Optimization of the reaction conditions. We examined the depolymerization of insoluble polyetheretherketone (PEEK) powder (Mw ~ 20800 and Mm ~ 10300 as catalog specifications) (1) with 2-ethyl-1-hexanethiol (2a) (2 equiv. relative to monomer unit) in 1,3-dimethyl-2-imidazolidinone (DMI) under various conditions (Table 1). The depolymerization was first performed using KOH, K3PO4, KOtBu, and Cs2CO3 as catalysts (10 mol% relative to monomer unit) at 150 °C to form the corresponding depolymerized products, dithiobenzophenone 4a, 1,4-hydroquinone (5) and a benzophenone-hydroquinone-type dimer intermediate 3 (Table 1, Entries 1-4). The use of Cs2CO3 was especially effective to form the final depolymerization monomers, 4a and 5, in good yields (Table 1, Entry 4), indicating that large countercation sizes as well as basicity promote the depolymerization. Encouraged by these results, we expected that bulky and strongly basic organic phosphazene bases such as P4-tBu (pKBH+ 30.25 in dimethylsulfoxide (DMSO))36-41 would be promising catalysts for this depolymerization (Fig. 2), which enhances the nucleophilicity of the counteranions.42-55 For example, Shigeno, Korenaga, and Kondo recently reported that P4-tBu activates an alkanethiol (pKa of n-BuSH: 17.0 in DMSO).56 In this study, highly basic phosphazene bases P4-tBu and P2-tBu (pKBH+ of P2-Et: 21.15 in DMSO) exhibited good catalytic activity in comparison with weaker bases such as DBU (pKBH+ 13.9 in DMSO) and P1-tBu-TP (pKBH+ 17.4 ± 1.2 in DMSO) (Table 1, Entries 5-8). Thus, the basicity and size of the catalysts are important for this reaction for enhancing the nucleophilicity of the counteranion. Increasing the amount of 2a from 2 equiv. to 2.5 equiv. enhanced the yield of 4a and 5 (Table 1, Entry 9). The P4-tBu catalyst loading was successfully reduced to 5 mol%, albeit with slightly decreased yield (Table 1, Entry 10). On the other hand, 20 mol% loading of P4-tBu had little effect (see Supplementary Information, Table S1, Entry 12). The reaction at lower temperatures (120 and 100 °C) decreased the yield (Table 1, Entries 11 and 12). Different solvents were surveyed and N,N-dimethylacetamide (DMAc) was effective whereas benzonitrile (PhCN) and xylene decreased the yield (Table 1, Entries 13-15). Finally, we found that the catalyst combination of P4-tBu (10 mol%) and K3PO4 (5 mol%) in DMAc gave 4a and 5 in excellent yields (Table 1, Entry 16).
Experimental mechanistic studies. To evaluate the present catalytic depolymerization reactivity, we monitored the yields of 4a and 5 during the reaction of PEEK powder 1 with 2a catalyzed by P4-tBu (10 mol%) and K3PO4 (5 mol%), P4-tBu (10 mol%), and K3PO4 (10 mol%) (Fig. 3, Supplementary Table S5). When P4-tBu catalyst was only used, depolymerization proceeded faster than when K3PO4 catalyst was used. The catalyst combination of P4-tBu and K3PO4 increased the rate of formation of 4a and 5 compared to the use of P4-tBu alone. These results indicate that the use of the P4-tBu catalyst allowed for rapid depolymerization. The K3PO4 assisted this catalytic activity of P4-tBu.
Next, we carried out NMR experiments to shed light on the combination of the thiol, P4-tBu, and K3PO4. The reaction of 4-tert-butylphenylthiol (0.02 mmol) and P4-tBu (0.02 mmol) in the presence of K3PO4 (0.02 mmol) was examined in DMF-d7 (0.5 mL) at 25 °C (see Supplementary Information, pages S20-S23). As a result, a 31P{1H} NMR spectrum suggested the formation of [P4-tBu-H]+ (see Supplementary Fig. S4) and mass peaks were also observed at m/z 634 in ESI-TOF-(+)-MS and m/z 165 in ESI-TOF-(-)-MS mass spectra, confirming the generation of [P4-tBu-H]+·[S(C6H4-tBu)]-. The same results were observed in the absence of K3PO4. On the other hand, in 1H NMR spectrum, the resonances for the aryl doublets (δ 6.69 and δ 7.14) were broadened in comparison with the case in the absence of K3PO4 (Fig. 4a). In addition, these signals were different from the combination of the thiol and K3PO4 (see Supplementary Fig. S3). These results indicated that [P4-tBu-H]+·[S(C6H4-tBu)]- was initially formed and the [S(C6H4-tBu)] anion coordinated to K3PO4 in the equilibrium state. Density functional theory (DFT) calculations suggested that the NBO charge of the phenylthiolate coordinating to K3PO4 is more nucleophilic than the non-coordinating one (see Supplementary Information, page S24). We assumed that this catalyst combination activates the thiol for the smooth depolymerization of PEEK.
Aromatic nucleophilic substitution with thiolate anions is known to proceed via the SNAr or SRA1 mechanism.57-59 In the SRA1 mechanism, thiyl radicals are thought to be involved. However, this catalytic depolymerization of PEEK gives hydroquinone which inhibits the generation of free radicals. We examined the depolymerization with 2a under P4-tBu/K3PO4 catalyst with 3,5-di-tert-butyl-4-hydroxytoluene (BHT, 2.5 equiv.), a radical inhibitor, at 150 °C for 16 h and observed the formation of 4a and 5 in high yields (Fig. 4b). These results ruled out the possibility of a radical pathway for the depolymerization. Of note, 2,2,6,6-tetramethylpiperidine 1-oxyl (TEMPO) as a typical radical scavenger was not suitable for this experiment, which converted 2a into the corresponding disulfide in the absence of PEEK (see Supplementary Information, page S25).
Proposed mechanism. A plausible pathway for the depolymerization catalyzed by P4-tBu and K3PO4 is shown in Fig. 5. The thiol is initially activated by P4-tBu to form a thiolate that interacts with K3PO4 in the equilibrium state. The sulfur center of the thiolate attacks the ipso-carbon bound to oxygen in the benzophenone unit in PEEK to form an anionic intermediate. The aryloxy anion is released to complete carbon–sulfur bond formation. K3PO4 may enhance the reactivity of the thiolate and assist in the release of the aryloxy anion. The generated aryloxy anion activates the thiol to form the organic thiolate and arenols. In fact, the basicity of arenols (pKa in DMSO of PhOH: 18.0; p-MeC6H4OH: 18.9)60 is higher than that of thiols (pKa in DMSO of n-BuSH: 17.0; PhSH: 10.3).56 This series of processes occurs repeatedly to generate the dithiobenzophenone 4 and hydroquinone (5).
Substrate scope.
With the optimum conditions using both P4-tBu and K3PO4 in hand, we examined the depolymerization of other super engineering plastics such as PSU, PEES, PPSU, PESU, and PEI which were analyzed by high-temperature GPC analysis prior to use (see Supplementary Information, page S26). These resins have cleavable aryl-oxygen bonds affected by electron-withdrawing groups in a manner similar to PEEK. Polysulfone (PSU) is composed of diphenylsulfone and bisphenol A. Since thiolate anions can cleave aryl-SO2 bonds,61-65 we were concerned that the present catalytic method may cleave the aryl-SO2 bond in the diphenylsulfone unit as well as the target C–O main chain. However, we found that polysulfone (PSU) pellets 6 (purchased from Sigma-Aldrich) and 6’ (purchased from Acros Organics) with different Mw (Mw 35000 and Mw 60000) in each of the catalog specifications underwent depolymerization with 2a via selective C–O bond cleavage to furnish the corresponding 4,4’-dialkylthiobenzosulfone (7a) and bisphenol A (8) in high yields (Table 2, Entries 1 and 2). In addition, there was no clear difference in the reaction rate between 6 and 6’ (see Supplementary Table S2). In the same way, polyetherethersulfone (PEES) pellets (9) or polyphenylsulfone (PPSU) powder (10) could be converted into 7a and hydroquinone (5) or 4,4’-dihydroxybiphenyl (11) in high yields (Table 2, Entries 3 and 4). In the case of polyethersulfone (PESU) (12) consisting of repeating oxy-diphenylsulfone units, three products 7a, 4-alkylthio-4’-hydroxy-diphenylsulfone 13, and bisphenol S (14) were obtained (Table 2, Entry 5). Polyetherimide (PEI) is composed of repeating structures of phenylene-1,3-bisphthalimide and bisphenol A. In this case, imide bonds in the phthalimide units may be cleaved by sulfur nucleophiles.66 Nevertheless, the C–O main chains were successfully cleaved selectively in the depolymerization of PEI pellets 15 with 2a, giving dithiofunctionalized phenylene-1,3-bisphthalimide 16 and 8 in good yields (Table 2, Entry 6).
We then explored the scope of thiols under the catalytic depolymerization of PSU pellets 6 (Fig. 6). 2-Phenylethanethiol or 2-mercaptoethanol underwent depolymerization at 100 °C to form a monomer-type product 7b or 7c and bisphenol A (8) in good yields. Triethoxysilyl-substituted propanethiol and cyclopentanethiol were used in the depolymerization and the corresponding depolymerization products 7d and 7e were obtained. Trimethylsilylmethylthiol gave 4,4’-dimethylthiodiphenylsulfone 7f and 8 in high yields via desilylation. Not only alkanethiols but also 4-tert-butylbenzenethiol could be utilized for depolymerization with only NaOtBu catalyst (20 mol%) to form the corresponding monomer 7g in 98% yield together with 8 quantitatively (see Supplementary Table S3). Instead of PSU, we attempted the depolymerization of PEEK powder with 4-tert-butylbenzenethiol under the P4-tBu/K3PO4 or NaOtBu catalyst in DMAc but the yield of the product, 4,4’-di(arylthiol)benzophenone 4b, was low (see Supplementary Table S4). At that time, a suspension containing precipitated 4b and its intermediates were obtained. Considering that the poor solubility of the products may have decreased the reactivity, we modified the conditions using a P4-tBu/Cs2CO3 catalytic combination in DMI to enhance the solubility. As a result, 4b was obtained in high yield, albeit with a long reaction time.
Utility of depolymerization method. To demonstrate the scalability of the depolymerization method, a gram-scale reaction of PSU pellets (6) with cyclopentanethiol catalyzed by 5 mol% of P4-tBu and K3PO4 was carried out. The desired products 7e and 8 were isolated in 78% and 75% yields, respectively (Fig. 7a). It is worth noting that this catalytic method was applicable to composite materials. Shaved powder of 30 wt% carbon-fiber reinforced PEEK (1’) underwent depolymerization with 2a to form 4a and 5 in good yields comparable to those obtained from neat PEEK powder (Fig. 7b). 30 wt% Glass-fiber reinforced PEEK (1’’’) was converted into 4a and 5 in the same way. In addition, small pieces of a baby bottle made up of PPSU (10’) as a representative consumer resin were transformed into depolymerized products, 7a and 11, in high yields.
Utility of products. Sulfur functional groups in the products can be utilized in various transformations to yield functional molecules. For example, 7e was applicable to the double cross-coupling with 4-decylaniline under palladium-catalyzed conditions67 to give the corresponding double amination product 17 (Fig. 8a). Double phenylation of 4b using diphenyliodonium trifluoromethanesulfonate and copper acetate catalyst in 1,2-dichloroethane at 100 °C, based on a reported method,68 gave benzophenone 4,4′-bis(diarylsulfonium) salt 18 in excellent yield (Fig. 8b). Such sulfonium groups are more reactive leaving groups than their parent sulfur functional groups. Thus, the sulfonium groups in 18 could be converted into fluorine by potassium fluoride and Kryptofix® 222 (4,7,13,16,21,24-hexaoxa-1,10-diazabicyclo[8.8.8]hexacosane) in N,N-dimethylformamide at 60 °C.69 Of note, the product, 4,4’-difluorobenzophenone (19), is used as a monomer for PEEK.70,71 PSU-depolymerized product 7g was also applicable to this transformation sequence. Double phenylation of 7g afforded diphenylsulfone 4,4′-bis(diarylsulfonium) salt 20 (Fig. 8c). Subsequent fluorination of 20 gave bis(4-fluorophenyl)sulfone (21) in 87% yield, which is a monomer of diphenylsulfone-based polymers such as PSU,72-75 PPSU,76-80 PESU,81,82 and PEES.83 In addition, 20 reacted with p-methoxyphenol in the presence of Cs2CO3 to give 4,4’-bis(p-anisyloxy)diphenylsulfone (22) in 79% yield.