Introduction

Hydrogenation of alkenes is one of the most fundamental transformations in organic synthesis, typically requiring both a catalyst and a hydrogen source1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19. The most common hydrogen source is hydrogen gas (H₂), known for its 100% atom efficiency. However, safety concerns related to handling flammable and potentially explosive H₂ have led to the exploration of alternative hydrogen sources20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38.

Photochemistry offers a novel approach for converting organic substrates into alternative hydrogen sources. One notable strategy involves photocatalytic reduction of alkenes to radical anions, which subsequently react with Brønsted-Lowry acids or solvents as proton donors to facilitate catalytic hydrogenation39. For instance, Guo and Houk et al. employed the strong acid TfOH and thioxanthone (TX) to photo-catalytically reduce α,β-unsaturated carbonyl compounds by using p-xylene (solvent) as the proton source (Fig. 1a)40. Similarly, Polyzos et al. described an iridium-catalyzed reduction of alkenes, followed by a reaction with Brønsted-Lowry acids to produce alkanes (Fig. 1b)41. However, this technique is limited by the reduction potential of the targeted alkenes. An alternative strategy was then developed, where iridium complexes were usually used to photo-catalytically generate radical cations to undergo hydrogen atom transfer (HAT) reactions with alkenes to form hydrogen-adduct carbon radicals. This is followed by either protonation with a proton source or addition with a hydrogen radical source to achieve catalytic hydrogenation42,43,44,45,46. For example, Wenger et al. demonstrated that an iridium photocatalyst mediated the HAT reaction between triethylamine (TEA) and alkenes, with subsequent radical addition completing alkene hydrogenation (Fig. 1c)45. Moreover, Studer and coworkers developed an iridium-catalyzed system for generating phosphine radical cations that activated water molecules to form a PR₃–H₂O radical cation intermediate, effectively utilizing both hydrogen atoms from H₂O in the alkene hydrogenation through sequential heterolytic (H⁺) and homolytic (H•) O–H bond cleavages (Fig. 1d)46. Despite these advancements, replacing these rare and costly iridium photocatalysts with more abundant and inexpensive alternatives would significantly enhance the sustainability of hydrogenation processes.

Fig. 1: The previous work on photocatalytic transfer hydrogenation reactions.
figure 1

Strategies involve (a, b) photocatalytic reduction of alkenes and (c, d) hydrogen atom transfer (HAT) reactions with alkenes.

Full size image

Alcohol is earth-abundant and sustainable, which would be a good candidate as a transfer hydrogenation reagent. However, the main difficulty for its use in the hydrogenation of alkenes originates from challenging homolytic cleavage of O–H and α-C–H bonds due to their high bond dissociation energies (BDE = ca 100 kcal/mol)47. This is supported by currently reported photocatalysts, which solely mediate homolytic cleavage of the α-C–H bond in alcohols, generating reactive nucleophilic α-C-centered radicals (•CR2OH) for conjugate additions with Michael acceptors (Fig. 2)48,49,50,51,52,53. We propose that if the homolytic cleavage of the α-C(sp³)–H bond of alcohol [R₂C(H)OH] generates a hydrogen radical (H•) and a relatively stable alcohol radical (•CR₂OH), this alcohol radical could undergo O–H bond cleavage, rather than reacting with the alkene, producing a C = O double bond in aldehydes or ketones (R₂C = O) and a second H•. These two hydrogen radicals could readily add across the C = C double bond, enabling effective alkene hydrogenation (Fig. 2).

Fig. 2: The designed photocatalytic transfer hydrogenation of alkenes using alcohol as the hydrogen source.
figure 2

Previous work involves alcohol radicals as C-radical donor, whereas our work involves alcohol radicals as H radical donor.

Full size image

Transition metal-polyoxometalate hybrid clusters demonstrate good activity in mediating direct hydrogenation54,55,56, prompting us to explore whether a low-cost and earth-abundant decatungstate anion, [W10O32]4- alone can be used as a photocatalyst for transfer hydrogenation. Upon irradiation, the decatungstate anion is excited to a reactive state capable of abstracting α-H atoms from alcohols, generating alcohol radicals (•CR₂OH) and reduced decatungstate species, H+[W10O32]⁵-. The alcohol radical undergoes homolytic cleavage of the O–H bond, forming an aldehyde or ketone and a hydrogen radical (H•). This hydrogen radical then reacts with the C = C double bond of an alkene (R₂C = CR₂), producing a carbon-centered radical (R₂(H)–C(•)R₂). The photocatalytic cycle is completed when this alkyl radical undergoes HAT with H+[W10O32]⁵-, yielding the hydrogenated alkene [R₂(H)–C(H)R₂] and regenerating the decatungstate anion [W10O32]⁴-. Herein, we report a simple, mild, and efficient strategy for unsaturated compounds hydrogenation using alcohol as the hydrogen source and low-cost decatungstate as the photocatalyst. To our knowledge, this study represents the first instance of using tungsten-based catalysts for the transfer hydrogenation of unsaturated compounds. Notably, the decatungstate/alcohol catalytic system exhibits excellent chemoselectivity, enabling selective hydrogenation of alkenes and alkynes even in the presence of reactive groups like ketones and carboxylic acids. These features highlight the significant potential of this approach for practical and sustainable applications in organic synthesis.

Results

Catalysis and Substrate Scope

We initially investigated prenol 2a, a natural alcohol derived from biomass, as a hydrogen source for the transfer hydrogenation of 1,2-dibenzoylethylene (1a) (Fig. 3a). It is because the prenol-derived α-C radical (Me₂C = CHCH(•)OH) formed during α-C(sp³)–H bond activation can be stabilized by delocalization with the C = C double bond57,58,59. This stabilization likely reduces the reactivity of the radical toward 5a, thereby enabling the homolytic cleavage of the O–H bond to proceed efficiently. Upon 390 nm LED irradiation and with 5 mol% TBADT (tetrabutylammonium decatungstate) as the photocatalyst, the reaction yielded four types of products: the hydrogenated compound 4a (isolated yield: 86%), aldehyde 3a, α-C(sp³)–H bond addition compound 5a (isolated yield: 5%), and a mixture of dimeric by-products arising from coupling of C = C double bond (Figure S3 for their X-ray crystal structure). When TBADT was replaced with 2,4,6-triphenylpyrylium, a lower yield of compound 4a was observed. In contrast, using anthraquinone as a photocatalyst led to the formation of compound 5a (67%) as the major product. These findings speculate that the stability of the reduced photocatalyst, H+[photocatalyst]–, plays a key role in determining whether the prenol-derived α-C radical (Me2C=CHCH(•)OH) is capable of undergoing rearrangement to form 3a and H• before the HAT step by H+[photocatalyst]– occurs. The use of EosinY as a photocatalyst resulted in a mixture of dimeric by-products as major products, while 4a and 5a are minor products. EosinY was capable of photocatalytic C–H bond addition of 2-benzylidenemalononitrile with simple alcohols in quantitative conversion49. As such, we replaced compound 1a with 2-benzylidenemalononitrile, where EosinY catalyzed hydrogenation of 2-benzylidenemalononitrile with 2a to form [PhCH2CH(CN)2] in 21% yield, along with a C–H bond addition product in 73% yield (see the Supplementary Information Section 3 for details). This supports our abovementioned hypothesis that prenol-derived α-C radical (Me₂C = CHCH(•)OH) has relatively higher stability, which allows hydrogenation to occur, even EosinY prefers to mediate C–H bond addition.

Fig. 3: Optimization of reaction conditions.
figure 3

a Photosensitizers catalyzed hydrogenation of 1,2-dibenzoylethylene 1a with prenol 2a as the hydrogen source. The yields of the target product 4a are highlighted in bold corresponding to the photocatalysts examined. The optimized result is indicated. b Scope of alcohols used for the formation of the target product 4a. Reaction conditions: olefin (0.1 mmol), alcohol (5 equiv.), CH3CN (2 mL), TBADT (5 mol%), 390 nm LEDs (40 W), 6 h reaction time, room temperature. Yields of 4a are isolated yields. The highest yield of 4a resulting from the use of 2a as the alcohol source is indicated.

Full size image

Next, we evaluate the performance of primary and secondary alcohols in the TBADT-catalyzed transfer hydrogenation of 1,2-dibenzoylethylene (Fig. 3b). Prenol is the most effective, yielding 4a with the highest isolated yield. Allyl alcohol (2b) also performed well, producing 4a in 79% isolated yield, while natural alcohol geraniol (2c) afforded a comparable yield of 81%. Benzyl alcohol (2d) and pyridin-2-ylmethanol (2e) were efficient hydrogen donors, delivering isolated yields of 72% and 76%, respectively. Secondary alcohols such as 1-phenylethanol (2f) and 1-(pyridin-4-yl)ethan-1-ol (2g) achieved similarly high isolated yields (75% and 86%, respectively). Furan-2-ylmethanol (2h) is cumbersome in the catalysis based on the moderate isolated yield of 4a (48%). Conversely, simple alcohols such as ethanol (2i) and 2-propanol (2j) produced lower isolated yields of 4a (26% and 29%). These findings show that the nucleophilic alcohol α-C-centered radicals (•CR2OH) contain an unsaturated moiety (R= benzene, pyridine or alkene), having high efficiency in the hydrogenation of alkene, probably due to the stabilization of the radical, allowing homolytic O–H bond cleavage. Overall, these results demonstrate the feasibility of using a diverse array of primary and secondary alcohols as hydrogen sources in TBADT-mediated transfer hydrogenation, emphasizing the robustness and versatility of this approach.

With optimized reaction conditions established, we explored the substrate scope of olefins, as summarized in Fig. 4. The hydrogenation of 1,2-dibenzoylethylene (1a) with prenol produced 4a in an excellent isolated yield (86%). Encouraged by this result, we explored α,β-unsaturated carbonyl compounds. To our delight, the 4-phenyl-3-buten-2-one (1b) reacted efficiently to produce 4-phenylbutan-2-one (4b) in 76% isolated yield. Derivatives of 1b with electron-donating and electron-withdrawing substituents at different positions ortho- (1c, o-Me; 1d, o-F), meta- (1e, m-OMe), and para- (1f, p-OMe; 1g, p-OH; 1h, p-Cl; 1i, p-F; 1j, p-CF3) were successfully hydrogenated, yielding the corresponding products (4c-4j) in moderate to high isolated yields (54-85%). The results revealed that the electron-rich substituents at the β position (1c, 1e, 1f, 1g) led to slightly reduced isolated yields (54–63%). Replacing the phenyl group with a methyl (1k) or α-naphthyl substituent (1l) did not significantly affect product yields, producing 4k and 4l in 66% and 71% isolated yields, respectively. However, the hydrogenation of 1m, which contains a pyridine substituent, led to reductive coupling instead of hydrogenation to form 6m in 91% isolated yield. Moreover, 1-phenylbuten-1-one (1n) and 1,3-diphenylpropen-1-one (1o) were hydrogenated, giving a moderate isolated yield of 41% and 32%, respectively. Carboxylic acid and ester substituents enhanced reactivity, yielding 4p-4s in 67-87% isolated yields. Importantly, gram-scale hydrogenation of 1s (1.589 g, 7.4 mmol) afforded 4s in 82% yield (1.251 g), underscoring the robustness and practicality of this catalytic protocol for larger-scale synthesis. Terminal olefins, namely 3-buten-2-one (1t) and 1-phenylprop-2-en-1-one (1u) were effectively hydrogenated leading to 4t (86%) and 4u (84%) in good isolated yields, respectively. When more substituted terminal olefin, namely, 3-methylbut-3-en-2-one (1v) was hydrogenated, the isolated yield of 4v dropped to 64%. Cycloenone 1w and more substituted derivative 1x also reacted to form cyclohexanones 4w (76%) and 4x (43%), respectively. Next, reactive functional groups namely ester, nitro, and cyano groups in 1y-1aa were well compatible in catalysis (38-46%). The 1,1-disubstituted olefins with electron-withdrawing substituents were well compatible in catalysis (1ab, 43%), however, those with electron-donating groups were not tolerated. Pyridine substituent promoted the hydrogenation, where 1,2-di(4-pyridyl)ethylene (1ac) was converted into 4ac with an excellent isolated yield of 94%. Remarkably, the anti-lymphoma and leukemia cancer drug, namely Ibrutinib (1ad) was hydrogenated to produce 4ad in 56% isolated yield, demonstrating the utility of this catalytic method in the late-stage functionalization of pharmaceutical agents.

Fig. 4: TBADT photocatalyzed hydrogenation of unsaturated organic compounds with prenol as the hydrogen source.
figure 4

Reaction conditions: substrate a alkenes, b diazo compounds, c, d alkynes (0.1 mmol), prenol (5 equiv.), CH3CN (2 mL), TBADT (5 mol%), 390 nm LEDs (40 W), 24 h, room temperature; yields are isolated yields. *6 h reaction time. †400 nm LEDs (40 W). ‡Reaction at 70 °C. #Yields determined from crude 1H NMR spectra using 1,3,5-trimethoxybenzene as the internal standard.

Full size image

Subsequently, we investigated the hydrogenation of other types of unsaturated bonds. N = N double bonds encompassing both symmetrical and asymmetrical azo compounds (1ae-1al) were efficiently reduced under the optimized conditions, yielding 4ae-4al in good isolated yields (82–89%). It appears that the N lone pair of electrons does not affect the reaction mechanism. Phenylacetylene and its derivatives with various substituents (9a-9h) were efficiently converted to the corresponding terminal olefins (10a-10h) in 61-91% isolated yields. Furthermore, 4-phenyl-1-butyne and alkyl alkynes (9i-9l) gave hydrogenated products (10i-10l) in 73–83% yields, demonstrating compatibility beyond aromatic substrates. Alkynes bearing carbonyl substituents (9a, 9b, 9t, and 9u) underwent double hydrogenation to directly form alkanes (4a-4u, 69-83% isolated yields). The reduction proceeds in a stepwise manner: initial hydrogenation affords electron-deficient olefins (1a, 1b, 1t, and 1u), which remain highly reactive due to the strong activating effect exerted by the carbonyl groups. As a consequence, these olefins serve as transient intermediates and promptly undergo further hydrogenation under the catalytic conditions, ultimately affording the fully reduced alkane products.

In addition, we explored the hydrogenation of a variety of nitroarenes (12a-12i, Fig. 5). All substrates were efficiently reduced, affording the corresponding amines in good yields (61–83%). Notably, the antiparasitic drug Niclosamide (11h) and the calcium channel antagonist p-Nifedipine (11i) were successfully hydrogenated to 12h and 12i in 73% and 79% isolated yield, respectively. These results highlight the practicality of our method for the late-stage functionalization of pharmaceutically relevant molecules.

Fig. 5: TBADT photocatalyzed hydrogenation of nitroarenes with prenol as the hydrogen source.
figure 5

Reaction conditions: nitroarene (0.1 mmol), prenol (5 equiv.), CH3CN (2 mL), TBADT (5 mol%), 390 nm LEDs (40 W), 70 °C, 12 h; all yields are isolated yields.

Full size image

Mechanistic Investigations

To clarify the catalytic mechanism, deuterated benzyl alcohols PhCH2OD and PhCD2OH were reacted with 1ac (Fig. 6a). These reactions resulted in a 1:1 molar ratio of 4ac-d and 3d, and a 1:1 molar ratio of 4ac-d and 3d-d, respectively (Figs. S7 and S8), confirming that alcohols are the sole hydrogen source. Moreover, introducing the radical scavenger TEMPO (1 equiv.) significantly inhibited the formation of 4ac as the H• and alcohol α-C centered radicals were quenched by TEMPO (Figs. S10 and S11 showing the presence of the quenched radicals in HRMS). These findings indicate that the mechanism involves the activation of the α-C(sp3)–H bond of alcohols to form alcohol α-C-centered radicals in the first step. To further understand why the alcohol α-C-centered radical undergoes O–H bond cleavage rather than directly adding to the alkene during catalysis, density functional theory (DFT) calculations were performed using butenone 1t and prenol α-C-centered radical R2a as model substrates. Three potential hydrogenation pathways were explored.

Fig. 6: Experimental mechanistic studies.
figure 6

a Deuteration and radical capture experiments; b proposed mechanism for the transfer hydrogenation of alkenes.

Full size image

In Path I (Fig. 7), the R2a undergoes homolytic cleavage of its O–H bond and addition of H• at the β-position of 1t via TS-A (ΔGǂ = 23.3 kcal/mol) to form intermediate A and 3a (ΔG = −8.3 kcal/mol). In contrast, the addition of H• at the α-position of 1t is infeasible due to the high kinetic barrier (Path II, TS-B: ΔGǂ = 43.8 kcal/mol, see Fig. S16 in Supplementary Information). Intermediate A then undergoes HAT with H+[W10O32]⁵- via TS-A2 (ΔGǂ = 23.3 kcal/mol) to form 4t·[W10O32]⁴- complex Prod-A2 (ΔG = −76.9 kcal/mol). 4t is then dissociated from [W10O32]⁴- to complete a catalytic cycle (ΔG = −66.3 kcal/mol). Finally, Path I is exergonic by 66.3 kcal/mol.

Fig. 7: Reaction mechanism studied by DFT calculations.
figure 7

Calculated Gibbs free energy profile (kcal/mol) illustrating the reaction pathways of the prenol radical in hydrogenation Path I (red) and α-C(sp³)–H functionalization with butanone Path III (green).

Full size image

In the C–C bond formation pathway (Path III, Fig. 7), R2a adds to 1t via TS-C (ΔG = 13.6 kcal/mol), forming intermediate C (ΔG = −5.8 kcal/mol). It then undergoes HAT with H+[W10O32]⁵- via TS-C2 (ΔGǂ = −67.8 kcal/mol) to form [MeC(O)CH2CH2CH(OH)CH = C(Me)2]·[W10O32]⁴- complex Prod-C2 (ΔG = −84.3 kcal/mol). However, the C–H bond addition product [MeC(O)CH2CH2CH(OH)CH = C(Me)2] (Prod-C3) requires a higher activation barrier of 20.4 kcal/mol to dissociate from [W10O32]⁴- to complete the catalytic cycle (ΔG = −63.9 kcal/mol). Finally, Path III is exergonic by 63.9 kcal/mol. Overall, these results indicate that the catalytic system preferentially follows the hydrogenation pathway (Path I) rather than the α-C(sp³)–H functionalization pathway (Path III).

Based on the experimental data and DFT calculations, a plausible mechanism for decatungstate-photocatalyzed transfer hydrogenation of alkenes using alcohol as the hydrogen source is proposed (Fig. 6b). Upon 390 nm LED light irradiation, decatungstate is excited to [W10O32]4-*, and rapidly relaxes to its reactive state wO60,61. This wO species initiates the process by abstracting an α-hydrogen atom from alcohol to produce an α-carbon-centered alcohol radical (•CROH I)62,63,64 and H+[W10O32]5-, due to the high redox potential of wO/[W10O32]5- (E = +2.44 V vs SCE)65. The resulting radical undergoes first hydrogen atom transfer (Path I) to alkene, forming a radical intermediate III and aldehyde/ketone (II). The intermediate III subsequently undergoes a back-HAT with H+[W10O32]5-, producing the final alkane product (IV) and regenerating the starting catalyst [W10O32]4- to close the catalytic cycle. In addition, the alcohol radical •CROH (I) can undergo conjugate addition with alkenes via Path III to generate intermediate V, which reacts with H+[W10O32]5- to yield a byproduct VI.

In summary, we have developed a simple, mild, and efficient photocatalytic method for the transfer hydrogenation of unsaturated compounds using alcohols as the hydrogen source and low-cost decatungstate as the photocatalyst. This method demonstrates broad substrate compatibility, high chemoselectivity, and practicality under ambient conditions. Mechanistic studies reveal that the hydrogenation proceeds via a radical-mediated pathway involving the sequential homolytic cleavage of α-C(sp³)–H and O–H bonds in alcohols. DFT calculations further confirmed that the hydrogenation pathway is more exergonic than the side α-C(sp³)–H functionalization route, indicating it is thermodynamically favored, consistent with experimental observations. These findings highlight the significant potential of the decatungstate/alcohol catalytic system in practical organic synthesis, offering a sustainable approach for unsaturated compounds hydrogenation.

Methods

General procedures for transfer hydrogenation of unsaturated compounds

In a glovebox, unsaturated compounds (0.1 mmol), TBADT (5 mol%), prenol (0.5 mmol, 5 equiv.), and anhydrous acetonitrile (2 mL) were placed in a 10 mL Schlenk tube equipped with a magnetic stir bar. The reaction mixture was subsequently irradiated with an LED light (Kessil 40 W) at room temperature (unless otherwise noted) for 6–24 h, with 24 h being sufficient for all substrates. Then, the solvent was removed under reduced pressure using a rotary evaporator, and the resulting residue was purified by silica gel column chromatography. The appropriate eluent system (hexane, hexane/ethyl acetate, DCM/methanol or ethyl acetate/methanol) was selected based on the polarity of the target product, yielding the corresponding products For the detailed experimental methods described in this manuscript, please refer to the Supplementary Information.