Conformation-driven reversibility control in ring-opening metathesis polymerization of non-bicyclic cyclooctenes

Abstract

Reversible ring-opening metathesis polymerization (ROMP) of cyclooctene (COE) derivatives remains a challenge due to their high ring strain energies (RSEs). While previous strategies rely on fused bicyclic systems to reduce RSEs, depolymerization of non-bicyclic COE polymers has proven difficult. Here, we demonstrate efficient reversible ROMP of non-bicyclic COE derivatives by introducing conformational constraints via geminal tert-butyl and hydroxy substituents. These tailored monomers enable depolymerization efficiencies exceeding 90%, achieving near-quantitative monomer recovery under optimized conditions. Experimental and computational analyses reveal that intramolecular OH–π interactions stabilize cyclic conformations in hydroxy-containing monomers, while steric hindrance in the ring-opened form is also critical for efficient depolymerization. This work highlights the interplay between substituent effects, steric control, and molecular conformation in enabling chemical recyclability. It offers an alternative molecular design strategy for developing sustainable materials through reversible polymerization, challenging the conventional reliance on bicyclic systems.

High ring strains hinder the reversible ring-opening metathesis polymerization (ROMP) of cyclooctenes, requiring fused bicyclic monomers. Here, the authors functionalise cyclooctene with tert-butyl and hydroxy substituents to introduce conformational constraints and enable reversible ROMP of non-bicyclic cyclooctenes.

Introduction

Ring-opening metathesis polymerization (ROMP) is a versatile and powerful tool for synthesizing advanced polymeric materials, particularly in developing recyclable polymers^1–5. Its inherent reversibility with ring-closing metathesis depolymerization (RCMD) makes it an attractive method for creating sustainable materials with tailored properties. Among ROMP-compatible monomers, cyclooctene derivatives have drawn considerable attention due to their potential to produce functional polymers with diverse applications^6–12. However, the relatively high ethenolysis ring strain energy (ERSE) of cyclooctenes (8.2 kcal/mol)⁶, compared to those of cyclopentene (5.2 kcal/mol)⁶ and cycloheptene (6.26 kcal/mol)¹³, poses a significant barrier to depolymerization, limiting their chemical recyclability (Fig. 1a).

Fig. 1. Ethenolysis ring-strain energy and reversible polymerization of cycloalkenes.

a ERSE and reversible polymerization of five- to eight-membered cyclic olefins via ROMP and RCMD. b Reversibly polymerizable C6 and C8 bicyclic monomers with controlled ERSEs. c Dienes and their RCM ability for eight-membered ring formation. d This work: Non-bicyclic monomers enabling reversible ROMP and RCMD with controlled ERSEs.

Recent advances have focused on modifying cyclooctene structures with fused bicyclic systems to address this challenge by reducing their RSEs. Notably, Wang and coworkers introduced a trans-cyclobutane fused at the 5- and 6-positions of cyclooctene (Fig. 1b), which reduced the ERSE from 8.2 kcal/mol to 4.9 kcal/mol, enabling efficient depolymerization of the resulting polymers⁶. Similarly, Lamb and coworkers demonstrated that polymers synthesized from the ROMP of oxa-fused cyclooctene derivatives could undergo quantitative depolymerization at an elevated temperature of 70 °C (Fig. 1b)⁸. These strategies for modulating ring strain using bicyclic constraints have also been successfully used for cyclohexene derivatives, where increasing their RSEs enabled ROMP, as reported by Chen¹⁴ and Hong¹⁵ (Fig. 1b).

Despite these advances, chemically recyclable polymers synthesized from simpler non-bicyclic cyclooctenes remain elusive. Previous studies have demonstrated that even commonly employed strategies for five- to seven-membered rings fail to drive RCM for cyclooctene derivatives. For example, Grubbs and Kirkland reported that geminal di-(CO₂Et)-functionalized dienes 1 and 2 (Fig. 1c) yielded only dimeric products under standard RCM conditions¹⁶. Similarly, Grubbs and coworkers found that geminal Me- and OTES-functionalized acyclic diene 3 (Fig. 1c) did not yield any eight-membered-ring product under standard RCM conditions but instead formed some dimeric products via intermolecular metathesis reactions¹⁷. These findings suggest that the Thorpe-Ingold effect, which facilitates the formation of five- to seven-membered rings via RCM, is ineffective for eight-membered rings. However, when Grubbs and colleagues targeted benzene- or cyclohexene-fused bicyclic cyclooctenes 4a–6a (Fig. 1c), RCM proceeded smoothly¹⁶, as further demonstrated in the studies of Wang⁶ and Lamb⁸ (Fig. 1b). These results underscore the critical role of strong conformational constraints, such as those imposed by bicyclic systems, in enabling the successful formation of eight-membered cyclic alkenes via RCM.

Although previous reports indicated that achieving cyclooctene formation from a completely linear diene may not be feasible, appropriate conformational control of a non-bicyclic cyclooctene through a strategically designed conformational constraint could enable reversible ROMP and RCMD. It is well established that sterically bulky tert-butyl groups can restrict the conformation of cyclic^18,19 and linear²⁰ compounds. Thus, we hypothesized that replacing a fused ring with a tert-butyl group could induce a conformational change in the eight-membered cyclic olefin, thereby enabling reversible polymerization via ROMP and RCMD. Herein, we demonstrate that appropriate functionalization of the cyclooctene ring—rather than relying on a bicyclic constraint—can induce a conformational shift that facilitates reversible polymerization and depolymerization under practical conditions.

Results

Monomer design and polymerization study

To achieve the goal of RCMD for non-bicyclic cyclooctenes, we synthesized cyclooctene monomers bearing a tert-butyl substituent: (Z)−8-(tert-butyl)cyclooct-4-en-1-ol (M1), featuring vicinal tert-butyl and hydroxy substituents; (Z)−1-(tert-butyl)cyclooct-4-en-1-ol (M2), featuring geminal tert-butyl and hydroxy substituents; and additional monomers (M3–M6) with substituents other than the tert-butyl group positioned geminal to the hydroxyl group (Fig. 2a). This approach allows for a comparative study of the depolymerization behavior of the resulting polymers with different substituents. Following this plan, monomer M1 was synthesized via epoxidation of 1,5-cyclooctadiene, followed by epoxide opening using tert-BuLi (Supplementary Fig. 1). Monomers M2–M6 were prepared through epoxide opening with LiAlH₄, followed by Swern oxidation, and subsequent treatment with RLi or RMgX, affording the desired products in good yields (Supplementary Fig. 1).

Fig. 2. Designed monomers and optimization of polymerization conditions.

a Designed monomers. b Optimization of polymerization conditions for M2. ^a Reaction conditions: (Z)−1-(tert-butyl)cyclooct-4-en-1-ol (M2, 0.55 mmol), G2 or G3 (M-to-C ratio = 1000:1) in CH₂Cl₂ (0.55 mL for 1.0 M monomer concentration and 1.1 mL for 0.50 M monomer concentration) at 25 °C for 1 h under inert conditions. ^b Conversion determined by ¹H NMR analysis of the crude reaction mixture, based on olefin peaks in the monomer and polymer. ^c M_n (number-average molecular weight) calculated from the initial monomer-to-initiator ratio, multiplied by conversion, plus the contribution from end groups. The end groups were assumed to consist of the catalyst ligand and the terminal olefin from ethyl vinyl ether (EVE) quenching. ^d M_n and Ð (dispersity, M_w/M_n) determined by gel permeation chromatography (GPC) in THF using a polystyrene standard.

With these monomers in hand, we optimized the reaction conditions for polymerization using monomer M2. Polymerization was carried out with the Grubbs second-generation (G2) and third-generation (G3) catalysts at a monomer-to-catalyst ratio of 1000:1, varying the monomer concentration in CH₂Cl₂ (Fig. 2b, entries 1–4). Both catalysts achieved excellent conversions (>96%) within 1 h at a monomer concentration of 0.50 M. However, G2 afforded a high-molecular-weight polymer P2 with better polydispersity (M_n up to 179.2 kDa, Đ ∼1.12; Fig. 2b, entry 1) compared to G3, which produced P2 with a lower molecular weight and slightly higher polydispersity (M_n up to 141.5 kDa, Đ ∼1.2; Fig. 2b, entry 2). Increasing the monomer concentration to 1.0 M with G2 resulted in a polymer with a marginally higher molecular weight and narrower polydispersity (M_n up to 182 kDa, Đ ∼1.11; Fig. 2b, entry 3). In contrast, G3 produced a polymer with a more pronounced increase in molecular weight but only moderately improved polydispersity (M_n up to 161.1 kDa, Đ ~ 1.16; Fig. 2b, entry 4). This trend, where G2 performs slightly better than G3 in molecular weight control and polydispersity, aligns with the findings reported by Chen¹⁴ and Lamb⁸. Considering its superior performance, cost, and stability, we selected G2 at a monomer concentration of 1.0 M (Fig. 2b, entry 3) for further polymerization studies.

After optimizing the reaction conditions, we conducted a time-dependent study to evaluate the progress and controllability of the polymerization. We performed the experiments in triplicate and observed that as the reaction time increased, the conversion of monomer M2 to polymer P2 also increased. However, the M_n value did not consistently correlate with conversion (Fig. 3a and Supplementary Table 1). Specifically, we noted slow initiation, an initially higher M_n than the theoretical M_n, followed by a decrease in M_n with increasing the conversion and reaction time. The observed decrease in M_n over time suggests secondary metathesis pathways, such as chain transfer or backbone scission, which may also generate cyclic polymers or oligomers. To probe this, MALDI-TOF analysis was performed, but the poor ionization efficiency of these polymers hindered conclusive detection of cyclic species. These results indicate that (Z)−1-(tert-butyl)cyclooct-4-en-1-ol (M2) is not suitable for living polymerization. This finding aligns with previous reports highlighting the challenges of achieving living polymerization with cis-cyclooctene^8,21. The moderate RSE of cis-cyclooctene reduces its reactivity for living ROMP, likely due to competition between cross metathesis and ring-opening metathesis. In contrast, Walker et al. demonstrated that the more strained trans-cyclooctene can undergo living ROMP²¹.

Fig. 3. Polymerization study.

a Conversion vs. M_n for the polymerization of M2 using G2. b Dependence of molecular weight (M_n) of P2 on the M-to-C ratio. c van’t Hoff plot for the polymerization thermodynamics of M2 (reaction conditions: monomer M2 (1.0 equiv.) and G2 (0.20 mol%) in CDCl₃ (1.0 M)). d ΔH and ΔS of polymerization (reaction conditions: M1–M6 (1.0 equiv.) and G2 (0.40 mol%) in CDCl₃ (0.050 M)). Molecular weights were measured by GPC using THF as the eluent. Each data point represents the average of three independent experiments, and error bars indicate the standard error (SE).

Next, to confirm the molecular weight control, we conducted polymerization experiments with monomer M2, varying the monomer-to-catalyst ratio under the optimized reaction conditions. Experiments were performed at monomer-to-catalyst ratios of 250:1, 500:1, 750:1, and 1000:1, with at least three replicate experiments for each condition. We observed a consistent increase in the molecular weight (M_n) of polymer P2 as the monomer-to-catalyst ratio increased (Supplementary Table 2). A plot of the M_n values as a function of the effective degree of polymerization (Fig. 3b) exhibited a linear relationship (r² = 0.99, indicated by the red line). At higher monomer-to-catalyst ratios, high-molecular-weight polymers with narrow dispersity (Ð) were obtained. For example, at a monomer-to-catalyst ratio of 1000:1, P2 had an M_n of 168.1 kDa and a Ð of 1.13 (Supplementary Table 2, entry 4). These results confirm that M_n can be effectively controlled by adjusting the monomer-to-catalyst ratio.

Because the monomers are unsymmetrical, the resulting polymers likely contain head–head (H–H), head–tail (H–T), and tail–tail (T–T) linkages. To examine this, we analyzed polymer P2 by 2D NMR (COSY, HSQC, HMBC; Supplementary Figs. 133–135). The spectra confirm the presence of all three linkage types, though quantification of their relative proportions is difficult due to significant peak overlap in both olefinic and aliphatic regions.

Finally, polymerization of other monomers (M1, M3–M6) was carried out under the optimized reaction conditions using a monomer-to-catalyst ratio of 1000:1. The resulting polymers (P1, P3–P6) were obtained in moderate to good yields, with molecular weights (M_n) ranging from 78 to 168 kDa and polydispersities (Ð) between 1.15 and 1.47 (Supplementary Fig. 4).

We investigated the thermodynamics of the polymerization of M2 by conducting experiments at six different temperatures, ranging from 20 to 45 °C. The initial monomer concentration was maintained at 1.0 M in CDCl₃ with 0.20 mol% of G2 for 12 h to ensure equilibrium was reached (Supplementary Fig. 7). The equilibrium concentration of the remaining monomer, [M]_e, was determined via ¹H NMR (ESI Supplementary Fig. 8). We then plotted the natural logarithm of [M]_e against the inverse of temperature (1/T, in K⁻¹) (van’t Hoff plot) and fit the data linearly using the equation ln[M]_e = ∆H/RT – ∆S/R (Fig. 3c). From this analysis, we extracted the enthalpy (ΔH) and entropy (ΔS) of polymerization, obtaining ∆H = −6.12 kcal/mol and ∆S = −13.17 cal/(mol·K). Additionally, we determined the ceiling temperature (T_c) to be 192 °C.

We next investigated the thermodynamics of the polymerization for monomer M3 under conditions similar to those used for M2, employing six different temperatures ranging from 20 to 45 °C, 1.0 M initial monomer concentration, and 0.20 mol% of G2 in CDCl₃. At all six temperatures, no remaining monomer was detected in the reaction mixtures by ¹H NMR (Supplementary Fig. 10). To further study the thermodynamics of polymerization for M3 and M5, we extended the temperature range to 40–60 °C while maintaining a 1.0 M initial monomer concentration and 0.20 mol% of G2 in CDCl₃ (Supplementary Figs. 11 and 12). At elevated temperatures, we observed isomerization of the residual monomer at equilibrium. To mitigate this, we repeated the experiments in benzene (45–70 °C) under identical conditions (Supplementary Fig. 13), but isomerization persisted even in rigorously dried solvents. Since benzoquinone is known to suppress metathesis isomerization^22,23, we added it as an additive. However, ROMP efficiency decreased sharply, with only 40% conversion after 48 h (Supplementary Table 3 and Fig. 14).

To determine ΔH and ΔS without interference from isomerization, we reduced the initial monomer concentrations to 0.050 M and performed polymerization thermodynamics experiments for other monomers (M1–M6) under these conditions. We conducted experiments using 0.40 mol% G2 in CDCl₃, at five different temperatures ranging from 20 to 60 °C. The resulting data were used to construct van’t Hoff plots, from which the ΔH and ΔS values for the polymerization were extracted (Fig. 3d and Supplementary Figs. 15–24). For M6, no remaining monomer was detected at equilibrium (Supplementary Fig. 23), while for M1, isomerization of the residual monomer was observed (Supplementary Fig. 24). Consequently, we were unable to extract ΔH and ΔS values for M1 and M6.

To evaluate thermal and mechanical properties, we conducted TGA, DSC, and UTM testing (Supplementary Table 4). All polymers exhibited 5% weight loss temperatures above 280 °C (Supplementary Fig. 25). Glass transition temperatures (T_g) ranged from −46.1 to 43.7 °C, varying with substituents (Supplementary Fig. 26). Hydroxy-substituted polymers (P2–P6) showed higher T_g than methoxy analogues (P7–P11), while aromatic groups (P6, P11) further increased T_g via π–π interactions. Alkyl substituents decreased T_g with increasing bulk (P5 > P4 > P3), except tert-butyl derivatives (P2, P7), which showed unexpectedly high T_g due to restricted backbone mobility. P10 exhibited the lowest T_g, likely due to the flexible methoxy substituent.

Mechanical testing (Supplementary Fig. 27) revealed that P6 and P11 were tougher than alkyl-substituted polymers, with P6 showing higher strength and modulus due to hydrogen bonding, while P11 was softer and more ductile. P2 showed higher strength than P7, but P7 had lower modulus and toughness. P4 displayed the highest strength, attributed to enhanced chain entanglement and hydrogen bonding. In contrast, P3, P8, P9, and P10 were too weak and sticky to test reliably.

Depolymerization studies

To evaluate the depolymerization capability of polymers (P1–P6), we first optimized the reaction conditions for the depolymerization of poly(Z)−1-(tert-butyl)cyclooct-4-en-1-ol (P2), which contains geminal tert-butyl and hydroxy substituents. Initially, we conducted the reaction in CHCl₃ using G2 (1 mol%) at 25 °C with a polymer concentration of 10 mM (with respect to the double bond). However, after stirring for 24 h, no monomer formation was observed (Supplementary Table 5, entry 1). Increasing the temperature to 40 °C resulted in 71% conversion after 24 h, as determined by ¹H NMR spectroscopy. The reaction yielded a mixture of monomer M2 and its isomer M2’, in a 95:5 ratio (Supplementary Table 5, entry 2). Further raising the temperature to 50 °C increased the conversion to 83%, with the similar M2-to-M2’ ratio of 94:6 (Supplementary Table 5, entry 3 and Supplementary Fig. 28).

Encouraged by these results, we increased the temperature to 60 °C and monitored the conversion at regular intervals using ¹H NMR spectroscopy. At this temperature, conversion reached 88% after 1 h, 90% after 4 h, and 87% after 18 h. While the higher temperature accelerated the reaction, it also led to increased formation of the isomerized monomer M2’, reaching 18% conversion (Supplementary Table 5, entries 4–6 and Supplementary Fig. 28). These findings suggest that the reaction reaches equilibrium at approximately 90% conversion within 1 h at 60 °C. To minimize the formation of the isomerized product M2’, we diluted the reaction solution to a 5.0 mM concentration, which resulted in a 90% product yield with <1% M2’ formation (Supplementary Table 5, entries 7–8 and Supplementary Fig. 28). To further enhance the conversion rate at this lower concentration, we increased the catalyst loading to 2.0 mol%. While this adjustment raised the conversion to 96%, it also led to over 50% formation of M2’ (Supplementary Table 5, entries 9–10 and Supplementary Fig. 28). In contrast, reducing the catalyst loading decreased the conversion to 67% but maintained excellent selectivity (>99%) for M2 (Supplementary Table 5, entries 11–12 and Supplementary Fig. 28). Based on these experiments, the optimal depolymerization conditions were determined to be 1.0 mol% G2, a polymer concentration of 5.0 mM (with respect to the double bond), a reaction temperature of 60 °C, and a reaction time of 1 h (Supplementary Table 5, entries 7–8).

To evaluate the depolymerization efficiency of other polymers (P1, P3–P6), we conducted experiments under the optimized conditions established for the depolymerization of P2 (Fig. 4a). Polymers with relatively less bulky substituents geminal to the hydroxyl group, such as isopropyl (P3), ethyl (P4), methyl (P5), and phenyl (P6), exhibited significantly lower depolymerization efficiencies than P2. Specifically, polymers P3, P5, and P6 yielded 30%, 19%, and 9% of monomers M3, M5, and M6, respectively (Fig. 4a, c and Supplementary Figs. 31, 33, 34), while P4 did not produce any detectable monomer (Fig. 4a and Supplementary Fig. 32). We also examined the depolymerization of P1, which features vicinal tert-butyl and hydroxy substituents. Despite the presence of a tert-butyl group, P1 exhibited minimal depolymerization, yielding only 3% of M1 under the optimized reaction conditions (Fig. 4a and Supplementary Fig. 35).

Fig. 4. Depolymerization study of polymers (P1–P11, P7’ and P8’).

a Depolymerization of polymers P1–P11, P7’, and P8’. b–g ¹H NMR spectra for P2 (b), P3 (c), P7 (d), P8 (e), P7’ (f), and P8’ (g) before (blue) and after (green) depolymerization with G2 (chloroform; [olefin] = 5.0 mM) at 60 °C. The ¹H NMR spectra of the corresponding monomers (M2, M3, M7, M8, M7’, M8’) are shown in red and for formed isomers (in case of (d) and (e) in black as references) in CDCl₃.

The depolymerization results highlight the interplay of enthalpy and entropy in governing thermodynamics²⁴. The negative ΔH reflects the thermodynamic preference for the polymer state, whereas the negative ΔS reflects the loss of disorder upon polymerization. Both factors influence depolymerization efficiency. For monomers M2, M3, and M5, the observed depolymerization efficiencies correlate more strongly with ΔH than with ΔS. For example, M2, with the smallest decreases in enthalpy and entropy (ΔH = −2.84 kcal/mol, ΔS = −0.72 cal/(mol·K)), undergoes the most efficient depolymerization (91% conversion). In contrast, M5, which has the most negative ΔH (−7.21 kcal/mol) and ΔS (−6.83 cal/(mol·K)), shows the lowest conversion (19%). This apparent contradiction against ΔS arises because, at ~300 K, the enthalpic contribution dominates over the entropic contribution, making ΔH the primary determinant of depolymerization efficiency.

These depolymerization studies of polymers P1–P6 clearly demonstrate that while the presence of a geminal substituent on the cyclooctene ring is necessary, the bulkiness of the substituent plays a crucial role in determining the depolymerization efficiency. Specifically, P1, which contains vicinal tert-butyl and OH groups, is non-depolymerizable under the optimized conditions (Fig. 4a and Supplementary Fig. 35). In contrast, P2, which features geminal tert-butyl and OH substituents, undergoes efficient depolymerization, yielding 91% of monomer M2 (Fig. 4a, b). This finding indicates that for efficient reversible polymerization, the tert-butyl and OH groups must be geminal.

To further investigate the potential critical role of the geminal hydroxyl (OH) group in reversible polymerization (vide infra), we converted the OH group to a methoxy (OMe) group. Methoxy-protected monomers M7–M11 were synthesized by treating monomers M2–M6 with methyl iodide (MeI) in the presence of sodium hydride (NaH) in tetrahydrofuran (THF), affording very good to excellent yields (Supplementary Fig. 2). These modified monomers were then polymerized under optimized conditions, using a monomer-to-catalyst ratio of 1000:1, yielding the corresponding polymers P7–P11 in very good yields (Supplementary Fig. 5).

Depolymerization of P7 under the optimized conditions for P2 produced mainly the isomer M7′ alongside M7 (Supplementary Fig. 37). Due to extensive NMR overlap with oligomers, accurate quantification was difficult, and attempts at chromatographic separation failed. We therefore sought cleaner conditions using additives. With G2 and benzoquinone, a monomer–isomer mixture was still observed, while G2 with acetic acid yielded almost exclusively M7′ (>96%) in 57% isolated yield (Fig. 4a, d)²². The structure of M7′ was confirmed by independent RuHCl(CO)(PPh₃)₃-catalyzed isomerization of M7, which afforded the same product without side isomers (Supplementary Fig. 3).

Similarly, P8 afforded a mixture of M8 and M8′ (M8’ as the major species); under G2/acetic acid conditions, the mixture was obtained in 52% isolated yield with an M8:M8’ ratio of 15:85. (Fig. 4a, e). M8′ was also confirmed by independent isomerization of M8 (> 96% conversion, Supplementary Fig. 3). Computational studies indicate that M7′ and M8′ are thermodynamically more stable than M7 and M8 (Supplementary Table 8), consistent with the observed isomerization. No significant isomerization of P7 and P8 was observed with [Ru-H], indicating that isomerization occurs at the monomer stage. This conclusion is further supported by recent depolymerization studies in which cyclooctene-based polymers produced cyclohexene derivatives²⁵, which we never observed in our system—consistent with isomerization proceeding through monomer equilibria rather than polymer backbones.

Because M7′ and M8′ formed in significant quantities, we evaluated their reactivity. Both polymerized more slowly than M7/M8 but reached >90% conversion after 12 h, affording P7′ and P8′. These polymers efficiently depolymerized back to M7′ (76%) and M8′ (78%) (Fig. 4a, f, g), establishing closed-loop cycles.

Beyond these systems, P9 depolymerized to M9 in 13% yield, whereas P4 did not depolymerize. P10 gave 13% yield, lower than P5 (19%), and P11 showed no depolymerization, unlike its hydroxy analogue P6 (9%) (Fig. 4a and Supplementary Figs. 42–44).

Overall, the results show that while hydroxyl groups are essential for efficient depolymerization in P2, their influence is less pronounced in other monomers. Hydroxy- and methoxy-substituted series also display distinct reactivity trends.

Experimental and theoretical conformational analyses

With the observed trends in depolymerization efficiency, we investigated the conformation of monomer M2. A detailed analysis of the ¹H NMR spectra for all monomers revealed that the tert-butyl-containing monomer (M2) exhibits a distinct splitting pattern with greater peak separation (Supplementary Fig. 45). To assign the peaks for each proton, we recorded COSY and HSQC spectra of M2 (Supplementary Figs. 46 and 47). These assignments identified distinct peaks for the axial and equatorial protons at C3, C8, and C7 (Supplementary Fig. 46), indicating that the ring adopts a rigid conformation and does not undergo flipping. This observation is consistent with previous studies suggesting that bulky tert-butyl groups rigidify ring conformations of cyclic compounds^18,19.

We then conducted NOESY analysis to further investigate the conformation. While the 2D NOESY spectrum showed excessive noise (Supplementary Fig. 48), 1D selective gradient NOESY provided clearer insights (Fig. 5a and Supplementary Figs. 49–55). For the olefinic protons H4 at 5.99 ppm and H5 at 4.04 ppm, NOE interactions were observed with protons H13, while proton H13 at 1.85 ppm displayed NOEs with H4, H5, and H10–H12 (Fig. 5a). These interactions suggest that the hydroxyl proton (H13) is in close proximity to the double bond at C4 and C5. Additional NOE interactions further clarified the molecular conformation: proton H3’ at 2.36 ppm interacted with H8” (Supplementary Fig. 54); proton H8” at 1.50 ppm showed NOEs with H3’ and H10–H12 (Supplementary Fig. 55); proton H10–H12 at 0.92 ppm exhibited interactions with H8’, H13, and H2 (Supplementary Fig. 51); proton H8’ at 1.98 ppm displayed NOEs with H10–H12 and H7’ (Supplementary Fig. 52).

Fig. 5. Experimental and computational conformational studies of M2.

a 1D selective gradient NOESY for H13, H4, and H5 (The atom labels can be found in this figure (c)) in CDCl₃. b IR spectra of monomers M2–M6 (neat). c Lowest-energy conformation of M2.

These selective gradient NOESY results, particularly the interactions involving protons H4 (5.99 ppm), H5 (4.04 ppm), and H13 (1.85 ppm) (Fig. 5a), suggest that the hydroxyl proton in M2 forms an OH–π hydrogen bond with the olefin moiety. This type of OH–π interaction has been well documented²⁶. For example, Lectka and coworkers studied the OH–π interaction using an elegantly designed sesquinorbornane scaffold²⁷. For comparison, similar analyses for monomer M5 revealed weak or no NOE interactions between the olefinic protons and the hydroxyl proton (Supplementary Figs. 56–62), suggesting that M5 forms either very weak OH–π hydrogen bond or none at all.

Infrared spectroscopy further supported these findings. M2 displayed a sharp O–H stretching peak, whereas monomers M3–M6 exhibited broader peaks (Fig. 5b). The sharp peak in M2 indicates selective intramolecular OH–π hydrogen bonding, with minimal contribution from intermolecular hydrogen bonding. Notably, in contrast to Lectka’s study, where stronger OH–π interactions led to a red shift in the IR frequency, our system exhibited the opposite trend. The O–H stretching frequencies for M2 (3691 cm⁻¹) and M5 (3694 cm⁻¹), determined using density functional theory (DFT) at the B3LYP-D3BJ/6-31G* level, were comparable and thus do not account for the observed blue shift. Instead, this shift may result from differences in intermolecular hydrogen bonding. The less bulky methyl group in M5 may permit a greater degree of intermolecular hydrogen bonding, which could red-shift the O–H stretching frequency relative to that of M2.

Based on the COSY, HSQC, 1D selective gradient NOESY, and IR studies, we proposed a stable conformation of M2 (Fig. 5c), where intramolecular OH–π hydrogen bonding plays a key role in stabilizing the eight-membered ring in M2. Consequently, polymer P2, synthesized from M2, undergoes depolymerization more readily than the polymers P3–P6. However, in monomer M7, where the hydroxyl hydrogen is replaced by a methyl group, eliminating the possibility of OH–π hydrogen bonding, polymer P7 still exhibits moderate depolymerization. This suggests that OH–π hydrogen bonding is not the primary factor driving depolymerization.

The independently conducted computational study agreed well with the experimental conformational analysis of M2 (Figs. 5c and 6). In this analysis, stable conformers of the cyclic monomers and their ring-opened forms were identified through a systematic conformer search, followed by refinement using xTB and B3LYP-D3BJ. For cyclic monomers, high-temperature molecular dynamics (MD) simulations at 600 K and manual modeling were also employed to assess their stability. ERSE calculations were performed at the B3LYP-D3BJ/def2-TZVP//B3LYP-D3BJ/6-31G* level.

Fig. 6. DFT-calculated structures and ERSEs of the monomers.

Optimized geometries of the monomers and their corresponding ring-opened forms, along with the calculated relative enthalpies, are shown. Geometry optimizations were performed using the GFN2-xTB method, followed by DFT calculations at the B3LYP-D3BJ/6-31G* level. Single-point energies were obtained at the B3LYP-D3BJ/def2-TZVP level of theory.

The calculated ERSE values of the monomers (Fig. 6 and Supplementary Table 6) correlated well with the RCMD results of the polymers. P1, synthesized from M1 had a high ERSE of 9.96 kcal/mol and exhibited only 3% depolymerization. P3 and P5, derived from M3 and M5 with moderate ERSE values of 7.15 and 7.12 kcal/mol, respectively, showed low depolymerization efficiencies (30% and 19%, respectively. In contrast, P2 and P7, derived from M2 and M7 with the lowest ERSE values (5.97 and 6.41 kcal/mol, respectively), showed different depolymerization behaviors: P2 underwent efficient depolymerization with 91% conversion, whereas P7 showed a relatively lower conversion of 61%, consistent with the ERSE trend.

ERSE represents the energy gap between the cyclic form (plus ethylene) and its corresponding ring-opened form. Lowering ERSE can be achieved by stabilizing the cyclic form or destabilizing the ring-opened form. Previous studies on bicyclic cyclooctene monomers by Wang and coworkers demonstrated that a fused cyclobutane ring at the 5,6-position reduces ERSE by forcing two bulky alkyl chains into a gauche conformation, thereby destabilizing the ring-opened form via increased steric hindrance⁶.

In our non-bicyclic cyclooctene system, steric hindrance of the ring-opened form plays a similarly crucial role (Fig. 6). For instance, M1, which features vicinal substituents, has a high ERSE of 9.96 kcal/mol. In contrast, monomers with geminal substituents exhibit significantly lower ERSE values (5.97–7.15 kcal/mol), reflecting a reduction of over 2 kcal/mol. Among these, M2 experiences the greatest destabilization of the ring-opened form, attributed to additional steric repulsion between the tert-butyl group and the 1-pentenyl chain.

The OH–π interaction, suggested by ¹H NMR and IR spectroscopy, stabilizes the cyclic form and contributes to the decrease in ERSE. DFT calculations further confirmed this interaction by measuring the distances (r₁ and r₂, Fig. 5c) between the hydroxyl proton and the two olefinic carbons (Supplementary Fig. 95 and Supplementary Table 9). In M2, r₁ and r₂ were 2.24 and 2.32 Å, respectively, indicating a strong OH–π interaction. In contrast, M5, which exhibited minimal OH–π interaction in ¹H NMR and IR studies, had longer distances of 2.29 and 2.35 Å. Similarly, M3 and M4 exhibited intermediate r₁ and r₂ values, aligning well with the IR results.

To further investigate substituent effects, Mulliken population analysis was conducted at the B3LYP-D3BJ/def2-TZVP//B3LYP-D3BJ/6-31G* level using model compounds 2,3,3-trimethylbutan-2-ol (M2 model) and 2-methyl-2-propanol (M5 model), which feature the tBu and Me groups geminal to hydroxyl group, respectively (Supplementary Table 10). The atomic charges of the hydroxyl moiety in the M2 model were q(O) = −0.472 and q(H) = 0.288, whereas the corresponding charges for the M5 model were q(O) = −0.454 and q(H) = 0.287. These results suggest a greater polarization of the O–H bond in M2, leading to a larger dipole, which in turn enhances the intramolecular OH–π interaction.

To further analyze stabilization and destabilization effects, we examined monomer M7, in which the hydroxyl group is replaced by a methoxy group. DFT calculations revealed an ERSE of 6.41 kcal/mol for M7, lower than those of M3 and M6 (Fig. 6 and Supplementary Table 6). Interestingly, the lowest-energy conformer of M7 adopts a conformation distinct from that of M2, particularly in the relative orientation of the methoxy (or hydroxyl) group and the olefin moiety. In M7, the methoxy and olefin moieties are oriented away from each other to minimize steric repulsion (Supplementary Fig. 94). This conformation, combined with the absence of a hydroxyl group, prevents the formation of OH–π hydrogen bonding in M7. Despite this, the increased steric hindrance from the geminal tert-butyl and methoxy groups on the tetra-substituted carbon significantly destabilizes the ring-opened form, resulting in moderate depolymerization of P7.

These results highlight the importance of conformational control in both the cyclic and ring-opened forms for efficient reversible ROMP. While intramolecular OH–π hydrogen bonding stabilizes the cyclic form, steric hindrance in the ring-opened form is another key factor driving depolymerization in non-bicyclic cyclooctene systems such as M2 and M7.

Discussion

This study successfully demonstrated the design and synthesis of non-bicyclic cyclooctene derivatives with tailored substituents, enabling reversible polymerization via ROMP and efficient depolymerization via RCMD. Through a systematic investigation of substituent effects, we identified key structural features that enhance depolymerization efficiency. Specifically, while the presence of a tert-butyl group is necessary for reversible polymerization, our findings reveal that its depolymerization efficiency depends significantly on the presence of a secondary substituent, such as a geminal hydroxyl or methoxy group.

Conformational analyses revealed that intramolecular OH–π interactions in hydroxy-substituted monomers stabilize the cyclic form, contributing to lower ERSE values. However, steric hindrance in the ring-opened form also plays a dominant role in driving efficient depolymerization. The high depolymerization efficiency of P7, despite the absence of OH–π hydrogen bonding, underscores the critical role of destabilization effects caused by steric hindrance in the ring-opened form. These findings challenge the conventional reliance on covalently fused bicyclic systems for RSE control, demonstrating the potential of non-bicyclic monomer designs for achieving reversible polymerization. This work provides valuable insights into the interplay between molecular conformation, substituent effects, and steric hindrance in the strategic design of sustainable polymers.

Methods

General procedure for synthesis of poly(Z)−1-(tert-butyl)cyclooct-4-en-1-ol (P2)

In a glass vial containing a solution of (Z)−1-(tert-butyl)cyclooct-4-en-1-ol (M2) (100 mg, 0.55 mmol) in CH₂Cl₂ (0.4 mL), G2 catalyst (0.466 mg, 0.10 mol%) in 0.15 mL CH₂Cl₂ from the stock solution was added under an inert atmosphere. The mixture was stirred at 25 °C for 1 h (monomer concentration 1 M). Gel formation was observed after 30 min, and the reaction continued for 60 min. TLC and proton NMR analysis showed almost complete conversion of the starting material. The reaction was quenched by adding 0.10 mL of ethyl vinyl ether and stirring for an additional 30 min. The product was precipitated by dropwise addition of the quenched reaction mixture into an excess of cold methanol (50 mL). The product was separated by centrifugation and decantation, then dried under high vacuum to obtain poly(Z)−1-(tert-butyl)cyclooct-4-en-1-ol (P2) as a white solid (81 mg, 0.44 mmol, 81%).

General procedure for the depolymerization of poly(Z)−1-(tert-butyl)cyclooct-4-en-1-ol (P2)

In a dry 100 mL round-bottom flask, poly(Z)−1-(tert-butyl)cyclooct-4-en-1-ol (P2) (40 mg) was dissolved in 44 mL of CHCl₃ (5.0 mM concentration with respect to the double bonds in the polymer) under inert conditions and heated to 60 °C. Then, 1.86 mg (1.0 mol%) of G2 in 2 mL of CHCl₃ was added, and the mixture was stirred at 60 °C for 1 h. After 1 h, an aliquot (5 mL) was withdrawn by syringe, quenched with ethyl vinyl ether, concentrated, and analyzed by ¹H NMR. Conversion was determined by comparing the integrals of olefinic signals from monomer versus oligomers, which indicated 91% conversion to monomer M2. The reaction was then quenched by the addition of ethyl vinyl ether and stirred for 30 min. The solvent was evaporated using a rotary evaporator, and the crude product was purified by column chromatography (silica gel, pentane/diethyl ether, 97:3) to afford (Z)−1-(tert-butyl)cyclooct-4-en-1-ol (M2) as a colorless oil (34 mg, 85%).

Author contributions

T.B.M.: Methodology, investigation, conceptualization, writing–original draft. K.C.: Formal analysis, validation. H.Z.: Formal analysis, methodology. S.L.: Investigation, methodology. J.H.P.: Formal analysis. Y.K.K.: Formal analysis, supervision. H.H.: Formal analysis, methodology, supervision. S.H.H.: Conceptualization, methodology, supervision. All authors: Writing–Review & editing.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewers for their contribution to the peer review of this work. A peer review file is available.

Data availability

Detailed experimental procedures, computational details, characterization data for new compounds, and Cartesian coordinates of the calculated structures are available from the Supplementary Data files and source data. The authors declare that the data supporting the manuscript are included in the manuscript, supplementary information, and supplementary data. All data are available from the corresponding author upon request. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-025-66835-0.