Bifunctional and recyclable polyesters by chemoselective ring-opening polymerization of a δ-lactone derived from CO₂ and butadiene

Abstract

When aiming at the direct use of CO₂ for the preparation of advanced/value-added materials, the synthesis of CO₂/olefin copolymers is very appealing but challenging. The δ-lactone 3-ethylidene-6-vinyltetrahydro-2H-pyran-2-one (EVP), synthesized by telomerization of CO₂ with 1,3-butadiene, is a promising monomer. However, its chemoselective ring-opening polymerization (ROP) is hampered by unfavorable thermodynamics and the competitive polymerization of highly reactive C=C double bonds under usual conditions. Herein, we report the chemoselective ROP of EVP using a phosphazene/urea binary catalyst, affording exclusively a linear unsaturated polyester poly(EVP)_ROP, with a molar mass (M_n) up to 16.1 kg·mol⁻¹ and a narrow distribution (Ð < 1.6), which can be fully recycled back to the pristine monomer, thus establishing a monomer-polymer-monomer closed-loop life cycle. In these polyesters, the CO₂ content reaches 33 mol% (29 wt%). The reasons for the unexpected chemoselectivity were investigated by Density-functional theory (DFT) calculations. The poly(EVP)_ROP features two pendent C=C double bonds per repeating unit, which show distinct reactivity and thus can be properly engaged in sequential functionalizations towards the synthesis of bifunctional polyesters. We disclose here a methodology providing a facile access to bifunctional and recyclable polyesters from readily available feedstocks.

Introduction

The direct utilization of carbon dioxide (CO₂) to produce valuable chemicals represents a most relevant and timely objective, because CO₂ is not only an abundant and inexpensive C1-resource but also a massive contributor to the greenhouse effect^1,2. The importance of accessing high value-added products from CO₂ has been emphasized by manufacturing degradable polymers^3,4. For example, the ring-opening alternating copolymerization of CO₂ with epoxides produces degradable aliphatic polycarbonates^5,6, which are the only industrial CO₂-based polymers. Noteworthy, the large-scale industrial availability of olefins has allowed polyolefins to become the most common synthetic polymers, representing ca. half of the 300 million tons of plastics produced worldwide⁷. Achieving efficient copolymerization of CO₂ with olefins hence represents a major target in terms of sustainability and economics^8,9,10. However, little success has been achieved so far in this direction, owing to unfavorable reaction thermodynamics and kinetics under mild conditions^11,12,13,14.

Transition metal complexes can catalyze reactions of CO₂ with C-C bond formation¹⁵ and in the late 1970s, Inoue and coworkers discovered that coupling of CO₂ with 1,3-butadiene using a Pd-catalyst produces esters and lactones, in particular the six-membered unsaturated δ-lactone 3-ethylidene-6-vinyltetrahydro-2H-pyran-2-one (EVP, Fig. 1)¹⁶, of which the yield and selectivity have been significantly improved over the years^{17,18,19,20,21,22,23}. This multifunctional δ-lactone appears as a promising potential monomer to construct CO₂-based polymers²⁴, since it contains 1) a cyclic ester unit that can undergo ring-opening polymerization (ROP) to give polyesters, 2) a terminal olefin that could be engaged in coordination addition polymerization, and 3) an α,β-unsaturated ester subunit that could afford polyacrylates by radical polymerization. However, EVP has received very little attention in polymer chemistry until almost 40 years after its discovery. In 2014, Nozaki and co-workers reported a milestone work with the free-radical homopolymerization of EVP to afford polyacrylates (Fig. 1a)²⁵. Later, Lin and coworkers synthesized high-molar-mass CO₂-derived polymers by O₂-initiated radical polymerization of EVP under additive/solvent-free conditions²⁶. More recently, the Pd-catalyzed copolymerization of EVP with ethylene was reported (Fig. 1b)²⁷.

Fig. 1: Synthesis of EVP and its derivatives and polymerization reactions.

a Free-radical polymerization, b coordination copolymerization, c partial hydrogenation to form EtVP and further ring-opening polymerization, d ring-opening homopolymerization, e complete hydrogenation to form DEtP and further ring-opening polymerization, f thiol-ene click polymerization, g cationic copolymerization, h scrambling polymerization, i methanolysis/esterification, j ring-opening/conjugate polymerization.

Although the selective ROP of EVP appears most attractive in view of the degradability and possible functionalization of the resulting aliphatic polyesters, all previous attempts have failed, owing to the stability of a six-membered disubstituted δ-lactone with low ring strain. Moreover, the conjugated double bonds could undergo Michael-addition reactions that would compete against or suppress ROP^{26,27,28,29,30,31,32}. Therefore, previous reports on ROP focused on the hydrogenated derivatives of EVP, including 3-ethyl-6-vinyltetrahydro-2H-pyran-2-one (EtVP, Fig. 1c) and 3,6-diethyl-tetrahydro-2H-pyran-2-one (DEtP, Fig. 1e). For instance, Tonks and coworkers reported the ROP of EtVP or DEtP to form recyclable polyesters, utilizing 1,5,7-triazabicyclo[4.4.0]dec-5-ene (TBD) as the catalyst^9,33. At almost the same time, Lin and coworkers synthesized polyesters with high-molar-mass, up to 613.8 kg·mol^-1, by ROP of DEtP³⁴. Other strategies include the copolymerization of EVP with other monomers or the transformation of EVP to other reactive intermediates³⁵. For example, the step-growth polymerization of EVP with di- or poly-thiols was developed to synthesize linear or network copolymers containing the intact lactone ring (Fig. 1f)^32,36, whereas copolymerization of EVP with β-butyrolactone (BBL) or ε-caprolactone (ε-CL) produced copolyesters (Figs. 1g and 1h, respectively)^37,38. Ni and coworkers reported the transformation of EVP to methyl-2-ethylidene-5-hydroxyhept-6-enoate methacrylate (Fig. 1i, MEDMA), a highly reactive trivinyl intermediate, by sequential methanolysis and esterification with methacryloyl chloride^31,39. Recently, Eagan and coworkers achieved the polymerization of EVP by ring-opening and 1,4-conjugate reactions (Fig. 1j). The dimer (Di-EVP) was first formed by vinylogous 1,4-conjugate addition and then polymerized to degradable polyesters⁴⁰.

These recent achievements prompted us to examine the direct ring-opening homopolymerization of EVP, which does not require prior hydrogenation to prepare EtVP or DEtP (Fig. 1c, e). It is noteworthy that the polyesters resulting from direct and selective ROP of EVP (Fig. 1d) retain two different types of C=C double bonds (terminal and internal) with distinct reactivity. Consequently, chemoselective ring-opening homopolymerization of EVP should provide opportunities for a post-functionalization of polyesters, via hydrolysis, click reaction, and cross-linking, and thus open new avenues for the synthesis of CO₂-based polymers with tailor-made structures and properties^41,42.

Herein, we report the chemoselective ROP of EVP, producing exclusively poly(EVP)_ROP (Fig. 1d), catalyzed by a phosphazene/urea binary system. After optimization of the catalyst structure and reaction conditions, a 43% conversion of EVP was achieved and a linear poly(EVP)_ROP with a M_n as high as 16.1 kg·mol^-1 was successfully synthesized. These polyesters have a high CO₂ content (29 wt%) and can be depolymerized back to the pristine monomer. The terminal and internal alkenes of poly(EVP)_ROP can be engaged in different post-functionalization reactions, allowing the synthesis of bifunctional CO₂-based polyesters.

Results and discussion

Chemoselective ROP of EVP under different conditions

A phosphazene/urea binary catalyst, known to be effective for the selective ROP of α-methylene-γ-butyrolactone (MBL) and α-methylene-δ-valerolactone (MVL)^43,44, was used for screening the ROP of EVP. The vinyl-addition polymerization (VAP) of α,β-unsaturated ester has been proposed to proceed via the generation of a carbanion intermediate, which requires a catalyst with a much stronger basicity than that for the ROP pathway. It has been reported that in a base/urea catalytic system, a hydrogen bond donor bis-urea usually shows higher catalytic activity toward the ROP of lactones than a mono-urea under otherwise identical conditions^45,46. We initially selected a cyclic trimeric phosphazene base (CTPB) of moderate basicity (pK_a = 33.3 in MeCN) and the bis-urea DU8 (Fig. 2) to form the binary catalyst.

Fig. 2: Phosphazene/urea binary system.

Chemical structures of phosphazenes (pK_a values in MeCN^64,65) and ureas (pK_a values determined experimentally^a in DMSO^66,67 or by calculation^b) used in this study.

The ROP reactions were first conducted between 25 °C and −70 °C to investigate the effect of temperature on the chemoselectivity, with ratios EVP/CTPB/DU8/BnOH = 50/1/1.5/1 and [EVP]₀ = 5.0 M in THF (Supplementary Table 1, runs 1−5). ¹H NMR data were collected on crude mixtures to monitor the conversion of EVP, the formation of ring-opening products (oligo(EVP)_ROP for M_n < 2.0 kg·mol^-1 or poly(EVP)_ROP for M_n ≥ 2.0 kg·mol^-1) and other by-products at different reaction temperatures (Fig. 3). As anticipated, rapid vinylogous 1,4-conjugate Michael addition occurred at room temperature (25 °C), resulting in the selective formation of Di-EVP (Fig. 1j), as indicated by the typical ¹H NMR signal at 7.07 ppm of the reaction mixture (Fig. 3b) and purified product (Supplementary Fig. 1). To realize the selective ROP of EVP, subsequent polymerization reactions were conducted at lower temperatures. At 0 °C (Supplementary Table 1, run 2), both ¹H NMR resonances at 7.07 and 6.91 ppm were observed (Fig. 3c), suggesting the presence of a mixture of Di-EVP, formed by 1,4-conjugate Michael addition, and of oligo(EVP)_ROP from ROP (vide infra), respectively. If exclusive formation of oligo(EVP)_ROP with 13% conversion was achieved at −20 °C within 8 h (Supplementary Table 1, run 3; Fig. 3d), a further decrease of the temperature to −50 °C resulted in selective ROP and a 30% conversion of EVP within 8 h (Table 1, run 1; Fig. 3e). Under such conditions, the oligo(EVP)_ROP obtained had a molar mass (M_n) of 1.5 kg·mol^-1, which was lower than the theoretical value (2.4 kg·mol^-1), probably because of the inevitable occurrence of transesterification during polymerization, as evidenced by the relatively broad dispersity (Đ = 1.42; Supplementary Fig. 2a). However, only a 13% conversion of EVP was observed at −70 °C (Supplementary Table 1, run 5; Fig. 3f), which was significantly lower than that obtained at −50 °C. This phenomenon is probably due to the decreasing solubility of the CTPB/DU8 mixture at −70 °C, which limits the interactions between phosphazene and urea⁴⁷. In addition, the polymerization reaction was also carried out for 16 h at −70 °C, and the conversion of EVP was still only 13% (Supplementary Table 1, run 13), as observed after 8 h reaction time (Supplementary Table 1, run 5). As shown in Supplementary Fig. 18, the polymerization solution of EVP with a 13% conversion is frozen at −70 °C, which limits the further polymerization. These results suggested that performing the reaction below 0 °C is a prerequisite for the chemoselective ROP of EVP⁴⁸. To gain a deeper insight into this reaction, the thermodynamics for ROP of EVP were studied with ratios EVP/CTPB/DU8/BnOH = 50/1/1.5/1 and [EVP]₀ = 5.0 M in THF at different temperatures (from –10 to –50 °C), and the results are summarized in Supplementary Table 2. The equilibrium monomer concentration, [EVP]_eq, was measured to be 4.60, 4.35, 4.10, 3.80, and 3.50 M at –10 °C, –20 °C, –30 °C, –40 °C, and –50 °C, respectively. The Van’t Hoff plot of ln[EVP]_eq versus 1/T yields a straight line (R² = 0.998) with a slope of –0.40 and an intercept of 3.05 (Supplementary Fig. 3). The enthalpy change (ΔH_p) and entropy change (ΔS_p) for the ROP of EVP are −0.80 kcal·mol^-1 and −6.1 cal·mol^-1·K^-1, respectively, resulting in a ceiling temperature (T_c) of –141 °C at [EVP]₀ = 1 M and of 5 °C at [EVP]₀ = 5 M according to the Dainton’s equation (T_c = ΔH_p /(ΔS_p + Rln[EVP]₀), where R is the gas constant)⁴⁹. It is surprising that the thermodynamics for the ROP of EVP are more similar for the five-membered γ-butyrolactone (γ-BL: ΔH_p = −1.29 kcal·mol^-1, ΔS_p = −9.5 cal·mol^-1·K^-1, and T_c (1 M) = − 136 °C) and α-methylene-γ-butyrolactone (MBL: ΔH_p = −1.41 kcal·mol^-1, ΔS_p = −9.6 cal·mol^-1·K^-1, and T_c (1 M) = −126 °C)^48,50, than for the six-membered hydrogenated EtVP (ΔH_p = −2.7 kcal·mol^-1, ΔS_p = −9.2 cal·mol^-1·K^-1, and T_c (1 M) = 18.3 °C) and DEtP (ΔH_p = −3.1 kcal·mol^-1, ΔS_p = −11.7 cal·mol^-1·K^-1, and T_c (1 M) = 6 °C)^34,51. These results further confirm that ROPs of EVP need to be conducted at low temperature, similarly to γ-BL and MBL. Furthermore, we also evaluated the propensity of EVP to undergo a thermodynamically favorable ROP using a methodology recently reported by Hoye⁵², where an equilibrium for ring-opening of EVP in 2 vol% CD₃OD in CDCl₃ was monitored by ¹H NMR (Supplementary Fig. 4). The percentage of ring-opened product at equilibrium is 17%, which is similar to the case of γ-BL monomer (18.7%) and consistent with the measured thermodynamic parameters in this study.

Fig. 3: Stacked ¹H NMR spectra (400 MHz, CDCl₃).

a EVP monomer and b−f crude ROP mixtures obtained using the CTPB/DU8 binary catalyst at different temperatures (Supplementary Table 1, runs 1−5). The calculations of selectivity are based on the integrations for peaks at 6.91 ppm (poly(EVP_ROP)) and 7.07 ppm (Di-EVP).

Table 1 ROP of EVP using phosphazene/urea binary organocatalysts under different reaction conditions.^a

Not only the temperature but also the molar ratio phosphazene/urea in the binary catalytic system were found to be critical for the chemoselective ROP of EVP (Table 1, run 1 and Supplementary Table 1, runs 6−8). The ratio CTPB/DU8 influences the monomer conversion and the selectivity of the polymerization (Supplementary Fig. 5). For instance, polymerizations conducted with a ratio CTPB/DU8 = 1/0 or 1/0.7 produced mixtures of Di-EVP and ROP products, as shown by the ¹H NMR resonances at 7.07 and 6.91 ppm, and the selectivity for ROP was only 36% or 57%, respectively (Supplementary Fig. 5, blue and red; Supplementary Table 1, runs 6 and 7). To achieve the exclusive ROP of EVP (100% through ROP pathway), at least 1.5 equiv. of DU8 to CTPB was required (Table 1, run 1). However, a large excess of urea relative to base (CTPB/DU8 = 1/3; Supplementary Table 1, run 8) resulted in a decreased conversion, which is consistent with previous work⁵³, although the selectivity for the ROP of EVP remained high.

Encouraged by these results, combinations of CTPB with various ureas (Fig. 2, U4, U7−U9 with known pK_a values) were further tested as binary organocatalysts for the ROP of EVP (Table 1, runs 5, and 8−10). When U4 with a pK_a value of 22.8 in DMSO was used (Table 1, run 5), a mixture of Di-EVP and oligo(EVP)_ROP was obtained (Supplementary Fig. 6). Exclusive ROP of EVP was achieved with U7 (pK_a = 18.7 in DMSO) as a cocatalyst (Supplementary Fig. 7), and a 20% conversion of EVP was obtained within 8 h, producing oligo(EVP)_ROP with a M_n = 0.9 kg·mol^-1 (Table 1, run 8). As shown in Supplementary Fig. 8, U8 with a pK_a value of 17.8 led to a conversion and selectivity (Table 1, run 9) similar to U7. However, a further increase in the acidity of the urea led to a significant decrease in the conversion of EVP, and U9 (pK_a = 13.8 in DMSO) combined with CTPB only resulted in a 6% conversion under otherwise identical conditions (Table 1, run 10; Supplementary Fig. 9). It is assumed that the more acidic U9 formed with CTPB a relatively robust adduct that reduced the catalytic capability^6,54. In addition to CTPB, the commercially available phosphazenes t-BuP₄ (pK_a = 42.7 in MeCN) and t-BuP₂ (pK_a = 33.5 in MeCN) (Fig. 2) were tested in combination with DU8, and chemoselective ROPs of EVP (Supplementary Table 1, runs 9 and 10) were observed. We also found that the monomer concentration dramatically influences the activity in the ROP of EVP. Decreasing the concentration of EVP resulted in lower conversions (Supplementary Table 3, runs 5 and 6), whereas attempts to perform bulk polymerization failed, because the pure EVP monomer immediately became very viscous at −50 °C. The oligo(EVP)_ROP or poly(EVP)_ROP with different molar masses were successfully obtained by adjusting the molar ratio of EVP to BnOH. ROP with EVP/BnOH ratios varying from 50/1, 50/0.5 to 50/0.25 produced oligo(EVP)_ROP or poly(EVP)_ROP with increased M_ns from 1.5, 3.6 to 6.5 kg·mol^-1, respectively (Table 1, runs 1, 12 and 13; Supplementary Fig. 2). However, further decreasing the initiator loading (EVP/BnOH = 50/0.125; Table 1, run 14) indeed resulted in a decreased conversion, probably because of the lower concentration of anionic active species. The lower conversion and the more significant transesterification (Ð = 1.58) led to a low M_n of 2.4 kg·mol^-1.

The poly(EVP)_ROP and oligo(EVP)_ROP samples were carefully analyzed by NMR spectroscopy (Fig. 4 and Supplementary Fig. 10), which revealed that homopolymerization of EVP occurred by a selective ROP process. The assignments were based on ¹H–¹H COSY, ¹H–¹³C HSQC and DEPT experiments (Supplementary Figs. 11−13). In the typical ¹H NMR spectrum of the poly(EVP)_ROP obtained in Table 1 run 13 (Fig. 4a), the signals at 6.91 (H_c), 5.81 (H_h) and 5.26 (H_i) ppm are characteristic of olefinic protons, and are associated with resonances at 138.02 (C_c), 136.40 (C_h) and 116.57 (C_i) ppm in the ¹³C NMR spectrum (Fig. 4b). The MALDI-TOF spectrum of the same poly(EVP)_ROP sample is shown in Fig. 4c, which displays only one series of signals assigned to a linear chain structure with BnO/H end groups. The two neighboring molecular ion peaks are separated by 152.2 g·mol^-1, which is fully consistent with the exact molar mass of the EVP repeating unit. The NMR and MALDI-TOF MS spectra clearly confirmed the selective ROP of EVP and the production of linear poly(EVP)_ROP polyester. No phosphazene/water initiation is observed in the MALDI-TOF spectra (Supplementary Figs. 14 and 15). To the best of our knowledge^8,9,10, this is the only report of a direct and selective ROP of EVP. A CO₂-based polymer with two distinct unsaturated functional groups (terminal and internal alkenes) has thus been developed (Fig. 1d).

Fig. 4: Characterization of poly(EVP)_ROP.

a ¹H NMR (400 MHz, CDCl₃), b ¹³C NMR (100 MHz, CDCl₃), and c MALDI-TOF spectra of poly(EVP)_ROP obtained in Table 1 run 13 (M_n = 6.5 kg·mol^-1, Ð = 1.24).

Computational Investigation of the ROP Mechanism

To elucidate the origin of the chemoselective ROP of EVP, both ROP and 1,4-conjugate addition dimerization pathways were studied and compared by DFT calculations. To simplify the calculations, the urea was considered to be completely deprotonated in the model reaction, and the chain initiations involved in the ROP and 1,4-conjugate addition dimerization pathways were mediated by the complex (INT0) resulting from the interaction of benzyl alcohol with deprotonated DU8, which was well documented in previous reports^43,53. As shown in Fig. 5, nucleophilic attack of methoxy-anion chain end BnO⁻ of BnOH on the carbonyl carbon atom of EVP takes place via a transition state (TS) TS1_ROP with an energy barrier (ΔG^‡_ROP) of 11.4 kcal·mol^-1 to generate an intermediate (INT) INT1_ROP. Then, ring-opening of EVP occurs via TS2_ROP to form INT2_ROP, with an exothermal energy of 4.3 (11.1−6.8) kcal·mol^-1, in which a new methoxy-anion chain end resulting from the ring-opening of EVP is stabilized by DU8 molecule through hydrogen-bonding interactions, being similar to the bisthiourea-nitrate complex formed in the bisthiourea-catalyzed Henry reactions⁵⁵.

Fig. 5: Computational investigation of the reaction mechanism.

DFT-computed energy profiles (ΔG_sol, 223.15 K, at the SMD(THF)//B3LYP-D3(BJ)/6-311 + G(2d,p)//B3LYP/6-31 G(d,p) level of theory) for ring-opening polymerization (ROP) and 1,4-conjugate addition dimerization (dim) pathways of EVP by binary CTPB/urea (DU8).

For the 1,4-conjugate addition dimerization pathway, a hydrogen from the methyl of EVP is captured by BnO⁻ of BnOH to surmount a TS1_dim with an energy barrier of 16.2 kcal·mol^-1, which further removes a benzyl alcohol and rearranges to afford INT1_dim. This INT1_dim attacks the methine carbon atom of the second EVP molecule through TS2_dim to form a dimer INT2_dim with an energy barrier of 11.7 (20.6−8.9) kcal·mol^-1. Furthermore, the complex INT3_dim containing a dimer of EVP is generated by addition of benzyl alcohol via TS3_dim with an energy barrier of 2.6 (8.2−5.6) kcal·mol^-1. By contrast, the ROP pathway overcomes a free-energy barrier of 11.4 kcal·mol^-1, which is more kinetically favorable than a 1,4-conjugate addition dimerization (with the whole energy barrier of 20.6 kcal·mol^-1, ΔG^‡_dim). The rate-determining step of the ROP or 1,4-conjugate addition dimerization pathway by CTPB/DU8 (intramolecular bis-urea mechanism) is nucleophilic attack by the BnO⁻ anion (TS1_ROP) or by carbanion (TS2_dim), respectively.

As previously reported, two possible mechanisms, intramolecular bis-urea mechanism and activated-U mechanism^45,46, can be considered for the DU8 mediated polymerization of EVP. We also considered these two pathways computationally (Supplementary Fig. 17). In the activated-U mechanism, DU8 is firstly activated by itself through intramolecular hydrogen-bond interactions as suggested by the H4···O2 (1.75 Å) and H3···O2 (2.33 Å) contacts. In contrast, the bis-urea activated intermediate INT0 appears lower in energy (0.0 vs 5.5 kcal·mol^-1) than the activated-U intermediate INT0_iso. Further nucleophilic attack on EVP takes place via TS1_ROPiso with an energy barrier of 12.3 kcal·mol^-1, which is higher than that (11.4 kcal·mol^-1) of the bis-urea activated pathway. Therefore, the greater stability of INT0 and TS1_ROP (11.4 vs 17.8 kcal·mol^-1) and easier nucleophilic attack make the bis-urea activation mechanism more favorable in the DU8-assisted EVP polymerization. To further verify this bis-urea effect, another bis-urea DU6 (Fig. 2) was prepared and used for the ROP of EVP (Table 1, runs 11, and 16−19). DU6 can be seen as the bis-analogue of mono-urea U6. Although DU6 has a computed pK_a of 19.6 similar to that (19.4) of U6, the former led to a significantly higher conversion of EVP (43%) than the latter (22%) under otherwise identical conditions (Table 1, run 11 vs run 7). The resulting poly(EVP)_ROP obtained by DU6 has a M_n of 2.2 kg·mol^-1, which is also higher than that by U6 (1.3 kg·mol^-1). These results are consistent with the results from mono-urea U8 or bis-urea DU8 and previous reports^43,45,46. It is interesting that a conversion of EVP as high as 43% was obtained with DU6 (Table 1, run 11), because precipitation of the resulting polymer from solution under these conditions (Supplementary Fig. 18) shifts the equilibrium towards propagation. It is noteworthy that poly(EVP)_ROP with a M_n as high as 16.1 kg·mol^-1 was successfully synthesized using this highly reactive CTPB/DU6 as the catalyst and BDM as a bifunctional initiator for 24 h (Table 1, run 19; EVP/CTPB/DU6/BDM = 50/1/1.5/0.125). This M_n of 16.1 kg·mol^-1 was relatively lower than the theoretical one (M_n,theor = M_EVP (152.19 g·mol^-1) × ([EVP]₀:[I]₀) × conversion + M_BDM (138.07 g·mol^-1) = 20.8 kg·mol^-1), probably because the precipitation of the resulting polymer from solution resulted in heterogeneous polymerization conditions and more significant transesterification during polymerization (Đ = 1.55). It is noteworthy that the CO₂-based polymers with relatively low M_n and terminal hydroxyl groups represent a class of high value-added polyols and are widely used in the manufacturing of degradable polyurethanes^56,57.

In the case of mono-ureas, a reaction mechanism involving only one urea was considered, according to previous research^53,58. As shown in Supplementary Fig. 16, nucleophilic attack of the BnO⁻ of BnOH on the carbonyl C atom of EVP occurs via a TS1_ROP (U8) with an energy barrier of 12.3 kcal·mol^-1 to INT1_ROP (U8), being accompanied by a proton (H1) transfer from BnOH to deprotonated U8. Then, the ring-opening of EVP occurs via TS2_ROP (U8) with an energy barrier of 2.6 kcal·mol^-1 to form INT3_ROP (U8) in which a new methoxy-anion chain end from the ring-opening of EVP is stabilized by one U8 molecule through hydrogen-bonding interactions. Finally, a proton is transferred from U8 to the oxygen atom of the new methoxy-anion chain end to accomplish the ring-opening polymerization cycle. Therefore, the rate-determining step of ROP by mono-urea is the ring-opening process with the ΔG^‡_ROP of 14.7 kcal·mol^-1.

In the 1,4-conjugate addition dimerization pathway, a hydrogen from the methyl of EVP is captured by the BnO⁻ of BnOH via TS1_dim (U8) with an energy barrier of 18.6 kcal·mol^-1 to generate INT1_dim (U8) while a proton is transferred from BnOH to the deprotonated U8. Then, benzyl alcohol is removed and a rearrangement leads to the more stable INT2_dim (U8) with a carbanion. The carbanion attacks the methine carbon atom of the second EVP through TS2_dim (U8) to provide a dimer INT3_dim (U8) with an energy barrier of 11.7 (20.3−8.6) kcal·mol^-1. An electronic rearrangement leads to a more stable INT4_dim (U8). Furthermore, the complex INT5_dim (U8) containing a dimer of EVP is generated via TS3_dim (U8) by addition of benzyl alcohol with an energy barrier of 5.7 (8.8−3.1) kcal·mol^-1. Finally, a proton transfers from U8 to a BnO⁻ by releasing an energy of 2.7 kcal·mol^-1. By contrast, the ROP pathway overcomes a free-energy barrier of 14.7 kcal·mol^-1, which is kinetically more favorable than the 1,4-conjugate addition dimerization (ΔG^‡_dim: 20.3 kcal·mol^-1). The rate-determining step of the ROP or 1,4-conjugate addition dimerization pathway by CTPB/U8 (mono-urea mechanism) is ring-opening (TS2_ROP) or nucleophilic attack by carbanion (TS2_dim), respectively.

To gain a deeper insight into the effect of ureas on the activity and selectivity of EVP polymerization, the pK_a values of different ureas and the rate-determining steps of both the ROP and 1,4-conjugate addition dimerization pathways mediated by CTPB/ureas (U4, U7, and U9) were also examined computationally, and the results are collected in Supplementary Table 4. In comparison with U8, the less acidic U4 (pK_a,exp = 22.8 and pK_a,cal = 22.0) shows a smaller energy-barrier gap (ΔG^‡_dim − ΔG^‡_ROP = 3.9 vs 5.6 (U8) kcal·mol^-1) between the EVP ROP pathway and the 1,4-conjugate addition dimerization pathway, which is consistent with the experimentally observed low selectivity (47%). Conversely, the more acidic U9 (pK_a,exp = 13.8 and pK_a,cal = 13.9) shows a larger ΔG^‡_dim − ΔG^‡_ROP (8.8 kcal·mol^-1), again consistent with the selectivity of 100% observed experimentally. For comparison, U7, U8, and DU8 with similar acidity (pK_a,cal = 19.0, 17.9, and 18.1) all show large ΔG^‡_dim − ΔG^‡_ROP (5.0, 5.6, and 9.2 kcal·mol^-1), thus resulting in high selectivity of ROP. However, the lower ΔG^‡_ROP (11.4 vs 14.1 (U7) and 14.7 (U8)) in the case of DU8 is consistent with higher conversion (30%) in comparison with U7 (20%) and U8 (22%). This could be ascribed to intramolecular interactions involving the bis-urea.

To expand the range of catalytic systems for EVP polymerization, we computationally tested the catalytic performances of other urea derivatives (Fig. 2, U1−U3, U5, and U6) for this reaction. The results are summarized in Supplementary Table 4, and indicate that U1−U3 with low acidities (pK_a,cal > 21) showed small energy-barrier gaps (ΔG^‡_dim − ΔG^‡_ROP < 4.0 kcal·mol^-1) and thus low selectivity for EVP ROP. In the case of U5, both low conversion and selectivity were observed experimentally because of its low solubility under polymerization conditions. Furthermore, calculations suggested that U6 (pK_a,cal = 19.4) had a similar acidity to U7 (pK_a,cal = 19.0) and also exhibited similar ΔG^‡_ROP (13.6 and 14.1 kcal mol^-1) and ΔG^‡_dim − ΔG^‡_ROP (5.0 kcal·mol^-1). Gratifyingly, similar catalytic activities and selectivities were observed experimentally for U6 and U7 (Supplementary Table 4, runs 6 and 8), illustrating the potential of DFT calculations for developing high-performance catalysts. In summary, the chemoselectivity for ROP of EVP should primarily depend upon the acidities of the ureas, namely their pK_a values, which determines the energy-barrier gap (ΔG^‡_dim − ΔG^‡_ROP) between the ROP pathway and the 1,4-conjugate addition dimerization pathway. As shown in Supplementary Table 4, all the ureas with a pK_a value lower than 19.6 are highly chemoselective for ROP of EVP. In addition, the tethered structure of bis-urea (DU6 and DU8) could favor the catalytic activity, probably because of the implication of a bis-urea assisted intramolecular mechanism.

Post-functionalization of poly(EVP)_ROP

Poly(EVP)_ROP features two C=C double bonds per repeating unit with distinct reactivity, which can be properly engineered by post-functionalization to generate bifunctional polyesters with tunable properties. For instance, the internal alkene of poly(EVP)_ROP was selectively functionalized by Michael addition, leaving the terminal alkene unreacted (Fig. 6a1). Thus, benzyl mercaptan was used as a representative thiol and reacted with poly(EVP)_ROP in DMF at 20 °C in the presence of DBU as the catalyst^59,60. The selective functionalization of the internal alkene was confirmed by NMR spectroscopy (Fig. 6b, Supplementary Figs. 19 and 20) and GPC (Supplementary Fig. 23) of the resulting polyester. In comparison with Fig. 6b1, b2, the signal at 6.91 ppm (H_a) assigned to the internal alkene has disappeared, while the original resonances at 5.81 (H_c) and 5.26 (H_b) ppm for the terminal alkene remain but are shifted to 5.67 (H_c’) and 5.25 (H_b’) ppm. In addition, two new resonances at 3.70 (H_e) and 1.24 (H_f) ppm are observed for the −S–CH₂–(H_e) and −S–CH–CH₃ (H_f) protons, respectively, which confirmed the selective Michael addition of the internal double bond affording poly(EVP)_S1 (Fig. 6a1).

Fig. 6: Post-functionalization, characterization, and chemical recycling of poly(EVP)_ROP.

a Post-functionalization of poly(EVP)_ROP via sequential (1) Michael addition reaction and (2) photoinduced thiol-ene click reaction. (b) Stacked ¹H NMR spectra (400 MHz, CDCl₃) of (1) poly(EVP)_ROP (M_n = 6.5 kg·mol^-1; Table 1 run 13), (2) poly(EVP)_S1 produced via selective Michael addition reaction, and (3) poly(EVP)_S1+S2 produced via two sequential Michael addition and click reactions. (c) Stacked ¹H NMR spectra (400 MHz, CDCl₃) of (1) pristine EVP monomer for comparison; (2) recovered EVP monomer from recycling; (3) poly(EVP)_ROP synthesized from recycled monomer (Supplementary Table 1 run 12).

Next, the terminal alkene was reacted with isobutyl mercaptan via photoinduced thiol-ene click reaction in the presence of 2,2-dimethoxy-2-phenylacetophenone (DMPA) to install the second functional group and yield a bifunctional polyester (Fig. 6a2, Supplementary Figs. 21 and 22)⁶¹, as indicated by the disappearance of the original resonances at 5.67 (H_c’) and 5.25 (H_b’) ppm of the terminal alkene, in comparison with Fig. 6b2, b3. The post-functionalization of poly(EVP)_ROP and the successful preparation of a thiolated polyester were further supported by GPC analysis, which indicated a significantly increased M_n from 6.5 to 8.5 to 14.3 kg·mol^-1 after post-functionalization (Supplementary Fig. 23). This stepwise strategy allows the preparation of a wide range of bifunctional polymeric structures, which are not accessible by ROP of the hydrogenated derivatives EtVP and DEtP^33,34.

Chemical recycling and thermal properties of poly(EVP)_ROP

As aforementioned, a ceiling temperature (T_c) of −141 °C at [EVP]₀ = 1 mol·L^-1 or 5 °C at [EVP]₀ = 5 mol·L^-1 was calculated according to the Dainton’s equation⁴⁹. This low T_c value provides the possibility for facile chemical recycling of the polyesters. For example, the bulk thermolysis of poly(EVP)_ROP with an M_n of 6.5 kg·mol^-1 was investigated for the recovery of the EVP monomer. Indeed, its heating at 150 °C under reduced pressure using 5 wt% Sn(Oct)₂ as the catalyst afforded the pure EVP monomer. Vacuum distillation allowed easy recovery of the EVP monomer in high yield (96%) and high purity (>99.5%), as confirmed by both ¹H NMR (Fig. 6c2) and gas chromatography (Supplementary Fig. 24). After simple drying over CaH₂, the recovered EVP monomer could be repolymerized with a conversion of 29% to oligo(EVP)_ROP with M_n = 1.4 kg·mol^-1 and Ð = 1.46 (Supplementary Table 1, run 12, Fig. 6c3 and Supplementary Fig. 25), values which are very similar to those for the polymer obtained from the original monomer (Table 1, run 1, M_n = 1.5 kg·mol^-1 and Ð = 1.42). These results successfully establish the “monomer-polymer-monomer” closed-loop life cycle for EVP and demonstrate that poly(EVP)_ROP is a highly promising sustainable polymeric material^62,63.

The thermal stability of poly(EVP)_ROP was analyzed by differential scanning calorimetry (DSC; Supplementary Fig. 26) and thermogravimetry (TGA; Supplementary Fig. 27). As shown in Supplementary Fig. 26, poly(EVP)_ROP is amorphous, as indicated by the absence of crystallization and melting transition peaks in the DSC curve. Poly(EVP)_ROP with a M_n of 16.1 kg·mol^-1 exhibited a T_g value of −18.3 °C, which is higher than reported for hydrogenated EVP-derivatives (−38.8 °C and −42.0 °C for poly(EtVP) and poly(DEtP), respectively)³³. In addition, DSC analyses showed that the T_gs can be easily tuned by functionalization, and poly(EVP)_S1 and poly(EVP)_S1+S2 have a T_gs of −7.1 (Supplementary Fig. 28) and −17.6 °C (Supplementary Fig. 29), respectively. As show in Supplementary Fig. 27, the TGA of poly(EVP)_ROP with a M_n of 6.5 kg·mol^-1 displays three successive steps, the initial extrusion of EVP, followed by the cross-linking of the pendant double bonds and the final decomposition⁵⁰. The relatively low temperature of 74.4 °C at 5% weight loss (T_d,5%) may be ascribed to the presence of hydroxyl-terminated polymer chains susceptible of backbiting reactions at high temperature to afford the EVP monomer, which was verified by ¹H NMR analysis (Supplementary Fig. 34). To improve the polymer thermal stability, acryloyl chloride was introduced and reacted with the terminal hydroxyl group of the polymer. The ω-terminal acrylate group capped poly(EVP)_ROP (Supplementary Figs. 30 and 31) showed good thermal stability with only a 5% weight loss at 291 °C (Supplementary Fig. 32). In contrast, the mono-functionalized poly(EVP)_S1 displayed a one-step decomposition profile with a T_d,5% at 270.4 °C and T_d,95% at 430.9 °C (Supplementary Fig. 28), while the bifunctionalized poly(EVP)_S1+S2 showed a T_d,5% at 303.6 °C and T_d,95% at 392.0 °C (Supplementary Fig. 29). Note that the mono-functionalized poly(EVP)_S1 has a broader decomposition profile than bifunctionalized poly(EVP)_S1+S2. This result is consistent with cross-linking of the pendant double bonds occurs at high temperature⁵⁰.

In summary, EVP obtained by telomerization of CO₂ and 1,3-butadiene has been selectively homopolymerized by ring-opening using a phosphazene/urea binary catalyst. A class of bifunctional aliphatic polyesters poly(EVP)_ROP was thus successfully developed. The catalytic activity and chemoselectivity were highly dependent upon the nature of the ureas and the polymerization conditions. Comparative DFT calculations on ROP and 1,4-conjugate addition dimerization pathways were performed to elucidate the details of the reactions and the origin of the chemoselectivity of the ROP of EVP. The resulting poly(EVP)_ROP can be chemically recycled back to the monomer at 150 °C in the presence of Sn(Oct)₂, establishing the complete polymer recyclability. Moreover, poly(EVP)_ROP containing two pendent C=C double bonds with distinct reactivity could be easily and selectively post-functionalized to produce bifunctional polyesters. This study will not only provide a promising route to prepare bifunctional and recyclable CO₂-based aliphatic polyesters with diverse properties, but also highlight the progress that can be made in using CO₂ and 1,3-butadiene as feedstocks for the synthesis of more added-value chemicals.

Methods

Materials

The monomer EVP dried over CaH₂ for 24 h and distilled under reduced pressure. THF was purified by first purging with dry nitrogen, then by passing through columns of activated alumina. 3-Chloroperoxybenzoic acid (75%) and benzyl mercaptan (98%) were obtained from Aladdin and 2,2-dimethoxy-2-phenylacetophenone (DMPA, 98%) from Energy Chemical Co. Benzyl alcohol (BnOH, 99%) was purchased from J&K Scientific. Solutions of 1-tert-Butyl-2,2,4,4,4-pentakis(dimethylamino)-2λ⁵,4λ⁵-catenadi(phosphazene) (t-BuP₂) in THF (2.0 mol/L) and of 1-tert-butyl-4,4,4-tris(dimethylamino)-2,2-bis[tris(dimethylamino)phosphoranylidenamino]-2λ⁵,4λ⁵-catenadi(phosphazene) (t-BuP₄) in n-hexane (0.8 mol/L) were purchased from Sigma-Aldrich and used as received. All other chemicals were purchased from commercial suppliers and used without further purification unless otherwise noted.

Instruments

Nuclear magnetic resonance (NMR) spectra were recorded on a Bruker AVANCE NEO 400 MHz NMR spectrometer (400 MHz for ¹H NMR, 100 MHz for ¹³C NMR). Correlation spectroscopy (COSY) and heteronuclear single-quantum coherence (HSQC) techniques were used to assist in the assignment of ¹H and ¹³C NMR spectra. The molar mass (M_n) and dispersity (Ð = M_w/M_n) were determined by Gel Permeation Chromatography (GPC) equipped with an Agilent HPLC system, a 1260 Hip degasser, a 1260 Iso pump and a 1260 differential refractometer detector. One PLgel 5 μm guard column and three Mz-Gel SD_plus (10³Å, 10⁴Å, and 10⁵ Å) columns were connected in series. The GPC columns were eluted with THF at 1.0 mL/min at 40 °C. The sample concentration used for GPC analysis was 5 mg/mL. Matrix-assisted laser desorption/ionization time-of-flight mass spectroscopy (MALDI-TOF MS) analyses were conducted on a Bruker Microflex MALDI-TOF MS spectrometer equipped with a 337 nm nitrogen laser. Trans-2-[3-(4-tert-butylphenyl)-2-methyl-2-propenylidene]-malononitrile (DCTB) was used as the matrix and CF₃COONa as the ionization agent. Differential scanning calorimetry (DSC) was performed using a TA differential scanning calorimeter DSC 25 that was calibrated using high purity indium. Measurements were performed under N₂ atmosphere with a flow rate of 50 mL/min. Each sample with a mass of 5–10 mg was used for the measurement. T_gs were determined from the second scan at a heating rate of 5 °C/min following a slow cooling rate of 5 °C/min to remove the influence of thermal history. Thermogravimetric analysis (TGA) measurements were performed on a STA 800 thermogravimetric analyzer. The samples were heated from room temperature to 600 °C at a heating rate of 10 °C/min under N₂ atmosphere with a flow rate of 40 mL/min. The post-functionalization experiments were carried out in an ultraviolet lamp box, using an ultraviolet high-pressure mercury lamp (GGZ250) with a wavelength of 365 nm and a power of 250 W. All polymerizations were performed using a cooling bath (EYELA PSL-2000, –80 °C to 20 °C, 100–800 rpm, TOKYO RIKAKIKAI CO., LTD) to control the reaction temperature.

General procedure for the polymerization of EVP (Table 1, run 1)

A flame-dried Schlenk tube was charged with BnOH (0.04 mmol, 4.2 μL), CTPB (0.04 mmol, 48 mg), DU8 (0.06 mmol, 36 mg), and THF (0.14 mL) in a glove box. The Schlenk tube was sealed with a septum and immersed into a cooling bath set to −50 °C. After equilibrium at −50 °C for 10 min under stirring, EVP (2.0 mmol, 0.26 mL) was injected into the Schlenk tube to begin the polymerization. After the desired reaction time, a few drops of acetic acid in trichloromethane were added to quench the reaction and a small portion of the solution was taken and used to determine the conversion of EVP by ¹H NMR. The remaining solution was purified by column chromatography (5:1 hexane: ethyl acetate) to remove the monomer and CTPB, then precipitated in DCM at −10 °C to remove urea. A sticky oil was obtained after concentration in vacuo (73 mg, yield: 24%). The resulting polymer was analyzed by NMR, and GPC. ¹H NMR (400 MHz, CDCl₃): δ 7.36 (s, 4 H, Ar-H of initiator), 6.96-6.84 (m, 13 H), 5.89-5.75 (m, 13 H), 5.34-5.08 (m, 39 H), 2.49-2.27 (m, 26 H), 1.89−1.69 (m, 63 H) ppm; ¹³C NMR (100 MHz, CDCl₃): δ 167.30, 166.67, 141.06, 138.53, 138.03, 136.46, 132.92, 132.51, 128.53, 128.07, 127.97, 116.58, 114.57, 74.13, 72.09, 66.21, 36.23, 33.35, 22.25, 14.13 ppm. M_n = 1.5 kg·mol^-1, Ð = 1.42.

General procedure for the polymerization of EVP (Table 1, run 13)

A flame-dried Schlenk tube was charged with BnOH (0.01 mmol, 1.0 μL), CTPB (0.04 mmol, 48 mg), DU8 (0.06 mmol, 36 mg), and THF (0.14 mL) in a glove box. The Schlenk tube was sealed with a septum and immersed into a cooling bath set to −50 °C. After equilibrium at −50 °C for 10 min under stirring, EVP (2.0 mmol, 0.26 mL) was injected into the Schlenk tube to begin the polymerization. After the desired reaction time, a few drops of acetic acid in trichloromethane were added to quench the reaction and a small portion of the solution was taken and used to determine the conversion of EVP by ¹H NMR. The remaining solution was poured into methanol (50 mL) at −10 °C to precipitate the polymer, which was dried overnight under vacuum in an oven (yield: 82 mg, 27%). The resulting polymer was analyzed by GPC, DSC, and TGA. M_n = 6.5 kg·mol^-1, Ð = 1.24. T_g = −21.9 °C, T_d, _5% = 74.4 °C, T_d, _95% = 486.5 °C.

Depolymerization of poly(EVP) catalyzed by Sn(Oct)₂

A single neck flask was charged with 1.3 g poly(EVP)_ROP (M_n = 6.5 kg·mol^-1) and (173 mg, 5 mol%) Sn(Oct)₂, and heated to 150 °C for 10 min. The recycled EVP was collected by distillation under reduced pressure, purified by passing through a silica-packed short column and collected in 96% yield.

Preparation of poly(EVP)_S1 via Michael addition reaction

A mixture of 120 mg poly(EVP)_ROP (M_n = 6.5 kg·mol^-1), 926 μL benzyl mercaptan (7.9 mmol, 10 equiv.), and 12 μL DBU (0.08 mmol, 0.1 equiv.) was dissolved in 0.2 mL DMF. The reaction mixture was stirred at 20 °C for 6 h. An aliquot of the solution was withdrawn and used to determine the conversion by ¹H NMR. The remaining mixture was poured into excess cold methanol ( − 10 °C). Poly(EVP)_S1 was obtained as a precipitate, which was washed twice with cold methanol and then dried under vacuum at room temperature (196 mg, yield: 90%). The resulting polymer was analyzed by NMR, GPC, DSC and TGA. ¹H NMR (400 MHz, CDCl₃): δ 7.28-7.21 (m, 5 H), 5.80-5.50 (m, 1 H), 5.33-5.02 (m, 3 H), 3.70 (s, 2 H), 2.87-2.69 (m, 1 H), 2.48-2.30 (m, 1 H), 1.84-1.32 (m, 4 H), 1.31-1.18 (m, 3 H) ppm; ¹³C NMR (100 MHz, CDCl₃): δ 172.08, 137.04, 134.99, 127.85, 127.50, 126.02, 116.67, 73.44, 50.05, 40.60, 34.26, 30.83, 25.08, 18.30 ppm. M_n = 8.5 kg·mol^-1, Ð = 1.22. T_g = -38.8 °C, T_d, _5% = 270.4 °C, T_d, _95% = 430.9 °C.

Preparation of poly(EVP)_S1+S2 via thiol-ene click reaction

A flame-dried Schlenk tube was charged with 85 mg poly(EVP)_S1, 3.4 mg (0.013 mmol, 4 wt%) DMPA and 0.5 mL chloroform, then 100 μL isobutyl mercaptan (0.9 mmol, 3 equiv.) was added into the Schlenk tube via a gastight syringe. The tube was then irradiated by UV light at 365 nm for 4 h. The reaction mixture was poured into excess cold methanol (−10 °C) and the precipitate obtained was washed twice with excess cold methanol (−10 °C) and then dried under vacuum at room temperature (106 mg, yield: 94%). The resulting polymer was analyzed by NMR, GPC, DSC and TGA. ¹H NMR (400 MHz, CDCl₃): δ 7.29-7.21 (m, 5 H), 5.07-4.77 (m, 1 H), 3.72 (s, 2 H), 2.90-2.71 (m, 1 H), 2.60-2.24 (m, 5 H), 1.92-1.36 (m, 7 H), 1.30-1.20 (m, 3 H), 0.97 (s, 6 H) ppm; ¹³C NMR (100 MHz, CDCl₃): δ 172.38, 137.09, 127.86, 127.53, 126.04, 116.92, 71.99, 49.67, 40.42, 34.22, 33.05, 30.78, 27.54, 25.23, 21.09, 18.16 ppm. M_n = 14.3 kg·mol^-1, Ð = 1.28. T_g = -42.0 °C, T_d, _5% = 303.6 °C, T_d, _95% = 392.0 °C.

Preparation of ω-terminal acrylate-capped poly(EVP)_ROP

200 mg (∼0.031 mmol of hydroxy end groups) poly(EVP)_ROP (Table 1 run 13) was dissolved in 3 mL anhydrous dichloromethane. The solution was cooled using an ice/NaCl bath, and 43 μL of triethylamine (0.31 mmol, 10.0 equiv) was added. Then, an acryloyl chloride (0.31 mmol, 10.0 equiv) solution (25 μL mL of acryloyl chloride in 3 mL dichloromethane) was added dropwise. The reaction mixture was stirred for 2 h at 0 °C and for another 24 h at room temperature. The reaction mixture was then precipitated by pouring into excess cold methanol (−10 °C). The solid was washed twice with excess cold methanol (−10 °C) and then dried under vacuum at room temperature (190 mg, yield: 95%). The resulting polymer was analyzed by NMR. ¹H NMR (400 MHz, CDCl₃): δ 7.35 (s, 5 H, Ar-H of initiator), 6.44-6.40 (m, 1 H, CH = CH₂ of end group), 6.17-6.11 (m, 2 H, CH = CH₂ of end group), 6.96-6.84 (m, 42 H), 5.89-5.75 (m, 44 H), 5.34-5.08 (m, 131 H), 2.49-2.27 (m, 86 H), 1.89-1.69 (m, 222 H).