Design of a rhenium-decorated mesoporous nickel phyllosilicate-derived Ni–Re/MCM-41 catalyst for efficient hydrogenation of levulinic acid to γ-valerolactone

Abstract

Herein, Ni and NiRe catalysts supported on mesoporous MCM-41 were synthesized through ammonia evaporation (AE) and impregnation (IM) routes to explore structure–activity correlations in the hydrogenation of levulinic acid (LA) to γ-valerolactone (GVL). The AE-derived nickel phyllosilicate (Ni-PS) framework provided strong interactions through Ni–O–Si linkages, leading to high dispersion and stabilization of Ni species. Incorporation of Re significantly improved reducibility, hydrogen activation, and the balance between acidic and metallic sites, resulting in enhanced catalytic efficiency. The optimized NiRe-PS catalyst exhibited a uniform nanostructure, strong Ni–Re synergy, and the highest metallic Ni fraction, which collectively promoted superior activity and stability. Under mild conditions (140 °C, 10 bar H₂), NiRe-PS achieved complete LA conversion and ∼96% GVL yield within 4 h, with a turnover frequency of 26.3 h⁻¹ (160 °C, 10 bar H₂) and with an apparent rate constant of 0.0059 min⁻¹. Mechanistic and isotopic investigations confirmed that both molecular and solvent-derived hydrogen contributed to the hydrogenation pathway. The exceptional activity, recyclability, and structural robustness of NiRe-PS demonstrate the potential of phyllosilicate-based bimetallic systems as efficient, non-noble catalysts for sustainable biomass valorization.

---

Green foundation

1. This work advances green chemistry by developing non-noble metal catalysts for efficient biomass conversion into biofuels, reducing dependence on precious metals and ensuring high catalytic activity and selectivity.

2. The NiRe-PS catalyst achieves 100% LA conversion and ∼96% GVL yield under mild conditions (140 °C, 10 bar H₂), with a high turnover frequency (TOF) of 26.3 h⁻¹. Isotopic studies confirm dual hydrogenation pathways, improving sustainability and efficiency.

3. Future research could explore renewable solvents, continuous-flow processes, and improved catalyst regeneration, reducing energy consumption and waste. Extending the catalyst's application to other biomass feedstocks will enhance scalability and environmental sustainability, further advancing biofuel production.

1 Introduction

The increasing world demand for renewable energy and sustainable chemicals has encouraged the development of the catalytic upgrading of biomass-derived feedstocks to fuel and value-added chemicals.^1–3 Among the various platform molecules derived from lignocellulosic biomass, levulinic acid (LA) is one of the most versatile intermediates due to its facile availability through acid-catalyzed hydrolysis of cellulose.^4–8 Its catalytic conversion into γ-valerolactone (GVL) has drawn particular interest because GVL is an important green solvent, fuel additive, and intermediate for the preparation of alkanes, pentanoic acid derivatives, and biodegradable polymers.^9–11 The efficient and selective hydrogenation of LA to GVL is therefore of exceedingly significant importance in the process development of integrated and sustainable biorefineries.^12,13

The hydrogenation of LA involves stepwise processes, namely the hydrogenation of the carbonyl group to 4-hydroxypentanoic acid (HPA) and subsequent intramolecular esterification to form GVL.^14–16 Noble metal catalysts such as Ru, Pt, Pd, and Ir are highly effective for this reaction; however, their high cost and limited availability restrict large-scale applications.^17–20 Conversely, Ni-based catalysts offer an economically favorable alternative due to their strong hydrogenation activity and abundance.^21,22 Nonetheless, conventional Ni catalysts suffer from sintering, oxidation, and leaching during hydrothermal reaction, which gradually results in loss of activity.^23–25 These hindrances serve to highlight the necessity for the formation of Ni catalysts with improved dispersion, reducibility, and metal–support interaction to maintain high activity and long lifetime under mild reaction conditions.^26,27

In this context, the selection of the support material is critical for stability and dispersion control of the Ni species.^28–30 MCM-41 was chosen as the support because of its high surface area, ordered mesoporous structure, and abundant silanol groups, which promote uniform metal dispersion and strong metal–support interactions. These properties enhance catalyst stability and accessibility of active sites during biomass hydrogenation.^31–34 The well-defined mesopores ensure not only uniform active metal dispersion but also prevent agglomeration during reduction and reaction.^35,36 In particular, when Ni is incorporated into the silica framework through the ammonia evaporation (AE) method, a unique nickel phyllosilicate (Ni-PS) structure is formed.^37,38 During this process, Ni²⁺ ions are slowly released from [Ni(NH₃)₆]²⁺ complexes and interact with silicate species to produce robust Ni–O–Si bonds.^39,40 These bonds provide robust strong interactions between Ni²⁺ and SiO₂, enabling progressive reduction of the Ni species to well-dispersed Ni⁰ nanoparticles and preventing sintering.^41,42 Therefore, Ni-PS catalysts have improved metal dispersion, structural stability, and thermal stability compared to conventional IM-derived Ni systems.

While phyllosilicate formation enhances the physical stability of Ni catalysts, their catalytic activity is also optimized by bimetallic modification.^43–47 The incorporation of a secondary metal can tune the electronic environment and surface acidity, and generate synergistic active sites.⁴⁸ Rhenium (Re) is a good promoter for Ni catalysts due to its oxophilic nature, variable oxidation states (Re⁷⁺ to Re⁰), and ability to promote hydrogen spillover.^49–52 In recent studies, Re species have been found to promote NiO reduction, generate additional acid sites, and increase the Ni–ReO_x interfacial area, which improve both hydrogenation and dehydration reactions during LA conversion.^53–55 While Re is relatively expensive, its low loading (<3 wt%) provides significant catalytic benefit, which justifies its use for high-value reactions. The resulting Ni–Re bimetallic systems are found to possess improved hydrogen activation, C=O bond polarization, and higher selectivity toward GVL.⁵⁶ Although several studies have examined Ni–Re catalysts for biomass hydrogenation, no systematic investigation has yet explored the incorporation of Re within a nickel phyllosilicate framework. Synergies between Ni–Re interactions and Ni–O–Si bonds from phyllosilicate might give an extremely efficient method for achieving high activity with long term durability in non-noble metal hydrogenation catalysts.

In this work, we report the synthesis of Ni and Ni–Re catalysts supported on MCM-41 via the simple ammonia evaporation (AE) and impregnation (IM) methods for investigating the role of phyllosilicate formation and rhenium modification in the hydrogenation of LA to GVL. Catalysts were extensively characterized using XRD, BET, TEM, H₂-TPR/TPD, NH₃-TPD, pyridine-DRIFTS, XANES/EXAFS, CO-DRIFTS, and XPS, and isotopic and kinetic measurements were performed to elucidate the molecular-level catalytic cycle. The NiRe-PS catalyst synthesized via the AE pathway possessed the lowest particle size, highest surface Ni⁰ content, and strongest Ni–ReO_x interfacial interaction, leading to outstanding catalytic performance. Isotopic experiments with 2-PrOH/2-PrOD and H₂/D₂ involved a bifunctional hydrogenation via both molecular and solvent-provided hydrogen pathways. This study highlights how the combination of the nickel phyllosilicate structure and Re modification provides an extremely active and stable non-noble catalytic system. The strong interaction between Ni²⁺ and SiO₂ and the Ni–Re synergistic effect provide enhanced reducibility, dispersion, and bifunctional activity, which renders the NiRe-PS catalyst to be a promising platform for biomass valorization and green hydrogenation reactions.

2 Experimental section

2.1 Catalyst preparation

Nickel phyllosilicate (Ni-PS) was synthesized using the ammonia evaporation (AE) method, as schematically demonstrated in Scheme 1. In a typical procedure, 6 mL of 25 wt% aqueous ammonia solution was mixed with 44 mL of deionized (DI) water under continuous stirring at 30 °C. Nickel nitrate hexahydrate (Ni(NO₃)₂·6H₂O) was then added to the mixture, forming a deep blue solution due to the generation of the [Ni(NH₃)₆]²⁺ complex. Subsequently, MCM-41 was introduced as a support and uniformly dispersed in the solution. Pristine MCM-41 exhibited a high BET surface area of 1056 m² g⁻¹, a pore volume of 1.1 cm³ g⁻¹, and an average pore diameter of 3.2 nm, which provided an ideal mesoporous framework for homogeneous dispersion of Ni species and facilitated the formation of nickel phyllosilicate during synthesis. The suspension was gradually heated to 80 °C to remove ammonia slowly, targeting a theoretical Ni loading of 10 wt% relative to the MCM-41 support, corresponding to a Ni/Si molar ratio of 0.1 : 1.

Scheme 1 Schematic representation of NiRe-PS and NiRe-IM synthesis routes showing nickel phyllosilicate formation and Re incorporation.

As ammonia evaporated, the pH of the solution decreased, leading to the controlled release of Ni²⁺ ions from the ammine complex. These Ni²⁺ species interacted with silicate groups from the MCM-41 framework, facilitating the in-situ formation of nickel phyllosilicate via the coordination of Ni with Si–O⁻ moieties. During this process, the solution color transitioned from dark blue to green, indicating the transformation of [Ni(NH₃)₆]²⁺ into the layered nickel phyllosilicate structure. When the pH reached approximately 7, the mixture was cooled to room temperature. The resulting solid was filtered, thoroughly washed with DI water to remove residual nitrates and ammonia, and then dried overnight at 110 °C. Finally, calcination at 500 °C for 4 h in air yielded the Ni-PS catalyst, in which nickel species are highly dispersed within the silicate framework as Ni–O–Si linkages.

For comparison, Ni/MCM-41 (Ni-IM) was prepared by the conventional impregnation (IM) method. An aqueous solution of Ni(NO₃)₂·6H₂O was impregnated onto MCM-41, followed by drying at 110 °C overnight and calcination at 500 °C for 4 h in air. Both catalysts were designed to achieve a nominal Ni loading of approximately 10 wt%. Subsequently, rhenium (Re) was introduced (2.5 wt%) onto both Ni-PS and Ni-IM supports via impregnation using an aqueous NH₄ReO₄ solution, which was performed according to the optimized Ni to Re molar ratio in our previously reported work in the Ni–Re system for the LA hydrogenation to GVL.⁵³ The obtained materials, denoted as NiRe-PS and NiRe-IM, were dried at 110 °C overnight and calcined at 300 °C for 0.5 h in air. The actual metal loadings of Ni and Re in the calcined catalysts were quantified by ICP-OES analysis (Table 1). Prior to catalytic evaluation, all samples were pre-reduced in flowing H₂ at 600 °C for 3 h to generate the active metallic phases.

Table 1 Physicochemical properties of Ni and Ni–Re catalysts

Sample	S BET ^a (m^2 g^−1)	V p ^b (cm^3 g^−1)	D p ^c (nm)	Elemental composition^d (wt%)		H2 consumption^e		H2 uptake^f (µmol g^−1)	NH3 uptake^g (mmol g^−1)	Crystallite size of Ni^0 ^h (nm)	Particle size^i (nm)
				Ni	Re	Actual	Theoretical
						(mmol g^−1)
a S BET obtained from the adsorption branch of the N2 isotherm. b V p calculated from N2 adsorption at a relative pressure of ∼0.99. c D p obtained from the desorption branch using the BJH method. d Elemental composition measured using ICP-OES. e H2 consumption calculated from the H2-TPR results in actual and theoretical calculations in temperature range of 100–800 °C. f H2 uptake calculated based on the H2-TPD profiles in the temperature range of 50–550 °C. g NH3 uptake calculated based on the NH3-TPD profiles in the temperature range of 50–550 °C. h Crystallite size calculated based on the XRD of the reduced catalyst metallic Ni^0(111) using the Scherrer equation. i Particle size obtained from TEM analysis.
Ni-IM	797	0.7	3.0	8.9	—	1.43	1.51	108	21	19.6	25.7 ± 9
NiRe-IM	738	0.6	3.0	8.1	2.6	1.44	1.87	145	25	17.9	15.0 ± 7.2
Ni-PS	649	0.7	3.7	8.0	—	1.62	1.36	314	23	3.9	3.8 ± 0.5
NiRe-PS	629	0.7	3.8	8.1	2.3	1.43	1.81	355	44	4.4	2.9 ± 0.5

---

2.2 Hydrogenation of levulinic acid and kinetic studies

The catalytic hydrogenation of levulinic acid (LA) to γ-valerolactone (GVL) was performed in a high-pressure stainless-steel batch reactor equipped with a glass liner (100 mL internal volume). The reaction temperature was monitored using a thermowell inserted into the reaction mixture and controlled by an external electric furnace fitted with a digital temperature controller. In a typical reaction, 1.0 g of LA (≥98%, Sigma-Aldrich) and 40 mL of 2-propanol (≥99.5%, Merck) were mixed with a specified amount of pre-reduced catalyst and loaded into the reactor. The reactor was tightly sealed and purged three times with H₂ to remove residual air before being pressurized to the desired H₂ pressure. In catalyst screening, the reaction temperature (120 °C), reaction time (2 h), H₂ pressure (10 bar), and catalyst loading (0.1 g) were systematically investigated to evaluate the superior activity of the NiRe-PS catalyst. After the reaction, the system was rapidly cooled to room temperature in an ice bath, and the liquid product was separated by filtration. Product analysis was carried out using a Shimadzu GC-2014 gas chromatograph equipped with a DB-5 capillary column (50 m × 0.32 mm × 0.52 μm) and a flame ionization detector (FID). The conversion of LA and yield of GVL were determined using calibration curves derived from authentic standards according to the following equations:

Kinetic studies were performed using the NiRe-PS catalyst and compared to the monometallic Ni-PS benchmark to evaluate intrinsic activity. Reactions were conducted at 120 °C and 10 bar H₂ using the catalyst in a powder form under vigorous stirring to eliminate internal and external mass transfer limitations. Kinetic measurements were intentionally carried out under mild reaction conditions at a low conversion level to ensure that the hydrogenation of levulinic acid proceeded in a kinetically controlled product, thereby reflecting the intrinsic catalytic properties and Ni–Re synergy. The reaction kinetics followed a pseudo-first-order model with respect to levulinic acid (LA) concentration. Under the applied reaction conditions, hydrogen was present in large excess, allowing the rate expression to be simplified to a pseudo first-order reaction represented by plotting ln(1 − X_LA) as a function of reaction time (t), which was used to determine the kinetic constant (k) and was calculated using eqn (3) and (4) as follows:

−ln(1 − X_LA) = kt (4)

To determine the activation energies (E_a) of the Ni-PS and NiRe-PS catalysts, the reaction was evaluated using temperature-dependent experiments conducted at 10 bar of H₂ for 2 h using 0.1 g catalyst, ensuring operation within a kinetically controlled product. The activation energy (E_a) was calculated by plotting ln r_LA as a function of and is mathematically equivalent to the conventional Arrhenius plot of ln k versus. Therefore, the activation energy was determined according to the Arrhenius relationship expressed in eqn (5) and (6) as follows:

The TOF was calculated, using eqn (7), based on GVL production adapted from previous studies^57,58 as follows:

All reactions were reproducible within ±3% variation in GVL yield, ensuring high experimental accuracy.

3 Results and discussion

3.1 Catalyst characterization

XRD patterns of calcined and reduced catalysts are shown in Fig. 1A and B, respectively. In calcined samples (Fig. 1A), Ni-IM and NiRe-IM exhibited typical reflections of NiO (2θ = 37.3°, 43.3°, and 62.9°).⁵⁹ On the other hand, Ni-PS and NiRe-PS showed broad diffraction bands typical of amorphous SiO₂ and the nickel phyllosilicate structure, with the absence of distinct NiO reflections, showing high dispersion and Ni species insertion into the silicate framework. Upon H₂ reduction (Fig. 1B), Ni-IM and NiRe-IM showed metallic Ni sharp peaks (2θ = 44.5°, 51.8°, 76.4°; JCPDS 44-0850), indicating the formation of large Ni⁰ crystals.⁶⁰ In contrast, Ni-PS and NiRe-PS exhibited very weak Ni⁰ reflections, consistent with the stabilization of the small crystallite size of Ni⁰ derived from the phyllosilicate structure after H₂ reduction treatment. Correspondingly, the calculated crystallite sizes were 21–25 nm for the impregnated catalysts, while those of the phyllosilicate-derived catalysts were significantly smaller (3–4 nm), reflecting the strong confinement effect of the AE method (Table 1).

Fig. 1 XRD patterns of (A) calcined and (B) reduced Ni-IM, NiRe-IM, Ni-PS, and NiRe-PS catalysts; (C) N₂ adsorption–desorption isotherms; and (D) pore size distributions derived from the BJH model of the calcined Ni-IM, NiRe-IM, Ni-PS, and NiRe-PS catalysts.

Nitrogen sorption data showed that all catalysts (Fig. 1C and D) retained the mesoporous structure of MCM-41 with surface areas of 629–797 m² g⁻¹ (Table 1). The S_BET for the Ni-IM catalyst was the highest (797 m² g⁻¹) and had a narrow pore diameter of 3.0 nm, typical of well-retained MCM-41 channels.^61,62 Upon Re addition, S_BET decreased slightly to 738 m² g⁻¹, showing some pore blocking due to Re oxide deposition. However, the Ni-PS catalyst had a lower surface area (649 m² g⁻¹) with a wider pore size (3.7 nm), consistent with in-situ growth of nickel phyllosilicate species onto mesopores of MCM-41. The incorporation of Re into Ni-PS further decreased the S_BET to 629 m² g⁻¹ and increased the pore diameter to 3.8 nm, and this may indicate more uniform dispersion of species in the channels. The pore volume remained relatively unchanged (0.7 cm³ g⁻¹) across all samples, confirming that the mesoporous framework of MCM-41 was preserved after metal incorporation.

ICP-OES analysis revealed that the Ni contents were close to the theoretical 10 wt% loading, with measured values of 8.9 wt% (Ni-IM), 8.1 wt% (NiRe-IM), 8.0 wt% (Ni-PS), and 8.1 wt% (NiRe-PS). This indicates high metal incorporation efficiency for both the AE and IM synthesis methods. Rhenium was also deposited efficiently, yielding 2.3–2.6 wt%, close to the nominal 2.5 wt% loading, confirming the effectiveness of the secondary impregnation step. These ICP results demonstrate that the synthesis procedures are well controlled and reproducible.

The distribution and morphology of Ni species after H₂ reduction were further investigated using TEM (Fig. 2a–d). The Ni-IM sample consisted of relatively large Ni nanoparticles with an average size of 25.7 ± 9.0 nm (Fig. 2e). Meanwhile, NiRe-IM consisted of smaller metal particles (15.0 ± 7.2 nm, Fig. 2f) because the Re particles were much smaller than the Ni particles, resulting in a reduced overall average particle size. In comparison, the AE-derived catalysts (Ni-PS and NiRe-PS) exhibited highly dispersed nanoparticles with average particle diameters of 3.8 ± 0.5 and 2.9 ± 0.5 nm, respectively (Fig. 2g and h), being well consistent with the XRD-derived crystallite sizes. HR-TEM characterization (Fig. 2i–l) revealed lattice fringes with interplanar distances of 0.204–0.219 nm, which are assignable to Ni(111) planes, signifying the presence of metallic Ni.^63–65 The SAED patterns (Fig. 2m–p) further confirmed the polycrystalline nature of reduced Ni nanoparticles, showing diffraction rings corresponding to Ni(111) and Ni(200). Elemental mapping (Fig. 2q–t) revealed a homogeneous distribution of Ni on the SiO₂ support derived from the Ni-PS structure, compared with the Ni-IM and NiRe-IM catalysts. The NiRe-PS sample exhibited a particularly uniform distribution, confirming the stabilizing effect of Ni particles derived from the phyllosilicate structure.

Fig. 2 (a–d) TEM micrographs; (e–h) particle size distribution profiles; (i–l) high-resolution TEM (HR-TEM) images; (m–p) selected-area electron diffraction (SAED) patterns; and (q–t) energy-dispersive X-ray spectroscopy (EDS) elemental mapping of the reduced (a, e, i, m, and q) Ni-IM; (b, f, j, n, and r) NiRe-IM; (c, g, k, o, and s) Ni-PS; and (d, h, l, p and t) NiRe-PS catalysts, respectively.

The H₂-TPR profiles (Fig. 3(a)) indicated the reduction behavior of Ni and Ni–Re catalysts. Ni-IM exhibited a broad reduction peak at 300–550 °C because of the stepwise reduction of NiO species with weak interaction with the SiO₂ support. NiRe-IM exhibited a more complex reduction profile with multiple peaks in the 300–450 °C range. The first, less intense peak is attributed to the reduction of Re oxides, whereas the latter, broader peak corresponds to the reduction of NiO species. This suggested the coexistence of NiO–ReO_x species and possible Ni–Re interactions that facilitated the reduction process.^66–68 In comparison, Ni-PS showed a high temperature reduction peak (>500 °C), characteristic of strongly bound nickel phyllosilicate species, which are more resistant to reduction than NiO species. It is interesting to note that the NiRe-PS sample showed a shift of the main peak to a slightly lower temperature (∼450–650 °C) than that of Ni-PS, suggesting that the addition of Re enhanced the reducibility of Ni by promoting the hydrogen spillover effect. In addition, the first peak (∼300–400 °C) observed in the NiRe-PS sample is ascribed to the reduction of rhenium oxides. The calculated H₂ consumption values are summarized in Table 1. Ni-PS exhibited the highest experimental H₂ consumption (1.62 mmol g⁻¹), which was significantly higher than the theoretical value. This result confirms the difficulty of the reduction process in the Ni-PS structure, likely due to the limitation of H₂ diffusion to the Ni species anchored within the silica framework. In contrast, after incorporating Re oxides into the Ni-PS and Ni-IM samples, the actual H₂ consumption was lower than the theoretical value, implying that the Re oxides were not completely reduced to metallic Re or were strongly interacting with Ni to form Ni–ReO_x species.^69,70 The H₂-TPD profiles presented in Fig. 3(b) clearly reveal the differences in hydrogen adsorption and metal–hydrogen interactions among the studied catalysts. The conventional Ni-IM catalyst exhibited only a weak and broad desorption peak, consistent with large Ni particles possessing fewer active surface sites, with a total H₂ uptake of 108 µmol g⁻¹ (Table 1).⁷¹ After Re incorporation, the NiRe-IM catalyst showed an increased uptake of 145 µmol g⁻¹, thus evidencing enhanced hydrogen activation through spillover between Ni and Re species along with improved metal dispersion. In contrast, the AE-derived Ni-PS catalyst exhibited broader desorption features at higher temperatures, suggesting stronger hydrogen binding and the presence of a wider range of surface metal sites. For Ni-PS, the total H₂ uptake was 314 µmol g⁻¹, nearly three times higher than that of Ni-IM. Further confirmation was obtained that the phyllosilicate framework stabilizes highly dispersed metallic Ni species after H₂ reduction. Among the catalysts examined, the Re-modified phyllosilicate catalyst (NiRe-PS) exhibited the highest H₂ uptake, amounting to 355 µmol g⁻¹. This was indicative of a synergistic enhancement of hydrogen adsorption and activation caused by the intimate Ni–Re interaction. The NH₃-TPD profiles (Fig. 3(c)) revealed two main desorption regions corresponding to weakly acidic sites (<200 °C) and strongly acidic sites (>300 °C). The Ni-IM catalyst had the lowest total acidity at 21 mmol g⁻¹, while that of Ni-PS was considerably higher (23 mmol g⁻¹).⁷² Incorporation of Re further enhanced the overall acidity: NiRe-IM reached 25 mmol g⁻¹, while NiRe-PS exhibited the highest acidity of 44 mmol g⁻¹. This substantial increase in the acid amount and density of NiRe-PS further supports the synergistic interaction between Ni and ReO_x species, which is favorable for bifunctional catalysis. The abundant accessible hydrogen sites and strong surface acidity of NiRe-PS account for its exceptionally high catalytic activity in the hydrogenation of levulinic acid to γ-valerolactone. Pyridine-DRIFTS (Fig. 3(d)) revealed the coexistence of Brønsted and Lewis acid sites. The band at 1450 cm⁻¹ is used for assignment to Lewis acid sites, and the band at nearly 1540 cm⁻¹ is attributed to Brønsted acidity. The most intense Lewis acid bands were observed for the NiRe-PS, indicating a selective enhancement of Lewis acidity upon Re incorporation, which is highly beneficial for hydrogenolysis and dehydration reactions in biomass conversion.⁷³ Overall, the combined characterization results found that the AE process gives rise to highly dispersed Ni nanoparticles derived from the phyllosilicate matrix with smaller crystallite size, higher surface hydrogen adsorption, and enhanced surface acidity compared to regular impregnation. In addition, Re incorporation also enhances further metal dispersion, reducibility, and acidity, wherein the optimum physicochemical properties are exhibited by NiRe-PS among the series.

Fig. 3 (a) Temperature-programmed reduction (H₂-TPR); (b) temperature programmed desorption (H₂-TPD); (c) temperature programmed desorption (NH₃-TPD) profiles; and (d) in situ pyridine-DRIFT spectra of the Ni-IM, NiRe-IM, Ni-PS, and NiRe-PS catalysts.

To gain additional information on the local coordination structure and oxidation state of the Ni and Re species, in-situ experiments were performed after H₂ reduction at 600 °C for 3 h (Fig. 4). Normalized XANES spectra at the Ni K-edge (Fig. 4(a)) similarly indicated clear differences between impregnated and phyllosilicate-derived catalysts. Ni-IM exhibited a white-line feature and NiO absorption edge position that indicates the presence of oxidic Ni prior to reduction. Upon H₂ reduction, the edge position moved to the lower energies towards that of metallic Ni foil, indicating reduction to Ni⁰.⁷⁴ In contrast, Ni-PS showed a higher edge energy and a feature broadening in the calcined state, reflecting the strong coordination of Ni in the silicate framework as Ni²⁺ species.⁷⁵ Upon reduction, the absorption edge of Ni in Ni-PS shifted completely to lower energy, overlapping with that of the Ni foil, confirming the full reduction of Ni²⁺ to metallic Ni⁰. This transformation indicates that the Ni²⁺ species in the phyllosilicate framework were completely converted to Ni⁰ under the applied reduction conditions, while the residual Ni–O–Si interactions likely contributed to the structural stabilization of the reduced nanoparticles, in good agreement with the TPR results (Fig. 4(a)). Interestingly, the NiRe-PS catalyst exhibited a greater edge shift towards Ni⁰ compared to Ni-PS, evidencing that Re facilitated the reduction of Ni via hydrogen spillover. This finding agreed well with the enhanced reducibility that had been observed in the H₂-TPR profiles. The first-derivative spectra (Fig. 4b) further support these findings, showing that all the calcined samples exhibited well-resolved Ni²⁺ features, whereas all the reduced Ni containing catalysts revealed the characteristics of the Ni⁰ state, indicating a complete reduction process. XANES spectra (Fig. 4(d)) at the Re L₃-edge indicated that calcined samples consisted primarily of Re in Re⁷⁺ or the NH₄ReO₄ stucture.⁷⁶ Upon H₂ reduction, partial reduction occurred, with the absorption edge shifting to lower energies, suggesting the coexistence of Re⁴⁺ and Re⁰ species (Fig. 4(e)). Notably, NiRe-PS exhibited a higher reduction shift compared to NiRe-IM, confirming that the phyllosilicate confinement favors intimate Ni–Re interactions, which stabilize the reduced Re species under reducing conditions. This dual reduction behavior is evidence of strong metal–metal synergy in NiRe-PS. The Fourier-transform Ni K-edge EXAFS spectra (Fig. 4(c)) for all reduced samples displayed a dominant coordination peak at approximately 2.2 Å (uncorrected), corresponding to Ni–Ni scattering, confirming the metallic nature of Ni species after H₂ reduction.⁷⁷ No significant Ni–O contributions were detected, indicating that the Ni²⁺ species originally integrated within the silicate framework were completely reduced to Ni⁰. This is consistent with the XANES results, which showed that the absorption edge of Ni in both Ni-PS and NiRe-PS overlapped with that of Ni foil, confirming the full reduction of Ni²⁺. The wavelet transform (WT) plots (Fig. 4(c-i–v)) further validated these findings, revealing single, intense maxima associated with Ni–Ni coordination and no visible features corresponding to Ni–O scattering. Compared with Ni-IM, the AE-derived Ni-PS and NiRe-PS catalysts exhibited stronger Ni–Ni signals with slightly shifted WT intensity toward higher k values, indicative of smaller, well-dispersed metallic clusters confined within the phyllosilicate matrix. These results confirm that the AE method promotes homogeneous Ni distribution and complete reduction while maintaining structural confinement that prevents excessive particle growth. At the Re L₃-edge, the EXAFS spectra (Fig. 4(f)) of the reduced catalysts exhibited a dominant Re–Re coordination peak centered around 2.4 Å (uncorrected), characteristic of metallic Re species. The detection of a bond distance in NiRe-PS near 2.1 Å confirmed that the ReO_x species present in the calcined precursors were partially reduced under the applied H₂ treatment.⁷⁸ This observation is further supported by the corresponding XANES spectra, where the absorption edge of the reduced catalysts shifted toward lower energy, approaching that of Re foil. The slightly weaker Re–Re amplitude in NiRe-PS compared to that in NiRe-IM suggests that Re nanoparticles were smaller and more finely dispersed within the phyllosilicate matrix, resulting in partial coordination with Ni atoms. These results imply that Re in NiRe-PS exists predominantly in a metallic state with strong Ni–Re interactions, contributing to its superior hydrogen activation and catalytic performance.⁷⁹ After H₂ reduction, NiRe-IM retained mainly Re–O features, but NiRe-PS had low-R features of Re–Ni scattering corresponding to Ni–Re alloy-like interactions or close interfacial proximity. The lower WT maxima observed for reduced NiRe-IM and NiRe-PS compared with the Re–Re reference can be assigned to Re–Ni coordination, suggesting strong interactions between Re and Ni and the formation of a Ni–Re interfacial synergy (Fig. 4(f-i–vi)). Collectively, these XANES/EXAFS findings complement the earlier TPR, TPD, and TEM findings. The synergism of Ni and Re is most prominent in NiRe-PS, where Re facilitates reduction of Ni and stabilizes highly dispersed metallic Ni and, simultaneously, forms intimate Ni–Re contacts. This dual role of Re not only enhances reducibility but also regulates acid–metal balance by generating additional Lewis acid sites, as confirmed by pyridine-DRIFTS. On the other hand, NiRe-IM catalysts exhibit weaker metal–metal interactions, leading to increased Ni particle size, weaker reducibility, and lowered surface acidity. Overall, atomic-scale structural evidence strongly confirms the higher catalytic performance of NiRe-PS, which synergistically integrates Ni phyllosilicate confinement and Re induced enhancement of reducibility and Ni–Re interfacial synergy.

Fig. 4 In-situ XANES and EXAFS analyses of catalysts after H₂ reduction at 600 °C for 3 h: ((a) and (d)) normalized and ((b) and (e)) first-derivative XANES spectra of ((a) and (b)) Ni K and ((d) and (e)) Re L₃-edges of the calcined and reduced NiRe-IM, Ni-PS, and NiRe-PS catalysts, respectively, and all the possible Ni and Re standards; in-situ Fourier transform of k²-weighted (c) Ni K-edge and (f) Re L₃-edge EXAFS spectra and wavelet-transform (c-i–c-v) Ni K-edge and (f-i–f-vi) Re L₃-edge EXAFS plots (no phase corrected distance) of the reduced NiRe-IM, Ni-PS, and NiRe-PS catalysts and all the possible Ni and Re standards. The spectra of the samples were recorded after the catalysts were reduced using H₂ at 600 °C for 3 h.

Time-resolved XANES experiments were performed in-situ during H₂ introduction in reduction treatment to monitor the evolution of the oxidation states of Ni and Re species (Fig. 6). Linear combination fitting (LCF) of the Ni K-edge and Re L₃-edge XANES spectra was performed to determine the evolving fractions of Ni and Re species, thereby identifying the temperature and holding time at which reduction begins (Fig. 5(a–h)). For the NiRe-IM catalyst, the time-resolved Ni K-edge XANES spectra (Fig. 5(a)) revealed a gradual shift of the absorption edge toward lower energy during H₂ reduction, accompanied by the progressive growth of the pre-edge feature typical of metallic Ni⁰. Linear combination fitting (LCF) analysis (Fig. 5(b)) showed that the fraction of Ni⁰ steadily increased with reduction temperature, confirming the continuous conversion of Ni²⁺ to Ni⁰ under the applied conditions after 450 °C.⁸⁰ The phyllosilicate-derived samples show a reversed trend. For Ni-PS (Fig. 5(c) and (d)), the Ni K-edge shifts only gradually toward the metallic reference during the temperature ramp, and the LCF fraction plots indicate that a significant portion of Ni remains in an oxidic (Ni²⁺) environment until considerably higher temperatures (and longer reduction times) are reached, particularly at 600 °C. This occurs due to the stable Ni–Si–O bonding and Ni incorporation in the phyllosilicate matrix, making the Ni species more resistant to reduction (consistent with the high-temperature TPR feature and the persistent Ni–O signal in EXAFS). Interestingly, the NiRe-PS sample (Fig. 5(e) and (f)) exhibits an intermediate kinetic behavior and the Ni reduction is shifted to a lower temperature relative to Ni-PS, with a faster accumulation of the Ni⁰ fraction, although not as rapid as in NiRe-IM. The evolution profile indicates that the presence of Re promotes Ni reduction, enabling faster transformation to the metallic Ni state compared with Ni-PS alone. This enhancement can be attributed to hydrogen spillover facilitated by Re, which accelerates the reduction of Ni species. This acceleration of Ni reduction in NiRe-PS is consistent with the H₂-TPR results (Fig. 3(a)), revealing a clear shift of the main reduction peak to lower temperature upon Re addition, confirming that Re facilitates hydrogen activation and promotes the reduction of Ni²⁺ species strongly bound to the support. The agreement between TR-XANES and TPR results demonstrates that Re acts as an efficient promoter by enhancing hydrogen spillover and electronic interaction with Ni, thereby accelerating the overall reduction kinetics. In addition, the presence of Re within the PS matrix facilitates Ni reduction over the inherently stable PS material; however, the confinement effect of the phyllosilicate continues to moderate the reduction kinetics compared with the IM samples. This observation is fully consistent with the EXAFS/WT analysis, which reveals a mixed Ni environment in NiRe-PS (with both Ni–O and Ni–Ni contributions), as well as with the TEM results showing very small, highly dispersed Ni particles rather than large metallic domains.

Fig. 5 Stacked plots of time-resolved in-situ XANES data collected during H₂ reduction: normalized Ni ((a), (c) and (e)) K-edges and (g) Re L₃-edge XANES spectra, and corresponding fractions of Ni ((b), (d) and (f)) and Re (h) components calculated by linear combination fitting (LCF) of the ((a) and (b)) NiRe-IM, ((c) and (d)) Ni-PS and (e)–(h) NiRe-PS catalysts.

Fig. 6 Time-resolved CO-DRIFTS spectra of the reduced (a) Ni-IM, (b) NiRe-IM, (c) Ni-PS, and (d) NiRe-PS.

The TR-XANES for Re (Fig. 5(g) and (h)) reveals a multistep reduction pathway for Re species in the NiRe samples. The Re is initially present mostly in high oxidation states (Re⁷⁺) in the calcined materials; an intermediate reduced Re species (consistent with Re⁴⁺ or lower-valent ReO_x) develops during the H₂ ramp, and a Re⁰ component is developed at the highest temperatures and times. Remarkably, the extent of reduction of Re to lower oxidation states is method dependent and NiRe-PS shows a greater extent of partial Re reduction and a greater percentage of reduced Re species at comparable reduction times and temperatures (Fig. 5(h)). The lower and broader partial reduction of Re in NiRe-PS agrees with the hypothesis that the phyllosilicate environment promotes closer Ni–Re contact and more effective hydrogen spillover between the two metals, enabling tandem reduction of Re and Ni to mixed oxidation states and generating active interfacial sites (Re–Ni) observed by EXAFS. Several mechanistic and practical implications can be derived from the TR-XANES results. First, the observed two-step Ni reduction process (rapid at the surface and slower in the bulk) demonstrates that catalytic activation is not simply an “on/off” phenomenon. Surface Ni species can be reduced to the metallic state and become catalytically active sites. Second, Re selectively lowers the barrier for the surface reduction step by providing sites for H₂ dissociation and facilitating hydrogen spillover, thereby shifting the Ni⁰ fraction curves toward earlier times and temperatures in the LCF plots. Third, although Re in the PS matrix promotes complete reduction of Ni²⁺ to metallic Ni⁰, partially reduced ReO_x species stabilized within the PS host provide the catalytically relevant Lewis acid sites. As a result, NiRe-PS contains metallic Ni⁰ nanoparticles in close proximity to ReO_x species, and this intimate metal–acid arrangement is responsible for the favorable acid–metal bifunctionality observed in NiRe-PS. Finally, the partial reduction of Re to lower oxidation states in NiRe-PS indicates that Re is not merely a spectator oxide but becomes redox-active during reduction, forming intimate ReO_x–Ni contacts that are expected to alter electronic properties and reaction pathways at the nanoscale, as evidenced by the EXAFS WT maps.

To gain a deeper insight into the surface metal dispersion and electronic characteristics of the reduced catalysts, time-resolved CO-DRIFTS measurements were conducted on Ni-IM, NiRe-IM, Ni-PS, and NiRe-PS (Fig. 6). For the Ni-IM catalyst (Fig. 6(a)), two distinct adsorption features existed, a broad band from 2050 to 2070 cm⁻¹ that was assigned to linearly adsorbed CO on metal Ni⁰ centers, and weaker bands from 1900 to 1950 cm⁻¹ that are assigned to bridged CO species adsorbed on multi-coordinated Ni ensembles.⁸¹ The relatively low Kubelka–Munk intensity of the linear band indicates that the Ni-IM catalyst possesses a low amount of exposed Ni⁰ surface sites consistent with the larger Ni particle sizes revealed by TEM and XRD. Rapid stabilization of the CO bands also suggests that CO adsorption primarily occurs at pre-existing metal Ni sites without substantial surface restructuring throughout measurement. Incorporation of Re into the impregnated sample (NiRe-IM, Fig. 6(b)) produced a dramatic increase in the intensity and sharpness of the linear CO band, accompanied by a considerable suppression of the contribution from bridged CO as compared with that with Ni-IM. This trend suggests a higher dispersion of Ni and a greater number of isolated Ni sites, which tend to selectively adsorb linear CO. The downshift of the frequency of ν(CO) by a few wavenumbers from the value in Ni-IM implies enhanced π-backdonation to electron-rich Ni sites, indicating that Re not only enables Ni reduction (as indicated by H₂-TPR evidence) but also electronically reorganizes the Ni centers via Ni–ReO_x interactions. This tuning increases the electron density on Ni, improving the interaction with CO and confirming the synergistic role of Re in modifying both the dispersion and electronic state of Ni. Very different CO adsorption behavior was observed for the phyllosilicate-derived Ni-PS sample (Fig. 6(c)). In this case, the spectrum was dominated by a single strong and sharp band at ∼2060 cm⁻¹ with negligible contribution from linear CO species. The absence of multi-coordinated CO adsorption strongly suggests that the phyllosilicate synthesis route is extremely effective in stabilizing Ni as highly dispersed and isolated nanoparticles in the silicate lattice. This is consistent with TEM sub-5 nm Ni particle analysis and with EXAFS data which showed persistence of Ni–Ni coordination after reduction, due to the confinement effect of the silicate matrix. The occurrence of the sharp linear CO band confirms the presence of a large number of homogeneous Ni⁰ sites in the Ni-PS sample, which are required for selective hydrogenation reactions. The most prominent enhancement was seen in the case of the NiRe-PS catalyst (Fig. 6(d)). The linear CO band not only intensified compared to that of Ni-PS but also shifted slightly downwards in wavenumbers (∼2055 cm⁻¹), pointing towards increased backdonation from Ni to CO due to electronic modification by Re. Additionally, the CO-DRIFTS spectrum revealed distinct features attributable to CO adsorption on metallic Re sites, consistent with XAS evidence of partially reduced Re species on the NiRe-PS surface. The coexistence of CO bands associated with both Ni⁰ and Re⁰ indicates that Re actively participates in CO chemisorption, reinforcing its role in modifying the electronic environment and enhancing Ni–Re interfacial synergy. The very high Kubelka–Munk intensity indicates the high number of accessible Ni and Re sites, and the more gradual stabilization of the CO bands compared to those of Ni-PS indicates dynamic reconstruction at the Ni–Re interface on exposure to CO. This rearrangement most likely involves the formation of Ni^δ+–Re^δ− pairs, which enable CO stabilization on adsorption and provide bifunctional surface sites beneficial for catalytic hydrogenation and hydrogenolysis.

Collectively, the CO-DRIFTS spectra indicate that the Ni-IM catalyst contains larger Ni ensembles, as reflected by the bridged CO adsorption, whereas the phyllosilicate route of synthesis effectively suppresses agglomeration and stabilizes highly dispersed Ni⁰ sites with exclusively linear CO adsorption. Introduction of Re further promotes Ni dispersion, eliminates bridged adsorption features, and reduces CO bands to lower wavenumbers, indicating enhanced electron donation from Ni to CO due to electronic alteration at the Ni–Re interface. The combination of the phyllosilicate framework with Re introduction, such as in NiRe-PS, produces the most favorable surface state, characterized by highly dispersed electron-rich Ni centers intimately coupled with Re species. These findings, complemented by TR-XANES and H₂-TPR/TPD investigations, conclusively support that NiRe-PS offers the optimal synergy of structural confinement, electronic modulation, and surface accessibility that lies behind its outstanding performance for biomass-derived hydrogenation reactions.

X-ray photoelectron spectroscopy (XPS) was performed to examine the oxidation states and surface compositions of Ni and Re in the reduced and passivated catalysts (Fig. S1 and Table S1). In the typical experiments, it should be noted that the catalysts analyzed by XPS were first reduced under H₂ and subsequently passivated prior to ex-situ measurement. As a result, partial surface oxidation of metallic Ni⁰ species inevitably occurred upon controlled exposure to air, leading to an apparently lower Ni⁰ surface fraction in the XPS spectra, particularly for the phyllosilicate-derived samples with highly dispersed Ni nanoparticles. In contrast, in-situ XAS measurements were conducted under reducing atmospheres and therefore reflect the actual bulk reduction state of Ni during catalyst activation. The observed difference between XPS and XAS results thus arises from the intrinsic surface sensitivity of XPS combined with post-reduction passivation effects, rather than incomplete reduction of Ni species. The Ni 2p spectra of the reduced and passivated catalysts (Fig. S1(a)) display characteristic signals for both metallic and oxidized Ni species. For Ni-IM, distinct peaks were observed at 852.3 eV (Ni⁰), 854.0 eV (Ni²⁺), and 855.5–856.5 eV (Ni²⁺/Ni^δ+), along with a satellite feature near 860.5 eV, typical of divalent nickel compounds.^38,82 Deconvolution shows that the surface is dominated by oxidized Ni species, with only a moderate contribution from metallic Ni⁰, corresponding to a Ni⁰ fraction of 0.24 (Table S1). Upon Re addition (NiRe-IM), the main Ni⁰ peak slightly shifts to 852.9 eV, while the Ni²⁺ component appears at 855.7 eV and the shake-up satellite remains near 860.4 eV. The Ni⁰ surface fraction decreases to 0.11, suggesting partial surface oxidation or coverage of Ni by ReO_x species after passivation. Although the apparent metallic Ni signal is lower, the Re promotion effect is still evident in the enhanced overall reducibility confirmed by H₂-TPR and XANES data. The small decrease in Ni surface atomic concentration (1.1 at% for NiRe-IM vs. 1.2 at% for Ni-IM) further supports the conclusion that Re modifies the surface distribution rather than reducing the bulk Ni content. The phyllosilicate-derived catalysts show markedly different behavior (Fig. S1(b)). In Ni-PS, the Ni⁰ feature appears at 852.8 eV, accompanied by NiO and Ni²⁺/Ni^δ+ peaks at 854.5–857.3 eV, with a strong satellite near 861.4 eV. The calculated Ni⁰ fraction is only 0.05, reflecting that most Ni species remain in the Ni²⁺ state. Introduction of Re in the phyllosilicate framework (NiRe-PS) causes noticeable spectral changes: the Ni⁰ peak at 852.3 eV becomes more intense, the Ni²⁺ contribution (855.9 eV) weakens, and the satellite intensity at 861.5 eV decreases. These modifications demonstrate that Re facilitates the reduction of Ni²⁺ to Ni⁰ even in the confined phyllosilicate structure. Quantitatively, the Ni⁰ fraction increases from 0.05 (Ni-PS) to 0.09 (NiRe-PS), while the surface Ni atomic content remains comparable (4.3–4.7 at%), confirming that Re enhances Ni reducibility and electronic interaction rather than simply increasing surface Ni exposure. In summary, the Ni 2p spectra reveal that impregnated catalysts contain mostly oxidized Ni with partial reduction to Ni⁰, the phyllosilicate formed strongly binds Ni to the support, lowering the surface metallic fraction, and Re addition promotes Ni reduction and modifies the electronic structure, especially in the phyllosilicate system, yielding a more balanced distribution of metallic and oxidized Ni species.

The Re 4f spectra (Fig. S1(c)) provide complementary information about the oxidation state of Re in NiRe-IM and NiRe-PS. Deconvolution revealed the existence of more than one oxidation state, with features at 41.8–42.9 eV attributable to Re⁰, those at 44.0–46.0 eV to Re⁴⁺/Re⁶⁺, and those at higher binding energy at 48.2–49.5 eV to Re⁷⁺.⁸³ Metallic Re⁰ was clearly identified in both catalysts, though its surface fraction was low compared to that of the oxidized species. NiRe-IM showed only a 0.08 metallic fraction of Re, while NiRe-PS showed a slightly higher metallic fraction of 0.17, indicating that Re is more reduced in the phyllosilicate-derived catalyst. The surface persistence of Re⁴⁺ and Re⁶⁺ species suggests that Re exists as partially reduced oxides or oxyanions, which should electronically stabilize Ni nanoparticles and provide bifunctional sites. In particular, the higher metallic Re content in NiRe-PS predicts stronger Ni–Re electronic synergy in this system, consistent with the more evident CO adsorption backdonation expressed in CO-DRIFTS and the better reducibility expressed by H₂-TPR. In general, the XPS data show that the incorporation of Re strongly alters the electronic state of Ni by enhancing its reduction and suppressing oxidized species. Such an influence is especially important in the Ni-PS system, where intense Ni–silicate interactions initially limit Ni reduction but are overcome due to the presence of Re, resulting in a high surface Ni⁰ fraction comparable to that of NiRe-IM. The observation of mixed Re oxidation states is also indicative of partially reduced ReO_x species retained on the surface, contributing to the electronic modulation of Ni and forming Ni–Re interfacial sites.

The physicochemical characterization of the Ni- and NiRe-based catalysts highlights the key importance of Re in regulating the structure, dispersion, and surface chemistry with direct effects on their catalytic activity for the hydrogenation of levulinic acid (LA) to γ-valerolactone (GVL). The textural investigation indicated that the phyllosilicate-derived NiRe/MCM-41 catalyst possessed an ordered mesoporous structure with good surface area and pore accessibility, and the surface promoted homogeneous dispersion of Ni and Re species. TEM and XRD confirmed that the addition of Re slowed Ni particle growth to smaller and more homogeneously metallic nanoparticles. H₂-TPR and in-situ TR-XANES showed that Re dramatically enhanced the reducibility of Ni and allowed metallic Ni⁰ to be stabilized under reaction conditions. CO-DRIFTS spectra also detected increased back-donation of Ni d-orbitals to CO π* orbitals in the bimetallic system, an indication of electron-rich Ni sites created by Ni–Re electronic synergy. XPS analysis further confirmed an increase in surface Ni⁰ species together with the presence of partially reduced ReO_x states, indicating strong Ni–Re electronic interactions and the formation of bifunctional interfacial sites that stabilize active Ni centers. Collectively, these electronic and structural properties are responsible for the high catalytic activity of NiRe/MCM-41, where there is high surface density of Ni⁰ sites along with the Re-mediated adjustment of electronic and acid–metal properties.

3.2 Activities of catalysts

To elucidate the intrinsic catalytic contribution of rhenium, monometallic Re catalysts supported on MCM-41 were evaluated under identical reaction conditions (120 °C, 10 bar H₂, 2 h). The Re/MCM-41-IM catalyst exhibited a levulinic acid conversion of 6.8 ± 0.9% without detectable γ-valerolactone formation. These results indicate that Re species alone possess very limited hydrogenation activity toward LA under the studied conditions. The absence of GVL formation further confirms that Re does not function as the primary hydrogenation center. Instead, the markedly superior performance of NiRe-PS originates from Ni-based hydrogenation sites, with Re acting as a promoter that enhances activity and selectivity through electronic and interfacial effects. The catalytic performance comparison between NiRe-PS-MCM-41 and NiRe-PS-SiO₂ under identical reaction conditions (1 g LA, 40 mL 2-propanol, 0.1 g catalyst, 120 °C, 120 min, 10 bar H₂) revealed a pronounced support effect. NiRe-PS-MCM-41 exhibited the highest activity with 49.5% LA conversion and 31.2% GVL yield, whereas NiRe-PS-SiO₂ achieved only 42.0% conversion and 20.1% yield. The superior performance of the MCM-41-supported catalyst is attributed to its high surface area, ordered mesoporosity, and abundant silanol groups, which facilitate uniform Ni–Re dispersion and stronger Ni–O–Si interactions, thereby enhancing hydrogenation efficiency and selectivity toward GVL. In contrast, the SiO₂-supported counterpart, with a less ordered and lower surface area structure, exhibits limited metal dispersion and weaker metal–support interactions, resulting in lower catalytic performance.

The catalytic performances of the synthesized catalysts for the hydrogenation of levulinic acid (LA) to γ-valerolactone (GVL) exhibited significant differences with respect to the preparation method and the presence of rhenium. As reflected in Fig. 7(a), the NiRe-PS catalyst achieved the maximum LA conversion (49.5%) and GVL yield (31.2%), which was about 4.1-fold and 6.2-fold higher, respectively, compared to that over Ni-PS (12% conversion and 5% yield). On the other hand, the impregnated catalysts (Ni-IM and NiRe-IM) were observed to be slightly active with less than 10% conversion and less than 3% GVL yields, along with the good characteristic activity of Ni/MCM-41 and NiRe/MCM-41 derived from the phyllosilicate-based material. This activity enhancement was also reflected in the reaction rates and turn-over frequency (TOF) (Fig. 7(b)), where NiRe-PS achieved a TOF of 8.9 h⁻¹ and a r_GVL of 13.4 mmol g_cat⁻¹ h⁻¹, whereas Ni-PS achieved only 1.7 h⁻¹ and 2.2 mmol g_cat⁻¹ h⁻¹. Kinetic evaluation through pseudo-first-order plots (Fig. 7(c)) confirmed that the rate constant (k) with the apparent value for NiRe-PS (0.0059 min⁻¹) was almost 5 times higher than that of Ni-PS (0.0012 min⁻¹). The experimental results provided in Table S2 substantiate these results with R² > 0.98 and standard deviation ≤0.02, ensuring the validity of fitting. Furthermore, Arrhenius plot analysis (Fig. 7(d)) shown that the NiRe-PS catalyst possessed a lower apparent activation energy of 64.0 kJ mol⁻¹ compared to Ni-PS (67.4 kJ mol⁻¹). As summarized in Table S3, ln(r_GVL) values with temperatures had a more sloping curve for Ni-PS in line with its higher energy barrier, while NiRe-PS maintained more stable activity over the temperature range under investigation. This shows that Re introduction not only enhanced hydrogenation activity but also provided efficient energetics towards LA to GVL conversion. These kinetic and catalytic data correlate well with the Ni–Re synergistic effect, in which electron transfer from ReO_x to Ni enhances H₂ activation and stabilizes key carbonyl intermediates, thereby promoting GVL formation at higher rates. Taken together, the combined evidence from catalytic activity, kinetic behavior, and catalyst structure analyses confirms that NiRe-PS exhibits a unique bifunctional nature: Ni sites provide hydrogenation capability, while partially reduced ReO_x species stabilize reaction intermediates, leading to enhanced GVL selectivity and lower activation energy barriers.

Fig. 7 Catalytic performance of Ni-IM, NiRe-IM, Ni-PS, and NiRe-PS catalysts for levulinic acid (LA) hydrogenation: (a) LA conversions and GVL yields; (b) turnover frequencies (TOFs) and reaction rate; (c) kinetic investigation; and (d) Arrhenius plots for LA hydrogenation over Ni-PS and NiRe-PS catalysts (reaction conditions: 1 g LA, 31.4 g 2-propanol, 2 h, 120 °C, and 10 bar H₂).

The effect of reaction temperature on the hydrogenation of LA to GVL over the NiRe-PS catalyst was examined (Fig. 8(a)). At 100 °C, the catalyst showed low activity with approximately 24% LA conversion and 5% GVL yield, indicating limited hydrogenation capability at lower temperatures. A sharp increase in both conversion and yield was observed between 120 °C and 160 °C, where the LA conversion rose from 50% at 120 °C to nearly 92% at 160 °C, accompanied by a corresponding increase in GVL yield from 32% to around 78%. Further raising the temperature to 180 °C resulted in complete conversion but only a marginal improvement in GVL yield (∼92%), suggesting that high temperature may promote secondary reactions such as condensation or over-hydrogenation. Notably, the pronounced enhancement in both conversion and GVL yield within the 140–160 °C range highlights the optimal operating range of the NiRe-PS catalyst and underscores its superior hydrogenation efficiency under elevated-temperature conditions. Similarly, the time-course profile at 140 °C (Fig. 8(b)) revealed a steady rise in conversion and yield with increasing reaction time. After 60 min, LA conversion reached 70% with a GVL yield of about 50%. Extending the reaction to 120 min and 240 min enhanced conversion to 92% and 100%, respectively, while GVL yield increased to 81% and 96%. The increase of reaction time to 360 min led to complete conversion but not to any significant improvement in GVL yield, establishing that 140 °C for 4 h is the optimum reaction condition for this system. To thoroughly evaluate the catalytic activity of NiRe-PS, that of its LA hydrogenation to GVL was compared with those of a series of literature Ni-based catalysts (Fig. 8(c) and Table S4). NiRe-PS exhibited 100% LA conversion and 91.6% yield of GVL under the optimal conditions (160 °C, 2 h, 10 bar H₂, and 2-propanol), with TOF = 26.3 h⁻¹ and reaction rate = 39.4 mmol GVL per g_cat per h. This activity is much higher than those of monometallic Ni catalysts such as 30% Ni/O-clay450N (TOF = 1.3 h⁻¹, 100% yield),⁸⁴ S3M1.5@NiPS-600 (TOF = 1.9 h⁻¹, 91% yield),⁸⁵ 1-Ni/C-400 (TOF = 7.2 h⁻¹, 95% yield),⁸⁶ and Ni@FLG-600 (TOF = 0.2 h⁻¹, 99% yield).⁸⁷ The same was noted with catalysts such as Ni–Al (TOF = 4.5 h⁻¹, 97% yield),⁸⁸ which also had relatively lower intrinsic activity. Significantly, when compared to a related bimetallic system, 500-NiFe NPs@C (3 : 1) (TOF = 2.2 h⁻¹, 90% yield),⁸⁹ Ni₃Zn₁@OMC (TOF = 2.7 h⁻¹, 61% yield),⁹⁰ Ni/Ru@WOMC (TOF = 5.0 h⁻¹, 100% yield),⁹¹ and NiRe_0.075/Al₂O₃ (TOF = 10.8 h⁻¹, rate = 20.1 mmol GVL per g_cat per h),⁵³ the NiRe-PS catalyst possessed more than double the turnover frequency. The increased TOF and reaction rate reflect the enhanced hydrogenation activity and the optimal utilization of active sites, resulting from the increased electronic interaction between the two metals derived from the phyllosilicate structure. Overall, the results place NiRe-PS among the most active non-noble metal catalysts for LA hydrogenation to GVL, with performance equating or even exceeding that of some noble-metal-based systems. The superior performance of NiRe-PS compared with those of previously reported catalysts can be attributed to the unique phyllosilicate-MCM-41 architecture, which enables higher dispersion and structural stabilization of Ni–Re species. The incorporation of Ni within the phyllosilicate framework suppresses metal aggregation by enhancing Ni²⁺ and SiO₂ interactions, while the mesoporous MCM-41 support provides high surface area and improved accessibility of active sites. This synergistic structural environment promotes more efficient H₂ activation and optimized metal–acid cooperation, resulting in higher catalytic activity and selectivity than those of conventional supported systems. The reusability of NiRe-PS was examined by conducting three successive catalytic runs at 140 °C at different reaction times (Fig. 8(d)). Even though the structure of the catalyst remained intact with its activity, a gradual decline in the LA conversion at 60 min was observed from 64% in the initial run to 58% and 44% in the second and third runs, respectively. However, even after the third run, the catalyst retained more than half of its initial activity, implying that simple regeneration methods, such as calcination–reduction treatments, could restore its performance. Catalyst stability was further evaluated through extended recycling experiments performed over five consecutive runs at 160 °C under 10 bar H₂ for 2 h using 0.1 g of catalyst. As shown in Fig. S2, the NiRe-PS catalyst exhibits high and reproducible performance over repeated cycles, with levulinic acid conversion decreasing gradually from 100% in the first run to 84% after the fifth run, while the γ-valerolactone yield decreases from 90% to 68%. Despite this moderate decline, the catalyst retains a substantial fraction of its initial activity and selectivity, demonstrating the high structural and compositional stability of the phyllosilicate-derived Ni–Re catalyst under prolonged reaction conditions. Overall, the NiRe-PS catalyst has superior catalytic activity, with near-quantitative conversion and high selectivity toward GVL under optimal conditions, and TOF and productivity values much higher than those for other Ni-based catalysts reported. These results further establish the feasibility of using NiRe-PS as an effective practical and scalable non-noble metal catalyst for green production of GVL from levulinic acid.

Fig. 8 Catalytic performance of the NiRe-PS catalyst for levulinic acid (LA) hydrogenation: (a) effects of reaction temperature (reaction conditions: LA and 2-PrOH were fixed at 1 g and 31.4 g, 120 min, 10 bar H₂ pressure, and 0.1 g catalyst); (b) effects of reaction time (reaction conditions: LA and 2-PrOH were fixed at 1 g and 31.4 g, 140 °C, 10 bar H₂ pressure, and 0.1 g catalyst); (c) comparison of catalytic activities (TOFs) of NiRe-PS and previous Ni-based heterogeneous catalysts; and (d) reusability of the NiRe-PS catalyst over three consecutive reactions for hydrogenating LA to GVL at 140 °C and an initial H₂ pressure of 10 bar for 1 h, using 0.1 g of the catalyst.

3.3 Isotopic studies for mechanistic insights

For mechanistic insight into the hydrogenation of LA to GVL, isotopic labeling experiments with 2-propanol (2-PrOH) and its deuterated analogue (2-PrOD) as hydrogen donors under different gas atmospheres (N₂, H₂, and D₂) at 140 °C (Fig. 9(a–f)) were performed. The catalytic activity trends distinctly establish the key role played by molecular hydrogen in facilitating LA conversion (as shown in Fig. 9(a)). 72% LA conversion and 49% GVL yield were achieved with the NiRe-PS catalyst under 2-PrOH/H₂ conditions, while under 2-PrOH/D₂ conditions the activity decreased to 40% conversion and 33% yield. The same occurred when using 2-PrOD was the solvent: the catalyst gave 55% conversion and 45% yield in H₂, while 35% conversion and 32% yield in D₂. Notably, the reactions conducted under N₂ atmospheres showed much lower activity, which confirms that neither 2-PrOH nor 2-PrOD can provide a sufficient hydrogen source for efficient transfer hydrogenation in the absence of pressurized H₂ or D₂. Isotopic and control experiments confirm that levulinic acid hydrogenation over NiRe-PS proceeds via parallel pathways involving direct molecular H₂ activation and solvent-derived hydrogen transfer. In isopropanol, levulinic acid can undergo esterification to form isopropyl levulinate, which was detected as a transient intermediate, particularly under a N₂ atmosphere and at short reaction times. This ester intermediate is rapidly hydrogenated to γ-valerolactone under H₂, resulting in its low steady-state concentration.⁹² Turnover frequency (TOF) and reaction rate data (Fig. 9(b)) corroborated these findings. The TOF became 2.3 h⁻¹ in 2-PrOH/H₂, but 1.5 h⁻¹ in 2-PrOH/D₂, indicating a drastic kinetic isotope effect (KIE) for C–H vs. C–D bond cleavage. The rate of GVL formation dropped from 3.5 mmol g_cat⁻¹ h⁻¹ in H₂ to 2.2 mmol g_cat⁻¹ h⁻¹ in D₂, further establishing that hydrogen transfer is the rate-determining step. When 2-PrOD was used, TOF values decreased even further (2.1 h⁻¹ in H₂ and 1.5 h⁻¹ in D₂), which means that solvent deuteration adds an additional kinetic barrier due to the lower reactivity of the O–D bond in the alcohol-mediated transfer hydrogenation. The GC-MS fragmentation patterns of GVL under different isotopic conditions (Fig. 9(c–f)) exhibited significant mass-to-charge (m/z) signal shifts, offering molecular-level confirmation of H/D incorporation. For the standard 2-PrOH/H₂ system, typical fragments were observed at m/z = 56, 85, and 100, consistent with those of non-deuterated GVL, but in 2-PrOD/H₂, 2-PrOH/D₂, and 2-PrOD/D₂, isotopic peaks at m/z = 56–57, 85–86, and 101–103 appeared, typical of mono- and di-deuterated GVL species. This result evidently demonstrates that the solvent and molecular hydrogen both provide the hydrogen atoms in the reaction, with a partial substitution with deuterium when D₂ or 2-PrOD is used. The simultaneous appearance of multiple isotopologues suggests simultaneous hydrogenation pathways involving both direct H₂/D₂ activation on Ni–Re sites and transfer hydrogenation from the alcohol solvent.

Fig. 9 Catalytic and mechanistic studies of the NiRe-PS catalyst for levulinic acid (LA) hydrogenation to γ-valerolactone (GVL): (a) catalytic activity, (b) turnover frequency (TOF), reaction rate, and ((c)–(f)) mass spectra of the GVL fragmentation patterns obtained by GC-MS analysis of the NiRe-PS catalyst for the LA hydrogenation to GVL using 2-propanol (2-PrOH) and 2-propanol-D (2-PrOD) as solvents under N₂, H₂, and D₂ atmospheres (reaction conditions: 140 °C, 60 min, 10 bar N₂/H₂/D₂, 0.02 g catalyst, 0.2 g LA, and 7 mL 2-PrOH or 2-PrOD).

The ¹H NMR spectra provide informative mechanistic data for levulinic acid (LA) hydrogenation to γ-valerolactone (GVL) using the NiRe-PS catalyst under diverse isotopic conditions (Fig. S3). Under the normal H₂/2-propanol system, characteristic GVL resonance peaks were clearly observed at δ 1.35–1.40 ppm for methyl protons of the lactone ring, corresponding well to the reference GVL spectrum. Remarkably, with the use of deuterated solvents (2-PrOD) or isotopic gases (D₂), significant changes in peak intensity and splitting were found, reflecting the H/D exchange process during the catalysis. For instance, in the case of 2-PrOH/D₂ and 2-PrOD/H₂ systems, partial line broadening and suppression of the methyl proton resonance suggested isotopic substitution during hydrogenation, whereas in the completely deuterated 2-PrOD/D₂ system, the resonances were suppressed to a greater extent and shifted, establishing extensive deuterium incorporation into the GVL framework. These changes in the spectra, together with the kinetic isotope effect on catalytic activity, indicate that both molecular hydrogen and protic solvent are active hydrogen donors, and that the route involving transfer via the solvent is of particular significance. This mechanistic profile conforms to the proposed route wherein LA is hydrogenated via the intermediate 4-hydroxypentanoic acid (HPA), followed by intramolecular cyclization to GVL. Thus, NMR data validate the bimolecular H₂/solvent role in hydrogen transfer and underscore the solvent isotopic composition effect on product isotopomer distribution, providing direct spectroscopic evidence for the molecular-level catalytic cycle of LA to GVL over NiRe-PS.

Based on the isotopic labeling and spectroscopic results, a plausible molecular-level catalytic cycle is proposed (Scheme 2). Structural analyses confirm that Ni and Re exist as spatially separated but electronically interactive phases in metallic Ni⁰ nanoparticles and dispersed ReO_x species rather than as a true Ni–Re alloy. In this configuration, Ni⁰ sites primarily dissociate molecular H₂, generating surface hydride species that migrate toward nearby ReO_x centers through hydrogen spillover. Simultaneously, ReO_x species act as Lewis acidic centers, facilitating adsorption and activation of the carbonyl group in levulinic acid (LA). The transferred hydrides from Ni sites hydrogenate the activated carbonyl intermediate on ReO_x, forming 4-hydroxypentanoic acid (HPA), which then cyclizes intramolecularly to produce γ-valerolactone (GVL).⁹³ Isotopic labeling results confirm that hydrogen atoms from both molecular H₂ and 2-propanol participate in the hydrogenation process, indicating a dual hydrogenation pathway involving direct H₂ dissociation and solvent-assisted hydrogen transfer. The observed kinetic isotope effect (KIE) suggests that the cleavage of the C–H (or C–D) bond from 2-propanol is at least partially rate-determining. Therefore, the synergistic cooperation between metallic Ni⁰ (for H₂ activation) and ReO_x (for carbonyl activation and intermediate stabilization) rather than alloy formation accounts for the high activity and selectivity of NiRe/MCM-41 derived from the phyllosilicate structure catalyst.

Scheme 2 Proposed molecular-level catalytic cycle for levulinic acid (LA) hydrogenation to γ-valerolactone (GVL) over the NiRe-PS catalyst, illustrating hydrogen activation and transfer processes on the Ni–Re interfacial active sites.

3.4 Characterization of the spent catalyst

To identify the structural and chemical stability of the NiRe-PS catalyst after application in the hydrogenation of levulinic acid to γ-valerolactone, a thorough post-reaction characterization was performed using TEM, XPS, XRD, and TGA measurements (as shown in Fig. S4(a), (b), (c) and (d), respectively) under reaction conditions of 160 °C, 10 bar H₂, 2 h, and 0.1 g catalyst. TEM images of the spent catalyst indicated that the overall support structure was retained, but there was extreme particle agglomeration relative to the reduced catalyst. Elemental mapping revealed that Ni (green) and Re (red) remained well-dispersed but local agglomeration was more evident, reflecting the occurrence of partial sintering of the catalyst under hydrogenation conditions. XPS analysis provided complementary data on the surface chemical states of Ni and Re before and after the reaction. For the fresh reduced catalyst, Ni predominantly existed in the Ni⁰ state (852.6 eV) with minimal contributions from the Ni²⁺ and Ni^δ+ species. After the first catalytic run, the Ni⁰ component was significantly decreased, and the relative intensity of the oxidized Ni²⁺ and NiO species increased, indicating that the partial oxidation of Ni was preferred under the reaction conditions. An identical trend was followed by Re, wherein the reduced catalyst had mixed valence states of Re⁷⁺, Re⁶⁺, Re⁴⁺, and Re⁰, wherein Re⁰ and Re⁴⁺ were catalytically active. The spent surface showed reduced Re⁰ and Re⁴⁺ fractions with enrichment of the higher oxidation states (Re⁶⁺ and Re⁷⁺), indicating that the Re species were more susceptible to oxidation under the hydrogenation process. The results reveal that both Re and Ni underwent partial surface oxidation and redistribution, which would have affected the bifunctional synergy of the catalyst. The crystallographic characteristics of the catalyst were also investigated by XRD. The used catalyst contained sharp peaks at 2θ ≈ 44.5°, 51.8°, and 76.3° corresponding to the (111), (200), and (220) planes of metallic Ni and a broad halo at ∼22° corresponding to amorphous SiO₂. After the catalytic reaction, the intensity of Ni reflections increases substantially and peaks become sharp, indicative of change in metallic crystallinity and probable formation of Ni metallic phases. Such structural evolution suggests that Ni nanoparticles change some of their metallic nature upon hydrogenation, which could contribute to decreased catalytic activity during recycling. Thermal stability and carbon deposition were investigated by TGA. The used catalyst exhibited a gradual weight loss of 5% up to 600 °C that mainly occurred due to desorption of physisorbed water and surface hydroxyl groups. In contrast, the spent catalyst experienced a significant weight loss of 10% in the same temperature range with distinct decreases within 300–500 °C. These are characteristic of coke deposition or intermediate deposition during carbonaceous deposit formation, confirming that coke deposition actually occurred under levulinic acid hydrogenation. To quantitatively assess metal stability, ICP-OES analysis was conducted on the fresh catalyst and on spent catalysts recovered after the 1^st and 5^th reaction cycles (Table S5). The Ni content decreased slightly from 8.1 wt% in the fresh catalyst to 6.9 wt% after the first run and 6.8 wt% after the fifth run, while the Re content showed only a minor decrease from 2.3 wt% to 1.9 wt% after five cycles. These results indicate that metal leaching is minimal and that both Ni and Re species are effectively stabilized within the MCM-41 framework. The modest activity loss observed during recycling is therefore attributed primarily to surface deactivation effects rather than the loss of active metal species. Characterization of the spent NiRe-PS catalyst indicates the presence of slight carbonaceous deposits, metal leaching, and partial surface oxidation of metallic species, which may reduce the number of accessible active sites. Importantly, no significant Ni particle sintering or structural degradation is observed, confirming the high structural stability of NiRe/MCM-41 derived from the phyllosilicate framework and its ability to preserve active-site dispersion over multiple catalytic cycles. Overall, spent catalyst characterization clearly illustrates that the NiRe-PS catalyst retains its structural framework but undergoes surface oxidation, metal particle aggregation, and deposition of coke or intermediate under levulinic acid hydrogenation conditions. These activities result in the gradual loss of catalytic activity when reused and point towards the necessity of optimizing reduction treatments and regeneration processes in order to impart catalyst stability for long-term use.

4 Conclusion

In this study, Ni and NiRe catalysts supported on MCM-41 were prepared systematically using ammonia evaporation (AE) and impregnation (IM) methods to investigate the influence of phyllosilicate formation and Re modification on levulinic acid (LA) hydrogenation to γ-valerolactone (GVL). The AE route successfully developed a nickel phyllosilicate (Ni-PS) framework with high Ni–O–Si anchoring bonds and with the addition of Re produced a highly dispersed Ni–ReO_x interface with improved reducibility and acid sites. The particle size of the NiRe-PS catalyst was the smallest (2.9 nm), and the surface Ni⁰ content (4.3%), H₂ uptake and acid density were higher compared to those of all catalysts. Catalytic tests showed that NiRe-PS achieved 100% conversion of LA and 96% yield of GVL at 140 °C and 10 bar H₂ with a 26.3 h⁻¹ turnover frequency (TOF) (160 °C, 10 bar H₂), which was superior to those of Ni-PS and NiRe-IM catalysts. Kinetic assessment confirmed pseudo-first-order kinetics with an apparent rate constant of 0.0059 min⁻¹ and lower activation energy (64 kJ mol⁻¹) for NiRe-PS, indicating faster hydrogenation kinetics. Isotopic experiments (2-PrOH/2-PrOD, H₂/D₂) showed a kinetic isotope effect, proving that molecular hydrogen and solvent-borne hydrogen are both active species in the reaction. Post-reaction characterization confirmed excellent structural integrity and reusability. In conclusion, this study demonstrates that the combination of the nickel phyllosilicate structure and rhenium modification produces an extremely active and durable bifunctional catalyst. The findings establish NiRe/MCM-41 derived from the phyllosilicate structure as a promising non-noble system for sustainable biomass valorization and green fuel synthesis.

Author contributions

Yupawan Maneewong and Pratikkumar Lakhani: investigation, validation, and writing – original draft; Sakhon Ratchahat, Chularat Sakdaronnarong, Wanwisa Limphirat, and Bunyarat Rungtaweevoranitc: resources and validation; Kanyanat Khosukwiwat: investigation; Suttichai Assabumrungrat and Kittisak Choojun: writing – review & editing; Tawan Sooknoi and Keiichi Tomishige: supervision and writing – review & editing; Atthapon Srifa: supervision, conceptualization, methodology, writing – review & editing, visualization, and funding acquisition.

Conflicts of interest

There are no conflicts to declare.

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Supplementary information (SI): general information on materials, experimental procedures, and characterization techniques; additional tables and figures supporting the main text, including XPS results of reduced catalysts (Table S1), kinetic and activation energy data (Tables S2 and S3), catalytic performance comparison of Ni-based catalysts (Table S4) and ICP-OES analysis of fresh and spent catalysts (Table S5). SI figures present the XPS spectra of Ni and Re regions (Fig. S1), reusability of the NiRe-PS catalyst shown as a bar chart (Fig. S2), ¹H NMR spectra of reaction products (Fig. S3), and characterization of the spent NiRe-PS catalyst, including TEM mapping, XPS, XRD, and TGA analyses (Fig. S4). See DOI: https://doi.org/10.1039/d5gc06171g.

Acknowledgements

This research project is supported by Mahidol University through the Strategic Research Fund (2024). In addition, this project is partially funded by the National Research Council of Thailand (NRCT) and Mahidol University under a Mid-Career Research Grant (grant number: N42A680336). It is also partially funded by the National Research Council of Thailand (NRCT): contract number N42A680527. Furthermore, this work is financially supported by the International Postdoctoral Fellowship for the fiscal year B.E. 2568 (A.D. 2025), Mahidol University International Relations Division (MUIR), Mahidol University.

References

1. P. Lakhani and A. Srifa, Fuel Process. Technol., 2025, 278, 108326.
2. B. Zhang, B. K. Biswal, J. Zhang and R. Balasubramanian, Chem. Rev., 2023, 123, 7193–7294.
3. F. Ali, A. Dawood, A. Hussain, M. H. Alnasir, M. A. Khan, T. M. Butt, N. K. Janjua and A. Hamid, Discover Sustainability, 2024, 5, 156.
4. C. Árvai, Z. Medgyesi, M. Y. Lui and L. T. Mika, Adv. Synth. Catal., 2024, 366, 4846–4888.
5. K. Yan, C. Jarvis, J. Gu and Y. Yan, Renewable Sustainable Energy Rev., 2015, 51, 986–997.
6. L. Jeanmard, W. Rongwong and Y. Chisti, Biomass Bioenergy, 2025, 195, 107683.
7. K. Tomishige, M. Honda, H. Sugimoto, L. Liu, M. Yabushita and Y. Nakagawa, Carbon Neutrality, 2024, 3, 17.
8. X. Su, T. Xiao, Q. Gong, H. Cheng and F. Zhao, Green Chem., 2025, 27, 9846–9861.
9. A. Hijazi, N. Khalaf, W. Kwapinski and J. J. Leahy, RSC Adv., 2022, 12, 13673–13694.
10. M. Khalid, M. G. Mesa, D. Scapens and A. Osatiashtiani, ACS Sustainable Chem. Eng., 2024, 12, 16494–16517.
11. S. G. Wettstein, D. M. Alonso, Y. Chong and J. A. Dumesic, Energy Environ. Sci., 2012, 5, 8199–8203.
12. R. Sakdee, S. Ratchahat, C. Sakdaronnarong, W. Koo-amornpattana, W. Limphirat, S. Mahakot, S. Assabumrungrat and A. Srifa, Fuel Process. Technol., 2025, 276, 108276.
13. A. Osatiashtiani, A. F. Lee and K. Wilson, J. Chem. Technol. Biotechnol., 2017, 92, 1125–1135.
14. C. A. S. Lanziano, C. B. Rodella and R. Guirardello, Molecules, 2025, 30, 1400.
15. O. A. Abdelrahman, A. Heyden and J. Q. Bond, ACS Catal., 2014, 4, 1171–1181.
16. P. Lakhani, K. Kliengklao, C. Pratchayakul, R. Sakdee, S. Ratchahat, C. Sakdaronnarong, W. Koo-amornpattana, W. Limphirat, S. Assabumrungrat and A. Srifa, Biomass Bioenergy, 2026, 205, 108563.
17. K. Malins and I. Malina, Biomass Bioenergy, 2023, 171, 106732.
18. W. Xu, M. Liu, K. Xu and B. Li, Coord. Chem. Rev., 2025, 529, 216445.
19. Y. Takeda, M. Tamura, Y. Nakagawa, K. Okumura and K. Tomishige, Catal. Sci. Technol., 2016, 6, 5668–5683.
20. Y. Nie, M. Uddin, Q. Wang, X. Li, N. Liu, S. Yu, P. Keomeesay, X. Zhao, Y. Chen and Z. Zheng, Green Chem., 2025, 27, 12070–12106.
21. N. Yamanaka and S. Shimazu, Eng, 2022, 3, 60–77.
22. S. De, J. Zhang, R. Luque and N. Yan, Energy Environ. Sci., 2016, 9, 3314–3347.
23. P. T. Pham, C. Q. Pham, T.-T. Dam, Q.-A. Nguyen and T. M. Nguyen, Fuel Process. Technol., 2025, 267, 108170.
24. G. Pipitone, G. Zoppi, R. Pirone and S. Bensaid, Int. J. Hydrogen Energy, 2022, 47, 151–180.
25. D. Bhanderi, P. Lakhani, H. Srivastava, A. Sharma, S. S. Soni and C. K. Modi, ACS Sustainable Resour. Manage., 2025, 2, 2157–2167.
26. F. Qi, J. Peng, Z. Liang, J. Guo, J. Liu, T. Fang and H. Mao, Environ. Sci. Ecotechnology, 2024, 22, 100443.
27. O. U. Osazuwa and K. H. Ng, Alexandria Eng. J., 2025, 127, 309–335.
28. P. Lakhani, D. Bhanderi and C. K. Modi, Discover Catal., 2024, 1, 2.
29. R. Kumar, K. Kumar, N. Choudary and K. Pant, Fuel Process. Technol., 2019, 186, 40–52.
30. J. Zhang, Y. Li, H. Song, L. Zhang, Y. Wu, Y. He, L. Ma, J. Hong, A. Tayal and N. Marinkovic, Nat. Commun., 2024, 15, 8566.
31. P. Lakhani and C. K. Modi, Mol. Catal., 2022, 525, 112359.
32. P. Lakhani and C. K. Modi, New J. Chem., 2023, 47, 8767–8775.
33. P. Lakhani, D. Bhanderi and C. K. Modi, J. Mol. Liq., 2024, 408, 125306.
34. P. Lakhani and C. K. Modi, Mol. Catal., 2024, 559, 114080.
35. P. Lakhani, S. Kane, H. Srivastava, U. Goutam and C. K. Modi, RSC Sustainability, 2023, 1, 1773–1782.
36. P. Lakhani, D. Chodvadiya, P. K. Jha, V. K. Gupta, D. Trzybiński, K. Wozniak, K. Kurzydłowski, U. Goutam, H. Srivastava and C. K. Modi, Phys. Chem. Chem. Phys., 2023, 25, 14374–14386.
37. Y. Zhang and Q. Liu, J. CO2 Util., 2021, 50, 101587.
38. S. Kuhaudomlap, O. Mekasuwandumrong, P. Praserthdam, K. M. Lee, C. W. Jones and J. Panpranot, ACS Omega, 2022, 8, 249–261.
39. Z. Yang, B. Niu, X. Duan, Z. Wei, Z. Shui, G. Jiang, F. Zhang and Z. Hao, Chem. Eng. Sci., 2025, 122457.
40. L. Liao, L. Chen, R. P. Ye, X. Tang and J. Liu, Chem. – Asian J., 2021, 16, 678–689.
41. P. Lakhani, D. Bhanderi and C. K. Modi, J. Nanopart. Res., 2024, 26, 148.
42. D. Leybo, U. J. Etim, M. Monai, S. R. Bare, Z. Zhong and C. Vogt, Chem. Soc. Rev., 2024, 53, 10450–10490.
43. R. Parmar, P. Lakhani, D. Bhanderi, S. Kane, U. Goutam and C. K. Modi, J. Organomet. Chem., 2024, 1008, 123073.
44. T. Shoosri, P. Chotiwilaiwan, T. Rattanapornchaiwat, T. Teerawatananond, T. Miyake, J. Panpranot and P. Weerachawanasak, RSC Adv., 2024, 14, 38232–38244.
45. R. Nakamura, H. Minamisawa and T. Okada, Catal. Sci. Technol., 2025, 15, 5857–5863.
46. L. Liu, B. Liu, Y. Nakagawa, S. Liu, L. Wang, M. Yabushita and K. Tomishige, Chin. J. Catal., 2024, 62, 1–31.
47. T. Zanette, A. García-Zaragoza, J. Mazarío, J. S. Martinez, B. Chaudret, C. Cerezo-Navarrete and P. Oña-Burgos, Green Chem., 2025, 27, 11438–11454.
48. P. Yuan, C. K. T. Wun and T. W. B. Lo, Adv. Mater., 2025, 2501960.
49. L. Xie, J. Liang, L. Jiang and W. Huang, Chem. Sci., 2025, 16, 3408–3429.
50. L. Ma, L. Yan, A.-H. Lu and Y. Ding, RSC Adv., 2018, 8, 8152–8163.
51. K. Tomishige, Y. Nakagawa and M. Tamura, Chin. Chem. Lett., 2020, 31, 1071–1077.
52. J. Luo, W. Yang, T. Wang, H. Liu, X. Mao and C. Liang, Green Chem., 2025, 27, 14873–14882.
53. Y. Maneewong, P. Nimmanterdwong, S. Ratchahat, C. Sakdaronnarong, W. Limphirat, P. Khemthong, B. Rungtaweevoranit, K. Faungnawakij, S. Assabumrungrat and Y.-C. Lin, Chem. Eng. J., 2025, 508, 160969.
54. K. Tomishige, M. Tamura and Y. Nakagawa, Chem. Rec., 2014, 14, 1041–1054.
55. J. S. Reinhold, J. Pang, B. Zhang, F. E. Kühn and T. Zhang, Green Chem., 2024, 26, 10661–10686.
56. P. Lakhani, R. Sakdee, S. Ratchahat, C. Sakdaronnarong, W. Koo-amornpattana, W. Limphirat, S. Assabumrungrat and A. Srifa, Fuel Process. Technol., 2026, 282, 108397.
57. Y. Wang, P. Prinsen, K. S. Triantafyllidis, S. A. Karakoulia, P. N. Trikalitis, A. Yepez, C. Len and R. Luque, ACS Sustainable Chem. Eng., 2018, 6, 9831–9844.
58. K. Zhou, J. Chen, Y. Cheng, Z. Chen, S. Kang, Z. Cai, Y. Xu and J. Wei, ACS Sustainable Chem. Eng., 2020, 8, 16624–16636.
59. N. Dzakaria, A. H. Lahuri, S. Samidin, S. S. Sulhadi, T. S. T. Saharuddin, F. Salleh, A. Samsuri, M. R. Yusop and A. Yarmo, AIP Conf. Proc., 2024, 2925, 020013.
60. J. Li, P. Li, J. Li, Z. Tian and F. Yu, Catalysts, 2019, 9, 506.
61. W. Uttamaprakrom, P. Reubroycharoen, P. Charoensiritanasin, J. Tatiyapantarak, A. Srifa, W. Koo-Amornpattana, W. Chaiwat, C. Sakdaronnarong, M. Sudoh and R. Watanabe, J. Environ. Chem. Eng., 2021, 9, 106150.
62. Y. Shi, C. Liu, J. Zhuo and Q. Yao, ACS Omega, 2020, 5, 20299–20310.
63. Z. Liu, W. Bi, W. Zhao and Q. Liu, Int. J. Hydrogen Energy, 2024, 78, 1183–1191.
64. Q. Fan, H. Li, H. Chen, L. Cheng and Q. Liu, Fuel, 2023, 333, 126440.
65. J. C. Park, H. J. Lee, J. U. Bang, K. H. Park and H. Song, Chem. Commun., 2009, 7345–7347.
66. A. Kittisabhorn, I. Ahmed, W. Pornputtapitak, S. Ratchahat, W. Chaiwat, W. Koo-Amornpattana, W. Klysubun, W. Limphirat, S. Assabumrungrat and A. Srifa, ACS Omega, 2024, 9, 28637–28647.
67. P. Huang, Z. Lv, J. Zhang, T. Niu, Y. Leng, M. Fan and P. Zhang, Fuel, 2025, 379, 133031.
68. W. Lin, Y. Chen, Y. Zhang, Y. Zhang, J. Wang, L. Wang, C. C. Xu and R. Nie, ACS Catal., 2023, 13, 11256–11267.
69. S. Koso, H. Watanabe, K. Okumura, Y. Nakagawa and K. Tomishige, J. Phys. Chem. C, 2012, 116, 3079–3090.
70. Y. Amada, H. Watanabe, M. Tamura, Y. Nakagawa, K. Okumura and K. Tomishige, J. Phys. Chem. C, 2012, 116, 23503–23514.
71. N. Meeboonanake, A. Pengsawang, W. Kraithong, A. Jiratanachotikul, P. Khemthong, T. Butburee, S. Kiatphuengporn, S. Thongratkaew, A. Junkaew and W. Donphai, Int. J. Hydrogen Energy, 2025, 138, 368–378.
72. H. Yang, J. Li, Q. Chang, Q.-c. Zhao, M.-n. Zhang, H.-j. Wang, C.-h. Zhang, F. Wang and X.-d. Wen, Fuel, 2025, 394, 135119.
73. N. Mora, E. Blanco, J. Seguel, J. D. de León and N. Escalona, Appl. Catal., A, 2025, 691, 120033.
74. A. Mansour and C. Melendres, J. Phys. Chem. A, 1998, 102, 65–81.
75. R. Dähn, M. Jullien, A. M. Scheidegger, C. Poinssot, B. Baeyens and M. H. Bradbury, Clays Clay Miner., 2006, 54, 209–219.
76. A. Tougerti, S. Cristol, E. Berrier, V. Briois, C. La Fontaine, F. Villain and Y. Joly, Phys. Rev. B: Condens. Matter Mater. Phys., 2012, 85, 125136.
77. R. Dähn, A. M. Scheidegger, A. Manceau, M. L. Schlegel, B. Baeyens, M. H. Bradbury and M. Morales, Geochim. Cosmochim. Acta, 2002, 66, 2335–2347.
78. J. Timoshenko, A. Kuzmin and J. Purans, J. Phys.: Condens. Matter, 2014, 26, 055401.
79. K. Yamaguchi, Y. Nakagawa, C. Li, M. Yabushita and K. Tomishige, ACS Catal., 2022, 12, 12582–12595.
80. Z. Luo, X. Ge, D. Fang, X. Xu, D. Zhang, Y. Cao, X. Duan, W. Li, J. Zhou and X. Zhou, J. Catal., 2024, 434, 115528.
81. Q. Xue, S. Qian, K. H. Lim, M. X. J. Wee, K. Nie, S. Cheng, X. Chen, P. Chirawatkul, B. Johannessen, B. Yan, Z. Yin, Y. Wang, G. Luo, B. C. Gates and S. Kawi, J. Am. Chem. Soc., 2025, 147, 38141–38157.
82. Y. Park, D. Chakraborty and E.-B. Cho, J. Ind. Eng. Chem., 2025, 141, 521–539.
83. A. Kittisabhorn, I. Ahmed, S. Ratchahat, W. Chaiwat, W. Koo-amornpattana, W. Klysubun, Y. Poo-arporn, W. Limphirat, P. Khemthong and S. Assabumrungrat, ChemCatChem, 2024, 16, e202400614.
84. P. A. Kamble, C. Vinod, V. K. Rathod and M. L. Kantam, Catal. Today, 2023, 408, 36–49.
85. C.-C. Li, C.-H. Hsieh and Y.-C. Lin, Mol. Catal., 2022, 523, 111720.
86. S. Fang, Z. Cui, Y. Zhu, C. Wang, J. Bai, X. Zhang, Y. Xu, Q. Liu, L. Chen and Q. Zhang, J. Energy Chem., 2019, 37, 204–214.
87. Y. Zhang, Y. Li, C. Shen, C. Yang, H. Wu, C. Jiang, S. Chen, M. Li, Y. Li and W. Ding, Appl. Catal., B, 2025, 361, 124595.
88. Y. Shao, K. Sun, M. Fan, J. Wang, G. Gao, L. Zhang, S. Zhang and X. Hu, Chem. Eng. Sci., 2022, 248, 117258.
89. H. Wang, C. Chen, H. Zhang, G. Wang and H. Zhao, Chin. J. Catal., 2018, 39, 1599–1607.
90. Y. Tang, J. Fu, Y. Wang, H. Guo and X. Qi, Fuel Process. Technol., 2023, 240, 107559.
91. X. Wang, X. Qi, M. Qiu, F. Shen, J. Yang and B. Shen, Fuel, 2023, 341, 127720.
92. X. Chen, H. Tian, L. Deng, S. Zhou, X. Yu, Z. Wang, Y. Li and Y. Tian, Chem. Eng. J., 2025, 508, 160796.
93. Q. Yuan, H. H. Van De Bovenkamp, Z. Zhang, A. S. Piskun, S. Sami, R. W. Havenith, H. J. Heeres and P. J. Deuss, ACS Catal., 2021, 11, 10467–10477.

---

Footnote

† These authors contributed equally to this work.

---