstrated that mixing an ionic liquid and gadolinium hexahydrate salt through a solvent-assisted method led to a metal-containing ionic liquid (MIL) with minimal water present. When used with rare earth metals, ionic liquids have been employed in various chemical transformations (e. g. Friedel-Crafts alkylation, ring-opening of cyclic esters [13,14] ). However, the ionic liquid in these reports served as a method for catalyst recovery and as a solvent. Some studies did find that the use of ionic liquid also improved some of the chemical transformations in comparison to when no ionic liquid was present. [14,15] These prior studies demonstrate the catalytic value of ionic liquids for rare earth metal catalysis, and support the hypothesis that MILs can limit water access around the metal ion.
If designed carefully, the IL that is paired with the rare earth metal salt could serve three roles; a solubilizing agent for the metal salt, a coordinating anion to prevent water coordination and a cocatalyst. Herein, we report the use of a phosphoniumbased ionic liquid, mixed with YCl 3 · 6H 2 O, to synthesize MIL catalysts that can maintain or increase high polymerization activities and improved ability to achieve high molecular weight polymers than the previous YCl 3 · 6H 2 O/[PPN]Cl cocatalyst system ( Figure 2C). MIL synthetic conditions, drying and storage all made a difference on the rate and molecular weight control of ROCOP. To our knowledge, this is the first example of a rare earth MIL being used as a cooperative catalyst, in which both ionic liquid and metal salt are important for the reactivity, for any polymerization.

Results and Discussion
First, it was important to identify whether ionic liquids could form MILs with YCl 3 · 6H 2 O, and if these MILs were then active for ROCOP of epoxides and cyclic anhydrides. As previously identified, YCl 3 (THF) 3.5 shows very slow activity and poor control for the target ROCOP without the presence of a cocatalyst, so it is important to identify if the ionic liquid can serve as a cocatalyst. At a catalytic scale, YCl 3 · 6H 2 O was mixed with varying ILs at different ratios (1-6 equivalent IL to metal salt) to identify what IL could convert the metal salt to a low melting point MIL. If the mixture of YCl 3 · 6H 2 O and IL showed successful formation of MIL at 60°C, it was then used to copolymerize six different monomer pairs. ROCOP of epoxides and cyclic anhydrides offers the advantage of having a wide monomer scope than the ringopening polymerization of cyclic esters. However, catalysts in the literature are not often universally active or controlled for the ROCOP of all the monomers commercially available. Therefore, initial screening needs a variety of monomers to identify the versatility of the catalyst system. For epoxides, 1-butene oxide (BO) was used as a common monosubstituted epoxide with a higher boiling point than the more frequently used propylene oxide, while cyclohexene oxide (CHO) is used as the most common disubstituted epoxide ( Figure 3). Glutaric anhydride (GA), phthalic anhydride (PA), and carbic anhydride (CPMA) were selected as representative monocyclic, bicyclic and tricyclic anhydrides, respectively.
The initial screening identified phosphonium-based ILs (trihexyltetradecylphosphonium chloride ([H3DP]Cl), tributyltetradecylphosphonium chloride ([B3DP]Cl)) to be the most promising candidates for the targeted copolymerization (see SI for further details). YCl 3 · 6H 2 O was mixed with one equivalent at a time of each phosphonium-based IL at 95°C until the mixture reached a homogeneous liquid. In the case of [B3DP]Cl, the IL is a solid at room temperature and the mixture at 95°C proved to be extremely viscous and the reaction required five equivalents of IL to make a homogeneous mixture. With [H3DP]Cl, which is a viscous liquid at room temperature, the reaction only required four equivalents to reach a homogeneous liquid. Since [H3DP]Cl offers the advantage of already being a liquid and added atom economy through needing less equivalents to reach a homogenous liquid, MILs with [H3DP]Cl were further pursued for this study.
Since the screening reactions required very small weights of metal salt and IL, it was important to scale up the synthesis of the MIL catalyst for more consistent comparisons of reactions. First, four rare earth metal salts (LnCl 3 · 6H 2 O, Ln = Gd, Nd, Ho, and Y) were mixed with four equivalents of [H3DP]Cl neat at 95°C and stirred until the formation of a homogeneous melt that was liquid at room temperature ( Figure 4). The different rare earth metals allow for the tunability in the size and Lewis acidity of the metal in the MILs, potentially impacting the catalyst reactivity during polymerization. Additionally, three paramagnetic lanthanides were selected due to the possible catalyst recovery through use of a magnet, as has been seen for iron-containing MILs. [12] Since no efforts were made to drive off water, these MILs were defined as Ln-WET for clarity.
The polymerization reactions were done at a scale of  Figure 5). The reactivity of each MIL for all six monomer pairs was compared using single point studies, which were converted to turnover frequency with respect to the rare earth metal ion (h À 1 , TOF). While single point studies are not a perfect comparison of reaction rates, the conditions for each monomer pair were kept constant to allow for the best comparison between metal ions. Interestingly, the MILs with different metals showed similar TOFs for the same monomer pair, in most cases all four metals studied were within error of each other. We hypothesize that the anions from the IL saturate the coordination sphere for all the rare earth metals used, limiting the impact of Lewis acidity or metal ion size on polymerization activity. For all MILs, the rates of polymerization were fastest for monomer pairs with the PA anhydride, with polymerization rates for monomer pairs with GA and CPMA being much slower. These trends were similar to those observed in the previously reported YCl 3 · 6H 2 O/ [PPN]Cl and YCl 3 (THF) 3.5 /[PPN]Cl systems showing these MILs as promising catalysts for the targeted polymerization. [4] Owing to yttrium's similar reactivity to the other metals, its lower cost, and diamagnetic properties, yttrium MILs were pursued for the remaining studies presented herein.
Initial characterization of the molecular weight of the polymers produced for Figure 5 revealed smaller molecular weights than would be expected for seven chloride initiators (Table S11). Through Attenuated Total Reflectance-Fourier Transform Infrared Spectroscopy (ATR-FTIR) studies of the yttrium MIL, water was revealed to be present in Y-WET. Thermal gravimetric analysis (TGA) was then used to characterize a rough estimate of how much water was remaining in the MIL (see SI for full details of these studies). By ramping up the temperature to 100°C and maintaining this temperature for 5 hours revealed a drop in mass in the sample that correlates to approximately six equivalents of water for every yttrium center. This suggested that this water was likely acting as a chain transfer agent, leading to more initiators than expected and therefore, lower molecular weight than theoretically calculated. The resultant small polymer molecular weights suggested water content must be minimized in the catalytic system to achieve better molecular weight control ( Figure 6).
The first aim to minimize water in the MILs was to dry Y-WET by heating the MIL at 50°C for 36 hours under dynamic vacuum (10 À 3 Torr). Upon drying, this MIL, labelled as Y-DRY, was transferred to and stored in a glovebox. When quickly removed from the glovebox, Y-DRY showed no observable water present by ATR-FTIR. Using the TGA method described above, only 2.5 equivalents of water were identified for Y-DRY, suggesting success in removing water from Y-WET. Alternatively, a prior report combined GdCl 3 · 6H 2 O with three equiv-   alents of [H3DP]Cl through a DCM/water solvent assisted synthesis method, which led to a MIL with no observable water by FTIR. [12] It was rationalized that while the neat method is more atom economical, the solvent-assisted method may ensure a more homogeneous mixture, while making it easier to remove water. This solvent-assisted method was used for YCl 3 · 6H 2 O, which was mixed with four equivalents of [H3DP]Cl at room temperature in DCM and water, with the resulting MIL being labelled Y-DCMWET (as it was only dried for a couple hours on the Schlenk line to evaporate any remaining solvent). While water was also not observed in the ATR-FTIR spectrum of Y-DCMWET, the TGA method described above identified 2.5 equivalents of water per yttrium center. This suggested that the solventassisted method removed as much water as the neat synthesis after days of drying Y-WET on the Schlenk line. Drying Y-DCMWET on the Schlenk line, analogous to that for Y-WET, only led to minor water removal for what is labelled as Y-DCMDRY, with TGA studies finding just 2 equivalents of water for each yttrium ion. Synthesis of a second batch of Y-WET and Y-DCMWET revealed lower water content than the first synthesis (2.8 and 1.4 equivalents of water, respectively), which indicates variability batch to batch. Therefore, single batches were used to compare reactivity, discussed below. The first question to address was how the synthesis (neat or solventassisted), as well as the dryness (synthesized as is or dried further), impact the rate of ROCOP for the targeted monomer pairs.
The "wet" and "dry" MILs were used for the targeted polymerization with the same six monomer pairs used above. All polymerizations were run at 110°C under neat conditions for 10-30 minutes, where the epoxide served as the solvent. The polymerizations carried out using the "wet" catalyst used monomers that were dried and kept under atmospheric conditions, while those that used "dry" catalysts used monomers that were dried, purified, and stored under a nitrogen glovebox atmosphere. The monomer conversion for the polymerizations was quantified by taking an aliquot from the reaction and analyzed via 1 H NMR spectroscopy. The polymerizations were done in triplicate unless otherwise stated with the turnover frequencies seen in Figure 7.
When comparing the TOFs, the "wet" catalysts resulted in higher reactivity than their representative "dry" catalysts across all monomer pairs where the highest TOFs were observed for monomer pairs using PA. This could be attributed to the residual water driving reaction rates, as indicated in prior work comparing YCl 3 · 6H 2 O and YCl 3 (THF) 3.5 . These rapid turnover frequencies for both Y-WET and Y-DCMWET are all faster than the YCl 3 · 6H 2 O/[PPN]Cl system, now representing the fastest reported TOFs for BO/PA, BO/GA and CHO/GA monomer pairs. Interestingly, the catalysts made through the solvent assisted methods (Y-DCMWET and Y-DCMDRY) showed similar reactivity to the comparative neat MILs when considering reactions conducted in butylene oxide (BO). However, reactions conducted in cyclohexene oxide (CHO) showed a dramatic difference in reactivity between the neat catalysts and the solvent assisted catalysts, with the solvent assisted MILs showing slower rates of polymerization. These trends do not track with water content found from TGA studies, suggesting other reasons for this reactivity difference. A plausible hypothesis is that the coordination environment of Y is not the same between the neat and solvent-assisted method, which could alter how well the epoxide can be activated for ring-opening. As shown in previous literature, the speciation of rare earth metal-containing ILs can be dependent on stoichiometry or reagents and reaction conditions. [16,17] All MILs studied showed suppression of epoxide homopolymerization (< 6 % for all conditions, see SI for details). Contrary to the YCl 3 · 6H 2 O/[PPN]Cl system, epimerization of CPMA-based polymers was observed prior to reaction completion.
To further understand the impact of the rare earth metal in the MILs, control experiments were conducted where the YCl 3 · 6H 2 O and 4 equivalents of [H3DP]Cl were weighed and added separately into the reaction vial as cocatalysts rather than together as an MIL (Figure 7). The series of these controls was found to have TOFs comparable to the "wet" catalysts where in the case of CHO/PA, the control displayed even faster rates than Y-WET. These results suggest that the MIL does show similar reactivity to cases where the metal salt and phosphonium salt are used as cocatalysts. An additional control study was performed where only 4 equivalents of [H3DP]Cl was used as the catalyst for a set of polymerizations ( Figure S13). The TOFs for those reactions only using [H3DP]Cl as the catalyst showed slower rates than the control of adding the metal salt and IL separately, thus demonstrating that the presence of the rare earth metal does aid the rate of polymerization.
Since reactions with the wet catalysts used dried monomers stored outside the glovebox, while reactions with dry catalysts used dried monomers stored inside the glovebox, it was important to identify if the catalyst, reaction conditions or monomer stock were responsible for the dramatic differences in rate between the wet and dry catalysts. Initial studies were performed with Y-DRY and Y-DCMDRY for all six monomer pairs, using the monomers that were dried and stored outside the glovebox ( Figure S13). In all cases, these reactions showed faster TOFs than the analogous dry conditions, some of which reached comparable rates to those of the "wet" catalysts with the same monomers. These results identified that monomer and reaction conditions did impact the rate of polymerization.
Since the BO/PA monomer pair showed the biggest difference in reactivity between catalysts, this pair was used for more in-depth studies (Figure 8). When using monomers dried and stored outside the glovebox, both Y-DRY and Y-DCMDRY showed comparable rates to reactions previously conducted with Y-WET and Y-DCMWET, respectively. In the case where polymerizations were prepared in the glovebox with "dry" catalysts and monomers stored in the glovebox, a decrease in reactivity for the "dry" catalysts is observed in comparison to the reactions being done with wet monomers with the "dry" catalysts. The hypothesis is that presence of water in the monomers could be driving the rate to compare more with the wet catalysts than the dry catalysts. Therefore, polymerizations were run with "dry" catalysts using monomers dried and stored in the glovebox with the reaction mixture exposed to air for 5 minutes before the polymerizations were run. In these conditions, a slightly faster rate was observed for Y-DCMDRY, while Y-DRY showed comparable results to the reaction with no air exposure, indicating the atmospheric conditions were not greatly impacting the catalyst activity. An additional set of reactions performed was the use of monomers stored and dried in the glovebox with the "wet" catalysts as well. Results obtained from these control experiments showed that the dryness of the monomers led to slower rates for the polymerizations. For instance, both Y-WET and Y-DCMWET showed a decrease in rate when reacted with the dry monomers from the glovebox. This demonstrates that the water found in the monomers is contributing to the increase in rates for these reactions. The aforementioned observations further support that the presence of water contributes to the higher conversions observed, which is hypothesized to occur through hydrogen bonding throughout the rate determining step of epoxide ring-opening.
Since water has been shown to act as a chain transfer agent, it is important to improve the catalyst systems that have limited water exposure. Although the "dry" MIL catalysts showed slower polymerization than their "wet" counterparts, these MILs still showed significant improvements in the turnover frequency for yttrium simple salts. For example, the turnover frequencies for all 6 monomer pairs for Y-DRY and Y-DCMDRY were faster or comparable to those of the YCl 3 · 6H 2 O/[PPN]Cl system. The difference in rate is likely due to the added equivalents of [H3DP]Cl (4 equiv) vs. 1 equiv. [PPN]Cl or due to the charge attraction between the cations and anion of the MIL, which keeps the catalyst more closely associated than with the parent system. These much-improved rates with less presence of water could lead to rapid synthesis of high molecular weight polymers, as originally hypothesized.
To gain further insight into the impact of the dry-ness of the catalysts on the resulting polymer, the polymer samples were analyzed through gel permeation chromatography (GPC). The BO/CPMA monomer pair was selected as the ideal candidate to study any trends observed in the data as high molecular weight polymers for the BO/CPMA monomer pair were synthesized with the previously reported yttrium salts. [4] In most cases, reactions run with monomers stored outside the glovebox showed low molecular weights with the standard conditions of 1 : 200 : 1000 ratio of [MIL] : [anhydride] : [epoxide]. In many cases, the molecular weight was between what would be expected for seven chloride initiators with and without the added water determined by TGA studies acting as initiators as well (Table S12). This suggested that not all initiators present in solution are initiating polymer chains. Interestingly, several reactions with "dry" catalysts showed much higher molecular weights than expected. Since these MILs had the least amount of water, it could be possible that aggregation of the metal ions creates bridging chloride interactions that are inactive for polymerization.
To lower the impact of water in the monomers, reactions were scaled up and used dried monomers stored in the glovebox ( Table 1). The polymer obtained with Y-WET, the catalyst with the most water present, was close to what is expected for 7 chloride initiators, achieving a 13.4 kDa polymer in just 75 minutes ( Table 1, entry 1). In comparison, when the monomers stored outside the glovebox are used, Y-WET only formed polymers with a molecular weight of 4.2 kDa (Table 1, entry 2). Alternatively, Y-DCMWET, which showed much less water in the MIL than Y-WET, was able to achieve a high molecular weight polymer of 27.5 kDa with a reasonably controlled dispersity of 1.31 within the same 75 minutes (Table 1, entry 3a). A duplicate reaction produced a polymer with even higher molecular weight polymer of 34.3 kDa with slightly higher dispersity (1.55) ( Table 1, entry 3b). Analogous reactions with Y-DRY and Y-DCMDRY, conducted under a nitrogen atmosphere, showed similar high molecular weight  (Table 1, entries 4 and 5). These results suggest that not all chlorides are initiating polymerization for MILs that contain less water. As described above, we hypothesize that lack of water could lead to bridging chloride interactions, which could alter how easily the anions can initiate polymerization. Since the MIL is a dynamic liquid catalyst, the metal anion could aggregate differently depending on conditions, explaining the variability in molecular weight and dispersity between duplicate reactions. Since the dispersity remains reasonably low for most cases, it is expected that these chlorides are not being initiated slowly over time.
When reactions with Y-DRY and Y-DCMDRY are exposed to air at this larger scale, even higher molecular weight polymers are obtained in 72 or 60 minutes, respectively (Table 1, entries 6 and 7). In the case of Y-DRY, duplicate reactions lead to polymers of approximately 80 kDa, with moderate dispersities near 1.5. At just over 50 % conversion, these polymers are more than ten times larger than expected for the initiators present in Y-DRY. Shockingly, the Y-DCMDRY MIL with the least amount of water showed an even higher molecular weight polymer of 185 kDa in just one hour with a dispersity of 1.43. It is clear that the exposure to air has a significant impact on the initiation of the polymer chains, even if it doesn't impact the rate of polymerization.
In order to compare the overall impact of the MIL, in comparison to the [H3DP]Cl IL itself, control reactions were done at the same scale with four equivalents of [H3DP]Cl. With monomers stored outside the glovebox, the resulting polymer is much lower in molecular weight than that expected for 4 initiators, as would be expected with water present (Table 1, entry 8). However, when monomers are used from inside the glovebox, a 76 kDa polymer is obtained within 70 minutes (Table 1, entry 9). These results are similar to that of Y-DRY when exposed to air (Table 1, entry 6), however the higher dispersity of 1.74 reveals the value of the metal ion in the MIL.
It is surprising that the IL by itself also produced a polymer with a higher molecular weight than expected. It is unclear if this is due to anions being inaccessible for initiation due to the IL network, similar to that hypothesized for the MILs.
Although these results suggest that it is difficult to predict the polymer molecular weight, based on the expected initiators in the MIL catalyst, they suggest that higher molecular weight polymers can be achieved with less monomer excess and in shorter times than prior literature. Described as polymer growth rate in Table 1, optimized catalyst systems show a range of 11-186 kDa/h. The MILs demonstrate a drastic improvement in obtaining high molecular weight polymers, in comparison to the (YCl 3 · 6H 2 O or YCl 3 (THF) 3.5 )/[PPN]Cl catalyst systems, where 30 + kDa polymers could only be achieved after 8 hours under the same reaction conditions (temperature, monomer ratio and catalyst loading), which is approximately 5 kDa/h. With this significant increase in reaction rate and polymer growth rate, these catalysts maintain polyester selectivity with no evidence of epoxide homopolymerization. Even for other monomer pairs, catalysts in the literature are not able to achieve high molecular weight polymers without extreme monomer excess and their polymer growth rates are commonly < 10 kDa/h. [18][19][20][21][22]

Conclusion
Combination of phosphonium ionic liquids and yttrium trichloride salts are reported to form metal-ionic liquids (MIL) that are active and controlled for the ring-opening polymerization of epoxides and cyclic anhydrides. Synthetic methods were found to impact the amount of water in the resulting MIL, with solvent-assisted methods more efficiently eliminating water than mixing the two components via neat conditions. Analogous to prior studies, dryness of both the catalyst and monomers proved to impact the rate and growth of polymer molecular weight of these polymerizations in conflicting ways. Presence of water increases the rate of propagation, while also acting as a chain transfer agent, lowering the polymer molecular weight from expected. Nevertheless, this study represents the first example of a rare earth metal MIL conducting ring-opening polymerization. Additionally, this system can achieve high molecular weight polymers much faster than the original (YCl 3 · 6H 2 O or YCl 3 (THF) 3 and reagents were purchased from commercial sources (io-lo-tec, Aldrich, TCI, Alfa Aesar, Acros, Fisher, and VWR) and used without further purification.
Solvent-assisted MIL. Prepared according to literature procedure. [12] In a 250 mL round bottom flask, 3.48 g (11.48 mmol) YCl 3 · 6H 2 O and 23.85 g (45.92 mmol) [H3DP]Cl was stirred with~100 mL dichloromethane and~50 mL DI H 2 O. The mixture was stirred overnight (1 4 hours) at room temperature to form two layers. The organic MIL layer was collected and dried in vacuo for approximately 2 hours to evaporate any remaining dichloromethane. Epoxide was added to the vial that was then capped, taped closed with electrical tape, and heated in a heating block, pre-heated to 110°C, for 10-30 minutes. The samples were then quenched with approximately 1 mL chloroform followed by the removal of an aliquot for conversion analysis by 1 H NMR spectroscopy. The remaining chloroform solution was charged with~10 mL of pentane to form a polymer precipitate. Lower conversion polymer samples were further purified with a Et 2 O wash. The resulting polymer was collected via decantation and dried in a vacuum oven at 65°C for BO polymers and 110°C for CHO polymers for 14 hours.
Final polymers were characterized by 1 H NMR spectroscopy and SEC-MALS. In the case where polymer was not obtained due to low polymer yield or impurities were found in the isolated polymer, the % ester was calculated from the in situ aliquot taken for 1 H NMR and is denoted (*) in Tables S1-S11.