Introduction

With the continuous emission of greenhouse gases, the situation of global warming is becoming increasingly severe [1, 2]. Electrochemical CO2 reduction reaction (CO2RR), characterized by low energy consumption, low cost, and high conversion efficiency, is a feasible technology for transforming greenhouse gases into valuable chemicals and feedstock through renewable and clean energy sources that stands out among many methods [3,4,5,6]. Moreover, it is equally important to seek cheap and abundant materials to serve as electrocatalysts that possess excellent activity, stability, and selectivity for transforming CO2. In this case, numerous catalysts have been developed, for instance, metal [7, 8], metallic oxide [9, 10], nonmetal material [11, 12], and heterogenous molecular complex [13, 14]. Considering the high cost and the low quantity of metal and its oxide limiting their extensive application, heteroatomic material, and molecular complex have aroused widespread interest.

Metalloporphyrin-centered molecular catalysts caught the researcher's eye because of the macrocyclic ligand framework with a conjugated structure and the regulable central metal ions with adjustable oxidation state [15, 16]. Therefore, they have been preferred in numerous redox reactions. The iron porphyrin has been reported to catalyze the CO2 to CO electrochemically, which represented high catalytic efficiency and selectivity [17,18,19]. Heterogenization of the molecular catalysts is capable of further enhancing the catalytic activity, which provided more active sites with controllable chemical surroundings [20,21,22]. Specifically, immobilizing the iron porphyrin-related molecular catalysts onto carbonic materials has achieved remarkable results in CO2RR. For instance, Maurin et al. [23] attached an iron molecular catalyst named \({\text{CAT}}_{{{\text{CO}}_{2} {\text{H}}}}\) covalently on the CNTs, the complex of which is highly selective and active for CO2RR in neutral water at low overpotential. Zhao et al. [24] combined Fe-porphyrin (FeTPPCl) and MWCNTs on glassy carbon electrodes, which exhibited a favorable synergistic effect and decreased overpotential of CO2RR.

Metal–organic frameworks (MOFs), composed of inorganic metal ions/clusters and organic ligands, are porous crystalline materials that possess structural tunability, stable porosity, and adjustable functionality [25, 26]. If well modified, the composite material is expected to exhibit the combined superiority of both the molecular catalyst and the porous architecture. For instance, Hod et al. [27] employed Fe-porphyrin-based MOFs that produced considerable catalytic sites (∼1015 sites/cm2) and 100% Faraday efficiency (FE) of the mixture of H2 and CO. Kornienko et al. [28] employed the Al oxide rods to anchor the active TCPPCo molecules into a MOF, which revealed a FECO of 76% and a turnover number (TON) of 1400.

It is potentially beneficial for CO2RR to employ MOFs to support the molecular catalysts (iron porphyrins) and immobilize the combination with carbonic materials. The introduction of molecular falsework makes it feasible to precisely control the spatial phase of the active catalytic centers [16]. Profiting from the flexible network structure and tunable pore surroundings, which may adsorb CO2 to increase its concentration, the activity and selectivity of CO2RR can be markedly promoted [29, 30]. Moreover, the addition of carbonic materials is advantageous to improve the conductivity of the catalysts, which meaning lower impedance and more effective electron transport [31, 32].

Carbon nanotubes, which are of great potential to CO2RR, thanks to their excellent conductivity, high stability, and considerable surface area, are ideal objects for the composition of MOFs [23, 33,34,35]. Herein, we obtained a composite catalyst named PCN-222(Fe)/CNTs, which was in situ synthesized through the solvothermal method that loads PCN-222(Fe) molecules onto CNTs. Thereinto, PCN-222(Fe) is the MOF that is centered on iron porphyrins named Fe-TCPP (5,10,15,20-tetrakis (4-methoxycarbonylphenyl) porphyrinato]-Fe (III) chloride) (Figure S1) and extended by Zirconium oxide clusters. By virtue of different structural characterizations, it was demonstrated that the PCN-222(Fe) molecules are successfully synthesized and uniformly dispersed on the surface of the CNTs. The consequence of 2 h chronopotentiometry analysis indicated that PCN-222(Fe)/CNTs has remarkable catalytic performance for transferring CO2 to CO with 95.5% FECO at −0.6 V vs. RHE, under a massive proportion Fe-TCPP: CNTs = 1:30. In addition, the turnover frequency (TOF) is as high as 448.76 h−1, and the hydrogen evolution reaction (HER) is indistinctive.

Experimental

Preparation of materials

MWCNTs (XFNANO, 95%), DUPONT Nafion PFSA Polymer dispersions (DuPont Engineering Polymers, 99.7%), KHCO3 (Aladdin, 99.5%), FeCl2·4H2O (Aladdin, 99%), pyrrole (Macklin, 99%), propionic acid (Macklin, 99.5%), ZrCl4 (99%). N,N-dimethylformamide (99.5%), acetone (99.5%), benzoic acid (99.5%), trichloromethane (99%), tetrahydrofuran (99%), methanol (99%), methyl p-formylbenzoate (99%), and hydrochloric acid (37%) were purchased from Sinopharm Chemical reagent. The materials are all employed without refinement, and the water used was deionized in the experiments.

Synthesis of PCN-222(Fe)/CNTs

Fe-TCPP was firstly synthesized according to the method as reported [36]. The composite PCN-222(Fe)/CNTs was synthesized by a traditional one-pot solvothermal method. Specifically, we state the process by using the 1:30 portion (m(Fe-TCPP):m(CNTs) = 1:30) as an example. ZrCl4 (70.0 mg), Fe-TCPP (50.0 mg), benzoic acid (2.7 g), and CNTs (1.5 g) were dissolved in DMF (8.0 mL) in a glass vial, then sealed in a reaction kettle and heated at 120 °C for 48 h. After cooling down to room temperature, the mixture was centrifuged. Afterward, 40.0 mL of DMF and 1.5 mL of HCl (8 mol L−1) were added to the mixture. The reaction system was transferred to the oil bath pan and kept at 120 °C for 12 h. After cooling down, the mixture was washed with DMF and acetone twice before centrifugation and drying. The catalysts with different proportions of 1:1, 1:2, 1:3, 1:5, 1:20, and 1:40 were synthesized through a similar synthetic process except that the contents of CNTs were 50.0, 100.0, 150.0, 250.0, 1000.0, and 2000.0 mg, respectively.

Preparation of PCN-222(Fe)/CNTs electrodes

Specifically, we still state the process by using the 1:30 portion as an example. 30.0 mg of dry catalyst material was ultrasonically treated in 2.0 mL of acetone (0.5% Nafion) for 2 h to obtain a well-dispersed black slurry. The carbon paper (99.5%) was cut into 1.0 × 1.0 cm2 pieces and pretreated overnight with 6.0 M HCl to remove residual metal impurities. The piece was then thoroughly rinsed with ultrapure water and dried before use. 40 μL slurry was dropped onto both sides of the chopped carbon paper, and the PCN-222(Fe)/CNTs electrode was obtained with a load of 3.75 mg cm−2.

Results and discussion

Microstructure characterization

The catalysts at 1:1, 1:2, 1:3, 1:5, 1:20, 1:30, and 1:40 we synthesized were characterized via PXRD, FT-IR, SEM, etc., and the microstructural discussion below is focused on the PCN-222(Fe)/CNTs-30 which performed the best catalytic activity to CO2RR. The crystalline phase of PCN-222(Fe)/CNTs at different proportions were confirmed through powder X-ray diffraction (PXRD). Figure 1a exhibits the PXRD pattern of PCN-222(Fe)/CNTs-30, pure CNTs, and the as-synthesized PCN-222(Fe). The diffraction peaks around 26.12°, 42.91°, 53.90°, and 78.04° are described as the signals of the (002), (100), (004), and (110) crystal planes of graphite [37]. Specifically, PCN-222(Fe)/CNTs-30 exhibits the characteristic peaks just as the pure CNTs, which is attributed to the coverage of the CNTs. As shown in Fig. S2a, when the load of CNTs is low, the catalysts represent favorable crystallinity of PCN-222(Fe), since the diffraction peaks appearing around 6.62°, 7.03°, 8.18°, and 9.58°, respectively, correspond to the previously reported PCN-222(Fe) [36]. With the load of CNTs increased, the characteristic peaks of PCN-222(Fe) disappear. The phenomenon demonstrates the successful combination of PCN-222(Fe) and CNTs since the diffraction peaks of the composite are in accord with CNTs and as-synthesized PCN-222(Fe).

Figure 1
figure 1

a The PXRD pattern of PCN-222(Fe), CNTs, and PCN-222(Fe)/CNTs-30. b The FT-IR spectra of PCN-222(Fe), CNTs, and PCN-222(Fe)/CNTs-30

Full size image

Functional groups on CNTs, PCN-222(Fe), and PCN-222(Fe)/CNTs-30 were verified by FT-IR spectroscopy. As shown in Fig. 1b, the pattern of PCN-222(Fe)/CNTs-30 is entirely in accord with PCN-222(Fe) and CNTs. The symmetric and asymmetric stretching vibrations of carboxylic groups in PCN-222(Fe) appear at 1417 cm−1 and 1606 cm−1, respectively. The characteristic peaks ranged in 2920–3090 cm−1 correspond to the C–H bond of pyrrole and benzene, and the C=C bonds of them are confirmed by the peaks around 1550–1603 cm−1 [38]. Moreover, the peaks of the composite at 3460 cm−1 and 1648 cm−1 are provided by CNTs. The peaks above demonstrate the successful combination of PCN-222(Fe) and CNTs. As shown in Fig. 1b and Fig. S2b, the absorption peaks of Fe–N bonds in Fe-TCPP centered at 1000 cm−1 are weakened to disappear [39], owing to the addition of the increasing load of CNTs, proving a successful synthesis of the composite catalyst as well.

As shown in Fig. 2, the pore size distribution and the specific surface area of PCN-222(Fe), CNTs, and PCN-222(Fe)/CNTs-30 were confirmed by N2 adsorption–desorption tests. According to the IUPAC regulation, the isotherm of PCN-222(Fe) is classified as type I, indicating micropore structure, while a turning point at P/P0 = 0.3 implies mesoporosity of PCN-222(Fe) [40]. As shown in Fig. 2d, the pores of 1.2 and 2.4 nm suggest triangular microchannels and hexagonal mesochannels, respectively, according to the density functional theory calculation (DFT). However, the isotherm of PCN-222(Fe)/CNTs-30 is displayed as type IV with H3 type hysteresis loop that indicated mesopore structure [41], which can be further verified from Fig. 2c. The mesoporosity of the catalyst can be attributed to the composition of CNTs, which exhibits a type IV isotherm and mesoporosity as well. Besides, the measured BET specific surface area of the composite catalyst is 109.61 m2 g1, which is relatively lower than PCN-222(Fe) of 1749.01 m2 g−1 but approximate to CNTs of 208.56 m2 g−1 (Table S1). The discussion above demonstrates the successful combination of PCN-222(Fe) and CNTs and the mesopore structure of the composite catalyst at a ratio of 1:30.

Figure 2
figure 2

a N2 sorption isotherms of PCN-222(Fe), CNTs, and PCN-222(Fe)/CNTs-30. b–d DFT pore size distribution of PCN-222(Fe)/CNTs-30, CNTs, and PCN-222(Fe), respectively

Full size image

The textural microstructure of the catalyst was illustrated by scanning electron microscopy (SEM) and transmission electron microscopy (TEM). Primitive CNTs exhibit one-dimensional fibrous morphology, and the as-synthesized PCN-222(Fe) displays a regular crystalline structure of a blocky morphology, as shown in Fig. S3. The composite PCN-222(Fe)/CNTs-30 reserves the fibrous morphology of CNTs but appears a rougher surface and a wider diameter since the CNTs are surrounded by PCN-222(Fe) (Fig. 3a, b). Furthermore, high-angle annular dark-field scanning transmission electron microscopy (HAADF–STEM) and energy-dispersive X-ray spectroscopy (EDS) were conducted to explore the surface structure of the catalyst. As shown in Fig. 3c, the composite catalyst exhibits linear morphology as displayed by TEM and SEM. Homogeneous distributions of C, N, and Fe elements can be observed on the surface of the well-defined fibrous CNTs, demonstrating a successful dispersion of PCN-222(Fe) on CNTs (Fig. 3d–f). The measurements above further confirmed the successful synthesis of the composite catalyst.

Figure 3
figure 3

a SEM of PCN-222(Fe)/CNTs-30. b TEM of PCN-222(Fe)/CNTs-30. c HAADF–STEM of PCN-222(Fe)/CNTs-30. d–f EDS of C, N, and Fe

Full size image

Measurement of electrocatalytic performance

The catalysts were loaded on carbon paper (CP), and the optimum load of the catalysts was explored by cyclic voltammetry (CV) and potentiostatic test whose results are shown in Fig. S4. It is evident that the load of the catalyst exerts an influence on the performance of CO2RR. Concretely, the samples with a low load of CNTs exhibit generally poor FE for CO and prominent HER (Fig. S4a–c); with a high load of CNTs at the ratio of 1:5 to 1:40, the FE for CO display maximum value varies with the loading of catalysts (Fig. S4d–g). The phenomenon is probably attributed to the fact that the increasing load of the catalysts can improve the conductivity of the electrodes but over thickened coating of the catalysts make side-effect on the conductivity and the charge transfer of the electrode. The concrete data are displayed in Table S2, and the specific calculating example is shown in Supplementary information. In the following discussion, we emphasize the optimum load of the catalysts with a higher FECO and relatively suppressed HER.

CO2RR was conducted in a 0.5 M CO2-saturated KHCO3 aqueous solution. Figure 4a exhibits the chronoamperograms of catalysts at different proportions at −0.6 V vs. RHE. All of them show a high initial current density and reach stability after 1 h. Prior to being steady, the current density on the catalyst experienced two sections of decline and we suppose a phenomenological explanation. In section A (marked in Fig. 4a), CO2RR didn’t occur instantaneously but over a short period, and the resistance of the system was solution resistance (Rcell); at the end of section A, CO2RR was activated, the resistance of the system was turned into charge transfer resistance (Rct), which was much higher than Rcell (Table S6). The numerical variation of current density was negatively correlated to Rct, which was in accord with Ohm’s Law. In section B, as CO2RR starting, numerous bubbles are generated which protrude the mass transfer process, leading to the increase in resistance and the decrease in current; Besides, being combined with CNTs decrease the excellent porosity of PCN-222(Fe), which may impede the gas transport to a certain extent. When the transportation of the gas from catalyst to solution achieved a dynamic equilibrium, the resistance of the system realized stability, resulting in an invariable current density. Gas chromatography (GC) was further employed to examine the gaseous product, and the calculation is shown in Fig. 4b. Thereinto, PCN-222(Fe)/CNTs-30 exhibits the highest FE for CO (95.5%) with 494 mV overpotential and attains a TOF of 448.76 h−1 (3.011 site−1 s−1) (calculation method is elucidated in Supplementary information). For primitive CNTs, HER dominated, where the FE for H2 is high as 70%, and the FE for CO is considerably low (less than 1%). The results above demonstrate not only excellent catalytic performance but also high selectivity of the composite catalyst. Detailed information is elucidated in Table S3.

Figure 4
figure 4

a I–T curve of catalysts at different ratios. b FE of catalysts at different ratios for CO and H2

Full size image

After clarifying the conclusive role that PCN-222(Fe) played in the selectivity to CO, we explored the effect of different potentials used through the CO2RR. Figure 5a shows the chronoamperograms of catalysts at different potentials and each of them maintains constant during the catalysis, which also exhibits short-term stability. An increase for the current density at −0.7 and −0.8 V vs. RHE are observed, and we ascribe the increase to the dominant HER [42]. Namely, the larger overpotential induces a more significant response to HER instead of CO2RR, the consequence was further verified through the calculation of FE for H2 and CO2, which are plotted in Fig. 5b, and the detailed information is elucidated in Table S3. The FE for H2 at −0.7 V vs. RHE and −0.8 V vs. RHE is approximately 70% or even more while the FE for CO is considerably low (less than 20%). At a lower potential (−0.5 V vs. RHE), there is still no improvement in the selectivity of products, which performs similar FEs for CO (59% ± 2%) and H2 (46% ± 3%). It is safe to conclude that −0.6 V vs. RHE is the optimal potential for CO2RR with PCN-222(Fe)/CNTs-30 served as the catalyst.

Figure 5
figure 5

a I–T curve of PCN-222(Fe)/CNTs-30 at different potentials. b FE for CO and H2 at different potentials

Full size image

To explore the reason for the high catalytic activity of the PCN-222(Fe)/CNTs-30, the concentration of surface electrochemically active sites (Γ) was calculated. As shown in Fig. 6a, the CV pattern of the PCN-222(Fe)/CNTs-30 exhibits one redox wave at −0.844 V vs. Ag/AgCl, which is related to the FeII/FeI redox couple [27, 43]. The electron transfer of the oxidation of Fe(I) to Fe (II) results in a peak area and the integration of which is calculated as QCV = 0.0003766 (already divided scan rate) (Fig. 6b). Γ's calculation is based on the equation: Γ = QCV/nFA [44]. In this specific example, ‘A’ refers to the catalytic surface area, ‘n’ refers to the electron transfer of the oxidation of Fe (I) to Fe (II) (n = 1). ‘F’ refers to the Faraday constant. It is calculated that the Γ is 1.95 × 10−9 mol cm−2. Moreover, the number of surface-active PCN-222 (Fe) sites is calculated by assuming a one-electron redox process: n′ = Q/F = 3.9 × 10−9 mol and the surface fraction of electrochemically active PCN-222 (Fe) sites is calculated as χ = n′/ntotal = 2.7%. More calculating details are in the Supplementary information.

Figure 6
figure 6

a CVs of the PCN-222(Fe)/CNTs-30, scan rate = 50 mV s−1. b Integration of the FeII/FeI redox couple peak area

Full size image

ECSA (electrochemical surface area) was evaluated by the corresponding double-layer capacitance (Cdl) [45, 46]. The CVs of PCN-222(Fe)/CNTs loaded on CP in CO2-saturated electrolyte (vs. Ag/AgCl) are detailed in Fig. S5. The Cdl values of PCN-222(Fe)/CNTs at different ratios are fitted in Fig. 7a. With the increase in the load of CNTs, the ECSA of the catalysts is promoted, which is probably attributed to the fact that as the load of CNTs increased, more active metal centers are exposed to the high conducting CNTs (Table S4). The increased ECSA facilitates rapid electron transfer and the kinetics of the electrochemical reaction [47].

Figure 7
figure 7

a Fitting Cdl values and b EIS of catalysts at different ratios

Full size image

Moreover, to explore the influence of different CNTs contents on the conductivity of catalysts, we conducted the electrochemical impedance measurement on all samples at −1.235 V vs. Ag/AgCl. As shown in Fig. 7b, the Nyquist plots are fitted with the equivalent circuit (Rcell(CRct)W) (inset in Fig. 7b and Table S5), which exhibits the solution resistance (Rcell) and the charge transfer resistance (Rct) of the reaction. The Rcell of all samples is relatively constant, ranging from 3 to 6 Ω since the same electrolyte was employed. With the increasing load of CNTs, the radius of the semicircles accordingly declined, indicating the enhancement of the charge transferability because of the correlation of radius with Rct. PCN-222(Fe)/CNTs-30 exhibits a value of 85.48 Ω, reflecting favorable electron transfer, less polarization loss, and outstanding kinetic of CO2RR [48]. Nevertheless, when the load of CNTs is excessive, the Rct increased again, which may be ascribed to an inferior synergetic effect.

Long-term chronoamperometric measurement was conducted to investigate the stability of the well-performed PCN-222(Fe)/CNTs-30 (where E = −0.6 V vs. RHE). Figure 8a shows the 10 h i–t curve of PCN-222(Fe)/CNTs-30, which stayed almost constant during electrocatalysis. As shown in Fig. 8b, FE for H2 and CO are depicted. In the beginning 4 h, PCN-222(Fe)/CNTs-30 exhibits the maximum FECO of about 95.5%, and the entire electrocatalysis produced 143 μmol of CO with an average FECO of 90%, demonstrating the excellent high stability of the catalyst. The result is superior to our previous work with an average FECO of 80.4% [16]. Nevertheless, an abnormal phenomenon occurs in Fig. 8b that the total FE for CO and H2 is higher than 100%. Considering errors in many aspects, we proposed a probable explanation in Supplementary information. Ultimately, the performance parameters of the catalyst synthesized in this work and other similar catalytic materials are compared in Table S6.

Figure 8
figure 8

a Long-term I–T curve of PCN-222(Fe)/CNTs-30 and b FE for CO and H2

Full size image

Additionally, Tafel plot was measured to explore the mechanism of the CO2RR on the surface of the synthesized PCN-222(Fe)/CNTs [49]. Specifically in this report, FE for CO and correspond current density from −0.4 V vs. RHE to −0.6 V vs. RHE (higher potentials lose their linearity) were diffusely tested (Fig. S6a), and partial current density of CO was obtained through jCO = j × FE(CO). Tafel slope was fitted by regarding lg(jCO) as x-axis and overpotential (η) as y-axis with a result of 137 mV dec−1 (Fig. S6b), which is closed to the calculated value of about 118 mV dec−1 [50]. A possible CO2RR mechanism of PCN-222(Fe)/CNTs-30 is explained in Eq. 1–4 and Eq. 2 is the possible rate-determining step (RDS), which is supported by Tafel analysis. Firstly, the dissolved CO2 molecules are absorbed by the porous PCN-222(Fe)/CNTs-30/CP electrode (Eq. 1). Subsequently, the absorbed CO2 molecule gets an electron and yield \({\text{CO}}_{2}^{ - }\). Then in Eq. 3, the \({\text{CO}}_{2}^{ - }\) transfers into H2O and CO by taking two protons and another electron. Ultimately, the absorbed CO molecule is released from the catalyst (Eq. 4).

$${\text{CO}}_{2} \,\left( {{\text{solution}}} \right) \to {\text{CO}}_{2} \,\left( {{\text{adsorbed}}} \right)$$
(1)
$${\text{CO}}_{2} \,\left( {{\text{adsorbed}}} \right) + e \to {\text{CO}}_{2}^{ - } \,\left( {{\text{adsorbed}}} \right)$$
(2)
$${\text{CO}}_{2}^{ - } \,\left( {{\text{adsorbed}}} \right) + e + 2{\text{H}}^{ + } \to {\text{CO}}\,\left( {{\text{adsorbed}}} \right) + {\text{H}}_{2} {\text{O}}$$
(3)
$${\text{CO}}\,\left( {{\text{adsorbed}}} \right) \to {\text{CO}}\,\left( {{\text{released}}} \right)$$
(4)

Conclusion

In general, we obtained a composite PCN-222(Fe)/CNTs catalyst through in situ solvothermal synthesis process that loads active PCN-222(Fe) molecules onto CNTs. Thanks to the favorable synergy of PCN-222(Fe) and CNTs, the catalyst PCN-222(Fe)/CNTs-30 exhibits excellent electrocatalytic performance for CO2RR with a FECO of 95.5%, where η = 494 mV. In addition, the TOF is high as 448.76 h−1, and the HER is well suppressed. The catalyst is found a high concentration of surface electrochemically active sites, a large ECSA, and a comparatively low charge transfer resistance. After long-term electrocatalysis, the catalyst exhibits high stability with an average FECO of 90%. In short, the composite catalyst PCN-222(Fe)/CNTs combines the advantages of molecular complex centered MOFs and the conducting material (CNTs). The method of combining molecular complex centered MOFs and the conducting CNTs is promising for transforming greenhouse gas into valuable chemicals and feedstock, which provides an expectation for further exploration in CO2RR.