Polymers from sugars and CS 2 : ring opening copolymerisation of a D -xylose anhydrosugar oxetane †

A D -xylose anhydro sugar derivative ( 1 ) has been applied in the ring-opening copolymerisation (ROCOP) with CS 2 to form a polythiocarbonate (poly( CS 2 - co - 1 )) with high head-head/tail-tail regioselectivity towards alternating thiono- and trithiocarbonate linkages (up to 95%). Through variation of the reaction parameters ( e.g. temperature and CS 2 stoichiometry), some control over the regioselectivity (head-head/ tail-tail linkages 57 – 95%) and the nature of the polymer linkages is possible. Conditions can also be tailored to enable the facile isolation of a polymerisable cyclic xanthate, 2 . Kinetic experiments suggest that across the range of temperatures studied, the formation of poly( CS 2 - co - 1 ) proceeds at least partially by direct copolymerisation of 1 and CS 2 , without necessarily going through the ring-opening polymerisation (ROP) of 2 . Poly( CS 2 - co - 1 ) exhibits partial chemical recyclability into cyclic monomer 2 (up to 45% after 20 h at 110 °C with [poly( CS 2 - co - 1 )] 0 = 1.34 mol L − L ). Finally, rapid degradation (<1 h) of poly( CS 2 - co - 1 ) is possible under UV radiation ( λ = 365 nm) and is accelerated in the presence of tris(trimethylsilyl)silane (TTMSS).


Introduction
Owing to the environmental issues surrounding plastic use, a societal shift away from traditional petroleum-derived feedstocks is driving innovation in the field of sustainable polymer chemistry. Polymers derived from sugars have huge potential as sustainable alternatives to current commodity plastics. [1][2][3][4][5] For instance, polymers which incorporate pyranose or furanose motifs typically exhibit remarkably high glass transition temperatures (T g ) and the availability of hydroxyl groups of sugar derivatives significantly broadens the scope for prospective material functionalisation. [6][7][8][9] It is known that the replacement of oxygen atoms with sulfur ones in oxygenated polymers can give the resulting polythioethers, polythiocarbonates and polythioesters enhanced physical and thermal properties, [10][11][12][13][14][15][16] as well as additional advanced optical 17 and electrical characteristics. 18 Moreover, sulfur-containing polymers exhibit biocompatibility and have shown potential in metal and bacterial adhesion. 19,20 The presence of sulfur atoms in the backbone of polymers may also accelerate degradation under appropriate conditions (e.g. under UV radiation). 19,21 The synthesis of polythiocarbonates is possible through polycondensation, [22][23][24][25] the ring-opening polymerisation (ROP) of cyclic thiocarbonates, 26,27 polyalkylation of trithiocarbonates 28,29 or the ring-opening copolymerisation (ROCOP) of CS 2 15,30-33 and COS 11,[34][35][36][37][38] with cyclic ethers (epoxides and oxetane). ROCOP methods are particularly interesting given the array of synthetic possibilities provided by the pool of cyclic ethers usable, coupled with the polymerisation control brought by existing ROCOP catalysts. 39,40 To date, both heterogeneous (e.g. Zn-Co(III) doublemetal cyanide complexes) 41 and homogeneous catalysts/ initiators (e.g. metal salen/onium catalysts 30,34,42 and LiO t Bu 31 ) have been reported for CS 2 /cyclic ether ROCOP. However, reports of ROCOP between an oxetane and CS 2 remain rare. 32 CS 2 is manufactured by reaction of charcoal or natural gas with sulfur, and despite its known toxicity, incorporating it into polymers has some benefits in terms of waste valorisation. Indeed, sulfur is an abundant by-product of the oil and chemical industry. [43][44][45] Our group has previously used CS 2 to synthesise a series of cyclic thiocarbonates for ROP towards sugar-derived polythiocarbonates. 27 However, the synthesis of these monomers was challenging, necessitating extensive purification (e.g. successive column chromatography) to be suitable for ROP techniques. For xylose-derived xanthate monomer, 2, ROP towards poly(2) formed a regioregular polymer with up to 87% head-head/tail-tail (HH/TT), trithiocarbonate/thionocarbonate linkages, although no control over the polymer microstructure was demonstrated.
It was envisioned that the ROCOP of CS 2 with oxetane-functionalised xylofuranose derivative, 1, may expediate the polymer synthesis given its ease of preparation in purities suit-able for ROP 46 and ROCOP. 47 Moreover, we hypothesised that alternative polymer sequences may be accessible through ROCOP. Lastly, through judicious choice of conditions, it was hoped that the reaction between CS 2 and 1 could be tailored towards cycloaddition over copolymerisation, to access 2 with fewer purification steps.
We have previously used 1,2-cyclohexanediamino-N,N′-bis (3,5-di-t-butylsalicyilidene)-chromium(III), CrSalen and bis(triphenylphosphine)iminium chloride (PPNCl) to catalyse the ROCOP of cyclic anhydrides and 1 for the preparation of polyesters. 47 CrSalen/onium salts (PPNN 3 ) binary catalytic system has also been applied in the ROCOP of oxetane with CS 2 . Detailed study revealed the presence of multiple polymer linkages indicative of sulfur/oxygen scrambling which became more prevalent at higher temperatures. 32 Werner and coworkers have reported on the ROCOP of terminal epoxides with CS 2 initiated by Li t OBu. 31 The polythiocarbonates were found to be highly regioregular with up to 94% head-head/tailtail alternating thionocarbonate/trithiocarbonate linkages.
Herein we describe the ROCOP of 1 with CS 2 to form sugarbased sulfur-containing polymers with very high regioselectivity towards HH/TT, trithiocarbonate/thionocarbonate linkages (up to 95%), marking an improvement on the similar polymer synthesised through the ROP of 2. The head/tail configuration of poly(CS 2 -co-1) can be varied through modulation of the reaction temperature and CS 2 stoichiometry. Mechanistic investigations show that ROCOP proceeds at least partially directly. Lastly, poly(CS 2 -co-1) exhibits partial chemical recyclability into 2, as well as full degradation under UV light.

Results and discussion
Poly(CS 2 -co-1) made by ROCOP: identification of polymer linkages Oxetane 1 was synthesised in three steps from D-xylose in accordance with previous reports (51% overall yield). 46 [46][47][48][49] the reaction proceeded readily at relatively mild temperatures, with 33% conversion of 1 after 4.5 h at 80°C to yield poly(CS 2 -co-1) of 2800 g mol −1 (Đ M = 1.54; measured by size-exclusion chromatography (SEC)). The formation of resonances at 192.9 ppm (S1) and 222.4 ppm (S2) in the 13 C{ 1 H} NMR spectrum and 5.52 ppm and 5.80-6.03 ppm in the 1 H NMR spectrum suggested the incorporation of CS 2 within the polymer backbone, later confirmed by 1 H-13 C{ 1 H} HSQC and HMBC experiments (Scheme 1, Fig. 1 and 2). 1 H and 1 H-13 C { 1 H} HMBC NMR spectra indicated that the ROCOP of CS 2 and 1 was ring-selective, with CS 2 incorporation across the oxetane moiety only. Furthermore, the 1 H-13 C{ 1 H} HMBC NMR spectrum of isolated and crude poly(CS 2 -co-1) revealed correlations between carbon resonance S1 and proton environment c (i.e. the CH in position 3 on the xylofuranose core), and between carbon resonance S2 and proton environment e (i.e. the CH 2 in position 6 on the xylofuranose core), consistent with HH/TT configuration and alternating thiono-and trithiocarbonate linkages. Overall, the NMR spectroscopic data for poly(CS 2 -co-1) was similar to that of the regioregular polymer obtained previously by ROP of 2 (Table 2, entry 1). 27 Minor resonances associated with the e environment were also observed between δ = 2.65-3.13 ppm and δ = Scheme 1 Possible products and linkages formed during the ROCOP of 1 and CS 2 . 3.42-3.56 ppm, suggesting the occurrence of other polymer linkages (Fig. 2). While only resonances S1 and S2 were observed in the thiocarbonyl region of the 13 C{ 1 H} NMR spectrum of isolated poly(CS 2 -co-1) (Fig. 1e), a weak correlation between a xanthate-like (δ = 212.8 ppm) and an e (δ = 3.42-3.56 ppm) proton environment was detected by 1 H-13 C { 1 H} NMR HMBC spectroscopy (Fig. 2c), revealing the presence of xanthate polymer linkages (e xanthate ). Conversely, no 1 H-13 C { 1 H} NMR HMBC correlation was observed for the e proton environment detected at δ = 2.65-3.13 ppm. Comparison with literature data 32 revealed that the 1 H NMR and 13 C{ 1 H} NMR shift values of this resonance was comparable to known polythioethers (Fig. 2b), 36 corroborating the absence of a nearby quaternary carbon. This resonance was therefore attributed to a thioether polymer linkage (e thioether ). The complexity of the signals seen may stem from random position of the thioether linkages in the polymer sequence. Regardless of the origins of those various linkages, the e environments in the 1 H NMR spectra were thus identified as an easy way to assess the selectivity of the ROCOP process.
Encouraged by this preliminary data, further catalytic ROCOP experiments were performed to study the effect of concentration, temperature, CS 2 stoichiometry and nature of the catalyst and of co-catalyst on the products of the ROCOP reaction ( Table 1).

Impact of temperature on polymer linkages
Polymerisations were carried out at 25°C, 60°C, 80°C, 100°C and 140°C at [1] [1][2][3][4][5]. CrSalen was found to be active for 1/CS 2 ROCOP at all temperatures tested, including at 25°C (Table 1, entry 1). The room temperature reactivity of the CS 2 /1 ROCOP contrasts strongly with analogous reactions with cyclic anhydrides which necessitate high temperatures and long reaction times (100°C, 2-6 days) to give similar oxetane conversions. Overall, the reactivity of 1 with CS 2 seems comparable with that of cyclohexene oxide, 30 cyclopentene oxide 42 and oxetane, 32 although direct comparisons of TOFs are not appropriate due to substantial differences in the reaction conditions used. The enhanced activity of CrSalen in the ROCOP of CS 2 /1 as compared with cyclic anhydrides may be a result of two effects. Firstly, CS 2 is a stronger electrophile than cyclic anhydrides owing to the lower enthalpy of the CvS bond vs. the CvO bond (573 and 799 kJ mol −1 , respectively). 50,51 Secondly, following CS 2 insertion, the resulting thiocarbonate species are more nucleophilic than a carboxylate and the Cr-S bond is likely less strong than a Cr-O bond, which promotes propagation.
Within the limits of 1 H NMR spectroscopy, in all the experiments performed, all resonances seen could be assigned to poly(CS 2 -co-1) (as described above), cyclic xanthate 2 or oxetane monomer 1 (Fig. 1). However, varying the reaction temperature caused significant divergences in product distributions. Increasing the temperature resulted in a loss of regioselectivity (  Table 1, entries 4 and 5) with an increase in thioether linkages detected (16% and 31%, respectively). The increased occurrence of thioether linkages at elevated temperatures likely arises from the elimination of COS from propagation intermediates, favoured by entropy (Scheme S2 †) and as noted in previous studies. 30,32,42 The ratio of xanthate links also increased with temperature, but to a lesser extent than for the thioethers (6% at 25°C up to 12% at 140°C, Table 1 entries 1-5).The formation of xanthate linkages may be a result of several mechanisms: the 'expected' CS 2 /1 ROCOP, but

Polymer Chemistry
Paper also various S/O rearrangements occurring during ROCOP, 27 including transthiocarbonation between polymer chains, as well as the regioregular ROP of 2 (Schemes S1 and S2 †). At 80°C, quantitative conversion of 1 was observed after 20 h, yielding the maximum molar mass of polymer achieved under the conditions tested (15 000 g mol −1 , Đ M = 2.13, Table 1, entry 3). More generally, obtained M n,SEC values were far from M n,theo , likely due to residual protic impurities (e.g. 1,2-O-isopropylidene-D-xylofuranose) in the monomer samples which may act as chain transfer agents to decrease the molar mass of the polymers. This is typical within the field of cyclic ether/CS 2 ROCOP. 30,32,42 Further purification of 1, including via successive or reactive distillation with NaH/MeI, 52 unfortunately did not increase the polymer molar masses. While analysis of the polymer by MALDI ToF mass spectrometry proved unsuccessful, end-group titration by 31 P NMR spectroscopy indicated that some of the polymer chains were linear and terminated by secondary alcohol end groups ( Fig. S24 and 25 †).  Increasing the temperature to 100°C and 140°C, also resulted in quantitative conversion of 1, although the polymer molar masses were lower (7700 g mol −1 , Đ M = 1.63 and 6300, Đ M = 1.93, Table 1 entries 4 and 5, respectively). At these temperatures, several factors may be responsible for a decrease in molar mass, including the increased occurrence of thioether linkages (concurrent with loss of COS) in the polymer, and the decreased reaction selectivity towards polymer (in favour of cyclic monomer 2), in line with thermodynamics principles.
The reaction at 60°C was monitored for up to 5 days to assess the impact of longer reaction times on the polymer linkages and molar mass ( Table 1, entry 2). A decrease in molar mass and increasing thioether linkages were observed, suggesting decarbonylsulfonation of the polymer chains upon extended heating.

Impact of CS 2 stoichiometry and co-catalyst
The impact of CS 2 loadings on product distributions and ROCOP regioselectivity was next investigated (

Impact of the nature of ROCOP catalyst and co-catalyst
The use of AlSalen (Al(III)) and CoSalen (Co(II)) complexes failed to give any conversion of 1 under the screened conditions (Table 1, entries 9 and 10). Conversely, the trisphenolate complex, AlTris (Al(III), was found to be active at 25°C, 80°C and 100°C, although, as compared with the analogous CrSalen reactions, the product microstructure was less regioregular and the molar masses of the polymer were lower ( Table 1, entries [11][12][13]. Dizinc complex, LZn 2 Ph 2 (with 1,2cyclohexandiol (CHD) chain transfer agent) and alkoxide/ crown-ether initiator gave no conversion of 1 (Table 1, entries  14 and 15). Lastly, no conversion of 1 was noted in when using only the PPNCl co-catalyst, whilst when performing the reaction with just CrSalen, poor conversion of 1 was obtained as compared with the standard conditions (  1, entry 20). An increased amount of 2 was also observed suggesting that iodide may promote backbiting.
Collectively, these data thus identified CrSalen/PPNCl as the optimal binary catalytic system for 1/CS 2 ROCOP so far, in terms of activity, polymer molar mass, and regioselectivity of the polymer linkages.

Optimisation of reaction conditions towards formation of 2
Although the reaction selectivity could not be tailored towards exclusive formation of 2, 1/CS 2 cycloaddition could be promoted over ROCOP at high temperatures and low [1] 0 . At 110°C and [1] 0 = 0.335 mol L −1 , formation of up to 65% of 2 was observed (Table 1 entries [21][22][23]. Monomer grade xanthate 2 could then be easily isolated using a simple filtration of the reaction mixture on silica, leading to isolated yields of up to 60%, a considerable improvement on the previous report (15% yield). 27 Unfortunately, a further decrease in [1] 0 to 0.08 mol L −1 led to a significant drop in catalyst activity and poor conversions of 1 (33%; Table 1, entry 24).

Kinetic and mechanistic studies
Considering the presence of 2 in all ROCOP reactions performed, it can be envisaged that the formation of poly(CS 2 -co-1) does not proceed directly but stepwise, first by the cycloaddition of CS 2 and 1 into 2, then ROP of 2 (Fig. 3). This would be analogous to what has been reported in the coupling of CO 2 and oxetanes by Darensbourg, 53-56 and by Dove and Coulembier. 57 A reaction performed at [2] (Fig. 4). At 80°C, inspection of the early stages (<1 h) of the reaction revealed that polymer and 2 formed at approximately the same rate (Fig. 4a). After 1-2 h, the concentration of 2 plateaued as the polymer concentration increased.
At 110°C, simultaneous rapid formation of both the polymer and cyclic xanthate was also observed (Fig. 4c), whereas at 25°C, minimal xanthate was detected across several days (Fig. 4d). An experiment to monitor the ROP of 2   Assuming first order kinetics in monomer, the rate of polymer formation was slower in the ROP of 2 than in the 1/CS 2 ROCOP (Fig. 4b). With no build-up of 2 seen during ROCOP, this data therefore suggest that polymer is formed at least partially via direct ROCOP.
2 could also be formed via back-biting of the polymer chains (Fig. 3) Table 2, entry 4). The polymer molar mass also decreased from 8500 g mol −1 to 3600 g mol −1 , corroborating the loss of 2 from polymer chains. Increasing the temperature further to 140°C led to formation of a black, insoluble residue ( Table 2, entry 5). SEC analysis of the THF-soluble fraction indicated the presence of oligomers only. These data suggest that at elevated temperatures, alternative depolymerisation mechanisms are operative leading to significant, uncontrolled, polymer degradation.

Thermal properties of the polymers
Thermal analysis of the polymers with a range of n : m : l ratios was undertaken to understand the impact of varying polymer linkage on properties (Table 3, Fig. S7-15 †). All polymers analysed were found to have temperatures of onset of degradation (T d,onset ) of between 167-208°C. Glass transition temperatures (T g ) were between 108-137°C, higher than those reported previously, 27 but mostly lower than those reported for the analogous, fully oxygenated polycarbonate developed by Gross and co-workers (T g = 128°C). 58 Contrasting with earlier studies which showed a clear impact of sulfur content on polymer properties, 30,42 no general trend between the polymer microstructure and thermal properties was observed. However, as samples of similar molar masses could not be obtained and compared, any existing relationship may be overshadowed by the effect of chain length variations. No crystallinity was detected by differential scanning calorimetry (DSC) or wideangle X-ray scattering (WAXS) analysis (Fig. S16 †).

Degradability of the polymers
A potential advantage of sulfur-containing analogues of polycarbonates is that their degradation into small molecules may be possible under UV light. For example, a catalyst-free method for removal of trithiocarbonate RAFT chain transfer agents from poly(vinylpyridine)s has recently been developed a Calculated by 1 H NMR spectroscopy in CDCl 3 using relative integration of e environments (CH 2 ) assigned to HH/TT trithiocarbonate linkages n (δ = 3.68 ppm (h, J = 6.9 Hz, 4H)), HT xanthate linkages m (δ = 3.51 ppm (t, J = 6.6 Hz, 2H).and thioether linkages l (δ = 3.04-2.74 ppm (m, 2H)). b Calculated by SEC relative to polystyrene standards in THF eluent; Đ M = M w /M n . by Kennemur and co-workers. 21 It was envisioned that such method may also affect the trithiocarbonate linkages in poly (CS 2 -co-1). A reaction performed in THF with 7.5 monomer equivalents of tris(trimethylsilyl)silane (TTMSS) led to oligomeric products after approximately 10 min of UV irradiation (λ = 365 nm; Fig. 5a). Further analysis of the degradation products by 1 H NMR spectroscopy proved challenging (Fig. S20 †). A control experiment performed in the absence of TTMSS resulted in degradation of poly(CS 2 -co-1) too, albeit at a slower rate, as inferred from SEC traces (Fig. 5b). Under these conditions, up to 12% formation of xanthate 2 was also observed by 1 H NMR spectroscopy after 1 hour, before degrading progressively along with poly(CS 2 -co-1).

Conclusions
The ROCOP of CS 2 with an anhydro-functionalised xylofuranose derivative has been reported. Through variation of the reaction parameters (e.g. temperature and CS 2 stoichiometry), some control over the regioselectivity and the nature of the polymer linkages is possible. Conditions can also be tailored to enable the isolation of a polymerisable cyclic xanthate with good yields. Chemical recycling and degradation of the polymers have also been demonstrated. Further investigations are ongoing to develop a deeper understanding of the polymerisation mechanism, of the sulfur/oxygen exchange reactions, and of the impact of the polymer sequence on physical properties.

Conflicts of interest
There are no conflicts to declare.