Selective Electrochemical Hydrogenation of Phenol with Earth‐abundant Ni−MoO₂ Heterostructured Catalysts: Effect of Oxygen Vacancy on Product Selectivity

Abstract

Herein, we report highly efficient carbon supported Ni−MoO₂ heterostructured catalysts for the electrochemical hydrogenation (ECH) of phenol in 0.10 M aqueous sulfuric acid (pH 0.7) at 60 °C. Highest yields for cyclohexanol and cyclohexanone of 95 % and 86 % with faradaic efficiencies of ∼50 % are obtained with catalysts bearing high and low densities of oxygen vacancy (O_v) sites, respectively. In situ diffuse reflectance infrared spectroscopy and density functional theory calculations reveal that the enhanced phenol adsorption strength is responsible for the superior catalytic efficiency. Furthermore, 1‐cyclohexene‐1‐ol is an important intermediate. Its hydrogenation route and hence the final product are affected by the O_v density. This work opens a promising avenue to the rational design of advanced electrocatalysts for the upgrading of phenolic compounds.

Incorporation of MoO₂ enhances the catalytic efficiency of Ni in selective electrochemical hydrogenation of phenol in aqueous 0.1 M H₂SO₄ solutions. The origin of the superior performance is revealed by in situ diffuse reflectance infrared spectroscopy and density functional theory calculations.

Introduction

With the increasing demands for sustainable carbon sources and the rapid depletion of fossil feedstocks, it has received increasing attention in the energy and chemical industry to replace fossil sources with renewable biomass in order to combat climate change and energy crisis. ^[1] Conversion of widely available underutilized lignin biomass into value‐added chemicals and fuels is key for the development of holistic biorefinery. ^[2] For example, cyclohexanone, a critical raw material and high‐volume industrial feedstock for the manufacturing of nylon, is usually produced from the selective hydrogenation of the lignin‐derived phenol. ^[3] Other value‐added products, such as cyclohexane and cyclohexanol that are widely used in the chemical and pharmaceutical industry, can also be obtained from phenol hydrogenation (Scheme 1).[ ^2a , ⁴ ]

Scheme 1.

Transformation of phenol to value added products.

The electrochemical hydrogenation (ECH) method, which is undertaken under mild conditions, provides a greener and more sustainable route for phenol hydrogenation in comparison with the traditional thermal catalytic one, which utilizes explosive H₂ or other hydrogen donors (such as HCOOH, NaBH₄). ^[5] During ECH, energetic electrons are used to generate adsorbed H (H_ads) from water and for the activation of phenol.[ ^5c , ⁶ ] Despite the promising feature, the ECH route for phenol hydrogenation still suffers from high‐cost associated with noble metal catalysts, such as Pd, ^[7] Pt, ^[7] Ru ^[8] and Rh.[ ⁷ , ⁹ ] Furthermore, since a large overpotential is needed to activate the stable aromatic ring structure in phenol and to provide sufficient H_ads for the hydrogenation, ^[10] faradaic efficiency (FE) for the ECH products is poor due to severe competition from the kinetically more favourable hydrogen evolution reaction (HER). Thus, there is a considerable room for improvement in terms of electrocatalyst development.

To date, several effective catalyst design strategies have been reported to improve the FE of the ECH products. One of them is to optimize the adsorption of substrates and reaction intermediates on the catalyst. For example, Zhou et al. reported a boron doping strategy for a PtNi alloy catalyst to enhance the FE of cyclohexanol by 13.7 times to 86.2 % in a 0.2 M HClO₄ electrolyte at 60 °C. ^[10a] This enhancement was attributed to the stronger adsorption of phenol after the boron doping, which was consistent with the findings of Cirtiu et al. in their study of phenol ECH on alumina‐supported Pd. ^[11] This strategy was also adopted by Peng et al. who developed a ternary alloy (PtRhAu) electrocatalyst for selective hydrogenating lignin‐derived methoxylated monomers to methoxylated cyclohexanes with a FE of 58 %. ^[12] Further investigation revealed that Rh and Au modulate the electronic structure of Pt, steering intermediate adsorption energetics on the electrocatalyst surface to facilitate the hydrogenation of lignin monomers and suppress C−OCH₃ bond cleavage.

Earth‐abundant Ni catalysts were also used in this reaction to form cyclohexanol with excellent ECH selectivity but low FE.[ ^10b , ¹³ ] For instance, Jackson's group reported the ECH of alkoxyphenols and aryl alkyl ethers with the Raney Ni catalyst.[ ^10b , ¹³ ] After the formation of phenol via cleavage of sp² C−O aryl ether or sp³ C−O ether bond (mainly found in α‐ketone ethers), cyclohexanol was obtained as the sole product from phenol with a FE of ca. 20 %. The authors also found that phenol binds very weakly to the catalyst surface compared to ketones, which may be responsible for the poor FE of cyclohexanol. Metal oxide supported Ni catalysts are widely used for thermal catalytic phenol hydrogenation. The Brønsted acid site of metal oxides can enhance the binding strength of phenol on the catalyst surface via the donation of a lone electron pair from oxygen. ^[14] This strategy can also be useful in phenol ECH reaction.

Herein, we report a carbon supported non‐noble Ni−MoO₂ catalyst (Ni_w@MoO_2−x /C; subscript w represents the weight percentage of Ni used in the synthesis) for ECH of phenol with controllable product selectivity. Defective MoO₂, a semiconductor, is chosen as the support for its low electrical resistance and good chemical stability over a wide pH range. ^[15] The O_v in MoO₂ can also introduce new electron acceptor levels, which facilitate charge transfer from Ni, promote phenol adsorption[ ¹⁶ , ¹⁷ ] and enhance electronic interaction between Ni and MoO₂, thus preventing the aggregation of Ni nanoparticles.[ ¹⁷ , ¹⁸ ] Electron paramagnetic resonance spectroscopy (EPR) data confirms the presence of enriched O_v sites in MoO₂, which decreases with the increasing Ni loading. The results obtained from Fourier‐transform infrared spectroscopy (FTIR) and X‐ray photoelectron spectroscopy (XPS) experiments suggest that MoO_2−x binds firmly with Ni and acts as an electron acceptor. The catalytic efficiency of the catalysts with different Ni loading is investigated with bulk electrolysis. Under optimal conditions (0.10 M H₂SO₄, 60 °C), cyclohexanol and cyclohexanone are obtained with high yields of 95 % and 86 %, with Ni₁₀@MoO_2−x /C and Ni₂₀@MoO_2−x /C respectively. Control experiment confirmed that Ni is the active site and Ni−MoO₂ heterostructure is essential for the high activity. Higher harmonic components of the Fourier transformed alternating current voltammetric (FTacV) data reveal that no underlying electron transfer process associated with Mo is involved in the ECH of phenol. Density functional theory (DFT) calculations and in situ diffuse reflectance infrared spectroscopy (DRIFTS) investigation reveal that the strong interaction between phenol and catalyst surface is responsible for its superior catalytic efficiency and the density of O_v sites affects the product distribution by influencing the hydrogenation route associated with the enol intermediate, 1‐cyclohexen‐1‐ol.

Results and Discussion

The MoO_2−x /C nanowire was synthesized using a previously reported procedure (Scheme 2). ^[19] Anilinium molybdate (Mo₈O₂₆(C₆N₈H₂)₄⋅2 H₂O) was first synthesized (powder X‐ray diffraction (PXRD) patterns shown in Figure S1a, scanning electron microscopy (SEM) shown in Figures S1b&c). Carbon‐supported MoO_2−x (MoO_2−x /C) was obtained by the pyrolysis of anilinium molybdate at 650 °C under an Ar flow. PXRD pattern shows a dominant presence of crystalline MoO₂ in MoO_2−x /C (Joint Committee on Powder Diffraction Standards (JCPDS) No. 01‐086‐0135, Figure 1a). ^[15a] SEM and transmission electron microscopy (TEM) images of the synthesized MoO_2−x /C (Figures S2a&b) show that the nanowire structure is well preserved after pyrolysis, and MoO_2−x nanoparticles are decorated on the nanowire with an average diameter of 48±9 nm, consistent with the value of 52 nm derived from the PXRD pattern using the Scherrer equation. High‐resolution TEM (HR‐TEM) (Figure S2c) shows the lattice fringes of 0.34 nm and 0.28 nm corresponding to the (110) and (101) facets of MoO₂, respectively. A 100 mg mL⁻¹ Ni(NO₃)₂ aqueous solution was added dropwise to an aqueous suspension of MoO_2−x /C. As the isoelectric point for MoO₂ is between 4 and 5, ^[20] the surface of MoO₂ is negatively charged in a neutral solution, benefiting the adsorption of positively charged Ni²⁺ ions. The solid product was collected after centrifugation. After drying, it was reduced in a H₂/Ar (5/95 vol %) atmosphere at 400 °C to obtain Ni_w@MoO_2−x /C. At this pyrolysis temperature, Ni²⁺ was reduced to metallic Ni by H₂. Due to the strong electronic interaction between Ni and MoO₂, Ni nanoparticles were trapped by the oxygen vacancy sites in MoO₂, ^[21] forming a Ni−MoO₂ heterostructure. ^[15a] New Bragg peaks at 44.5° and 51.8° associated with the PXRD patterns for Ni_w@MoO_2−x /C (Figure 1a) match well with those of metallic Ni (JCPDS No. 01‐71‐3740). ^[22] SEM and TEM images of Ni_w@MoO_2−x /C (Figure S3, Figure 1 b&c) illustrate that the wire structure of MoO_2−x /C is preserved during pyrolysis. HR‐TEM images of Ni_w@MoO_2−x /C (Figures 1d&e) show the intimate contact between MoO₂ and Ni. The lattice fringes of 0.34 nm and 0.20 nm are well‐indexed to the (110) plane of MoO₂ and the (111) plane of Ni, respectively. Energy dispersed spectroscopy (EDS) mapping (Figure 1f) and spectrum (Figure 1g) further confirm that the majority of Ni in Ni₁₀@MoO_2−x /C are in contact with MoO_2−x particles. In contrast, the EDS mapping for Ni₂₀@MoO_2−x /C (Figure S4) indicates that some Ni is loaded on the carbon support and agglomerated, which is due to excessive Ni loading amount. The exact amounts of Ni and Mo were determined using inductively coupled plasma optical emission spectroscopy (ICP‐OES). The results are summarized in Table S1. A control sample containing carbon only (C_p) was also prepared from aniline and loaded with Ni (Ni₁₀@C_p, Ni₂₀@C_p) following the same procedure.

Scheme 2.

Illustration of synthesis process for the preparation of Ni_w@MoO_2−x /C from anilinium molybdate.

Figure 1.

Characterization of Ni_w@MoO_2−x /C: a) PXRD patterns for MoO_2−x /C and Ni_w@MoO_2−x /C; TEM images for b) Ni₁₀@MoO_2−x /C and c) Ni₂₀@MoO_2−x /C; HR‐TEM images for d) Ni₁₀@MoO_2−x /C and e) Ni₂₀@MoO_2−x /C; f) HAADF‐TEM image and EDS mapping images for Ni₁₀@MoO_2−x /C, scale bar 1 μm; and g) EDX spectra for Ni₁₀@MoO_2−x /C.

XPS measurements were conducted to reveal the surface composition of the Ni_w@MoO_2−x /C catalyst. Figure S5a shows that the Mo 3d region consists of the contribution from three Mo species, namely Mo^VI, Mo^V, and Mo^IV, indicating a mixed‐valence state of Mo on MoO_2−x /C surface.[ ^15a , ²³ ] Mo^VI at 232.9 and 236.1 eV is the most dominant surface species in all samples, which accounts for more than 40 % of the Mo signal. After Ni deposition, the peaks at 236.1 eV associated with Mo^VI shift to lower binding energy (235.7 eV), indicating the electron redistribution at the interface between Ni and MoO₂. ^[24] Moreover, an increasing content of Mo in lower valance states (IV and V) is observed in Ni_w@MoO_2−x /C, owing to further reduction of Mo in the process of Ni reduction. ^[25] The spectra of O 1s orbital can be deconvoluted to 2 peaks located at 530.8 eV (Metal‐O) and 533.5 eV (C−O). The binding energy for metal‐O shifts to the lower binding energy of 530.4 eV in Ni_w@MoO_2−x /C (Figure S5b), confirming strong electronic interaction between Ni and MoO₂. High‐resolution XPS in the Ni 2p region is shown in Figure S6. The characteristic binding energy at 856.0 eV in Ni_w@MoO_2−x /C can be assigned to Ni²⁺, generated from the oxidation of surface Ni⁰ during storage. ^[15a] In summary, the above‐mentioned results suggest that Ni acts as an electron donor while MoO₂ functions as an electron acceptor in Ni₁₀@MoO_2−x /C, which results in the electron redistribution at the interface between Ni and MoO₂. ^[24] High‐resolution XPS in C 1s region (Figure S7) can be deconvoluted into two peaks, 284.9 eV (C−C) and 288.9 eV (O−C). ^[26]

EPR spectroscopy was then employed to evaluate the surface defects and electron configuration of MoO_2−x /C and Ni_w@MoO_2−x /C (Figure 2a). The strong signal at g=2.004, manifesting the electrons trapping at the O_v sites, which corroborates the presence of abundant O_v.[ ¹⁹ , ²⁷ ] The signal intensity indicates that Mo_2−x /C possesses the highest O_v concentration followed by Ni₁₀@Mo_2−x /C, which may be due to the occupation of Ni in the O_v sites. ^[28] The much lower signal intensity of O_v found in Ni₂₀@MoO_2−x /C (Figure 2a, inset) indicates that Ni may occupy most of the surface O_v sites, consistent with the fact that more and larger Ni nanoparticles were found in Ni₂₀@MoO_2−x /C (Figure S4 vs. Figure 1f). The FT‐IR spectra of MoO_2−x /C and Ni_w@MoO_2−x /C are shown in Figure 2b. The adsorption peak at 990 cm⁻¹, 927 cm⁻¹, and 825 cm⁻¹ can be assigned to the vibration of Mo=O, Mo−O−Mo, and Mo−O species, respectively, confirming the presence of MoO₂ in MoO_2−x /C.[ ²³ , ²⁹ ] After introducing Ni, ν_Mo=O and ν_Mo‐O‐Mo shift to lower wavenumbers, indicating a weaker Mo−O bond after loading of Ni. This is consistent with the lower binding energy of metal‐O for Ni_w@MoO_2−x /C found in HR‐XPS (Figure S5b). Combining the information above, it is concluded that 1) there is a strong interaction between Ni and MoO_2−x particles in Ni_w@MoO_2−x /C, and 2) Ni in Ni₂₀@MoO_2−x /C is excessive, leading to agglomeration of Ni nanoparticles on the surface of carbon wires.

Figure 2.

a) EPR spectra (inset: enlarged EPR spectra in the range of 3300–3600 Gauss) and b) FT‐IR spectra for MoO_2−x /C and Ni_w@MoO_2−x /C.

The catalytic property of the Ni_w@MoO_2−x /C catalyst for ECH of phenol was then evaluated. The cyclic voltammogram (CV) was first recorded in a 0.10 M H₂SO₄ aqueous solution (pH 0.7). The experimental details for the CV measurements are provided in Supporting Information. The potential scale reported in this paper was converted to the reversible hydrogen electrode (RHE) scale using Equation (S1). The HER activity of MoO_2−x /C with different loading of Ni was first tested. The current density (j) at −0.7 V increases from 1.8 mA cm⁻² for MoO_2−x /C to 5.4 mA cm⁻² for Ni₁₀@MoO_2−x /C and 7.3 mA cm⁻² for Ni₂₀@MoO_2−x /C (Figure 3a), suggesting that the HER activity is promoted with increasing Ni loading. Upon addition of 5 mM phenol to the electrolyte, the current density at −0.7 V increases to 9.6 mA cm⁻² and 10.9 mA cm⁻² for Ni₁₀@MoO_2−x /C and Ni₂₀@MoO_2−x /C, respectively (Figures 3c&d). In contrast, the current density remains essentially unaltered for MoO_2−x /C under these conditions (Figure 3b). This enhanced current density associated with Ni₁₀@MoO_2−x /C and Ni₂₀@MoO_2−x /C may be attributed to the contribution from phenol ECH catalysed by Ni, indicating that both catalysts may be promising for the ECH of phenol.

Figure 3.

Cyclic voltammograms recorded at 20 °C in 0.1 M H₂SO₄: a) using MoO_2−x /C, Ni₁₀@MoO_2−x /C and Ni₂₀@MoO_2−x /C modified electrodes; b‐d) in the presence and absence of 5 mM phenol using (b) MoO_2−x /C, (c) Ni₁₀@MoO_2−x /C and d) Ni₂₀@MoO_2−x /C modified electrodes. Scan rate was set at 50 mV s⁻¹.

Controlled potential bulk electrolysis of phenol was conducted at 60 °C in an airtight H‐cell in a solution containing 0.1 M H₂SO₄ and 20 mM phenol. The electrolysis products were analysed after passing 20 C of charge. The results are summarized in Figure 4. High product selectivity towards cyclohexanol is observed with the Ni₁₀@MoO_2−x /C catalyst. The highest FE for cyclohexanol reaches 76 % at −0.7 V vs. RHE. Another product from ECH of phenol is cyclohexane with a FE of ca. 2 %. A further increase in driving force (i.e. more negative applied potential) leads to a decreased FE for cyclohexanol from 76 % at −0.7 V vs. RHE to 30 % at −0.9 V vs. RHE. ^[6b] It also diminishes the FE for the total ECH products. This trend is consistent with the previous studies employing noble metal catalysts and can be attributed to the insufficient H_ads coverage at −0.6 V and more severe HER competition at highly negative potentials.[ ⁷ , ^10a ] ECH reaction competes with HER for surface H_ads. Insufficient H_ads at −0.6 V limits the phenol hydrogenation to cyclohexanol as it requires 6 H_ads per phenol (Scheme 1), leading to a low FE for this product. ^[7] At −0.7 V, more H_ads atoms are available at the catalyst surface enhancing the kinetics of phenol hydrogenation. As the potential goes further negative, the HER becomes more dominant, diminishing the FE for the ECH products. Phenol hydrogenation competes for H_ads with the HER and shows a stronger dependence on the concentration of H_ads than HER since it requires more H_ads to form an ECH product (Scheme 1).[ ^6b , ⁷ ] Therefore, the lower H_ads concentration at −0.6 V vs. RHE benefits the HER, leading to a lower FE for ECH. The enhanced H_ads concentration at −0.7 V vs. RHE promotes the ECH reaction, showing a higher FE for ECH. However, a more negative applied potential (<−0.8 V vs. RHE) results in a diminished FE for ECH due to severe HER competition, which has been commonly observed in ECH of biomass molecules.[ ^5c , ^5d ] While for Ni₂₀@MoO_2−x /C, high ECH product selectivity toward cyclohexanone is found with a small fraction of cyclohexanol as a by‐product (ca. 2 % FE), giving the highest FE for cyclohexanone of 71 % at −0.6 V vs. RHE. This observation is in a huge contrast to Ni₁₀@MoO_2−x /C, where cyclohexanol is a major ECH product. The less negative optimal potential required for Ni₂₀@Mo_2−x /C compared with Ni₁₀@MoO_2−x /C can be attributed to the higher HER activity, providing sufficient H_ads for ECH product formation at a less negative potential. As a more negative potential is applied, the FE for cyclohexanone continually drops with a value of 16 % found at −0.9 V due to the HER competition. Since the characterisation data, including the PXRD patterns for the crystal structure (Figures 1a and S8a), HR XPS of the Ni 2p orbital for the valence state (Figure S6) and HR TEM images for the exposed crystal facet (Figures 1d&e), do not show distinct difference in Ni, the drastic difference in product selectivity for Ni₁₀@MoO_2−x /C and Ni₂₀@MoO_2−x /C is tentatively attributed to the difference in their oxygen vacancy densities based on the EPR data (Figure S8b). To verify this hypothesis, two additional catalysts with different Ni loadings (5 % and 15 %) were prepared (XRD shown in Figure S8a). Ni₅@ MoO_2−x /C with a high O_V density (Figure S8b) shows similar product selectivity as Ni₁₀@MoO_2−x /C, producing cyclohexanol as the main product but with a lower current density (Figure S9a), which may be due to a lower amount of surface Ni. Ni₁₅@MoO_2−x /C with a considerable amount of O_V sites (Figure S8b) also shows excellent selectivity for cyclohexanol, while the FE is diminished due to severe HER competition (Figure S9b). These results confirm that the O_V site density plays a significant role in determining the product selectivity of the Ni_w@MoO_2−x /C catalyst.

Figure 4.

Summary of bulk electrolysis results in 0.1 M H₂SO₄ with 20 mM phenol with different catalysts: a) Ni₁₀@MoO_2−x /C b)  Ni₂₀@MoO_2−x /C. 20 C of charge were collected. A carbon paper (2 cm×1 cm) with a catalyst loading of 1 mg cm⁻² was used as the working electrode. Temperature=60 °C.

The influence of reaction temperature on the catalytic efficiency of both catalysts was further investigated. Decreasing the temperature causes a significant drop of the total current density and FE for the ECH products for both catalysts while the ECH product selectivity remains almost unaltered (Figure S10). Taking the result from Ni₁₀@MoO_2−x /C as an example (Figure S10a), only 3 % of cyclohexanol is obtained at 20 °C after consuming 20 C of charge. When the temperature is increased to 40 °C, the FE for cyclohexanol increases to 30 %. In both cases, cyclohexane is obtained as a by‐product with low FE. Higher temperature enhances the reaction rate, and hence the current density. As the ECH of phenol is more sensitive to the temperature than the HER, higher temperature favours the ECH reaction, ^[30] leading to a higher FE for the ECH products. This effect is consistent with the results reported by other groups.[ ^6b , ^10a , ³¹ ] The effect of pH was also evaluated in 0.50 M H₂SO₄ at 60 °C. The results are shown in Figure S11. At a lower pH, the ECH product selectivity doesn't change for both Ni₁₀@MoO_2−x /C and Ni₂₀@Mo_2−x /C while lower FEs for ECH products are observed, which may be due to more severe HER competition.

Long‐term electrolysis experiments (ca. 2 h) were conducted with both catalysts to assess their stability. The results are summarized in Figure 5. As more charges are passed, the selectivity of the ECH products for both catalysts, calculated using Equations (S7) and (S8), remains essentially unchanged. High yields of 95 % of cyclohexanol and 86 % of cyclohexanone are obtained with Ni₁₀MoO_2−x /C and Ni₂₀MoO_2−x /C, respectively. The production rate is also calculated using Equations (S9) and (S10) ^[32] and summarized in Table S2. The FE for the ECH products keeps decreasing as more phenol was consumed, which could be explained by the competition for the reaction site between phenol and H⁺. The continuous decrease in phenol concentration favours H₂O adsorption. This is also the origin of the departure from the linearity in the Yield of cyclohexanol vs Charge collected relationship when 250 C of charges are collected at −0.7 V (Figure 5a). The performance of our non‐noble metal Ni_w@MoO_2−x /C catalysts are comparable to those of the state‐of‐the‐art noble metal catalysts, taking both FE and ECH product selectivity into consideration although the ECH partial current density is lower compared with those obtained with noble metal (Table S3).

Figure 5.

Summary of long‐term electrolysis results obtained at 60 °C in a solution containing 0.10 M H₂SO₄ and 20 mM phenol with a) Ni₁₀@MoO_2−x /C (at −0.7 V vs. RHE), b) Ni₂₀@MoO_2−x /C (at −0.6 V vs. RHE). A 4 cm×2 cm carbon paper with a catalyst loading of 1 mg cm⁻² was used as the working electrode.

The catalyst recycling tests were further performed (Figure S12). In this test, bulk electrolysis experiments were repeated 5 times. Each time 20 C of charge were collected. The results suggest that Ni₁₀@MoO_2−x /C and Ni₂₀@MoO_2−x /C remain stable. The content of Ni in the catalysts was determined by ICP‐OES after electrolysis. The value is nearly the same as that found in freshly prepared samples (Table S1). The electrolyte solution in the cathode chamber was also analysed by ICP‐OES after electrolysis. The content of Ni is below the detection limit (0.01 ppm). Therefore, the possibility of Ni leakage can be ruled out. This is not surprising since under the ECH conditions, the oxidation state of Ni is 0 based on the standard electrode potential of the Ni^2+/0 process. ^[33] The ESR spectrum taken before and after the recyclability test indicates an increase in oxygen vacancy sites during electrolysis (Figure S13). This increase is due to the generation of new oxygen vacancy sites while applying negative potential, which is supported by the FTacV results (see below).

Experiments were conducted to reveal the active site for this multicomponent catalyst. MoO_2−x /C was first investigated in bulk electrolysis under the optimal conditions. Only H₂ is obtained (Figure S14a), indicating that MoO_2−x /C alone is not active for the ECH of phenol. A Mo free Ni catalyst (Ni₁₀@C_p, a XRD spectrum shown in Figure S15) was also prepared and tested in the bulk electrolysis experiment. HER is the dominant reaction and the highest FE for cyclohexanol obtained at −0.7 V vs. RHE is only 19 %, far less than the value of 76 % obtained with Ni₁₀@MoO_2−x /C (Figure S14b).

The above results suggest that both MoO₂ and Ni sites are essential for high ECH efficiency. The effect of different Mo−Ni interfaces on catalytic efficiency was therefore investigated. The Ni−MoO₂ heterostructure material can be transformed to NiMo alloy (Ni−Mo interface without O) by increasing the pyrolysis temperature. ^[15a] Therefore, Ni (10 wt %) loaded MoO_2−x /C was treated at higher temperatures under a H₂/Ar (5 : 95 vol %) atmosphere to obtain NiMo materials (XRD shown in Figure S16), and the effect of the pyrolysis temperature on the structure and the performance of the NiMo materials for ECH of phenol was evaluated. Significant loss of activity and selectivity is only observed when the pyrolysis temperature is increased above 600 °C (Figure S17) when a stable NiMo alloy was formed. ^[15a] These results suggest that the Ni−MoO₂ heterostructure is vital for the superior ECH efficiency.

In situ DRIFTS was then employed to confirm the reaction route and intermediates. Figure 6a shows the potential‐dependent DRIFTS spectra obtained with Ni₁₀MoO_2−x /C and Ni₂₀MoO_2−x /C catalysts in a 0.1 M H₂SO₄ solution containing 10 mM phenol. The spectra at open circuit potential (OCP) exhibit bands at 1540 cm⁻¹, 3150 cm⁻¹, 3270 cm⁻¹, 3670 cm⁻¹ corresponding to the different vibrational modes of phenoxy species, confirmed by comparing to the blank without phenol (Figure S18). The bands at 1540 cm⁻¹ and 3670 cm ⁻¹ are observed with both catalysts and can be assigned to the vibrations of the aromatic ring ν(C=C_ring) and hydroxyl group ν(O‐H_aromatic), respectively. ^[10a] Whereas the bands at 3150 cm⁻¹ and 3270 cm⁻¹, which are only obvious with Ni₂₀MoO_2−x /C, correspond to the ν (C−H) stretching mode of the aromatic ring of phenol. ^[34] The DRIFTS spectra for phenol adsorbed on Ni₁₀@C_p catalyst without MoO₂ were also collected. No characteristic peaks for phenol are found at OCP, indicating a weak adsorption of phenol on Ni₁₀@C_p. ^[35] These observations indicate that the adsorption of phenol is enhanced in the presence of MoO₂ and phenol is adsorbed via aromatic ring and hydroxyl group. ^[35b] As a negative potential is applied, a new peak appears at 1600–1700 cm⁻¹ for both catalysts which can be assigned to the carbonyl group. Moreover, the peak observed in Ni₁₀@MoO_2−x /C blue‐shifts to 1680 cm⁻¹ compared to that found in Ni₂₀@MoO_2−x /C (1720 cm⁻¹), which is the typical band for the stretching mode of carbonyl groups (Figure S19). This is due to the interaction between C=O bond and surface H, indicating that the enol type intermediate is in a protonated form ^[36] and likely to be further hydrogenated to cyclohexanol (Figure 6b, Route a). The appearance of a new peak at 3550 cm⁻¹ at negative potentials, which is a characteristic peak for O−H in cyclohexanol, further confirms the formation of cyclohexanol from enol intermediate. While for Ni₂₀@MoO_2−x /C, no new peak is observed at 3550 cm⁻¹, indicating little cyclohexanol was formed. Moreover, the increase intensity at 1720 cm⁻¹ confirms the formation of cyclohexanone (Figure 6b, Route b), consistent with the electrolysis results (Figure 4b & Figure 5b).

Figure 6.

a) Potential‐dependent in situ DRIFTS obtained in a 0.1 M H₂SO₄ solution containing 10 mM phenol in the potential range from OCP to −0.9 V. Spectra were collected 10 min after a certain potential was applied to obtain a stable signal; b) Proposed reaction route via 1‐cyclohexen‐1‐ol intermediate.

MoO _x based materials can undergo an extensive series of electron transfer reactions. ^[15a] To investigate whether electron transfer processes involving MoO _x occur under the ECH conditions, FTacV experiments were undertaken. In FTacV, a large amplitude periodic ac waveform is superimposed onto the direct current (dc) ramp to generate higher harmonic components that are devoid of background charging current and insensitive to catalytic process, allowing fast underlying electron transfer processes to be detected. ^[37] This technique has been used previously to detect the electron transfer process that directly underpinned the electrocatalytic water oxidation, ^[38] hydrogen evolution, ^[39] and CO₂ reduction. ^[37c] Figure S20 shows the fifth harmonic component of the FT ac voltammograms of C_p, MoO_2−x /C, Ni₁₀@MoO_2−x /C and Ni₂₀@MoO_2−x /C obtained in a solution containing 0.10 M H₂SO₄. Two well‐defined electron transfer processes are observed at the potentials of −0.18 and −0.40 V with MoO_2−x /C, while these processes are not found with C_p suggesting that they are associated with MoO_2−x . These two electron transfer processes are also observed with Ni₁₀@MoO_2−x /C and Ni₂₀@MoO_2−x /C, with the peak potential for the first reduction process remaining essentially unchanged, while that for the second process shifts positively by ∼40 mV to −0.36 V for Ni₁₀@MoO_2−x /C and further more positively to −0.28 V for Ni₂₀@MoO_2−x /C. These results could be indicative of charge delocalization due to incorporation of Ni, as found from the XPS measurements (Figure S5). Since these processes occur prior to the onset potential of the ECH process, they are not expected to directly involve in ECH. However, more oxygen vacancies could be generated from these redox processes to promote the selective hydrogenation of phenol to cyclohexanol with Ni₁₀@MoO_2−x /C, as found from the EPR measurements (Figure S13). As for Ni₂₀@MoO_2−x /C, the second electron transfer process shifted further more positively, while the magnitude of current for both processes decrease significantly.

Phenolic compounds with other functionals groups were also investigated under optimal conditions using the Ni₁₀@MoO_2−x /C catalyst. Alkyloxy groups at para or ortho positions were removed when the aromatic ring was reduced (Table S4, Entry 1&2). Noting that cyclohexanol was inert under this condition, the alkyloxy group may be removed via the hydrogenolysis route. A lignin model compound, diphenyl ether, was further used as substrate, which produced cyclohexanol with a yield of 81 % (Table S4, Entry 3). Benzene was also detected as another C−O cleavage product of diphenyl ether. Cyclohexanol was produced when halogen substituted phenols were used as substrate (Table S4, Entry 4&5), while amide and carboxylic acid groups remained intact (Table S4, Entry 6&7), producing substituted cyclohexanol with high yields (ca. 80 %). In conclusion, besides hydrogenation of lignin derived phenolic compounds, our method is also applicable in producing other substituted cyclohexanols bearing reducible groups except halogens.

Computational simulations are also beneficial to identify the origin of the superior catalytic efficiency of Ni_w@Mo_2−x /C catalysts. DFT calculations were carried out by constructing Ni (111) (Figure S21) and MoO₂ (110) slabs as model systems of Ni and MoO₂ respectively according to the XRD results (Figure 1a). To simplify the models of Ni₁₀@MoO_2−x /C and Ni₂₀@MoO_2−x /C (Figures S22 and S23), we constructed Ni−MoO₂ slabs with oxygen vacancy (Ni/MoO₂−O_V) and without oxygen vacancy (Ni/MoO₂) according to the EPR results (Figure 2). The different adsorption sites of phenol, 1,3‐cyclohexadien‐1‐ol, 1‐cyclohexen‐1‐ol, cyclohexanol and cyclohexanone on the three surfaces were calculated and the most stable adsorption configurations of all species are shown in Figure 7. Phenol prefers to adsorb on Ni (111) with its aromatic ring lying flat on the surface (E _ad=−0.83 eV), which agrees with the previous reports. ^[8] After introducing MoO₂, the adsorption of phenol via aromatic ring is still preferred, but the adsorption is significantly enhanced (E _ad=−2.10 eV on Ni/MoO₂ and E _ad=−2.16 eV on Ni/MoO₂−O_V), indicating a stronger interaction with phenol after introducing MoO₂. Moreover, the adsorption of intermediates are also notably enhanced due to the strong electronic interaction between Ni and MoO₂. Therefore, ECH reaction is more likely to occur on Ni−MoO₂−O_v and Ni−MoO₂ surfaces than on Ni surface and initial reduction on the benzene ring is preferred, ^[10a] consistent with our experimental results. In addition, the adsorption of cyclohexanone is relatively weak (E _ad=−0.12 eV) on the Ni surface and moderately enhanced after introducing MoO₂ (E _ad=−0.64 eV). It is worth noting that the presence of oxygen vacancies in MoO₂ adjacent to Ni can greatly enhance the adsorption of cyclohexanone via oxygen in cyclohexanone (E _ad=−1.37 eV). Such bonding mode might affect the product distribution.

Figure 7.

a) The most stable structure of adsorbates (phenol, intermediates) on Ni, Ni/MoO₂−O_V and Ni/MoO₂, and corresponding adsorption energy. b) Probable pathways for the reaction involving 1‐cyclohexen‐ol on Ni, Ni/MoO₂−O_V and Ni/MoO₂, and corresponding activation energy data (inset Table).

The free energy profiles of ECH of phenol to cyclohexanone/cyclohexanol at pH 0 and potential (U vs. RHE)=0 on Ni, Ni/MoO₂−O_V and Ni/MoO₂ surfaces were then calculated and compared (Figure S24). The formation of 1,3‐cyclohexadien‐1‐ol is the rate‐determining step with the highest energy barriers (0.60, 0.62 and 0.80 eV) for all three surfaces. For Ni surface, formation of cyclohexanol via direct reduction of benzene ring instead of via cyclohexanone as an intermediate is preferred from the thermodynamic point of view. For Ni/MoO₂ and Ni/MoO₂−O_V, once 1‐cyclohexen‐1‐ol was generated, the production of cyclohexanol and cyclohexanone are all thermodynamically possible due to the subtle difference between the energy barriers for the production of cyclohexanol or cyclohexanone. However, the oxygen in cyclohexanone can occupy the oxygen vacancy in MoO₂, making the desorption of cyclohexanone very difficult. Therefore, it is hard to reveal the origin of different production distribution observed in experiment from the thermodynamic aspect. Therefore, kinetics (i.e., reaction activation energy) associated with the reduction of 1‐cyclohexen‐ol on these three surfaces were calculated. Two possible routes starting from 1‐cyclohexen‐1‐ol were calculated (Figure 7, Figure S25–S27): Route a) Direct hydrogenation: the double bond in 1‐cyclohexen‐1‐ol is hydrogenated to form cyclohexanol. Route b) Keto‐enol tautomerization: H in hydroxyl of 1‐cyclohexen‐ol is abstracted by the catalyst to form cyclohexanone. For Ni/MoO₂, Route b is favored with a lower activation energy (E _a ¹: 0.40 eV/E_a ²: 0.63 eV vs. E _a ⁴: 0.80 eV for Route a). Oxygen in MoO₂ can facilitate H abstraction from 1‐cyclohexen‐ol, lowering the activation energy (E _a ¹). Furthermore, the higher activation energy for cyclohexanone hydrogenation (E _a ³: 0.98 eV) inhibits the further hydrogenation of cyclohexanone to cyclohexanol on Ni/MoO₂, in line with the in situ DRIFTS and electrolysis results in which only cyclohexanone was found with the Ni₂₀@MoO_2−x /C catalyst. While for Ni/MoO₂−O_V, the activation energy for H abstraction and further hydrogenation increased significantly (E _a ¹: 0.70 eV/E _a ²: 0.71 eV) compared to that of Ni/MoO₂, and the activation energy for Route a decreases slightly (E _a ⁴: 0.63 eV), smaller than that of Route b, indicating that the formation of cyclohexanol via Route a is kinetically favored. In conclusion, MoO₂ significantly enhances the adsorption of phenol and intermediates, accelerates the enol‐keto tautomerization by facilitating H abstraction, leading to kinetically favored cyclohexanone. The presence of O_v sites further enhances the adsorption of cyclohexanone and increases the energy needed for the enol‐keto tautomerization (E _a ¹=0.70 eV, E _a ²=0.71 eV), benefiting the direct hydrogenation of phenol to cyclohexanol.

Conclusion

Non‐noble metal Ni_w@MoO_2−x /C catalysts were developed for selective ECH of phenol. MoO₂ was introduced to enhance the adsorption of phenol on the catalyst surface. The O_v rich Ni₁₀@MoO_2−x /C catalyst exhibited a yield of 95 % for cyclohexanol with a moderate FE of 53 % at −0.7 V vs. RHE in 0.1 M H₂SO₄, while increasing the Ni content (Ni₂₀@MoO_2−x /C) diminished the density of O_v sites, leading to a high yield of 86 % for cyclohexanone. Control experiments indicated that the Ni−MoO_2−x heterogeneous structure was responsible for the superior ECH efficiency. FTacV experiments revealed the presence of two well‐defined electron transfer processes associated with Mo in the potential region prior to the ECH of phenol suggesting that these electron transfer processes were not directly involved in the ECH of phenol. Nevertheless, these electron transfer processes could increase the number of O_v sites. DFT calculation and in situ DRIFTS experiment confirmed the strong interaction between phenol/intermediates and catalyst surface. The oxygen vacancy was responsible for the different selectivity of catalysts with different Ni loadings. This work sheds light on the rational design of non‐noble metal catalysts for ECH of phenol.

Conflict of interest

The authors declare no conflict of interest.

Supporting information

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer reviewed and may be re‐organized for online delivery, but are not copy‐edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors.

Supporting Information

Zhou P., Guo S.-X., Li L., Ueda T., Nishiwaki Y., Huang L., Zhang Z., Zhang J., Angew. Chem. Int. Ed. 2023, 62, e202214881; Angew. Chem. 2023, 135, e202214881.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.