Steering carbon dioxide reduction toward C-C coupling using copper electrodes modified with porous molecular films

Electrochemical CO2 reduction (CO2RR) offers a carbon‑neutral route to produce chemical feedstocks and store renewable electricity. Copper is uniquely capable of forming hydrocarbons and oxygenates from CO2, yielding both single-carbon (C1) products (formate, CO, methane) and multicarbon (C2+) products (acetate, ethylene, ethanol, isopropanol). The higher energy density and applicability of C2+ products make steering selectivity from C1 to C2+ highly desirable, while also suppressing the competing hydrogen evolution reaction (HER). Prior strategies include tuning Cu morphology, oxidation state, alloying/dopants, and electrolyte composition. A complementary approach is to tailor the catalyst microenvironment with molecular or polymeric films on Cu. This study investigates how the physical structure—specifically thickness and porosity—of a bipyridine-based organic film on Cu affects CO2RR selectivity and intrinsic activity, with the working hypothesis that porous, thicker films retain in situ–generated CO, elevate local CO partial pressure and surface coverage, and thereby enhance C–C coupling toward C2+ products.

Previous reports demonstrate that organic modifiers on Cu can steer CO2RR via altered electric fields, intermediate stabilization, hydrophobicity, and increased local CO2 concentration. For example, N‑arylpyridinium modifiers correlated ethylene selectivity with atop vs bridge‑bound CO on Cu, while hydrophobic polymers altered H2O/CO2 transport to affect activity and selectivity. A tricomponent copolymer on Cu achieved state‑of‑the‑art performance, attributed to both polymer‑induced electric fields and increased film porosity facilitating CO2 diffusion. Beyond organic modification, Cu morphology (size, grain boundaries, facets), oxidation state, alloying, and electrolyte cations/anions also impact CO2RR pathways. Collectively, these studies motivate systematic probing of film thickness and porosity as physical parameters to control mass transport of CO2/CO/H+, local coverages, and ultimately C–C coupling.

Electrode preparation: Polycrystalline Cu foils (2.5 × 2.5 cm2, 99.99%) were mechanically polished and electrochemically polished (85 wt% H3PO4, 2.1 V vs graphite, 5 min), rinsed, and dried. T‑bipyridine films were formed by electroreductive dimerization of 10 mM 1‑(4‑tolyl)pyridinium triflate (T‑Pyr) in CO2‑saturated 0.1 M KHCO3 (pH 6.8), using deposition potentials between −1.25 and −0.55 V vs RHE to tune film thickness and porosity. Five modified electrodes (Cu‑1 to Cu‑5) were prepared under specified potentials (Table S1). A thin‑film variant (Cu‑5thin) used 300 s deposition; a reconstructed, nonporous film (Cu‑5block) was produced by dissolving Cu‑5’s film with acetone and allowing it to re‑form upon solvent evaporation.

Film characterization: AFM measured film thickness (d) via step edges produced by masking during deposition; d increased from 0.10 ± 0.01 to 0.79 ± 0.08 µm as deposition potential was made more negative. SEM assessed morphology; films became rougher and more porous at more negative potentials. Porosity (P) was quantified by an acetone reconstruction method, assuming the rebuilt film is dense and uniformly distributed; P = (d − drecon)/d × 100%. P rose from 55% (Cu‑1) to 83% (Cu‑5). XPS (Cu 2p, Cu LMM) verified no significant change in Cu valence relative to pristine Cu; Cu0/Cu+ species consistent with native Cu2O were present on both. AFM roughness after film removal showed no nanostructuring. Water contact angles assessed hydrophobicity (pristine Cu ≈ 22°, modified Cu‑n ≈ 75–115°).

Electrochemistry and product analysis: CO2RR was evaluated in a PEEK flow cell with a Sustainion membrane separating chambers; Cu or T‑bipyridine/Cu as working electrode (exposed area 1.22 cm2), Pt counter, leak‑free Ag/AgCl reference. Electrolyte: CO2‑ or Ar‑saturated 0.1 M KHCO3. iR‑corrected potentials converted to RHE. Bulk electrolysis was performed at −0.96 V vs RHE for 60 min with gas flow (2.3 mL min−1). Gaseous products (H2, CO, CH4, C2H4) were quantified by GC every 15 min; liquid products by 1H NMR with DMSO internal standard. Faradaic efficiencies (FEs) were averaged over four injections. ECSA and roughness factor (RF) were determined via double‑layer capacitance from non‑Faradaic CVs (−0.05 to 0.05 V vs RHE; 20–120 mV s−1); specific capacitance of electropolished Cu: 29 µF cm−2. Partial current densities were normalized to geometric area and ECSA. HER under Ar was measured as a control. Methyl viologen redox probe CVs evaluated film transport characteristics.

Operando Raman: A three‑electrode cell with quartz window was used under CO2‑saturated 0.1 M KHCO3 at −0.9 V vs RHE. Spectra were collected (785 nm laser) to monitor surface species (Cu–CO modes and CO stretching) on pristine Cu and Cu‑5.

Microkinetic modeling: A simplified Langmuir adsorption‑based microkinetic framework described competition between hydrogen (H) and carbon (C, representing CO‑derived species) at the surface. Surface coverages follow Langmuir expressions with partial pressures PH and PC. Rates for H2, C1, C2, C3 formation were expressed as power‑law functions of PH, PC, and coverages. The model probes qualitative trends in FE as a function of PC, serving as a descriptor for local carbon reactant availability within the porous film.

• Film growth and porosity: Film thickness increased from 0.10 ± 0.01 to 0.79 ± 0.08 µm as deposition potential was made more negative (−0.55 to −1.25 V vs RHE). Porosity P increased from 55% (Cu‑1) to 83% (Cu‑5). • Selectivity shift: At −0.96 V vs RHE (60 min, CO2‑saturated 0.1 M KHCO3), pristine Cu favored H2 (FEH2 = 52.8%) and C1 products (FECH4 = 11.8%, FECO = 9.4%) with minor C2+ (FEC2H4 = 6.1%, FEC2H5OH = 1.4%). Cu‑5 (d = 0.79 µm, P ≈ 83%) achieved FEC2H4 = 46.1% with low CH4 (FECH4 = 2.6%) and no detectable CO; FEH2 = 24.7%. Total C2+ selectivity increased from 9.7% (pristine Cu) to 61.9% (Cu‑5). • Intrinsic activity: ECSA‑normalized ethylene partial current density increased 8.7×: JECSA,C2H4 = 0.52 mA cm−2 (Cu‑5) vs 0.06 mA cm−2 (pristine). JECSA,C2+ reached 0.68 mA cm−2 (6.8× pristine). The ratio JECSA,C2H4/JECSA,CH4 rose from 0.4 (Cu‑1) to 18.2 (Cu‑5). Total JECSA remained largely unchanged across films, indicating selectivity rather than total activity changes. • HER behavior: Under CO2, JECSA,H2 decreased from 0.51 mA cm−2 (pristine) to 0.28 mA cm−2 (Cu‑5), implying suppression of HER coincident with enhanced CO coverage. Under Ar (no CO2RR), JECSA,H2 was 0.74–1.07 mA cm−2 and insensitive to the film, indicating unaffected proton transport in the absence of CO2. • Thickness vs porosity controls: Reducing thickness while retaining a porous morphology (Cu‑5thin, d = 0.15 ± 0.03 µm) decreased FEC2H4 from 46.1% to 26.3% and total FEC2+ from 61.9% to 34.7%, with FECH4 increasing to 5.5%. Rendering the film lamellar and nonporous (Cu‑5block, d = 0.13 ± 0.02 µm) drastically reduced FEC2H4 to 7.8% and FEC2+ to <10%, with JECSA,C2H4 5.7× smaller than Cu‑5, underscoring the critical role of porosity. • Surface state and morphology: XPS and operando Raman indicated metallic Cu active sites (Cu0/Cu+ from native oxide reduction) unchanged by film; no evidence of halide‑induced Cu nanostructuring. AFM/SEM showed increased film roughness/porosity but no Cu nanostructure. Contact angles increased (75–115°) on modified surfaces; however, hydrophobicity alone did not account for performance gains. • Spectroscopy: Operando Raman at −0.9 V vs RHE showed stronger Cu–CO modes (≈280, 366 cm−1) and C=O stretching (≈1950–2100 cm−1) on Cu‑5 vs pristine, indicating higher CO surface coverage without changes in Cu–CO bond strength. • Transport probing: Methyl viologen CVs on Cu‑5 exhibited S‑shaped waves suggestive of diffusion through a porous network resembling an ultramicroelectrode array, consistent with selective permeability that still allows efficient CO2/H2O access. • Modeling: A microkinetic model using PC (local carbon reactant partial pressure) as a descriptor reproduced experimental trends: increasing PC increases FEC2 and FEC3 while decreasing FEH2; FE C1 exhibits a volcano‑type trend. The qualitative agreement supports the hypothesis that porous, thick films elevate local CO partial pressure and coverage to promote C–C coupling.

The results support the central hypothesis that tailoring the physical structure of molecular films on Cu—specifically increasing thickness and porosity—modulates the local reaction microenvironment to favor multicarbon formation. Porous, thicker T‑bipyridine films retain in situ–generated CO, increasing its residence time and effective local partial pressure, which elevates CO surface coverage. This enhances C–C coupling rates and redirects selectivity from C1 (e.g., CH4) toward C2+ (e.g., C2H4, ethanol, acetate), while modestly suppressing HER under CO2. The nearly constant total ECSA‑normalized current across samples shows that the modifier does not increase overall activity but reshapes product distribution by redistributing active‑site utilization. Operando Raman confirms higher CO coverage on modified electrodes without altering Cu–CO bond strength, and XPS/AFM/SEM rule out changes in Cu oxidation state or nanostructuring as the cause. Transport studies and contact angle measurements indicate that while hydrophobicity increases, mass transport through the porous network remains sufficient for CO2 and H2O, and hydrophobicity alone cannot explain the trends. The microkinetic model, using local carbon reactant partial pressure as a descriptor, captures the experimentally observed evolution of FE with film structure, reinforcing the mechanistic picture.

Engineering the thickness and porosity of bipyridine‑based molecular films on polycrystalline Cu provides a viable route to steer CO2RR selectivity toward C2+ products without increasing total intrinsic activity. A thick, highly porous T‑bipyridine film achieves up to 46.1% FE for C2H4 and 61.9% total C2+ selectivity, with an 8.7‑fold increase in intrinsic ethylene activity relative to bare Cu. Spectroscopic and modeling evidence indicates that such films create CO‑rich microenvironments, elevating local CO partial pressure and surface coverage to promote C–C coupling, while leaving Cu’s chemical state unchanged. These insights highlight microenvironmental control via molecular film architecture as a powerful design lever for CO2‑to‑C2+ conversion. Future work should address stabilizing even thicker porous films, quantifying the relationship between film structure and local CO partial pressure, and exploring electrolyte composition (e.g., cations) and alternative molecular frameworks to further optimize selectivity and durability.

• Stability of physically adsorbed films limits exploration of thicker films; portions detached during deposition/electrolysis at more negative potentials. • The microkinetic model is qualitative; a direct quantitative relationship between film thickness/porosity and local CO partial pressure was not established. • Electrolyte effects (notably cation identity/concentration) were not systematically varied, though known to influence CO2RR. • ECSA estimation via double‑layer capacitance assumes a constant specific capacitance; Pb UPD could not be applied due to interference from the film. • Long‑term durability and performance under industrially relevant current densities and gas‑diffusion configurations were not evaluated.