CO₂ reduction to acetate in mixtures of ultrasmall (Cu)_n,(Ag)_m bimetallic nanoparticles

Significance

Efficient reduction of carbon dioxide to useful fuels and chemicals is an important research goal in artificial photosynthesis. Significant progress has been made for the C1 products, CO and HCOO⁻. We report here a procedure based on the use of ultrasmall, monodispersed Cu and Ag bimetallic nanoparticles on thin, electrochemically polymerized poly-Fe(vbpy)₃(PF₆)₂ films. They reduce CO₂ to acetate at pH 7 in aqueous HCO₃⁻ solutions at relatively high efficiencies with significant rate enhancements with added benzotriazole. In the sequence of clusters, the most efficient results for acetate production were obtained in films of (Cu)₂,(Ag)₃ with a faradaic efficiency of 21.2% for acetate from CO₂ at −1.33 V vs. reversible hydrogen electrode in 0.5 M KHCO₃ with 8 ppm of added benzotriazole at 0 °C.

Abstract

Monodispersed mixtures of 6-nm Cu and Ag nanoparticles were prepared by electrochemical reduction on electrochemically polymerized poly-Fe(vbpy)₃(PF₆)₂ film electrodes on glassy carbon. Conversion of the complex to poly-Fe(vbpy)₂(CN)₂ followed by surface binding of salts of the cations and electrochemical reduction gave a mixture of chemically distinct clusters on the surface, (Cu)_m,(Ag)_n|polymer|glassy carbon electrode (GCE), as shown by X-ray photoelectron spectroscopy (XPS) measurements. A (Cu)₂,(Ag)₃|(80-monolayer-poly-Fe(vbpy)₃²⁺|GCE electrode at −1.33 V vs. reversible hydrogen electrode (RHE) in 0.5 M KHCO₃, with 8 ppm added benzotriazole (BTA) at 0 °C, gave acetate with a faradaic efficiency of 21.2%.

Efficient reduction of carbon dioxide to useful fuels and chemicals is an important research goal of artificial photosynthesis (1–7). Solar or electrochemical reduction of CO₂ is complicated by the large overpotentials required for one-electron transfer reduction to CO₂⁻ [with E(CO₂/CO₂⁻) = −1.9 V vs. normal hydrogen electrode (NHE) at pH 7] and a high reorganizational energy (8). Subsequent steps to give formate or CO are energetically more favorable. A growing number of catalytic systems for CO₂ reduction have been identified that give C₁ products (9–13) but obtaining multicarbon (C₂₊) products has proven more difficult due to the high barrier for C–C bond formation on electrode surfaces (14).

Current research has led to robust catalysts for CO₂ reduction, including molecular (15, 16), metallic (10, 17, 18), and nonmetallic catalysts (9) with an extensive literature on metal catalysts (19, 20). Cu has proven to be the most effective metallic CO₂ reduction catalyst with an ability to form a variety of high-energy-density products including high-value-added, multiple carbon, C₂₊, products (20). At Cu electrodes, Jaramillo and coworkers (21) have identified 16 CO₂ reduction products, including 12 multicarbon products. In follow-up experiments, the lack of selectivity by Cu has led to the synthesis of Cu-based bimetallic catalysts (3, 22, 23) or to decreases in the size of the Cu catalysts (24). In these experiments, difficulties arise from complicated and time-consuming synthetic procedures by contamination by added surfactants and catalysts (25).

Electrochemical deposition provides a fast and convenient approach for depositing nanoparticles on electrode surfaces (26). The procedure can be less time-consuming than wet chemical synthesis, but there are difficulties in obtaining small, confined nanoparticles and an extended range in size from hundreds of nanometers to micrometers in diameter (27). Here we utilize a procedure described earlier for surface binding and cluster formation (24) to prepare ∼6-nm clusters of Cu and Ag on the surfaces of electropolymerized films. We also report that surface-dependent, electrochemical reduction of CO₂ occurs on these surfaces with C–C coupling to give acetate as a significant product.

Results and Discussion

A general procedure for preparation of the electrodes has been described previously (2, 28) and is shown in an adapted form in Scheme 1. In the procedure, the vinyl-derivatized salt [Fe(vbpy)₃][PF₆]₂ (vbpy is 4-methyl-4′-vinyl-2,2′-bipyridine) is electropolymerized on the surface of a glassy carbon electrode (GCE). Subsequent addition of CN⁻ to an external solution causes displacement of a –vbpy ligand by CN⁻ by immersing the polymerized electrode in 0.1 M [TBA][CN] for 20 min (28). The CN⁻ ligands act as a bridge to added cations in solution and exposure of the films gives the ligand-bridged film adducts Fe^II-CN-M^m+ (2, 28). Further reduction of the films at potentials sufficient to reduce the coordinated metal ions results in reduction and transfer of the reduced metals to the surface of the films, Eq. 1, where nanoparticles form. For films loaded with multiple metal ions, composite nanoclusters form on the electrode surface providing a basis for multiple surface interactions and cluster formation. On surfaces on which multiple nanoparticles are added, the ratio of available metals can be tuned by the metal ion composition in the precursor solution.

$$n{\text{Fe}}^{\text{II}}{\text{-}\mathrm{CN}\text{-}\mathrm{M}}^{m+},\text{bpy}{\hspace{0.17em}}^{\stackrel{+m{e}^{-},-{\text{CN}}^{-}}{\to }}\hspace{0.17em}n{\text{Fe}}^{\text{II}}{\left(\text{bpy}\right)}^{2+},{\left({\text{M}}_n\right)}^0.$$ [1]

Polymerization of [Fe(vbpy)₃]²⁺ onto a 0.071 cm² GCE in 1 mM solution, in 0.1 M TBAPF₆ MeCN, was monitored by cyclic voltammogram (CV) scanning at 250 mV s⁻¹ (Fig. 1A). Well-defined redox waves for the couples poly-[Fe(vbpy)₃]^2+/1+ and poly-[Fe(vbpy)₃]^1+/0 at E_1/2 = −1.28 V and −1.45 V vs. Ag/Ag⁺ were observed. As shown, the peak currents increase with the number of cycles, consistent with the growth of poly-Fe(vbpy)₃²⁺ on the electrode surface. The latter occurs by reduction of the vbpy ligands to vinyl radicals, initiating C–C bond formation (2, 28, 29). Given the presence of multiple polymerizable vinyl substituents, polymerization is accompanied by extensive cross-linking as the polymer deposits on the electrode surface to form the porous open structure shown in Fig. 1E. The surface coverage of poly-Fe(vbpy)₃²⁺ was estimated to be 6.44 ± 0.48 nmol/cm² from peak current integrations of the poly-[Fe(vbpy)₃]^3+/2+ wave at E_1/2 = 1.03 V vs. Ag/Ag⁺ (Fig. 1B). Based on this analysis, the coverage used in this study was ∼80 molecular layers with a monolayer surface coverage of 8 × 10⁻¹¹ mol/cm² (2, 28, 29). The thickness of the polymer film on the electrode was ∼250 nm (Fig. S1).

Scheme 1.

Fabrication of Cu,Ag-polymer-GCE electrodes.

Fig. 1.

(A) Room-temperature CVs showing the electropolymerization of 1 mM [Fe(vbpy)₃](PF₆)₂ on a GCE in 0.1 M TBAPF₆/MeCN at 250 mV s⁻¹. (B) CV of the poly-[Fe(vbpy)₃]^3+/2+ couple before replacement of -vbpy (black solid line) and after replacement (gray solid line) at 100 mV s⁻¹ in 0.1 M TBAPF₆/MeCN, room temperature. (C) SEM image for a (Cu)_m,(Ag)_n/80-layer-poly[Fe(vbpy)₃](PF₆)₂/GCE surface. (D) TEM image and size distribution (Inset) for (Cu)₂,(Ag)₃ nanoparticles; and XPS spectra for Ag (E) and Cu (F) in (Cu)₂,(Ag)₃-polymer-GCE.

Following the −CN displacement step, the poly-[Fe(vbpy)₃]^3+/2+ couple shifts cathodically by 0.51–0.52 V vs. Ag/Ag⁺, coincident with −bpy replacement by –CN⁻ (28). The change in film structure is also shown in UV-vis spectra (Fig. S2). There is a red shift in the visible absorption bands for the dπ(Fe^II) → π*(bpy) metal-to-ligand charge-transfer transitions following –CN replacement. The metal ions were added by immersing poly[Fe(vbpy)₃](CN)₂|GCE in solutions containing different ratios of Cu²⁺ and Ag⁺ in CH₃CN which leads to the rapid incorporation of Cu(II) and Ag(I). Addition of the metal ions was evident by shifts in UV-vis spectra with a blue shift observed upon addition of the ions. After binding to the metal ions, the cyano group becomes a better π-acceptor, stabilizing the dπ levels, decreasing the dπ → π*energy gap (28). Following addition of the ions, a single-pulse electrodeposition at −1.7 V vs. Ag⁺ for 30 s was carried out which led to rapid reduction of the cyano-bound Cu(II)/Ag(I) ions to Cu(0)/Ag(0).

Fig. 1C characterizes the surface morphology of the polymerized electrode. The polymer film has an open porous structure which presumably enhances mass transport during electrochemical reactions (30). The nanoparticulate structure is not obvious in the cross-sectional image in Fig. S1 because the particle sizes were too small to resolve by SEM. The average (Cu)₂,(Ag)₃ particle diameter was 6 ± 2 nm as characterized by the transmission electron microscope (TEM) images in Fig. 1D. This small particle size is documented in uniform distributions of particles prepared by direct electrodeposition (Table 1). The X-ray photoelectron spectroscopy (XPS) spectrum in Fig. 1E shows the characteristic metallic Ag 3d 2/5 band at 366.8 eV and a mix of metallic Cu and its oxide form by XPS in Fig. 1F (31, 32). The appearance of oxide in the Cu sample is not surprising because the Cu nanoparticles are easily oxidized, especially when small (31, 33). Based on the binding energies for Cu and Ag from XPS measurements, compared with their monometallic forms (Fig. S3), in the Cu–Ag particle mixtures, there are separate Cu and Ag nanoparticle clusters rather than mixed-metal alloys. This in contrast to earlier results described for clusters of Cu_xPd_y where electrochemical formation of clusters of the alloy was observed (2). The difference in behavior may be due to differences in the electrochemical deposition methods. Single-pulse electrodeposition provides a large driving force over a short period of time, initially with selective deposition of the more reducible metal. The surface atomic ratio of Cu and Ag was determined by XPS with at least three parallel measurements for each sample.

Table 1.

Properties of electrodeposited of Cu and Ag nanoparticles

Material	d/nm	Condition	Substrate	Technique	Ref.
Cu	25.1	0.029 g Cu(BF4)2·xH2O and 0.058 g of dodecylbenzene sulfonic acid sodium salt	Au	Electrochemical cycling	(41)
	51	1 mM CuSO4 in 0.1 M H2SO4	HOPG	Double-pulse deposition	(42)
	20–80	2.0 mM CuSO4 in 0.1 M Na2SO4	GCE/CNTs	Electrochemical cycling	(43)
	50–80	10 g/L CuSO4·5H2O, 12 g/L NaCl and 20 g/L H3BO3, 50 °C	TiO2-NT	Direct electrodeposition	(44)
Ag	50–150	1.0 mM AgNO3 in 0.1 M KNO3	GCE/ZnO	Double-pulse deposition	(45)
	50–80	0.43 mg/mL AgNO3 in 0.1 M NaNO3	Au	Electrochemical cycling	(46)
	16.6	5 mM AgNO3 in 0.1 M KNO3 with poly (N-vinylpyrrolidone) as the additive	Pt	Electrodeposition	(47)

During the electrodeposition process, nucleation and growth from within the particle occur sufficiently rapidly to separate the two effects completely (34). Under normal conditions, with high concentrations of metal ions in the external solution at high applied potentials, large particle sizes with a broad size distribution would be expected (34). In the strategy adopted here, CN⁻ coordination plays a critical role in dictating particle size and distribution. Electrodeposition, under the same condition but without the CN⁻ ligand, resulted in the appearance of a film for Cu and a large flake for Ag as shown in Fig. S4. Presumably, the diluted concentration of metal ions in the electrodes, by coordination to the CN⁻ ligand, decreases local interactions resulting in a decrease in particle size (34).

The electrocatalytic reactivities of the nanoparticles of the polymer-stabilized, (Cu)_m,(Ag)_n/GCE electrodes for CO₂ reduction were evaluated in 0.5 M KHCO₃ with 8 ppm of benzotriazole (BTA, Scheme 2) added. To increase the solubility of CO₂, all experiments were performed at 0 °C (35). The products of CO₂ reduction were identified and quantified by ¹H NMR in the liquid phase, and by GC for the headspace gaseous products. This (Cu)_m,(Ag)_n/polymer/GCE assembly showed a stable catalytic current over extended electrolysis periods as shown in Fig. S5.

Scheme 2.

Structure of BTA.

The 8e⁻ product (Table S1), CH₃COO⁻ (Fig. 2A), is a highly reduced product. Its appearance is unusual because of the difficulties in overcoming high-energy barriers for C–C formation on surfaces at low temperatures. Its appearance as the major liquid product at low temperatures is one of the few examples in CO₂ reduction chemistry (21). Other products that appear include formate in the liquid phase and CO and methane in the gas phase. A control experiment was conducted using the identical parameters except N₂ was used in place of CO₂. Under this condition, no product peaks were observed in the 1H NMR (Fig. S6). The major product in the background is hydrogen from reduction of the solvent.

Fig. 2.

(A) NMR spectrum of acetate (δ = 1.83 ppm) with DMSO (δ = 2.5 ppm) as the internal standard. (B) Efficiency of acetate formation as a function of the atom ratios of Cu and Ag in (Cu)_m,(Ag)_n/polymer/GCE in CO₂-saturated 0.5 M KHCO₃, −1.53 V vs. RHE after 3,600 s, 0 °C. (C) Product distribution with (solid) without (empty) 8 ppm BTA on (Cu)₂, (Ag)₃|80-layer-poly[Fe(vbpy)₃](PF₆)₂|GCE electrode, condition as above. (D) As above, influence of the applied potential on acetate production for a (Cu)₂,(Ag)₃/polymer/GCE electrode.

The appearance of acetate by the Cu_m,Ag_n clusters led to a more detailed investigation. The results of a study in which the surface metal atom composition was varied are shown in Fig. 2B. Data in the figure appear for acetate generation from the clusters Cu, Cu₃–Ag, Cu₂–Ag, Cu₂–Ag₃, and Ag in CO₂ saturated 0.5 M KHCO₃ at 0 °C at E = −1.53 vs. RHE over periods of 1 h. Based on the data in the figure, there was an increase in the appearance of acetate as Ag was added to the nanoparticles with a maximum reached at 5.5% for Cu₂–Ag₃ clearly showing that surface metal composition plays an important role in acetate formation.

The influence of BTA on CO₂ reduction was also investigated with the reactive Cu₂–Ag₃ cluster mixture. As shown in Fig. 2C, CO₂ reduction by the Cu₂–Ag₃/polymer/GCE electrode, at 0 C at −1.53 V vs. RHE for 1 h, gave acetate and methane at 5.5% and 1.8% with 8 ppm added BTA. Without BTA the values were 0% acetate and 0.3% methane. Similar results were obtained for a Cu/polymer/GCE (Fig. S7) film. Without added BTA, acetate 0.23% and methane appeared in 23% yields and with acetate the yields were 0.7% with methane at 3.10%. The role of BTA was selective. For example, there was no impact of added BTA on Ag/polymer/GCE nanoparticle reductions. Its role may arise from its adsorption on Cu with chemical changes at the electrode interface. BTA has been shown to adsorb on metallic Cu surfaces by physical and chemical adsorption (36).

The potential dependence of acetate generation by the (Cu)₂,(Ag)₃ cluster was also investigated. Faradaic efficiencies were evaluated as a function of applied potential with results shown in Fig. S8. At −1.33 V vs. RHE, acetate was generated in 21.2% yield. Decrease of the surface potential to −0.93 V resulted in no acetate. The appearance of acetate increases with applied potential with the highest efficiency of 21.2% found at −1.33 V vs. RHE. Further decreases in potential resulted in decreased formation of acetate with no acetate formed by −1.73 V. A related behavior has been reported by Jaramillo and coworkers at Cu electrodes with CH₃COO⁻ maximized as a product at −1.1 V vs. RHE and H₂ the dominant product at more negative potentials.

The mechanism of C–C formation is still debatable. From the data in Fig. 2B, addition of Ag clusters to the surface plays an important role with acetate maximized as a product for the cluster pair Cu₂–Ag₃. Ag is a known catalyst for CO₂ reduction to CO with insight on mechanism from calculations by Peterson and Nørskov (37). Once formed by loss of CO₂ from the surface, CO(gas) is favored thermodynamically as shown in Scheme 3. Cu, on the other hand, has a moderate ability to adsorb and desorb CO (37). A possible contributor to acetate production may be Ag-catalyzed formation of CO followed by its capture on neighboring Cu sites for C–C bond formation, Scheme 3. A similar observation was made by Yeo and coworkers (3) for phase-segregated CuZn nanoparticles.

Scheme 3.

Possible reaction pathways for CO₂ reduction to acetate on (Cu_m),(Ag)_n/polymer/GCE electrodes.

In the mechanism in the scheme, acetate could be produced after a series of hydroxylation and electron transfer steps (21). In an alternate route, CO insertion (3, 38, 39) as observed by Kenis and coworkers (38) on pure Cu surfaces could also occur in ethanol generation. Following CO₂ reduction to –CH₂ on the surface of Cu, dimerization of the latter with free CO from Ag would occur to give the C–C bond on the surface. Once formed on the surface, further reduction of acetaldehyde would give acetate as the final product (40).

Conclusion

We have introduced here the details of a well-defined electrochemical procedure for preparing a family of (Cu)_m,(Ag)_n clusters on an underlayer of poly[Fe(vbpy)₃](PF₆)₂|GCE. Monodispersed, ultrasmall, ∼6-nm particles were obtained by electrodeposition within the films to give nanoparticles that are among the smallest particle sizes available by this technique. In their reduction chemistry, a key factor is reduction of CO₂ with an enhancement by added BTA. The ability of clusters to undergo CO₂ reduction to acetate is highly dependent on the mole fraction ratio in the (Cu)_m,(Ag)_n/polymer/GCE mixtures. In the sequence of clusters, the most efficient results for acetate were obtained for (Cu)₂,(Ag)₃/polymer/GCE films with a faradaic efficiency of 21.2% for acetate at −1.3 V vs. RHE in 0.5 M KHCO₃ with 8 ppm of BTA at 0 °C.

Methods

Electrochemical Measurements.

All of the electrochemical measurements were performed on a CHI601 D potentiostat station (CH Instruments, Inc.) with a three-electrode setup by using a Pt mash as the counterelectrode, a Ag (for nonaqueous system) and a saturated calomel electrode (for aqueous system) as the reference electrode, and a GCE as the working electrode. The GCE was first polished with a decreasing size of 1.0- and 0.05-μm alumina lapping compounds for 5 min each and followed by sonication in an ultrasound bath before use. The potential used for CO₂ reduction reaction (CO2RR) is corrected to RHE to make comparison with literature.

Electrode Fabrication.

The polymerization process was carried out in a nitrogen-saturated solution containing 1 mM [Fe(vbpy)3](PF6)2 in 0.1 M TBAPF6/CH3CN. The GCE was then immersed into this well degassed solution to perform consecutive scans to get poly-[Fe(vbpy)3]/GCE. This electrode was then transferred into 0.1 M TBACN/MeCN solution and left for 20 min to allow full replacement of –CN with –bpy. Before being put into solution with Cu/Ag, the formed dicyano film/GCE was washed with MeCN and dried. After 20 min of incorporation in the metal ion precursor solution, the electrode was cleaned with MeCN and transferred into a nitrogen-saturated 0.1 M TBAPF6/MeCN to run electrodeposition under −1.7 V vs. Ag for 30 s. Aggregation of particles were observed with longer electrodeposition time.

Detailed product analysis and CVs are presented in Supporting Information.