Efficient electron transmission in covalent organic framework nanosheets for highly active electrocatalytic carbon dioxide reduction

Efficient conversion of carbon dioxide (CO2) into value-added products is essential for clean energy research. Design of stable, selective, and powerful electrocatalysts for CO2 reduction reaction (CO2RR) is highly desirable yet largely unmet. In this work, a series of metalloporphyrin-tetrathiafulvalene based covalent organic frameworks (M-TTCOFs) are designed. Tetrathiafulvalene, serving as electron donator or carrier, can construct an oriented electron transmission pathway with metalloporphyrin. Thus-obtained M-TTCOFs can serve as electrocatalysts with high FECO (91.3%, −0.7 V) and possess high cycling stability (>40 h). In addition, after exfoliation, the FECO value of Co-TTCOF nanosheets (~5 nm) is higher than 90% in a wide potential range from −0.6 to −0.9 V and the maximum FECO can reach up to almost 100% (99.7%, −0.8 V). The electrocatalytic CO2RR mechanisms are discussed and revealed by density functional theory calculations. This work paves a new way in exploring porous crystalline materials in electrocatalytic CO2RR.


Responses to the Reviewers' Comments Reviewer 1:
Comments to the Author: In this manuscript Zhu et al. reported the synthesis of electroactive metal-porphyrin-Tetrathiafulvalene based Covalent Organic Framework and study the electrocatalytic activity for CO 2 reduction under basic conditions. On electrocatalytic CO 2 reduction can get published in Nature Communications after substantial more work has been performed. In particular control experiments/a fair comparison to other systems are missing, which makes it really hard to estimate how well the system performs. Also the extremely high metal loading is a concern which needs to be addressed. Another concern is the lack of details in the experimental section, which will make it very hard for others to reproduce these results. Overall the research is interesting and suitable for the readership of Nature Communications. However, further revision is needed to make this work publishable in Nature Communications. The authors should clarify the following points. 1. Donor-acceptor concept is not suitable for this type of COF. Tetrathiafulvalene and metallated porphyrin both can act as electron donors. The authors should explain this in detail what they meant for donor-acceptor polymer in this case and relative energetics by measuring electrochemistry of monomers.
Based on these experimental results, donor-acceptor concept might be suitable for this type of COFs. Metalloporphyrin and TTF are connected to construct metalloporphyrin-TTF based COFs. In the structure, TTF is a sulfur-rich conjugated molecule with two reversible and easily accessible oxidation states (i.e., radical TTF + cation and TTF 2+ dication) that have been widely studied as a kind of electron donor (ref. J. Am. Chem. Soc. 2012, 134, 12932). Co-porphyrin with inherent macrocycle conjugated π-electron system is very beneficial for electron mobility and Co(II) enables to be reduced to Co(I) during the process in many references (refs. Nat. Commun. 2018, 9, 4466;Science 2015Science , 349, 1208. The connection of tetrathiafulvalene and metalloporphyrin will presumably create an oriented electron transmission pathway under the motivation of electric field and thus obtained COF based materials exhibits excellent CO 2 RR performances (FE CO % = 91.3, -0.7V; long-time durability, 40 h). 2. For the xps, do they observe any shift in the binding energy of cobalt, Nickel once they are in COF structures in comparison to cobalt porphyrin monomer? The same thing goes for N as well.

Response:
Thanks for your insightful suggestion. We have observed the shift in the binding energy of cobalt, nickel and N in COF structures compared with metallated porphyrin monomer. Co-TTCOF/Co-TAPP and Ni-TTCOF/Ni-TAPP have been selected as target pairs to demonstrate the possible binding energy shift in XPS spectra. The spectra of Co-TTCOF and Co-TAPP display one pair of peaks arising from the spin-orbit doublet of Co2p, which can be assigned to the Co2p 3/2 and Co2p 1/2 ( Supplementary Fig. 28 a, b). The Co2p 3/2 and Co2p 1/2 peaks of Co-TTCOF locate at 780.84 eV and 796.14 eV, which presents apparent positive shift of ~0.25 eV and ~0.24 eV compared with pristine Co-TAPP (Co2p 3/2 , 780.59 eV and Co2p 1/2 , 795.90 eV). Besides, the binding energy of N1s for Co-TTCOF (Co-N bond, 398.64 eV) displays a negative shift of ~0.1 eV when compared with Co-TAPP (Co-N bond, 398.54 eV). Similar phenomenon is also detected for Ni-TTCOF, in which 0.07 and 0.06 eV shift are observed for Ni2p 3/2 and Ni2p 1/2 when compared with Ni-TAPP ( Supplementary Fig. 28c, d). Also for N1s, a 0.03 eV shift is detected for Ni-TTCOF in contrast to Ni-TAPP. The change of binding energy provides direct evidence that the charge carrier migration pathway might be from TTF to M-TAPP (M = Co, Ni) as supported by the HOMO-LUMO results in Response 1 and are also verified by other important works (refs. Adv. Mater. 2019, 31, 1802981;Nat. Commun. 2018, 9, 1425Angew. Chem. Int. Ed. 10  . In this paper, upon light irradiation, there was a slight positive shift (by 0.3 eV) in the Ti 2p binding energy, suggesting a decrease in its electron density under light irradiation. Meanwhile, two characteristic peaks attributed to CdS at Cd 3d 5/2 and Cd 3d 3/2 were observed without light, which underwent negative shift (by -0.2 eV) under light irradiation, suggesting an increase in the electron density on the CdS. These binding energy shifts provide direct evidence of the charge carrier migration pathway across the TiO 2 /CdS interface. In detail, the photogenerated electrons migrate from TiO 2 to CdS. Figure 28. High-resolution XPS spectrum of Co-TTCOF and Ni-TTCOF. a Co2p for Co-TTCOF. b N1s for Co-TTCOF. c Ni2p for Ni-TTCOF. d N1s for Ni-TTCOF. 3. What is the thickness of the working electrode after deposition of COF materials? (red). XANES spectra of b Co L-edge, d Ni L-edge, and f S L-edge for NiCo 2 S 4 (blue) and N-NiCo 2 S 4 (red) (ref. Nat. Commun. 2018, 9, 1425. Figure R1-2 shows the XPS Co 2p core-level spectra of NiCo 2 S 4 and N-NiCo 2 S 4 NWs. Both spectra display one pair of peaks arising from the spin-orbit doublet of Co 2p, which can be assigned to Co 2p 3/2 and Co 2p 1/2 . The binding energies of Co 2p 3/2 and Co 2p 1/2 display a positive shift of ~0.4 eV after nitrogen incorporation. The electrons of the metal atoms prefer to flow towards nitrogen atoms.

Response:
Thanks for your suggestion. We have measured the thicknesses of working electrodes before and after the deposition of COF materials using a vernier caliper, but there is no obviously change (~0.30 mm) in the macroscopic range. Therefore, we intend to detect the thickness of the working electrode from microscopic perspective through SEM test. Taking Co-TTCOF based working electrode for example, the average diameter of sample coated carbon cloth is about 9.8 μm compared with the bare carbon cloth (7.8 μm). Based on the result, the thickness of the coating is estimated to be about 1 μm (Supplementary Fig. 42). This might be ascribed to the special fabrication procedure of working electrode: M-TTCOFs (10 mg) was well-mixed with acetylene black (3 mg) and 0.5% Nafion solution (1 mL) to form a homogeneous ink. Then the mixture was uniformly dropped onto a carbon cloth (1 cm × 1 cm) and the working electrode was obtained after drying. Now we have added relative discussion in the revised manuscript (page 14, line 26) and Supporting Information (page 42, Supplementary Fig. 42).

Response:
Thanks for your insightful suggestion. To calculate the percent of electrochemically active cobalt or nickel, cyclic voltammogram tests of Co-TTCOF and Ni-TTCOF are conducted. Peak current and scan rate as two important parameters are detected to reveal the electrochemically active sites. The peak current shows a linear dependence on the scan rate (tested from 20 to 120 mV s -1 ) both for Co-TTCOF and Ni-TTCOF ( Supplementary Fig. 23, refs. Nat. Commun. 2018, 9, 4466;Chem. Commun. 2019, 55, 11634). Regression of the linear regime between 20 and 120 mV s -1 with equation: slope = n 2 F 2 A τ o /4 R T (n = number of electrons involved; F = Faraday constant in C mol -1 ; A = geometrical surface area of the electrode (0.071 cm 2 ); τ o = surface coverage; R = gas constant; T = temperature (298 K)) gives the surface concentrations (τ o ) of electroactive Co-TTCOF and Ni-TTCOF to be 7.05×10 -9 and 1.50×10 -9 mol cm -2 (ref. 5. They should also perform the catalysis with cobalt porphyrin only and explain the advantages of using it in this COFs support.

Response:
Thanks for your kind suggestion. According to your suggestion, the catalysis performance of cobalt porphyrin has been evaluated as comparison. Linear sweep voltammetry (LSV) curves (without iR compensation) show that the onset potential of Co-TTCOF (-0.45 V) is much more positive than that of Co-TAPP (-0.51 V) in CO 2 -saturated KHCO 3 solution ( Supplementary Fig. 33a). Furthermore, Co-TAPP exhibits a FE CO of 69% at -0.9 V, which is inferior to Co-TTCOF (91.3%, -0.7 V) ( Supplementary Fig. 33b, d). Besides, Co-TAPP gives a partial CO current density of 0.48 mA cm -2 , which is much less than that of Co-TTCOF (1.84 mA cm -2 ) at -0.7 V ( Supplementary Fig. 33c). As shown in Response 1, Co-TAPP (0.69 V vs. Ag/AgCl in CH 3 CN) is electron acceptor as indicated by its higher oxidation potential than 4-formyl-TTF (0.48 V vs. Ag/AgCl in CH 3 CN) ( Supplementary Fig. 26). In this work, Co-porphyrin and TTF are connected by robust covalent bonds to construct Co-TTCOF. Co-porphyrin with inherent macrocycle conjugated π-electron system is very beneficial for electron mobility and Co(II) enables to be reduced to Co(I) during the process as revealed in many references (refs. Nat. Commun. 2018, 9, 4466; Science 2015, 349, 1208). Thus-obtained Co-TTCOF exhibits excellent CO 2 RR performances (FE CO % = 91.3, -0.7 V; long-time durability, 40 h), which can further verify the advantages of using cobalt porphyrin in this COFs system. Now we have added relative discussion in the revised manuscript (page 8, line 27; page 7, line 39) and Supporting Information (page 33, line 3, Supplementary Fig. 33; page 26, Supplementary Fig. 26). Figure 33. Electrocatalytic performances of Co-TAPP. a LSV curves. b Faradaic efficiencies for CO. c Partial current density for CO. d FE CO calculated over potential range from -0.5 to -0.9 V.

Supplementary Figure 26. The cyclic voltammogram and optical tests of Co-TAPP and 4-formyl-TTF. a Cyclic voltammograms of Co-TAPP. b Solid state UV of Co-TAPP (inset Tauc plot). c Cyclic voltammograms of 4-formyl-TTF. d Solid state UV of 4-formyl-TTF (inset Tauc plot).
6. The authors should show the pH dependence of overpotential values.

Response:
Per suggestion, we have tested the pH dependence of overpotential values. To test it, electrolyte with various pH values (from 4.8 to 13.8) are applied and tested. An interesting phenomenon is observed that the alkaline solutions (pH = 7.8-13.8) are acidified by the saturated CO 2 and the final pH values are reduced to ~6.8. After numerous trial-and-error processes, acidic electrolyte with pH values of 4.8, 5.8 and 6.8 are picked as three representative ones to investigate the pH dependence of overpotential values (Supplementary Fig. 39, ref. Chem. Sci. 2016, 7, 1521). Linear sweep voltammetry tests (LSV, without iR compensation) (ref. ACS Appl. Mater. Interfaces 2019, 11, 1520) and electrocatalytic CO 2 RR performances of Co-TTCOF (ref. Nat. Energy. 2019, 4, 732) as two kinds of powerful methods are applied to reveal the pH dependence of overpotential values from different aspects. In LSV curves, the overpotential at 1 mA cm -2 decreases from ~410 mV (pH, 4.8) to ~340 mV (pH, 5.8) and finally slightly increases to ~350 mV (pH, 6.8) with the increase of pH values ( Supplementary Figs. 39a, b). While for the electrocatalytic performances of Co-TTCOF, the overpotential (based on the highest FE CO %) of Co-TTCOF reaches to ~790 mV both for pH = 4.8 and pH = 5.8, then the value decreases to ~590 mV (pH, 6.8). The results of these two methods indicate that the overpotential value is closely related to the pH of electrolyte ( Supplementary Figs. 39c, d).
Now we have added relative references (refs. 41 and 47) and discussion in the revised manuscript (page 9, line 8) and Supporting Information (page 39, line 4, Supplementary Fig. 39). Figure 39. The pH dependence of overpotential values for Co-TTCOF. a LSV curves. b Overpotential (η @ 1 mA cm -2 ) vs pH plot. c FE CO calculated over potential range from -0.5 to -0.9 V in various solution. d Overpotential (η @ highest FE CO %) vs pH plot. 7. The authors can also perform solid state UV to confirm the presence of cobalt porphyrin.

Response:
Thanks for your kind suggestion. According to your suggestion, the solid state UV tests of Co-TTCOF is performed. The electronic absorption spectrum of Co-TTCOF displays a solid-state broad absorption giving four maxima (Q-band, 546, 593  8. The authors can also perform TGA analysis for all the COFs materials.

Response:
Per suggestion, we have performed TGA analyses under N 2 atmosphere for all of the COFs (i.e. H 2 -TTCOF, Co-TTCOF and Ni-TTCOF). Taking Co-TTCOF for example, the TGA plot shows a slight weight loss of 6.3%, which might be attributed to the loss of solvent ( Supplementary Fig. 18a). Then the material remains stable up to ∼350 o C, followed by a sharply weight loss (~74.6%) until 550 o C and finally the plateau remains stable without weight loss up to 800 o C. The remaining mass is about 14.0 wt%, which might be ascribed to the formation of cobalt oxide. Similar result is detected for Ni-TTCOF ( Supplementary Fig. 18b, there plateaus: 4.9%, 63.8 % and 27.7%). For H 2 -TTCOF without metal doping, about 3.7% mass loss at temperature range from 25 to 200 o C is attributed to the loss of guest molecules. After 350 o C, the framework of H 2 -TTCOF starts to collapse and ends at about 600 o C ( Supplementary  Fig. 18c). Now we have added relative discussion in the revised manuscript (page 5, line 49) and Supporting Information (page 18, Supplementary Fig. 18). Figure 18. TGA analyses of M-TTCOFs under N 2 atmosphere. a Co-TTCOF. b Ni-TTCOF. c H 2 -TTCOF. 9. Does cobalt or Nickel leach into the solution after the run? Again this needs to be measured via ICP (post experiment measurement).

Response:
Thanks for your insightful comment. We have conducted the ICP tests of Co-TTCOF and Ni-TTCOF after long-time durability tests (> 40 h). Electrocatalysis was conducted (Co-TTCOF, -0.7 V (vs. RHE) and Ni-TTCOF, -0.9 V (vs. RHE)) after 40 h. After tests, negligible leaching of metal ions are detected for both Co or Ni in the electrolyte. This might be attributed to the strong covalent bond (C=N) form in the COF structures that can endow these materials with high durability. Now we have added relative discussion in the revised manuscript (page 9, line 4).
10. Manuscript writing should be improved; the authors have failed to cite recent references in introduction part, In the introduction: ACS Catal. 2017, 7, 6120, ACS Appl. Mater. Interfaces, 2019, 11, 1520and ACS Appl. Mater. Interfaces 2017 are relevant under COF, CMP catalysis with metal porphyrin CMPs, COFs should be cited.

Response:
Thanks for your useful suggestion. We have carefully checked and improved the manuscript writing through the manuscript. Besides, we have properly cited relevant references in the revised manuscript and highlighted them in yellow (page 17, refs. 40, 11. Electrocatalytic properties of COFs, which composed of free-base porphyrin, should be checked to verify the key role of Co 2+ and Ni 2+ as the active site.

Response:
Per suggestion, we have tested the electrocatalytic performances of H 2 -TTCOF (composed of free-base porphyrin). The gas chromatography (GC) analysis shows that H 2 is the primary reduction product over a wide potential range (-0.5 to -0.9 V vs. RHE), and only a small amount of CO is detected (FE CO , 4.22%, -0.7 V). The performance is inferior to Co-TTCOF (FE CO , 91.3%, -0.7 V) and Ni-TTCOF (FE CO , 20.9%, -0.9 V) (Fig. 2c). This implies Co 2+ or Ni 2+ as the active sites in the porphyrin center indeed plays a key role during the reaction process. Now we have added relative discussion in the revised manuscript (page 8, line 18).

Fig. 2c
Electrocatalytic performances of M-TTCOFs. FE CO is calculated over potential range from -0.5 to -0.9 V.
12. To established durability of your catalyst, you should mention material characterization data like HRTEM, FESEM, XPS after the CO 2 reduction experiment (post characterization data).

Response:
Thanks for your kind suggestion. According to your suggestion, we have added the HRTEM, FESEM, XPS tests of Co-TTCOF after electrocatalysis. To test it, the electrocatalyst is scraped from the surface of carbon cloth after electrocatalysis and characterized. The SEM and TEM images of Co-TTCOF agree well with the state before electrocatalysis, indicating Co-TTCOF can maintain its morphology after electrocatalysis ( Supplementary Fig. 37). Specially, some inevitable acetylene black particles are detected on the surface of Co-TTCOF owing to the fabrication procedure of working electrode. Besides, the XPS tests after electrocatalysis show that the valence state of Co(II) remains almost unchanged (Co2p 3/2 , 780.86 eV and Co2p 1/2 , 760.16 eV) when compared with that of Co-TTCOF (Co2p 3/2 , 780.84 eV and Co2p 1/2 , 796.14 eV) before electrocatalysis ( Supplementary Fig. 38). These results indicate Co-TTCOF to be excellent electrocatalyst with high durability, which might be attributed to the strong covalent bonds generated in COFs.
We have now included relative discussion in the revised manuscript (page 8, line 46) and Supporting Information (page 37, Supplementary Fig. 37; page 38, Supplementary Fig. 38). 13. The efficacy of the catalyst is well represented by TON and this value should be given. What is TON for this process?

Response:
Thanks for your useful suggestion. Turnover number (TON) is defined as the mole of reduction product generated per electrocatalytic active site over a given period of time, which is an important parameter to evaluate the catalysis performance of electrocatalyst. We have calculated the TON of Co-TTCOF for this process. Notably, the TON (CO) of Co-TTCOF as high as 40142 in just 10 h and can reach up to 141479 after 40 h (Supplementary Fig. 36).
The TON for CO was calculated as follows:

TON =
where Q is the total charge passed in time, E F is the Faradaic efficiency for the desired product, N is the number of electrons in the half reaction (N = 2 for CO 2 to CO conversion), F is the Faraday constant (F = 96485 C mol -1 electrons), and n tot is the total moles of catalyst employed in the electrolysis. The TON is calculated on the basis of the actually catalytic activity.
We have now included the relative discussion in the revised manuscript (page 8, line 44) and Supporting Information (page 36, Supplementary Fig. 36). Figure 36. Plots of CO and H 2 evolving turnover number versus time for Co-TTCOF. As shown in the images, the TON (CO) is as high as 40142 in just 10 h and can reach up to 141479 after 40 h.

Supplementary
14. The NMR is too broad and unresolved. Deconvolution of the solid state 13 C NMR spectra would be needed to state this COFs structure (e.g. Macromolecules, 2018, 51, 3088).

Response:
Thanks for your insightful suggestion. According to your suggestion, we have tested the solid state 13 C NMR spectra of Co-TTCOF with modified test parameters (now we use 600 MHz (400 MHz in our previous test, which caused too broad peak) 13 C NMR instrument (Bruker AVANCE III 600)) to improve the resolution. The formation of C=N bond is proved by the existence of resonance signal around 157.3 ppm (peak a) ( Supplementary Fig. 4a, b). The α-pyrrolic carbon present in the porphyrin moiety shows a peak at ∼144.5 ppm (peak b1). The dithiole carbon present in the tetrathiafulvalene shows a peak at ~113.8 ppm (peak d1). The peak around ∼115.8 ppm (peak d2) is attributed to β-pyrrolic carbon and methine carbon of the porphyrin macrocycle ( Supplementary Fig. 4d). Besides, the peaks among the range from 124 to 136 ppm (peak c) can reflect the existence of phenyl moiety in Co-TTCOF.
We have properly cited relevant references (ref. 39) and added relative discussion in the revised manuscript (page 4, line 22) and Supporting Information (page 4, Supplementary Fig. 4).