Mechanism of Electrocatalytic H2 Evolution, Carbonyl Hydrogenation, and Carbon–Carbon Coupling on Cu

Aqueous-phase electrocatalytic hydrogenation of benzaldehyde on Cu leads not only to benzyl alcohol (the carbonyl hydrogenation product), but Cu also catalyzes carbon–carbon coupling to hydrobenzoin. In the absence of an organic substrate, H2 evolution proceeds via the Volmer–Tafel mechanism on Cu/C, with the Tafel step being rate-determining. In the presence of benzaldehyde, the catalyst surface is primarily covered with the organic substrate, while H* coverage is low. Mechanistically, the first H addition to the carbonyl O of an adsorbed benzaldehyde molecule leads to a surface-bound hydroxy intermediate. The hydroxy intermediate then undergoes a second and rate-determining H addition to its α-C to form benzyl alcohol. The H additions occur predominantly via the proton-coupled electron transfer mechanism. In a parallel reaction, the radical α-C of the hydroxy intermediate attacks the electrophilic carbonyl C of a physisorbed benzaldehyde molecule to form the C–C bond, which is rate-determining. The C–C coupling is accompanied by the protonation of the formed alkoxy radical intermediate, coupled with electron transfer from the surface of Cu, to form hydrobenzoin.


S1. Additional Catalyst Characterization
The N2 adsorption-desorption isotherms were determined at 77 K on a PMI automated BET sorptometer.Prior to the measurements, the sample was outgassed at 523 K for 2 h.The surface area of the catalysts was estimated to be equal to ~223 m 2 •gcat -1 by the Brunauer-Emmett-Teller (BET) method.This parameter was fixed during the fit.
The Cu K-edge X-ray absorption spectroscopy (XAS) measurements were performed on an easyXAFS300+ spectrometer system equipped with a ProtoXRD XRT60 X-ray tube and AXAS-M silicon drift detector (SDD) from KETEK GmbH.Monochromatic X-rays were obtained using a Si553 spherically bent crystal analyzer.The electron consumption (and the corresponding H + consumption) towards H2 evolution (  2 ) was calculated using the following relation.

S3. Additional Calculation Details
The conversion of a reactant  at a given time  was estimated by: where   () is the amount of reactant detected at time  and   0 is the initial amount of reactant in the electrolyte.
Similarly, the yield of a product  with respect to a reactant  at any given time  was estimated by: where   () is the amount of product  detected at time  and   0 is the initial amount of reactant  in the electrolyte solution.
The Faradaic efficiency towards organic conversion was estimated by: where   is the electrons consumed towards the conversion of the organic substrate and   is the total electrons consumed.
The Faradaic selectivity towards the formation of a product  was estimated by: where   is the electrons consumed towards formation of product , and   is the total electrons consumed.

S4. Additional Computational Details
][3][4] The single-electron wavefunctions were estimated using the employed basis set truncated at an energy cut-off of 400 eV.The k-points for these wavefunction calculations were sampled from a grid size 2 × 2 × 1 in the Brillouin zone.The partial occupancies were set for each orbital using the Methfessel-Paxton scheme to smoothen the wavefunctions. 5Using projector augmented-wave (PAW) method, the core electrons were treated with the frozen-core approximations, while the valence electron wavefunctions were calculated using the plane wave basis set. 6urther, to determine electron density from these wavefunctions, the electron self-interaction contribution in the Hartree potential was corrected using generalized gradient approximation (GGA) Perdew-Burke-Ernzerhof (PBE) exchange correlation functional with additional dispersion corrections including using the D3 method with the Becke-Johnson (BJ) damping scheme. 7,8 he open-source VASPsol package was used to implement implicit solvent in all ab initio simulations with spin-polarized calculations. 9The Bader charge analysis was performed using the code available from Henkelman and co-workers. 10l simulations were performed on a 4 × 4 × 4 supercell of the Cu(111) surface, cleaved from bulk Cu with lattice constant of 3.59 Å, and a vacuum of at least 10 Å above the water layer.The top two layers of the Cu(111) slab, representing the surface atoms were not constrained during simulations, while the bottom two layers were fixed representing bulk Cu.The electrodeelectrolyte interface was modelled using 15 explicit H2O molecules in addition to one proton and reaction intermediates above the Cu(111) surface.The system was first relaxed using ab initio molecular dynamics (AIMD) simulations.The AIMD simulations were performed using a canonical ensemble at constant temperature (300 K) and volume (NVT).To maintain constant temperature, a Nosé-Hoover thermostat was employed with a time constant of 0.01 ps.AIMD simulation were carried out for at least 5 ps with a time step of 0.5 fs while maintaining the energy and force convergence at 1 × 10 -5 eV and 5 × 10 -2 eV•Å -1 , respectively.The selected configurations from the AIMD simulations were further optimized using density functional theory (DFT).Three different reaction pathways were investigated: first H addition, second H addition, and C-C coupling.The H addition via PCET was modelled by including an additional H to the system that provided an H + to the water layer near the interface and an e -to the metal surface, while the Langmuir-Hinshelwood type surface hydrogenation was modeled using H atoms (H*) adsorbed on Cu.The transition states (TS) for each step were identified using the climbing image-nudged elastic band technique (cl-NEB). 11For this, 8 -16 frames were developed between the initial state (IS) and the final state (FS) to observe a minimum energy path along the reaction coordinate.
We note that the work function (), or the surface potential (), of the metal surface varies between the IS, TS, and FS, especially in the PCET steps.The obtained potentials were, therefore, corrected using the method developed by Chan and Nørskov. 12Based on this method, the difference in the energy between state 1 and state 2 at a given potential can be estimated by:  2 ( 1 ) −  1 ( 1 ) =  2 ( 2 ) −  1 ( 1 ) + ( 2 −  1 )( 2 −  1 ) 2 where   is the work function at state  and   is the charge on the surface (including the adsorbate) at state .For this, the charge was estimated using Bader charge analysis.
Furthermore, the work function (  ) at state  can be related to the surface potential of the Cu(111) surface using the following relation:

Figure S2 .
Figure S2.(a) X-ray absorption near edge structure (XANES) of Cu/C.The XANES of Cu reference foil is also shown for comparison.(b) Fourier-transformed extended X-ray absorption fine structure (FT-EXAFS) of Cu/C.Experimental data are shown as closed symbols and the corresponding fits are shown as solid lines.

Figure S6 .
Figure S6.Yield versus conversion (on a carbon basis) plots during BZ ECH on Cu/C.Reaction conditions: 20 mM BZ,  = -0.5 V vs RHE, 1.5 M acetate buffer solution (pH ~ 4.6), room temperature, ambient pressure.The dashed lines are linear fits.

Figure S7 .
Figure S7.Concentration profiles of reactants and products during BZ ECH on the carbon black support.Reaction conditions: 20 mM BZ,  = -0.5 V vs RHE, 1.5 M acetate buffer solution (pH ~ 4.6), room temperature, ambient pressure.The dashed lines are guides to the eye.The reported number is the Faradaic efficiency (FE) towards BZ conversion.

Figure S9 .
Figure S9.BZ conversion as a function of reaction time during BZ ECH with or without tbutanol (t-BuOH) on Cu/C with initial BZ concentration in the electrolyte equal to (a) 5 mM, (b) 10 mM, (c) 20 mM, and (d) 40 mM.Reaction conditions: 5 -40 mM BZ, 0 mM or 200 mM t-BuOH,  = -0.5 V vs RHE, 1.5 M acetate buffer solution (pH ~ 4.6), room temperature, ambient pressure.The dashed lines are linear fits, and the reported numbers are the initial BZ conversion rates in mmolBZ•gCu -1 •s -1 .

Figure S10 .
Figure S10.H2 evolution as a function of reaction time during HER with or without t-BuOH on Cu/C.Reaction conditions: 0 mM or 200 mM t-BuOH,  = -0.5 V vs RHE, 1.5 M acetate buffer solution (pH ~ 4.6), room temperature, ambient pressure.The dashed lines are linear fits, and the reported numbers are the initial HER rates in mmolH₂•gCu -1 •s -1 .

Figure S12 .
Figure S12.Initial (IS), transition (TS ‡ ), and final (FS) states of first H addition to an adsorbed BZ molecule via the Langmuir-Hinshelwood-type surface reaction between BZ* and H* on the Cu(111) surface.The reported number are interatomic distances in Å. Cu: grey, C: orange, O: red, H: white/purple.H2O molecules have been removed for clarity.

Figure S15 .
Figure S15.Electron consumption (in A•s) during BZ ECH on Cu/C.Reaction conditions: 20 mM BZ,  = -0.5 V vs RHE on Cu/C, 1.5 M acetate buffer solution (pH ~ 4.6), room temperature, ambient pressure.The dashed lines are guides to the eye.
, =   −    where  , is the surface potential of Cu(111) relative to SHE at state , and   and   are the work functions of Cu(111) at state  and that of SHE (equal to 4.44 eV), respectively.Finally, the calculated surface potential is reported relative to the RHE ( , ), as per the following equation:  , =  , + 0.0591 •