Electro-activated indigos intensify ampere-level CO₂ reduction to CO on silver catalysts

Abstract

The electrochemical reduction of carbon dioxide (CO₂) to carbon monoxide (CO) is challenged by a selectivity decline at high current densities. Here we report a class of indigo-based molecular promoters with redox-active CO₂ binding sites to enhance the high-rate conversion of CO₂ to CO on silver (Ag) catalysts. Theoretical calculations and in situ spectroscopy analyses demonstrate that the synergistic effect at the interface of indigo-derived compounds and Ag nanoparticles could activate CO₂ molecules and accelerate the formation of key intermediates (*CO₂^– and *COOH) in the CO pathway. Indigo derivatives with electron-withdrawing groups further reduce the overpotential for CO production upon optimizing the interfacial CO₂ binding affinity. By integrating the molecular design of redox-active centres with the defect engineering of Ag structures, we achieve a Faradaic efficiency for CO exceeding 90% across a current density range of 0.10 − 1.20 A cm^–2. The Ag mass activity toward CO increases to 174 A mg^–1_Ag. This work showcases that employing redox-active CO₂ sorbents as surface modification agents is a highly effective strategy to intensify the reactivity of electrochemical CO₂ reduction.

Introduction

The electrocatalytic carbon dioxide reduction reaction (CO₂RR) powered by renewable electricity offers an attractive carbon-negative route for producing industrial chemicals and fuels¹. Among the various products of CO₂RR, carbon monoxide (CO) stands out as the most competitive target when considering the market value and cost efficiency^2,3,4. To render CO₂ electrolysis profitable compared to conventional petrochemical processes, techno-economic analyses indicate that the Faradaic efficiency (FE) for CO should exceed 90% at current densities above 500 mA cm^–2 (refs ^3,4). However, achieving high-rate, high-selectivity CO production in CO₂RR remains challenging, as near-unity selectivity is only observed at limited partial current densities². This limitation arises from the large overpotential required to initiate CO₂ conversion, coupled with the competing hydrogen evolution reaction (HER)².

The linear CO₂ molecule features a robust C=O bond of 1.163 Å in length and 750 kJ mol^–1 in dissociation energy, making it thermodynamically stable and difficult to activate⁵. The two-electron, two-proton conversion of CO₂ to CO typically involves the adsorption and activation of CO₂ to form the *CO₂^– intermediate, which is subsequently hydrogenated to produce the carbon-bound *COOH species, followed by further reduction and desorption steps to yield CO (refs. ^1,2). The initial activation of CO₂ is demonstrated to be the possible rate-determining step, especially for silver (Ag) catalysts, due to their weak interaction with CO₂ and the sluggish kinetics toward *CO₂^– and *COOH formation^1,6,7. Therefore, ample efforts have been devoted to promoting CO₂ stabilization and activation on Ag-based catalysts through approaches such as nanostructuring, doping, and alloying^2,8.

Beyond solely tuning metal catalysts, surface modification with organic compounds has emerged as a promising alternative strategy^9,10. This strategy leverages synergistic effects at the interface between organic molecules and heterogeneous active sites, which regulates the interfacial reactant concentrations, intermediates adsorption, and reaction energy barriers¹⁰. In this context, compounds exhibiting reactive interactions with CO₂ have been explored for modifying Ag surfaces in CO₂-to-CO conversion^11,12. For instance, amine-confined Ag has been developed to enrich local CO₂ concentration during oxygen-containing flue gas electro-reduction¹³, and grafting pyridine groups onto Ag electrodes has been shown to induce a ten-fold activity enhancement toward CO by stabilizing the *COOH intermediate¹⁴. Similar improvement has also been observed with Ag catalysts upon the addition of CO₂-binding imidazolium ionic liquids in nonaqueous electrolytes¹⁵. These studies demonstrate the potential of using organic molecules to modulate the reactivity profiles of heterogeneous CO₂RR. However, a comprehensive understanding of the structure-function relationships in molecularly enhanced catalysts remains elusive, as precise synthetic controls over their interactions with CO₂ and/or relevant intermediates are still lacking¹⁶.

In our recent study, we discovered indigo (Id), a redox-active Lewis base, as an effective CO₂ sorbent for electrochemically mediated carbon capture (Fig. 1a)¹⁷. Specifically, the carbonyl groups in Id can undergo electro-reduction to form a nucleophilic dianion (Id^2–), which then reacts with electrophilic CO₂ to afford a carbonate adduct (Id-2CO₂^2–)^17,18,19,20. Inspired by the ability of redox-active Id to capture CO₂ via carbonate formation, we seek to explore the possibility of employing indigo-based compounds as molecular promoters to accelerate CO₂-to-CO conversion in conjunction with Ag catalysts. We envisage that the unique binding interaction between electrochemically reduced Id and CO₂ could facilitate the activation of linear CO₂ molecule into a bent conformation, consequently stabilizing key reaction intermediates to lower the energy barriers for protonation steps⁶. Importantly, molecular engineering allows us to finetune the chemical properties of the CO₂-binding sites, resulting in more favorable intermediate adsorption behaviors¹⁸.

Fig. 1: CO₂RR performance and mechanistic insight of AgNP+Id hybrid.

a Reaction mechanism of redox-active Id for CO₂ capture. b CV of Id under N₂ and CO₂ atmospheres. The experiments were conducted by dissolving 2.5 mM Id in dimethyl sulfoxide (DMSO) with 0.1 M tetrabutylammonium hexafluorophosphate (TBAPF₆) at a scan rate of −20 mV s^–1. Ferrocene/ferrocenium (Fc^0/+) was used as an internal reference. Potentials are non-iR corrected. c DFT-optimized structure of Id-2CO₂^2– adduct showing the bent CO₂ configuration at the redox-active oxygen center. The corresponding bond angles and lengths are listed next to the structure. The black, red, blue, and white spheres represent C, O, N, and H atoms, respectively. d, e Comparison of CO FE as a function of j_total (d), and j_CO as a function of potential (e) for AgNP and AgNP+Id. Potentials have been 100% iR corrected. The flow cell was operated with 1 M KOH (pH = 14.0 ± 0.2). Data are presented as mean ± s.d. Error bars represent s.d. from measurements of three independent electrodes. f Adsorption configuration of the *CO₂^– intermediate at the Ag/Id interface and comparison of *CO₂^– adsorption energies for AgNP and AgNP+Id. The black, red, blue, white, and purple spheres represent C, O, N, H, and Ag atoms, respectively. g Potential-dependent ATR-SEIRAS contour map for AgNP and AgNP/Id. ΔR/R₀  =  (R  −  R₀)/R₀, where R and R₀ are spectra collected at the sample potential and the open circuit potential, respectively. Source data are provided as a Source Data file.

Herein, through a combination of precise molecular design, in situ spectroscopy characterizations, and theoretical calculations, we systematically investigate a series of indigo derivatives with varying CO₂ affinities to unravel the interplay between electro-activated CO₂ sorbents and the catalytic performance at the Ag/organic interface. Moreover, by further immobilizing CO₂-binding moieties into a macromolecular structure and interfacing them with defect-rich Ag particles on a carbon support, the resulting optimized hybrid catalyst markedly improves CO₂RR reactivity and Ag mass activity toward CO. In particular, steady CO FEs of over 90% can be reached at ampere-level current densities up to nearly 1.2 A cm^–2, accompanied by a notable Ag mass activity of 174 A mg^–1_Ag toward CO production.

Results

Probing the promotional effect of indigo on Ag catalyst for CO₂RR toward CO

To confirm the reactivity of reduced Id toward CO₂, we conducted cyclic voltammetry (CV) studies of Id in an aprotic electrolyte under nitrogen (N₂) and CO₂ atmospheres. Analogous to other redox-active CO₂ sorbents, the positive shift in the reduction onset potential and the merging of the two redox waves in the presence of CO₂ strongly imply a chemical interaction between reduced Id and CO₂ (Fig. 1b)^17,18,19,20. The formation of the Id-2CO₂^2– adduct was also validated in our precious study¹⁷. Similarly, an anodic shift in redox potential under CO₂ was observed in an aqueous electrolyte (Supplementary Fig. 1), suggesting that Id^2– is also capable of binding with CO₂ under typical CO₂RR conditions^21,22. Density functional theory (DFT) calculation further supports that the complexation with Id^2– induces considerable restructuring of CO₂ molecules (Fig. 1c and Supplementary data 1). The CO₂ adduct adopts a bent configuration with a bond angle of 132.4° (O–C–O), and the two C–O bonds (bonds a⁽′⁾and b⁽′⁾) are elongated to 1.236–1.261 Å. The bent CO₂ and the redox-active oxygen center form a C–O bond (bonds c⁽′⁾) with a length of ~ 1.486 Å. Given that the onset potential for CO₂ capture by Id (0.25 V vs. the reversible hydrogen electrode (RHE), Supplementary Fig. 1) is more positive than the equilibrium potential for CO₂RR-to-CO conversion (–0.10 V vs. RHE)⁷, we hypothesize that the introduction of CO₂-binding Id to Ag catalysts could effectively active CO₂ by weakening its C=O bond strength during CO₂RR, thereby facilitating the subsequent reductive transformation.

To test our hypothesis, we ultrasonically mixed Id molecules with commercial Ag nanoparticles (AgNP, 20–40 nm in diameter) in isopropanol and spray-coated the resulting ink onto a carbon gas diffusion layer (GDL) to obtain the Id-modified Ag electrode (AgNP+Id). Scanning electron microscopy (SEM) images reveal that Id crystals are interspersed among agglomerated AgNP, and energy-dispersive X-ray spectroscopy (EDS) elemental mapping confirms the uniform distribution of both components across the electrode (Supplementary Fig. 2). The successful modification of the AgNP electrode with Id is also verified by X-ray photoelectron spectroscopy (XPS), which shows strong N 1s signals (Supplementary Fig. 3a). The Ag 3d XPS spectra indicate the chemical state of metallic Ag⁰ for both bare AgNP and AgNP+Id (Supplementary Fig. 3b). These results suggest that Id modification by physical mixing has minimal impact on the morphology or electronic structure of AgNP.

The CO₂RR performance was accessed in a flow cell using 1 M potassium hydroxide (KOH) aqueous electrolyte. Compared to bare AgNP, the AgNP+Id electrode promotes CO selectivity by suppressing the competing HER to below 5%, especially at higher current densities (Supplementary Figs. 4 and 5). Specifically, AgNP+Id achieves a CO FE of 92% at a total current density (j_total) of 283 mA cm^–2, whereas CO FEs for AgNP rapidly drop to below 87% when j_total exceeds 173 mA cm^–2 (Fig. 1d). AgNP+Id also exhibits enhanced CO activity, as manifested by a ~ 70 mV decrease in overpotential at a j_total of 30 mA cm^–2 and a ~ 2.7-fold increase in CO partial current density ( j_CO) at –0.57 V vs. RHE, compared to AgNP under the same conditions (Fig. 1e). Control experiments without AgNP indicate that improved reactivity upon the addition of Id shall be attributed to a synergistic effect between Id and AgNP since the Id-only sample predominantly generates hydrogen (H₂) with negligible CO production (Supplementary Fig. 6). The incorporation of Id molecules slightly decreases the electrochemically active surface area (ECSA), likely due to the blocking of the Ag surface by organic moieties (Supplementary Fig. 7 and Supplementary Table 1)²³. Therefore, AgNP+Id delivers an even higher ECSA-normalized j_CO compared to bare AgNP (Supplementary Fig. 7d), excluding surface roughness as the contributor to the improved CO₂RR performance.

We conducted DFT calculations to gain a deeper understanding of the cooperativity between Id and Ag in CO₂ activation (Supplementary data 1). The results demonstrate that the configuration of adsorbed CO₂ at the Ag/Id interface resembles that of the Id-2CO₂^2– adduct (Fig. 1f). The bent CO₂ molecule is stabilized through the redox-active oxygen center of reduced Id, with a *CO₂^– adsorption energy of –0.635 eV. By contrast, *CO₂^– exhibits much weaker adsorption energy of merely –0.211 eV on a pristine Ag surface, which is insufficient to initiate the CO₂ reduction pathway²⁴. These preliminary findings indicate that the electro-activated Id moiety is a promising promoter to collectively stabilize CO₂ near the Ag surface, which is pivotal for boosting CO₂-to-CO conversion²⁴.

Building on the insights from DFT calculations, we then employed in situ attenuated total reflection surface-enhanced infrared absorbance spectroscopy (ATR-SEIRAS) to identify intermediate adsorption features during CO₂RR. Potential-dependent spectra were collected in CO₂-saturated D₂O with 0.1 M potassium phosphate as the electrolyte to minimize interference from adsorbed H₂O and (bi)carbonate ions (Fig. 1g and Supplementary Fig. 8). Both the AgNP and AgNP+Id samples show strong bands at ~ 1550 cm^–1, ascribed to the adsorbed CO₃^2– species^25,26. The stronger intensity of this band for AgNP+Id could originate from the faster local proton consumption during CO₂RR. Bare AgNP displays a very weak band at ~ 1640 cm^–1 (Fig. 1g), which corresponds to the C=O stretching of the *COOD intermediate (ν(*COOD))²⁷. The redshift of this band relative to the reported wavenumber for *COOH might be due to the deuteration effect^25,28. By contrast, we observe intensified ν(*COOD) bands emerging at lower overpotentials for AgNP+Id, implying accelerated kinetics of *COOD formation.

Importantly, additional bands arise exclusively for the Id-modified sample (Fig. 1g). We assign the band at ~1698 cm^–1 to the carbonate C=O stretching of the Id-2CO₂^2– adduct (ν(C=O)), which closely matches the DFT-simulated vibration of 1675 cm^–1 at the catalytic interface (Supplementary Fig. 9). The band intensity peaks at around –0.5 V vs. RHE and gradually decreases as the potential becomes more negative. Such a phenomenon reflects the progressive accumulation and conversion of activated *CO₂^– species. In addition, the band at ~ 1590 cm^–1 can likely be attributed to the asymmetric O−C−O stretching of *CO₂^– evoked by the resonance structure of Id-2CO₂^2– (ν_as(O−C−O)), which merges with the CO₃^2– band at more cathodic potentials. It is noteworthy that the ν_as(O−C−O) band has been reported as weak or even undetectable on bare Ag, especially at low overpotentials²⁷. Our observation of a strong ν_as(O−C−O) band implies the existence of an unconventional adsorption configuration of *CO₂^– at the Ag/Id interface, such as partial Ag−O coordination^28,29, as also suggested by our DFT calculations (Fig. 1f). We also observe broader bands in the range of 1460–1530 cm^–1, though their precise assignments remain debatable. In some reports, these bands are associated with adsorbed CO₂ and/or protonated carbonate species^25,26. Collectively, the in situ ATR-SEIRAS results support that the redox-active Id molecule can significantly facilitate CO₂ activation to form key intermediates at the Ag/Id interface, namely *CO₂^– and *COOH, ultimately leading to promoted CO generation.

Optimizing CO₂ binding affinity on indigo derivatives

Our combined experimental and theoretical results confirmed the hypothesis that modifying Ag catalysts with CO₂-binding Id moieties can create reactive interfaces conducive to CO₂ activation and subsequent conversion to CO. Notably, the advantage of employing redox-active CO₂-binding organic molecules as promoters lies in the ability to leverage precise synthetic control to finetune their CO₂ affinities^18,20. This enables the insightful interrogation of the correlations between molecular properties and catalytic performance and thus directs the optimization of intermediate adsorptions in the CO₂RR-to-CO pathway. Our previous studies demonstrated that introducing electron-withdrawing groups (EWGs, such as −Br and −COOH) to the structure of redox-active compounds weakens their interaction with CO₂ by reducing the electron density at the binding sites while introducing electron-donating groups (EDGs, such as −OCH₃ and −OC₃H₇) enhances CO₂ affinity¹⁸. Therefore, we synthesized a series of indigo derivatives with different substituent groups (see Supplementary Note 1 for synthesis details), including 5,5′,6,6′-tetramethoxylindigo (TMId), 5,5′-dipropoxyindigo (DPId), 6,6′-dibromoindigo (DBId), and indigo-6,6′-dicarboxylic acid (DCId), to investigate their impact on Ag-catalyzed CO₂RR.

CV curves of the synthesized indigo derivatives were examined in both aprotic and aqueous electrolytes to assess their CO₂ complexation behaviors (Supplementary Figs. 10 and 11). Similar to Id, all of these indigo derivatives exhibit chemical reactions with CO₂ upon electro-reduction in both environments, as evidenced by the positive shifts in redox potential under CO₂ compared to N₂ atmospheres. It is well-established that, in aprotic solvents, a linear scaling relationship exists between the reduction potential and the CO₂ affinity of redox-active CO₂-binding molecules³⁰. Specifically, species with more anodic reduction potentials generally show weaker CO₂ affinities. Here, we find that the onset potentials for CO₂ capture in aprotic and aqueous electrolytes also display a strong linear correlation (Fig. 2a). This finding implies that molecular engineering remains effective in tuning the interaction strength between CO₂ and indigo derivatives in aqueous electrolytes, which is also supported by our DFT calculation (Supplementary Table 2). Thus, the CO₂ affinity of indigo derivatives gradually decreases across the series of TMId, DPId, Id, DBId, and DCId. Furthermore, it is reasonable to infer that this trend in CO₂ affinity can extend to the corresponding hybrid interface with Ag catalysts for CO₂RR, as corroborated by the AgNP+Id case discussed above.

Fig. 2: Electrochemical properties of indigo derivatives and their promotional effects on CO₂RR when modified on AgNP.

a Linear relationship between the onset potentials for CO₂ capture in aprotic and aqueous electrolytes for indigos functionalized with EDGs or EWGs. b, c Comparison of CO FE at different j_total (b), and j_CO at different potentials (c) for AgNP modified with various indigos. d Correlation between the CO₂RR potential at a j_total of ~ 30 mA cm^–2 for modified AgNP catalysts and the onset potential for CO₂ capture by various indigos. Potentials have been 100% iR corrected. The flow cell was operated with 1 M KOH (pH = 14.0 ± 0.2). Data are presented as mean ± s.d. Error bars represent s.d. from measurements of three independent electrodes. Source data are provided as a Source Data file.

The indigo derivatives were then used to modify AgNP catalysts following the same procedure as that used for AgNP+Id. Again, there are no apparent changes in the structure or chemical state of AgNP after modification (Supplementary Figs. 12–17). The as-prepared hybrids are denoted as AgNP+TMId, AgNP+DPId, AgNP+DBId, and AgNP+DCId, respectively. Introducing these indigo derivatives to AgNP drastically promotes CO₂RR compared to bare AgNP (Supplementary Figs. 18 and 19). All modified samples display similar enhancement in CO selectivity, achieving FEs of ~ 93% across the investigated current densities (Fig. 2b). Nevertheless, pronounced variations in j_CO with respect to overpotential can be observed. Using the CO activity of AgNP+Id as a benchmark, we found that indigos functionalized with EWGs (DBId and DCId) can further accelerate CO generation on Ag, while those functionalized with EDGs (TMId and DPId) demand higher overpotentials for CO₂-to-CO conversion (Fig. 2c). The best CO₂RR performance is achieved with DCId, whose CO₂ affinity is the weakest among the indigo derivatives. Specifically, AgNP+DCId attains a j_CO of 273 mA cm^–2 at –0.54 V vs. RHE, which is ~ 2.6 times higher than that of AgNP+TMId (the derivative with the strongest CO₂ affinity) at a similar potential.

Intriguingly, we identified a linear relationship between the CO₂RR potential of the modified catalysts and the onset potential for CO₂ capture of various indigos (Fig. 2d). This indicates that the CO₂RR activity is strongly correlated with the varying *CO₂^– adsorption energy at the catalytic interface. DFT calculations reveal that the *CO₂^– adsorption energy at the Ag/DCId interface (–0.583 eV) falls between those of the Ag/Id interface (–0.635 eV) and the bare Ag surface (–0.211 eV, Supplementary Fig. 20). We postulate that, for an optimal CO₂RR performance, the CO₂-binding capabilities of molecular modifiers should be sufficiently strong to effectively stabilize the *CO₂^– intermediate at the molecule/catalyst interface, yet not too strong to potentially slow down the subsequent hydrogenation and/or *CO desorption steps in the CO pathway³¹. To validate this hypothesis, we evaluated the CO₂RR performance of thioindigo-modified AgNP. Thioindigo has a much weaker interaction with CO₂ due to its electron-deficient nature³², as evidenced by its nearly identical redox behavior under CO₂ and N₂ conditions in an aqueous electrolyte (Supplementary Fig. 21b and Supplementary Table 2). As a result, thioindigo shows no promotional effect on AgNP for CO₂-to-CO conversion (Supplementary Fig. 21). Plotting the CO₂RR potential against the onset potential for CO₂ capture over AgNP electrodes modified with various indigos affords a volcano-like trend (Supplementary Fig. 22), with AgNP+DCId exhibiting the most favorable *CO₂^– adsorption energy toward CO formation in our case.

As a final note, no trends in CO₂RR activity were observed in relation to surface hydrophobicity or roughness across the different modified electrodes (Supplementary Figs. 23–25). The comparable ECSAs of these samples also suggest that the availability of Ag active sites is not substantially affected by the modification (Supplementary Table 1). These results rule out other possible factors as main contributors to the enhanced CO₂RR performance of AgNP modified with indigo derivatives.

Immobilizing CO₂-binding moieties via polymerization

Albeit being the most effective promoter, molecular DCId is susceptible to dissolution into the electrolyte during CO₂ electrolysis at relatively high current densities (Supplementary Fig. 26). To achieve a more stable molecular modification, we immobilized the redox-active CO₂-binding moieties into a macromolecular structure. As shown in Fig. 3a, we synthesized an indigo-based polymer (P-Id) by amidation between DCId and 2,2′-(ethylenedioxy)diethylamine (DODA, see Methods for synthesis details). Fourier-transform infrared (FTIR) spectroscopy of P-Id displays a vibration band associated with the amide C−N stretching at ~ 1540 cm^–1, accompanied by the disappearance of the C=O stretching band of carboxylic acid from the DCId precursor at ~ 1680 cm^–1 (Fig. 3b). These findings indicate the successful amidation polymerization between the two precursors. The formation of the proposed P-Id structure is further verified by solid-state ¹³C nuclear magnetic resonance (NMR) spectroscopy (Fig. 3c), where the observed peaks can be assigned to the different types of carbon marked in Fig. 3a.

Fig. 3: Synthesis and characterization of P-Id for a stable modification on AgNP.

a Synthesis of P-Id. HBTU: 2-(1H-benzotriazol-1-yl)−1,1,3,3-tetramethyluronium hexafluorophosphate; DIPEA: N,N-diisopropylethylamine; DMF: N,N-dimethylformamide. b FTIR spectra of DCId, DODA, and P-Id. c Solid-state ¹³C NMR spectrum of P-Id. The peaks are assigned to the corresponding carbons labeled in (a). Asteroids denote spinning sidebands. d, e Comparison of CO FE as a function of j_total (d), and j_CO as a function of potential (e) for AgNP, AgNP+DCId, and AgNP+P-Id. Potentials have been 100% iR corrected. The flow cell was operated with 1 M KOH (pH = 14.0 ± 0.2). Data are presented as mean ± s.d. Error bars represent s.d. from measurements of three independent electrodes. Source data are provided as a Source Data file.

It is important to note that the onset potential for CO₂ capture by P-Id closely matches that of DCId, confirming that the polymer structure retains the redox-active oxygen center without compromising its favorable CO₂ binding affinity (Supplementary Figs. 27 and 28). As anticipated, modifying AgNP with P-Id (AgNP+P-Id) by ultrasonic mixing results in substantially enhanced CO₂RR reactivity, with the CO FEs exceeding 90% across a wide range of j_total from 30 to 850 mA cm^–2 (Fig. 3d and Supplementary Fig. 29). Particularly, AgNP+P-Id achieves a CO FE of 99% at a j_total of 100 mA cm^–2, slightly outperforming AgNP+DCId (95%) at a similar j_total. This additional improvement is likely due to further HER inhibition in the more hydrophobic microenvironment induced by P-Id (Supplementary Fig. 30). Given the comparable CO₂ affinity of the redox-active centers, AgNP+P-Id exhibits almost identical overpotentials to AgNP+DCId at relatively low current densities (Fig. 3e). However, AgNP+P-Id enables CO₂ electrolysis at much higher current densities (j_CO up to nearly 700 mA cm^–2), thanks to the suppressed solubility and enhanced robustness of P-Id, as substantiated by post-electrolytic characterizations (Supplementary Fig. 31).

Engineering Ag catalyst for enhanced utilization of Ag/indigo interfaces

To enhance the availability of catalytic interfaces, we sought to improve the dispersion and utilization of Ag active sites by incorporating highly porous carbon black as the catalyst support (Supplementary Fig. 32). Carbon supports typically offer a better conductive network, mitigate nanoparticle agglomeration, and increase mass activity by reducing catalyst loading^33,34. Therefore, we prepared a carbon-supported Ag catalyst with abundant surface defects (D-Ag/C) via in situ electrodeposition (see Methods for synthesis details).

The Ag loading on the GDL was 6 μg cm^–2 (equivalent to 2 wt% on carbon black) measured by the inductively coupled plasma optical emission spectrometry (ICP-OES). High-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) and corresponding STEM-EDS mapping show that the Ag particles obtained via in situ electrodeposition have an average diameter of ~28 nm and are primarily isolated and well dispersed on the carbon matrix (Fig. 4a and Supplementary Fig. 33). High-resolution transmission electron microscopy (HRTEM) images indicate the presence of planar defects on the Ag surface (Fig. 4b and Supplementary Fig. 34), which can provide uncoordinated sites to potentially enhance CO₂ conversion^35,36. X-ray absorption spectroscopy (XAS) was applied to probe the electronic state and coordination structure of the D-Ag/C catalyst. The normalized X-ray absorption near-edge structure (XANES) spectra at the Ag K-edge confirm the metallic Ag nature of D-Ag/C (Fig. 4c), consistent with the XPS result (Supplementary Fig. 35). The Ag K-edge extended X-ray absorption fine structure (EXAFS) spectra (R-space) show the lowest intensity of peaks between 1.5 and 3.3 Å, corresponding to the first nearest neighboring Ag−Ag bond for D-Ag/C, in comparison to Ag foil and commercial AgNP (Fig. 4d). Corresponding fitting analysis indicates that the coordination numbers (CNs) of Ag decrease in the order: Ag foil (12) > commercial AgNP (10.8 ± 0.4) > D-Ag/C (8.0 ± 0.6) (Fig. 4e, Supplementary Fig. 36 and Supplementary Table 3). The slight CN diminution in AgNP relative to Ag foil is ascribed to the small particle size, while the further diminution in CN for D-Ag/C shall be mainly driven by the existence of abundant planar defects³⁷.

Fig. 4: Morphological and structural characterizations of D-Ag/C catalyst.

a HAADF-STEM and STEM-EDS mapping of the D-Ag/C catalyst showing isolated Ag particles on carbon support. b HRTEM image of the D-Ag/C catalyst showing abundant planar defects (denoted by arrows). c, d XANES data (c) and Fourier transform magnitudes of the k²-weighted EXAFS spectra (d) at the Ag K-edge of AgNP and D-Ag/C. The spectra of Ag foil were shown as a reference. e The Ag−Ag coordination numbers obtained by theoretical fits to the EXAFS data. Source data are provided as a Source Data file.

We subsequently benchmarked the CO₂RR performance of D-Ag/C in a flow cell. D-Ag/C possesses superior CO₂RR reactivity compared to carbon-supported AgNP (AgNP/C) with the same Ag loading of 6 μg cm^–2 (Supplementary Fig. 37). Furthermore, D-Ag/C achieves a maximum CO FE of 94% and a j_CO of 279 mA cm^–2 at –0.73 V vs. RHE (Fig. 5a and b). These values are ~ 1.1- and ~ 1.5-fold higher, respectively than those of commercial AgNP with a much higher Ag loading of 0.4 mg cm^–2 at the same potential (Supplementary Fig. 37). The combination of the carbon support and the in situ electrodeposition method effectively prevents the aggregation of defective Ag particles and increases the density of highly active sites participating in CO₂RR, leading to a significant enhancement in CO production for D-Ag/C.

Fig. 5: CO₂RR performance of D-Ag/C and D-Ag/C+P-Id catalysts.

a, b Comparison of CO FE as a function of j_total (a), and j_CO as a function of potential (b) for the D-Ag/C and D-Ag/C + P-Id catalysts. c Ag mass activity toward CO product at different potentials for the D-Ag/C + P-Id catalyst. Potentials have been 100% iR corrected. The flow cell was operated with 1 M KOH (pH =14.0 ± 0.2). Data are presented as mean ± s.d. Error bars represent s.d. from measurements of three independent electrodes. d Performance comparison of D-Ag/C+P-Id with reported electrocatalysts for CO₂RR to CO in the gas diffusion electrode (GDE)-based electrolyzer. Details are given in Supplementary Table 4. Source data are provided as a Source Data file.

The highly dispersed D-Ag/C catalyst provides exciting opportunities for creating rich catalytic interfaces when modified with P-Id. Indeed, introducing P-Id to D-Ag/C (D-Ag/C+P-Id) increases the j_CO to 818 mA cm^–2 at –0.73 V vs. RHE, nearly tripling the performance of bare D-Ag/C at the same potential, meanwhile maintaining a high CO FE of 93% (Fig. 5a, b and Supplementary Fig. 38). Analogous to the AgNP+P-Id case, P-Id modification drastically extends the CO₂-to-CO conversion across a wider current density range (Fig. 5a). The FEs for CO exceed 95% at a j_total ranging from 188 to 780 mA cm^–2. Remarkably, the FE for CO remains at 91% when the j_total is ramped up to 1.17 A cm^–2 at –0.80 V vs. RHE, ranking among the highest performances reported to date (Fig. 5c). The low Ag loading gives rise to a high mass activity of 174 A mg^–1_Ag for CO production at –0.80 V vs. RHE (Fig. 5c), which is nearly two orders of magnitude higher than previously reported AgNP catalysts and even outperforms state-of-the-art single-atom catalysts (SACs) for the CO₂RR to CO (Fig. 5d and Supplementary Table 4). Note that H₂ is the only product from electrolysis under an argon atmosphere, which confirms that the carbon source of CO evolution originates from the CO₂ feedstock (Supplementary Fig. 39). Such a collective strategy could potentially pave the way for large-scale, cost-effective, and sustainable CO₂-to-CO conversion (see Supplementary Note 2 for more discussions).

To improve the full-cell energy efficiency (EE), the D-Ag/C+P-Id catalyst was integrated into a zero-gap membrane electrode assembly (MEA) cell. The catalyst-coated membrane (CCM) method was utilized on the cathode and paired with a highly efficient Ni foam-supported Ni-Fe oxide as the anode to minimize voltage loss (Supplementary Fig. 40)^38,39,40. The MEA cell, operated at 50 °C with 0.2 M cesium hydroxide (CsOH) as the anolyte, yields steady CO selectivity across the investigated current density range of 0.10−1.20 A cm^–2 at cell voltages between 2.34 and 3.89 V (Fig. 6a). A peak CO FE of 94% with a j_CO of 375 mA cm^–2 is achieved at a cell voltage of 2.89 V, corresponding to an EE of 43% (Fig. 6b). The MEA cell also exhibits marginal decay in CO production performance over at least 20 hours at a j_total of 400 mA cm^–2 (Fig. 6c). Post-reaction characterizations demonstrate the promising durability of both the D-Ag/C and P-Id components (Supplementary Figs. 41 and 42).

Fig. 6: CO₂RR performance of D-Ag/C+P-Id catalyst in MEA cell.

a FE and cell voltage at different j_total. b j_CO and EE for CO at different cell voltages without iR compensation. Data are presented as mean ± s.d. Error bars represent s.d. from three independent measurements. c CO₂RR stability at a j_total of 400 mA cm^–2 when operating the MEA cell at 50 °C using 0.2 M CsOH (pH = 13.2 ± 0.1) as the anolyte.

Discussion

In summary, using redox-active indigo molecules as a model system, we demonstrate that decorating Ag-based catalysts with electro-activated CO₂-binding organics creates synergistic interfaces that significantly enhance both the selectivity and activity of CO₂-to-CO conversion. The dynamic complexation interactions between these organic promoters and CO₂ readily activate CO₂ molecules and effectively enrich the *CO₂^– and *COOH intermediates at the nearby Ag catalytic sites. Importantly, by precisely tuning the CO₂ binding affinities of indigo derivatives via molecular engineering, we unravel a critical volcano-like relationship between the *CO₂^– adsorption energy induced by organic modifiers and their promotional effect on CO₂RR. This mechanistic insight culminates in the development of a hybrid catalyst that couples polymerized indigo moieties with the optimal CO₂ affinity and highly dispersed, defect-rich Ag particles, achieving impressive CO₂RR performance at ampere-level current densities. Furthermore, we anticipate that this redox-active molecular platform can be extended to benefit CO₂ electrolysis to high-value multi-carbon products when incorporated into copper-based catalysts. Our work opens new avenues for the rational design of highly efficient, molecularly tailored catalysts. Moreover, the integration of CO₂-binding species with CO₂RR catalysts could potentially enable a desirable reactive carbon capture scheme where CO₂ from dilute sources is directly converted into chemicals and fuels without prior concentration, offering advantages in process intensification and energy efficiency.

Methods

Chemicals and materials

Silver nanopowder (20−40 nm, 99.9%) and silver nitrate (AgNO₃, 99.9%) were obtained from Thermo Scientific. Indigo powder (97.0%) was purchased from TCI America. Chemicals for polymerization, including HBTU (99.0%), DIPEA (99.5%), DODA (98.0%), and DMF (99.9%), were purchased from Sigma Aldrich. All chemicals are used as received without purification.

Organic synthesis

Synthesis procedures for indigo derivatives are provided in Supplementary Note 1. P-Id was prepared via amidation polymerization. Specifically, to a mixture of DCId (0.7 g, 2 mmol) and HBTU (1.64 g, 2.16 mmol) in anhydrous DMF (30 ml) was added DIPEA (2.09 ml, 12 mmol) dropwise. The mixture was stirred for 30 mins before DODA (2 mmol, 292 μl) was added. The reaction was stirred for another 24 hours at room temperature (20 °C) and poured into a saturated NaHCO₃ aqueous solution (200 ml). The precipitate was collected by filtration, washed sequentially with saturated NaHCO₃ (aq), water, 0.2 M HCl (aq), cold methanol, and dichloromethane, and dried in vacuo to give a dark blue solid (784 mg, 85% yield).

D-Ag/C catalyst synthesis

The D-Ag/C catalyst was synthesized using the in situ electrodeposition method. Vulcan XC-72 carbon black suspension (10 mg ml^–1) was added to AgNO₃ aqueous solution (0.33 mg ml^–1) and sonicated in an ice bath for 2 h to allow the uniform dispersion of Ag⁺ on the carbon black support. The resulting slurry was added to isopropanol with 5 wt% Nafion solution to obtain the catalyst ink, which was then spray-coated on GDL (Sigracet 39BB) until reaching a total mass loading of 0.3 mg cm^–2. The D-Ag/C catalyst was formed in situ by reducing the as-prepared GDE at 30 mA cm^–2 for 15 min in the flow cell supplied with CO₂ gas and 1 M KOH. The final Ag loading on the GDE was ~ 6 μg cm^–2 measured by ICP-OES.

Material characterizations

FTIR spectra were collected on a Thermo Fischer Nicolet iS5 spectrometer. Solid-state ¹³C NMR spectrum was obtained from a Bruker Ascend 500 MHz Solids at a spin rate of 10000 Hz. Liquid ¹H NMR spectra were recorded on a Bruker Avance 400 MHz with solvent residue peak as the internal reference. SEM images were collected on a Helios G5-focused ion beam SEM. XPS measurements were carried out using a PHI Quantera XPS equipped with the Al Kα radiation source. HAADF-STEM, STEM-EDS elemental mapping, and HRTEM were collected from a FEI Titan Themis³ S/TEM. XAS measurements were performed on the Beamline 7-BM of the National Synchrotron Light Source II (NSLS II) at Brookhaven National Laboratory. The Athena and Artemis software in the Demeter package was applied for data processing and analysis⁴¹. The theoretical EXAFS signal was fitted to the experimental EXAFS data in R-space by Fourier transforming both the theoretical and experimental data.

Electrochemical measurements

CV measurements were carried out on a BioLogic VSP potentiostat (BioLogic Science Instruments). Glassy carbon (3 mm in diameter) was used as the working electrode, and Pt wire was used as the counter electrode. For measurements conducted in aprotic electrolytes, Ag wire was used as a pseudo-reference electrode, and ferrocene was used as an internal reference. The organic compound (2.5 mM) was dissolved in DMSO with 100 mM TBAPF₆ as the supporting salt. For measurements conducted in aqueous electrolytes, Ag/AgCl (3 M KCl) was used as the reference electrode. The reference electrode was calibrated using a standard hydrogen electrode before measurements. The compound was mixed with Vulcan XC-72 (mass ratio of 1:1) in isopropanol and drop-casted onto the glassy carbon electrode. The pH of the electrolyte (1 M KHCO₃ saturated with either N₂ or CO₂) was measured by a pH meter (SevenCompact). The potential was converted to the RHE scale using E_RHE = E_Ag/AgCl + 0.209 V + 0.0591 × pH. CV curves were collected at the scan rate of –20 and –10 mV s^–1 for the aprotic and aqueous conditions, respectively.

Electrocatalytic CO₂RR measurements

CO₂RR performances were evaluated in a flow cell at 20 °C with 1 M KOH (pH = 14.0 ± 0.2) as the electrolyte. The fresh electrolyte was prepared using a volumetric flask before the testing. For the GDE fabrication, AgNP or AgNP mixed with organic species of interest were dispersed in isopropanol with 5 wt% Nafion. The concentration of AgNP was 5 mg ml^–1. The molar ratio of small molecules to AgNP was kept at 10%. The P-Id content was 50 wt% relative to AgNP. After sonication for 2 h, the ink was spray-coated onto the GDL. The Ag loading was controlled to be ~ 0.4 mg cm^–2 for all samples. The GDE cathode and Ni foam anode were separated by an anion exchange membrane (AEM; Sustainion X37-50, 50 μm, 3 × 3 cm²). AEM was submerged into 1 M KOH for 12 h and rinsed with deionized water before assembling into the cell. A multichannel peristaltic pump (Ismatec) was applied to feed the electrolyte (30 ml) at 1 ml min^–1 and 2 ml min^–1 to the cathodic and anodic compartments, respectively. A mass flow controller (Alicat Scientific) was employed to control the CO₂ gas flow rate at 50 standard cubic centimeters per minute (sccm) to the cathode. The CO₂RR experiments were operated in the chronoamperometry mode under ambient conditions on a BioLogic VSP potentiostat. When the current density exceeded 1 A cm^–2, a d.c. power supply (B&K Precision) was used instead of the potentiostat. The cathodic potential was determined relative to the Ag/AgCl (3 M) reference electrode and converted to the RHE scale accordingly. Because of the rapid dynamic evolutions of local environments, the resistance (R) between the cathode and reference electrode was immediately measured using potentiostatic electrochemical impedance spectroscopy (PEIS) after each electrolysis, and then the ohmic drop (iR) was 100% manually compensated to each potential. The R was around 3.3 ± 0.3 Ω for an electrode area of 1 cm². Non-iR corrected data were provided in Supplementary Table 5.

In the MEA setup, a more robust AEM (PiperION, 40 μm, 3 × 3 cm²)⁴² was used for the CCM fabrication. In brief, the catalyst was airbrushed onto the AEM directly, and combined with a highly hydrophobic GDL (Tory 090) to form the cathode³⁸. The anode, Ni-Fe oxide, was prepared using a reported electrodeposition method on Ni foam³⁹. The MEA cell was operated at 50 °C with 0.2 M CsOH (pH = 13.2 ± 0.1) as the anolyte to mitigate salt precipitation and the consequent electrode flooding⁴³. The catalyst, membrane, and cell setup were initially activated at 30 mA cm^–2 for 2 h before starting the performance measurements. The cell voltage was recorded without the iR correction.

During CO₂ electrolysis, an online gas chromatograph (GC, Shimadzu GC-2014) equipped with a thermal conductivity detector was employed to monitor the gas products. The FEs of gas products were calculated as follows:

$${FE}\left(\%\right)=\frac{{zFxV}}{{j}_{{total}}}\times 100\%$$

where z is the number of electrons transferred to form a target product; F is the Faraday constant; x is the molar fraction of a target product determined by GC; V is the molar flow rate of effluent gas measured using a digital flow meter (Omega); and j_total is the total current density.

The EEs of CO were calculated as follows:

$${EE}\left(\%\right)=\frac{1.23-{E}_{{{{\rm{CO}}}}}^{0}}{{E}_{{{{\rm{cell}}}}}}\times {{FE}}_{{{{\rm{CO}}}}}$$

where ${E}_{{{{\rm{CO}}}}}^{0}$ is the equilibrium potential for CO₂RR to CO (– 0.10 V vs. RHE)⁷; ${E}_{{{{\rm{cell}}}}}$ is the applied cell voltage; and ${{FE}}_{{{{\rm{CO}}}}}$ is the FE for CO.

In situ ATR-SEIRAS measurements

The ATR-SEIRAS setup was assembled according to prior works^44,45. The catalyst ink was drop-casted onto a silicon ATR wafer with a thermal-evaporated Ag film (30 nm). The catalyst loading was controlled to be 0.4 mg cm^–2. Graphite rod and Ag/AgCl (3 M KCl) were used as the counter and reference electrodes, respectively. The SEIRAS spectra were recorded using a Thermo Fischer Nicolet iS50 spectrometer equipped with an N₂ cooled HgCdTe (MCT) detector and a Veemax III IR attachment from PIKE. The spectrometer was operated at a scan rate of 30 kHz. Spectra were acquired with a spectral resolution of 4 cm^–1, and 16 interferograms were coadded for each spectrum. During in situ measurements, the electrolyte (0.1 M potassium phosphate in D₂O) was continuously sparged with CO₂. CO₂ electrolysis was carried out at potentials ranging from –0.1 to –1.0 V vs. RHE. The spectrum collected at open circuit potential was used as the reference.

DFT calculations

DFT calculations were performed using the Vienna Ab initio Simulation Package (VASP)⁴⁶. The Perdew-Burke-Ernzerhof (PBE) functional⁴⁷ was applied, with a plane wave energy cutoff of 400 eV. Van der Waals interactions were accounted for using the DFT-D3 dispersion correction⁴⁸. A constant potential of –0.3 V vs. RHE was used in the simulations, and the pH was set to 14, reflecting the alkaline conditions of the electrolyte. Silver relaxation was modeled using a (6 × 6 × 6) supercell with a 15 Å vacuum gap, and the Gamma point was employed for k-point sampling. The adsorption of CO₂ and indigo on the Ag(111) surface was studied using the constant-potential model (CPM)⁴⁹ with a VASPsol patch to consider the implicit solvation^50,51. The adsorption of two CO₂ molecules on the indigo and its derivatives was calculated with two extra electrons. Considering the increased complexity of the derivative molecules and the need for a much larger Ag surface, we calculated the adsorption on isolated molecules for qualitative analysis (Supplementary Table 2).