Expanding the Scope of Boron-Based Ate Complexes by Manipulating Their Reactivity: The Case of Cyclic Esters and Their (Co)Polymers

The value and merit of triethylborane (TEB)-based ate complexes for the synthesis of various oxygenated polymers have been recently illustrated through successful examples of homopolymerization of epoxides and of their copolymerization with CO2, CS2, COS, anhydrides, isocyanates, etc. To further expand the scope of TEB-based initiating systems to a broader family of oxygenated polymers, they were used in this study to anionically polymerize cyclic esters and to copolymerize the latter monomers with propylene oxide (PO). To promote a fast and controlled ring-opening polymerization (ROP) of cyclic esters, hydrogen-bonding donors such as amines and (thio)ureas were added to TEB-based ate complexes used as initiators. Only under these conditions could the ROP of ε-caprolactone (CL), δ-valerolactone (VL), and Llactide (LLA) proceed under controlled conditions with hardly detectable intraor intermolecular transesterifications. The role of amines and (thio)ureas when used alone or together in association with these boron-based initiators is discussed in detail. The controlled character of the polymerization of CL, VL, and LLA is attested by matrix-assisted laser desorption ionization time-offlight mass spectrometry (MALDI-ToF MS) and other characterization techniques. As examples of the far-reaching potential of these borane-based ate complexes, random P(PO-co-VL) copolymers were derived by statistical copolymerization of VL with PO and various well-defined block copolymersPPO-b-PVL, PVL-b-PPO, PPO-b-PCL, PPO-b-PLLA, and PPO-b-PVL-b-PPOwere grown by sequential polymerization of the corresponding monomers.


■ INTRODUCTION
Over the past two decades, organocatalysis has emerged as a powerful alternative to metal-based catalysis for the ringopening polymerization (ROP) of cyclic esters, 1−5 paving the way for the synthesis of metal-free aliphatic polyesters aimed at biomedical and microelectronic applications. However, a common feature of most of the organocatalysts described so far is their lack of versatility: for instance, organocatalysts that would work for CL would not be necessarily appropriate for cyclic ethers or even lactides. On the other hand, boron-based ate complexes, which also fall into the category of organocatalysts, have shown exceptional versatility by enabling the synthesis of miscellaneous oxygenated polymers, including polyethers, polyesters, polycarbonates, polythiocarbonates, polyurethanes, etc. 6−15 One of the alkyl boron compounds used in these ate complexes, triethyl boron (TEB), is a mild Lewis acid that is oxyphilic and of nonmetallic character. When utilized in combination with an oxyanion or a Lewis base, TEB-based Lewis pairs demonstrate reversible interaction and an extraordinary degree of tunability with respect to their reactivity. Unlike classical Lewis pairs that form stable Lewis adducts 16 or frustrated Lewis pairs that have not found yet an opportunity for application in polymer chemistry, 17 such TEBbased Lewis pairs were indeed shown to bring about the "living" (co)polymerizations of the following (co)monomers: epoxides/CO 2 , 6,7 epoxides/COS, 8 epoxides/anhydrides, 9−11 and epoxides/isocyanates 12 (Scheme 1). To the best of our knowledge, there is no study and report on the living/ controlled ROP of lactones and lactides in the presence of TEB-based ate complexes. One team, however, described the preparation in one step of multiblock copolymers involving epoxides, anhydrides, and L-lactide (LLA) in the presence of a TEB/DBU system but did not elaborate on the living character of LLA polymerization or on the scope of their initiating system for other cyclic esters. 18,19 On the other hand, Zhao, Ling, and their co-workers combined TEB with growing oxyanions to generate (polyether-b-polyester) n multiblock copolymers by using alternatively an excess of TEB and an excess of oxyanion to alternatively grow their polyether and polyester blocks. 20 However, the aim of these authors and their study was to design an initiating system appropriate for the synthesis of polyether-b-polyester block copolymers but not necessarily applicable to the preparation of other types of oxygenated polymers as those mentioned above.
In an effort to advance an easy access to a broader range of oxygenated polymers by TEB-based ate complexes, 7,12−14,21−23 in this work, we report and focus on the anionic ring-opening polymerization (AROP) of three cyclic estersCL, VL, and LLAby a TEB/onium salts system. To suppress transesterification reactions occurring during AROP, various strategies have been attempted including the association of alkoxides with bulky and complex ligands, 24,25 with CO 2 in dormant/active exchange, 26 or the selection of weak bases such as carbenes 27 or t-BuP 2 . 28 Another approach promoted by Waymouth and co-workers was to generate active species with a bifunctional role of a monomer and a chain-end activator to favor propagation over scrambling reactions. 29−31 To this end, they utilized (thio)ureas that get deprotonated in the presence of alkoxides and could thus play the above-mentioned bifunctional role to bring about a living polymerization of cyclic esters. 29,30 In our investigation, we found that in the presence of TEB-based ate complexes used as initiators, the polymerization of lactones is very sluggish. We then resorted to various amines and (thio)ureas used alone or together with TEB-based initiating species and eventually achieved the control of the AROP of the above-mentioned cyclic esters. We found that the role of amines and (thio)ureas used as additives is not the same: amines preferentially activate the monomer whereas (thio)ureas give rise to distinct TEB-based active species in the presence of TEB-based initiators. In both cases, the ROP of the tested cyclic esters was only moderately boosted; however, when used together, a synergistic effect was observed, illustrated by their fast and yet controlled polymer-ization of all three cyclic esters. In the case of LLA, the presence of amines was even found unnecessary. As shown in Scheme 2, the three cyclic esters, CL, VL, and LLA, were polymerized in the presence of phosphazenium t-Bu-P 4 H + benzoxide/TEB serving as TEB-based initiators with or without the addition of amines and (thio)ureas used alone or synergistically. As a demonstration of the versatility of these TEB-based systems, PO and VL were randomly copolymerized; PO, CL, VL, and LLA were also sequentially copolymerized affording an array of statistical, diblock, and triblock copolymers.

■ RESULTS AND DISCUSSION
Controlled AROP of Cyclic Esters by the TEB/Amine Strategy. When investigating the polymerization of cyclic esters carried out in the presence of the TEB/onium salts system, we chose to first study the case of CL. As expected, the AROP of CL proceeded very fast when initiated by [t-BuP 4 H + ][BnO − ] that was obtained by the deprotonation of BnOH by a superbase t-BuP 4 . The produced polyester exhibited, as expected, a broad polydispersity due to the occurrence of transesterification reactions (entry 1, Table S1). Upon adding one equivalent of TEB, an ate complex made from the above-mentioned alkoxide initiator [t-BuP 4 H + ]-[BnO − ] and TEB was formed, which totally quenched polymer growth: no monomer conversion was detected even after 24 h (entry 2, Table S1) and only a modest conversion of 1% was observed after 48 h (entry 3, Table S1). In contrast to the case of epoxides, 13−15 adding TEB in excess to the alkoxide did not boost the polymerization neither at room temperature nor at 50°C (entries 4 and 5,  Figure S1). On the other hand, the 13 C NMR spectra of CL/amines clearly show a shift of 0.16 ppm, indicative of CL activation by amines ( Figure S2). Such a result clearly indicates that the reactivity of CL toward benzoxide borate anions (entry 7, Table S1) can be enhanced by amine addition. An increase of the polymerization temperature helped to further enhance the rate of polymerization (similar conversion in 96 h), which was still rather sluggish (entry 8, Table S1); in any case, the polymers obtained exhibited expected molar masses and narrow polydispersities.
Besides BuA, other amines including isobutylamine (IBuA), isopropylamine (IPA), cyclohexanamine (CyHA), 1,3-xylenediamine (XyDA), ethylenediamine (EtDA), diethylamine (DEA), diisobutylamine (DIBA), diisopropylamine (DIPA), and triethylamine (TEA) were tried in association with [t-BuP 4 H + ][BnO − ]/TEB and their impact on the rate of polymerization was evaluated (entries 10−19, Table S1). Based on the obtained results, the various amines tried can be ranked in the following order for their ability to activate CL through hydrogen bonding and thus for their influence on the rate of polymerization: normal primary amines > isopropylamine and isobutylamine > secondary amines. XyDA demonstrated the highest impact on the rate of polymerization and on the control of CL ROP: 89% conversion of CL could be reached at room temperature in 120 h, and the polymers generated exhibited molar masses very close to the theoretical value and a narrow polydispersity (1.06).
On the other hand, amines with their lone pair of electrons are also known for their affinity toward boron compounds and for their propensity to form dative B−N bonds in the presence of the latter Lewis acids. 21 After mixing TEB with an equal equivalent of amines, the initial 11 B NMR chemical shift in ppmcould not be seen, indicating that amines stayed put and did not disrupt the above-mentioned ate complex and rather activated CL (Figures S1 and S2). Similar profiles were observed in toluene-d 8 although chemical shifts were different from those obtained in THF-d 8 ( Figure S4). The addition of amines indeed helped to enhance the rate of polymerization of CL; however, the latter was deemed still rather slow as compared to the rates of polymerization observed with other systems. We then turned our attention to other potential additives, namely, (thio)ureas, which were recently shown to promote a fast and controlled polymerization of cyclic esters in the presence of alkoxides. 29 Table 1), which is dramatically different from the 89% conversion seen in 120 h without UPh alone (entry 19, Table S1). As shown in the 1 H NMR spectrum ( Figure S6), benzoxide is the actual initiator and neither UPh nor amine is incorporated into the PCL chains that were generated. In the presence of these two additives, the polymerization is not only fast but it also exhibits a controlled character; upon increasing the ratio of TEB to benzoxide from 1 to 2, a PCL sample with a narrower polydispersity could be obtained and a very minor transesterification reaction could be detected (entry 6, Table 1).  With the addition of 3 equiv of UPh, the polymerization proceeded faster and yet exhibited a controlled character (entry 10,  Figure S7). In summary, the rate of polymerization of CL could be varied at will upon varying the amount of TEB, amine, and (thio)urea; in the presence of TEB, transesterification reactions were thus almost suppressed; the role of amine was thus to activate the monomer (as confirmed by 13 C NMR) although some of the amine was lost in the formation of the TEB−BuA adduct ( 11 B δ = −3.9 ppm. vide infra); the role of (thio)urea was to trap the borate anions and generate species that activated both the hydroxyl carried by chain ends and the monomer carbonyl function by the same mechanism as the one unveiled by Waymouth. 29,30 The ability of other (thio)ureas to play a similar role was also checked. The conversion of CL reached 4  Table 1). In the presence of TUPh (pK a 13.4 in dimethylsulfoxide, DMSO), 32 no polymerization occurred (entry 7, Table 1). Under optimized conditions, PCLs of different molar masses (up to 27.4 kg mol −1 ) and narrow polydispersities could be obtained with different feeding ratios of monomer to initiator from 100 to 500 (entries 10−13, Table 1). In all cases, the molar masses were very close to the theoretical values, and in addition, narrowly distributed molar masses were obtained with polydispersities in the range of 1.07−1.12 ( Figure 1B). The chemical structures of the obtained PCLs were analyzed by NMR and MALDI-ToF measurements ( Figures 1A and S6). Subsequently, another lactone, namely, VL, and also LLA were polymerized using TEB-based ate complexes in the presence of both an amine and a (thio)urea. In the presence of [t-BuP 4 H + ][BnO − ]/TEB/amine/urea, VL polymerized faster than CL due to its larger ring strain: 30 Table 1), and the molar mass of the obtained PVL sample (M n(SEC) = 9.3 kg mol −1 ) agreed well with the theoretical value (M n(theo) 9.4 kg mol −1 ) and a narrow distribution of molar masses could be achieved (Đ m = 1.14) attesting to the controlled character of the AROP of VL. Attempts at synthesizing higher molar masses (entries 15−16, Table 1) produced PVL samples with molar mass up to 30.8 kg mol −1 and a polydispersity index of 1.09. The chemical structures of the PVL samples obtained were also characterized by NMR and MALDI-ToF MS, which indicated an efficient polymerization by phosphazenium benzoxide (Figures S8 and  S9).
The AROP of LLA in the presence of TEB was tried last. Unlike the case of lactones such as CL and VL (entries 2−4, Table S1), more active LLA could be anionically polymerized in the presence of TEB-based ate complexes without the need for activation by an amine. The reaction was carried out at 50°C in THF, and the rate of polymerization was rather slow (21% conversion in 24 h, entry 17, Table 1). The obtained PLLA sample exhibited expected molar masses and a narrow distribution of molar masses confirming the controlled character of LLA polymerization. Upon using TUPh, monomer conversion could reach 92% in 3.5 h even at room temperature (entry 18, Table 1). The presence of the benzoxy moiety due to the initiator in the PLLA sample synthesized was confirmed by 1 H NMR spectroscopy ( Figure S10). The incorporation of BnOH into the PLLA chains formed was also confirmed by MALDI-ToF MS; the presence of one set of peaks where each peak corresponds to an even number of LA units suggested negligible transesterification if any during the AROP of LLA ( Figure 1A). In addition, the epimerization of LLA during AROP is also minor; only one peak (69−70 ppm) corresponding to stereosensitive methine carbon in comparison to isotactic-enriched poly(lactide) was detected in the 13 C NMR spectrum ( Figure S11). The above-mentioned polymerization conditions could be applied to synthesize PLLA samples of higher molar masses in a reasonable time without any difficulty (entries 18−20, Table 1). In all cases, the SEC traces were narrow and unimodal ( Figure 1B). In addition, the AROP of rac-lactide (rac-LA) afforded a isotactic-enriched poly(lactide) ( Figure S11). The homonuclear decoupled 1 H NMR spectrum indicates a probability of isotactic propagation (P i ) equal to 0.74 through a chain-end control mechanism ( Figure S12), which is consistent with an isotactic enchainment of rac-LA observed with other organocatalysts. 33,34 Random and Block Copolymerization of Epoxide and Cyclic Esters. As mentioned in Introduction, one of the aims of this study was to investigate the scope and limits of TEBbased ate complexes when used to initiate the polymerization of both lactones and lactides. We found that these systems, which were successfully used in the synthesis of a number of oxygenated polymers, require additional activation by (thio)ureas and amines for lactones and by (thio)ureas alone for lactides, inducing in both cases a fast polymerization and producing well-defined polyester samples free of epimerization. From the above investigation, it appears that the TEB-based ate complexes, by their versatility, provide an opportunity to copolymerize either randomly or sequentially more than one family of monomers (Scheme 3). To take the example of cyclic esters, there is a wealth of enzymatic, 35 organic, 36 and metalbased catalysts/initiators 37,38 that bring about a living/ controlled polymerization of these monomers free of To expand further the already broad scope of the TEB-based ate initiators, we thus attempted to use them to copolymerize randomly and also sequentially both cyclic esters and propylene oxide: this is the second aim of this study. We first tried to copolymerize VL and PO. To the best of our knowledge, only two systems were shown to bring about the statistical copolymerization of lactones/lactides and epoxides; Lynd and co-workers resorted to the classical Vandenberg catalyst to copolymerize the two families of monomers, but the samples were rather ill-defined. 39 More recently, Coates and Waymouth and their co-workers combined their respective catalystsCr-salen catalyst of Coates and DBU of Way-  Table S2), PPO-b-PVL block polymer (entry 2, Table S3), homopolymer PVL (entry 14, Table 1), and homopolymer PPO (entry 2, Table S3). (B) polymerization kinetics of the copolymerization of PO and VL (Table S2). (C) DOSY NMR spectrum of the P(PO-co-VL) copolymer (entry 4, Table S2). mouthto achieve a dual and concomitant catalysis of PO and VL and eventually obtain statistical poly(ether-co-ester) copolymers. 40 In both cases, the two systems mentioned above could not be utilized to generate other families of oxygenated polymers. For the copolymerization of PO and VL, we thus resorted to the same initiator, BnOH, that was deprotonated using t-BuP 4 and used TEB to generate the corresponding ate complex and EtDA as an amine (Table S2). Under these conditions, a smooth copolymerization of the two monomers occurred with a faster incorporation of PO compared to that of VL ( Figure  2B). In comparison to the spectra of PPO, PVL, and PPO-b-PVL, the newly appeared peaks e and f at 5.0 and 1.2 ppm correspond to methane and methyl groups of PO next to the ester of VL and peak d at 3.5 ppm corresponds to methylene protons of VL next to PO ether oxygen, respectively ( Figure  2A). 1 H− 1 H COSY and 1 H− 13 C HSQC confirm such assignation of the connection between PO and VL units (Figures S13−S15). In addition, the same diffusion coefficients from diffusion-ordered spectroscopy (DOSY) NMR characterization of these different samples further attested to the random incorporation of the two monomers and to the formation of statistical copolymers ( Figure 2C). To further demonstrate the versatility of the TEB-based ate complexes, we used the TEB-based initiation systems to synthesize block copolymers made of poly(propylene oxide) blocks and polyester blocks. For example, after obtaining the first ester PVL block through ROP of VL as described in the previous section, PO along with 3 more equiv of TEB was added for sequential ROP of PO: a PVL-b-PPO diblock copolymer with PVL as the first block was obtained with a well-defined structure (Scheme 3B, Figures S16 and S17). Likewise, starting from polyether blocks, which were successfully grown using tetrabutylammonium chloride as an initiator, we synthesized poly(propylene oxide)-b-polylactone and poly(propylene oxide)-b-polylactide block copolymers (Scheme 3C). Tetrabutylammonium chloride (TBACl) was added to TEB before initiating the polymerization of PO. After complete AROP of PO in the presence of 2 equiv of TEB with respect to TBACl, lactones and lactide monomers were then sequentially polymerized after introducing in the reaction medium either an amine or a urea under optimized conditions as previously shown (TEB/XyDA/UPh = 2/0.5/3/1). In all cases (Table  S3), narrow and unimodal SEC traces were detected, and a clear shift of traces with respect to the traces of the PPO precursor was observed, indicating the formation of the expected diblock copolymers ( Figure 3A). In addition to the characteristic peaks due to PPO and polyester blocks in 1 H NMR spectra (Figures 2A and S18−S20), the diffusionordered (DOSY) NMR spectrum shows the same diffusion coefficients for the two block segments ( Figure 3B), confirming the diblock nature of the copolymers formed, PPO-b-PVL, PPO-b-PCL, and PPO-b-PLLA. Similar to the preparation of the PVL-b-PPO diblock copolymer, sequential additions of PO and the use of more TEB afforded the PPO-b-PVL-b-PPO triblock copolymer ( Figure S21).
Discussion and the Proposed Mechanism. When trying to utilize Lewis pairs made of onium salts or oxy-anions and TEB to polymerize cyclic esters as we successfully did for the (co)polymerization of epoxides with a number of comonomers (CO 2 , etc.), we were confronted with a very slow polymerization of all three cyclic esters tried (Table S1). Even in the presence of an excess of TEB, the latter monomers could not be activated enough to trigger a reasonably fast polymerization.
Considering that primary amines when mixed with boronbased ate complexes can weaken the BnO − /TEB Lewis pair and thus activate it, we resorted to 11 B NMR to verify this hypothesis and used 11 B as a probe to understand the exact role played by these amines. Actually, we found that they did not disrupt the Lewis pair made of oxy-anion (BnO − ) and TEB (4) but they activated cyclic esters through hydrogen-bonding interaction with the monomer carbonyl, allowing a rather slow but well-controlled ROP of CL, VL, and LLA to occur (route A, Scheme 4).
On the other hand, one of the most remarkable additions to the topic of controlled ROP of cyclic esters is the investigation led by Waymouth who unveiled all of the benefits to the use of (thio)ureas. In the presence of alkoxides, the latter compounds get deprotonated and appear to activate both the hydroxylterminated chain end and the monomer. The second discovery of the current study was to demonstrate that our boron-based ate complexes and Waymouth's (thio)ureas are perfectly compatible and that their combination produces novel reactive species (5) that are distinct from the initial ate complexes (3). As evidenced by 11 B NMR, these reactive species are more nucleophilic than their parent boron-based ate complex, but unlike the latter species they are able to bring about the living/ controlled ROP of cyclic esters. The propagation proceeds at a moderate rate in the presence of these boron-based urea anions (5) or thioimidolate but is free of intramolecular/ transesterification reactions (route B, Scheme 4). The third discovery of the current investigation is that a synergistic effect is observed when alkoxide/urea/TEB systems are combined with amines: that synergy likely entails monomer activation by both amine and urea through hydrogen bonding with the monomer carbonyl (6 and 6′, TEB complexed and uncomplexed urea anions), bringing about a rather fast ROP of cyclic esters (route C, Scheme 4; entry 4 in Table 1). Yet the corresponding sample with rather broader dispersity (Đ = 1.18) also indicates that transesterification occurred after the addition of BuA due to the presence of species 6′. The latter species that are found in experiments corresponding to entries 1 and 2 of Table 1 are obviously responsible for transesterification reactions. In other words, Waymouth's species when combined with t-BuP 4 H + cations do not bring about a living/controlled polymerization of cyclic esters. Upon adding an excess of TEB (entry 6, Table 1), we, as expected, favored the formation of species 6 at the expense of species 6′ and eventually obtained a rather fast and yet controlled ROP of cyclic esters. 1 H NMR characterizations give similar results, where only urea disrupts the ate complex 3 into 5, with two groups of peaks appearing at 0.82, 0.28 ppm and 0.73, 0.15 ppm corresponding to methyl and methylene protons of TEB, revealing the presence of two different species 6 and 6′ after addition of both urea and amine into the ate complex 3 ( Figure  S22).
The other main advantage of using urea anions as the Lewis base in combination with TEB as the Lewis acid is that the Lewis pair formed can be used to polymerize not only cyclic esters but also an epoxide such as PO. Note that Waymouth's (thio)urea anions can only catalyze the ROP of cyclic esters but are ineffective for epoxides. When combined with amines, urea anions/TEB Lewis pairs demonstrate enough adaptability and adjustability of their reactivity to allow the successful (co)polymerization of both cyclic esters and epoxides.

■ CONCLUSIONS
First successfully employed to copolymerize CO 2 with various epoxides under metal-free conditions, TEB-based Lewis pairs have been progressively and successfully applied to the homopolymerization and copolymerization of epoxides with various comonomers, including anhydrides, COS, CS 2 , isocyanates, and the likes.
In an attempt to further expand the utilization of these TEBbased Lewis pairs as initiators, we demonstrate here their ability to multitask by adding to the above list of (co)polymerized monomers that of cyclic esters. Unlike other initiating systems that were mainly designed for one family of monomers, these TEB-based Lewis pairs can also serve to copolymerize either randomly or sequentially more than one type of monomer, as illustrated here by the synthesis of polyester−polyether random and block copolymers.
■ ASSOCIATED CONTENT * sı Supporting Information