Closing the “One Monomer–Two Polymers–One Monomer” Loop via Orthogonal (De)polymerization of a Lactone/Olefin Hybrid

■ INTRODUCTION

As part of large worldwide efforts [1][2][3][4][5][6][7][8][9][10][11][12][13][14] combating the worsening plastics problem that has already caused not only plastics pollution crisis but also tremendous energy and materials value loss in the economy, [15][16][17][18][19][20] the development of next-generation, chemically recyclable polymers [21][22][23][24][25][26][27][28][29][30][31] represents one of the approaches addressing these complex issues. [32][33][34] However, redesign of tomorrow's polymers with chemical recyclability or biodegradability must consider not only their closed-loop life cycles but also their performance properties. 35,36 The strategy starts from monomer design [36][37][38] the key to discovering new intrinsically circular polymers. The monomer is specifically designed to overcome typical trade-offs between monomer's polymerization/depolymerization thermodynamics and recyclability as well as material performance. 36 As this design is required to meet stringent thermodynamic, kinetic, and realworld performance requirements for ultimately developing circular plastics that exhibit not only full chemical recyclability but also high-performance properties, it still presents a formidable challenge.

With increasing demands for diverse polymer applications, effective methods have been developed to manipulate comonomer composition, topology, and functionality for targeted specific properties. 39 While well-established synthetic techniques exist for the facile and precision synthesis of conventional/nondegradable polymers, the same methodologies are rare in the realm of degradable/recyclable polymers where a higher level thermodynamic and chemoselection challenges persist. 40,41 Among several strategies, the ringopening polymerization (ROP) of functional cyclic monomers is an effective approach to synthesize well-defined degradable/ recyclable materials with excellent control over topology, tacticity, and sequence, and so on. [42][43][44] In this context, tremendous progress has recently been made in developing various classes of degradable/recyclable functional polymers 44,45 such as polyesters, [46][47][48][49][50][51][52][53] poly(ester-amide)s, 54 polythioesters, [55][56][57] and polycarbonates. [58][59][60] In spite of these advances, many synthetic approaches suffer from challenges arising from multistep routes for the synthesis of monomers 40 and the sensitivity of catalytic systems. 7 Additionally, the introduction of complexities, such as functional groups, often significantly decreases the recyclability of the polymer. With this in mind, the strategies that allow high levels of topological and functional group tolerance must be compatibilized with the strategies that allow recyclability.

Ring-opening metathesis polymerization (ROMP) [61][62][63][64][65] is a powerful method for preparing diverse poly(cyclic olefin)s. Strained cycloalkenes, such as norbornene and cyclooctene, are highly polymerizable because thermodynamics of their ringopening favor polymerization over depolymerization and thus have been frequently used as monomers for poly(cyclic olefin) synthesis. Most recently, Wang et al. redesigned a cyclooctene monomer by transfusing a cyclobutane ring and successfully rendered (de)polymerization reversibility of the otherwise nonreversible, highly strained parent cyclooctene. 66,67 While ROMP of low-strain cycloalkenes gives a similar recycling strategy to nonstrained lactones such as γ-butyrolactone (γ-BL) 47 wherein a polymerization-depolymerization equilibrium can be regulated by reaction temperature and monomer concentration, historically, low-strain cycloalkenes have remained largely underutilized. 68,69 Nonetheless, poly(cyclic olefin)s derived from cyclopentene derivatives [70][71][72][73][74][75] and a fivemembered cyclic enol ether (2,3-dihydrofuran) 76 are recyclable through ring-closing metathesis to re-form their low-strain starting monomers. Among the low-strain cycloalkenes, cyclohexene has essentially no ring strain and thus hardly undergoes ROMP due to a lack of thermodynamic driving force, [77][78][79][80] analogous to the ROP of γ-BL. While several alternative strategies have been developed, such as ringopening metathesis copolymerization of low to moderately strained cycloalkenes with highly strained monomers [81][82][83][84][85][86][87] and tandem ring-opening/ring-closing metathesis polymerization based on a selective cascade reaction between a terminal alkyne and a cyclohexene or cascade polymerization of bicycloalkene monomers, [88][89][90][91][92] the ROMP of cyclohexene still remains a challenge, and its potential as a circular poly(cyclic olefin) strategy to deliver high-performance materials with complete chemical recyclability has not been realized.

In our continued efforts toward developing chemically recyclable polymers, 36,47,48,50,56 we recently conceived a novel hybrid monomer design strategy that synergistically couples a high-ceiling temperature (HCT) substructure (such as εcaprolactone) for high polymerizability/performance properties with a low-ceiling-temperature (LCT) substructure (such as γ-BL) for high depolymerizability/recyclability within the same monomer structure. 50 The offspring bicyclic lactone (BiL) can be readily polymerized under ambient conditions to high-molecular-weight poly(BiL) (PBiL) materials that exhibit both high-performance properties and complete chemical recyclability at 120 °C in the presence of a catalyst. In this work, we designed a bicyclic lactone/olefin bifunctional monomer (BiL = ) comprising "nonpolymerizable", LCT substructures of γ-BL toward ROP and cyclohexene toward ROMP, which could potentially not only be polymerizable but also render orthogonal mechanistic pathways (i.e., ROP vs ROMP, Scheme 1). Excitingly, the new hybrid monomer BiL = can undergo orthogonal polymerization between ROP and ROMP. In other words, we can produce two totally different classes of polymeric materials from one single olefin/lactone monomer BiL = : polyester P(BiL = ) ROP via ROP and functionalized poly(cyclic olefin) P(BiL = ) ROMP via ROMP (Scheme 1). Importantly, both P(BiL = ) ROP and P(BiL = ) ROMP are chemically recyclable under mild conditions (25-40 °C). Furthermore, the intact functional group during orthogonal polymerization (i.e., the double bond in ROP and the lactone in ROMP) offers opportunity to further functionalize these two resulting polymers for tuning performance property variations. Several systems were reported wherein one monomer could undergo two different polymerization pathways to afford two different polymers 41 or one identical polymer; 93 however, those systems had to compromise between several trade-offs and did not achieve chemical recyclability for both polymers.

■ RESULTS AND DISCUSSION

There are four design elements in the olefin/lactone bifunctional monomer BiL = (6-oxabicyclo[3.2.1]oct-3-en-7- Journal of the American Chemical Society one), reflected on by the four unique built-in substructural units (Scheme 1): in the ROP manifold, it has both the sevenmembered lactone for high polymerizability and the fivemembered lactone for high depolymerizability, while in the ROMP manifold, it has the appropriately strained cyclohexene for balanced (de)polymerizability due to the bridging ester moiety in the monomer and in-chain five-membered lactone ring in the polymer. We hypothesized that by this design the hybrid BiL = monomer would exhibit orthogonal ROP and ROMP (de)polymerizability although both parents γ-BL and cyclohexene are notoriously "nonpolymerizable" or only polymerizable under harsh conditions in the corresponding ROP and ROMP method. The hybrid BiL = can be synthesized from 3-cyclohexene-1-carboxylic acid via a scalable (60 g), two-step reaction with a high overall isolated yield ∼94%. Worth noting here is that the carboxylic acid is readily obtainable from acrylic acid and 1,3-butadiene via the quantitative Diels-Alder reaction.

ROP Manifold. Both the widely practiced, commercially available organic catalyst 1,5,7-triazabicyclo[4.4.0]dec-5-ene (TBD) and metal-based catalyst La[N(SiMe 3 ) 2 ] 3 (La-1) were employed for initial screening of polymerization conditions. Specially, the TBD-catalyzed ROP in toluene (6.0 M) at room temperature (RT, ∼25 °C) with [BiL = ]/[TBD]/[BnOH] = 100/1/1 (BnOH = benzyl alcohol, as the initiator) achieved 98% conversion after 12 h, affording P(BiL = ) ROP with a low number-average molecular weight (M n ) of 8.8 kg mol -1 and a relatively broad dispersity (Đ) of 1.41 (Table S1, run 1). Further exploring the ROP with [BiL = ]/[TBD]/[BnOH] = 500/1/1 in toluene (6.0 M) at RT, the monomer conversion reached to 69% after 48 h, affording P(BiL = ) ROP with M n = 31.3 kg mol -1 and Đ = 1.33 (Table S1, run 2). The resulting unsaturated polyester P(BiL = ) ROP is an amorphous material, with a relatively high T g of 103 °C as a polyester. Switching to metal-catalyzed coordinative-insertion ROP, the La-1 catalyzed polymerization achieved 89% conversion after 8 h with [BiL = ]/ [La-1]/[BnOH] = 300/1/3 (RT, 6.0 M in toluene) (Table S1, run 3).

Surprisingly, 1 H NMR of the P(BiL = ) ROP produced by TBD revealed epimerization at the stereogenic carbon adjacent to the carbonyl carbon which took place during the ROP, thus affording the P(BiL = ) ROP containing both cis (69%) and trans (36%) stereoconfigurations (Figure S3A). The 1 H NMR spectrum exhibited two sets of peaks at both the alkoxy methine proton [-CHO-] region (labels c and c′ for cis-and trans-configurations, respectively) and the [-CHCH-] methine proton regions (labels a,b and a′,b′ for cis-and transconfigurations, respectively). In contrast, the P(BiL = ) ROP produced by the metal-catalyzed coordinative-insertion ROP completely retains the cis-configuration without noticeable epimerization, based on 1 H NMR analysis (Figure S3A vs Figure S3B). Specifically, the 1 H NMR spectrum of the P(BiL = ) ROP produced by La-1 displayed no peaks associated with the trans-configuration (i.e., a′, b′, and c′ labeled peaks Journal of the American Chemical Society attributed to epimerization, Figure S3A). Additionally, the resulting polymer obtained by La-1 had a narrow Đ of 1.17 as compared to the considerably higher Đ value of 1.41 for the TBD-derived P(BiL = ) ROP . These results are consistent with literature reports that strongly basic TBD often causes side reactions during polymerization, such as transesterification and epimerization. [94][95][96] Controlling the Topology of P(BiL = ) ROP . On the basis of our prior observations that La-1, when used alone (i.e., in the absence of an alcohol co-initiator), produces cyclic polyesters in the ROP of lactones, 47,48,50 we investigated the ROP of BiL = by La-1 alone, aiming to produce cyclic P(BiL = ) ROP . With a [BiL = ]/[La-1] ratio of 100/1 in toluene, 60% conversion was achieved at RT after 8 h (Table S1, run 4), affording P(BiL = ) ROP with M n = 20.1 kg mol -1 and Đ = 1.74. The targeted cyclic topology was confirmed by matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) and gel-permeation chromatography (GPC) with light scattering, refractive index, and viscosity triple detection in chloroform (Figure 1). Specifically, the MS of the P(BiL = ) ROP produced by La-1 without the BnOH initiation consisted of only one series of molecular ion peaks, with the identical spacings between the two neighboring molecular ion peaks being that of the exact molar mass of the repeat unit (m/z = 124.1), as shown by the slope (124.0) of the linear plot of m/z values (y) vs the number of BiL = repeat units (x). The intercept of 22.8 accounts for the mass of Na + , revealing no end groups for the cyclic P(BiL = ) ROP structure (Figure 1A,B). To provide further experimental evidence to distinguish between the linear and cyclic P(BiL = ) ROP topologies, a Mark-Houwink plot (i.e., double-logarithmic plots of intrinsic viscosity [η] vs absolute weight-average molecular weight (M w ) determined by light scattering detection) of the linear P(BiL = ) ROP and the cyclic P(BiL = ) ROP is presented in Figure 1C. As expected, the cyclic P(BiL = ) ROP exhibited a lower intrinsic viscosity than its linear analogue, consistent with the theoretically predicted value for cyclic polymers. When the M w 's of these linear and cyclic analogues are compared per elution volume as functions of hydrodynamic radius, the cyclic P(BiL = ) ROP has higher M w than its linear analogue, which revealed the cyclic P(BiL = ) ROP has a smaller hydrodynamic volume (Figure 1D).

Controlling the Stereomicrostructure of P(BiL = ) ROP . The stereoretention afforded by La-catalyzed ROP provides the opportunity to produce perfectly isotactic P(BiL = ) ROP starting from chiral monomer (1S,5S)-6-oxabicyclo[3. As anticipated, the resulting polymer is purely isotactic, displaying no observable stereoerrors on its 13 C NMR (P m = 1.00, Figure S13). This chiral polymer exhibited a very high T m of 239 °C on the differentiation scanning calorimetry (DSC) second heating scan with a crystallization temperature of 212 °C (Figure S14). When rac-BiL = was polymerized by achiral catalyst La-1, only iso-enriched P(BiL = ) ROP was produced with a low tacticity of P r = 0.35 (Figure S9). The fact that the stereocenters contained in the monomer are in a cis-configuration, which remains unperturbed by the metal-catalyzed ROP, implies that isotactic (R,R-R,R and S,S-S,S) and heterotactic (R,R-S,S) sequences are the only possible products from the rac-BiL = ROP because the syndiotactic sequence (R,S-R,S) is only obtainable by epimerization. Thus, to improve the stereoselectivity of the ROP of rac-BiL = , we next employed discrete yttrium silylamido {N(SiHMe 2 ) 2 } complexes supported by C 2symmetric N,N′-bis(salicylidene)cyclohexanediimine (salcy) ligands since such catalysts have been shown to mediate sitecontrolled highly isoselective polymerization of eight-membered rac-diolides by enantiomorphic-site control. 97,98 However, the use of [Y]{N(SiHMe 2 ) 2 } complexes supported by 3,5-CMe 3 -substituted salcy and 3-CPh 3 -5-Me-substituted salcy ligands only afforded an amorphous material with low tacticity (P r = 0.35-0.45) (Table S1, runs 5-7, and Figures S16-S18).

Subsequently, more sterically encumbered discrete yttrium alkyl complexes supported by tetradentate amino-bisphenolate ligands 99,49 with β-OMe (Y-1) or β-NMe 2 (Y-2) side-arm Journal of the American Chemical Society donor (Figure 2D) were employed for the ROP of rac-BiL = to induce chain-end controlled heteroselectivity. At the outset, tert-butyl-substituted complex Y-1(a) in toluene also afforded an atactic polymer (P r = 0.35) (Table S2, run 1). Tuning the reaction conditions by using dichloromethane (DCM) instead of toluene further enhanced the P r value to 0.49 (Table S2, run 2), which was assessed by 13 C NMR spectra (Figures S19 andS20). Further increasing the steric hindrance of the bisphenolate ligand to ortho-cumyl [Y-1(b)] in toluene and DCM still afforded atactic polymers but with increased P r value to 0.59 and 0.74 (Table S2, runs 3 and 4; Figures S21 andS22), respectively, which indicated that DCM is more effective for producing high heterotacticity. We also conducted the ROP without solvent, which gave a P r value of 0.66 (Table S2, run 5, and Figure S23), midway between toluene (P r = 0.59) and DCM (P r = 0.74). Thus, the solvent has a significant impact on the chain end selectivity of rac-BiL = . ROP of rac-BiL = at -30 °C in toluene by Y-1(b) gave a decreased P r of 0.45 (Table S2, run 6, and Figure S24).

Manipulating the catalyst ligand side arm (the pendant donor group) has been shown to be an effective strategy for the design of organometallic catalysts for asymmetric organic reactions. 49,100 Indeed, changing the side-arm donor from β-OMe in Y-1(b) to β-NMe 2 generated Y-2(b) which drastically increased selectivity and produced heterotactic-enriched P-(BiL = ) ROP (P r = 0.71 in toluene or P r = 0.78 in DCM; Table S2, runs 9 and 10; Figures S27 andS28). Excitingly, with the increased heterotacticity, the P(BiL = ) ROP material derived from Y-2(b) in DCM finally became semicrystalline, as evidenced by their endothermic first-order melting transition with a high T m value of 145 °C as shown in Figure 2B.

With the above informative observations, we envisioned that complex Y-2(c) with even bulkier -CMePh 2 substituents would be more selective to realize highly heteroselective ROP, Indeed, when the ROP of rac-BiL = was catalyzed by Y-2(c) at [rac-BiL = ]/[Y] ratio of 100/1, 90% conversion was achieved after 12 h, affording semicrystalline P(BiL = ) ROP with M n = 16.0 kg mol -1 , Đ = 1.09, and P r = 0.93 (Table S2, run 11, and Figure S29). The heterotactic microstructure of the resulting P(BiL = ) ROP can be confirmed by comparing with the purely isotactic microstructure obtained from the stereoretention ROP of enantiopure monomer (S,S)-BiL = (Figures 2A(1) and 2C). Intriguingly, the heterotactic-enriched P(BiL = ) ROP exhibited two melting temperatures (T m1 = 141 °C, T m2 = 195 °C) on the DSC first heating scan (Figure 2B). When precatalyst Y-2(c) was combined with 1 equiv of BnOH initiator for a more controlled polymerization, the heterotacticity of the resulting P(BiL = ) ROP decreased slightly to 0.91 (Table S2, run 12, and Figure S30). However, the BnOH initiation system is well-controlled and follows strictly firstorder kinetics (R 2 = 0.998) up to high conversions with molecular weight increasing linearly with monomer conversion while maintaining low dispersity (Đ < 1.1) (Figure S32).

ROMP Manifold. After realizing the topologically controlled and stereoselective ROP of BiL = , we subsequently turned our attention to the ROMP pathway of BiL = . In contrast to the ROP product P(BiL = ) ROP , which is a polyester with cyclohexene incorporated in the repeating unit, the ROMP product P(BiL = ) ROMP is a poly(cyclic olefin) with C-C and CC bonds in the main-chain backbone (Scheme 1).

Several Grubbs catalysts were screened for the ROMP of BiL = . When ROMP of BiL = was performed with the Grubbs third-generation catalyst (G3) (Table S3), the polymerization proceeded rapidly at RT with varied monomer concentrations (1.9 to 5.0 M) and [BiL = ]/[G3] feed ratios (1000/1 to 40000/ 1). However, the measured M n values of the resulting poly(cyclic olefin)s deviated significantly from the theoretical values. Next, the Grubbs second-generation catalyst (G2) was used (Table S4), which afforded high-molecular-weight P(BiL = ) ROMP (M n up to 340.2 kg mol -1 , Đ ∼ 1.3). Specifically, the ROMP of BiL = performed in DCM (5.0 M) at RT with [BiL = ]/[G2] = 10000/1 gelled after 1 h and achieved 80% monomer conversion after 24 h. The resulting P(BiL = ) ROMP had a higher M n of 58.7 kg mol -1 than the P(BiL = ) ROMP produced from G3, and the dispersity was relatively narrow (Đ = 1.23) (Table S4, run 1). Increasing the [BiL = ]/[G2] ratio to 20000/1 gave a higher M n (89.6 kg mol -1 , Đ = 1.29), while the monomer conversion decreased to 51% (Table S4, run 2). Further lowering the catalyst loading to 25 ppm ([BiL = ]/[G2] = 40000/1), the monomer conversion reached only 45% after 24 h, affording a higher molecular weight P(BiL = ) ROMP with M n = 96.6 kg mol -1 and Đ = 1.26 (Table S4, run 3). These results indicated that the G2 catalyst exhibited better control for the ROMP of BiL = , and the catalyst loading can be as low as 25 ppm. Heterogeneous polymerizations of BiL = were performed in toluene and solvent free conditions to yield higher molecular weight P(BiL = ) ROMP with M n up to 340 kg mol -1 and unimodal distribution (Table S4, Figures S37 andS38).

The ROMP of enantiopure (S,S)-BiL = exhibited similar polymerization behavior as rac-BiL = . Specifically, the P[(S,S)-BiL = ] ROMP produced in DCM had a medium M n of 23.1 kg mol -1 (Table S4, run 16), while ROMP of (S,S)-BiL = in toluene afforded a high M n of 276 kg mol -1 with a relatively narrow dispersity of 1.28 (Table S4, run 17). The structural differences can be observed by comparing the 13 C NMR of the P(BiL = ) ROMP and P[(S,S)-BiL = ] ROMP . The peak splitting of P[(S,S)-BiL = ] ROMP is clearer than that of P(BiL = ) ROMP ; however, the connection between two adjacent units is not regioselective, with both head-to-tail and head-to-head connections and cis/trans alkene configurations possible (Figures S34 andS36). As a result, the P[(S,S)-BiL = ] ROMP derived from the enantiopure monomer is an amorphous material displaying a similar T g of 111 °C to that of the P(BiL = ) ROMP (Figures S39 andS40).

Chemical Recyclability of P(BiL = ) ROMP and P(BiL = ) ROP . To quantify the polymerizability of BiL = and depolymerizability of P(BiL = ) ROMP as a function of reaction conditions, the ROMP thermodynamics of the BiL = polymerization was probed by using [BiL = ]/[G3] = 100/1 and [BiL = ] 0 = 0.5 mol L -1 in CD 2 Cl 2 via a variable-temperature NMR study. The equilibrium monomer concentration, [BiL = ] eq(ROMP) , obtained by plotting [BiL = ] t as a function of time until [BiL = ] became constant, was measured to be 0.380, 0.285, 0.245, and 0.215 mol L -1 for 25, 15, 10, and 5 °C, respectively. The van't Hoff plot of ln[BiL = ] eq vs 1/T gave a straight line (R 2 = 0.998, Figures S41 andS42), from which the thermodynamic parameters were calculated to be ΔH (Figures S43 andS44). These results indicate both good polymerizability of BiL = and medium depolymerizability of the functionalized poly(cyclic olefin) P(BiL = ) ROMP and polyester P(BiL = ) ROP .

Indeed, the ROMP and ROP results summarized in Table S1-S4 are consistent with the above-derived thermodynamic parameters. Next, we examined the depolymerizability of P(BiL = ) ROMP and P(BiL = ) ROP . We first probed the depolymerizability of the preformed P(BiL = ) ROMP (M n = 96.6 kg mol -1 , Đ = 1.26) using G2 at RT under dilute conditions (10 mg/mL), which is limited by the solubility of P(BiL = ) ROMP . The depolymerization of P(BiL = ) ROMP in CD 2 Cl 2 was monitored in situ by 1 H NMR, revealing that the depolymerization was almost complete after 24 h to cleanly regenerate the monomer BiL = (Figure S45). This depolymerization was also monitored in situ by GPC; after stirring with 0.2 mol % G2 catalyst at RT for 10 min, the trace of the initial unimodal P(BiL = ) ROMP turned into a broad peak with lower molecular weight fractions appearing, at which point only 2% BiL = was regenerated (by 1 H NMR integration), indicating a random chain scission process in the depolymerization of P(BiL = ) ROMP catalyzed by G2. After 24 h, the P(BiL = ) ROMP was nearly completely depolymerized with the disappearance of the polymer peak and increase of the monomer peak (Figure 3D). Increasing the depolymerization temperature and catalyst loading significantly shortened the depolymerization time (Figure S46). In addition, we also conducted the P(BiL = ) ROMP depolymerization on the 28 g scale, wherein the depolymerization was performed with 0.1 mol % G2 catalyst in DCM; after refluxing for 24 h in open air, the catalyst was removed by flash column chromatography, and the pure BiL = monomer was recovered in 93% yield. These results demonstrate the excellent recyclability of P(BiL = ) ROMP .

With the success of chemical recycling P(BiL = ) ROMP under mild conditions, we then set out to determine whether the resulting polyester P(BiL = ) ROP could also be recycled. Journal of the American Chemical Society = -55.8 J mol -1 K -1 , T c = 106 °C), which corresponds to the higher ring strain due to the introduction of the CC double bond in the monomer structure. 50 Thus, we first tried the exactly the same conditions as those employed for the depolymerization of the saturated analogue. However, the depolymerization of P(BiL = ) ROP actually proved to be difficult, as P(BiL = ) ROP contains a reactive site CC bond that induces side reactions during the high-temperature depolymerization conditions. Subsequently, we explored the depolymerization of P-(BiL = ) ROP catalyzed by ZnCl 2 in toluene-d 8 (10 mg/mL) at 120 °C but observed only around 20% conversion back to monomer due to side reactions that decomposed the majority (∼80%) of P(BiL = ) ROP after 24 h. When we turned to TBD under the same conditions, an increase in monomer recovery up to 40% after 24 h was achieved, but still the same decomposition was observed. Thus, side reactions outcompete the depolymerization at high temperatures, but such conditions are necessary for most polyesters to overcome the depolymerization energy barrier. 48,50 To suppress or avoid the temperature-induced side reactions, we turned our attention to low-temperature depolymerization aided by a Lewis acidic catalyst. Excitingly, depolymerization of P(BiL = ) ROP with ZnCl 2 (∼50 mg) in DCM at 40 °C regenerated 94% monomer after 12 h (Figure S47). Even an epimerized mixture of cis-and trans-P(BiL = ) ROP materials produced from TBD was completely depolymerized with slightly lower selectivity (∼87%, Figure S48). Fascinatingly, using the exact same condition to depolymerize the saturated analogue (PBiL) without the double bond, which has an even lower T c value, 50 we observed no depolymerization after 24 h.

To gain further insight into the mechanism of ZnCl 2catalyzed depolymerization of P(BiL = ) ROP , we ascertained two working hypotheses. First, ZnCl 2 might actually react with P(BiL = ) ROP to form a specific active depolymerization complex.

Alternatively, simple interactions between ZnCl 2 and P-(BiL = ) ROP alkene/carbonyl might augment the conformation of the P(BiL = ) ROP so that the energy barrier for ring-closing depolymerization is lowered. To this end, we employed ZnCl 2 to depolymerize a mixture of P(BiL = ) ROP and PBiL under the same conditions. If a specific zinc complex formed, it should also be active for the depolymerization of lower ring-strain analogue PBiL. However, only P(BiL = ) ROP was selectively depolymerized while PBiL was left untouched (Figure S49), which indicates the hypothesis of an active catalyst formed in situ is less probable. Furthermore, additional zinc compounds were tested, such as zinc acetate, zinc acetylacetonate, and [(βdiketiminate)Zn(TMS) 2 ], but none of these compounds catalyzed the depolymerization of P(BiL = ) ROP . Subsequently, we explored the ZnCl 2 particle size effect on P(BiL = ) ROP depolymerization, with the as-received crystalline ZnCl 2 having a diameter of ∼200 μm, as estimated by scanning electron microscopy (SEM), and the grinded powder ZnCl 2 of ∼100 μm (Figure S50). Depolymerization kinetics with the same amount (20 mg) of crystalline and powdery ZnCl 2 were conducted. The powdery ZnCl 2 showed higher activity, achieving 72% monomer recovery in 12 h, while the crystalline ZnCl 2 only afforded 15% monomer recovery in 12 h (Figure 3A). These results indicated the higher the surface area of ZnCl 2 , the higher the depolymerization rate is. As additional corroborative evidence, the ZnCl 2 loading also showed a significant effect on P(BiL = ) ROP depolymerization: the depolymerization rate increased with the increasing of the powdery ZnCl 2 loading, and the kinetics data showed the monomer recovery yield was increased linearly with the time (Figure 3B). These results indicated the depolymerization follows zero-order kinetics and is consistent with the interfacial catalytic mechanism. Furthermore, in contrast to the depolymerization of P(BiL = ) ROMP , in situ monitoring of the depolymerization of P(BiL = ) ROP (M n = 24.7 kg mol -1 , Đ = Journal of the American Chemical Society 1.15) by GPC indicated that the depolymerization follows an unzipping mechanism rather than a random chain scission process (Figure 3C). Lastly, changing the solvent from CD 2 Cl 2 to a coordinating solvent (THF), which would compete with the P(BiL = ) ROP alkene and carbonyl for interaction with ZnCl 2 , gave no depolymerization of P(BiL = ) ROP .

The success of achieving the monomer recovery of both P(BiL = ) ROMP and P(BiL = ) ROP further inspired us to pursue the orthogonality on the depolymerization of P(BiL = ) ROMP and P(BiL = ) ROP . First, a 1/1 mixture of P(BiL = ) ROMP and P(BiL = ) ROP (a physical blend) was dissolved in CD 2 Cl 2 with 1 wt % G2 catalyst added, and the ring-closing metathesis depolymerization of P(BiL = ) ROMP went smoothly and selectively, without touching P(BiL = ) ROP . Then, ZnCl 2 was added to start the P(BiL = ) ROP depolymerization once the P(BiL = ) ROMP was completely depolymerized into the monomer BiL = . The depolymerization results showed excellent orthogonality on the selective depolymerization of P(BiL = ) ROMP and P(BiL = ) ROP , depending on the catalyst added (Figure 3E and Figure S51). The same strategy also successfully applied to the depolymerization of the copolymer of P(BiL = ) ROMP and P(BiL = ) ROP (Figure 3E and Figure S52).

Thermomechanical Properties of P(BiL = ) ROMP and P(BiL = ) ROP . The differences in material properties between P(BiL = ) ROMP and P(BiL = ) ROP were studied next. The thermomechanical properties of P(BiL = ) ROP prepared with TBD (M n = 31.3 kg mol -1 , Đ = 1.33) and P(BiL = ) ROMP prepared with G2 (M n = 334 kg mol -1 , Đ = 1.31) were examined by dynamic mechanical analysis (DMA) in a tension film mode. The thermomechanical spectra of P(BiL = ) ROP (Figure 4A) and P(BiL = ) ROMP (Figure 4B) show that at room temperature (the glassy state) both P(BiL = ) ROP and P(BiL = ) ROMP exhibit high storage modulus (E′) values, although E′ (2.16 ± 0.29 GPa) of P(BiL = ) ROP was higher than that (1.65 ± 0.13 GPa) of P(BiL = ) ROMP .

However, after the glass transition region [T g ∼ 74-90 °C, as defined by the peak maxima of tan δ, the loss modulus/ storage modulus ratio (E″/E′)] (Figure 4A,B), E′ of P(BiL = ) ROP dropped and then quickly went to the viscous flow state. In contrast, E′ of P(BiL = ) ROMP also dropped significantly after the glass transition region but did not completely lose the strength and went to the rubbery plateau, suggesting that the P(BiL = ) ROMP was cross-linked during the thermomechanical testing. Mechanical properties of P-(BiL = ) ROMP are discussed in the following section where direct comparisons are made between the materials before and after postfunctionalization.

Thermal property differences can be found in their thermal data. P(BiL = ) ROMP is an amorphous material but exhibits a high T g of 113 °C by DSC (Figure S39), which can be ascribed to the incorporation of the five-membered γ-BL ring into the main chain. The P(BiL = ) ROP also showed a high T g between 103 and 115 °C, depending on the microstructure and molecular weight (Figure S7 and Figure 2B). Noteworthy also is the perfectly isotactic P(BiL = ) ROP and heterotactic-enriched P(BiL = ) ROP display high T m values between 141 and 239 °C, depending on the tacticity. P(BiL = ) ROMP displays a high decomposition temperature (T d,5% , defined by the temperature at 5% weight loss) of 361 °C and a high maximum rate decomposition temperature (T max ) of 459 °C, as measured by thermogravimetric analysis (TGA) and derivative thermogravimetric analysis (DTG) (Figure 4E). This high T d value is consistent with the character of poly(cyclic olefin)s. In comparison, as a polyester, P(BiL = ) ROP exhibits a much lower T d,5% of 278 °C (Figure 4D).

Postfunctionalization of P(BiL = ) ROP and P(BiL = ) ROMP . The built-in functionalities of P(BiL = ) ROP (the CC bond) and P(BiL = ) ROMP (the lactone) allow for their respective postfunctionalization. As expected, Fourier transform infrared (FTIR) spectra of P(BiL = ) ROP and P(BiL = ) ROMP showed the disubstituted CC stretching for both polymers. The νCC for P(BiL = ) ROP is at 1653 cm -1 , where the disubstituted CC is in the cyclohexene form (cis-), while the CC stretching frequency for P(BiL = ) ROMP is 22 cm -1 higher at 1675 cm -1 , which indicates that the CC bond in the main chain of P(BiL = ) ROMP mainly adopts the trans-configuration. Meanwhile, the red-shift of the CO stretching frequency (νC O) for P(BiL = ) ROP (1723 cm -1 ) to a wavenumber of 37 cm -1 lower than that for P(BiL = ) ROMP is also consistent with structural differences between the ester carbonyl in P(BiL = ) ROP and the lactone carbonyl in P(BiL = ) ROMP (1760 cm -1 , Figure 4C).

Through the thiol-ene click reaction, we grafted both linear and cyclic P(BiL = ) ROP with 1-octadecanethiol to give the corresponding brush polyesters. This postfunctionalization made it possible for the direct observation of their respective linear and cyclic topology (Figure 5A) by high-resolution transmission electron microscopy (TEM). On the other hand, postfunctionalization of P(BiL = ) ROMP utilized its built-in γ-BL ring in the main chain, which was readily hydrolyzed by NaOH to afford the carboxylate ionomer (Figure 4C, green line, and Figure 5B). This ionomer is soluble in water and stable in open air for over six months. Mechanically, P(BiL = ) ROMP is best descried as a hard, strong, and tough material, as characterized by a high Young's modulus (E) of 3.8 ± 0.6 GPa, a high ultimate tensile strength (σ B ) of 40.0 ± 10.9 MPa, and a modest elongation at break (ε B ) of 76 ± 9%, attributable to the in-chain incorporated γ-BL ring. In comparison, while the ionomer produced from the hydrolysis of P(BiL = ) ROMP exhibited a lower E (0.71 ± 0.15 GPa) and σ B (18.5 ± 3.2 MPa), it is much more flexible with a high ε B value of 445 ± 91% (Figure 4F). More excitingly, once neutralized with HCl, the carboxylate ionomer collapses back to the stable γ-BL to regenerate P(BiL = ) ROMP (due to the solubility limitation, once ∼50% γ-BL was regenerated it precipitated out), demonstrating an intriguing reversible modification strategy (Figure 5B).