Introduction

The renewable energy-driven electrochemical carbon dioxide reduction reaction (CO2RR) to valuable chemical feedstocks provides a sustainable opportunity to reduce greenhouse gas and store chemical energy1,2,3,4,5,6. Among potential products of electrochemical CO2RR, carbon monoxide (CO) and formate (HCOOH) have been routinely achieved7,8, with Faradaic efficiencies (FEs) of >95% and partial current densities of well above 100 mA cm−2. Several C1 and C2 deep-reduction (>2e transfer) products9,10,11, including methane (CH4), ethylene (C2H4), and ethanol (CH3CH2OH), have also been reported with FEs over 50%. However, a C3 product, n-propanol (aka. n-PrOH), despite its large energy content and high market price1,5, has only been exhibited with limited selectivity (FEs < 10%) and low activity in CO2RR12,13,14,15. The low selectivity of producing n-propanol has been attributed to the challenges of both creating electrocatalytic sites for stabilizing key C2 intermediates (*C2) to form n-propanol1,16,17, and the complicated coupling mechanisms between active C1 and C2 species, which have previously been suggested with two different modes as CO–CH2CHO15,18,19 and CO–OCCO16,20,21, respectively. OCCOCO* is the most suggested *C3 intermediate in CO2RR (widely used in theoretical calculations), not only toward n-propanol, but also other products such as acetone22. Another mechanism of n-propanol formation was also reported by using CO reaction with acetaldehyde23. Recently, the electrochemical CO2RR to isopropanol was also reported using carbonized copper metal organic framework-derived electrodes, in which a high reaction temperature (400‒800 °C) was required during the electrocatalysis24.

Ion vacancies have been demonstrated as a category of catalytic sites for enhancing CO2 activation and *C1 adsorption to generate C≤2 products25,26,27. For instance, copper oxide with surface oxygen vacancies created by electrochemical reduction showed a ~63% FE of converting CO2 into C2H427. Sargent and coworkers reported that the modification of Cu2S core with Cu surface vacancies led to FEs of 8 ± 0.7% for n-propanol26. Nonetheless, the negligible selectivity of C3 products in those previous works suggests the insufficiency of stabilizing *C2 intermediates and further promoting C1–C2 coupling to generate C3+ products. It is reasonable to hypothesize that an effective catalytic site for C3+ production should present a high density of negative charges, while conventional approaches of forming vacancies, such as high-temperature annealing25 and wet chemical reduction28, tend to form dispersed vacancies with limited hierarchical vacancy structures.

Copper sulfide (CuS) has previously been reported to electrochemically reduce CO2 toward major products of HCOOH (FE = ~60%)29 or CH4 (FE = ~73%)30, and its selectivity tuning was ascribed to the catalyst morphologies and the local electronic structures of copper around sulfur atoms. Nevertheless, as a two-dimensional (2D) transition metal dichalcogenides, hexagonal phase CuS possesses abundant surface sulfur atoms along (001) and (100) directions in its layered structure (Supplementary Fig. 1) and relatively low dissociation energy of Cu–S bonds31,32. These surface sulfur atoms can be removed to form sulfur vacancies, designated as CuSx (0 < x < 1)31,33. The controlled sulfur vacancies can lower the oxidation of copper between 0 and +2, which has been proved to promote C–C coupling34. Herein, we developed a double-sulfur vacancy (DSV) structure using a lithium electrochemical tuning approach, which allowed for stabilization of both CO* and a *C2 dimer (i.e., OCCO*), as well as subsequent coupling with a third *CO via CO–OCCO (Fig. 1a). The DSV-rich CuSx catalyst exhibited a much-improved FEn-PrOH of 15.4 ± 1% for n-propanol production at −1.05 V versus reversible hydrogen electrode (vs. RHE) in 0.1 M KHCO3, corresponding to one of the highest CO2RR selectivities to n-propanol reported in H-cells to date. Using a flow cell with 1 M KOH electrolyte, a high partial current density for n-propanol production (jn-PrOH) was obtained as 9.9 mA cm−2 at −0.85 V vs. RHE without ohmic correction.

Fig. 1: Schematics of CuSx with double sulfur vacancies for converting CO2 to n-propanol, and the corresponding calculations.
figure 1

a Mechanism of n-propanol formation on adjacent CuSx-DSV, showing the dimerization of CO–CO followed by CO–OCCO coupling. b, c Top views of the optimized OCCOCO* intermediate configurations on (100) surface of b CuSx-SSV and c CuSx-DSV. The arrows indicate the positions of sulfur vacancies. d Corresponding energy diagrams for CuSx-SSV (blue curve) and CuSx-DSV (red curve) at 0 V vs. RHE. The pink, yellow, gray, red spheres and red wireframe in a–c represent copper, sulfur, carbon, oxygen atoms, and water molecules, respectively.

Full size image

Results and discussion

DFT calculations of sulfur vacancies

Density functional theory (DFT) calculations were first conducted to investigate the effective vacancy structures as catalytic sites (Methods). CO*, OCCO*, and OCCOCO* are the critical *C1, *C2 and *C3 intermediates adsorbed on Cu atoms, respectively, and the energy barrier and energy change are used to assess the capability of C–C coupling. The adsorption energies of those intermediates were first calculated on hexagonal CuS(001) surfaces (Supplementary Fig. 2 and Supplementary Table 1). It is found that no matter zero, single, double, triple, or quadruple sulfur vacancies exist, the distance between two neighboring Cu atoms around sulfur vacancies is still relatively far (>3 Å), thus not capable of inducing coupling of multiple adsorbed CO* toward *C2 or *C3 intermediates.

Unlike (001) planes, the CuS(100) surface presents closer distance between neighboring Cu atoms, due to the shorter interlayered S–S distance (<3 Å) (Supplementary Fig. 1). As the rate-limiting steps for n-propanol formation16,20,21, the coupling energies of two CO* (to form an OCCO* dimer) and between the OCCO* dimer and a third CO* (to form an OCCOCO* trimer) were first calculated, respectively. Our results show that although a single sulfur vacancy (SSV) on CuS(100) can dimerize two CO* into OCCO* with an adsorption energy decrease of –0.24 eV (Supplementary Fig. 3 and Supplementary Table 2), the addition of a third CO* destabilizes the system with an increased energy of 0.1 eV (Fig. 1b, d). The reason why SSV cannot achieve the CO–OCCO coupling is the lack of adjacent sulfur vacancy centers, and thus cannot accommodate the strong electrostatic repulsion of three CO* that are all adsorbed at a single sulfur vacancy center.

In contrast, the DSV on CuS(100) allow to enrich negative charges near two adjacent co-planar vacancies (Supplementary Fig. 4), and can stabilize the OCCO* dimer with an adsorption energy decrease of –0.35 eV (Supplementary Fig. 5 and Supplementary Table 2). Moreover, it also enables further coupling with a third CO* to form a polyline OC–COCO trimer, with a decreased energy of –0.23 eV (Fig. 1c, d). Thus, the reason that DSV-rich CuSx(100) site can stabilize OCCO* and promote CO–OCCO coupling is attributed to the synergetic effect of two adjacent sulfur vacancies (i.e., DSVs), including the enriched negative charge density to adsorb three CO*, close distance of between adjacent Cu atoms (<3 Å) to couple CO–CO, and suitable space to relax the concentrated charges of OCCOCO* (Supplementary Table 3).

Lithium electrochemical tuning for sulfur vacancies

CuS nanocrystals were synthesized by hydrolysis of CuCl2·2H2O and thioacetamide (C2H5NS) at 60 °C (Methods)35. The as-made product presented a nanoflower morphology assembled by 2D nanosheets (Supplementary Fig. 6a). X-ray diffraction (XRD, Supplementary Fig. 6b) analysis revealed a hexagonal phase with a space group of P63/mmc; a = b = 3.7938 Å, c = 16.3410 Å; α = β = 90°, γ = 120° (JCPDS# 03-065-3588). The high-angle annular dark-field scanning transmission electron microscopy (HAADF-STEM) mapping further confirmed the existence and uniform distribution of Cu and S elements in CuS nanosheets (Supplementary Fig. 7). The lattice spacing of 2D nanosheets was 0.281 nm (Supplementary Fig. 8), corresponding to the (103) planes of hexagonal CuS, consistent with the XRD result.

Afterward, the sulfur vacancy-rich CuSx was synthesized by an electrochemical lithium tuning strategy (Fig. 2a), in which the as-made CuS and a Li foil were used as the cathode and anode materials, respectively (Methods). Based on the partial conversion reaction of CuS + Li+ + e− → CuSx + Li2S (0 < x < 1), Li+ was intercalated into the CuS lattice to form Li2S and sulfur vacancies. The sulfur vacancy concentration was tuned by the charge/discharge cycle numbers, including 1, 10, and 100 cycles, whereas the corresponding samples were labeled as CuSx-1-cycle, CuSx-10-cycle, and CuSx-100-cycle, respectively. The HAADF-STEM images and corresponding mapping of all the obtained CuSx showed a uniform distribution of both Cu and S elements (Supplementary Fig. 9−11). The atomic ratio of S/Cu in CuSx was determined by energy-dispersive spectroscopy (EDS), which reduced from original 1.21 to 0.92 (for cycle number of 1), 0.67 (for cycle number of 10), and 0.64 (for cycle number of 100), (Fig. 2b, Supplementary Table 4), indicating the continuous removal of sulfur atoms and formation of sulfur vacancies. The charge/discharge curves versus time showed the decreased discharge capacity with increased cycle numbers (Supplementary Fig. 12), suggesting more and more sulfur atoms were removed by lithium ions.

Fig. 2: Lithium electrochemical tuning method and structural characterizations.
figure 2

a Schematic diagram of lithium-ion battery assembled with CuS (cathode) and Li metal (anode). b Discharge capacity of CuS (blue dots) and the atomic ratio of S/Cu (red stars) with respect to the cycle number at a constant current of 0.044 mA·cm-2 in the voltage range of 0.01–3 V. Sulfur atoms in the CuS lattice were controllably taken away to form Li2S, resulting in CuSx-DSV. c Spherical aberration corrected HRTEM, d HAADF-STEM image, and e corresponding intensity profiles extracted from the blue line of CuSx-DSV in d. The typical lattice arrangement of hexagonal phase CuS was presented by the higher contrast of Cu and the lower contrast of sulfur atoms. The intensity of lower left sulfur atoms was around 20 units, almost half of S atoms (~ 40 unit), suggesting the absence of single sulfur atoms. The pink, yellow spheres, and yellow dashed circles indicate copper, sulfur atoms, and sulfur vacancies, respectively. f ESR spectra of CuS (blue curve) and CuSx-DSV (red curve).

Full size image

Structure characterizations

The HAADF-STEM images showed that the spacing of CuSx-10-cycle nanosheets was 0.283 nm (Fig. 2c), corresponding the CuS(103) facets. The (103) planes on the edge of CuSx-1-cycle and CuSx-100-cycle were also observed (Supplementary Fig. 13). Spherical aberration corrected HAADF-STEM was conducted to investigate the detailed structure of sulfur vacancies. The CuSx-10-cycle nanosheets exhibited the typical hexagonal phase, where Cu and S atoms were displayed with high and low contrasts, respectively (Fig. 2d). DSV sites were observed at the edge of the nanosheets (highlighted as yellow dashed circles in Fig. 2d). The extracted intensities of two adjacent single sulfur vacancies were about half of two sulfur atoms (Fig. 2e), confirming the existence of double vacancies28,36. Furthermore, the electron spin resonance (ESR) spectra of the as-prepared CuS nanosheets did not present any unpaired electrons (Fig. 2f). In contrast, for CuSx-DSV, a pair of opposite peaks were clearly observed with a signal at g = 2.003, suggesting the existence of unpaired electrons due to the DSV25,28.

X-ray photoelectron spectroscopy (XPS) was conducted to study the oxidation states of Cu element on the catalyst surfaces (Fig. 3a). For the CuS nanosheets, two main peaks around 931.9 and 951.8 eV were observed, corresponding to Cu 2p3/2 and Cu 2p1/2 peaks, respectively. The two peaks were further deconvoluted into four sub-peaks, assigned to Cu2+ 2p3/2 (932.1 eV), Cu2+ 2p3/2 (933.4 eV), Cu2+ 2p1/2 (952 eV) and Cu2+ 2p1/2 (954.1 eV), respectively26. The peaks located at 933.4 and 954.1 eV were attributed to slight surface oxidation of CuS in air35. For CuSx-DSV, a new pair of shoulder peaks located at 930.3 and 950.2 eV were observed, corresponding to Cu+ or Cu0 2p3/2/2p1/2 peaks12. The appearance of Cu+ was attributed to the partial reduction of Cu2+ in CuS. The peaks centered around 942 and 962 eV were ascribed to satellite peaks of Cu+ or Cu0, respectively12.

Fig. 3: Electronic and fine structural characterizations.
figure 3

a XPS spectra of CuS (lower panel) and CuSx-DSV (higher panel). b, c Normalized Cu K-edge XANES and the first derivative μ(E)/dE of CuS, CuSx-DSV, Cu foil, standard CuS, and Cu2S samples. d Fourier-transformed k2χ(k) of CuS (lower panel) and CuSx-DSV (upper panel). The insets in d are (100) facets on CuS without (lower inset) or with (upper inset) adjacent double sulfur vacancies. The black dashed cycles represent an adjacent sulfur vacancy pair. The decreased coordination number and scattering Cu–S path length (the extracted fitting results in Supplementary Table 4) suggest the absence of sulfur atoms around copper.

Full size image

X-ray absorption spectroscopy (XAS) was conducted to investigate the electronic structure of CuSx. X-ray Absorption Near-edge Fine Structure (XANES) spectroscopy of Cu K-edge in CuS nanosheets showed a pre-edge feature around 8978.2 eV (Fig. 3b), similar to that of standard crystalline CuS powder37. However, compared to standard Cu2S and CuS, CuSx-DSV displayed a different shape of the rising edge and the post edge between 8990 and 9020 eV (Fig. 3b), suggesting their different chemical valance states and local coordination environments. In addition, compared to standard crystalline Cu powder, CuSx-DSV did not present any similar Cu0 feature based on the absorption edge or the location of white lines. The first derivatives of the normalized μ(E) spectra, dμ(E)/dE, were further extracted to highlight the difference (Fig. 3c). The energy value of the first large peak, E0, was defined as the value of absorption threshold or the lowest energy state reached by the core electron excitations38. The E0 value of CuSx-DSV was 8981.0 eV, which was located between standard Cu2S (8980.4 eV) and CuS (8981.4 eV), further indicating the co-existence of Cu+ and Cu2+ species.

Compared to the CuS nanosheets, CuSx-DSV showed a decreased intensity in k-space (Supplementary Fig. 14), indicating its lower coordination number (CN) and relatively higher disorder39. Fourier-transformed k2χ(k) of Cu K-edge Extended X-ray Absorption Fine Structure (EXAFS) spectra were further conducted to quantitatively assess the CN and bond length information of these samples (Fig. 3d), which can be used to investigate the local chemical environment of copper around double sulfur vacancies40. The strong peak at R = 1.81 and small envelope peak around R = 2.91 were assigned to the Cu–S and Cu–Cu scattering paths (without phase shifts correction) in the first shell37. After fitting, compared with the CuS nanosheets, the CN of the first shell Cu–S path in CuSx-DSV decreased from 3 to 1.0 ± 0.2 (Supplementary Table 5), suggesting the formation of abundant sulfur vacancies around the Cu centers40. According to the coodination number of Cu–S path and the ratio of sulfur vacancies/total sulfur atoms, the concentration of double sulfur vacancies was calculated as 2.78%, similar to the value of 2.53% calculated in our theoretical model of DSVs (Supplementary Table 6). The length of Cu–Cu path decreased from 3.23 ± 0.03 Å to 2.78 ± 0.02 Å, ascribed to the closer local distance between Cu atoms due to sulfur vacancies. The results were consistent with the optimized CuSx-DSV model in our DFT calculations, where the average length of Cu–Cu around DSV sites was below 3 Å, substantially shorter than the theoretical value of 3.23 Å in hexagonal CuS crystal (Supplementary Table 3 and Supplementary Table 5).

Electrochemical CO2 reduction

Based on the aforementioned hypothesis that the DSV-rich CuSx can stabilize OCCO* intermediates and further enhance the CO–OCCO coupling to generate n-propanol, the electrocatalytic CO2RR was first conducted using CuSx-DSV as the cathode catalyst in a CO2-saturated 0.1 M KHCO3 aqueous solution under ambient conditions (Methods in the Supplementary Information). Compared with those of the CuS nanosheets, the CuSx-1-cycle and CuSx-100-cycle catalysts, the CuSx-DSV catalyst presented the largest total current density in the linear sweep voltammetry (LSV) curves (Fig. 4a), suggesting its highest electrocatalytic activity. The gas and liquid products were determined by gas chromatography (GC) and 1H nuclear magnetic resonance (1H-NMR), respectively. Both CuS and CuSx-DSV catalysts presented a major product selectivity of HCOOH in the potential window between –0.85 and –1.25 V vs. RHE (Fig. 4b, Supplementary Fig. 15, Supplementary Table 7, 8). Importantly, the 1H-NMR spectra of liquid products of CuSx-DSV showed a clear triplet at a chemical shift around 0.77 (Supplementary Fig. 16), corresponding to the methyl hydrogen of n-propanol20. The CuSx-DSV catalyst exhibited a clearly enhanced selectivity of n-propanol, with a peak FEn-PrOH of 15.4 ± 1% and a corresponding partial current density (jn-PrOH) of 3.1 ± 0.2 mA cm−2 at −1.05 V vs. RHE (Fig. 4c). This jn-PrOH value was enhanced for almost 10 times compared with that of the CuS catalyst without sulfur vacancies (0.32 ± 0.14 mA cm−2). In addition, the gas chromatography–mass spectrometry (GC–MS) spectra showed the fragment ions matched well with the standard database of n-propanol (Supplementary Fig. 17).

Fig. 4: Electrocatalytic CO2RR in H-cells and flow-cells.
figure 4

a Linear sweep voltammetry curves of CuS, CuSx-1-cycle, CuSx-DSV, and CuSx-100-cycle catalysts in CO2-saturated 0.1 M KHCO3 aqueous electrolyte with the scan rate at 50 mV s‒1. b CO2RR product distribution using CuSx-DSV catalysts in H-cells. c FE of n-propanol on the four catalysts at different applied potentials. d FEn-PrOH and the ratio FEn-PrOH/FEC1+C2+C3 of the four catalysts at −1.05 V vs. RHE. e Linear sweep voltammetry curves of CuS and CuSx-DSV catalysts in 1 M KOH aqueous electrolyte in flow cells. f Partial current densities of n-propanol versus potentials using CuSx-DSV catalyst in H-cells (blue curve) and flow-cells (red curve). The data in a–d and e–f were tested in H-cells and flow-cells, respectively. Error bars in b–d and f correspond to mean ± standard deviation from three measurements.

Full size image

To further confirm the n-propanol formation, 13CO2 isotope labeling experiment was also conducted, and the products were analyzed by 1H-NMR. Compared with the original triplet peak at 0.77 and 0.83 ppm with 12CO2 as feedstock, another two weak peaks were observed around 0.64 and 0.84 ppm when using 13CO2 as feedstock23, as well as the doublet at 8.08 and 8.57 ppm (Supplementary Fig. 18a), indicating the existence of –13CH3 group of n-propanol and –13COOH group on formate41. The chemical shift centered at 1.03 and 1.43 corresponded to –13CH2– group on n-propanol and –13CH3 group on ethanol, respectively (Supplementary Fig. 18b)23. The relative low selectivity to C2H5OH (FE ~ 4%) for CuSx-DSV catalyst was ascribed to the consumption of *C2 intermediates by further C1–C2 coupling to form n-propanol, similar to the reported electrochemical results14,42.

By contrast, the CuS catalyst showed negligible selectivity toward n-propanol (2.3 ± 0.9%, Supplementary Fig. 15), as well as other possible products including CO, CH4, C2H4, CH3OH, CH3COOH (all below 1%). The CuSx-1-cycle catalyst showed a higher FEn-PrOH than CuS, but lower than CuSx-DSV (Supplementary Fig. 19), which was ascribed to the lower concentrations of DSV sites resulted from insufficient conversion reaction with Li+. By tuning the cycle number, the selectivity of n-propanol (i.e., the percentage in all CO2RR products, C1 + C2 + C3) was tuned from 4% to 22% with a 5.5-fold increase (Fig. 4d). However, when more cycles of electrochemical lithium tuning were applied, i.e., 100 cycles, both the FEn-PrOH and FEn-PrOH/FE(C1+C2+C3) values dropped down, which was ascribed to the damage of the formed DSV sites (Supplementary Fig. 20).

In addition, after 10 h of continuous electrochemical CO2RR test at −1.05 V vs. RHE, the FEn-PrOH of CuSx-DSV was maintained at 12.1%, corresponding to ~ 78% retention (Supplementary Fig. 21), suggesting its excellent stability. After CO2RR, the CuSx-DSV catalyst still preserved a similar morphology as that before electrochemical measurement (Supplementary Fig. 22). The Cu K-edge XANES spectra (Supplementary Fig. 23a, b) showed that the position of the white line was 8981.8 eV, higher than the Cu foil of 8981.3 eV. The μ(E)/dE spectra showed its E0 (8980.9 eV) was downshifted about 0.1 eV, suggesting a slight reduction of copper. Fourier-transformed k2χ(k) (Supplementary Fig. 23c, d) showed that the relative intensity ratio between Cu–Cu and Cu–S paths were slightly increased compared to that before CO2RR, suggesting that more sulfur atoms were taken away during the CO2RR test. After fitting, the coordination numbers of Cu–Cu and Cu–S paths were 4.2 and 0.6, respectively (Supplementary Table 9). The double sulfur vacancies were calculated as 3.89%.

As HCOOH is the major product on all those CuS and CuSx catalysts, DFT calculations were further conducted to study the relationship between HCOOH, CO and n-propanol products. On Cu-based electrocatalysts, *OCHO and *HCOOH are the widely accepted intermediates to produce HCOOH43, while *COOH and CO* are the intermediates toward CO43. First, it was found that all CuS(100), CuSx-SSV(100), and CuSx-DSV(100) surface have relatively lower free energy barriers to form *HCOOH (Supplementary Figs. 24 and 26a) than CO* (Supplementary Figs. 25 and 26b), in accord with our electrochemical CO2RR results that HCOOH is the main product. Second, for the CO* formation pathway, the desorption of CO molecule is spontaneous on both CuS and CuSx-SSV surfaces, but nonspontaneous on CuSx-DSV surface (Supplementary Fig. 26b), which allows for subsequent CO–CO coupling. Our calculations indicate that the formation pathways of *HCOOH and CO* are competing with each other, and the existence of sulfur vacancies helps to enhance the probability of forming CO* on catalyst surface. Combined with our aforementioned calculations of CO–OCCO dimerization in Fig. 1d, it can be concluded that CuSx-DSV allowed to produce n-propanol via both CO* and OCCO* intermediates, much more significant than on either CuS or CuSx-SSV.

In order to promote the yield of n-propanol, the electrochemical CO2RR of the CuSx-DSV catalyst was further conducted in flow-cells with 1 M KOH aqueous electrolyte. Due to its greatly enhanced mass transport than that in H-cells10, the total current density was significantly increased from 20 to 390 mA cm−2 at the same potential of –1.05 V vs. RHE (Fig. 4e), indicating an almost 20-fold enhancement. A larger peak partial current density (jn-PrOH) of 9.9 ± 1 mA·cm−2 was achieved at an even lower applied potential of –0.85 V vs. RHE (Fig. 4f, Supplementary Fig. 27, Supplementary Table 10, 11), comparable to the best reported results for n-propanol production (Supplementary Fig. 28 and Supplementary Table 12)12,13,14,15,26,42,44,45. After 4500 s continuous electrolysis test at –0.85 V vs. RHE, the FEn-PrOH of CuSx-DSV was maintained at 3.1%, corresponding to about 80% performance retention (Supplementary Fig. 29).

Finally, to exclude the influence of poly(vinylidene fluoride) (PVDF) binder using in the lithium tuning process, PVDF was prepared as the electrode to perform CO2RR test, as a control. In either H-cells or flow cells, H2 was the major product with corresponding FE values over 85% in the whole range of voltage tested (Supplementary Fig. 30), while the total FEs of all carbon products were below 5 − 10%. These control experiments confirmed that PVDF had no catalytic selectivity to convert CO2 into n-propanol.

In summary, we have demonstrated that double sulfur vacancies formed in hexagonal CuS(100) planes can feature as active electrocatalytic centers for CO2RR, enabling the stabilization of CO* and OCCO* dimer, and further coupling CO–OCCO to form the key *C3 intermediate of n-propanol. Then, we developed a facile lithium electrochemical tuning approach to enable the formation of the high density of double sulfur vacancies and the decrease of Cu–Cu distance. The FE of n-propanol production was enhanced to 15.4% in H-cells, and the partial current density of n-propanol production was further increased to 9.9 mA cm−2 in flow cells, comparable to the best reported values for electrochemical CO2RR toward n-propanol. Our work suggests an attractive strategy of applying the lithium electrochemical tuning method to achieve a host of new structures with tailorable ion vacancies as active electrocatalytic sites.

Methods

Synthesis of CuS nanosheets

In a typical synthesis, 0.36 g of CuCl2·2H2O was dissolved in the mixed solution of 13 mL of ethanol and 26 mL of deionized (DI) water to form a green transparent solution A, and 0.18 g of thioacetamide (C2H5NS) was dissolved in the mixed solution of 10 mL of ethanol and 20 mL of DI water to form a colorless transparent solution B. Afterwards, the solution B was added dropwise into solution A with slow stirring to form a yellow suspension, which was immediately kept in 60 °C water bath for 24 h. Then, the mixture was cooled down to room temperature and washed with DI water and ethanol, each for three times, followed by drying in 60 °C vacuum oven overnight. The black powder of CuS was then obtained.

Synthesis of vacancy-rich CuSx

For introducing sulfur vacancies, the as-made CuS powder was mixed with poly(vinylidene fluoride) with a weight ratio of 7:1, and dispersed in N-methyl pyrrolidone (NMP) under ultrasound to form a black suspension. Afterwards, it was coated uniformly on the surface of a fresh copper foil and dried at 80 °C for about 4 h to remove the solvent. Next, a lithium-ion battery device, using the CuS as the cathode, a lithium foil as the anode, and 1 M LiPF6 in ethylene carbonate/dimethyl carbonate/ethyl methyl carbonate (1:1:1 vol%) as the electrolyte, was assembled in CR2016-type coin cells in a glovebox filled with pure argon gas (O2 < 1 ppm and H2O < 1 ppm). A current density of 0.044 mA cm−2 was applied in the voltage window of 0.01–3 V vs. Li+/Li. After 1, 10, and 100 cycle(s) of charge and discharge, the CuS cathode was dissembled, rinsed with water and then acetone for several hours to remove the electrolyte and Li2S by-product. The solvent of NMP was volatilized at 80 °C. The CuSx catalysts with different density of sulfur vacancies were then purified and obtained.