Due to fossil fuel combustion, the concentration of CO₂ in the atmosphere has recently reached 400 ppm and is expected to rise continuously, which can lead to numerous environmental problems such as global warming, rising sea levels, and more erratic weather patterns [1, 2]. Therefore, CO₂ reduction reaction (CORR) has become a research focus to control the atmospheric concentration of CO₂. Various CO₂ reduction approaches, including electrochemical [3], biochemical [4], photochemical [5, 6], and thermochemical [7] methods, have been proposed and explored in the past decades. Among them, electrocatalysis is a very promising method as it can be operated under ambient conditions.

During the past decades, heterogeneous electrochemical CO₂ reduction has been explicitly investigated, and various catalysts have been investigated to generate products such as carbon monoxide, formic acid, methane, and methanol [8-10]. Recently, Kanan et al. [11, 12] suggested that the activity of metal catalysts obtained by the reduction of metal oxides is usually higher than that in other methods. Xie et al. [13] investigated cobalt, associated with loosely bonded d electrons and high electrical conductivity, as a promising catalyst for the CORR. They synthesized two-dimensional cobalt/cobalt oxide hybrid by using a solvothermal method with high electrocatalytic activity for the CORR, and the main product was formate. It was proposed that the metal/metal oxide interface plays a key role in catalyzing the CORR.

In this work, we develop a green and facile synthetic strategy for unique nanostructured cobalt by using nontoxic capping and reductant agents. The electrochemistry of the CORR on the resultant Co is then explicitly investigated. The flower-like Co, which favors the exposure of the active sites and the accessibility of the species, is found to show superior electrocatalytic activity than the bulk Co. Nuclear magnetic resonance (NMR) results indicate that the main product is formate, and the yield is highly dependent on the applied potential.

Flower-like Co: First, CoCl₂·6H₂O powder was dissolved in a liquid mixture of 21 mL deionized water and 9.0 mL n-butylamine to yield a concentration of 50 mmol/L. Second, NaH₂PO₂·H₂O was added as a reductant with a concentration of 0.40 mol/L. After vigorous stirring, the mixture was transferred to a Teflon cup in a stainless steel-lined autoclave for conditioning at 140 ℃ for 24 h. Then, the system was allowed to naturally cool down to room temperature, and the black fluffy solid product was collected by centrifuging the mixture. The product was rinsed with deionized water and ethanol in sequence, and then dried in a vacuum oven at 60 ℃ for 4 h.

Bulk Co and Co(OH)₂: The preparation of bulk Co and Co(OH)₂ followed the abovementioned procedure, except that n-butylamine was not used for bulk Co and the hydrothermal treatment time was shortened to 3 h for Co(OH)₂.

X-ray diffraction (XRD) measurements were carried out on a TD-3500 X (Tongda Technology) diffractometer with a Cu K_α radiation source operated at 40 keV. The patterns were collected at a scan rate of 0.05°/s in the 2θ range 10° to 80°. Scanning electron microscopy (SEM) observations were carried out on a Nova NanoSEM 430 scanning electron microscope.

The electrochemical behavior was characterized by cyclic voltammetry and linear sweep voltammetry using a three-electrode cell on an electrochemical workstation (CHI660E), at room temperature (25 ℃). A gold wire and a saturated calomel electrode (SCE) were used as the counter and reference electrodes, respectively. The working electrode was a glassy carbon disk (5.0 mm in diameter, PINE) covered with a thin layer of Nafion-impregnated catalyst. Typically, the thin-film electrode was prepared as follows: 10 mg of the catalyst was dispersed in 1.0 mL Nafion/ethanol (0.84 wt% Nafion) by sonication for 120 min. Then, 10 μL of the dispersion was transferred onto the glassy carbon disk by using a pipette, yielding a loading of 0.50 mg/cm².

The electrolyte solution, 0.10 mol/L of KHCO₃, was bubbled with either Ar (99.999%) or CO₂ (99.999%) for 30 min before the test. The cyclic voltammograms (CVs) were collected at 20 mV/s in the potential range between -1.25 and -0.4 V (vs. SCE) in Ar-and CO₂-saturated electrolyte solution. The linear sweep voltammograms (LSV) were collected by scanning the potential from -0.6 down to -1.2 V in Ar-or CO₂-saturated electrolyte solution. All the potentials are referenced to SCE in this study.

For controlled-potential electrolysis, 50 μL of the dispersion was loaded onto a gas-diffusion carbon paper with an area of 1 cm² to form a working electrode. The counter and reference electrodes were gold wire and SCE, respectively. The electrolysis was performed in an H-type cell with a piece of Nafion^®117 cation-exchange membrane (H⁺ form) as a separator. 20 mL of KHCO₃ solution (0.10 mol/L) was introduced into the cathode chamber, and CO₂ was continuously bubbled throughout the experiment to ensure saturation. The electroreduction of CO₂ was performed at a constant potential of -0.75, -0.85, -0.95, -1.05, and -1.15 V. The electrolysis was prolonged for 10 h to enrich the reduction product.

The liquid products were quantified by ¹H NMR (Bruker AVANCE Ⅲ 400) spectroscopy, for which 0.25 mL electrolyte was mixed with 0.50 mL dimethyl sulfoxide-D6 and 5.0 mg methylbenzene was added as an internal standard. The faradaic efficiency of formate (η_formate) was calculated from the total amount of charge Q (C) passed through the material and the total amount of formate produced n_formate (mol).

where I (A) is the reduction current at a specific applied potential and t (s) is the time.

Fig. 1(a) reveals that cobalt hydroxide has a hexagonal sheet structure. After reduction, serious aggregation of Co, namely bulk Co, is observed in the absence of n-butylamine (see Fig. 1(b)); in comparison, a flower-like Co covered with curved nanosheets, which inherits the morphology features of Co(OH)₂, is synthesized by adding n-butylamine (see Fig. 1(c)). The difference in the morphology can be attributed to the chemical adsorption of n-butylamine on the facets of Co(OH)₂, thus preventing the aggregation of Co. The unique flower-like morphology of Co favors both the exposure of the active sites and the accessibility of the active species, thus facilitating the electrocatalysis (vide infra). The corresponding XRD patterns are shown in Fig. 2. As seen in Fig. 2(1), Co(OH)₂ shows the characteristic peaks of β-Co(OH)₂ (No. 45-0031). Both bulk Co and flower-like Co basically show the same patterns with the diffraction peaks at 41.68°, 44.76°, 47.57°, 62.73°, and 75.94° (see Fig. 2(2, 3)), indicating that they are bulk metal of the hexagonal close-packed (hcp) phase (No. 05-0727).

The CO₂ reduction activity of flower-like Co, bulk Co, and Co(OH)₂ is evaluated by collecting the LSVs. As seen in Fig. 3(a), for the three catalysts, a broad cathodic reduction wave is observed between -1.2 and -0.6 V, which is then followed by a more significant reduction wave below -1 V. The first cathodic wave is not seen in the Ar-saturated solution, intuitively indicating that this peak originates from the direct reduction of dissolved CO₂ molecules. To further investigate the origin of the wave, the peak current is plotted with respect to the scan rate, as seen in Fig. 3(b). The logarithm of the cathodic peak current shows a linear relationship with the scan rate, and the linear slope is close to 0.5, indicating that the cathodic peak is controlled by the CO₂ diffusion.

Among the three catalysts, the flower-like Co exhibits the most positive onset potential at -0.7 V and the highest reduction current than do the other two catalysts. This superior electrocatalytic activity should be attributed to the unique morphological features of the flower-like Co, as mentioned above. The current of the second reduction wave (< -1 V) increases with decreasing potential, which is much larger than that of the first reduction wave. This wave has been claimed to originate from the hydrogen evolution reaction (HER)[13]. However, this does not seem convincing because the background current is much smaller in Ar-saturated solution. Here, we propose that besides the hydrogen evolution reaction, this wave should also originate from both the direct reduction of HCO₃^- (vide infra) and reduction of CO₂ to different products.

Fig. 4 shows the CVs of flower-like cobalt in KHCO₃ (0.10 mol/L) in different potential windows. Fig. 4(a) shows that a negligible faradaic current response is seen in the range from -0.7 to -0.4 V (see black trace) under Ar atmosphere. The Pourbaix diagram reveals that Co remains in the divalent state and no charge transfer occurs in this potential range. Two reduction waves are seen in the negative-going scan when the lower limit potential is scanned to -1.25 V (green trace). The first wave, in the potential range from -1 to -0.85 V, can be attributed to the reduction of the divalent Co. Then, a dramatic increase in current is seen as the potential is further negatively scanned, and originates from both the HER and the direct reduction of HCO₃^-. In the positive-going scan, a remarkable anodic peak is observed at potentials greater than -1.1 V, which can be attributed to the electrochemical oxidation of metal Co. It is, thus, concluded that Co is in the metal state at potentials more negative than -0.85 V and is converted into oxide at more positive potentials, and this is understandable from the thermodynamics point of view. To make the transition clearer, the CV is collected in the potential range from -1 to -0.7 V (see red trace). One pair of cathodic/anodic waves corresponding to the electrochemical transition between the divalent state and the metal state can be clearly recognized. Finally, the CV in the potential range from -1.25 to -1 V (see blue trace) reveals that metal Co is active for the HER and the direct reduction of HCO₃^-.

Fig. 4(b) shows the CVs of CO₂-saturated KHCO₃ (0.10 mol/L) electrolytes. The black trace reveals negligible current in the potential range from -0.7 to -0.4 V. As the lower limit potential is scanned down to -1.25 V (see green trace), two reduction waves are seen in negative-going scan with the onset potential at -1 and -0.75 V, respectively. The two waves have been recognized in the LSVs (see Fig. 3), as discussed above. It is interesting to note that in the positive-going scan, the first wave is not seen, while the second wave remains unchanged. This finding indicates that in the potential range from -1 to -0.75 V, pure metal Co may not be inactive to the CORR, and divalent oxide plays a key role as the active sites. This analysis is further validated by the red trace. The first cathodic reduction wave is remarkably decreased when the upper limited potential is restricted to -0.7 V, at which much less divalent oxide is generated on the surface. Furthermore, the comparison of blue trace and green trace indicates that the upper limit potential has a negligible effect on the second cathodic reduction wave (< -1 V). This result is unsurprising as Co always remains in the metal state in this potential range. In the positive-going scan, a remarkable anodic wave above -0.7 V is seen in green trace, and the current is much larger than that in Fig. 4(a), suggesting that another electrochemical oxidation process occurs. It is proposed that the reduced product, generated in the negative-going scan, can be electro-oxidized at higher potentials in the positive-going scan on the Co catalyst. As such, this catalyst can be seen as a bifunctional one for the CORR.

To recognize the origin of the second reduction wave, the effect of KHCO₃ concentration on the LSVs was investigated, as shown in Fig. 5. It is seen that the cathodic current linearly increases with the concentration of the HCO₃^- ions. This result intuitively suggests that the reduction of HCO₃^- ions contributes to the second cathodic wave. Notably, the first wave gradually decreases as the concentration of KHCO₃ increases. This may be due to the fact that the solution viscosity increases in this course, and decreases the diffusion coefficient of CO₂ [14].

Controlled-potential electrolysis is conducted to recognize the product and evaluate the faradaic efficiency by using the flower-like Co catalyst. All experiments are performed in CO₂-saturated KHCO₃ (0.10 mol/L), and NMR is used to quantify the solution-phase products (see Fig. 6(a)). Formate is found to be the main product in the investigated potential range. Fig. 6(b) summarizes the yield of formate and the faradaic efficiency. Both the yield and the faradaic efficiency are strongly dependent on the given potential. The highest faradaic efficiency (63.4%) is achieved at -0.85 V, and the formate yield reaches at 136 mg/L. The formate yield decreases at more negative potentials, and this can be attributed to the serious concomitant reactions, including the reduction of H₂O and the reduction of CO₂ to generate other products.

In this work, a novel flower-like cobalt was synthesized through a facile hydrothermal method with the aid of n-butylamine. The morphological features facilitate both the exposure of the active sites and the mass transfer of the active species, favoring the electrocatalysis of the carbon dioxide reduction reaction. The electrocatalytic activity of flower-like cobalt is found to be 3.5 and 4 times higher than that of the Co(OH)₂ and bulk Co metal, respectively. As a result, high faradaic efficiency up to 63.4% is achieved and the yield of the product formate reaches 136 mg/L. Electrochemical analysis reveals that the divalent oxide is essential to constitute the active sites for the CO₂ reduction reaction. In addition, this catalyst is highly active for the oxidation of reduced product, thus acting as a bifunctional catalyst.