To solve the issue of excessive emission of CO2, many strategies have been proposed [1,2,3,4,5]. Electrochemical CO2 reduction reaction (CO2RR), driven by renewable electricity, could convert CO2 into value-added industrial feedstock [6,7,8,9,10,11]. However, the low selectivity, small current density, poor stability, etc., must be overcome [12,13,14,15]. As one of products in CO2RR, formate/formic acid (HCOO−/HCOOH) was important chemical intermediates in industrial process [16,17,18]. Many efforts had been devoted to improving selectivity of products. However, enhancement of current density was more urgent. Only with high current density of HCOO−/HCOOH production will CO2RR possess the potential for industrial application [1, 2, 8]. Compared with other metallic materials (e.g., Pd, In, Sn), Bi-based materials were one of the potential candidates for formic acid due to their low toxicity, earth- abundance and good selectivity [19,20,21,22,23,24]. Generally, Bi2S3 is prepared under complex conditions (e.g., multi-steps, usage of strong acids, high reaction temperature, long reaction time, etc.) (Table S1). Zhang et al. [25] reported the synthesis of urchin-like Bi2S3 using the solvothermal method holding at 240 °C for 24 h, the maximum Faradaic efficiency of HCOO− (FEHCOO-) was 84.0% at − 0.75 V versus reversible hydrogen electrode (RHE), but it suffered from low current density of HCOO− (JHCOO-) ≈ − 6 mA·cm−2. On the other hand, tripe-phase interfaces regulation could enhance mass transport and local concentration of reaction species [26,27,28,29]. Sun et al. studied the hydrophobicity of the interface of Cu nanoarray electrode for CO2RR, which could change the distribution of products [30].

Herein, we synthesized Bi2S3 nanorods under relatively mild conditions and tuned their hydropathy by adding polytetrafluoroethylene (PTFE) nanoparticles into catalyst ink. The optimized sample (BS-P1) achieved a FEHCOO- of 93.4% at  − 1.1 V, and the JHCOO- was 2.29 times larger than that of BS-P0 (without modification) at − 1.2 V. Also, the stability of Bi2S3 was extended from 12 to 20 h by tuning hydropathy. In flow cells, FEHCOO- > 90% at from − 0.5 to − 0.75 V with increased current density. Hydrophobicity of interfaces could accelerate CO2 mass transfer, thus enhancing CO2RR performance. The influence of temperature for CO2RR was also investigated.

Universally, the synthesis conditions of Bi2S3 were generally complex (e.g., strong acids and high temperatures, etc.) (Table S1). Then we prepared Bi2S3 under mild conditions. The Bi2S3 nanorods were synthesized from BiCl3 and TAA in ethanol at 150 °C (Figs. 1a and S1). First, we explored the ratio of BiCl3 to TAA, called as BS-5/2, BS-3/1, BS-10/3 (BS-P0) and BS-11/3, respectively. X-ray diffraction (XRD) was used to investigate the crystal structure for the catalysts. The main peaks of BS located at 24.9° and 28.6° were consistent with (130) and (211) planes in standard card of Bi2S3 (PDF No. 17-0320), respectively (Fig. 1b). Then PTFE nanoparticles were added in catalyst ink (see the detail in SI). According to the amounts of PTFE added, the catalysts were called as BS-Px (x = 0, 0.5, 1, 2). Peak at 18.1° was a characteristic peak of PTFE (PDF No. 47-2217) [31]. As the amount of added PTFE increased, characteristic peak of F was enhanced. Then the Bi 4f spectra of BS-P0 and BS-P1 are shown in Fig. 1c, and two peaks at 158.2 and 163.6 eV corresponded to Bi 4f7/2 and Bi 4f5/2, respectively. Also, peaks at 160.9 and 162.0 eV were related to Bi–S vibration [32]. Peaks at 158.6 and 164.3 eV would correspond to Bi–O vibration, which was mainly due to oxidation in air [33]. After adding PTFE, the peaks of FTFE were be detected, but the peaks intensity of Bi 4f spectrum become weaker (Fig. S2). X-ray photoelectron spectroscopy (XPS) is a surface detection technique. The addition of PTFE reduced the amount of Bi2S3 detectable on the electrode surface. XPS results showed that PTFE was successfully loaded on the surface of Bi2S3 nanorods.

Fig. 1
figure 1

a Schematic illustration for synthesis of Bi2S3; b XRD patterns; c Bi 4f XPS result; d FTIR of BS-P0; e Raman spectrum of BS-P0 (inset: Raman spectrum in range of 100–500 cm−1); f EPR of BS-P0; g HRTEM image of BS-P0 (inset: line intensity profile); h EDS elemental mappings images of BS-P1

Full size image

In the Fourier transform infrared (FTIR) spectra presented in Fig. 1d, the bands at 617 and 1114 cm−1 could be assigned to Bi–S vibration. The band at 1642 cm−1 was assigned to O–H stretching vibrations due to the adsorption of water on surface of BS-P0 [34]. The Raman spectrum of BS-P0 is provided in Fig. 1e, and Raman bands located at 230, 246, 300, 424, 608 and 962 cm−1 could be observed, matching with the Raman feature of Bi2S3 [35,36,37]. No carbon species were found in BS-P0. The isotropic electron paramagnetic resonance (EPR) signals at g = 2.003 in Fig. 1f were observed. They could be ascribed to the formation of sulfur vacancies, which may be good for CO2RR [7].

The crystal structure was revealed by high-resolution transmission electron microscope (HRTEM) [38]. In Fig. 1g, the crystalline phase with interatomic distance of 0.360 nm can be related to Bi2S3 (130) plane. The result of HRTEM was consistent with that of XRD patterns. Energy-dispersive X-ray spectroscopy (EDS) elemental mappings revealed that the width of the Bi2S3 nanorod was ~ 200 nm. The morphology of Bi2S3 did not change with the addition of PTFE. Also, EDS elemental mappings confirm a relatively uniform distribution of PTFE (Figs. 1h and S3).

The electrochemical performance of CO2RR was then investigated in H-type cells with 0.5 mol·L−1 KHCO3 [39]. The final products (included gas products and liquid products) were measured by gas chromatography and nuclear magnetic resonance (NMR) spectroscopy (Fig. S4) [40]. Considering that the optimal ratio of TAA and BiCl3 was 10/3 (Fig. 2a), the BS-P0 (10/3) was selected as the optimized sample. The current density in CO2 was larger than that in Ar, indicating the electrocatalytic activity toward CO2RR (Fig. S5) [41]. To compare their selectivity in CO2RR, the constant voltage electrolysis methods were used (Fig. S6). The final products were only HCOO−, CO and H2. Sum of FE was close to 100%, and no other products were detected. With adding PTFE nanoparticles (200 nm), JHCOO- increased. Excessive PTFE led to the reduction of JHCOO-. The maximum JHCOO- of BS-P1 could reach − 41.08 mA·cm−2 at − 1.2 V, but JHCOO- of BS-P0 only was − 17.45 mA·cm−2 (Fig. 2b). BS-P1 has the largest JHCOO- in the range from − 0.9 to − 1.2 V with FEHCOO- above 90%. Then FEHCOO- of BS-P1 could reach 93.4% at − 1.1 V. In comparison, for BS-P0, the maximum FEHCOO- was 84.64% at − 1.0 V (Fig. 2c). When the size of PTFE nanoparticles changed from 200 to 100 nm, FEHCOO- reduced (Figs. 2d and S7). Then FEHCOO- of BS-P1 still remained 88.5% at − 1.0 V after 20 h, while the FEHCOO- of BS-P0 was only 70.5% at − 1.0 V after 12 h (Fig. 2e). BS-P1 had a relatively good CO2RR performance in the H-type cells (Fig. S8 and Table S2) [42].

Fig. 2
figure 2

a FEHCOO- of BS with different amounts of TAA; b JHCOO- and c FEHCOO- of different samples; d comparison of FEHCOO- with different PTFE particle sizes; e stability at − 1.0 V of BS-P1 and BS-P0 in H-type cells

Full size image

Then we assembled the standard three-electrode flow cells with 1.0 mol·L−1 KOH as electrolyte [43]. A higher current density (200 mA·cm−2 at − 0.69 V) was obtained than that in H-type cells (Fig. 3a). FEHCOO- was larger than 93% in the potential range from − 0.50 to − 0.75 V. The maximum FEHCOO- reached 95.38% at − 0.6 V (Figs. 3b and S9).

Fig. 3
figure 3

a LSV curves of BS-P1 under different atmospheres and b FEs under different voltages of BS-P1 in standard three-electrode flow cells

Full size image

As shown in Figs. 4a and S10, adding 0.5% PTFE nanoparticles had a little effect on the electrochemical active surface areas (ECSA), which was evaluated by the electric double-layer capacitance (Cdl) method [44]. Compared to BS-P0 (581 μF·cm−2), ECSAs of BS-P1 and BS-P2 were significantly reduced to 526 and 503 μF·cm−2, respectively, because of hydropathy modulation of triple-phase interfaces by adding PTFE, which could retain gas and reduce contact between electrolyte and electrode surface [45]. On the other hand, the ECSA of BS-P2 reduced nearly 13.4% of BS-P0. It meant that less electrolyte contacted with the electrode surface, which was the reason of the reduced CO2 activity compared with BS-P1 (Fig. S5).

Fig. 4
figure 4

a ECSA measurement; b EIS results; c contact angle measurements on BS electrode with different PTFE loadings before and after CO2 electrolysis at − 1.0 V in H-type cells

Full size image

To investigate the intrinsic impedance of the catalyst, we tested the electrochemical impedance spectroscopy (EIS) without CO2. EIS results showed that adding PTFE increased charge transfer resistance due to the poor conductivity of PTFE [46]. However, PTFE could change the microenvironment of the triple-phase interfaces on the electrode, thus influencing the JHCOO- (Figs. 4b, 2b). A lower Tafel slope of 159.82 mV·dec−1 on BS-P1 than that on BS-P0 (209.69 mV·dec−1) suggested the accelerated reaction kinetics (Fig. S11) [47].

In Fig. 4c, the initial contact angles of BS-P0 (140.79°), BS-P0.5 (144.74°), BS-P1 (144.47°) and BS-P2 (143.82°) were similar. When BS with different PTFE amounts were electrolyzed at − 1.0 V for 1 h, the contact angle of BS-P0 decreased from 140.79° to 91.05° and the contact angle with 1% PTFE decreased from 144.47° to 130.36°. And with the increase in PTFE amounts, the contact angle after electrolysis also showed an increasing trend, which was larger than that of BS-P0. It meant that moderate hydrophobicity of triple-phase interfaces increased the CO2RR performance. The hydrophobicity of interfaces accelerated CO2 mass transfer and reduced the availability of electrolyte [48]. When 2% PTFE was added, the current density was dropped because of the lost balance between gas and liquid phases (Fig. S5).

Considering future extraterrestrial planet exploration, such as Mars, the performance of CO2RR at low temperature is worth studying. However, the research of CO2RR at low temperature was rare [49]. Here, as temperature decreased (25 to − 20 °C), the current density decreased (Fig. 5a). Compared with BS-P1 at 25 °C, the onset potential of BS-P1 at − 20 °C shifted to a more positive value, implying a slower CO2RR kinetics [50]. In Fig. 5b, the peak FEC1 of BS-P1 reached 99.42% at 2.8 V, while the competing hydrogen evolution reaction (HER) on BS-P1 was substantially suppressed. The inhibition of HER may be due to the inhibition of the Volmer reaction of HER under alkaline conditions [51]. FECO of BS-P1 was increased with temperature decreasing (Figs. S12, S13).

Fig. 5
figure 5

a LSV curves (inset: optical photograph of low-temperature control device) and b FE of BS-P1 at different temperatures in two-electrode flow cells

Full size image

In summary, Bi2S3 nanorods were synthesized under a mild condition. Then we added PTFE nanoparticles with different sizes (200 and 100 nm) into the catalyst ink to regulate the hydropathy of electrode surface. The JHCOO- and stability of Bi2S3 were improved more than two folds. Also, a high FEHCOO- was achieved (93.4% at − 1.2 V). Thus, this work not only provides a relatively mild route to synthesize Bi2S3, but also further improves the current density and stability of Bi2S3 through hydropathy modulation. By adding PTFE nanoparticles, the transition from hydrophobic to hydrophilic of electrode surface could be effectively slowed down during CO2RR, which was the reason of enhanced CO2RR performance. In addition, the influence of temperature in the CO2RR was studied, which could change the distribution of products. However, research on low-temperature CO2RR is rare. The mechanism of inhibition of HER at low temperature is explored underway.