Potential-mediated enol-keto equilibrium in phenol electro-hydrogenation on platinum

Abstract

The selective formation of cyclohexanone under cathodic potentials during electrocatalytic hydrogenation (ECH) of phenol on transition metal catalysts attracts extensive interest, yet the mechanistic understanding of this potential-dependent product distribution remains limited. By combining ab initio molecular dynamics with explicit solvation and electrode potential control, we demonstrate that cathodic polarization shifts the enol-keto equilibrium toward the keto form through surface-mediated charge transfer, thereby promoting aromatic ring hydrogenation to cyclohexanone. Moreover, the negatively charged surface electrostatically stabilizes cyclohexanone in a metastable “floating” state, which kinetically suppresses its further hydrogenation. This work addresses the question of cyclohexanone selectivity in phenol ECH by identifying potential-controlled tautomerization as the key step affecting selectivity, offering a clear mechanistic basis for the experimentally observed product distribution.

Electrocatalytic hydrogenation of phenol provides valuable chemicals, but the origin of selective cyclohexanone formation remains elusive. Here, the author report that cathodic polarization drives enol–keto tautomerization, trapping cyclohexanone in a floating state for selective accumulation.

Introduction

The growing dependence on fossil fuels has led to severe environmental challenges, driving urgent demands for sustainable alternatives^1–6. As a key lignin-derived platform molecule, phenol represents an ideal model system for studying biomass electrocatalytic upgrading, particularly through electrocatalytic hydrogenation (ECH)^7–9 - an emerging green approach that operates under ambient conditions with high efficiency. Unlike conventional thermocatalytic processes requiring harsh temperatures and pressures, ECH enables precise control over reaction pathways through electrode potential modulation^10–13. Electron transfer between organic species and catalyst surfaces has been extensively documented¹⁴. Crucially, such charge transfer could modify molecular bond configuration. In the domain of the single-molecule level, applied voltage or optical field variations have been shown to precisely regulate keto-enol tautomerization, enabling reversible switching between forms and modulating molecular conductance^15–18. Recent advancements have extended this principle to electrocatalytic processes, where charge-induced tautomerization has been experimentally confirmed, offering alternative avenues for controlling reaction pathways under electrochemical conditions¹⁹. Central to phenol ECH is the keto-enol tautomerism, which fundamentally dictates product selectivity. However, whether tautomerization of phenol, possibly initiated by electron transfer or regulated by varying the applied potential, subsequently influences the catalytic pathway remains an open question. The equilibrium between keto and enol intermediates directly determines the reaction network, making tautomerization control crucial for the targeted production of these high-value chemicals^20–28. While cyclohexanone serves as an essential precursor for nylon production and cyclohexanol finds wide applications in polymers and solvents, the mechanistic origins of their selective formation during phenol hydrogenation, particularly under electrochemical conditions, remain incompletely understood. The key challenge lies in understanding how electrochemical interfaces, particularly under applied potentials, regulate the tautomeric equilibrium and subsequent hydrogenation steps at the atomic scale - a knowledge gap that hinders the rational design of efficient ECH systems.

A variety of metal catalysts, such as Pt^29–34, Pd^34–38, and Rh^32,34,39,40 have been widely explored for the hydrogenation of phenol. Pt and Rh both exhibit high intrinsic activity and similar product distributions in phenol ECH, where cyclohexanone is the dominant product at low conversion, and cyclohexanol becomes prevalent at high conversion³⁴, while Pd-based catalysts are more commonly employed in thermocatalytic hydrogenation. Since Pt has been most widely employed as a benchmark electrocatalyst due to its well-defined surface structures and well-established performance in ECH, it serves as an ideal model catalyst for theoretical investigation of the keto–enol equilibrium and reaction mechanism in phenol ECH. The hydrogenation of phenol follows two competing pathways: (1) ring saturation containing keto-enol tautomerization to yield cyclohexanone, or (2) direct hydrogenation to form cyclohexanol. Despite substantial progress in elucidating the mechanisms of phenol hydrogenation, controversy persists over the precise stage at which keto–enol tautomerization occurs during the reduction process. Most mechanistic understanding favors initial hydrogenation over tautomerization due to thermodynamic and kinetic constraints, with the first hydrogenation step typically presenting the highest energy barrier^31,41–44. This view is supported by studies showing that phenol dissociation inhibits phenoxy hydrogenation⁴², and that tautomerization of hydrogenated phenol has generally been regarded as a viable route to cyclohexanone^41,44,45. However, emerging experimental evidence challenges this paradigm. Recent operando Raman spectroscopy revealed 2,4-cyclohexadienone as the primary intermediate in phenol ECH, leading to predominant cyclohexanone formation⁹. Similar observations were reported for hydrodeoxygenation (HDO) reactions, where tautomerization precedes hydrogenation on Pd-based catalysts, which exhibit high selectivity towards cyclohexanone^27,46. These studies highlight the importance of tautomerization in the formation of cyclohexanone. In addition, kinetic studies demonstrated that during phenol ECH, cyclohexanone generation precedes its further hydrogenation to cyclohexanol on Pt^34,47, featuring slower reaction kinetics compared to aromatic ring hydrogenation.

These observations were made in Pt-based ECH systems, where reactions occur at electrified interfaces, with the applied potential modulating the interfacial charge distribution and the electric field. Such effects are known to influence the stability and reactivity of adsorbed intermediates. Given that keto–enol tautomerization involves substantial charge redistribution, its energetics may be particularly sensitive to the interfacial electrostatic environment. This raises a central question: could electrode potential regulate the balance between keto–enol tautomerization, analogous to the regulation at the single-molecule level? Specifically, does cathodic polarization actively promote ketone formation through interfacial charge transfer? And if so, why does the resulting cyclohexanone resist further hydrogenation despite favorable thermodynamics? These questions lie at the heart of designing selective ECH processes, yet the atomic-scale dynamics at electrified interfaces have remained elusive, hindered by the challenge of modeling explicit solvation and potential effects in reactive simulations. To address these mechanistic questions, we employ ab initio molecular dynamics with explicit inclusion of solvation and potential effects to investigate phenol tautomerization on Pt under electrochemical conditions. Our simulations reveal that cathodic potential promotes enol-keto tautomerization via surface-mediated electron transfer, thereby activating the phenyl ring hydrogenation, while simultaneously suppressing cyclohexanone hydrogenation due to electrostatic repulsion-induced unstable adsorption. This potential-driven dual control mechanistically explains the formation and accumulation of cyclohexanone.

Results

Structural evolution and surface-mediated electron transfer of adsorbed species

Initial investigations examined the adsorption behaviour of phenol and its tautomer, 2,4-cyclohexadienone, on the Pt (111) surface under varying electrochemical conditions. Consistent with prior studies, phenol adopts a characteristic bowl-shaped geometry on Pt (111), with tilted C-H and O-H bonds, a structural feature mirrored in the C-H bonding of the adsorbed tautomer (tautomer*) (Fig. S1)^45,48. Detailed analysis of C-O bond length distribution provides critical mechanistic insight: phenol* exhibits a unimodal peak at 1.32 Å, indicative of a C-O single bond, whereas tautomer* displays a bimodal distribution with peaks at 1.26 Å (corresponding to an intact C = O double bond) and 1.32 Å (reflecting an activated C-O single bond, with 1.30 Å defined as the threshold for distinguishing single versus double bonds in this study, where bonds shorter than 1.30 Å are classified as double bonds), suggesting that the catalyst surface partially activates the C = O bond (Fig. 1a, raw data is shown in Fig. S2). This bond-length distribution shifts markedly with applied potential, as depicted in Fig. 1a, synchronized changes in carbonyl adsorption height (h_C=O) and bond length (d_C=O) reveal that a decrease in h_C=O correlates with an elongation of d_C=O, consistent with bond activation as the carbonyl group approaches the surface. Notably, this trend is attenuated at -0.37 V and -0.62 V, where d_C=O stabilizes at ~1.26 Å, h_C=O remains constant at ~2.7 Å and ~3 Å, respectively, and the population of the C-O single bond diminishes significantly (Fig. 1b). This passivation is attributed to electrostatic repulsion between the carbonyl group and the negatively charged Pt (111) surface, which redirects hydrogenation preferentially to the phenol ring rather than the carbonyl oxygen, as mechanistically substantiated in subsequent analyses. Bader charge analysis (Figs. 1c and S3) further demonstrates that tautomer* accumulates more electrons (~-0.07 e) compared to phenol* (~0.34 e), corresponding to a net electron gain of ~0.4 e per molecule during tautomerization. This electron accumulation, absent in the aqueous phase (Fig. S4), signifies a surface-mediated reduction process, consistent across applied potentials of -0.37 V and -0.62 V, which indicates that tautomerization on the catalyst involves electron transfer from the electrode to the adsorbate, underscoring its reductive character and suggesting that the reaction could be effectively modulated by varying the applied potential, as validated by the electrochemical responses detailed in the following section.

Fig. 1. Structural evolution and charge distribution of phenol and its tautomer on Pt(111).

a C-O bond length distributions of tautomer* at three conditions, i.e., 0 V, -0.37 V and -0.62 V from the final 5 ps AIMD trajectory, respectively. b Percentage of C-O bond types and corresponding adsorption height during AIMD simulations at three conditions. c Bader Charge of phenol* (red) and tautomer* (blue) on Pt (111). Atom colors: Pt (grey), C (brown), O (red). Source data for Fig. 1 are provided as a Source Data file.

Mechanistic pathways and potential-dependent energetics of tautomerization

To validate the proposed hypothesis, we investigated the tautomerization of adsorbed phenol (phenol*) on the Pt(111) surface at three distinct applied potentials. The tautomerization mechanism proceeds via two sequential elementary steps: (i) dissociation of the hydroxyl group from phenol* to form adsorbed phenoxy (phenoxy*) and (ii) hydrogen transfer to the ortho-carbon position, yielding the keto-tautomer^49,50. Prior studies suggest that this process may generate either adsorbed hydrogen (H*) or a hydronium ion (H₃O⁺), enabling subsequent hydrogenation via the Langmuir-Hinshelwood (LH) or Eley-Rideal (ER) mechanisms, respectively^42,51. To elucidate the structural dynamics, we performed molecular dynamics simulations (10 ps) of phenol* and phenoxy* species, with detailed structural analysis conducted on the final 5 ps of the trajectories. This analysis revealed a critical structural reorganization following hydroxyl dissociation: the average C-O bond length contracts by approximately 0.1 Å to 1.25 Å, as shown in Fig. S5, consistent with the formation of a carbonyl (C = O) group characteristic of the keto-tautomer. This structural evolution strongly suggests that keto-tautomer formation is the dominant outcome of dissociation. The energetics of two competing dissociation pathways were calculated, as detailed in Fig. 2. (i) The water-assisted dissociation pathway exhibits negligible dependence on applied potential, maintaining a consistently low activation free energy barrier (ΔG^‡) of ~0.1 eV across the studied potential range, indicative of its thermodynamic accessibility. (ii) In stark contrast, the surface-mediated dissociation pathway demonstrates pronounced cathodic activation, with its ΔG^‡ decreasing significantly from 0.83 eV at 0 V to 0.46 eV at -0.37 V and further to 0.18 eV at -0.62 V. Concurrently, the reaction free energy (ΔG) for this pathway transitions from endothermic (0.35 eV at 0 V) to exothermic (0.04 eV at -0.37 V and -0.46 eV at -0.62 V), underscoring the thermodynamic favorability induced by negative potentials. The energetics of hydrogenation at the ortho-carbon position via the LH and ER mechanisms were further analyzed, as depicted in Fig. 2^44,45,52. The LH pathway exhibits potential-independent activation barriers of ~1.0 eV. In contrast, the ER pathway shows a moderate reduction in its activation barrier of ~0.20 eV at -0.62 V compared to 0 V. However, it remains energetically disfavored due to restricted water molecule access at the hydrophobic aromatic interface, which hinders efficient proton delivery. Significantly, ΔG for tautomerization decreases from 0.75 eV to 0.17 eV with increasingly negative potentials, exhibiting a pronounced linear correlation with the applied potential, as illustrated in Fig. 3a. This relationship, which also extends to the ΔG of direct phenol* hydrogenation via the LH mechanism, demonstrates that tautomerization becomes thermodynamically more favorable than hydrogenation at lower potentials. These findings corroborate the hypothesis that the reaction proceeds via an electron-accepting reduction process, with cathodic polarization preferentially driving keto-tautomer formation over direct hydrogenation.

Fig. 2. Energy profiles of the tautomerization of phenol at various electrode potentials.

a 0 V; b -0.37 V; c -0.62 V. The tautomerization of phenol* via the Langmuir-Hinshelwood (LH) mechanism and Eley-Rideal (ER) mechanism is presented in purple and blue, respectively. Atom colors: Pt (grey), C (brown), H (white), O (red). Source data for Fig. 2 are provided as a Source Data file. Source data for Fig. 2 are provided as a Source Data file.

Fig. 3. Thermodynamic and kinetic analysis of phenol hydrogenation and water structure at the Pt(111)/electrolyte interface.

a Linear relationship between the applied potential and ΔG of tautomerization (in blue) and of direct hydrogenation (in orange). b Free energy gradient data along the reaction coordinate of surficial dissociation of phenol* and graphic illustrations of the metastable state during the reaction. Error bars represent 90% confidence intervals estimated by bootstrap resampling (n = 10,000 iterations) over the last 2 ps of AIMD trajectories. c Potential-dependent free energy profiles of the Volmer step at three applied potentials. d The distribution of the dipole angle of water molecules from the last 5 ps AIMD of phenol adsorbed on Pt (111). Atom colors: Pt (grey), C (brown), H (white), O (red). Source data for Fig. 3 are provided as a Source Data file.

We elucidate the observed decrease in ΔG at negative cathodic potentials, primarily attributed to the formation of the adsorbed hydrogen intermediate (H*). As depicted in the free energy profiles in Fig. 2 and the energy gradient data from constrained molecular dynamics in Fig. 3b, the dissociation of phenol* via the surface-mediated pathway encounters a significantly higher activation free energy barrier (ΔG^‡ = 0.83 eV) at 0 V compared to -0.37 V (ΔG^‡ = 0.46 eV) and -0.62 V (ΔG^‡ = 0.18 eV). This difference arises from the presence of a metastable intermediate state at -0.62 V, stabilized by enhanced Coulombic interactions between the negatively charged electrode surface and the positively charged hydrogen atom, reducing the energy barrier for this reaction. Concurrently, a negatively charged surface enhances the thermodynamic favorability of the Volmer step (H₃O⁺ + e^- + * → H* + H₂O), as shown in Fig. 3c, reducing both the ΔG^‡ and ΔG, rendering the reaction more exergonic at increasingly negative potentials. Furthermore, the cathodic potential induces a reorientation of interfacial water molecules, favouring the ‘H-down’ configuration, as illustrated in Fig. 3d. This configuration lowers the energy barrier for O-H bond cleavage in H₃O⁺, facilitating direct proton transfer to the Pt (111) surface kinetically. Consequently, these changes enhance the formation of H*, which serves as the dominant hydrogen source for the hydrogenation of phenoxy* (formed via phenol* dissociation), despite the low kinetic barrier for H₃O⁺ formation. The strengthened Pt -H* binding under cathodic polarization thermodynamically favours H* formation⁵³, thereby promoting the surface-mediated dissociation pathway and altering the overall electrocatalytic pathway.

Keto-intermediate formation and the dynamic evolution of cyclohexanone

As mentioned in previous sections, increasing cathodic potential favours tautomerization over hydrogenation both kinetically and thermodynamically. Post-tautomerization hydrogenation selectively targets the benzene ring over the carbonyl group, with the former being exothermic (-0.16 eV). The latter being endothermic (0.40 eV), forming ketone intermediates (Fig. 4a). This potential-driven selectivity shift enhances keto-intermediate formation, altering phenol hydrogenation product distributions; hence, it is predictable that once tautomerization of phenol* becomes more favourable than direct hydrogenation, further hydrogenation would lead to saturation on the phenol ring rather than reformation of the enol intermediate. Moreover, at -0.62 V, keto tautomers (2-cyclohexenone and 3-cyclohexenone) are more energetically favourable than the enol form (1,3-cyclohexadienol) (Figs. 4b and S6), indicating irreversible keto formation at this stage; thus, 3-cyclohexenone may be the precursor of cyclohexanone. To verify this hypothesis, it is necessary to investigate its evolution towards cyclohexanone.

Fig. 4. Following hydrogenation on tautomer and relative stability of cyclohexenone tautomers.

a ΔG and ΔG^⧧ of hydrogenation of tautomer* on the carbon atom of the phenyl ring (C-site) and on the oxygen atom of the carbonyl group (O-site) via the LH mechanism. b Relative free energies of 1,3-cyclohexadienol, 2-cyclohexanone, and 3-cyclohexenone averaged over the final 5 ps AIMD trajectory. Energies (eV) are referenced to 1,3-cyclohexadienol. Atom colors: Pt (grey), C (brown), H (white), O (red). Source data for Fig. 4 are provided as a Source Data file.

The dynamic behaviour of cyclohexanone (the product formed upon complete addition of four hydrogen atoms) under electrochemical conditions was investigated to elucidate trends in product distribution. Cyclohexanone adopts two distinct adsorption configurations (Figs. 5a and S7) showing both stable adsorption in the gas phase: (i) a carbonyl-bound planar (side-on) configuration, and (ii) a vertical configuration, with only the oxygen atom adsorbed on the surface (end-on), consistent with prior theoretical studies^41,54. However, AIMD simulations reveal that the end-on configuration spontaneously transitions to a stable ‘floating’ state at the water/Pt interface, characterized by ~1 Å increase in carbonyl oxygen-surface distance (h_c-o) relative to initial adsorption (Fig. 5b). This floating state persists throughout the simulation trajectory. Furthermore, charge-density difference analysis indicates negligible electron transfer between cyclohexanone and the Pt catalyst (Fig. S8). Similarly, the side-on configuration remains thermodynamically stable throughout the AIMD simulation.

Fig. 5. Adsorption configurations and dynamic evolution of cyclohexanone on the Pt(111) surface.

a Adsorption energies of cyclohexanone on Pt (111) and corresponding optimized structure in ‘side-on’ and ‘end-on’ forms. b The evolution of the C-O bond height of cyclohexanone in two different adsorption configurations (upper: end-on adsorption and lower: side-on) from 10 ps of AIMD trajectories with corresponding bond length. Snapshots of ‘end-on’ and ‘floating’ were shown on the right. c Energy profiles of cyclohexanone transiting from ‘floating’ to ‘side-on’ configuration. d C-O bond height distributions for cyclohexanone (‘side-on’ and ‘floating’) and 3-cyclohexenone at -0.62 V (final 5 ps). Atom colors: Pt (grey), C (brown), H (white), O (red), and K (purple). Source data for Fig. 5 are provided as a Source Data file. Source data for Fig. 5 are provided as a Source Data file.

As shown in Fig. 5c, we present the free energy pathway for the transformation between the side-on and floating configurations. The floating configuration is found to be more stable than the side-on one at both 0 V and -0.62 V. At 0 V, the side-on configuration can readily convert to the floating configuration with low energy barriers around 0.1 eV, while the energy barrier increases to 0.3 eV for the floating configuration transferring back to the side-on configuration, suggesting the floating state of cyclohexanone would be stabilized at a more negative potential due to the greater repulsion between the carbonyl group and charged surface. Interestingly, by comparing the adsorption configuration of the stable intermediate (3-cyclohexenone) discussed in the previous section with the two adsorption configurations of cyclohexanone, we observe that the floating state aligns well with the tilted geometry of the phenyl ring in 3-cyclohexenone (Fig. 5d), suggesting that 3-cyclohexenone acts as a direct precursor. Upon saturation, cyclohexanone tends to repel from the surface rather than adopting the side-on configuration required for further hydrogenation.

We then investigate the energy difference between cyclohexanone and its enol tautomer at -0.62 V. We found that 1-cyclohexenol is thermodynamically more stable than cyclohexanone (Table S1), indicating that enol formation is spontaneous, whereas the reverse transformation from 1-cyclohexenol back to cyclohexanone is energetically unfavourable. However, due to repulsion between cyclohexanone and the surface, it would be unfeasible for cyclohexanone to be activated for tautomerization or further hydrogenation to cyclohexanol, leading to cyclohexanone accumulation and a more sluggish hydrogenation rate observed in experiments.

Herein, we propose a reaction pathway for electrocatalytic phenol hydrogenation on the Pt catalyst (Fig. 6), wherein tautomerization activates the benzene ring for saturation, forming 3-cyclohexenone, which exhibits greater thermodynamic stability than its enol tautomer (Table S1) and adopts a surface-floating configuration structurally analogous to cyclohexanone (Fig. 5d), facilitating hydrogenation to cyclohexanone. The latter may undergo tautomerization to form 1-cyclohexenol, which is further hydrogenated to yield cyclohexanol. However, this final tautomerization faces an energetic barrier despite exothermicity⁴¹, primarily due to electrostatic repulsion between cyclohexanone’s carbonyl group and the negatively charged surface at -0.62 V (Fig. 5c). This repulsion kinetically impedes cyclohexanone hydrogenation relative to aromatic ring saturation³⁴. Once formed, surface repulsion forces cyclohexanone into a non-activated floating state (Fig. 5b), necessitating external energy for surface approach and further hydrogenation or tautomerization. Consequently, cyclohexanone accumulates as the dominant hydrogenation product under cathodic potential primarily, and the further hydrogenation of cyclohexanone would be sluggish (either hydrogenation on the carbonyl group or tautomerization into the corresponding enol form) due to the absence of activation from the catalyst surface.

Fig. 6. Proposed reaction mechanism for the electrocatalytic hydrogenation of phenol on Pt (111).

Solid lines denote the dominant tautomerization pathway towards cyclohexanone; dashed lines show the direct hydrogenation pathway towards cyclohexanol.

Discussion

This study establishes that electrode potential serves as a dynamic redox switch that controls the enol-keto equilibrium on Pt(111). At cathodic polarization, surface-mediated electron transfer activates phenol tautomerization both kinetically and thermodynamically. This process is also driven by the optimized Volmer thermodynamics, which favor the formation of phenoxy* to facilitate tautomerization. The applied potential further dictates product selectivity by electrostatically confining cyclohexanone in a metastable floating state, kinetically hindering C = O activation and promoting its accumulation. Coupled with the stabilization of keto intermediates, these potential-dependent effects collectively identify tautomerization as the key step in determining selectivity. By resolving these mechanisms with AIMD and explicit solvation, our work highlights the role of electrode potential in modulating both tautomerization and product distribution, offering a conceptual basis for the design of selective biomass upgrading strategies.

Our findings highlight the necessity of explicitly incorporating both electrode potential and solvent dynamics to describe the electrochemical interface. Nevertheless, a fully realistic description of the electrochemical reaction environment, including reactant coverage, pH effects and a more accurate free energy sampling, remains challenging due to the limitations of current computational models and the associated computational cost. Future developments in machine learning-accelerated molecular dynamics simulations are expected to address these challenges and provide deeper insights into phenol ECH under operando conditions.

Methods

Computational details

The model in this work was constructed using a four-layer 4 × 6 Pt (111) supercell (dimensions: 11.10 × 14.42 × 26.80 ų), with a 20 Å vacuum layer containing 80 explicit water molecules (density ≈1.0 g/cm³, Fig. S9). Static DFT calculations were carried out by Vienna ab initio Simulation Package (VASP)^55,56 employing the Perdew−Burke−Ernzerhof (PBE) exchange-correlation functional within the generalized gradient approximation (GGA)⁵⁷. A plane-wave cutoff energy of 520 eV⁵⁸ and Γ-point Brillouin zone sampling were used. Convergence thresholds were set to 0.05 eV/Å for ionic relaxation and 10⁻⁵eV for electronic self-consistency. Dynamic simulations utilized the CP2K/Quickstep package⁵⁹ under the canonical ensemble (NVT) conditions with Nosé−Hoover thermostats at 300 K and a 1.0 fs timestep^60,61. The mixed double-ζ Gaussian and plane-wave (GPW) basis sets with 400 Ry cutoff were applied using Goedecker−Teter−Hutter (GTH) pseudopotentials for core electrons (valence electrons: 18, 4, 6, 1, and 9 for Pt, C, O, H, and K, respectively). Grimme’s DFT-D3 correction accounted for noncovalent interactions in all calculations^62,63.

All electrode potentials mentioned in this study are referred to the reversible hydrogen electrode (RHE), the electrode potentials (Φ) were regulated via introducing K⁺ as counterions and calculated as⁶⁴:

$$\Phi =\frac{\sigma }{C_{\text{dl}}}+{\Phi }_{\text{PZC}}$$ 1

Where σ is the surface charge density obtained by Bader charge analysis⁶⁵, C_dl corresponding to 24 μF/cm² (Pt (111) double-layer capacitance), and Φ_PZC is the potential of zero charge (0.37 V vs. RHE)⁶⁶. The radial distribution functions (RDFs) and the coordination number analysis of O–O and O–H for the liquid water structure are provided in Fig. S10. The distribution of solvent water in our systems and the distribution of K⁺ in counterions added system are shown in Figs. S11 and S12.

Reaction free energies and barriers were computed using constrained molecular dynamics (cMD) coupled with thermodynamic integration (TI). Slow-growth simulations were employed to obtain initial structures for TI calculations. A holonomic constraint was applied to the reaction coordinate (ζ)⁶⁷, and the free energy $\Delta A\left(\zeta \right)$ was obtained by integrating the average force along ζ:

$$\Delta A\left({\zeta }_a,{\zeta }_b\right)=-{\int }_{{\zeta }_a}^{{\zeta }_b}F\left(\zeta \right)d\zeta$$ 2

where $\Delta A\left({\zeta }_a,{\zeta }_b\right)$ denotes the free-energy difference between two reaction coordinates (${\zeta }_{a}and{\zeta }_b$), and $F\left(\zeta \right)$ refers to the ensemble-averaged constrained force. Detailed definitions of collective variables (CVs) for all elementary steps are listed in Table S2; the procedure for error estimation and the data of free energy gradient (Figs. S14-S16) are provided in Supplementary Information.

Author contributions

Z.Y. and Y.-G.W. supervised the project and designed the idea. Q.-Y. L. performed the density functional theory (DFT) calculations and ab initio molecular dynamics (AIMD) simulations, analyzed the data, and drafted the manuscript. C.-H. J. contributed to data visualisation and assisted in the AIMD trajectory analysis.

Peer review

Peer review information

Nature Communications thanks the anonymous reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

This work was financially supported by the National Natural Science Foundation of China (NSFC, No. 22022504, 220330005, and 22250710677 to Y.-G.W.), the NSFC Center for Single-Atom Catalysis (22388102), the Guangdong “Pearl River” Talent Plan (No. 2019QN01L353 to Y.-G.W.), and the Science, Technology and Innovation Commission of Shenzhen Municipality (No. JCYJ20210324103608023, KCXST20221021111207017, and JCYJ20220818100410023 to Y.-G.W.).

Data availability

Source Data are provided with this paper. All computational structural data are available at (https://github.com/Yanggang-Wang-Group/LQY_2026_Phenol_Pt) and have been made citable via Zenodo (10.5281/zenodo.20728980)⁶⁸. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41467-026-75401-1.