Hydrogenative Depolymerization of Nylons

The widespread crisis of plastic pollution demands discovery of new and sustainable approaches to degrade robust plastics such as nylons. Using a green and sustainable approach based on hydrogenation, in the presence of a ruthenium pincer catalyst at 150 °C and 70 bar H2, we report here the first example of hydrogenative depolymerization of conventional, widely used nylons and polyamides, in general. Under the same catalytic conditions, we also demonstrate the hydrogenation of a polyurethane to produce diol, diamine, and methanol. Additionally, we demonstrate an example where monomers (and oligomers) obtained from the hydrogenation process can be dehydrogenated back to a poly(oligo)amide of approximately similar molecular weight, thus completing a closed loop cycle for recycling of polyamides. Based on the experimental and density functional theory studies, we propose a catalytic cycle for the process that is facilitated by metal–ligand cooperativity. Overall, this unprecedented transformation, albeit at the proof of concept level, offers a new approach toward a cleaner route to recycling nylons.


■ INTRODUCTION
Nylons are commonly used plastics with a current global market of more than 8 million tons per annum, which is expected to grow at the rate of 2.2%, reaching up to 10.4 million tons, equivalent to 47.0 billion USD by 2027. 1 Inevitably, the ubiquity of nylons and their nonbiodegradable nature have resulted in increased land and ocean pollution, posing a severe threat to our ecosystem. 2 Therefore, it is critical that efficient and sustainable technologies for the recycling of plastics such as nylons are developed. The current plastic recycling technologies are mostly based on mechanical recyclingconverting one form of plastic to another by mechanical reprocessing (melting and re-extrusion): for example, converting plastic bottles to fibers. However, mechanical recycling produces poor quality of plastic, and the number of times a plastic can be mechanically recycled is also limited. 3 Distinct from mechanical recycling, chemical recycling involves depolymerization of a plastic to produce the monomer raw material from which the same virgin plastic can be made. 3 Thus, only chemical recycling closes the loop of plastic production and is a sustainable mode of recycling. The reported chemical recycling methods for nylons are mainly based on pyrolysis, hydrolysis, or aminolysisall of which require harsh reaction conditions (such as temperatures ≥250−300°C). 4 Other methods based on glycolysis or aminoglycolysis do not result in the efficient regeneration of nylons from the degraded polymers. 5 Thus, development of new, efficient, and sustainable methods to depolymerize nylons for the purpose of recycling is highly desirable.
Catalytic hydrogenation is an atom-economic, green, and sustainable route for organic transformations as (a) it does not produce stoichiometric waste as in the case of conventional reducing agents and (b) hydrogen gas can potentially be produced from renewable sources. 6 In the past, we reported an efficient ruthenium pincer catalyst (1′) for the hydrogenation of amides via C−N bond cleavage to form alcohols and amines ( Figure 1Ai). 7 Several other homogeneous catalysts based on ruthenium, 8−17 iron, 18−21 and manganese 22 have also been reported for the hydrogenation of amides to form alcohols and amines. 23,24 Additionally, we also reported acceptorless dehydrogenative coupling of alcohols and amines to form amides catalyzed by ruthenium pincer complex 2′ ( Figure  1Aii). 25 In this direction, Guan 26 and we 27 independently reported the synthesis of polyamides via dehydrogenative coupling of diols and diamines using the ruthenium pincer catalyst 2′ (Figure 1Aiii). In order to develop a closed loop cycle, we envisioned that it might be possible to achieve the very challenging catalytic hydrogenative depolymerization of polyamides to their monomers such as diols and diamines or amino alcohols (Figure 1Aiv). Saito and co-workers have briefly mentioned an example of Ru-catalyzed hydrogenation of specialized, water-soluble functionalized polyamides, although the structures of the polyamides and the hydrogenation products thereof were not reported. 15 Hydrogenative depolymerization of polyesters and polycarbonates was also reported previously. 28 However, we are not aware of any report on the catalytic hydrogenation of conventional nylons such as nylon-6, nylon-66, or nylon-12 or related polyamides, in general, which are highly robust and resistant to solvents. We report here our findings on the depolymerization of nylons using a ruthenium-catalyzed hydrogenation process.

■ RESULTS AND DISCUSSION
A major challenge in the chemical reactivity of nylons is their resistance to solvents and reagents due to the multiple, strong intermolecular hydrogen bonding interactions between the polymer chains. We started our investigation by searching for a solvent that would (a) be capable of dissolving nylon-6 and (b) be compatible with Ru pincer complex-catalyzed hydrogenation. We previously reported that complex 1 ( Figure  1B) in the presence of 2 (or more) equiv (relative to 1) of base is a highly active catalyst for the hydrogenation of esters and amides. 29,30 Under the catalytic conditions using 1 (2 mol %), KO t Bu (8 mol %), and H 2 (60 bar) for 48 h at 135°C, we screened several solvents (2.5 mL) for the catalytic hydrogenation of nylon-6 (1 mmol, commercial resins with a size of 2−3 mm, M w = ∼10,000). As ruthenium PNN complexes react with acids, resulting in their deactivation as catalysts, 31 we avoided using acids such as formic acid or acetic acid as solvents even though they are known to dissolve nylons. No conversion of nylon-6 was observed when toluene, tetrahydrofuran (THF), 1,4-dioxane, water, or dimethylformamide was used under the catalytic conditions described above, and resins of nylon-6 were completely recovered unreacted, presumably due to the insolubility of nylon-6 in these solvents (Table S1; see the Supporting Information). We then employed dimethylsulfoxide (DMSO) as a solvent as it can dissolve nylon-6 at increased temperature (40 g/100 mL at 130°C). 32 In addition to the ability of solubilizing nylons, DMSO has also been claimed as an environmentally friendly solvent as it can be prepared from renewable sources, exhibits toxicity lower than that of several other polar solvents, and is biodegradable. 33 Because of such unique properties, coupled with low cost, DMSO has been utilized in several industrial processes as a solvent or reagent. 33 Encouragingly, in the case of DMSO solvent, a mixture of a solution and a white precipitate was obtained at the end of the hydrogenation, suggestive of a partial solubilization or degradation of nylon-6. Indeed, analysis of the reaction mixture by gas chromatography (GC) revealed the formation of 6-amino-1-hexanol in 14% yield. Additionally, some other signals suggestive of oligomers were observed by GC. High-resolution electrospray ionization mass spectrometry (HR-ESI-MS) and matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) spectrometry as well as NMR spectroscopy confirmed the presence of 6amino-1-hexanol and oligoamides (dimer−tetramer). Additionally, diffusion-ordered spectroscopy (DOSY) showed that the average size of the formed oligoamides is approximately 3 times that of 6-amino-1-hexanol. The conversion of nylon was determined by measuring the weights of the solid reaction mixture before and after the reaction, as nylon-6 has no solubility in DMSO at room temperature (see the Supporting Information for details). Based on this method, the conversion of nylon-6 was determined to be 65%. Additional solvents such as anisole, hexamethylphosphoramide, 1-butyl-3-methyl imidazolium octyl sulfate, diglyme, and meta-cresol were also screened under the same catalytic conditions (Table S1; see the Supporting Information), but we did not obtain any better results than when using DMSO.
Further, we optimized the catalytic conditions by varying base loading, temperature, pressure, reaction time, ruthenium complexes (see Table 1), and additives such as water, molecular sieves, and Lewis acids (see the Supporting Information for the table of optimization). Best results were obtained using 1 or 2 (2 mol %), KO t Bu (8 mol %), 70 bar of H 2 , and 150°C (oil bath temperature) for 48 h, which resulted in 77−80% conversion of nylon-6 (resins) and formation of 6amino-1-hexanol in 24−26% yield (Table 1, entries 1−2 and  Table 2, entry 1), with the remaining being oligomers (dimer− tetramer). As complexes 1 and 2 exhibited comparable catalytic activity (Table 1), we used complex 1 for further studies as it is relatively easier to synthesize compared to complex 2.
One of the reasons for the lower yield of 6-amino-1-hexanol could be catalyst deactivation over time. Hence, in order to achieve better yields of 6-amino-1-hexanol, we attempted sequential hydrogenation of nylon-6. After the first hydrogenation in DMSO under the optimized catalytic conditions, 2 mol % of complex 1 and 8 mol % of KO t Bu were added back to the mixture, and the autoclave was pressurized with 70 bar of H 2 and heated at 150°C for another 72 h. Following both hydrogenation steps, 99% of nylon-6 was depolymerized to form 6-amino-1-hexanol (37%) and a mixture of oligoamides (Table 2, entry 1 b ; see the Supporting Information, section 9). We found that if not for a matter of solubility, then THF or 1,4-dioxane is a better solvent than DMSO for hydrogenation of amides using complex 1 (see the Supporting Information, section 14.4). Thus, we developed an additional approach in which, after the first step of hydrogenation, we distilled off the DMSO and added 1,4-dioxane along with complex 1 (2 mol %), KO t Bu (8 mol %), and H 2 (70 bar), and the autoclave was heated at 150°C for an additional 72 h. Interestingly, this resulted in a higher yield of 6-amino-1-hexanol (48%), although a lower conversion of nylon-6 (80%) was observed compared to the aforementioned process when DMSO was used for both the steps, presumably due to a lower solubility of nylon-6 in dioxane ( Table 2, entry 1 c ; see the Supporting Information, section 9).
In order to explore the generality of this depolymerization reaction, we attempted the hydrogenation of other nylons. A powder form of commercial nylon-6 (M w ∼ 11,000) was also depolymerized (85% conversion) under the optimized catalytic conditions (2 mol % of 1, 8 mol % of KO t Bu, 70 bar of H 2 at 150°C) to produce 6-amino-1-hexanol in 32% yield, with the remainder being oligomers as observed for the hydrogenation of nylon-6 resins (Table 1, entry 2, top row). Under the same catalytic conditions, but using complex 2, a slightly better yield of 6-amino-1-hexanol (36%) was observed. Interestingly, when a catalytic combination of 2 (5 mol %) and KO t Bu (20 mol %) was used for 72 h, maintaining the remaining conditions, 99% conversion of nylon-6 was observed and 6-amino-1-hexanol was detected in 55% yield based on GC analysis, with the rest being the oligomers ( Table 2, entry 2 d ). The products were also confirmed by HR-ESI-MS and NMR spectroscopy. Liquid chromatography (LC)-ESI-MS was used to estimate the amount of formed oligomers, showing a ratio of monomer, dimer, and trimer of 3:4:2 (see the Supporting Information for yields of the formed oligomers estimated by the LC-ESI-MS).
Given the satisfactory result of the hydrogenative depolymerization of nylon-6, we explored the hydrogenation of other polyamides. Commercial resins of nylon-12 (purchased from Sigma-Aldrich) were also hydrogenated using the optimized catalytic condition to obtain a mixture of 12-aminododecan-1-ol and smaller oligomers (dimer and trimer, Table 2, entry 3). Additionally, we also prepared some polyamides using the dehydrogenative coupling route reported previously by us 27 and then attempted to depolymerize them. Under the optimized catalytic conditions, 60% conversion of nylon-66 was observed and 1,6-hexane diol and hexamethylene diamine were detected in 20 and 25% yields, respectively (entry 4). A low molecular weight polyamide prepared by the dehydrogenative coupling of 1,4-phenylenedimethanol and 1,4-phenylenedimethanamine was hydrogenated to afford 80% diol and 85% diamine (entry 5). Some other synthesized poly(oligo)amides with the combination of aliphatic and aromatic parts were also successfully hydrogenated to produce diols and diamines (entries 6−8). Further, we extended this concept to the hydrogenation of polyurethanes, which are also commonly used plastics and can be prepared from CO 2 . 34 Thus, hydrogenation of polyurethanes (to diols, diamines, and methanol) is beneficial not only for the recycling purpose but also for the indirect conversion of CO 2 to methanol. Under the same catalytic conditions ( Table 2, entry 9), using complex 1 (2 mol %) and KO t Bu (8 mol %), commercial resins of a polyurethane were hydrogenated to afford 1,4-butanediol (70%), 4,4′-methylenebis(cyclohexan-1-amine) (66%), and methanol (45%). With the recent growth in the notion of  35 we also demonstrated as a proof of concept that the catalytic hydrogenation/dehydrogenation process can be utilized for the development of a closed loop cycle for polyamide synthesis/degradation. The monomers and oligomers obtained (after removal of DMSO) from the reaction mixture described in Table 2, entry 5, were refluxed in 1,4-dioxane solvent containing complex 5 (1 mol %) and KO t Bu (1 mol %) for 48 h to produce the corresponding polyamide of 1600 mol wt, similar to that used for hydrogenation (entry 5; also see the Supporting Information, section 13). Intrigued by the catalytic activity of the ruthenium pincer complexes 1 and 2 for the hydrogenation of polyamides, we explored the mechanism of the hydrogenative depolymerization process. The excellent ability of DMSO to disrupt the hydrogen bonding in amides and peptides was reported previously using NMR spectroscopy and density functional theory (DFT) studies. 36,37 We found that using complex 1 (2 mol %) and KO t Bu (8 mol %), only a trace amount of DMSO is hydrogenated under 70 bar of H 2 in THF (150°C, 24 h). We detected Me 2 S only in a trace quantity under the reaction conditions of hydrogenation of nylons using DMSO as solvent. These experiments indicate that complex 1 is not a good catalyst for the hydrogenation of DMSO, availing the catalyst for the hydrogenation of polyamides in DMSO as solvent. Additionally, reaction of complex 1 with 4 equiv of KO t Bu in DMSO-d 6 shows formation of a doubly deprotonated anionic complex A in THF reported by us previously. 30 30 This indicates that DMSO is likely not coordinated to the ruthenium center, making the vacant coordinate site available for the hydrogenation reaction.
In order to gain further mechanistic insight, we studied the reaction of nylon-6 (powder form, M w = 11,000) with the in situ prepared complex A in DMSO-d 6 using variable-temperature NMR spectroscopy (Scheme 1ii). No reaction was observed when the temperature was gradually increased from 298 to 360 K, but interestingly, at 400 K, the hydride signal in the 1 H NMR spectrum completely disappeared and the corresponding 31 P{ 1 H} NMR spectrum showed a mixture of species. This might indicate insertion of one of the carbonyl groups of nylon-6 into the Ru−H bond. Addition of H 2 (5 bar) to the same NMR tube at 298 K resulted in the formation of multiple ruthenium hydride species in the 1 H NMR spectrum featured by broad signals at δ −5.56, −5.78, −10.30, and −11.34. The signals at δ −5.56 and −5.78 agree well with the ruthenium trans dihydride complex C that was independently prepared by the reaction of complex 1 with KO t Bu and H 2 (Scheme 1i, with hydride signals at δ −4.73 (d, 2 J H,P = 14.9 Hz) and −5.12 (d, 2 J H,P = 17.6 Hz) in THF-d 8  A similar observation was made upon reaction of complex A with N-benzyl benzamide (see the Supporting Information, section 14.4). We also studied the reactivity of complex A with a diamide (Scheme 1iii). Reaction of an in situ prepared complex A in THF or toluene-d 8 with N 1 ,N 2 -bis(4methoxybenzyl)oxalamide (4 equiv) resulted in the formation of a mixture of complexes (see the Supporting Information, section 14.5). Interestingly, a single crystal obtained from this reaction mixture revealed a diruthenium complex B (see the Supporting Information, Figure S56), featuring N−H activation of both amide groups. Significantly, both the C O groups of the two amide moieties were found to coordinate to the two ruthenium centers, opening the amine side arm of the pincer ligand. Hemilability of the amine side arms in the case of PNNH pincer complexes of ruthenium and manganese was proposed a few times in catalytic (de)hydrogenation reactions. 29,38−40 Notably, this is the first crystal structure providing direct evidence for the hemilability of a PNNH pincer complex relevant to catalysis. The size of complex B as a dimer was also confirmed in the solution by DOSY NMR spectroscopy.
To obtain further insights into the mechanism, we employed DFT calculations at the ωB97M-V/def2-TZVPP/RIJCOSX/ SMD//M06-L/def2TZVP/GD3/W06 level of theory. We suggest that the first step involves deprotonation of the precatalyst 1 at the benzylic position of the ligand P-arm to generate the dearomatized complex D (as observed experimentally). 30 In-depth experimental thermodynamic analysis showed that, for a similar PNP−Ru complex, the dearomatized form can be in equilibrium with different Lewis or Brønsted acids to undergo reversible rearomatization. 41 From a computational point of view, the amido N−H bond of a simple test amide seems not to be acidic enough to generate higher concentrations of F (+9.1 kcal/mol) nor to be a good chelate (E, +9.6 kcal/mol), in contrast to recent findings involving formamides. 42 On the other hand, other deprotonated species such as N-arm dearomatized D′ (+5.9 kcal/mol) or N−H deprotonated D″ (−0.3 kcal/mol) should be accessible under the employed reaction conditions.
Our calculations suggest that the hydrogenation reaction proceeds via two connected catalytic cycles involving the dearomatized complex D and the trans-dihydride complex C formed upon hydrogen activation of D (Scheme 2). Importantly, although the reaction is run at high temperature, which was shown to disfavor the H 2 -splitting thermodynam- As the reaction proceeds, both alcohols and amines are generated and are very likely involved in this step. Although not rate-determining and oversimplified, we wish to emphasize that we find the lowest barriers for H 2 splitting from D′ (TS1A, 19.4 kcal/mol), with the test amide not acidic/basic enough to facilitate this step (TS1B, 26.2 kcal/mol). In the first catalytic cycle, the hydrogen activation step is followed by the hydrogenation of an amide bond by the trans-dihydride complex C to form hemiaminal and regeneration of the dearomatized complex D. Given the experimental evidence for ligand hemilability in the presence of the diamide in B (Scheme 1iii), we probed if this would be a viable pathway for amide hydrogenation. However, we find a stepwise Noyoritype mechanism to be more favorable with hydride transfer (TS2A1, 24.1 kcal/mol) preceding proton transfer from the N−H group (TS2A2, 21.4 kcal/mol). Indeed, hydride transfer via ligand hemilability is around 9 kcal/mol higher in energy (TS2B, 33.0 kcal/mol).
An intriguing aspect of amide hydrogenation is the fate of the intermediate hemiaminal. We calculate its dissociation to the amine and aldehyde to be strongly favorable (ΔG = −7.1 kcal/mol). A common notion is thus that the hemiaminal dissociates easily, as also implied by a mechanistic proposal by Sanford and co-workers. 46 Similarly, Prakash and co-workers could not detect any hemiaminal intermediate during CO 2 hydrogenation via formamides. 47 In contrast, recent work from the Bernskoetter group indicates that hemiaminal dissociation is kinetically hindered and relies on proton-relay catalysis to occur readily. 48 This is in agreement with previous DFT calculations from the same group, highlighting the TOFlimiting nature of this step that is facilitated by the employed iron complex or via proton relay by a solvent/reactant molecule. 49 We were thus interested how these observations translate to our system. Again, the test amide seems not to be acidic enough to serve as an efficient proton relay for the generation of the labile zwitterion (TS3D1, 43.1 kcal/mol). Interestingly, for Fe and Mo complexes in the absence of alcohols, 50 a C−N bond cleavage might be promoted via a Ncoordinated intermediate. In our case, such a pathway can be clearly ruled out due to very high barriers (TS3B and TS3C, 39.7 and 37.4 kcal/mol, respectively). On the other hand, C− N bond cleavage via O-coordination (TS3A, 30.7 kcal/mol) positions the leaving amido group for protonation by the side arm CH 2 group, as observed also for the reverse reaction in a PNP−Ru complex. 42 Although significantly lower in energy than the other explored pathways, it would still be linked to an energetic span of around 40 kcal/mol. Anionic alkoxide species are amenable to expel very weak leaving groups (e.g., hydrides in the case of alcohol dehydrogenation). 51 As amines are poor leaving groups, it might be suggested that basic conditions will favor C−N bond cleavage. This is at least in qualitative agreement with the above-reported example of Beller's anionic Mo−alkoxo complex 50 that can deprotonate the hemiaminal and facilitate subsequent C−N cleavage via H-bonding or the example from the Bernskoetter group where the strongly basic cocatalyst TBD (1,5,7-triazabicyclo[4.4.0]dec-5-ene) generates the highest catalytic activity. 48 In another Ru-catalyzed amide hydrogenation system, basic conditions were needed to promote C−N cleavage. 52 Finally, an analogy to the cleavage to the hemiaminal might be the scission of C−N bonds in oxaziridines, which was shown to be base-catalyzed. 53 In our case, excess KO t Bu is needed for reaching high turnover numbers. Moreover, as the reaction proceeds, the reaction medium will become more basic. The generated amines and

Scheme 2. Proposed Mechanism for the Hydrogenation of Amides Using DFT Calculations
Journal of the American Chemical Society pubs.acs.org/JACS Article alcohols could further provide a stabilizing H-bond network. A preliminary surface scan of the potassium salt of the hemiaminal showed that, in the absence of an H-bonding network, the C−N cleavage is uphill in energy but without any sizable transition state (see the Supporting Information). Clearly a more profound analysis, possibly involving explicit solvation, would be needed to assess with certainty the true nature of the C−N bond-cleaving transition state in such amide hydrogenation reactions. Finally, the aldehyde from the decomposition of hemiaminal is hydrogenated by C to form alcohol and regenerates the dearomatized complex D″ that would again be in rapid equilibrium with D.

■ CONCLUSION
In conclusion, we report here the hydrogenative depolymerization of conventional nylons such as nylon-6, other polyamides, and a polyurethane. The reaction is catalyzed by ruthenium pincer complexes using DMSO as a solvent, which plays a critical role in the process by disrupting the hydrogen bonding of the polyamide and at the same time remaining uncoordinated to the metal center, thus allowing the catalysis to occur. We also demonstrate an example where the diols and diamines obtained from the hydrogenative depolymerization process are used to synthesize the original polyamide via a ruthenium-catalyzed dehydrogenative coupling process, thus closing the loop of polyamide production. DFT calculations confirm the important role of metal−ligand cooperativity where the hydrogenation of amide bonds takes place via an outer-sphere mechanism. Methods for chemical recycling of nylons have been reported previously; however, this is the first time that a process based on hydrogenation, which is a green, sustainable, and atom-economic reaction, has been used to depolymerize conventional and robust nylons such as nylon-6. We believe that this unprecedented transformation, although not at a practical level yet, provides a new promising direction aimed at the highly desirable depolymerization of waste nylons as well as hydrogenation of other polyamides. Weizmann Institute of Science, Rehovot 76100, Israel; orcid.org/0000-0002-2320-0262; Email: david.milstein@ weizmann.ac.il