Self-metathesis of IE and effect of the impuritiesTo enable olefin metathesis on unpurified lignin oil we decided to test ruthenium-based olefin metathesis catalysts with increased stability and rate of initiation, such as phosphine-free Hoveyda Grubbs II generation (HGII) and nitroGrela (nG). In addition, we evaluated a commercial cyclic alkyl amino carbene (CAAC)-based catalyst UltraNitroCat (UCN) for its known increased air and moisture tolerance (Fig. 2a–b). The reactions were performed with IE as a substrate. IE is a convenient substrate for initial screening and is of interest not only for lignin valorization but also for other types of biomasses as it is a major component of essential oils (e.g. clove oil). We have not performed a comprehensive optimization of the reaction conditions on IE, which would not be directly transferable to a realistic lignin-derived mixture, but rather limited ourselves to a short catalyst screening.Fig. 2: Self-metathesis of iso-eugenol (IE) in its pure form and in the presence of various additives.a A reaction scheme of self-metathesis of IE and structures of the screened catalysts: Hoveyda Grubbs 2nd generation catalyst (HGII), nitroGrela (nG), UltraNitroCat (UCN). b Catalytic performance of the catalysts. Reaction conditions: 0.01 mol% catalyst, room temperature, N2 atmosphere, presented yields (conversions) are average of at least two repetitions. c Effect of various additives on a catalytic performance of HGII in self-metathesis of IE. Reaction conditions: 0.01 mol% catalyst, 5 mol% additive, room temperature, N2 atmosphere, 1 h. Normalized conversion = (conversion in the presence of an additive)/(conversion in the absence of an additive)*100% (average of at least two repetitions). Sinapyl alcohol was added in 4.3 mol%. d Effect of various amounts of additives on a catalytic performance of HGII in self-metathesis of IE. Reaction conditions: 0.01 mol% catalyst, room temperature, N2 atmosphere. Normalized conversion based on at least two repetitions.The reaction was performed neat, at room temperature (N2 atmosphere, glove box) using 0.01 mol% of the catalyst with IE as purchased. Both HGII and nG showed comparable performance (although nG expectedly exhibited faster initiation), and the product was formed in 68% and 74% yield for HGII and nG respectively after 1 h (Fig. 2a). A faster initiation in case of nG is caused by the electron withdrawing effect of the nitro group, which facilitates the dissociation of isopropoxy group from the metal center, which is directly involved in the initiation step. When a freshly distilled IE was used the yield improved and reached 84-88%, which can be rationalized by the presence of oxygenated species in as purchased IE (Sigma-Aldrich, 98%, mixture of cis and trans isomers) acting as catalysts’ poisonings (Fig. 2a). No by-products were detected in the reaction mixture and the conversion of IE was equal to the yield of IE-IE. Unlike HGII and nG, under standard conditions UNC only allowed for the formation of trace amounts of the product. Using a freshly distilled substrate or significantly longer reaction time (36 h) did not lead to the formation of product in a desirable yield (the product was formed in 47% yield). While we have not performed any further experimental studies regarding the lower reactivity of UNC, based on previous literature reports, we believe steric factors might play a role in the observed catalytic behavior. UNC is more sterically congested compared to nG and HGII, which can slow down the initiation. For the initiation of the tested catalysts isopropoxy group of benzylidene moiety needs to dissociate from ruthenium, and the benzylidene moiety needs to acquire a position in a plane parallel to Ar-N bond in order to provide a space for the incoming olefin, which is sterically unfavorable in case of UNC34,35. The steric considerations might become even more pronounced for PS due to an additional methoxy group. Based on these results, HGII was chosen for further studies.To enable this reaction on unpurified lignin oil, the effects of other common lignin-derived monomers generally accompanying target IE and PS, featuring hydroxyl, carboxyl, aldehyde or olefinic functionalities were studied (Fig. 2c). Addition of 5 mol% of vanillin, vanillic acid or cinnamyl aldehyde to IE did not lead to any significant drop in the yield of the target homodimer (Fig. 2c). As primary alcohols are known to cause degradation of Grubbs 1st generation (GI)36,37 and Grubbs 2nd generation (GII) catalyst38, the effects of 3-phenyl propanol and furfuryl alcohol were tested. Neither of the alcohols caused a significant change in product yield (Fig. 2c). Even higher loadings of 3-phenyl propanol (20 mol%) did not cause significant drop in yield of IE-IE (Fig. 2d). In addition, the reaction mixture was analyzed by GC-MS and we did not observe any by-products derived from 3-phenyl propanol (Supplementary Fig. 1).While saturated primary alcohols did not cause a significant drop in yield of the desired product, allylic alcohols had a clear effect. Cinnamyl alcohol, allyl alcohol and sinapyl alcohols significantly decreased the yield of the product, when they were added in 4-5 mol% to the substrate (Fig. 2c). Allylic alcohols, thus, represent a main concern for the reaction applied to lignin-derived oils, as they were found to be primary monomeric species released into the solution during RCF of wood and with a high probability can be present in the final mixture of monomers39,40. Even as little as 0.5 mol% of cinnamyl alcohol was found to decrease the yield of the desired product by 82% (Fig. 2d). While there are examples in literature where metathesis of allyl alcohols was performed by GII41, as well as by HGII42, the reactions were usually carried out using high catalyst loadings (up to 5 wt%).To get an increased understanding of the effect of allylic alcohols on the transformation, their derivatives were tested. Cinnamyl methyl ether, cinnamyl acetate and cinnamyl aldehyde (5 mol%) had little effect on the yield of IE-IE (Fig. 2c). Larger quantities of cinnamyl methyl ether (20 mol%) slowed down the reaction resulting in 54% lower conversion after 1 h (compared to the reaction in the absence of an additive). However, the catalyst remained active, and the conversion reached the same level as in case of pure IE after 11 h (Fig. 2d). GC-MS analysis revealed the formation of products derived from self- and cross-metathesis of methyl cinnamyl ether and no products of the decomposition of cinnamyl methyl ether were observed (Supplementary Fig. 2). Thus, the lower reaction rate in this case can be attributed to lower total catalyst loading (since the total amount of olefinic species increased by 20 mol%) and different kinetics of the cross- and self-metathesis of methyl cinnamyl ether.Mechanistic investigation of HGII decomposition by allylic speciesTo gain a further insight into the influence of allylic alcohols and their derivatives the decomposition of HGII in their presence (4 mol% relative to the substrate in toluene-d8) was monitored by 1H NMR. The decomposition was tracked by the disappearance of the alkylidene signal at 16.51 ppm (Fig. 3a). In case of cinnamyl alcohol, the alkylidene signal was almost completely lost after 80 min (Fig. 3a). With cinnamyl methyl ether, the signal was still present after 260 min. However, this signal almost fully disappeared after 22 h, instead two new signals at  ̴ 13.3 ppm (minor) and at  ̴ 13.8 ppm (major) appeared. The signal at Ì´13.8 ppm most certainly corresponds to Fisher carbene species ([Ru] = CHOMe) (vide infra)43,44. In case of cinnamyl aldehyde alkylidene signal was still present after 48 h. Importantly, no signal in the region from -26 ppm to 0 ppm for any of the samples was observed, indicating that ruthenium hydride species were either not formed in detectable by 1H NMR amounts or were rapidly consumed.Fig. 3: NMR study of the deactivation of HGII (Hoveyda Grubbs 2nd generation catalyst) by cinnamyl alcohol, cinnamyl methyl ether and cinnamyl aldehyde.a From top to bottom: 1H NMR spectra of HGII in toluene-d8 and the reaction mixtures containing cinnamyl alcohol, cinnamyl aldehyde and cinnamyl methyl ether in the presence of 4 mol% HGII (N2 atmosphere, room temperature) after specified period of time. The results indicate a rapid decomposition of HGII (disappearance of alkylidene signal at 16.51 ppm) in case of cinnamyl alcohol and slower decomposition in case of methyl cinnamyl ether. The 1H NMR spectra of the reaction mixture containing methyl cinnamyl ether and HGII reveal the presence of enol ethers and Fisher carbenes. b 1H NMR spectra of the reaction mixtures of cinnamyl methyl ether (top) and cinnamyl alcohol (bottom) containing 4 mol% HGII, N2 atmosphere, 110 °C, 1 h. In both cases the catalyst was decomposed (alkylidene signal was lost) within 1 h.In a bid to get additional information regarding the decomposition pathway, the reaction mixtures were analyzed by GC-MS. In the reaction mixture containing cinnamyl alcohol a substantial amount of stilbene and cinnamyl aldehyde in addition to unreacted cinnamyl alcohol was observed (Supplementary Fig.3). In case of cinnamyl methyl ether, besides the peak corresponding to the starting material and stilbene two peaks with the same molecular weight as starting material (Mw = 148), but different retention times were observed (Supplementary Fig. 4). It is assumed that these signals correspond to cis and trans enol ethers formed from methyl cinnamyl ether via an isomerization. 1H NMR spectra of the reaction mixtures after 48 h supported this assumption; new signals were observed in the range characteristic to enol ethers: 4.5 ppm (cis-), 4.8 (trans-) ppm (PhCH2CHCHOMe), and 6.0-6.4 ppm (cis- and trans- PhCH2CHCHOMe) (Fig. 3a–b)45. This evidence is consistent with the isomerization of methyl cinnamyl ether to the corresponding cis and trans forms of 3-Methoxy-2-propenyl-benzene (MPB), which in turn can form Fisher carbenes with HGII. The reaction mixture containing cinnamyl aldehyde showed almost solely a peak of the starting material (Supplementary Fig. 5).Subsequently the decomposition of HGII by cinnamyl methyl ether and cinnamyl alcohol at higher temperatures (4 mol%, 110 °C) was studied. In this case, the decomposition was significantly faster, and a complete disappearance of alkylidene signal was observed for both substrates within an hour (Fig. 3b). In case of cinnamyl methyl ether, while a significant amount of cis- and trans-MPB was detected in the reaction mixture by 1H NMR, signals corresponding to Fisher carbenes (13-14 ppm) were absent, likely due to their more rapid decomposition (Fig. 3b). Analysis of the reaction mixture by GC-MS further confirmed a significant amount of cis- and trans-MPB (Supplementary Fig. 6). Important to mention, that under this conditions cinnamyl alcohol was partly isomerized to 3-phenyl propionaldehyde (Supplementary Fig. 7).To identify the fate of the decomposed catalyst, ESI-HRMS (high resolution electrospray ionization mass spectroscopy) analysis was attempted of the following reaction mixtures: HGII + IE, HGII+cinnamyl alcohol and HGII+allyl alcohol (Supplementary Fig. 8). The reaction mixtures containing cinnamyl and allyl alcohol revealed a presence of high molecular weight species (Mw in a range of 900–1400 Da), which were not observed in case of IE. Importantly, there is a great degree of similarity in ruthenium species found in the reaction containing allyl alcohol and cinnamyl alcohol, suggesting that both alcohols probably cause the decomposition via a similar pathway. While it is difficult to discern a precise structure of the products of the decomposition of the catalyst which will require further studies, it is likely that the decomposition involves the formation of higher molecular mass ruthenium species.The important implications of this part of the study are: (1) at room temperature allyl ethers cause the decomposition of HGII at significantly lower rate than allylic alcohols. However, under prolonged reaction times, they isomerize into enol ethers which cause the formation of catalytically inactive Fisher carbenes. (2) At high reaction temperatures (110 °C) and high catalyst loadings (4 mol%) the isomerization of allyl ethers proceeds significantly faster and causes a fast deactivation of HGII. (3) At higher temperature (110 °C) and high catalyst loading (4 mol% HGII) allylic alcohols convert into the corresponding saturated aldehydes. Taking into account previously proposed mechanism for the decomposition of GI by allylic species (Fig. 4b)46, an increased poisonous effect of allylic alcohols vs. allylic ethers may be due to the increased propensity of allylic alcohols for β-hydride shift and/or existence of alternative decomposition pathways. E.g. the ruthenium alkylidene can be lost via dehydrogenation of cinnamyl alcohol (Supplementary Fig. 35), which is in line with the observed formation of cinnamyl aldehyde and previous reports36. However, more investigations will be needed to draw final conclusions.Fig. 4: Analysis of the reaction mixtures of self-metathesis of iso-eugenol (IE) in the presence of cinnamyl alcohol (5 mol%).a Self-metathesis of IE in the presence of cinnamyl alcohol (5 mol%), stepwise vs. at once addition of Hoveyda Grubbs 2nd generation catalyst (HGII). Reaction conditions: room temperature, N2 atmosphere, presented yields are average of at least two repetitions. During a stepwise addition of HGII to the reaction mixture of IE and cinnamyl alcohol (5 mol%) very low conversion (6%) of IE was observed even when the total catalyst loading reached 0.5 mol%. In case, when the catalyst was added in a single portion (0.5 mol%) 83% conversion of IE was achieved within 1 h. b A reaction mechanism proposed by Werner’s group for the deactivation of Grubbs 1st generation catalyst (GI) by allylic alcohol46. c Proposed pathway for the formation of products 1 and 2 from cinnamyl alcohol during olefin metathesis of IE (5 mol% of cinnamyl alcohol, HGII 0.5 mol% added at once).It might be difficult to avoid a minor presence of allylic alcohols in lignin-derived steams (as they are the primary species released into the solution during lignin depolymerization). Allyl alcohols can significantly increase required catalyst loadings for the metathesis. E.g., to achieve 83 mol% conversion of IE in self-metathesis (1 h) in the presence of 5 mol% cinnamyl alcohol, 0.5 mol% of HGII was required, which is 50 times more compared to pure IE (Fig. 4a). Interestingly, when the same reaction was performed with the catalyst being added in a stepwise manner, a very low conversion was achieved when the total catalyst loading reached 0.5 mol% (6% vs. 83%). GC-MS analysis revealed that in case when HGII was added in one portion almost no cinnamyl alcohol was present in the final mixture (Supplementary Fig. 9). It was converted into stilbene (a product of self-metathesis), propenyl benzene (a product of cross metathesis with IE) and products 1 and 2 (Mw = 162 and Mw = 176 respectively, for proposed structures see Fig. 4a). The formation of these products can be rationalized based on the mechanism proposed by Werner’s group for the decomposition of GI by allylic species (Fig. 4b, c; for the general mechanism of olefin metathesis see Supplementary Fig. 34)46. On the contrary, for the mixture with a stepwise addition of HGII cinnamyl alcohol was still observed in the final mixture in large quantities (Supplementary Fig. 10). These observations are consistent with the conclusion that higher catalyst loadings not only increase the rate of IE metathesis, but also the rate of decomposition of cinnamyl alcohol, which is probably catalyzed by the decomposed HGII. These considerations prompted us to optimize in situ conversion of cinnamyl alcohol into catalytically inert species.A one-pot tandem catalytic procedure was envisioned where cinnamyl alcohol is converted into a saturated aldehyde via an isomerization or into a saturated alcohol and α, β-unsaturated aldehyde via a disproportionation followed by self-metathesis of IE. Tandem metathesis/isomerization of allylic alcohols was realized by Snapper and coworkers47 who performed the metathesis reaction at room temperature (0.5 mol% catalyst) followed by an isomerization of the product at higher temperatures (200 °C). It is hypothesized that these transformations can be realized in a reverse sequence, where the isomerization of allylic alcohols is taking place at elevated temperature followed by metathesis of IE at room temperature. This sequence was tested on a mixture of IE and cinnamyl alcohol (5 mol%), using 0.025 mol% HGII (relative to IE) at 90 °C. The conversion was monitored by GC-MS (Fig. 5a). When almost full conversion of cinnamyl alcohol (primarily to 3-phenyl propionaldehyde) was achieved (12–24 h), the reaction mixture was cooled to room temperature and an additional portion of HGII (0.025 mol%) was added to carry out metathesis of IE. The final product was formed in 89% yield in 1 h with a total catalyst loading of 0.05 mol%, which is 10 times less compared to a single olefin metathesis reaction (Fig. 4a). A slightly higher catalyst loading for the metathesis of IE (0.025 mol% vs. 0.01 mol%) compared to pure IE can be rationalized by the fact that minor amounts of cinnamyl alcohol were still present in the reaction mixture.Fig. 5: Tandem isomerization(disproportionation)/metathesis of iso-eugenol (IE)/cinnamyl alcohol (5 mol%) mixtures.a Conversion of cinnamyl alcohol (mainly via isomerization) (5 mol% in IE) in the presence of Hoveyda Grubbs 2nd generation (HGII) at 90 °C followed by the olefin metathesis of IE under standard conditions (room temperature, 0.025 mol% HGII, N2). b Conversion of cinnamyl alcohol (mainly disproportionation) in the presence of RuClH(CO)(PPh3)3 (RuH) at 90 °C followed by the olefin metathesis of IE under standard conditions (room temperature, 0.02 mol% HGII, N2). Both methods allowed to reach high in situ conversion of cinnamyl alcohol, which in turn allowed to perform metathesis of IE with low catalyst (HGII) loading in a second step.In addition, carbonyl-chloro-hydrido-tris-(triphenylphosphine)-ruthenium(II) (RuH) was tested for the tandem isomerization/metathesis sequence. We found that in this case cinnamyl alcohol primarily undergoes a disproportionation into 3-phenyl propanol and cinnamyl aldehyde (Fig. 5b). A 92% conversion of IE was achieved in 1 hour upon addition of 0.02 mol% HGII, which corresponds to 0.047 mol% total ruthenium catalyst loading. While RuH demonstrated a good performance in the isomerization of cinnamyl alcohol in the presence of IE, it is important to mention that RuH species can slow down metathesis of IE, when present in large amounts (see Section 1 in Supplementary Information).Thus, we have developed two alternative systems for the elimination of allylic alcohols from the reaction mixture: via either disproportionation or isomerization. The development of metal-free protocols for these pathways is of interest for future studies.The studies of the decomposition of HGII (vide supra) revealed that at higher temperature and higher catalyst loading cinnamyl alcohol ethers isomerize into the corresponding vinyl ethers, which in turn can poison the catalyst. Thus, we investigated if the designed tandem protocol is effected by the presence of cinnamyl ethers. The reaction was performed under the same reaction conditions as on Fig. 5a, but with an addition of 2.5 mol% of cinnamyl methyl ether. The isomerization of methyl cinnamyl ether was taking place, however, it was observed at a much lower extent than in case of cinnamyl alcohol (Supplementary Fig. 11). The subsequent metathesis of IE exhibited slower kinetics (the completion was reached over 20 h) and required slightly higher total catalyst loading (0.058 mol% total) to drive the reaction to 87% conversion of IE (74% was achieved with 0.05 mol%). Overall, this result demonstrated that the developed method was effective for the reaction mixtures containing allylic ethers, however, required a slightly higher catalyst loading.Self- and cross-metathesis of PSThe major product of the hydrogen-free RCF of hardwood is PS. Thus, after the reaction was optimized on easily available IE the reaction conditions were tested on PS. Under the reaction conditions developed for IE (0.01 mol% HGII, room temp., N2 atmosphere, neat) the product was formed in 23% yield after 1 h, after 3 h the yield increased to 37%, and reached 52% after 24 h (Fig. 6a). When the reaction was performed in toluene the conversion reached 68% after 3 h and 93% after 28 h (Fig. 6a). It is apparent that self-metathesis of PS proceeds significantly slower than self-metathesis of IE. A higher conversion of PS in a solution compared to neat conditions is very likely due to a better mass transfer at the later stages of the reaction, when the reaction mixture becomes viscous and eventually solidifies. In the case of IE, due to a rapid conversion, this effect is less pronounced.Fig. 6: Self- and cross-metathesis of propenyl syringol (PS).a Self-metathesis of PS. Reaction conditions: Hoveyda Grubbs 2nd generation catalyst (HGII), N2 atmosphere, room temperature. Addition of solvent allows to reach higher conversion due to a better mass transfer at the later stages of the reaction. Addition of equimolar amount of iso-eugenol (IE) significantly improves conversion of PS. b Self-and cross-metathesis of PS and IE. Reaction conditions: HGII (0.01 mol% relative to the total amount of IE and PS), N2 atmosphere, room temperature. The results clearly indicate a higher rate of PS cross-metathesis vs. self-metathesis, especially pronounced at the earlier stages of the reaction.With 0.02 mol% of HGII 93% yield can be achieved already in 1 h (Fig. 6a). Interestingly, upon addition of an equimolar amount of IE the conversion of PS improved, reaching 73% in 1 h, and 92% in 10 h with 0.01 mol% catalyst loading (relative to total amount of PS and IE).To understand the nature of the observed reactivity a series of reactions with a mixture of PS and IE in different ratios (Fig. 6b) were performed. In the equimolar PS/IE mixture (Table in Fig. 6b, entries 1–3), a slightly faster conversion of IE compared to PS was observed, but this difference was significantly diminished compared to pure substrates. Even more important observation is the evolution of the ratio between homo- and heterodimers of PS (PS-PS/PS-IE). The ratio of dimers was estimated via 1H NMR (Supplementary Fig. 12). At a lower conversion of starting materials there is a clear preference for cross-metathesis of PS with IE vs. self-metathesis, which is reflected in PS-IE/PS-PS ratio being > 2 times greater than the predicted ratio for random coupling (Random coupling was simulated in python, see Supplementary Information, Section 4) (Table in Fig. 6b, entry 1). Over the course of the reaction the proportion of PS-PS dimer increases getting closer to (but still below) the predicted ratio for the random coupling. The same trend was observed for the reactions performed at PS/IE ratios 0.66. 1.56 and 3.56 (Table in Fig. 6b, entries 4–7). It can be concluded that in all instances there is a preference towards the formation of a heterodimer (PS-IE) over the homodimer (PS-PS).The enhanced reactivity of PS in the presence of IE can be rationalized via both electronic and steric factors. An increased steric demand of PS may result in a slower initiation. In addition, because of the electron-rich nature of PS the corresponding ruthenium carbene complex is more stable and thus less reactive towards ruthenacycle formation. The presence of IE leads to a faster initiation and formation of the corresponding ruthenium carbene, which in turn can react with PS to give the heterodimer.Based on these observations, it is apparent that a presence of IE in the mixture of PS significantly facilitates the conversion of PS. As such an addition of IE to the lignin-derived oil with low IE/PS ratios is advantageous.Polyesters based on IE-IE (P-IE-IE), and IE-IE, IE-PS and PS-PS dimers (P-PS-IE)As a model reaction to convert the obtained bisphenols into final products polyesterification with succinyl chloride was chosen. While P-IE-IE was reported before26, (P-PS-IE) was not previously described. Figure 7a and Supplementary Fig. 28–29 show 1H and 13C NMR spectra of P-PS-IE in DMSO-d6. The PS/IE ratio (PS/IE = 1) in the starting mixture and in the final polymer stayed largely unchanged (Supplementary Fig. 13). This indicates that the reactivity of all three dimers towards polycondensation was similar, and the final ratio of PS and IE units can be predicted from the initial composition of the mixture.Fig. 7: Structural and thermal properties of P-IE-IE and P-PS-IE.a 1H NMR spectra of P-IE-IE and P-PS-IE in DMSO-d6. b Representative differential scanning calorimetry (DSC) curves of P-IE-IE and P-PS-IE. P-IE-IE exhibits three clear thermal transitions (glass transition, crystallization and melting), while P-PS-IE demonstrated only glass transition. This can be rationalized by a more regular structure of P-IE-IE compared to P-PS-IE allowing for its packing into crystals. c Representative thermogravimetric analysis (TGA) graphs (under N2) of P-IE-IE and P-PS-IE. TGA curves indicate two mass loss processes.Thermal properties of the polymers were assessed by differential scanning calorimetry (DSC) and thermogravimetric analysis (TGA) (Fig. 7b). The main drastic difference between P-IE-IE and P-PS-IE is the absence of crystallization peak in case of P-PS-IE, where only a clear glass transition occurs at 149.2 ± 0.7 °C (mid.point, Fig. 7c). In case of P-IE-IE there are three transitions: glass transition (95.3 ± 0.6 °C, mid.point; Fig. 7c), crystallization (186.2 ± 0.9 °C) and melting (232.9 ± 0.6 °C). This is expected from a more regular structure of P-IE-IE vs. P-PS-IE allowing for a better packing of polymer chains. P-PS-IE exhibited ca. 50 °C higher Tg, reflecting impeded mobility of the polymer chain due to an additional methoxy group in the aromatic ring of PS fragment. Both polymers exhibited two mass losses processes. P-PS-IE polymers demonstrated superior thermostability with Td5% being around 40 °C higher than for P-IE-IE (304 ± 3 °C vs. 263.4 ± 0.5 °C). The thermal properties of P-IE-IE are comparable (or even superior) to commercial polyesters (PET, Tg = 80 °C, Tm = 251 °C)26. The thermal properties of P-PS-IE, which exhibited high Tg = 149.2 ± 0.7 °C are matching commercial amorphous BPA-based polycarbonates (Tg = 150 °C)48.Molecular weight determination of the prepared polymers was hindered by their poor solubility in THF and DMF—solvents generally used for gel permeation chromatography (GPC). Moreover, while the polymers are soluble in DMSO, polystyrene standards needed for analysis show a poor solubility in DMSO at room temperature. These factors made the estimation of the molecular weight of the polyesters challenging and therefore, the molecular weights were estimated via Diffusion-Ordered NMR Spectroscopy (DOSY). The experiments were performed in DMSO-d6 at 100 °C (Supplementary Fig. 14). MW were estimated to be 13041 and 11450 Da for P-IE-IE and P-PS-IE respectively. However, both samples exhibited a high polydispersity expected for step-growth polymerization.Successful polymerization of the mixture consisting of PS-PS, PS-IE and IE-IE dimers suggests that polymerization of dimers obtained from lignin-derived oil is feasible. Moreover, it allows for potential additional tuning of the final properties upon adjustment of the initial PS/IE ratio.Preparation of lignin-derived oil enriched in PS and IEAfter getting a better understanding of the system using model mixtures of PS and IE, we focused on transferring the developed protocols on a lignin oil derived directly from wood. While an optimization of the reaction conditions for hydrogen-free RCF is outside the scope of the current study and can be found elsewhere24, in this work the relationships between the composition of the reaction mixture and its reactivity were explored. Specifically, a particular focus was placed on the effect of allylic species, as they were found to exert most significant effect on the catalytic conversion of target IE and PS. The screening was limited to the reaction conditions for the specific set-up used in this work (Supplementary Table 1). RCF was performed using birch sawdust (particle size < 0.3 mm). Qualitative analysis of the lignin oil (LO) revealed that performing RCF at lower temperature (190 °C) results in a mixture containing a significant amount of sinapyl and coniferyl alcohol and their derivatives, which is in a good agreement with the mechanistic proposal for lignin depolymerization established in previous studies (Supplementary Fig. 15)39,40. LO obtained at 210 °C already contained an increased amount of products of overreduction (e.g. propyl and ethyl syringol, Supplementary Fig. 16). For this set-up the amount of allylic species can be minimized, while the yield of PS and IE can be maximized by performing the RCF reaction at 200 °C for 4 h with a slow preheating from room temperature to 200 °C (Supplementary Figs. 17– 18). However, when sawdust with large particle size (non-sieved wood) was used for RCF at 200 °C, the obtained LO still contained minor amounts of sinapyl alcohol (Supplementary Fig. 33).A separation of monomers from lignin dimers and oligomers can be accomplished either via a distillation or via a solvent extraction. Due to a simplicity of the protocol on the laboratory scale, an isolation of monomers from the LO via an extraction with hexane (reflux, overnight) was performed, previously reported by the Sels research group20. In addition, extraction can help to minimize the amount of oxygenated species (including allylic alcohols) in the lignin monomers oil (LMO), due to their higher polarity and as such a lower solubility in hexane. At the same time purification of LMO from allylic species via distillation is challenging due to close b.p. of coniferyl alcohol—340 °C and PS—353 °C.In this work, two types of LO and their corresponding lignin monomers oils (LMO) were studied: obtained via RCF at 190 °C (LO-190, LMO-190) and at 200 °C (LO-200, LMO-200). LO-200 showed a minor presence of allylic species (sinapyl alcohol), while in case LO-190 both sinapyl alcohol and coniferyl alcohol were present in high amounts. An extraction with hexane (reflux, overnight) allowed to almost fully remove allylic species from LMO-200 and significantly reduce their amount in case of LMO-190 (Supplementary Fig. 19-22). However, sinapyl and coniferyl alcohols were still clearly present in LMO-190. Total content of IE and PS in LMO was determined by GC-FID (Supplementary Fig. 23) and 1H NMR (Supplementary Fig. 24). Both methods provided similar quantitative results. A total content of PS and IE was found to be 67-72 wt% (LMO-200) and 55-57 wt% (LMO-190) with a molar ratio of PS/IE of 2.3 (LMO-200) and 2 (LMO-190). A general pathway to LMO is schematically presented on Fig. 8a. 1H NMR of LMO-200 is shown on Fig. 8b. Even though the mixture consists of multiple components, signals corresponding to PS and IE are clearly distinguishable, which allows monitoring the transformation of PS and IE via 1H NMR (Fig. 8c). A full description of the identified compounds in LMO-200 are given in Supplementary Table 2.Fig. 8: Olefin metathesis of lignin monomers oil (LMO-190 and LMO-200) using Hoveyda Grubbs 2nd generation catalyst (HGII).a The transformation pathway from wood to LMO: reductive catalytic fractionation (RCF) of birch sawdust, hexane extraction of lignin monomers from lignin oil (reflux, 16 h). b 1H NMR of LMO-200 enriched in propenyl syringol (PS) and iso-eugenol (IE). Despite the complexity of the mixture the signals of PS and IE are clearly distinguishable, which allows monitoring the transformation by 1H NMR. c An example of the reaction mixture of the metathesis of LMO. Characteristic peaks corresponding to PS, PS-PS, PS-IE dimers. d Optimization of olefin metathesis of LMO. LMO-200 with minimized amount of allylic alcohols required significantly lower catalyst loading compared to LMO-190 containing higher amounts of sinapyl and coniferyl alcohols. e An example of 1H NMR spectrum of PS-enriched purified lignin dimers (PS-PS, PS-IE). A simple wash of the reaction mixture with Et2O allowed to obtain dimers of high purity.Metathesis of lignin-derived oilUpon obtaining the LMOs, their performance in metathesis were studied by performing the reactions under standard reaction conditions (room temperature, inert atmosphere (glovebox), HGII), and monitoring the progress of the reaction via conversion of PS (Fig. 8c). LMO-190 was tested first, where both coniferyl and sinapyl alcohols were detected. When the reaction was performed with 0.07 mol% of HGII the conversion of PS reached 21% after 22 h (Table in Fig. 8d, entry 1). Increasing the catalyst loading to 0.14 mol% resulted in improved, but still moderate conversion (50%, Table in Fig. 8d, entry 2). LMO-200 demonstrated a significantly better performance, where the conversion of PS reached 50% already with 0.07 mol% HGII and 79% with 0.14 mol% HGII (Table in Fig. 8d, entry 3–4). The conversion of PS in LMO-200 dropped only slightly when catalyst loading was lowered to 0.11 mol% (75 %, table in Fig. 8d, entry 6). These results are in good agreement with the studies performed on model compounds, which revealed that the presence of allylic species (even in minor amounts) exhibits a clear deleterious effect on the metathesis of IE.The developed protocol for in situ conversion of allylic alcohols was probed on LMO, obtained from non-sieved wood (RCF at 200 °C, Supplementary Fig. 33) containing a minor amount of sinapyl alcohol. Using 0.07 mol% of HGII led to 31% conversion of PS. When the reaction was performed in a two-step manner (90 °C, HGII 0.025 mol%, 14 h, followed by the reaction at room temperature with 0.07 mol% of HGII, 22 h), the conversion improved to 77%. Note, that while the total amount of HGII used was 0.095 mol%, the product was not formed during the pretreatment step. This result indicates that the developed tandem protocol can be applied to mixtures derived directly from wood, however, additional studies will be required to identify all the transformations occurring during the pretreatment step.Then we tested if an enhanced reactivity of PS in the presence of IE, discovered during model studies, is also observed in case of metathesis performed on lignin-derived mixture. IE was added to the reaction mixture in 60 wt% (50 mol%). Indeed, an addition of IE allowed to lower the catalyst loading more than twice resulting in 83% and 89% conversion of PS respectively with 0.055 mol% HGII and 0.065 mol% of HGII (relative to total amount of PS and IE) (Table in Fig. 8d, entries 7-8). Note, that the reaction was evaluated based on the conversion of PS, which solely comes from LMO, rather than on the conversion of IE, which was added to the reaction mixture in its pure form.A gaseous phase of the reaction mixture was also analyzed using head-space GC-MS method (Supplementary Fig. 25). Expectedly, in addition to toluene (used as a solvent) the chromatogram revealed the presence of butene as the only product.Overall, the results obtained from lignin-derived mixtures (LMO-190 and LMO-200) are in line with model studies signifying the importance of the elimination of allylic alcohols from the reaction mixture and an enhanced reactivity of PS in the presence of higher amount of IE.Isolation and photo-chemical properties of lignin derived dimersOne of the key advantages of the system is a simplicity of the isolation of the final products. Final dimers (PS-PS, PS-IE and IE-IE) possess a significantly higher polarity and boiling points compared to both unreacted starting PS and IE and other monomeric impurities, which offers an easy purification either via distillation or extraction with non-polar solvents. The obtained dimers in their pure form were isolated via two dispersion/centrifugation cycles using Et2O as a solvent (Fig. 8e).The presence of stilbene moiety in the structure of the dimers renders them photo-chemically active. Several transformations can occur upon their exposure to UV-light, including [2 + 2] cycloaddition, cis-trans isomerization, Mallory reaction, etc. This reactivity can be beneficial allowing for the diversification of the products, however, it can also compromise the stability of the reaction mixtures. To this end, a targeted study of the stability of the dimers upon the exposure to UV light (365 nm) in CDCl3 and CD3OD was performed. In CDCl3 upon exposure to UV light (365 nm) the initially yellow solution rapidly (within minutes) acquires a red color, which changes back to yellow upon exposure to white light (Supplementary Fig. 26a), indicating a photochromic reaction. However, 1H NMR of the reaction mixture revealed other irreversible and non-selective transformations (Supplementary Fig. 26b). Thus, the solutions of the dimers in CDCl3 need to be handled with caution in the presence of UV light. The dimers dissolved in CD3OD were stable upon exposure to UV light. No color change was noted and the 1H NMR spectrum revealed a clear cis-trans isomerization (Supplementary Fig. 27).Polymerization of lignin-derived dimers into polyestersAs a final part of this work, the preparation of polyesters from lignin-derived mixtures of PS-enriched dimers was explored. PS-enriched mixtures are more challenging with regards to the polymerization. This is due to higher steric demand of PS fragments compared to IE and higher pKa of the phenolic groups, which can slow down their nucleophilic attack on the succinyl chloride. In addition, due to a highly electron-rich nature of PS units they are especially prone towards oxidation under basic conditions. To probe if lignin-derived mixtures of dimers with a high PS content still can be successfully polymerized, a mixture with PS/IE ratio of 6.65 was chosen (Supplementary Fig. 30). PS/IE ratios can be easily tuned by an addition of IE or by choosing a different source of wood (e.g. soft wood). The polyester was prepared following the same procedure as in case of model compounds (Fig. 9a) via polycondensation with succinyl chloride. It is important to underline that a second product of the metathesis reaction – butene-2, is a precursor for the production of succinyl chloride via established synthetic routes (via maleic anhydride)49,50. Even though this transformation is outside the scope of the current study the final polymeric materials can thus be considered all-carbon bio-based. 1H NMR of the final polymer is shown on Fig. 9a (Supplementary Fig. 31), 13C NMR is shown on Supplementary Fig. 32. The thermal properties of the polymer using DSC and TGA (Fig. 9b, c) were evaluated. Tg, Td5% and Td50% were found to be 140.3 ± 0.7 °C, 231 ± 6°C and 330 ± 4 °C respectively. Mw of the polymer (16330 Da) was estimated using DOSY NMR and is in a similar range as Mw of P-PS-IE and P-IE-IE (13041 and 11450 Da). This part of the work demonstrates a final step of the proposed lignin monomers valorization strategy, which addresses the entire process from wood to final polymeric materials.Fig. 9: Structural and thermal properties of polyester prepared from lignin-derived mixture of dimers.a 1H NMR spectrum of the polyester. b A representative termogravimetric analysis (TGA) graph of the polyester (N2 atmosphere). The curve exhibits two major mass loss events. c A representative differential scanning calorimetry (DSC) curve of the polyester indicating only one phase transition – glass transition.
