Conjugated polyimides modified self-supported carbon electrodes for electrochemical conversion of CO₂ to CO

Synthesis of conjugated polyimide (CPI) powder

The CPI powder was synthesized following the literature procedure^[44] with slight modification, and denoted as CPI-A. First, 1,2,4,5-Benzenetetracarboxylic anhydride (PMDA, 981.5 mg, 4.5 mmol) and H₂O (162 μL) were added into 1-methayl-2-pyrrolidinone (NMP, 20 mL) in an autoclave. Then, melamine (MA, 378.4 mg, 3 mmol) dispersed in a 20 mL solution of NMP was dropwise added into the mixture under stirring for 1 h at 25 °C. After reacting at 210 °C for 24 h in the sealed autoclave, the mixture stayed still for another 24 h. After washing with NMP, N,N-dimethyl formamide (DMF), hot water, and ethanol to remove any impurities, the yielding dark brown solids were dried at 100 °C for 12 h in a vacuum oven. The obtained products were directly used without any further purification for characterization and subsequent experiments.

Synthesis of self-supported CPI/CP

The CPI/CP was fabricated following a procedure similar to the synthesis of CPI-A. At first, the CP (1 cm × 2 cm) was rinsed in concentrated HCl (12 M) solution for 10 min to remove surface impurities and enhance the hydrophilicity, which benefits the in-situ growth of CPI. After washing with deionized water, the CP was dried at 60 °C for 8 h in a vacuum oven. PMDA (981.5 mg, 4.5 mmol) and H₂O (162 μL) were first added into 20 mL NMP in an autoclave. Then, MA (378.4 mg, 3 mmol) dispersed in a 20 mL solution of NMP was dropwise added into the mixture under stirring for 1 h at 25 °C. Subsequently, a piece of pre-treated CP (1 cm × 1.5 cm) was placed in the autoclave. After reacting at 210 °C for 24 h in the sealed autoclave, the mixture stayed still for another 24 h. After washing with deionized water to get rid of the solvents, the CPI-coated CP was air-dried in the fume hood overnight. The resulting CPI/CP was directly used without any further purification.

Synthesis of CPI powder in sealed tubes

The CPI powder synthesis was performed in a vacuum-sealed Pyrex tube, denoted as CPI-S. Initially, 981.5 mg (4.5 mmol) of PMDA and 378.4 mg (3 mmol) of MA were dissolved in a 20 mL NMP solution within a 100 mL Pyrex tube. The tube was subsequently frozen using liquid nitrogen, evacuated, and flame-sealed. The reaction was then conducted at 120 °C for 120 h. The resulting solids were sequentially washed with NMP, DMF, hot water, and ethanol to remove impurities. The dark brown solid product was then dried at 100 °C for 12 h in a vacuum oven. The resulting CPI-S powder was directly used for characterization and subsequent experiments without further purification steps.

The additional detailed experimental data, such as reagents, characterizations, and electrochemical measurements, are provided in the Supplementary Materials.

Synthesis and characterization of catalysts

Scheme 1 illustrates the systematic procedure for preparing self-supported CPIs on CP substrate (CPI/CP). The CPI/CP composite was synthesized via solvothermal synthesis of CPI with PMDA and MA at 210 °C with pre-treated CPs additionally placed in the autoclave. Upon visual inspection of the fabricated CPI-CP, the surface color transitioned from dark grey to a bit yellowish, indicating the successful CPI growth on the CP surface. For a comprehensive investigation of CPI materials, CPI powders were also prepared via solvothermal condition in an autoclave (CPI-A) and a vacuumed sealed tube (CPI-S) without any heterogeneous substrate. Subsequently, the CPI-A and CPI-S powders were introduced onto the CP with the aid of Nafion solutions. All the three fabricated working electrodes were directly used in the following electrolysis after activation without any other operations.

Scheme 1. The synthetic diagram of in-situ growth of CPI nanosheets on CP thermodynamically.

The surface morphologies of these materials were investigated using scanning electron microscopy (SEM), atomic force microscopy (AFM), and transmission electron microscopy (TEM). SEM images clearly depicted the formation of CPI microspheres with diameters ranging from 2-10 µm, adhering to the fibers of CP [Figure 1A]. Additionally, as illustrated in Figure 1B, regular microspheres were constructed from CPI nanosheets, exhibiting solid structures with irregular stacking. Transverse surface and partially enlarged images are presented as Figure 1C and D, revealing CPI nanosheets stacked randomly with rough boundaries. SEM images labeled CPI-A and CPI-S are presented Figure 1E and F, respectively. A mixture of microspheres and microsheets is observed in the CPI-A sample, indicating shape transformation during the solvothermal process. Conversely, microsheets with smooth boundaries were the predominant morphology observed in the CPI-S sample. Moreover, TEM images of CPI dispersed ultrasonically in CPI-A [Figure 1G] and exfoliated from CPI/CP [Figure 1H] reveal the 2D nanoscale structure of CPI sheets, potentially enhancing electrical conductivity^[45]. AFM images also revealed the micrometer-scale dimensions of CP microspheres [Figure 1I and J]. As shown in Figure 1K, energy dispersive X-ray spectroscopy (EDS) mapping images confirmed the even distribution of each element, indicating the construction of CPI through covalent bonding of organic substrates.

Figure 1. SEM images of (A-D) CPI/CP, (E) CPI-A, and (F) CPI-S. TEM images of (G) CPI-A and (H) CPI/CP. (I) AFM measurements of CPI/CP. (J) Corresponding height curves for the selective areas in (I). (K) EDS mapping image of CPI/CP.

To obtain more detailed information on the structure and composition of CPI materials, additional characterization techniques were employed. In the Fourier transform infrared spectroscopy (FT-IR) spectrum comparison, the characteristic peaks around 1,362, 1,785, and 3,400 cm^-1 corresponded to the stretching vibrations of C-N-C, C=O, and N-H bonds, respectively [Figure 2A]^[46]. Noticeable N-H stretching bands were observed in the spectra of CPI-A and CPI-S, suggesting the presence of MA or its oligomers in these structures. Furthermore, the ultraviolet-visible (UV-Vis) diffuse reflectance spectra (DRS) showed that CPI-A and CPI-S shared similar absorption peaks with PMDA, albeit with a slight red shift, whereas CPI/CP exhibited a broad absorption spectrum [Figure 2B]. This confirmed the enhanced conjugated structure of CPI/CP, potentially facilitating rapid electron transport. Moreover, the X-ray diffraction (XRD) patterns of these materials revealed a more ordered structure for CPI/CP, characterized by fewer peaks between 10°-35° compared to CPI-A and CPI-S [Supplementary Figure 1]^[47]. Additionally, the peaks at 167.29, 157.02, 137.34, and 120.08 ppm in the cross-polarization magic angle spinning nuclear magnetic resonance (CP MAS NMR) spectrum can be attributed to hydrogen bonded benzylic, imide bonded benzylic, carbonyl, and triazine carbons in the CPI structure [Figure 2C]. Meanwhile, the X-ray photoelectron spectroscopy (XPS) demonstrated the absence of metal elements within the CPI material, consistent with the findings of EDS mapping^[46]. In the full spectrum, the peaks centered at 286.6, 399.4, and 531.0 eV were attributed to C 1s, N 1s, and O 1s, respectively [Figure 2D]. In the C 1s spectrum, the 284.8 eV represents graphite carbon, inferring the conjugated structure of CPI [Figure 2E]. Additionally, the 286.3 and 288.4 eV were mainly attributed to the generated imide and intrinsic triazine groups in the CPI. In N 1s spectrum, the peaks at 398.6 and 399.7 eV can be denoted as the triazine N and imide N, respectively, further certifying the successful synthesis of CPI in a bonding manner [Figure 2F]. While in the O 1s spectrum, apart from the peak at 531.4 eV for carbonyl O in the imide group, there was also a peak at 532.8 eV representing the adsorbed O [Figure 2G]. The signals in XPS collectively provide compelling evidence for the formation of CPI from PMDA and MA. Furthermore, thermogravimetric analysis (TGA) and corresponding derivative thermogravimetry (DTG) of CPI/CP revealed that the majority of CPI decomposed at 395 °C, confirming its relatively higher stability owing to the conjugated structure [Supplementary Figure 2]^[46,47]. To elucidate the specific surface areas and pore size distributions of CPI materials, N₂ adsorption-desorption tests were carried out for CPI-S and CPI/CP. A reversible type IV isotherm with an H4 hysteresis loop was presented by CPI-S, suggesting its mesoporous nature with a layered structure [Figure 2H]. The surface area of CPI-S was calculated to be 26.41 m²/g using the Brunauer-Emmett-Teller (BET) model and the pore size distribution was calculated using the Barrett-Joyner-Hallender method, of which the peak value was 3.84 nm. These mesoporous structures are generated by the regular arrangement of CPI layers. In contrast, CPI/CP exhibits a type IV isotherm with an H1 hysteresis loop, indicative of a spherical cluster morphology with a calculated surface area of 14.48 m²/g [Figure 2I]. The moderated surface area results from the random arrangement of CPI layers, forming microspheres that provide a reduced contact surface area, enhancing selectivity in production. The pore size distribution presented a wide range of 10-25 nm as mesoporous.

Figure 2. (A) FT-IR and (B) solid-state UV-Vis DRS spectra of CPI/CP, CPI-A, CPI-S and raw materials. (C) CP MAS NMR spectrum of CPI exfoliated from CPI/CP. (D) XPS spectra of CPI/CP. (E) C 1s, (F) N 1s, and (G) O 1s XPS spectra of CPI/CP. N₂ adsorption-desorption isotherms and pore size distribution of (H) CPI-S and (I) CPI/CP.

Electrochemical CO₂RR performance

Electrolysis was performed in a H-cell equipped with three electrodes: the fabricated working electrode, a Pt foil counter electrode, and an Ag/AgCl reference electrode. The catalytic activities of CO₂RR were initially assessed by comparing linear sweep voltammetry (LSV) measurements in an Ar/CO₂ saturated 0.5 KHCO₃ electrolyte. The on-site potential of CPI/CP in a CO₂ atmosphere is lower than that in an Ar atmosphere, resulting in a clear reductive potential window between -0.4 and -1.1 eV [Figure 3A]. While the CPI-A and CPI-S provide higher current density in Ar than in CO₂, which may be due to the inactive catalytic performance for the conversion of CO₂ molecules and the prohibition effect of HER from adhesive CO₂[Supplementary Figure 3]. Compared to CPI-A and CPI-S, CPI/CP exhibited significant suppression of HER and a remarkable enhancement of CO₂RR despite a slight decrease in current density. Subsequently, potentiostatic tests of CO₂RR were conducted at five potentials (-0.50, -0.55, -0.60, -0.65, and -0.70 V vs. RHE) over a 1-h period at each potential with constant CO₂ gas input [Supplementary Figure 4]. With potentials lower than -0.6 V vs. RHE, the current densities of electrolysis catalyzed by CPI/CP were generally lower than those of CPI-A and CPI-S. Notably, at -0.65 V vs. RHE, where the highest catalytic activities exhibited, the CPI/CP presented average current density at 2.8 mA cm^-2, while at 3.2 mA cm^-2 for those catalyzed by CPI-A and CPI-S. The output gases and electrolyte liquids were collected for further determination of products. Aside from CO and H₂ detected by specific gas chromatography equipped with Flame Ionization Detector (FID) and Thermal Conductivity Detector (TCD), no other obvious signals indicating the generation of alternative carbon products were observed in any of the collected samples [Supplementary Figure 5]. Furthermore, no liquid products were detected by the ¹H nuclear magnetic resonance (NMR) of the electrolytes using the Water Eliminated Fourier Transform (WEFT) technique [Supplementary Figure 6]. At each potential, the electrolysis was conducted three times independently under identical conditions, and the corresponding FEs and yield rates mentioned below refer to the average values of the calculated results. Throughout all the potentiostatic tests, CPI/CP consistently exhibited exceptional catalytic performance compared to the others at each examined potential. As depicted in Figure 3B, at -0.5 Vvs. RHE, CPI/CP demonstrated the highest FE_CO of 54.0%, while CPI-A exhibited a moderated FE_CO at 19.6% and CPI-S showed a negligible result. As the potential shifted to the negative, the FE_CO reached a maximum of 88.6% at -0.65 V vs. RHE under the catalysis of CPI/CP, followed by a decrease. Meanwhile, the FE_CO values of CPI-A and CPI-S followed a similar trend, reaching their highest values at -0.65 Vvs. RHE, with 59.2% and 29.6%, respectively. As shown in Figure 3C, the CO yield rate of CPI/CP reached 82.0 mmol g^-1 h^-1 at -0.65 V vs. RHE, which is 1.76 and 3.63 times higher than those of CPI-A and CPI-S. In contrast to the trend in FE_CO, the CO yield rates decreased dramatically at lower overpotentials, such as -0.55 and -0.50 V vs. RHE. This phenomenon indicates that the differences between the partial currents of CO and H₂ are relatively lower at low overpotentials than those of higher ones, in consistency with potential window discovered in LSV spectrum. Based on the characterization of CPI materials, the robust electrocatalytic activity and production selectivity of CPI/CP can be attributed to their extensive conjugative property and random stacking of CPI nanosheets, which balance electron transfer throughout the organic conjugated structure and CO₂ adsorption on the imide/triazine nitrogen containing sites^[48]. Despite the regularly arranged CPI nanosheets in CPI-S possessing a layer-by-layer nanostructure and an enlarged surface area with concentrated pore size distribution, the dominant HER process in the aqueous solution still significantly influences the CO₂RR.

Figure 3. Electrochemical CO₂RR performance of CPI/CP, CPI-A, CPI-S. (A) LSV curves in a CO₂-saturated 0.5 M KHCO₃ electrolyte. (B) FE_CO. (C) CO yield rates. (D) FE_CO with various counter electrodes. (E) FE_CO with various electrolysis media. (F) C_dl. (G) Stability test of CPI/CP at -0.65 V vs. RHE for 24 h. (H) FE_CO comparison of non-metal electrocatalysts.

Subsequent experiments were conducted using various counter electrodes to demonstrate the active role of Pt in the electrochemical CO₂RR process^[49]. As depicted in Figure 3D, neither the blank CP nor the graphite counter electrodes provided significant enhancement for CO production, likely due to their inherent inertness to the oxygen evolution reaction, which offers minimal promotion of the reduction overpotential [Supplementary Figure 7]. Furthermore, the influence of electrolyte pH on the reaction was examined in controlled tests [Supplementary Figure 8]. With equal molar of potassium ions, K₂SO₄ (0.25 M, pH = 7) and KHSO₄ (0.5 M, pH = 1.3) were selected as neutral and acidic electrolytes, respectively, because the alkaline conditions were neutralized by the CO₂ that was bubbled through the electrolyte. The FE and CO yield rate after 1 h of electrolysis are presented in Figure 3E. It is evident that the mild alkalinity of KHCO₃(0.5 M, pH = 8.2) is conducive to the reaction, whereas the increased hydrogen evolution in neutral and acidic electrolytes impedes the electrochemical conversion of CO₂, leading to reduced CO production. Besides, the pH value of KHCO₃ electrolyte remained the same after prolonged electrolysis, indicating the carbon source of product CO is the inlet CO₂ gas. Moreover, the electrochemical surface area (ECSA) of CPI electrocatalysts was determined by measuring the double-layer capacitance (C_dl) in the non-Faraday reaction region of the cyclic voltammetry (CV) curve at various scan rates, reflecting the quantity of active surface sites [Supplementary Figure 9]. After the data collection and linear correlation, the C_dl values for the CPI-S, CPI-A, and CPI/CP are 1.641, 0.543, and 0.766 mF cm^-2, respectively [Figure 3F]. Originating from the solvothermal method using autoclaves, CPI-A and CPI/CP exhibited C_dl values with minor difference under electrolytic circumstances. CPI-S obtained the highest C_dl value, indicating its excellent electron transportation ability. However, despite the higher C_dl value, larger ECSA, and more exposed active sites of CPI-S, these characteristics did not confer any advantages in selective electrocatalytic CO₂RR, as the inevitable HER process was improved with these conditions in aqueous electrolysis. Moreover, the CPI/CP electrocatalyst exhibited considerable durability and stability during the electrolysis at -0.65 V vs. RHE for continuous 24 h [Figure 3G]. Although the current density has dropped by approximately 40%, the FE_CO has not decreased much, from 93.1% to 78.3%. The surface morphology of CPI/CP was also examined after long-term electrocatalysis for CO₂RR. The enhanced electrocatalytic performance of CPI/CP compared to those of carbon material-based catalysts can be concluded in Figure 3H and Supplementary Table 1. Considering its low cost and pyrolysis-free synthetic route, this stable and active electrocatalyst has a good application prospect in the conversion of CO₂. The electrochemical assessments of CPI electrocatalysts elucidate their role in CO₂RR. While CPI-S exhibits superior electron transport, it does not confer selectivity advantages, highlighting the dominance of the HER process. Conversely, CPI/CP demonstrates durability and stability during prolonged electrolysis, indicating their potential for sustainable electrochemical CO₂ conversion.

To further assess the stability of polymer catalysts, we characterized catalysts after extended electrolysis and compared them with unexposed catalysts. The SEM image shows that the microspheres on the CP electrode surface almost disappeared compared to the fresh fabricated and activated ones [Figure 4A]. Furthermore, contact angle measurements of water droplets on CPI/CP before and after long-term electrolysis in aqueous solution were conducted, yielding values of 122.1° and 79.0°, respectively, indicating a change in hydrophobicity during the electrocatalytic process [Figure 4B]. Despite some catalyst detachment during the process, the TEM image of CPI nanosheets was still observable post-reaction, suggesting that the structural integrity of the catalyst was largely maintained [Supplementary Figure 10]. Additionally, the charge transfer resistance (R_ct) of CPI/CP was examined using electrochemical impedance spectroscopy (EIS) [Figure 4C]. Nyquist plots indicate that CPI/CP is controlled by both charge transfer and substance diffusion during electrochemical CO₂RR processes, where electrochemical polarization coexists with concentration polarization. After the electrolysis, the R_ct of CPI/CP reduced from 11.12 to 2.76, inferring swift electron transport on the catalyst compared to that before electrolysis. The increased slope (from 2.88 to 3.11) at the low frequency zone indicates the faster mass transport for the catalyst after the electrolysis. Considering the catalytic performance and the characterization of the CPI/CP before and after electrolysis, such a transformation of the catalyst occurring can be ascribed to the disappearance of the mesosphere morphology. Such rearrangement of the catalyst surface facilitates the electron transfer, mass transport, and hydrophilicity of the surface, which attenuates the selectivity for CO₂ conversion by emerging HER.

Figure 4. (A) SEM images of CPI/CP and (B) contact angle of water droplet on CPI/CP before (up) and after (down) long-term electrolysis. (C) Nyquist plots for CPI/CP before and after electrolysis. (D) Free energy diagram of CPI unit for CO₂RR.

Theoretical calculations

Density Functional Theory (DFT) calculations were employed to gain insights into the catalytic mechanism of CPI structure. The free energy diagrams for the CO₂RR demonstrate that CO₂ molecules are easily adsorbed onto the catalyst surface, a process that is enhanced by the activated CPI structure [Figure 4D]. The activation of adsorbed CO₂ to form *COOH is identified as the potential rate-determining step on CPI, with a relatively low energy barrier that implies fast reaction dynamics for CPI unit. Furthermore, the lower energy required to generate the intermediate *CO and the final product CO underscores the thermodynamic favorability of CO₂RR on CPI structure. The DFT analysis also suggests that the active sites on CPI play a crucial role in lowering the activation energy for CO₂ conversion, thereby facilitating the CO₂RR process. This enhanced activation capability, coupled with the thermodynamic advantage, indicates that CPI/CP can efficiently catalyze the conversion of CO₂ to CO, leading to improved CO₂RR performance. The results from DFT calculations thus provide a molecular-level understanding of the catalyst’s activity and its ability to promote the in-situ generation of CO, which is crucial for the development of more efficient CO₂RR catalysts.