Lignin is one of the major components of available biomass in nature1,2,3. In lignin, there are three major types of aryl ether bonds of α-O-4 (~218 kJ/mol), β-O-4 (~289 kJ/mol), and 4-O-5 (~314 kJ/mol)1,2,3. Cleavage of C–O bonds in lignin can afford renewable aryl sources for fine chemicals. However, the high bond energies of these C–O bonds, especially the 4-O-5-type diary ether C–O bonds, make the cleavage very challenging1,2,3,4,5,6,7,8,9,10. Therefore, in fundamental research, the cleavage of aryl C–O bonds has attracted much attention11,12,13,14,15,16,17,18.

Cleavage of the α-O-4 and β-O-4 types C–O bonds has been studied, even under mild conditions by visible-light photoredox catalysis19,20,21,22,23. For cleavage of the 4-O-5-type diary ether C–O bonds, classical studies focused on the hydrolysis by supercritical water, and the hydrogenolysis using the model of 4-O-5 lignin linkages with poor selectivity2,4,5,10. Many aryl skeletons were destroyed. In selective cleavage methods, the use of stoichiometric alkali metals24,25,26 or electrocatalytic hydrogenolysis27,28,29 limited the large-scale applications because of the associated high costs.

In recent years, selective hydrogenolysis under comparatively milder conditions was developed by Hartwig, Grubbs, Wang et al. (Fig. 1a)30,31,32,33,34,35,36,37,38,39,40,41. The hydrogenolysis was accomplished by using [Ni], [Fe], or [Co] as catalyst, with H2 (1–6 bar) or LiAlH4 (2.5 equiv.) as reductant at 120–180 °C30,31,32,33,34,35,36. A higher than the stoichiometric strong base, such as NaOtBu/KOtBu/KHMDS (2.5 equiv.), is important for the selectivity (Fig. 1a, conditions A). Without using a transition-metal catalyst, a combination of Et3SiH/NaH (≥2.5 equiv.) and KHMDS (2.5 equiv.) at 140–165 °C achieved the selective hydrogenolysis (Fig. 1a, conditions B)37,38. More importantly, by MOF or Pd/C as catalyst without a base, the selective hydrogenolysis with H2 (10–30 bar) at 120–200 °C resulted in no more than 55% yields (Fig. 1a, conditions C)39,40,41. In these studies, a large amount of reductant, strong base, and/or the low yield are/is still the limiting factor(s) for the applications.

Fig. 1: Selective C–O bond cleavage of diaryl ethers.
figure 1

a Typically selective hydrogenolysis of diaryl ethers30,31,32,33,34,35,36,37,38,39,40,41. b This work: visible-light photoredox-catalyzed acidolysis of diaryl ethers and the following hydrolysis.

Thus, there exists a strong incentive to develop more practical methods for diaryl ether C–O bonds cleavage, toward the utilization of lignin as renewable aryl sources.

We developed the C–O bond cleavage of diaryl ethers containing a carboxylic acid group on the ortho position by a visible-light photoredox-catalyzed intramolecular aryl migration from an aryl ether to the ortho carboxylic acid group at room temperature (rt), and a following one-pot hydrolysis42. Thus, we envisioned the possibility of C–O bond cleavage of general diaryl ethers by an aryl acidolysis with an aryl carboxylic acid followed by hydrolysis, which would expand the scope of the special diaryl ethers largely, even to the model of 4-O-5 lignin linkages.

Specifically, in the aryl acidolysis, photoredox catalysis affords an aryl carboxylic radical A, then followed by its electrophilic attack on the diphenyl ether, and a single electron transfer (SET) with a proton and an electron (Fig. 2). However, the two issues make the transformation more challenging. First, although some intramolecular electrophilic attack reactions of aryl carboxylic acid radicals to arenes have been reported42,43,44,45,46,47, the intermolecular electrophilic attack reaction of aryl carboxylic acid radicals to arenes has not been successfully explored. The intermolecular electrophilic attack of aryl carboxylic acid radicals to arenes was proposed in the thermal decomposition of substituted dibenzoyl peroxides in diphenyl ether, in which the corresponding aryl benzoates were obtained in less than 39% yields with low selectivity. In addition, in comparison with the substituted dibenzoyl peroxides, the amount of arenes was huge because diphenyl ether was used as solvent48. Second, similarly to the intramolecular reaction, the formation of the more stable ester C–O bond should be the driving force of the diaryl ether C–O cleavage, as the C–O bond energies of an aryl ether and an ester are about 78.8 and 87–93 kcal/mol, respectively49,50. However, the possible intermediate B may lack a stronger driving force of C–O bond cleavage, compared with the six-membered ring intermediates in the intramolecular reactions42,43,44,45,46,47. As mentioned below, less than 10% yields of the product were obtained with the remaining starting material under the optimized reaction conditions for intramolecular reactions (Table 1, entries 1, 2). Perhaps a Lewis acid could activate the aryl ether by coordination of the oxygen atom.

Fig. 2: Designed pathway of visible-light photoredox-catalyzed acidolysis of diaryl ethers with an aryl carboxylic acid.
figure 2

Photoredox catalysis affords an aryl carboxylic radical (A), then followed by its electrophilic attack on the diphenyl ether to afford B, and a single electron transfer (SET) with a proton and an electron.

Table 1 Optimization of the reaction conditions.

Herein, we report visible-light photoredox-catalyzed C–O bond cleavage of diaryl ethers by an acidolysis with an aryl carboxylic acid and a following one-pot hydrolysis at rt (Fig. 1b). Phenols with some advantages than arenes in transformations such as aminations, the functionalization, cross-coupling reactions51,52, are obtained. The aryl carboxylic acid used for the acidolysis can be recovered. The key to the success of the acidolysis is merging visible-light photoredox catalysis using an acridinium photocatalyst (PC) and Lewis acid catalysis using Cu(TMHD)2 (Fig. 1b). Inspired by the significant contributions of Fukuzumi53,54,55, Nicewicz56,57,58, Sparr59,60 et al.61,62,63. on acridinium PCs, we investigated the use of an acridinium catalyst, in which an aryl ring with electron-withdrawing groups, instead of mesitylene, typically used in other acridinium catalysts53,54,55,56,57,58,59,60,61,62,63, on the 9-position to give higher efficiency. Furthermore, with this method, the model of 4-O-5 lignin linkages afforded phenol and 2-methoxyphenol in high efficiency. Compared with the developed selective hydrogenolysis in recent years, using H2O instead of a large amount of reductant afforded two molecules of more valuable phenols at rt.


Optimization study

With these considerations in mind, diphenyl ether (1a) and 4-methylbenzoic acid (2a) were studied as model substrates. Under our developed conditions for the intramolecular C–O bond cleavage42, using PDI64 or Acr+-Mes ClO4 (PC 1)53,54,55 as PC, with 10 mol% K2HPO4 as a base, under 450–455 nm blue LEDs irradiation, only <10% yields of phenyl 4-methylbenzoate (3a) and phenol (4a) were obtained (Table 1, entries 1, 2). Thereafter, a series of Lewis acids such as Cu(OAc)2, Cu(acac)2, Cu(OTf)2, Ni(acac)2, Fe(acac)2, Zn(acac)2, and Cu(TMHD)2 were studied (Supplementary Table 1, entries 1–6 and Table 1, entry 3). Cu(TMHD)2 slightly promoted the transformation. Adjustment of the wavelength of the blue LEDs to the maximum absorption of PC 1 (425–430 nm) induced a slightly increased reactivity (Table 1, entry 4). Other solvents such as MeOH, DCE, EtOAc, and acetone did not give any better results (Supplementary Table 1, entries 7–10). As Acr+-Mes ClO4 is susceptible to degradation in the presence of oxygen-centered radicals57, and de-N-methylation is also possible56, it is deduced that the generated carboxylic acid radical may induce the degradation of Acr+-Mes ClO4.

Subsequently, PC 2 and PC 3 were tried57, in which the N-phenyl were used instead of the N-methyl, and also with the tert-butyl on the 3- and 6-positions of the acridinium in the latter case. Both factors induced distinct higher efficiencies (Table 1, entries 5 and 6, 3a in 29 and 55% yields, with 4a in 22 and 49% yields).

Furthermore, the influence of substitutes on the 9-aryl ring was investigated. Since the complex procedure for the synthesis of PC 357, PC 4PC 9, with a similar skeleton to that of PC 3 but with different substituents on the 9-aryl ring, by a two-step synthetic procedure61, were investigated (Table 1, entries 7–12). Notably, the aryl rings with electron-withdrawing groups instead of mesitylene, typically used in other acridinium catalysts53,54,55,56,57,58,59,60,61,62,63, resulted in noticeably high efficiencies. To our delight, PC 9 with 2′-Cl and 4′-F on the 9-aryl ring resulted in 80% of 3a with 71% of 4a (entry 12). Although the smaller group of 2′-Cl compared with the methyl groups in mesitylene was used, X-ray crystallography of PC 9 unambiguously confirmed the angle of torsion between the 9-aryl ring and the acridinium ring (Fig. 1b), which is closely related with a longer fluorescence lifetime53,56. In addition, a variation of the tert-butyl groups from the 2,7-positions to the 3,6-positions resulted in decreased efficiency (entry 13). A further variation of the substituents on the 10-aryl ring revealed that the unsubstituted phenyl gave a slightly higher yield (entry 15). Decreasing the amount of Cu(TMHD)2 to 5 mol% resulted in obviously lower efficiency (entry 16, 3a in 47% yield with 4a in 45% yield). Without Cu(TMHD)2, the transformation only gave 12% of 3a with 7% of 4a (entry 17). These results indicate the transformation is promoted by Cu(TMHD)2. The amount of base did not influence the reaction efficiency, even without base (entries 18, 19). Other Lewis acids, such as Cu(OAc)2, Cu(acac)2, and Fe(acac)2 instead of Cu(TMHD)2, were investigated once again (entries 20–23), as slight differences during the initial investigation. The distinct promotion of Cu(TMHD)2 (entry 18) was further confirmed in comparison with the results of other Lewis acids (entries 20–23). Without base and Cu(TMHD)2, no reactivity was observed (entry 24). Control experiments indicate that a PC and visible-light irradiation are essential (entries 25, 26). The fluorescence lifetime and the redox potentials of PC 1PC 12 were determined (Supplementary Table 2). The data do not provide clear insight regarding the higher efficiency achieved using PC 9.

Evaluation of substrate scope

With the optimized reaction conditions, the substrate scope was investigated (Fig. 3). First, the influence of various substituents on the benzoic acid was investigated. 4-Methoxyl, 4-tert-butyl gave decreased reaction efficiencies (3ab, 3ac in 61–65% yields with 4a in 56–60% yields). The benzoic acid afforded 3ad in 55% yield with 4a in 50% yield. 4-Fluoro, 4-chloro, and 4-bromo induced excellent yields (3ae3ag in 87–91% yields with 4a in 82–84% yields). 4-Nitro, 4-aldehyde also resulted in high efficiencies (3ah, 3ai in 71–77% yields with 4a in 65–70% yields). The substituents on the 3-position were also studied. Similar to the substituents on the 4-position, fluoro and chloro resulted in high yields (3ak, 3al in 81–86% yields with 4a in 74–76% yields). Methyl gave a decreased yield (3aj in 61% yield). When the methyl was installed on the 2-position, no product was observed. 2-Fluoro resulted in lower efficiency (3ap in 63% yield). The phenomenon should be influenced by the steric factor. A series of aromatic heterocycles carboxylic acids containing pyridinyl, furyl, and thienyl were tested. Only thienyl carboxylic acids resulted in moderate efficiencies (3aq, 3ar in 49–54% yields with 4a in 40–45% yields).

Fig. 3: Substrate scope of carboxylic acids.
figure 3

Reaction conditions: 1a (0.50 mmol), 2 (0.60 mmol), PC 9 (3 mol%), Cu(TMHD)2 (10 mol%), CH3CN (5 mL), irradiation with blue LEDs (425–430 nm, 10 W) for 30 h. Isolated yields were reported. a60 h.

Next, the substituents on the aryl ring of the diphenyl ethers were investigated (Fig. 4). With electron-withdrawing groups such as methyl ester, trifluoromethyl, nitro, cyan, or acetyl on the 4-position of the aryl ring, 3a (65–74%) and the corresponding phenols with these electron-withdrawing groups (4b-4f, 62–72%) were selectively obtained in high yields. The results agree with the designed pathway of the electrophilic attack of the generated carboxylic acid radical, including the selective attack on the electron-rich aryl ring of the diphenyl ethers. 4-Bromo afforded two esters 3a (22%) and 3as (64%), and two phenols 4g (21%) and 4a (54%). The reason for the result may be the synergic effect of induction and conjugation of the bromo. For the 4-phenyl group, 3at and 4a were selectively obtained in 84% yields. The reason for the selectivity may be the 4-phenyl group stabilizing the generated radical intermediate after the electrophilic attack. Under standard conditions, 4-methyl and 4-methoxyl resulted in very low efficiency (<10%). When the reactions were conducted by a stop-flow reactor, 4-methyl resulted in 3au in 64% yield with 4a in 54% yield. And 4-methoxyl still resulted in low efficiency with 3av in 24% yield, which should be caused by the lower oxidation potential of 4-methoxyl-diphenyl ether (+1.39 V vs SCE) than that of the anion of 2a (+1.45 V vs SCE) to inhibit the aryl carboxylic radical generation (Supplementary Table 3). Methyl, methoxyl, dimethyl substituents on other positions afforded 3aw3ay in 68–80% yields with comparable yields of 4a. Symmetric dimethyl, dimethoxyl, dibromo, and dichloro, and asymmetric dichloro on the 3- or 4-positions resulted in good to high efficiencies (3az, 3ay, 3as, 3aA, 52–90%, 4h4k, 50–86%). When the methyl and methoxyl on 2- or 3-position, with 4′-ester, 4′-cynao or 4′-trifluoro, were investigated, the esters with methyl or methoxyl, as well as phenols with these electron-withdrawing groups, were obtained selectively in high yields (3aw, 3az, 3aB, 72–82%, 4b, 4e, 4c, 70–82%).

Fig. 4: Substrate scope of diaryl ethers.
figure 4

Reaction conditions: 1 (0.5 mmol), 2a (0.6 mmol), PC 9 (3 mol%), Cu(TMHD)2 (10 mol%), CH3CN (5 mL), irradiation with blue LEDs (425–430 nm, 10 W) for 30 h. Isolated yields were reported. a60 h. bCH3CN (5 mL) and PhCF3 (2.5 mL), 60 h. c1 (0.2 mmol), 2a (0.24 mmol), PC 9 (3 mol%), Cu(TMHD)2 (10 mol%), CH3CN (2 mL) with PhCF3 (1 mL), stop-flow reactor was used with blue LEDs (420–430 nm, 25 W) irradiation for 60 h.

Synthetic application

To demonstrate the potential application, a gram-scale reaction of 1a with 2e in a flow reactor and following one-pot hydrolysis was conducted. 4a was obtained in 80% yield, with 2e in 88% recovery rate (Fig. 5a). Meanwhile, the model of 4-O-5 lignin linkage (1v)30 afforded 4a (71%) and 4l (75%) in high efficiency, with 2e in 82% recovery rate (Fig. 5b). A comparatively complex model of 4-O-5 lignin linkage, 2-methoxyl-4-ethyl-2′-methoxyl-5′-methyl diphenyl ether33, was tested. The transformation was totally inhibited, which should also be caused by its lower oxidation potential (+1.22 V vs SCE) than that of the anion of 2a (+1.45 V vs SCE) to inhibit the aryl carboxylic radical generation (Supplementary Table 3).

Fig. 5: Gram-scale reaction and its application.
figure 5

a Gram-scale reaction. b Application. Due to the difficult purification for the isolated yields of 4a and 4l, after the acidolysis, the reaction was worked-up to afford 3et and 4a in 81% and 71% yields, respectively.

Mechanism studies

To gain insight into the reaction mechanism, a series of experiments were conducted. First, UV–vis absorption spectra of each component and the reaction mixture confirmed that PC 9 acts as a PC (Fig. 6a). Second, luminescent quenching experiments were conducted (Fig. 6b). The anion of 2a (4-MePhCO2nBu4N+), 2a, 1a, Cu(TMHD)2, and the anion of 2a with Cu(TMHD)2 quenched the excited-state PC 9*. The anion of 2a displayed an obviously larger quenching rate. 1a, Cu(TMHD)2 displayed lower quenching rates. The anion of 2a with Cu(TMHD)2 displayed a lower quenching rate in comparison with that of the anion of 2a. These results suggest there is little possibility for Cu(TMHD)2 to participate in the catalytic cycle of acridinium catalyst. Third, the pH value of the reaction mixture was determined as 3.61 or 4.30, with or without 10 mol% Cu(TMHD)2. Based on these pH values, Cu(TMHD)2 should promote the ionization of 2a by the salt effect65. As mentioned above, the transformation is promoted by Cu(TMHD)2 under base conditions. In addition, under base free conditions, 80% ester 3a with Cu(TMHD)2 (entry 19), and 13–56% ester 3a with Cu(OAc)2, Cu(acac)2, Ni(acac)2, Fe(acac)2 (entries 20–23) in comparison with no production of 3a without any these metal salts and base (entry 23), these results suggest that Cu(TMHD)2 also acts as a Lewis acid to promote the transformation.

Fig. 6: UV–vis absorption and fluorescence quenching experiments.
figure 6

a UV–vis absorption spectra of 1a, 2a, PC 9, and the reaction mixture. b Emission-quenching experiments of the anion of 2a (4-MePhCO2nBu4N+), 2a, 1a, Cu(TMHD)2, and the anion of 2a with Cu(TMHD)2.

Furthermore, with the addition of 2 equiv. of TEMPO as oxidant, compounds 5a and 5b, instead of 3 and 4, were obtained (Fig. 7). This result and the standard reaction suggest that the reaction of 1a with 2a generates possible intermediates B′, C, and C′ in a reversible manner under the optimized conditions. Under the standard conditions without TEMPO, the diaryl ether bond of B′ is activated by the Cu2+ to promote the equilibrium shifting to give 3a and 4a in high efficiency. With TEMPO as oxidant, the equilibrium shifting is promoted by TEMPO to give 5a and 5b via C and C57.

Fig. 7: Deducing possible intermediate B′a.
figure 7

With the addition of 2 equiv. of TEMPO as oxidant, compounds 5a and 5b, instead of 3 and 4, were obtained. This result and the standard reaction suggest that the reaction of 1a with 2a generates possible intermediate B’.

The thermodynamic feasibility of the photo-induced SET was analyzed based on the oxidation–reduction potentials. The oxidation potential of E4-CH3PhCO2•/4-CH3PhCO2–, E1a+•/1a, and the reduction potential of EPC 9/PC 9−• in CH3CN were determined as +1.45 V vs. SCE, +1.86 V vs. SCE66, and −0.47 V vs. SCE (Supplementary Figs. 11, 12, and 36), respectively. The excited-state energy E0,0 of PC 9 was determined as 2.63 eV (Supplementary Fig. 24). Therefore, the reduction potential of EPC 9*/PC 9−• was calculated as +2.16 V vs. SCE (EPC*/PC–• = EPC/PC–• + E0,0) (Supplementary Fig. 36). These reduction potentials indicate the prior formation of PC–• and the carboxylic acid radical42,43,44,45,46,47 by a SET between PC* and the carboxylic acid anion. Based on the electrochemical potentials of phenolic products66 and PC 9, phenolic products can be readily oxidized by PC 9. The stability of phenolic products under the reaction conditions may be attribute to the back electron transfer67,68. Furthermore, a quantum yield value of φ = 0.20 was determined. Thus, at this stage, whether the reaction proceeds via a photoredox catalytic pathway or a radical chain pathway could not be reached69.

Based on these results, the reaction mechanism is proposed as shown in Fig. 8. First, Cu(TMHD)2 promotes the ionization of 2a to afford 2a and a proton. Meanwhile, irradiation of PC with blue LEDs leads to the excited-state PC*. A SET occurs between PC* and 2a to generate the carboxylic acid radical A′ and PC–•. An electrophilic attack of A′ occurs on the electron-rich aryl ring of diphenyl ethers to form intermediate B′. A SET between B′ and PC–• in the presence of a proton with the promotion of Cu(TMHD)2 as a Lewis acid affords 3a, 4a, with the regeneration of PC.

Fig. 8: Proposed mechanism for the acidolysis.
figure 8

A selective electrophilic attack of the generated aryl carboxylic radical on the electron-rich aryl ring of the diphenyl ether.

In summary, we have developed visible-light photoredox-catalyzed C–O bond cleavage of diaryl ethers by an acidolysis and a following one-pot hydrolysis at rt. Two molecules of phenols are obtained from one molecule of diaryl ether in high efficiency. The aryl carboxylic acid used for the acidolysis can be recovered. The transformation is applied to a gram-scale reaction and the model of 4-O-5 lignin linkages. The applications of this approach to more complex models of 4-O-5 lignin linkages and the linkages in native biomass for utilization of lignin as renewable aryl sources are in progress.


General procedure for the C–O bond cleavage of diaryl ethers

To a quartz tube equipped with a magnetic stirring bar, PC 9 (0.015 mmol, 3.0 mol%, 9.60 mg), compound 1 (0.50 mmol) compound 2 (0.60 mmol), and Cu(TMHD)2 (0.05 mmol, 10 mol%, 21.5 mg) were added. The tube was evacuated and filled with argon three times with each cycle in 15 min. The freshly distilled solvent was then added into the tube via a syringe under an argon atmosphere, then stirred and irradiated with 425–430 nm blue LEDs at ambient temperature (19–21 °C) in a Wattecs Parallel Reactor (Supplementary Fig. 1) for 30–60 h. After the reaction, the solvent was removed in vacuo and the residue was purified by column chromatography (petroleum ether/EtOAc = 200/1–5/1) to afford compounds 3 and 4.

Full experimental procedures are provided in the Supplementary Information.