Main

Selective oxyfunctionalization reactions of non-activated C–H bonds are the holy grail in synthetic chemistry because they require the reorganization of kinetically inert C–H bonds1. Haem-thiolate enzymes have garnered much interest because of their high reactivity, selectivity and broad substrate range2,3. Among them, unspecific peroxygenases (UPOs; International Union of Biochemistry and Molecular Biology (IUBMB) classification: EC 1.11.2.1) use only hydrogen peroxide (H2O2) to generate the catalytically active oxoferryl haem (Compound I) through the peroxide shunt pathway3,4, making UPOs a promising alternative to cytochrome P450 monooxygenases and their chemical counterparts. Nevertheless, UPOs suffer from the oxidative inactivation of their haem active sites at elevated concentrations of H2O2 (ref. 5). This instability has been addressed through the use of additional redox catalysts3,6,7,8 that facilitate in situ H2O2 supply in adequate concentrations. Most of these catalysts reduce O2 to H2O2 at the expense of artificial electron donors6,7,8 or co-substrates3. However, this necessity complicates the reaction schemes and causes serious issues (for example, poor atom economy and the accumulation of undesirable side products)3,6. These problems can be circumvented if the redox catalyst can use water as both electron donor and co-substrate in the reduction of O2 to H2O2 and oxidation of H2O to H2O2, respectively. This strategy is based on (1) water’s role as a solvent in UPO catalysis, (2) its abundance (55.55 M), (3) the near-unity atom economy of H2O2 production (94.44%) and (4) the formation of O2 molecules as a side product of water oxidation9. Despite these merits, water oxidation catalysts have the disadvantage of producing reactive oxygen species, which is detrimental to UPO-driven catalysis8,10,11.

Here, we report lignin as a multifunctional photocatalyst that accomplishes the challenging goal of sustainable UPO catalysis under solar visible light, that is, through in situ H2O2 formation by O2 reduction and H2O oxidation, the use of H2O as electron donor and the HO·-scavenging activity of lignin (Fig. 1). Solar energy holds great promise as an abundant, sustainable and clean source of chemical potential;12 hence, photocatalysts have been applied extensively to solar-to-chemical conversion13,14,15,16,17,18. Lignin is the second most earth-abundant biopolymer and is highly functionalized with various aromatic structures in lignocellulosic biomass19,20. To the best of the current knowledge, lignin lacks a well-defined primary structure, but rather represents a complex polymer of three phenylpropanoid (C9) units (that is, syringyl (S), guaiacyl (G) and p-hydroxyphenyl (H) units) that are primarily linked by multiple types of C–C/C–O bonds20,21,22. The pulp, paper and biofuel industries generate lignin as waste with an annual production22 of around 50 million metric tons. However, 95% of lignin is abandoned or combusted21,22 in biorefinery processes because of its complex, irregular and ill-defined chemical structure19,22,23. Recently, lignin materials have been studied (1) to prepare value-added aromatics24 and (2) to use them as building blocks19 for energy and environmental applications.

Fig. 1: Photoenzymatic oxyfunctionalization using lignin photocatalysts and peroxygenases.
figure 1

Illustration of a photoenzymatic oxyfunctionalization reaction through the synergetic integration of lignin photocatalysts and peroxygenase redox biocatalysts. The photoactivation of lignin energy materials (for example, lignosulfonate and kraft lignin) under visible light drives redox reactions, such as the reduction of O2 to H2O2 and the oxidation of H2O to H2O2 and O2. The in situ generated H2O2 reacts with peroxygenases, forming the oxoferryl haem (Compound I, redox centre) that catalyses stereoselective oxyfunctionalization reactions (for example, hydroxylation and epoxidation). Cys, cysteine.

Moving beyond these conventional approaches, we hypothesized that lignin polymers can perform photoredox reactions because they contain redox moieties that many molecular, organic photocatalysts share25,26. Inspired by their molecular structures, we investigated lignin’s electronic and photophysical properties and demonstrate here lignin-sensitized H2O2 formation (through O2 reduction and H2O oxidation) and O2 evolution (through H2O oxidation) under visible light. We further combined these photoredox reactions with peroxygenase biocatalysis to accomplish the photoenzymatic oxidation of non-activated C–H bonds and thus synthesize enantiopure alcohols and epoxides (that is, enantiomeric excess (e.e.) > 99%). Furthermore, we found that the lignin photocatalysts function as an antioxidant to suppress HO·-mediated UPO inactivation. This enabled the lignin–peroxygenase system to achieve record reaction time length and total turnover number (TTN) of 130 h and 81,000, respectively, among solar-assisted biocatalytic oxyfunctionalization studies.

Results

Electronic and photophysical properties of lignins

We chose lignosulfonate (LS) and kraft lignin (KL) as our model lignins because they are the two most available and commercialized lignin sources with annual production of 1.32 and 0.26 million metric tons, respectively22,27. These lignin macromolecules are isolated from lignocellulosic biomass by sulfite and kraft pulping processes, respectively19. The different chemical treatments lead to substantial structural changes in the native lignin22 and give LS and KL rather different chemical and redox properties. We elucidated their structural characteristics using 1H–13C heteronuclear single quantum coherence (HSQC) spectroscopy (Fig. 2). As shown in Fig. 2a,e, LS possesses a native β-aryl ether (β-O-4) linkage28 and sulfite-derived sulfonate functional groups28 adjacent to S and G aromatic units, whereas KL contains native structural units29 (for example, β-O-4, phenylcoumaran (β-5), resinol (β-β), dihydrocinnamyl alcohol and secoisolariciresinol) and kraft-derived moieties (for example, β-5 stilbene29, enol ether29, guaiacyl hydroxyethyl ketone30 and sulfhydryl groups30; Fig. 2b,e and Supplementary Figs. 1 and 2). Sulfur elemental analysis of LS and KL further supported the existence of sulfur functional groups (for example, SO3 and SH) in LS and KL (Supplementary Table 1). In addition to these different functional groups, LS and KL share common groups, including aliphatic hydroxy, phenolic hydroxy, C5-condensed hydroxy and carboxy groups (Supplementary Fig. 3), which we demonstrated by 31P nuclear magnetic resonance (NMR) spectroscopy31. In Supplementary Tables 2 and 3 we enumerate (1) the abundance of interunit bonding motifs per 100 C9 units, (2) the amount of functional groups (millimoles per gram lignin) and (3) the relative amounts of the three primary units in LS and KL, which we estimated from quick quantitative HSQC measurements30 and quantitative NMR analysis. We depict the plausible constitutional structures of LS and KL based on these multiple analytical results in Fig. 3a.

Fig. 2: Structural investigation of the lignin models.
figure 2

ad, 1H–13C HSQC spectroscopic analysis of lignosulfonate (a), kraft lignin (b), cellulolytic enzyme lignin (c) and lignin dehydrogenation polymer (d). 1H–13C HSQC spectra of the oxygenated alkyl region (left) and aromatic region (right) are displayed for each lignin model. See refs. 28,29,30,31,39 for peak assignments. Polysaccharides and unassigned moieties are shown in grey. e, The structural motifs present in the lignins are colour-coded to correspond to the assignments of the two-dimensional HSQC spectra in ad. A, β-aryl ether (β-O-4); A(G), β-O-4 linked to a G unit; A(S), β-O-4 linked to an S unit; B, phenylcoumaran (β-5); C, resinol (β-β); D, β-5 stilbene; X, cinnamyl alcohol; E, enol ether; H, p-hydroxyphenyl unit; G, G unit; G′, α-oxidized G unit; G″, α-sulfonated G unit; S, S unit; S′, α-oxidized S unit; S″, α-reduced S unit; S‴, α-sulfonated G unit.

Fig. 3: Electronic and photophysical properties of lignin photocatalysts.
figure 3

a, Energy diagram for lignin-sensitized O2 reduction to H2O2 and H2O oxidation to H2O2 and O2. Inset: plausible constitutional structures of lignosulfonate (top) and kraft lignin (bottom) based on multiple analytical results (including HSQC, 31P NMR, elemental analysis and Ellman’s assay) and the literature28,29,30. b, Transient photocurrent response of lignin photocatalysts at −0.028 V (vs RHE) under solar-simulated visible light. c, Electrochemical impedance spectroscopic analysis in the form of Nyquist plots under visible light at −0.028 V (vs RHE). Zre and Zim represent real and imaginary impedance, respectively. Light source: xenon lamp (λ > 400 nm, photon flux = 0.58 μE cm−2 s−1). The solid lines are fits to a Randles circuit model consisting of solution resistance (Rs), charge-transfer resistance (Rct) and a constant phase element (CPE).

To better understand the origin of lignin’s photocatalytic activity, we investigated its optical absorption using ultraviolet-visible (UV-Vis) spectroscopy. KL exhibited a stronger photoabsorption from the UV to visible light region compared with LS (Supplementary Fig. 4a). The energy gaps between the highest-occupied molecular orbital (HOMO) and the lowest-unoccupied molecular orbital (LUMO) of LS and KL were estimated to be around 2.74 and 2.93 eV, respectively, based on their Tauc plots (Supplementary Fig. 4b,c). Next, we used UV photoelectron spectroscopy to estimate the HOMO levels of the lignins. We found that the HOMO levels of LS and KL are 2.44 and 2.69 V (versus the reversible hydrogen electrode (RHE)), respectively (Supplementary Fig. 5). Thus, the LUMO energies of LS and KL were calculated to be −0.30 and −0.24 V (versus RHE), respectively, based on their HOMO−LUMO gaps (Fig. 3a).

In addition to molecular orbital levels as thermodynamic indices, the Gibbs energy of photoinduced electron transfer (ΔGPET) is instructive25 to evaluate the plausibility of a reaction between an excited photocatalyst and its substrate. Thus, we estimated the ground-state redox potentials, excited-state energies of the first singlet excited state (E0,0) and excited-state redox potentials of LS and KL. The ground-state oxidation potentials of LS and KL were determined to be 2.29 and 2.40 V (versus RHE), respectively (Supplementary Fig. 6 and Supplementary Table 4), and the E0,0 values were estimated to be 3.85 and 3.59 eV, respectively (Supplementary Table 4) from the intersection25 between the normalized absorption and photoluminescence (PL) spectra (Supplementary Fig. 7). Thus, the excited-state oxidation potentials of LS and KL were estimated to be −1.56 and −1.19 V (versus RHE), respectively (Supplementary Table 4), on the basis of their ground-state oxidation potentials and E0,0 values.

The separation/recombination and transfer of photoexcited charge carriers are key factors in photoredox reactions because charge carriers must be delivered to adjacent substrates. We performed steady-state and time-resolved PL spectroscopy on the lignins because these tools provide an approximate assessment of charge-separation properties25,32. As shown in Supplementary Fig. 8, the emission intensity of KL was much lower than that of LS. In addition, the time-resolved fluorescence decay profiles of LS and KL (Supplementary Fig. 9) show that the average fluorescence lifetime of KL (13.56 ns) was longer than that of LS (2.74 ns). These two results suggest that KL undergoes slower charge recombination than LS. Furthermore, we found that KL exhibits better charge-transfer ability than LS. According to our chopped-light chronoamperometric analysis (Fig. 3b), the photocurrent of KL was approximately four times higher than that of LS under visible light (λ > 400 nm). This photosensitization is a result of their HOMO–LUMO gaps in the visible range (Fig. 3a and Supplementary Fig. 4b,c). The difference in the charge-transport properties of the lignins is further supported by electrochemical impedance spectroscopy: the charge-transfer resistances of LS and KL were determined to be 25.4 and 12.0 kΩ, respectively, from their Nyquist plots, which were fitted to a Randles circuit model (Fig. 3c).

Lignin-sensitized reduction of O2 to H2O2

Building on the electronic and photophysical properties of the lignins, we investigated their capability to photocatalytically reduce O2 to H2O2 because (1) the reduction potential (Ered) of O2/H2O2 (Eq. (1))33 is more positive than the lignins’ LUMO levels (Fig. 3a) and (2) their negative ΔGPET indicates that the reduction reaction is an exergonic electron transfer event (Supplementary Table 4).

$${{{\mathrm{O}}}}_2 + 2({{{\mathrm{H}}}}^ + + {{{\mathrm{e}}}}^-) \to {{{\mathrm{H}}}}_2{{{\mathrm{O}}}}_2,E_{{{{\mathrm{red}}}}} = 0.68\,{{{\mathrm{V}}\,{\mathrm{vs}}\,{\mathrm{RHE}}}}$$
(1)

In addition to the thermodynamic requirement, the interaction between the electron acceptor (for example, dioxygen) and donor (for example, the lignins) plays an important role in redox catalysis. We confirmed the reversible interaction between O2 and the lignins using UV-Vis spectroscopy (Supplementary Fig. 10). To elucidate the origin of the interaction, we analysed LS and KL by 1H NMR spectroscopy. We observed the characteristic phenomenon of a paramagnetic shift of the signals of LS and KL only between 6.7 and 7.4 ppm upon purging with O2, which can be attributed to the aromatic hydrogen atoms of the lignins34 (Supplementary Fig. 11). Taken together, the UV-Vis and 1H NMR data support the reversibility of the lignin interaction with O2. We further observed that O2 molecules did not oxidize phenoxy-functionalized molecules to the corresponding quinone-based molecules (Supplementary Fig. 12).

Having substantiated the favourable non-covalent interaction between the lignins and O2, we exposed each lignin (1 mg ml−1) in O2-enriched potassium phosphate buffer (KPi, 100 mM, pH 7.0) to visible light from a solar simulator (λ > 400 nm). Photoactivated LS and KL gradually accumulated H2O2 at a rate of 80 ± 20 and 160 ± 30 mM per g catalyst per h (80 ± 20 and 160 ± 30 mM gcat−1 h−1), respectively (Fig. 4a). We attribute KL’s faster H2O2 production to its higher light absorption, charge-separation and charge-transfer abilities (Supplementary Figs. 4a, 8 and 9, and Fig. 3b,c). Control experiments in the absence of the lignins or light resulted in negligible H2O2 production (Fig. 4a and Supplementary Fig. 13). In addition, the action spectra of LS and KL were analogous to their absorption spectra (Supplementary Fig. 14), indicating that the photoactivation of LS and KL is the key step in H2O2 formation. We further found that LS and KL exhibited high stability under visible light using multiple spectroscopic tools (Supplementary Fig. 15). On the other hand, LS exhibited catalytic activity toward H2O2 decomposition under dark conditions, whereas KL showed negligible activity (Supplementary Fig. 16). This may seem unfavourable for the current system, but it keeps the in situ H2O2 concentration low, which is required for robust UPO catalysis3,5.

Fig. 4: Lignin-sensitized production of H2O2 under visible light and its mechanism.
figure 4

a, Visible light-driven generation of H2O2 by lignosulfonate and kraft lignin photocatalysts. Reaction conditions: lignin photocatalyst (1 mg ml−1) in O2-enriched KPi buffer (100 mM, pH 7.0) under dark or light conditions (λ > 400 nm, photon flux = 0.58 μE cm−2 s−1) at 298.2 K. b, Effect of 1,4-benzoquinone (left) and N2 (right) on rate of lignin-sensitized H2O2 production. Reaction conditions of the left panel: lignin photocatalyst (1 mg ml−1) with or without 1,4-benzoquinone (1 mM) in KPi solution (100 mM, pH 7.0) at 298.2 K. BQ, 1,4-benzoquinone. c, Photochemical formation of H2O2 driven by CEL, lignin DHP, guaiacylglycerol-β-guaiacyl ether (β-O-4 dimer), coniferyl alcohol and sinapyl alcohol. Reaction conditions: lignin (dimer) model or monolignol (1 mg ml−1) in O2- or N2-rich KPi buffer (100 mM, pH 7.0) under dark or light conditions (λ > 400 nm, photon flux = 0.58 μE cm−2 s−1). The error bars correspond to the standard deviation (n = 3). * and ** denote no statistically significant difference between groups according to one-way analysis of variance.

Source data

The reduction of O2 to H2O2 may proceed by (1) a two-step, single-electron reduction (that is, O2 → O2•− → H2O2) or (2) a one-step, two-electron reduction (that is, O2 → H2O2) route35. To elucidate the pathway of lignin-sensitized H2O2 production, we analysed the formation of superoxide ions (O2•−) by the nitro blue tetrazolium (NBT) assay9 (Supplementary Fig. 17a). The LS and KL photocatalysts produced O2•− under visible light in an O2-enriched environment, whereas a negligible amount of the radical ion was detected under N2-rich or dark conditions (Supplementary Fig. 17b). The addition of 1,4-benzoquinone (O2•− scavenger)36 to an O2-purged lignin solution decreased the rate of H2O2 formation by LS and KL photocatalysts (Fig. 4b, left), which supports the two-step reduction of O2 to H2O2.

Photocatalytic H2O oxidation to H2O2 and O2

We paid particular attention to the absence of artificial electron donors in the lignin-catalysed production of H2O2 because the incomplete depression of H2O2 formation (Fig. 4b, left) could originate from the oxidation of the H2O solvent to H2O2. This motivated us to evaluate the capability of photoactive lignins to oxidize H2O. Note that the oxidation of H2O to H2O2 (Eq. (2))9 induced by lignin photoexcitation is thermodynamically favourable because the HOMO levels of LS and KL are more positive than the oxidation potential (Eox) of H2O to H2O2 (Fig. 3a).

$$2{{{\mathrm{H}}}}_2{{{\mathrm{O}}}} \to {{{\mathrm{H}}}}_2{{{\mathrm{O}}}}_2 + 2({{{\mathrm{H}}}}^ + + {{{\mathrm{e}}}}^ - ),E_{{{{\mathrm{ox}}}}} = 1.76\,{{{\mathrm{V}}\,{\mathrm{vs}}\,{\mathrm{RHE}}}}$$
(2)

To exclude H2O2 production by the O2 reduction reaction, we purged a lignin solution with N2 gas before and during photocatalysis. The rate of H2O2 formation decreased to 25 and 66 mM gcat−1 h−1 for LS and KL, respectively (Fig. 4b, right). The rates are almost identical to those observed in the presence of 1,4-benzoquinone (O2•− scavenger) under O2-enriched conditions (Fig. 4b, left). These results suggest that the one-step reduction of O2 to H2O2 rarely occurs, and the oxidation of water is another photocatalytic pathway for H2O2 production on LS and KL. Furthermore, we confirmed the lignin-catalysed formation of O2 (Eq. (3)9 and Supplementary Fig. 18) using head-space gas chromatography.

$$2{{{\mathrm{H}}}}_2{{{\mathrm{O}}}} \to {{{\mathrm{O}}}}_2 + 4({{{\mathrm{H}}}}^ + + {{{\mathrm{e}}}}^ - ),E_{{{{\mathrm{ox}}}}} = 1.23\,{{{\mathrm{V}}\,{\mathrm{vs}}\,{\mathrm{RHE}}}}$$
(3)

O2 formation is thermodynamically favourable because Eox(O2/H2O) is less positive than the HOMO levels of LS and KL. Taken together, the lignin photocatalysts serve two functions: (1) the reduction of O2 to H2O2 and (2) the oxidation of H2O to H2O2 and O2.

Universality of lignin models for H2O2 formation

Because the molecular structures of LS and KL do not resemble that of native lignin in several ways, we investigated whether a native-like lignin model can execute the same photocatalytic reaction (that is, H2O2 production). Because cellulolytic enzyme lignin (CEL) is reported to be highly representative21 of native lignin, we synthesized CEL by the cellulase-driven hydrolysis reaction37 of lignocellulosic biomass feedstocks (for example, empty fruit bunches). 1H–13C HSQC and 31P NMR spectroscopy showed that CEL possesses three C9 units (that is, H, G and S) with β-O-4, β-5 and β-β linkages as well as carboxy groups (Fig. 2c,e, Supplementary Fig. 19 and Supplementary Tables 5 and 6). The exposure of a CEL-containing solution to visible light triggered the formation of H2O2 under O2-enriched/depleted conditions (Fig. 4c), but the reaction did not proceed in the dark.

We next explored the structural role of lignin in H2O2 production. We synthesized a lignin dehydrogenation polymer (DHP)38, an artificial lignin model, by peroxidase-mediated dehydrogenation reactions. Note that we gradually added monolignols to a peroxidase-containing solution for preferential end-wise polymerization38, thereby solving existing synthetic issues (for example, severely low numbers of β-O-4 linkages). 1H–13C HSQC and 31P NMR spectroscopy revealed that the DHP contains (1) β-O-4, β-5 and β-β linkages (in a ratio of 37:15:13), (2) S and G units, and (3) cinnamyl alcohol39, C5-condensed hydroxy and carboxy groups (Fig. 2d,e, Supplementary Fig. 19 and Supplementary Tables 5 and 6). The as-synthesized DHP exhibited the same photochemical behaviour as observed for LS, KL and CEL, that is, (1) light was an essential requisite for the reaction and (2) purging with O2 facilitated H2O2 production (Fig. 4c).

Building on the common capability of multiple lignin models (for example, LS, KL, CEL and DHP) to photochemically reduce O2 to H2O2, we speculated that the β-O-4 linkage may contribute to the photoredox reaction because all these models share this linkage, which is the most dominant bond23,24 in lignin. We prepared guaiacylglycerol-β-guaiacyl ether as a phenolic β-O-4-type lignin dimer model (Supplementary Fig. 20) and verified that the dimer produced H2O2 under light conditions, not dark conditions (Fig. 4c). However, coniferyl alcohol (4-hydroxy-3-methoxycinnamyl alcohol) and sinapyl alcohol (4-hydroxy-3,5-dimethoxycinnamyl alcohol), which are major monolignols in lignin synthesis and do not have a β-O-4 bond, did not accomplish the photochemical reaction (Fig. 4c). We suspected that the Cα–OH moiety of β-O-4 might be involved in O2 reduction because benzylic alcohols (that is, Cα–OH) have been reported21,22,40 to be oxidized to carbonyl groups (that is, Cα=O). We observed the formation of C=O bonds in the dimer model under light conditions, not dark conditions (Supplementary Fig. 21).

Enantioselective photoenzymatic oxyfunctionalization

Having substantiated in situ H2O2 generation by the lignin photocatalysts, we next investigated its potential for photobiocatalytic oxyfunctionalization reactions by coupling lignin-driven photocatalysis with peroxygenase-mediated biocatalysis in a one-pot process. We chose the UPO from Agrocybe aegerita, which was expressed recombinantly in Pichia pastoris (rAaeUPO)3, because of its versatility36 and high activity toward inert C–H bonds41. As shown in Supplementary Fig. 22, the LS–rAaeUPO and KL–rAaeUPO systems converted ethylbenzene into (R)-1-phenylethanol enantioselectively (>99% e.e.) under visible light (λ > 400 nm); we identified and quantified the product by gas chromatography–mass spectrometry (GC–MS) and GC, respectively (Supplementary Figs. 23a and 24a). CEL, DHP and guaiacylglycerol-β-guaiacyl ether also drove the rAaeUPO-driven production of enantiopure (R)-1-phenylethanol under light conditions (Supplementary Fig. 25). However, coniferyl alcohol and sinapyl alcohol did not prompt the enzymatic reaction (Supplementary Fig. 25), which we attribute to their negligible photochemical generation of H2O2 (Fig. 4c). It is noteworthy that the stereo- and chemoselectivity of the overall reaction were very high, in contrast to reported photoenzymatic reactions;3,6,8 in the previous studies the non-enzymatic oxidation of the substrate to a racemic product and overoxidation to acetophenone, respectively, were specified as reasons. The omission of lignin, light or substrate did not lead to synthesis of 1-phenylethanol (Supplementary Fig. 22) because of the imperceptible formation of H2O2 under the depleted conditions (Fig. 4a and Supplementary Fig. 13). However, the photobiocatalytic reaction occurred under O2-depleted conditions (Supplementary Fig. 22) because of in situ oxidation of H2O to H2O2 (Fig. 4b); oxygen isotopic analysis confirmed the formation of 18O-labelled (R)-1-phenylethanol in N2-enriched H218O solvent (Supplementary Fig. 26). Furthermore, we confirmed the influence of lignin concentration and photon flux on the kinetics of the photoenzymatic reaction (Supplementary Fig. 27).

We verified the interaction between the lignin photocatalysts and rAaeUPO enzyme by UV-Vis spectroscopy. As shown in Supplementary Fig. 28, the absorbance of rAaeUPO between 250 and 450 nm gradually decreased with increasing concentrations of LS and KL, which indicates that the lignins alter the electronic states of the aromatic functional groups (~280 nm) and haem prosthetic group (~418 nm) of the rAaeUPO. The rates of the photoenzymatic reactions were higher than those of lignin-sensitized H2O2 production (Supplementary Fig. 22 and Fig. 4b). This suggests that the interaction accelerates H2O2 consumption by UPOs because the concentration of H2O2 adjacent to lignin photocatalysts is higher than that away from the photocatalysts.

Next, we conducted a long-term photobiocatalytic oxyfunctionalization reaction of ethylbenzene using 8 mg ml−1 lignin photocatalyst, 50 nM rAaeUPO and 1.74 μE cm−2 s−1 visible light. The LS–rAaeUPO and KL–rAaeUPO hybrids produced the enantiopure product for at least 130 h (Supplementary Fig. 29), recording total turnover numbers of rAaeUPO (TTNrAaeUPO) of 81,000 and 72,600, respectively (Table 1, entry 1). Furthermore, the lignin–rAaeUPO couples were applied to other enantioselective oxyfunctionalization reactions (Supplementary Figs. 23 and 24 and Supplementary Table 7): as summarized in Table 1, the light-driven enzymatic systems achieved (1) the hydroxylation of propylbenzene, tetralin and cyclohexane (entries 2–4) and (2) the epoxidation of cis-β-methylstyrene (entry 5).

Table 1 Substrate scope of the photobiocatalytic oxyfunctionalization reaction using lignin photocatalysts and rAaeUPO biocatalysts

The lignin–rAaeUPO hybrids are much more robust than other water oxidation catalyst–UPO systems;8,10,36 these reported abiotic catalysts generate hydroxyl radicals that oxidatively inactivate the corresponding UPO, thus ceasing their biocatalytic reactions within 35 h. Thus, we hypothesized that water-oxidizing LS and KL energy materials do not produce hydroxyl radicals. To verify this we investigated the possible formation of HO· by the terephthalic acid (TA) assay42: photoactivated LS and KL produced a negligible amount of the radical under visible light (Supplementary Fig. 30), even though they thermodynamically favour HO· formation by two pathways, that is, H2O oxidation (Eq. (4)) and H2O2 reduction11 (Eq. (5)).

$${{{\mathrm{H}}}}_2{{{\mathrm{O}}}} \to {{{\mathrm{HO}}}}^ \bullet + ({{{\mathrm{H}}}}^ + + {{{\mathrm{e}}}}^ - ),E_{{{{\mathrm{ox}}}}} = 2.38\,{{{\mathrm{V}}\,{\mathrm{vs}}\,{\mathrm{RHE}}}}$$
(4)
$${{{\mathrm{H}}}}_2{{{\mathrm{O}}}}_2 + {{{\mathrm{e}}}}^ - \to {{{\mathrm{HO}}}}^ \bullet + {{{\mathrm{OH}}}}^ - ,E_{{{{\mathrm{red}}}}} = 0.79\,{{{\mathrm{V}}\,{\mathrm{vs}}\,{\mathrm{RHE}}}}$$
(5)

This result motivated us to investigate the HO·-scavenging activity of lignin photocatalysts. We synthesized a nanostructured haematite electrode (α-Fe2O3; Supplementary Fig. 31) to produce HO· radicals. As shown in Supplementary Fig. 32, the α-Fe2O3 photoanode generated HO· radicals in an O2-purged KPi buffer at 0.622 V versus RHE (0 V versus Ag/AgCl) under visible light (for a detailed explanation see the legend to Supplementary Fig. 32). In contrast, the radical was not detected when lignin photocatalysts were present in the buffer, which indicates the antioxidant properties of lignin materials.

Discussion and conclusion

In this study we identified the capability of lignin as a non-metallic photocatalyst. Based on the thermodynamic indices of lignin photocatalysts, we have unveiled lignin-sensitized O2 reduction to H2O2 and H2O oxidation to H2O2 and O2 under solar irradiation (λ > 400 nm). Identification of H218O2 molecules and high photostability of lignin further support that lignin acts as a photocatalyst. H2O is the most desirable electron donor in aqueous redox chemistry because it is abundant, biocompatible and simplifies reaction schemes; the thermodynamic favourability of the lignin photocatalysts for H2O oxidation distinguishes them from many other photocatalysts7,8,35 that rely on artificial electron suppliers (for example, formic acid and primary and secondary alcohols). In addition, lignin oxidizes H2O to H2O2 and O2; this leads to a build up of local concentrations of oxygen in the vicinity of lignin photocatalysts, which can escalate the rate of O2 reduction to H2O2. Based on the extensively accepted mechanism of photoredox catalysis25,43, we propose that a possible process for the formation of H2O2 and O2 in the presence of the LS and KL photocatalysts involves an oxidative quenching cycle, as depicted in Supplementary Fig. 33. Other lignin models (that is, native-like CEL and artificial DHP) and a β-O-4 dimer model (that is, guaiacylglycerol-β-guaiacyl ether) also achieve the photochemical reduction of O2 to H2O2. The negligible production of H2O2 driven by monolignols (that is, coniferyl alcohol and sinapyl alcohol) leads us to suspect that the β-O-4 linkage might play a role in the photocatalytic reaction. Follow-up studies should aim to (1) decipher detailed mechanisms of the role of lignin β-O-4 in the photochemical formation of H2O2 and (2) explore further the structure–reactivity relationship using, for example, operando spectroscopy and computational quantum mechanical modelling methods.

Furthermore, we have achieved UPO-catalysed enantioselective oxyfunctionalization reactions (for example, benzylic hydroxylation, alkane hydroxylation and styrene epoxidation) through the lignin-driven generation of H2O2 in situ under visible light. The lignin photocatalysts not only supply H2O2 at the requisite concentrations for the enzymatic activity of UPOs, they also interact with the UPOs, which means that the enzymes can readily consume H2O2 adjacent to the lignins, thus facilitating the oxyfunctionalization reactions. Among the photoenzymatic systems3,8 that use visible light and H2O, the lignin–UPO hybrids are more apposite for the production of enantiopure alcohols and epoxides (that is, >99% e.e.). Furthermore, lignin’s phenol-based monomeric units exhibit antioxidative function19,44, which addresses the inactivation of UPOs by HO· radicals; this property is a departure from the HO·-generating behaviour of other water oxidation catalysts8,10,36 that halts UPO-mediated biotransformations.

Lignin–UPO combinations compare favourably with cutting-edge photoenzymatic systems that use rAaeUPO under visible light (Fig. 5). Photoactive nanomaterials3,6,8 and π-conjugated acridine derivatives7 (for example, gold-loaded TiO2, graphitic C3N4, methylene blue, phenosafranine and flavin mononucleotide) require artificial electron donors (for example, methanol, formate or in situ regenerated 1,4-dihydronicotinamide adenine dinucleotide) to achieve meaningful TTNs (10,900–71,000). The TTNrAaeUPO values of the lignin photocatalysts are the highest even at the expense of H2O as a clean and desirable electron donor. Furthermore, the catalytic performances (for example, e.e. and catalytic turnover) of the lignin–rAaeUPO systems outweigh those of chemical catalysts and well-established P450 monooxygenases (Supplementary Table 8). Future challenges to enhance enzymatic productivity would be (1) the design of a biphasic aqueous–organic system to increase the concentration of hydrophobic substrates, (2) the chemical modification of lignins to control their energy levels and (3) the conjugation of lignin with redox mediators to boost H2O2 formation. In addition, in future work we will explore the photoenzymatic oxyfunctionalization reactions of complex molecules (for example, the synthesis of human drug metabolites45) using UPOs and other enzymes.

Fig. 5: TTNrAaeUPO values of photoenzymatic systems for ethylbenzene hydroxylation.
figure 5

Comparison of the TTN values achieved for ethylbenzene hydroxylation using the lignin photocatalysts of this study and state-of-the-art photoenzymatic systems that combine rAaeUPO and visible light. The photocatalysts include gold-loaded rutile TiO2 nanoparticles3,6, gold-loaded anatase TiO2 nanoparticles3, graphitic carbon nitride8, methylene blue7, phenosafranine7 and flavin mononucleotide7. CbFDH, formate dehydrogenase from Candida boidinii; NAD+, nicotinamide adenine dinucleotide.

In conclusion, in this work we have substantiated the multifunctional role of lignins, that is, synergistic H2O2 formation, H2O oxidation to O2, no necessity for artificial electron donors and HO·-scavenging activity, in a cascade process combining lignin photocatalysis and UPO biocatalysis. Currently, lignin is received as a waste product and combusted in refinery processes; however, shining light on lignin renders it productive in solar-to-chemical conversion. The renewable biopolymers absorb visible light to generate H2O2 and O2 through O2 reduction and H2O oxidation without requiring artificial electron donors. Furthermore, the enantioselective oxyfunctionalization of inert C–H bonds, a dream reaction in synthetic chemistry, is realized through the merger of lignin photocatalysts and peroxygenases. Lignin’s radical-scavenging activity plays a role in the protection of UPOs from HO·-mediated inactivation, leading to a new benchmark (TTNrAaeUPO = 81,000, >99% e.e.) Overall, this work establishes lignin as an energy conversion material for producing fuels and chemicals, presenting an example of waste-to-wealth conversion.

Methods

Lignin characterization

We recorded UV-Vis spectra using a V-650 UV-Vis absorption spectrophotometer (JASCO). To estimate the HOMO–LUMO gaps of the lignins, we used their absorption spectra to obtain their Tauc plots46, that is, (αhν)2 versus , where α is the absorption coefficient, h is Planck’s constant and ν is the photon frequency. UV photoelectron spectra were recorded using a Sigma Probe spectrometer (Thermo VG Scientific) with a photon energy of 21.2 eV (He I radiation) to estimate the HOMO levels of the lignins. Photoluminescence spectra were obtained using an RF-5301PC spectrofluorophotometer (Shimadzu). Time-resolved fluorescence decay profiles were acquired using a Fluorolog3 spectrofluorometer (HORIBA) with time-correlated single photon counting to determine the lignin fluorescence lifetimes. Fourier-transform infrared spectra were obtained using an FT-IR 200 spectrophotometer (JASCO). We employed a potentiostat/galvanostat (WMPG 1000, WonATech) to perform photoelectrochemical analyses of the lignin photocatalysts in a three-electrode configuration (fluorine-doped tin oxide glass working electrode (geometrical surface area = 2.52 cm2), Ag/AgCl reference electrode (3 M NaCl) and a stainless-steel counter electrode). A xenon lamp (Newport) equipped with an infrared water filter and 400 nm cut‐on optical filter was used as a solar-simulated light source. The electrolyte solution consisted of KPi buffer (100 mM, pH 7.0) containing 1 mg ml−1 lignosulfonate or kraft lignin. The lignin charge-transfer resistances were calculated by electrochemical impedance spectroscopic analysis under visible light using an impedance analyser (ZIVE SP1, WonATech).

Two-dimensional NMR spectroscopy

We recorded the 1H–13C HSQC spectra of the lignins using a 500 MHz AVANCE NEO NMR spectrometer (Bruker) equipped with an iProbe HR liquids probe, or a 400 MHz Fourier-transform NMR (FT-NMR) AVANCE III HD spectrometer (Bruker) equipped with a 5-mm multinuclear broadband observe (BBO) NMR probe. We dissolved lignosulfonate or kraft lignin (~50 mg) in [D6]dimethylsulfoxide ([D6]DMSO; 0.6 ml) in an NMR tube, selected hsqcedetgpsisp 2.3 as the HSQC pulse sequence, set the HSQC parameters and assigned the HSQC peaks according to the literature24. The central DMSO solvent peak (δH = 2.49 ppm, δC = 39.52 ppm) was used for calibration of the correlation peaks.

One-dimensional NMR spectroscopy

We recorded the 1H NMR spectra of lignins on a 400 MHz AVANCE III HD Nanobay spectrometer (Bruker) equipped with a 5-mm multinuclear broadband fluorine observe (BBFO) probe. We dissolved lignosulfonate or kraft lignin (2 mg) in a KPi D2O buffer (100 mM, pH 7.0, 1 ml) containing 3-(trimethylsilyl)-1-propanesulfonic acid. The lignin solutions were transferred to a screw-cap NMR tube and purged with O2 for 12 h. We used zg30 as the NMR pulse sequence, a recycle delay of 6 s and 256 scans. We recorded the 31P NMR spectra using a 500 MHz AVANCE NEO NMR spectrometer (Bruker) or a 400 MHz FT-NMR AVANCE III HD spectrometer (Bruker). A sample of lignin was dissolved in a dried solvent mixture consisting of anhydrous pyridine and deuterated chloroform (1.6:1, v/v) and activated molecular sieves (3 Å) to minimize the moisture content in the mixture. Next, a cyclohexanol solution (8 μl ml−1, internal standard) and chromium(III) acetylacetonate solution (10 mg ml−1, as relaxation agent to shorten the spin–lattice relaxation time of the phosphorus nuclei) were prepared using the dried solvent mixture. The internal standard solution (75 μl) and relaxation agent solution (75 μl) were then added to the lignin solution. Finally, 2-chloro-4,4,5,5-tetramethyl-1,3,2-dioxaphospholane (TMDP; 100 μl) was added to the mixture to drive the phosphitylation of lignin’s hydroxy groups. After shaking the mixture for several seconds, the mixture was transferred to a NMR tube and analysed immediately. We employed an inverse-gated decoupling pulse program (zgig) with an acquisition time of 1 s, a relaxation delay of 10 s, a scan number of ~128, a spectrum centre of 140 ppm and a spectral width of 100 ppm. We calibrated the 31P chemical shift by assigning the sharp signal arising from the product from the reaction between residual H2O and TMDP as 132.2 ppm.

H2O2 quantification

We dissolved a lignin model in KPi buffer (100 mM, pH 7.0) without any supplementary treatments (for example, filtering off insoluble residues of the lignin). We used a xenon arc lamp (equipped with an infrared water filter and 400 nm longpass filter) to irradiate the sample at 298.2 K. The amount of H2O2 formed by the lignin photocatalysts was determined spectrophotometrically using the 2,2′-azinobis(3-ethylbenzothiazoline-6-sulfonic acid) (ABTS) assay. The reaction sample was mixed with a colourimetric reagent solution (2.5 U horseradish peroxidase and 2 mM ABTS in KPi solution (100 mM, pH 5.0)). The absorbance of the mixture was monitored at 420 nm using a V-650 UV-Vis absorption spectrophotometer (JASCO). To obtain action spectra of the lignin photocatalysts, the incident light was monochromated using a 74004 Cornerstone 130 1/8 m monochromator (Newport).

NBT and TA assays

We used NBT and TA assays to detect O2•− and HO·, respectively36. We added 10 μM NBT or 300 μM TA to a lignin-containing solution. The solution was irradiated with solar-simulated visible light from a xenon arc lamp equipped with a 400 nm cut-off filter (Newport). After lignin-driven photocatalysis, we monitored the changes in the absorbance at 560 nm and the fluorescence intensity at 430 nm (λex = 315 nm) to detect NBT formazan and 2-hydroxyterephthalic acid (HTA), respectively.

Photobiocatalytic reaction and analysis

We prepared a reaction sample by dissolving lignin, rAaeUPO and substrate in KPi solution (100 mM, pH 7.0) in an Eppendorf tube (SPL Life Sciences). The tube was immersed in a water bath to maintain the reaction temperature at 298.2 K and irradiated with a xenon arc lamp (λ > 400 nm) to promote biocatalytic oxyfunctionalization reactions. After the photoenzymatic reactions, we extracted the oxyfunctionalized products using ethyl acetate, dried them over MgSO4 and quantified them using a 7890A gas chromatograph (Agilent) equipped with a flame ionization detector and a CP-Chirasil-Dex CB column (25 m × 0.32 mm × 0.25 μm). Detailed oven-temperature programmes are tabulated in Supplementary Table 3. We identified the enzymatic products using an ISQ QD300 gas chromatograph–mass spectrometer (Thermo Scientific) after extracting them using ethyl acetate. The e.e., turnover frequency (TOF) and TTN were calculated using Eqs. (6)–(8):

$${\mathrm{e.e.}}= \frac{{\left| {{{{\mathrm{Moles}}}}\;{{{\mathrm{of}}}}\;{{{\mathrm{an}}}}\;{{{\mathrm{enantiomer}}}} - {{{\mathrm{Moles}}}}\;{{{\mathrm{of}}}}\;{{{\mathrm{the}}}}\;{{{\mathrm{other}}}}\;{{{\mathrm{enantiomer}}}}} \right|}}{{{{{\mathrm{Total}}}}\;{{{\mathrm{moles}}}}\;{{{\mathrm{of}}}}\,{{{\mathrm{product}}}}}} \times {{{\mathrm{100}}}}$$
(6)
$${{{\mathrm{TOF}}}}_{{{{\mathrm{r}}Aae{\mathrm{UPO}}}}} = \frac{{{{{\mathrm{[Product]}}}}}}{{[{{{\mathrm{r}}Aae{\mathrm{UPO}}}}] \times {{{\mathrm{Time}}}}}}$$
(7)
$${{{\mathrm{TTN}}}}_{{{{\mathrm{r}}Aae{\mathrm{UPO}}}}} = \frac{{{{{\mathrm{Maximum}}}}\,{{{\mathrm{[Product]}}}}\,{{{\mathrm{at}}}}\,{{{\mathrm{a}}}}\,{{{\mathrm{given}}}}\,{{{\mathrm{time}}}}}}{[{{{{\mathrm{r}}Aae{\mathrm{UPO}}}}}]}$$
(8)

Antioxidant property of lignin materials

We carried out the photoelectrochemical reactions in a three-electrode configuration (α‐Fe2O3 as working electrode (geometrical surface area = 2.5 cm2), Ag/AgCl (3 M NaCl) as reference electrode and a stainless-steel counter electrode). These three electrodes were immersed in O2-purged KPi buffer (100 mM, pH 7.0) containing 0.3 mM TA with or without lignin energy materials (for example, kraft lignin and lignosulfonate). We conducted controlled potential photoelectrocatalysis at 0 V versus Ag/AgCl (0.622 V versus RHE) under solar-simulated visible light (λ > 400 nm, photon flux = 1.74 μE cm−2 s−1) to generate hydroxyl radicals. Note that the lignin was also irradiated with the visible light. After the catalytic reactions, we measured the fluorescence intensity of the electrolyte solution at 430 nm (λex = 315 nm) to quantify HTA.