The anthropic excessive emission of carbon dioxide (CO2) has resulted in serious climate change and environmental crisis, such as global warming, acid rain, and rising of sea level, etc1. Powerful methods like storage and conversion of CO2 have been developed to reduce atmospheric CO2 level2,3,4. In contrast, conversion of CO2 is regarded as more promising way, due to which can make CO2 transform into serviceable high-valued energy products or chemicals (e.g., CO, CH4, HCOOH and C2H5OH, etc.), that would simultaneously mitigate the greenhouse effect and energy crisis5,6,7. At present, the most representative strategy for implementing this conversion process is electrocatalytic CO2 reduction reaction (CO2RR), because of its characteristic advantages of easy processing, simple devices, and low energy consumption8,9,10,11,12,13. Nevertheless, the electrochemical conversion of CO2 still needs to conquer considerable kinetic barrier required by the intrinsically thermodynamic stability and low electron affinity of the CO2 molecule, and the accompanying competitive H2 evolution is always hindering the generation of aiming product, resulting in hard to discover suitable electrocatalysts to afford high CO2RR efficiency and selectivity14.

During past decades, researchers have explored different kinds of materials for CO2RR electrocatalysis, such as metals8,15, transition metal oxides9,16, transition metal chalcogenides17,18, metal-free 2D materials14, and metal-organic frameworks (MOFs)19,20,21,22,23, etc24,25,26,27, among which MOFs are regarded as ascendant CO2RR electrocatalysts owing to their structural merits commonly involve in open Lewis acid metal sites in favor of adsorbing CO2 molecule, as well as well-defined catalytic environments suitable for mechanism explanation28,29,30,31. However, it is a pity that their poor electrical conductivity and electron-donating capability are always the major constraints for MOFs being as efficient electrocatalysts, considering the accomplishment of necessary multiple electron transfer process for obtaining any reductive products in CO2RR. Consequently, integrating electron-rich unit, electron mobility and active component related to specifically electrocatalytic CO2 reduction product, into MOF will likely to be an effective approach to improve the selectivity and efficiency of CO2RR.

As a proof-of-concept, we decided that assembling reductive polyoxometalates (POMs) and metalloporphyrins to construct MOFs so as to verify the feasibility of this inference32,33. The corresponding considerations are listed as follows. Firstly, reductive POMs mainly composed of low valent metal ions, such as Zn-ε-Keggin cluster ({ε-PMo8VMo4VIO40Zn4}34,35, including eight MoV atoms), are usually electron-rich aggregates36 and can easily offer electrons when triggered by redox reaction or bias stimulus; secondly, metalloporphyrin, where inherent macrocycle conjugated π-electron system is very beneficial for electron mobility, has proven to be advantageous to produce CO in electrocatalytic CO2RR19,37,38,39; thirdly, the connection of POM and metalloporphyrin will presumably create an oriented electron transportation pathway under the motivation of electric field, abundant electrons flowing from POM cluster to metalloporphyrin motif can guarantee and facilitate the fulfillment of multiple electron migration process of CO2RR; and finally, polyoxometalate-based MOFs always performing good structural robustness and chemical stability that is essential to the durability test of CO2RR. From the above we deduce that reductive polyoxometalate-metalloporphyrin organic frameworks (PMOFs) will probably be promising candidates to enhance the efficiency and selectivity of CO2RR, while the assembly of this type of structures is still extremely challenging40.

Herein, we synthesize a series of stable PMOFs constructed by tetrakis[4-carboxyphenyl]-porphyrin-M (M-TCPP) linkers and reductive Zn-ε-Keggin clusters through an in situ hydrothermal method with similar formulae of [PMoV8MoVI4O35(OH)5Zn4]2[M-TCPP][2H2O][1.5TBAOH] (M = Fe, Co, Ni, and Zn). The synergistic combination of Zn-ε-Keggin and M-TCPP in these PMOFs can serve as the role of gathering electron donating, electron migration, and electrocatalytic active component in the CO2RR. The electroreduction performances of diverse transition metal (i.e., Co, Fe, Ni, and Zn) based PMOFs are tested and relative results are discussed in the single-crystal level. As expected, these PMOFs show excellent performances in electrocatalytic CO2RR, especially for Co-PMOF, which is able to selectively convert CO2 to CO with a superior faradaic efficiency of 99% (highest in reported MOFs), a high TOF of 1656 h−1 (−0.8 V) and excellent stability for more than 36 h. It is worth noting that this is the first case to explore polyoxometalate-based MOF in the application of electrochemical CO2RR.


Structure and characterization of M-PMOFs

M-PMOF (i.e., Co-PMOF, Fe-PMOF, Ni-PMOF, and Zn-PMOF) samples were synthesized under similar hydrothermal conditions. A mixture of M-TCPP, Na2MoO4·2H2O, H3PO3 and zinc chloride was treated at 180 °C for 72 h and thus-obtained crystals were washed and measured. Single-crystal X-ray diffraction studies reveal that M-PMOF crystals have iso-reticular structures and are crystallized in orthorhombic Fmmm space group (Fig. 1). In the structure, Zn-ε-Keggin and M-TCPP present tetrahedral and quadrilateral connection modes, respectively. One Zn-ε-Keggin cluster with four Zn (II) centers connects with adjacent Zn-ε-Keggin clusters with two Zn-O bonds to give cluster chain. Chains are further connected by M-TCPP ligands to provide a three-dimensional framework with two-fold interpenetrated mog topology (Supplementary Fig. 1). Typically, in the structure, four Zn-ε-Keggin chains (adjacent distance is about 15 Å) are linked by M-TCPP ligands through the coordination of Zn (II) and carboxylic groups. The distance between two nearest and parallel M-TCPP units in uninterpenetrated network is about 17 Å (Fig. 1). The interpenetrated topology might endow these structures with high chemical, thermal, or catalysis stability and the direct connection of Zn-ε-Keggin with M-TCPP ligands in a structure would presumably create an oriented electron transportation pathway that might be beneficial for efficient charge transfer.

Fig. 1
figure 1

Schematic illustration of the structures of M-PMOFs (M = Co, Fe, Ni, Zn). M-PMOF was constructed by the 4-connected TCPP linkers and zigzag POM chains

To investigate their stability, M-PMOF samples are successfully prepared and certified by the powder X-ray diffraction (PXRD) tests (Supplementary Figs. 36). The thermal stability of M-PMOF samples is studied by thermogravimetric analysis (TGA) in O2 atmosphere with a heating rate of 10 °C min−1. Under O2 atmosphere, all these M-PMOF samples can be stable at temperatures higher than 200 oC. Taking Co-PMOF for example, about 4% mass loss at temperature range from 0 to 100 oC is attributed to the loss of guest molecules. After 200 °C, the framework of Co-PMOF starts to collapse and ends at about 500 °C (Supplementary Fig. 8). To further certify the thermal stability, PXRD patterns of M-PMOF samples treated at 200 °C in the presence of ultrapure O2 were tested. As shown in Supplementary Fig. 9, the PXRD patterns of M-PMOF samples still agree well with the simulated ones. Furthermore, M-PMOF samples show high chemical stability. All the M-PMOF samples enable to be stable in KHCO3 aqueous (0.5 M) for more than 24 h, guaranteeing the further investigation of electroreduction performance (Supplementary Figs. 1013). Moreover, after 24 h test, all of them can maintain their structures at pH range from 5 to 11 proved by PXRD tests, IR, and mapping test, respectively (Supplementary Figs. 14, 15). The high chemical and thermal stability of M-PMOF samples set fundamental basis for further applications.

The electrocatalytic performance of M-PMOFs

The electrochemical CO2RR performance is evaluated in an airtight three electrodes electrochemical H-type cell (the compartments are separated by a proton exchange membrane (Nafion®212)). All the M-PMOFs (i.e., Co-PMOF, Fe-PMOF, Ni-PMOF, and Zn-PMOF) samples are packed in the cells and tested. In this work, all the potentials are measured vs. Ag/AgCl electrode and the results are reported vs. reversible hydrogen electrode (RHE).

Linear sweep voltammetry (LSV) curves (without iR compensation) show that Co-PMOF has a small onset potential of −0.35 V and is much more positive than that of Fe-PMOF (−0.53 V), Ni-PMOF (−0.58 V), and Zn-PMOF (−0.60 V) in CO2-saturated 0.5 M KHCO3 (pH = 7.2) solution (Fig. 2a). Co-PMOF exhibits a small overpotential of 0.24 V and the relative equilibrium potential of CO2/CO is −0.11 V. Besides, Co-PMOF presents a total current density of 38.9 mA cm−2 at −1.1 V and is superior to Fe-PMOF (25.1 mA cm−2), Ni-PMOF(20.02 mA cm−2), and Zn-PMOF (16 mA cm−2). Due to its excellent performances, this work will mainly focus on Co-PMOF to discuss the possible relations between its properties and well-defined structure.

Fig. 2
figure 2

Electrocatalytic performances of M-PMOFs. a Linear sweep voltammetric curves. b Faradaic efficiencies for CO. c Partial CO current density and TOFs. d Tafel plots. e Durability test of Co-PMOF at the potential of −0.8 V vs. RHE. These tests are performed in 0.5 M KHCO3 electrolyte and at the scan rate of 5 mV s−1

Combing POMs and M-TCPP as building units, the high electrochemical activity of Co-PMOF is supposed to come from CO2RR. To prove it, relative tests are conducted both in N2 and CO2 saturated KHCO3 solution. As expected, Co-PMOF exhibits a more negative onset potential and a smaller current density in N2-saturated KHCO3 solution than that in CO2-staurated KHCO3 solution (Supplementary Fig. 16). Furthermore, the GC analyses show that H2 and CO are the primary reduction products and there is no liquid product detected by 1H nuclear magnetic resonance spectroscopy (Supplementary Fig. 17). To further evaluate the selectivity of M-PMOF samples for CO2RR, the corresponding FE for CO and H2 (denoted as FECO and FEH2) are calculated over the entire potential range (Fig. 2b and Supplementary Fig. 18). Taking Co-PMOF for example, the initial formation of CO is detected by GC at the potential of −0.35 V with a small CO partial current density of 0.39 mA cm−2. With the increase of potential, the FECO continuously increases and reaches up to a maximum value of 98.7% at −0.8 V and keeps higher than 94% even at −1.0 V (Supplementary Fig. 19). The high FECO over such as wide potential range (from −0.8 V to −1.0 V) outperforms many reported materials (Supplementary Table 1)41,42,43. Noteworthy, the maximum FECO value (98.7%) of Co-PMOF is higher than other M-PMOFs (i.e., Fe-PMOF, 28.8%; Ni-PMOF, 18.5%, and Zn-PMOF, 1.2%, respectively) and is highest among reported MOFs (Supplementary Table 1)36,37. In addition, the same electrochemical experiment of Co-PMOF is carried out in N2-saturated 0.5 M KHCO3 to reveal the source of CO and there is no CO detected under this condition (Supplementary Fig. 21). Besides, an isotopic experiment that using 13CO2 as substrate was performed under identical reaction conditions to reveal the carbon source of CO. The products were analyzed by gas chromatography and mass spectra. As shown in Supplementary Fig. 22, the peak at m/z = 29 is assigned to 13CO, indicating that the carbon source of CO indeed derives from the CO2 used.

To further reveal the efficiency of the remarkable performances of Co-PMOF, partial current densities of CO and H2 at different potentials are detected (Fig. 2c and Supplementary Fig. 23). Co-PMOF gives a partial CO current density of 18.08 mA cm−2 at −0.8 V. This value is 30 times larger than that of Fe-PMOF (0.27 mA cm−2), Ni-PMOF (0.47 mA cm−2), and Zn-PMOF (0.02 mA cm−2). Besides, TOF of Co-PMOF is calculated to be 1656 h−1 at −0.8 V and the TON values of Co-PMOF following the change of time was shown in Supplementary Fig. 2437. To the best of our knowledge, this catalytic behavior of Co-PMOF outperforms most of MOF catalysts and is one of the best reported materials (Supplementary Table 1)43,44.

To elucidate the dynamics activity of Co-PMOF for electrochemical CO2RR, Tafel slopes are calculated based on Tafel equation (η = b log j + a, where η is the overpotential, j is the current density and b is the Tafel slope) and presented in Fig. 2d. Remarkably, the Tafel slope for Co-PMOF is 98 mV dec−1, which is much smaller than that of Fe-PMOF (211 mV dec−1), Ni-PMOF (675 mV dec−1), and Zn-PMOF (206 mV dec−1). This indicates the favorable kinetics of Co-PMOF for the formation of CO, which might be ascribed to the more efficient charge transfer and larger active surface in the catalytic process. To support the hypothesis, electrochemical double-layer capacitance (Cdl) is calculated to estimate the electrochemical active surface area (ECSA) to further discuss the potential influence factors (Supplementary Fig. 25)45. The result show that Co-PMOF indeed exhibits larger Cdl value (12.17 mF cm−2) than that of Fe-PMOF (10.26 mF cm−2), Ni-PMOF (10.16 mF cm−2), and Zn-PMOF (9.83 mF cm−2), which enable to provide more active sites in electrocatalyst to contact the electrolyte for being beneficial to increase the reaction speed of electrochemical CO2RR.

Moreover, electrochemical impedance spectroscopy (EIS) measurement is carried out to probe the electrocatalytic kinetics on the electrode/electrolyte surface. The Nyquist plots demonstrate that Co-PMOF has much smaller charge transfer resistance (9.83 Ω) than Fe-PMOF (10.26 Ω), Ni-PMOF (10.70 Ω), and Zn-PMOF (12.17 Ω) during the electrochemical CO2RR (Supplementary Fig. 26). This indicates Co-PMOF can provide faster electron transfer from the catalyst surface to the reactant (i.e., CO2) in intermediate (HCOO* and CO*) generation, eventually resulting in largely enhanced activity and selectivity. This conclusion will be further verified by the theoretical calculations and relative contents are discussed in the mechanism part.

Stability is a key factor to evaluate the durability properties of catalysts. To evaluate the CO2RR stability of Co-PMOF, long-time durability test is performed with chronoamperometric test at a fixed overpotential of 0.69 V in 0.5 M KHCO3 solution (Fig. 2e). After 36 h, negligible decay in activity and FECO is detected (the gaseous product is analyzed by GC every 5 h). During the process, the corresponding FECO can be retained at values more than 90% and the current density is about 17 mA cm−2 over the entire experiment. This stability measurement reveals that Co-PMOF is highly stable electrocatalyst, which has much potential to be used in efficient electrochemical CO2RR (Supplementary Figs. 27, 38).


To reveal the catalytic mechanism, a series of contrast samples are prepared and tested. The electrochemical CO2RR of contrast samples are measured adopting the same testing methods. To evaluate the impact of Co-TCPP in electrochemical CO2RR, NNU-12 {[PMo8VMo4VIO36(OH)4Zn4][BCPT]2∙xGuests} is prepared46. NNU-12 with dia topology is constructed by Zn-ε-Keggin and BCPT2- linker, which contains the same Zn-ε-Keggin unit as Co-PMOF while the ligand is different. NNU-12 shows poor current density (3.63 mA cm−2 at −0.8 V), low FECO (1.80% at −0.8 V), high onset potential (−0.6 V) with a Tafel slope of 413 mV dec−1, and has negligible CO2 reduction activity (Fig. 3, Supplementary Figs. 28, 30). This indicates the high CO2 reduction activity of Co-PMOF originates from Co-TCPP. In the process, Co-TCPP might act as electrocatalytic center and redox-hopping-based conduit, which is beneficial for charge transfer.

Fig. 3
figure 3

Electrocatalytic performances of contrast samples for Co-PMOF. a Linear sweep voltammetric curves. b Maximum FECO of TMCP (−0.6 V), NNU−12 (−0.6 V), Co-TMCP (−0.9 V) and Co-PMOF (−0.8 V). c Partial CO current density. d Tafel plots

The influence of reductive POM in the electrochemical CO2RR is also investigated. Tetramethyl 5, 10, 15, 20-tetrakis[4-carboxyphenyl]-porphyrin-Co (Co-TMCP) is prepared and shows a small onset potential of −0.4 V, high current density (11.60 mA cm−2 at −0.8 V) with a Tafel slope of 151 mV dec−1 (Supplementary Fig. 29 and Fig. 3). In contrast, Co-TMCP exhibits poorer CO2 reduction performance than Co-PMOF, which might be attributed to the lower proton and electron transfer efficiency (Fig. 3 and Supplementary Fig. 30). Zn-ε-Keggin with strong reducibility might provide additional electrons and facilitate the charge transfer for Co-PMOF in electrochemical CO2RR.

In addition, tetramethyl 5, 10, 15, 20-tetrakis[4-carboxyphenyl]-porphyrin (TMCP) is prepared to study the effect of Co in the reduction process. TMCP shows poorer current density (2.06 mA cm−2 at −0.8 V), lower FECO (0.77% at −0.8 V), higher onset potential (−0.67 V) with a Tafel slope of 552 mV dec−1 and has no CO2RR activity than Co-PMOF (Fig. 3, Supplementary Figs. 29, 30), implying Co is the electrochemically active site in Co-PMOF. To further support the results, ECSA and EIS measurements are conducted and relative results are calculated. These results give similar performance tendency in Co-PMOF as discussed above (Supplementary Fig. 34 and 35). Moreover, MOF-525(Co) was prepared as relevant comparison to further support the catalytic mechanism of M-PMOF (Supplementary Fig. 31). MOF-525(Co) has similar ligand as Co-PMOF and is constructed with Zr6O4(OH)4 unit without POM47. MOF-525(Co) is successfully prepared and certified by the PXRD tests (Supplementary Fig. 32). The electrochemical CO2RR of MOF-525(Co) is measured adopting the same testing methods. As shown in Supplementary Fig. 33, MOF-525(Co) shows lower FECO (47.9% at −0.8 V) than Co-PMOF (98.7% at −0.8 V), which might be attributed to the poorer proton and electron transfer efficiency. This indicates POM actually acts as electron-rich aggregates in the catalytic mechanism of M-PMOF.

To understand the highly active as well as selective reaction mechanism of Co-PMOF, DFT calculations are performed (Supplementary note 2). Generally, the electroreduction of CO2 to CO includes three elementary reactions, the formation of *COOH and *CO with one electron transfer for each of them, and the finally CO desorption process. The asterisk (*) represents the surface active sites for adsorption and reaction. From the calculated free energy diagrams shown in Fig. 4a, the rate-determining steps (RDS) for CO2 reduction on single Zn-ε-Keggin (POM) and Co-TCPP are the formation of adsorbed intermediates *COOH and *CO with rather high free energies of ΔG1 = 0.96 eV and ΔG2 = 0.53 eV, respectively. Expectedly, when assembling the Zn-ε-Keggin and Co-TCPP together, the final compound of Co-PMOF possesses remarkably reduced ΔG1 and ΔG2, and particularly the RDS of *COOH formation decrease to a much smaller free energy of ΔG1 = 0.34 eV, being consistent with the high electroreduction activity and selectivity of this PMOF. These results indicate not only the favorable active site is the Co in Co-TCPP instead of POM, but also the efficiently synergistic electron modulation of POM and the porphyrin metal center. Then, the reaction pathways are further calculated by replacing the Co in porphyrin center with Fe, Ni, and Zn (Fig. 4b and Supplementary Fig. 36). For Fe-PMOF, although the energy of *COOH and *CO formations have decreased when compared to that of Co-PMOF, the desorption of CO become harder (ΔG3 = 0.98 eV), which may be ascribed to the higher affinity of Fe. For the case of Ni and Zn, the RDS both are the formation of *COOH with much higher free energies of 1.12 and 1.86 eV, respectively. This indicates the initial proton-coupled electron transfer to CO2 forming an adsorbed *COOH intermediate is sluggish in these systems. In a word, the sequence of the reaction energies on different metal centers is well consistent with our experiments, which further confirms the high activity of the Co modified compounds and the modulation effect in the M-PMOF systems.

Fig. 4
figure 4

The DFT calculation and proposed reaction mechanism. a The free energy diagrams of CO2 reduction to CO for POM (orange), Co-TCPP (green), and Co-PMOF (pink). The rate-determining step and the corresponding free energy on each material are indicated (Note that asterisk represents a surface active site for reaction). b Comparison of the free energy of each elementary reaction (ΔG1, ΔG2, and ΔG3 represent the free energy of *COOH formation, *CO formation, and CO desorption process, respectively) in CO2RR for Co-PMOF, Fe-PMOF, Ni-PMOF, and Zn-PMOF, respectively. c, d Proposed mechanistic scheme for the CO2RR on Co-PMOF (POM as electron-donating group, porphyrins as charge transfer ligand and transition metals as electron-collecting group)

Based on the experiment results and theoretical calculations, we elaborate the reduction processes from CO2 to CO on Co-PMOF and the possible mechanism is summarized in Fig. 4d and Supplementary note 3. During the electrochemical process, Co centers are reduced from Co(II) to Co(I) proved by the cyclic voltammetry (CV) tests (Supplementary Fig. 37), which has been detected in many Co-centered macrocycles11,29,37,48. As proposed in the mechanism, Zn-ε-Keggin initially traps the electron from the electrodes and transfers it to the Co center (Fig. 4c). Meanwhile, the Co center is reduced from Co(II) to Co(I) during the process. Then Co(I) interacts with carbon monoxide to give Co(II)*COOH intermediate coupling with the proton-electron transfer. Subsequently, Co(II)*COOH convert to Co(II)*CO. Finally, CO is desorbed and produced (Fig. 4d).

In summary, we have designed and synthesized a series of stable PMOFs by applying reductive Zn-ε-Keggin cluster and metalloporphyrin as building block and linker, respectively. The direct communication of reductive POM unit and metalloporphyrin under the exertion of the electric field results in an oriented electronic transportation channel that is extremely conductive to the accomplishment of multiple electrons transfer process in electrocatalytic CO2RR. Obviously, these PMOFs show excellent electrochemical CO2 reduction performances at last. In particular for Co-PMOF, the best one of all target compounds, enables to selectively convert CO2 to CO with a remarkable faradaic efficiency of 99% (highest in reported MOFs), a high TOF of 1656 h−1 (−0.8 V) and excellent stability for more than 36 h. Specially, DFT calculation demonstrates the superior catalysis performances of Co-PMOF than other transition metals (i.e., Fe, Ni, and Zn) and further reveals the synergistic effect of reductive POM and Co-porphyrin. This strategy opens great perspectives in designing novel and efficient CO2RR electrocatalysts and might shed light on the exploration of powerful protocols to address CO2 problems.


Syntheses of various M-TMCPs

5, 10, 15, 20-Tetrakis(4-methoxycarbonylphenyl)porphyrin (TMCP) (1.7 g, 2.0 mmol), CoCl2·6H2O (6.1 g, 25.6 mmol), and DMF (200 mL) were mixed and refluxed for 12 h. A large amount of water was added after cooling down to room temperature. The as-synthesized material was filtered and washed with water (600 ml) for six times. The fuchsia solid sample was obtained (yield 90% based on TMCP). The preparation processes of Fe-TMCP, Ni-TMCP, and Zn-TMCP were similar to Co-TMCP except that CoCl2·6H2O (6.1 g, 25.6 mmol) was replaced by FeCl2·4H2O (5.0 g, 25.6 mmol), NiCl2·6H2O (6.2 g, 25.6 mmol), and ZnCl2 (3.5 g, 25.6 mmol), respectively49.

Syntheses of various M-TCPPs

The obtained ester, MeOH (50 mL) and THF (50 mL) were mixed and stirred. A volume of 50 mL KOH solution (5.2 g, 93.9 mmol KOH was dissolved in 50 mL H2O) was added into the mixture and was refluxed for one night. After reaction, the organic solvents were evaporated. To make the solid fully dissolve, moderate water was added to the mixed solution and heated. 2 M HCl was used to acidify the obtained fuchsia homogeneous solution until no further fuchsia precipitate was generated. The solid was collected and washed with deionized water until the pH of filtrate reaches about 5. Finally, the solid sample was dried in vacuum at 60 °C. The colors of Co-TCPP, Fe-TCPP, and Ni-TCPP were red, brown, and crimson, respectively49.

Syntheses of various M-PMOFs

Taking Co-PMOF for example, a mixture of H3PO3 (10 mg, 0.125 mmol), zinc chloride (68 mg, 0.50 mmol), Na2MoO4·2H2O (310 mg, 1.28 mmol), and tetrabutylammonium hydroxide (TBAOH) solution (250 μL, 10 wt % in water) was added into 3.5 mL H2O and stirred for 15 min. 2 mol L−1 HCl solution was used to adjust the pH of the mixture solution until the pH reached about 5. Then, 5,10,15,20-tetrakis(4-carboxyphenyl)porphyrinato-Co (Co-TCPP) (132 mg), Mo powder 99.99% (25 mg), and dimethylacetamide solvent (250 μL) were added into the mixed solution and stirred in a 15 mL Teflon-lined stainless steel autoclave and heated at 180 °C for 72 h. After temperature cooled down to room temperature with a rate of 15 °C h−1, the crystals were washed with water and collected (yield was about 75% based on Co-TCPP). The preparation processes of Ni-PMOF and Fe-PMOF were similar to Co-PMOF except that Co-TCPP (132 mg) was replaced by Ni-TCPP (132 mg) and Fe-TCPP (33 mg), respectively. It is worthy to note that the preparation process of Zn-PMOF was similar to Co-PMOF except that Co-TCPP (132 mg) is replaced by TCPP (118 mg) (Supplementary Fig. 2). The IR peaks are listed as follows (Supplementary Fig. 7, KBr pellets, ν cm−1): Co-PMOF: 3427 (s), 3122 (w), 2958 (w), 1699 (w), 1604 (s), 1400 (s), 1040 (m), 1271 (w), 1172 (w), 1027 (m), 941 (s), 869 (w), 796 (s), 713 (m), 594 (w), 495 (w); Fe-PMOF: 3446 (s), 3211 (w), 1701 (w), 1602 (s), 1404 (s), 1271 (w), 1043 (m), 931 (s), 864 (w), 783 (m), 714 (m), 595 (w), 489 (w); Ni-PMOF: 3436 (s), 3214 (w), 1704 (w), 1606 (s), 1399 (s), 1261 (w), 1041 (m), 930 (s), 861 (w), 785 (m), 712 (m), 597 (w), 490 (w) and Zn-PMOF: 3431 (s), 3188 (w), 1703 (w), 1600 (s), 1401 (s), 1272 (w), 1038 (m), 937 (s), 865 (w), 789 (m), 711 (m), 596 (w), 492 (w). For the formulae of M-PMOF, all of these M-PMOF are iso-reticular in structure and only have minor difference in centered metal ions. Their structures have been well-defined in the SXRD tests. Based on the elemental analyses and SXRD tests, the calculated formulae was [PMoV8MoVI4O35(OH)5Zn4]2[M-TCPP][2H2O][1.5TBAOH] (M = Fe, Co, Ni, and Zn). This result was further supported by TGA test (Supplementary Fig. 8).

Synthesis of MOF-525

ZrCl4·8H2O (12.5 mg, 0.037 mmol) and DMF (10 mL) were sonicated for 30 min. Then, tetrakis(4-carboxyphenyl)porphyrin (2.5 mg, 0.037 mmol) was added into the solution. After sonication for 10 min, acetic acid (2.5 mL) was added. The solution was placed in a 20 mL scintillation vial and heated at 65 °C for 3 days. The microcrystalline powder was filtered and washed with DMF, acetone for several times. Finally, the powder was dried in vacuum at 120 °C for 48 h47.

Synthesis of MOF-525(Co)

CoCl2·6H2O (100 mg, 0.62 mmol), MOF-525 (50 mg), and DMF (10 mL) were mixed and heated at 100 °C for 18 h. The microcrystalline powder were collected by filtration and washed with DMF, acetone for several times. Finally, the powder was dried in vacuum at 120 °C for 48 h47.

Characterizations and instruments

Powder X-Ray diffraction (PXRD) patterns of samples were carried out on a D/max 2500 VL/PC diffractometer (Japan) equipped with Cu Kα radiation (λ = 1.54060 Å). FTIR spectra were tested on a Bruker Tensor 27 FT/IR spectrophotometer. Thermogravimetric analyses of crystal samples were measured on a Netzsch STA449F3 analyzer under the oxygen atmosphere with a heating rate of 10 °C min−1. Single-crystal XRD data of Co-PMOF, Fe-PMOF, Ni-PMOF, and Zn-PMOF were measured on a Bruker APEXII CCD diffractometer with graphite-monochromated Mo Kα radiation (λ = 0.71073 Å) at 296 K. All the structures were solved by direct method using SHELXT program and refined on an Olex2 software. Multi-scan technique was used to absorption corrections. The TBA+ ion could not be detected in the structure and some other guest molecules were defined as disorder. The detailed crystallographic information of these crystals was listed in Supplementary Table 2. The CCDC number of Co-PMOF, Fe-PMOF, Ni-PMOF, and Zn-PMOF (NNU-13) were 1832596 [], 1832597 [], 1832598 [], and 1181860 [] respectively.

Electrochemical measurements

All electrocatalysis tests of the catalysts were performed at ambient environment on the electrochemical workstation (Bio-Logic) in a standard three-electrode configuration in 0.5 M KHCO3 (pH = 7.2). Pt wire and Ag/AgCl electrode were used as the counter and reference electrode, respectively. The working electrode was modified carbon cloth (1 cm × 1 cm). The experiment was performed in an airtight electrochemical H-type cell with a catalyst-modified carbon cloth electrode (denoted as CCE, 1 cm × 2 cm) as the work electrode.

Given the poor intrinsic electrical conductivity of MOFs, acetylene black (AB) was introduced to mix with the as-synthesized MOFs to improve the conductivity. Nafion solution was introduced as a kind of MOF dispersion solution generally applied in many reported works, which can form a homogeneous ink with MOF and further help to attach onto the surface of carbon cloth. The pure carbon cloth has no CO2RR activity (Supplementary Fig. 20). The preparation of the CCE working electrode was as follows. A volume of 10 mg electrocatalyst, 10 mg acetylene black (AB), and 1 mL 0.5% Nafion solution were grounded to form uniform catalyst ink. After sonication for 30 min, the ink was dropped directly onto a carbon cloth (1 cm × 1 cm) with a catalyst loading density of ~1 mg cm−2 and dried. In the H-type cell, two compartments were separated by an exchange membrane (Nafion®212).

During the CO2 reduction experiments, the polarization curves were performed by LSV mode at a scan rate of 5 mV s−1. Initially, polarization curves for the modified electrode were recorded under an inert N2 (gas) atmosphere. After that, the solution was bubbled with CO2 (99.999%) for at least 30 min to make the aqueous solution saturated and then the electrocatalytic CO2RR was conducted (Supplementary note 1). Potential was measured vs. Ag/AgCl electrode and the results were reported vs. reversible hydrogen electrode based on the Nernst equation: E (vs. RHE) = E (vs. Ag/AgCl) + 0.1989 V + 0.059 × pH.

EIS measurement was carried out by applying an AC voltage with 10 mV amplitude in a frequency range from 1000 kHz to 100 mHz at a overpotential of −0.8 V (vs. RHE). To estimate the ECSA, cyclic voltammograms (CV) were tested by measuring double-layer capacitance (Cdl) under the potential window of −0.45 V to −0.55 V (vs. Ag/AgCl) with various scan rates from 10 to 100 mV s−1. All the LSV curves were presented without iR compensation.