Porous Indium Nanocrystals on Conductive Carbon Nanotube Networks for High-Performance CO₂-to-Formate Electrocatalytic Conversion

2.1 Structural and Characterization of mp-In@MWCNTs Hydride Nanocatalysts

The mp-In@MWCNTs were synthesized in a single-step process based on the previously reported methods.^[39] Indium salt precursor (InCl₃) was dissolved in an aqueous solution containing uniformly dispersed MWCNTs. After the mixed solution was heated to 60 °C for 10 min, NaBH₄ aqueous solution was added as a reducing agent. As shown in Figure 1a and Figure S1, Supporting Information, the mesoporous In nanocrystals anchored on MWCNTs were obtained. The mp-In consisted of porous nanoplates and possessed hierarchical porous structures (Figure 1b), which offered more unsaturated bonds and active sites. The average length of indium nanocrystals was 54.8 ± 15 nm, with an average width of 29.6 ± 10 nm (Figure S2, Supporting Information). As confirmed by the selected area electron diffraction (SAED) in Figure 1c, the diffraction rings of 0.23, 0.16, and 0.11 nm correspond to (110), (200), and (220) crystal planes along the [001] band axis, respectively, suggesting that mp-In was composed of metal In with a polycrystalline structure. The inner diffraction rings at 0.34 nm are attributed to the crystal plane (002) of MWCNTs. Additionally, as revealed by the high-resolution TEM (HRTEM) in Figure 1d, the lattice fringes with distances of 0.23 nm are assigned to the crystallographic plane (110) of tetragonal indium metal. The space of 0.34 nm corresponds to the plane (002) of MWCNTs, indicating the successful synthesis of mp-In@MWCNTs composite nanomaterials. The energy-dispersive X-ray spectroscopy (EDS) analysis in Figure S3, Supporting Information is consistent with the results for the composite material made of carbon nanotubes and indium nanocrystals. As can be seen from the high-angle annular dark field scanning TEM (HAADF-STEM) and elemental mapping (Figure 1e–h) analyses, the homogeneous dispersion of In elements (cyan) decorated on the nanocrystals and C elements (red) is distributed uniformly on the MWCNTs support.

The obtained mp-In@MWCNTs nanocatalysts were further characterized using powder X-ray diffraction (XRD). In the XRD spectrum, the main diffraction peaks at 32.9, 36.3, and 39.2 (blue curve in Figure 2a) were assigned to the (101), (002), and (110) crystal planes in tetragonal indium [PDF # 85-1409], respectively. The peaks of indium oxide and hydroxide were not found in the XRD spectrum, revealing that the porous nanomaterials mainly contain indium metal, rather than indium oxide or hydroxide. The diffraction peak confirmed the existence of MWCNTs in the complex at 26.2°, which corresponds to the (002) plane of carbon. The sharp diffraction peaks indicated a high crystallinity, consistent with the HRTEM image in Figure 1d. The surface chemical composition of mp-In@MWCNTs was investigated by X-ray photoelectron spectroscopy (XPS). The nanocatalysts were composed of indium without impurity elements as inferred from a full-scan XPS spectrum from 0 to 900 eV binding energies in Figure 2c, which is consistent with the results of EDS analysis in Figure S3, Supporting Information. Furthermore, Figure 2d presents the In 3d XPS spectrum with In valence states. The deconvolution of In 3d XPS spectrum revealed two peaks at 443.7 and 451.4 eV, corresponding to In 3d_5/2 and In 3d_3/2 states, respectively, suggesting the existence of the In (0) state,^{[20, 39, 40]} rather than the other common valence state indium (III), further suggesting the absence of indium oxide or hydroxide in the nanocatalysts. This finding confirms that the as-prepared mp-In@MWCNTs nanocatalysts are more stable in the ambient atmosphere compared to commercial indium catalysts (Figure S4, Supporting Information).

Nitrogen gas adsorption–desorption isotherms were measured to investigate the specific surface areas and pore distribution of the as-prepared mp-In@MWCNTs nanocatalysts. As shown in Figure 2b, mp-In@MWCNTs showed a clear hysteresis loop, indicating the presence of well-defined mesoporous structures in the composite catalysts. The specific surface areas of mp-In@MWCNTs and commercial In were found to be 543 and 66 m² g⁻¹, respectively, based on the Brunauer–Emmett–Teller (BET) results. The maxima of the pore size distribution in mp-In@MWCNTs were approximately at 1.5 and 2 nm, as revealed using the Barrett–Joyner–Halenda (BJH) method. The porous structure of mp-In anchored on MWCNTs can be attributed to the hydrogen gas evolution from the catalytic hydrolysis of sodium borohydride serving as geometric templates.^[39] However, commercial In catalysts had almost no porous structures (Figure S5, Supporting Information). The surface dangling bonds and unsaturated sites on mp-In@MWCNTs would promote the catalytic performance.

2.2 CO₂ Reduction Reaction Performance of mp-In@MWCNTs Nanocatalysts

The CO₂ electroreduction performance using mp-In@MWCNTs composite nanocatalysts was evaluated in a Nafion-117 membrane-separated H-type cell. Linear sweep voltammetry (LSV) curves of mp-In@MWCNTs and commercial In electrocatalysts were measured in a CO₂-saturated 0.1 m KHCO₃ solution at room temperature (Figure 3a). The mp-In@MWCNTs exhibited a higher current density of −78.5 mA cm⁻² at −1.2 V, while commercial In showed a current density of −16.0 mA cm⁻² at −1.2 V under the same conditions. Meanwhile, a more positive onset potential at about −0.7 V for mp-In@MWCNTs was observed in LSV curves compared with the commercial In, indicating significantly higher electrocatalytic activities of CO₂RR for the mp-In@MWCNTs catalyst. The CO₂ electroreduction was also tested in a flow cell shown in Figure S6, Supporting Information, the LSV curve acquired in a flow cell was similar to the LSV curve acquired in a H-type cell, indicating negligible influence in cells and convincing results. The LSV curves were also obtained for an Ar-saturated 0.1 m KHCO₃ solution to evaluate the hydrogen evolution side reaction. Both mp-In@MWCNTs and commercial In nanocatalysts displayed lower current densities in Figure 3a, due to the intrinsic low activity toward HER of metal In.^[7] Nevertheless, mp-In@MWCNTs still showed a higher activity result from the porous structure of indium nanocatalysts and carbon nanotube framework.

To quantify the reaction products and assess the selectivity toward CO₂RR, the controlled electrolysis measurements were performed for 1 h under the potentials varied from −0.5 to −1.2 V in a CO₂-saturated 0.1 m KHCO₃ solution. Using nuclear magnetic resonance (NMR) analysis and gas chromatography (GC), the products generated from CO₂RR were detected and quantified. Formate was the predominant CO₂RR product for mp-In@MWCNTs, accompanied by a small amount of CO and H₂. Other carbonaceous chemicals were not detected, neither liquid nor gaseous products. As shown in Figure 3b, the Faradaic efficiencies of different products (HCOO⁻, H₂ and CO) were calculated at different potentials on mp-In@MWCNTs. At −0.5 V negative potential, the Faradaic efficiency of formate (${\mathrm{FE}}_{{\text{HCOO}}^{-}}$) was firstly measured to be 23.7%, while the FE of H₂ (FE_H2) was up to 69.2% due to the competing HER side reaction. When the potential was raised to −0.8 V, ${\mathrm{FE}}_{{\text{HCOO}}^{-}}$ rapidly increased to 91.0%. ${\mathrm{FE}}_{{\text{HCOO}}^{-}}$ maintained >90% in a broad potential range from −0.8 to −1.0 V accompanied by the suppressed HER with FE_H2 of 6%. When the applied potential was shifted to more negative values of −1.1 and −1.2 V, ${\mathrm{FE}}_{{\text{HCOO}}^{-}}$ gradually declined due to the diffusion-limited process of reactant CO₂ molecules and the associated mass transfer.^[41-43] During the electrolysis of CO₂, the FE of CO remained at a relatively low level of 6%. Additionally, to further distinguish the properties of mp-In@MWCNTs toward CO₂RR, the commercial In was also tested as a CO₂RR electrocatalyst. As shown in Figure S7, Supporting Information, the commercial In exhibited ~70% ${\mathrm{FE}}_{{\text{HCOO}}^{-}}$ (which is much lower than for the mp-In@MWCNTs) at potentials between −0.8 and −1.0 V and peaked at 71.2% at −0.9 V, indicating a much higher overpotential. These findings confirmed that mp-In@MWCNTs possessed a high activity and selectivity for converting CO₂ to formate in a wide potential range, which is consistent with the main group of metal-based electrocatalysts (e.g., Sn, Bi and Pb) for CO₂ conversion to formate.^[7]

In addition, the formate partial current density of mp-In@MWCNTs and commercial In was calculated and plotted against the applied potential, as shown in Figure 3c. A superior current density of ${\mathrm{j}}_{\text{HCOO}}$ = 51.9 mA cm⁻¹ was observed at −1.2 V on mp-In@MWCNTs, which is five times higher than 9.9 mA cm⁻¹ for the commercial In catalyst. Meanwhile, Figure 3d shows the formate production rate during CO₂RR. The mp-In@MWCNTs showed a higher production rate of 898.6 μmol cm⁻² h⁻¹ than 208.6 μmol cm⁻² h⁻¹ for the commercial In catalyst. The superior catalytic activity and excellent selectivity in a wide operating potential window for mp-In@MWCNTs demonstrated the competitive electrocatalytic performance of CO₂RR toward formate conversion compared with other In-based CO₂RR catalysts (Table S1, Supporting Information).

The electrochemical stability is also an important indicator of catalytic performance. To further evaluate the performance stability of the new mp-In@MWCNTs catalysts, the long-term CO₂RR tests were performed in 0.1 m KHCO₃ electrolyte with CO₂-saturation. A flow rate of CO₂ (20 mL min⁻¹) was let into the cathodic compartment. As shown in Figure 3e, during 30 h long-term CO₂RR electrolysis at −1.0 V, the current density demonstrated a negligible decay at approximately −20 mA cm⁻², and the FE of formate remained unchanged at over 90%. Meanwhile, the morphology and crystalline structure of mp-In@MWCNTs were characterized after the long-term tests, and maintained almost unchanged, as shown in Figure 4a,b. The valence state of metallic In(0) was determined from the XPS analysis of mp-In@MWCNTs in Figure 4c,d. The results indicated that the mp-In@MWCNTs catalyst possessed excellent electrochemical durability.

To understand the reaction kinetics of CO₂ reduction on the surface of mp-In@MWCNTs, the Tafel slope was calculated and is presented in Figure 3f. In the low-current-density regions, the CO₂ reduction reaction was mainly limited by electro-kinetics.^{[40, 42, 44]} The mp-In@MWCNTs featured a Tafel slope of 134 mV dec⁻¹, which was close to 118 mV dec⁻¹, suggesting that CO₂RR on mp-In@MWCNTs was limited by the initial one-electron transfer to convert an adsorbed ${{\mathrm{CO}}_2}^{*}$ intermediate. While the Tafel slope for commercial In was much higher at 150 mA cm⁻², which is inferior compared to the mp-In@MWCNTs. It is well known that the rate-determining step of CO₂RR is the initial electron transfer from the electrode to the ${{\mathrm{CO}}_2}^{*}$ for conversion of ${{\mathrm{CO}}_2}^{*-}$. The faster initial electron transfer for CO₂ molecules to form the adsorbed ${{\mathrm{CO}}_2}^{*-}$ intermediate on the surface of mp-In@MWCNTs ensured the improved CO₂RR performance.

2.3 Origin of the Improved Activity and Selectivity of mp-In@MWCNTs Nanocatalysts

The above results suggest that the outstanding CO₂RR performance is primarily related to the unique structural features of the mp-In@MWCNTs catalytic structures. Mesoporous In nanocatalysts possessed enriched unsaturated sites and abundant dangling bonds on the surfaces, promoting the adsorption and desorption of CO₂ molecules and intermediates and facilitating CO₂ reduction. This work used carboxylic acid-modified MWCNTs as a support network in mp-In@MWCNTs, thus ensuring good electrical conductivity. As presented in Figure 5a, electrochemical impedance spectra (EIS) analyses were carried out in 0.1 m KHCO₃ to measure the ohmic resistance over the catalysts. The Nyquist impedance plots showed smaller impedance arcs for the mp-In@MWCNTs than for the mp-In nanocatalysts, indicating the higher electron transfer rate from indium nanocrystals to CO₂. We emphasize that highly conductive MWCNTs promoted the electronic conductivity, catalytic performance, and electron-transfer kinetics on the prepared mp-In@MWCNTs electrodes. In addition, the composite In catalyst has a higher conductivity than commercial In. To confirm the increased current density on mp-In@MWCNTs, the LSV curves on mp-In without MWCNTs supports were obtained for CO₂-saturated 0.1 m KHCO₃ as control experiments. The current density of −41.8 mA cm⁻² at −1.2 V for mp-In was inferior to −78.5 mA cm⁻² on mp-In@MWCNTs in Figure 5b, confirming the reduced charge transfer resistance. These results suggest that a higher electronic conductivity leads to a higher current density and an improved electrochemical reduction on the mp-In@MWCNTs catalysts, allowing for reducing the amount of In used (containing 15.0% indium measured by ICP) while maintaining excellent catalytic activity.

Additionally, the structure of mp-In@MWCNTs was formed by In nanocrystals activating CO₂RR and a three-dimensional framework of carbon nanotubes. Figure S8, Supporting Information displays the Zeta potentials of In nanocrystals, MWCNTs modified by carboxylic acid, and mp-In@MWCNTs. The corresponding Zeta potential for MWCNTs, In nanocrystals and the composite mp-In@MWCNTs was −17.6, +1.6, and −9.2 mV, respectively. The observed Zeta potential in mp-In@MWCNTs suggested the strong electrostatic attraction between the In nanocrystals and MWCNTs ensuring the excellent durability of mp-In@MWCNTs.

Based on the previous studies,^{[34, 39, 45]} the mesoporous structure of mp-In@MWCNTs was another determining factor in enhancing the CO₂RR performance by improving the adsorption and desorption of intermediates and facilitating the CO₂RR. The mp-In@MWCNTs contained more edges, low coordinated surface sites and MWCNTs supported framework, thereby enhancing the adsorption of CO₂ molecules. The volumetric adsorption of CO₂ was performed at 273 K, and the adsorption capacity is shown in Figure 5c. The amount of CO₂ adsorption capacity reached 93.25 cm³ g⁻¹ on mp-In@MWCNTs, which is much higher than 66.28 cm³ g⁻¹ (mp-In), 33.27 cm³ g⁻¹ (commercial In@MWCNTs), and 16.46 cm³ g⁻¹ (commercial In). The significantly enhanced adsorption capacity is conducive to the lower potential and high FE of formate in the CO₂RR. Meanwhile, the abundant active sites of mp-In@MWCNTs contributed to the larger electrochemical surface area (ECSA).

To further assess the ECSA of electrocatalysts, the cyclic voltammograms (CV) were performed at different scanning rates based on the double-layer capacitance (C_dl) method.^[39] Figure 5d,e show the CV curves corresponding to mp-In@MWCNTs and commercial In catalysts. As shown in Figure 5f, C_dl of mp-In@MWCNTs was estimated to be 19.0 mF cm⁻², which is 3.5 times larger than 5.6 mF cm⁻² for the commercial In, indicating more electrochemically active sites available on the surface of the mp-In@MWCNT catalysts.

Furthermore, the unsaturated coordination sites and edge dangling bonds on the surface of mp-In@MWCNTs could stabilize reaction intermediates and decrease the activation energy Ea.^{[39, 46-48]} Compared with the uncatalyzed reactions, a reaction path with the lower activation energy is enabled by the mp-In@MWCNTs catalyst, thus raising the reaction rate. The value of Ea was estimated by the Arrhenius equation via the electrolysis reactions at different temperatures. As shown in Figure 5h, the mp-In@MWCNTs catalyst exhibited a lower value of Ea of 12.7 kJ mol⁻¹, while the calculated value of Ea was 25.1 kJ mol⁻¹ for the commercial In catalyst. These results suggest the lower activation barrier for the key step and CO₂ activation on the mp-In@MWCNTs catalyst due to the increased exposed active sites in the unique porous structure. In addition, the d-band center (relative to the Fermi level) of the electrocatalysts indicated the binding strength of the intermediate. As shown in Figure 5g, the d-band center of mp-In@MWCNTs after the integration with MWCNTs was −3.42 eV (up-shifted close to the Fermi level), indicating that the introduction of MWCNTs promoted the electron-donation effect from indium to CO₂ and intermediates. Compared with mp-In nanocatalysts with the d-band center of −4.99 eV, the upshift of the d-band for mp-In@MWCNTs structures suggested that the stronger binding of intermediate OCOH* promoted CO₂ reduction due to the shift of the anti-bonding states above the Fermi level.