Olefin metathesis-based chemically recyclable polymers enabled by fused-ring monomers

A promising solution to address the challenges in plastics sustainability is to replace current polymers with chemically recyclable ones that can depolymerize into their constituent monomers to enable the circular use of materials. Despite some progress, few depolymerizable polymers exhibit the desirable thermal stability and strong mechanical properties of traditional polymers. Here we report a series of chemically recyclable polymers that show excellent thermal stability (decomposition temperature >370 °C) and tunable mechanical properties. The polymers are formed through ring-opening metathesis polymerization of cyclooctene with a trans-cyclobutane installed at the 5 and 6 positions. The additional ring converts the non-depolymerizable polycyclooctene into a depolymerizable polymer by reducing the ring strain energy in the monomer (from 8.2 kcal mol–1 in unsubstituted cyclooctene to 4.9 kcal mol–1 in the fused ring). The fused-ring monomer enables a broad scope of functionalities to be incorporated, providing access to chemically recyclable elastomers and plastics that show promise as next-generation sustainable materials. Depolymerizable polymers can potentially address challenges in polymer sustainability, but most existing systems lack the useful thermomechanical properties of traditional ones. Now, it has been shown that depolymerizable polymers based on olefin metathesis show good thermal stability as well as versatile mechanical properties and that the monomers used to make them can be prepared from abundant materials.

S ynthetic polymers, including synthetic rubber and synthetic plastics, have been used in nearly every aspect of our daily lives. The dominance of synthetic polymers is largely driven by their excellent stability and processability as well as their versatile mechanical properties. However, due to their high durability, waste materials composed of these polymers have accumulated in the ocean and have caused serious concerns for marine ecosystems 1 . In addition, because 90% of these polymers are derived from finite fossil feedstocks, the production of these materials is unsustainable if they cannot be recycled and reused 2 . Efforts to address these issues include the development of biodegradable polymers and mechanical recycling. However, most biodegradable polymers that can be degraded in artificial environments do not undergo efficient degradation in seawater, giving rise to new environmental consequences 3 . Mechanical recycling of polymers typically results in substantial loss of quality, and the recycled material is not suitable for high-performance applications 4 . A particularly attractive strategy is to develop chemically recyclable polymers that can depolymerize into their constituent monomers for recycling and repolymerization 5,6 . Circular use of the materials not only will help to preserve finite natural resources but also can address end-of-life issues for such materials.
To replace currently available commercial polymers, depolymerizable polymers need to match or exceed the properties of the current ones. With exceptions, polymerization is typically favoured in enthalpy (ΔH < 0) and disfavoured in entropy (ΔS < 0). The temperature at which the entropic loss will offset the enthalpic gain is defined as the ceiling temperature T c , and depolymerization is favoured when the temperature is above T c . Common polymers such as polyolefins have high T c values, and their depolymerization is either costly in terms of energy or is susceptible to decomposition. Classical low-T c polymers such as poly(olefin sulfones) 7,8 , poly(α-methyl styrene) 9 and polyaldehydes 10 lack high thermal and chemical stability, and their use has been limited to certain specific applications such as transient electronics [11][12][13] . A promising strategy to address the stability issues is to develop polymers that only undergo efficient depolymerization in the presence of a catalyst. In other words, without the catalyst, the polymer is in a kinetic trap so that it will stay intact even when the temperature is above T c . Recently, it has been shown that ring-opening polymerization of certain cyclic monomers-such as cyclic esters, cyclic carbonates and cyclic olefins-can lead to polymers that can depolymerize into the corresponding monomers in the presence of catalysts but show high thermal stability when the catalysts are removed 6 . For example, Chen and coworkers have shown that a poly(γ-butyrolactone) required heating overnight at 300 °C to be depolymerized, but the depolymerization temperature was reduced to 120 °C when a ZnCl 2 catalyst was added 14 .
Among the catalytically depolymerizable polymers, a particularly attractive system is the one that is based on olefin metathesis 15 . Since olefin metathesis does not occur without a catalyst, unintended depolymerization can be prevented by removing the catalyst. In addition, metathesis is compatible with a wide variety of functional groups and can be conducted in mild or ambient reaction conditions 15 . Metathesis-based depolymerizable polymers are typically made via ring-opening metathesis polymerization (ROMP) 16 of cycloalkenes and can depolymerize through ring-closing metathesis (RCM) 17 to form the corresponding monomers. Compared to the ring-opening polymerization that is based on cyclic esters and cyclic carbonates, ROMP enables the production of polymers with hydrocarbon backbones, which have greater hydrolytic and thermal stability. However, up to now, demonstrations of depolymerizable ROMP polymers have been limited to polymers of five-membered cycloalkenes, namely the polypentenamers 18 . As polypentenamers are typically amorphous polymers with glass transition temperatures (T g ) below or near room temperature 18 , they are not suitable for plastics applications where glassy or semicrystalline polymers are needed. Furthermore, the controlled polymerization of cyclopentene is challenging: the reported cases of controlled polymerization of cyclopentene require that the conversion be limited to a low Olefin metathesis-based chemically recyclable polymers enabled by fused-ring monomers Devavrat Sathe 1 , Junfeng Zhou 1 , Hanlin Chen 1 , Hsin-Wei Su 1 , Wei Xie 1 , Tze-Gang Hsu 1 , Briana R. Schrage 2 , Travis Smith 1 , Christopher J. Ziegler 2 and Junpeng Wang 1 ✉ A promising solution to address the challenges in plastics sustainability is to replace current polymers with chemically recyclable ones that can depolymerize into their constituent monomers to enable the circular use of materials. Despite some progress, few depolymerizable polymers exhibit the desirable thermal stability and strong mechanical properties of traditional polymers. Here we report a series of chemically recyclable polymers that show excellent thermal stability (decomposition temperature >370 °C) and tunable mechanical properties. The polymers are formed through ring-opening metathesis polymerization of cyclooctene with a trans-cyclobutane installed at the 5 and 6 positions. The additional ring converts the non-depolymerizable polycyclooctene into a depolymerizable polymer by reducing the ring strain energy in the monomer (from 8.2 kcal mol -1 in unsubstituted cyclooctene to 4.9 kcal mol -1 in the fused ring). The fused-ring monomer enables a broad scope of functionalities to be incorporated, providing access to chemically recyclable elastomers and plastics that show promise as next-generation sustainable materials. threshold (<20%) 19 or that it be conducted in a variable-temperature fashion 20 .
To address the needs for more robust depolymerizable ROMP polymers, we looked into other cyclic olefins. Because cyclopropene and cyclobutene are highly strained, with reported ring strain energies (RSEs) of 54.5 and 30.6 kcal mol -1 , respectively 21 , the depolymerization of their ROMP polymers is not feasible. The ring strain of cyclohexene is too low to achieve a successful ROMP 22 . Of the seven-membered to twelve-membered cyclic olefins, the eight-membered one is particularly attractive because functionalized cyclooctenes can be made from 1,5-cyclooctadiene, an inexpensive, commercially available material. As a result, poly(cyclooctenes) have been one of the most extensively employed poly(cycloalkenes) 23 . In addition, the Z-alkene in cyclooctene can be inverted into an E-alkene to increase the RSE (the RSE of trans-alkene cyclooctene is 16.7 kcal mol -1 (ref. 21 )) and enable living polymerization 24,25 . However, compared to cyclopentene (RSE = 5.2 kcal mol -1 ), the higher ring strain of cyclooctene (RSE = 8.2 kcal mol -1 ) precludes the depolymerization of polycyclooctene; note that because different RSEs have been reported for cyclooctene, we use our calculated values throughout the text for consistency. We hypothesized that if the RSE of cyclooctene could be reduced to a level comparable to that of cyclopentene, the corresponding polymer would undergo similar depolymerization.
It has been shown that the ring strain of cyclooctene can be lowered by a fused ring at the 5 and 6 positions and that the stereochemistry of the fused ring plays a critical role. For example, Coates and coworkers showed that an imidazolium-fused cyclooctene monomer only afforded macrocyclic oligomers under typical ROMP conditions, suggesting that the imidazolium-fused ring lowers the ring strain of cyclooctene 26 . Scherman et al. showed that a cyclooctene monomer with a trans-fused acetonide afforded much lower yield in polymerization than a cis-acetonide cyclooctene monomer (15% versus 80% when the polymerization was conducted at 1.0 M monomer concentration), indicating that the trans-fused acetonide cyclooctene has a lower RSE than that of its cis analogue 27 . These studies suggest that an appropriate fused ring may lower the ring strain of cyclooctene to enable the depolymerization of the corresponding ROMP polymer (Fig. 1a).

Results and discussion
Calculations of RSEs. To identify a fused ring that can lower the ring strain of cyclooctene to the above-mentioned target, we calculated the RSEs of cyclooctenes with three-, four-, five-and sixmembered rings fused at the 5 and 6 positions, including both cis and trans isomers (Fig. 1b). The computation of RSE was conducted by calculating the enthalpy change of the RCM reaction shown in Fig. 1c using density functional theory at the B3LYP/6-31G(d,p) level, which provided reasonable predictions for the RSEs of cyclic olefins 22 .
Based on the difference in RSE between the fused rings and the unsubstituted cyclooctene (RSE = 8.2 kcal mol -1 ), the calculated results can be sorted into three groups: 1. RSEs that are higher than that of the virgin cyclooctene, including cis-cyclopropane-, trans-cyclopropane-and ciscyclobutane-fused cyclooctenes 2. Fused rings with slightly decreased ring strain, including cis-cyclopentane-, cis-cyclohexane-and trans-cyclohexanefused cyclooctenes 3. Cycloalkane-fused cyclooctenes having the lowest RSEs and having RSEs that are lower than or comparable to that of cyclopentene (RSE = 5.2 kcal mol -1 ), including trans-cyclobutaneand trans-cyclopentane-fused cyclooctenes The calculated RSEs suggest that the ROMP polymers of the trans-cyclobutane-and trans-cyclopentane-fused cyclooctenes would undergo catalytic depolymerization in a way similar to polypentenamers. Because of the synthetic convenience of accessing cyclobutane-and cyclopropane-fused cyclooctenes from 1,5-cyclooctadiene, we employed these cycloalkane-fused cyclooctenes for our experimental and control studies.

Monomer synthesis.
Recently, in an effort to control polymer degradation through mechanochemical activation, we synthesized M1 through a photochemical [2 + 2] cycloaddition of 1,5-cyclooctadiene and maleic anhydride followed by LiAlH 4 reduction (Fig. 2a) 28 . Note that the cyclobutane in M1 is trans-fused to the cyclooctene, indicating a cis-to-trans isomerization of the alkene prior to cycloaddition. Similar inversion of stereochemistry has been observed in other systems [29][30][31][32] . Importantly, the moiety of trans-cyclobutane-fused cyclooctene (tCBCO) is exactly the target moiety we are aiming for. Aside from lactone, tCBCOs with other functional groups-including the diesters M2a and M2b, the diether M3 and the imide M4were conveniently derived from the intermediate anhydride 1. a, Polycyclopentene is depolymerizable, whereas polycyclooctene is not. the difference is caused by a higher RSE in cyclooctene. An appropriate ring (shown as a dashed line) lowers the RSE of cyclooctene to a level that is lower than or comparable to that of cyclopentene, enabling depolymerization of the polymer. b, the calculated RSEs for cyclooctene with 3-, 4-, 5-and 6-membered rings both cis-and trans-fused at the 5 and 6 positions. c, the generalized ring-closing metathesis reaction used to calculate the RSE of a cyclic olefin: the RSE for the cyclic olefin is essentially the enthalpy change for the ring-closing metathesis that affords the cyclic olefin.

Polymerization thermodynamics.
A high monomer concentration was found to be essential for the ROMP of these monomers, as low monomer concentrations resulted in low conversions. In addition, the conversion was also influenced by the temperature. In addition, the entropy changes for these monomers are substantially lower than that for cyclopentene (ΔS = −18.5 cal mol −1 K −1 ). The reduced entropy change can be attributed to the lower loss in translational entropy for the larger-sized monomers and the higher gain in rotational entropy due to the higher number of freed bonds in the eight-membered ring that were attained through the ring-opening reaction. Compared to cyclopentene, the lower entropy loss for the polymerization of tCBCO monomers compensates the lower enthalpy gain and makes the polymerization favourable at high monomer concentrations. For example, at 25 °C and 1.0 M concentration, the Gibbs free energy values (ΔG) for polymerization of M2a and M4 are −0.63 kcal mol -1 and −1.33 kcal mol -1 , respectively, which are more favourable than that for cyclopentene (ΔG = −0.087 kcal mol -1 ). The ceiling temperatures for P2a, P4 and polypentenamer at 1.0 M are 199 °C, 295 °C and 30 °C, respectively. In addition, many functional groups on the 3 position of cyclopentene could reduce the ceiling temperature to values below room temperature, making the monomers challenging to polymerize 18 .
Compared to polypentenamer, the favourable ceiling temperatures of the tCBCO system can enable the preparation of tCBCO polymers of a variety of functionalities through ROMP.
Polymer synthesis. The monomers were subjected to ROMP conditions with ~2 M monomer concentration in dichloromethane and G2 as the initiator, yielding the polymers P1, P2a, P2b, P3 and P4 at high conversions (>80%; Fig. 2b). The number average molecular weight M n , dispersity Đ, glass transition temperature T g and decomposition temperature T d (defined as the temperature at which 5% weight loss occurs) for each polymer are shown in Fig. 2b. Polymers with high molecular weights (M n > 100 kDa) were obtained for all polymers. The high dispersities (Đ > 1.5) are comparable to those for other ROMP polymers of low-strain monomers 22,33 . As stated above, an advantage of the eight-membered cycloalkene is that the cis-alkene can be inverted to a trans-alkene, which is highly strained and is suitable for living ROMP. We converted M2a into E-M2a through photoisomerization by following the protocol demonstrated by Fox and coworkers 34 . 9) and MS ( Supplementary Fig. 10). In addition, monomer M2a, which was isolated from the depolymerization mixture of P2a (characterization data of the isolated products shown in Supplementary Figs. [11][12][13][14], was subjected to the ROMP conditions described above and yielded polymers of high molecular weight (M n = 71 kDa, Đ = 1.53; Supplementary Figs. 15 and 16).  By contrast, when exposed to the same conditions, polycyclooctene did not depolymerize into cyclooctene (Fig. 3d) despite a reduction in molecular weight ( Supplementary Fig. 17). Moreover, the ROMP polymers of both cis-and trans-cyclopropane-fused cyclooctenes at the 5 and 6 positions (cis-gDCC-CO and trans-gDCC-CO, respectively) did not depolymerize into the corresponding monomers (Fig. 3e,f), which is consistent with the high RSEs of these monomers (Fig. 1b). In addition, a 1,9-decadiene with trans-cyclobutane installed at its 5 and 6 positions underwent RCM at quantitative conversion to form tCBCO when it was refluxed in the presence of G2 (Supplementary Fig. 18). By contrast, when the cis-cyclobutane diene was subjected to the same conditions, no cis-cyclobutane-fused cyclooctene was observed in the mass spectrum, which instead showed a mixture of acyclic and cyclic oligomers ( Supplementary  Figs. 19-21). The depolymerization and RCM studies confirm the computational results that trans-cyclobutane appreciably reduces the RSE of cyclooctene and highlight the impact of the size and stereochemistry of the fused ring on the RSE of a cyclic molecule. The depolymerization conditions, including temperature and concentration, were screened in order to understand the thermodynamics and to identify the optimal conditions for depolymerization. Conversions of 94-98% were obtained when the depolymerization of P2a was conducted at concentrations of [olefin] ≤ 200 mM at 50 °C (Supplementary Table 1). In particular, the concentration of 200 mM corresponds to a weight concentration of 5% (w/w), meaning that 5 g of polymers can be depolymerized using 100 ml of Poly(cis-gDCC-CO) Poly(cis-gDCC-CO) treated with G2 for 2 h Cis-gDCC-CO   Figs. 10, 11 and 25). The fraction of the oligomers was obtained from the integration of the GPC refactive index (RI) traces (details in Supplementary Information). When the depolymerization was conducted at 50 °C, no residual polymer peak was observed after 3 min. This rapid depolymerization makes it inconvenient to obtain sufficient data points for kinetic studies ( Supplementary Fig. 22). The time after which no residual polymer was observed was 10 min at 45 °C (Supplementary Fig. 23) and 120 min at 19 °C ( Supplementary Fig. 24), which allowed sufficient data points to be obtained for kinetic analysis. The fractions of monomer, oligomers and residual polymer were plotted against depolymerization time for depolymerization at 45 °C ( Supplementary  Fig. 23) and 19 °C (Supplementary Fig. 24). As shown in the plots, the polymer fraction continuously decreased while the monomer fraction continuously increased throughout depolymerization. However, the fraction of oligomers first increased and reached a maximum in the range of ~30-40% before starting to decrease (Supplementary Fig. 23 and 24). Most of the oligomers eventually turned into the monomer. For example, no residual polymer was observed for the depolymerization of P2a at 45 °C at 10 min; at that time, the fractions of monomer and oligomer were 80% and 20%, respectively. The oligomers continued to convert into monomer until an equilibrium was reached at around 20 min. The equilibrium fractions for monomer and oligomers were 93% and 7%, respectively. In addition, when the depolymerization was conducted at 19 °C, the fractions of the oligomers were higher than those of the monomer at the early stage of depolymerization (0-30 min). These results suggest that while the monomer is the thermodynamic products, the formation of the oligomers is kinetically more favourable. The formation of cyclic oligomers was not observed for the depolymerization of polypentenamers 35 . This difference can be attributed to the difference between the RSEs of the monomer and dimer: the difference in RSE between the eight-membered and sixteen-membered rings might be less drastic than that between the five-membered and ten-membered rings.
Studies by Badamshina et al. showed that the depolymerization of polypentenamer proceeds through a random chain cleavage pathway 36 . Recently, Kennemur and coworkers studied the depolymerization mechanism of a bottlebrush polypentenamer and found that the chain cleavage occurs in an end-to-end fashion 35 . In this case, due to the steric hinderance of the side chain in the bottlebrush, the ruthenium catalyst preferentially coordinates to the terminal olefin over the internal olefin. Since there is no long side chain in the linear tCBCO polymers, it is expected that the depolymerization will proceed in a random fashion, similar to that for a linear polypentenamer. At the early stage of depolymerization, the polymer peak shifted to a longer retention time; meanwhile, a shoulder peak that connects the polymer peak and the oligomer peaks appeared . The shoulder peak eventually disappeared along with the polymer peak. The appearance of a shoulder peak supports random chain cleavage, since a shoulder peak is not expected for the end-to-end depolymerization 35 . In addition, end-to-end depolymerization would result in a linear reduction of molecular weight over the fraction of depolymerization; however, the plot of molecular weight against depolymerization conversion did not show a linear relationship (Supplementary Figs. 23 and 24), further suggesting that random chain cleavage occurred during the depolymerization.
Structure and RSE relationships. The finding that the ring strain of cyclooctene is reduced by a highly strained cyclobutane is somewhat counterintuitive. Transannular strain and torsional strain are important sources of ring strain in medium-sized (seven-to twelve-membered) rings 37,38 . To assess the impact of the cyclobutane-fused ring on the transannular interaction, we looked at all of the non-bonded H-H distances on the unsubstituted cyclooctene and tCBCO. The list of the H-H distances in Supplementary  Fig. 26 shows that only one notable transannular interaction is present in the unsubstituted cyclooctene: it occurs between H3a and H8a at a distance of 2.15 Å. Incorporation of a trans-cyclobutane-fused ring only slightly increases the H3a-H8a distance (2.18 Å), indicating that the transannular strain is not noticeably impacted by the trans-cyclobutane-fused ring. Since the ring strain of a cyclic molecule is the relative energy between the cyclic molecule and an acyclic reference molecule, to understand the impact of the fused rings  on the ring strain, we examined the structural difference between the cyclooctenes and the acyclic 1,9-decadienes (the structures of which were obtained for use in the ring strain calculations) via the Newman projection along the C5-C6 bond (Fig. 4). The dihedral angles H5-C5-C6-C7, H6-C6-C5-C4 and H5′-C5-C6-H6′ in cyclooctene are 25.8°, 24.6° and 19.4°, respectively, generating substantial torsional strain. The torsional strain becomes negligible in the acyclic 1,9-diene, as the dihedral angles are all close to 60°. We then looked at the trans-cyclobutane-fused cyclooctene, in which the H5′ and H6′ (of an unsubstituted cyclooctene) are replaced with C5′ and C6′. As shown in Fig. 4, the dihedral angles H5-C5-C6-C7, H6-C6-C5-C4 and C5′-C5-C6-C6′ in tCBCO are 30.8°, 29.2° and 17.7°, respectively. Notably, these dihedral angles remain identical in the corresponding acyclic diene. As a result, the relative energy between tCBCO and the corresponding acyclic diene is reduced. As for the cis-cyclobutane-fused system, while torsional strain is also present in the acyclic diene, the dihedral angles are markedly different from those in the cyclic form, indicating that the structure of cyclobutane is distorted from the acyclic to the cyclic olefin. The reduction of ring strain by maintaining the eclipsing effect in the acyclic form is therefore offset by the distortion energy in the case of cis-cyclobutane-fused cyclooctene. The compensation between these two effects can be generalized to other cycloalkane-fused cyclooctenes: while the cycloalkanes (from three-to six-membered rings) reduce the ring strain by locking the conformation of the acyclic form so that the eclipsing effect is present (Supplementary Figs. 27  and 28), the fusion of the two rings (the cycloalkane and cyclooctene) can distort the cycloalkane, which increases the ring strain. As shown in Supplementary Fig. 29, the RSEs of the cycloalkane-fused cyclooctenes correlate well with the distortion of the dihedral angles C4-C5-C5′-C6 and C7-C6-C6′-C5 from the acyclic diene to cycloalkane-fused cyclooctene. The trans-cyclobutane therefore represents an ideal scenario where the torsional strain is maintained in both the cyclic and acyclic forms at a low energy cost.
Thermal and mechanical properties. The mild depolymerization conditions along with the excellent thermal stability (T d > 370 °C) of the tCBCO polymers make them appealing for use as sustainable materials. Since a wide range of T g values (from −30 °C to 100 °C) can be accessed through functionalization of the tCBCO polymers (Fig. 2b), both elastomers and plastics can be prepared.
Using the monomer M2b (T g of P2b is −31 °C) and a tCBCO crosslinker XL1 (for the structure and characterization, Supplementary Information), we prepared an elastomer PN1 and conducted a tensile test (Fig. 5a). The stress-strain curve showed a Young's modulus of 0.61 ± 0.10 MPa, a maximum strain of 238 ± 15% and a tensile strength of 0.61 ± 0.10 MPa. Previous work by Moore and coworkers has shown that metathesis-based depolymerization can occur in bulk materials, suggesting that the crosslinked material here should still be depolymerizable 39 . A depolymerization conversion of 94% was observed when PN1 was immersed in a solution of G2 in CDCl 3 and was heated to 50 °C for 2 h (Supplementary Fig. 30). The imide-functionalized tCBCO polymer P4 has a T g of 100 °C (Fig. 2b), which is comparable to that of polystyrene (one of the most widely used plastics). The large thermal processing window (which is above the T g but below the T d ) of P4 allowed the melt-pressing of films of the material. From the melt-pressed film of P4, we prepared a dog-bone specimen and conducted tensile testing to obtain the stress-strain curve (Fig. 5b). A polystyrene sample was processed in the same manner and was subjected to tensile testing for comparison. As shown in Fig. 5b, P4 showed a Young's modulus of 1.40 ± 0.06 GPa, which is comparable to that of polystyrene (1.56 ± 0.12 GPa). The strain at break (3.4 ± 0.3%) and the tensile strength (39 ± 5 MPa) of P4 were also comparable to those of polystyrene (strain at break of 3.7 ± 0.4% and a tensile strength of 46 ± 1 MPa).

Conclusions
In principle, the tCBCO monomers can be obtained by the photochemical [2 + 2] cycloaddition of 1,5-cyclooctadiene with a wide variety of olefins, therefore providing numerous possibilities of functionalities and material properties. For example, this process would enable the preparation of chemically recyclable polymers using biobased olefins 40 , therefore reducing the reliance on fossil feedstocks. Another example is the use of fluorinated olefins for preparing chemically recyclable fluorinated polymers, which are highly desirable but particularly rare. Moreover, the cis-alkene in the cyclooctene monomer can be isomerized into trans-alkene, which substantially increases the ring strain of the monomer, allowing for living polymerization. This feature (temporarily increasing the monomer ring strain to assist polymerization) makes the tCBCO system unique compared to other depolymerization systems 6 . The living ROMP of norbornene-derived monomers has enabled the synthesis of diverse polymer architectures and has fundamentally impacted many fields 41 ; however, the resulting polynorbornenes are not depolymerizable. The tCBCO system will allow for living ROMP that produces depolymerizable polymers. Furthermore, the concept of controlling ring strain using an additional ring-as well as the insights gained in this study on how the size and stereochemistry of the fused ring impact the ring strain-can be useful for many other fields where ring strain plays an important role, including mechanochemistry 42 , molecular machines 43 , catalysis 44 and click chemistry 45 . For example, in the strain-promoted azide-alkyne cycloaddition demonstrated by Bertozzi and coworkers, the reactivity is substantially increased by the ring strain in cyclooctyne 45 . The fused-ring strategy used here could be applied to further control the ring strain of cyclooctyne and the strain-promoted azide-alkyne cycloaddition reactivity.

Online content
Any methods, additional references, Nature Research reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/ s41557-021-00748-5.

methods
Calculations of RSEs. The RSEs of the cyclic olefins were calculated by following a method demonstrated by Grubbs and coworkers 22 . This method involves the calculation of the enthalpy change for the ring-closing metathesis (Fig. 1c).
Since the numbers and types of bonds of the reactant are identical to those of the products (that is, an isodesmic reaction), the enthalpy change of the reaction should come solely from the change in ring strain (equation (1))-which is essentially the ring strain of the cyclic olefin, as there is no ring strain in the acyclic diene reactant or the ethylene products.
H ethene , H cyclic olefin and H diene are the calculated ethalpies at 298.15 K and 1.0 atm for ethene, the cyclic olefin and the diene, respectively. All calculations were performed using Gaussian 16 at the B3LYP/6-31G(d,p) level of theory. The structures were first optimized with the Universal Force Field as implemented in the auto-optimization tool in Avogadro, followed by a conformer search with Avogadro, and the optimized conformation was subjected to geometry optimization and vibrational frequency calculations using Gaussian 16.
Thermodynamics studies of polymerization. Thermodynamics studies were performed for the diester monomer M2a and imide monomer M4. The polymerization for M2a was conducted at five different temperatures, and the polymerization for M4 was conducted at four different temperatures. The equilibrium monomer concentration [M] e for each temperature was measured via 1 H NMR. The thermodynamic parameters were obtained by plotting ln[M] e against temperature 1/T (van't Hoff plot) according to equation (2): In equation (2), [M] e is the equilibrium monomer concentration in moles per litre, T is the reaction temperature in Kelvin (K), ΔH is the enthalpy of reaction in kilocalories per mole, ΔS is the entropy of reaction in calories per mole per Kelvin and R is the gas constant, 8.314 J mol -1 K -1 . The values for ΔH and ΔS can be obtained from the slope and the intercept of the van't Hoff plot, respectively. The following procedure is a representative example of the polymerization performed here.
M2a (252.3 mg, 1.0 mmol, 1 equiv.) was dissolved in 0.5 ml xylene in a 1 ml volumetric flask. Additional xylene was added to reach the 1 ml mark to form the 1 M monomer solution, which was transferred to a 1 dram vial. To the monomer solution was added G2 (8.5 mg, 0.01 mmol, 0.01 equiv.); 200 μl of the solution was added to each of another four vials, and the five vials were placed in five oil baths that were preheated to the desired temperatures. Since time studies of the polymerization show that the polymerization would have reached equilibrium before 7 h (Supplementary Figs. [32][33][34], the reactions were quenched at 7 h with ethyl vinyl ether (100 μl) and the conversion was measured with 1 H NMR. The olefinic peaks in the 1 H NMR for the monomer and polymer were integrated to determine the residual monomer content at equilibrium (Supplementary Fig. 83).
Depolymerization studies. A representative depolymerization procedure is given as follows: in a vial equipped with a stir bar, P2a (146.5 mg, 0.58 mmol, 1 equiv.) was dissolved in CHCl 3 (5 ml). To the polymer solution was added G2 (4.93 mg, 0.0058 mmol, 0.01 equiv.) from a stock solution. The volume of the added G2 solution was precalculated to obtain a concentration of 25 mM ([olefin] = 25 mM).
The vial was placed in an oil bath preheated to 50 °C and stirred for 2 h. Then 0.5 ml ethyl vinyl ether was added to quench the reaction, and the mixture was stirred for 15 min. The vial was allowed to cool to room temperature, and 930 mg of Quadrapure TU macroporous particles and 1 ml CHCl 3 were added. The mixture was stirred overnight, after which it was filtered through a Celite plug. The filtrate was concentrated on a rotary evaporator and characterized using 1 H NMR and GPC. The initial molecular weight and concentration as well as the extent of conversion to monomer for each polymer are shown in Supplementary Table 1.
Mechanical testing. The tensile testing experiments were performed on five parallel specimens for both the elastomer (PN1) and the two plastics (polystyrene and P4). The values of mechanical properties are reported as average plus or minus standard deviation.
Tensile testing of the elastomer sample PN1 was performed on a custom-built tensile tester controlled by an Arduino UNO board. A crosshead speed of 5 mm min -1 was used. For the tensile testing of P4, a dichloromethane solution of the polymer was allowed to dry under ambient conditions followed by drying in a vacuum oven at 110-120 °C for 2 days, until no further mass change occured. Dog-bone-shaped samples of gauge length 10 × 3 × 1.5 mm 3 were prepared by compression moulding using a Carver Lab Press at zero pressure for 3 min at 130-150 °C, followed by 9,000 kg pressure for 5 min at the same temperature. Samples were then quenched in water and allowed to equilibrate for an hour at room temperature prior to examination of the mechanical properties. Dog-bone specimens made from Styron 663 polystyrene (acquired from Dow Chemical) were prepared by following a similar procedure for comparison. Mechanical properties were determined at a constant speed of 5 mm min -1 using an Instron 5969 tensile tester, which gave an initial crosshead speed V/L 0 of 0.5 min −1 , where V is the pulling speed and L 0 is the initial length.