Kai
Zheng‡
a,
Ding-Yi
Hu‡
a,
Xue-Wen
Zhang
a,
Xian-Xian
Xiao
a,
Zi-Jun
Liang
a,
Jun-Xi
Wu
a,
Duo-Yu
Lin
a,
Lin-Ling
Zhuo
a,
Heng
Yi
a,
Li
Gong
b,
Dong-Dong
Zhou
*a and
Jie-Peng
Zhang
*a
aMOE Key Laboratory of Bioinorganic and Synthetic Chemistry, School of Chemistry, GBRCE for Functional Molecular Engineering, IGCME, Sun Yat-Sen University, Guangzhou 510275, China. E-mail: zhoudd3@mail.sysu.edu.cn; zhangjp7@mail.sysu.edu.cn
bInstrumental Analysis and Research Center, Sun Yat-Sen University, Guangzhou 510275, China
First published on 10th June 2024
Two-dimensional (2D) coordination polymers have attracted great attention for catalysis because of their abundant exposed active sites. Here, we show that bending the local structure of 2D coordination polymers can inverse the electrocatalytic selectivity. A series of ultrathin nanosheets based on isoreticular/isostructural/isomeric 2D Cu(I) triazolate coordination polymers were successfully prepared. By introducing an amino group on the triazolate ligand, the shape of the 2D layer transforms from planar into wavy, which inverses the electrocatalytic selectivity from the HER (selectivity ∼ 80%) to the CO2RR (selectivity ∼ 76%, C2H4 up to 52%). Computational simulations showed that the wavy structure allows the amino groups to form attractive hydrogen-bonding interactions with the key reaction intermediates of the CO2RR for C2H4 and steric hindrance with the key reaction intermediates of the HER, giving lower and higher reaction energy barriers, respectively.
Compared with three-dimensional (3D) bulk materials, 2D nanomaterials with abundant exposed active sites have demonstrated exceptional catalytic performances.11,23,24 Ultrathin nanosheets of 2D coordination polymers (CPs) combining the advantages of CPs and 2D materials have attracted increasing attention for catalysis.25–29 However, the supramolecular microenvironment of 2D CPs has been scarcely considered for catalysis. Because coordination bonds are weaker than conventional covalent bonds, to maintain the 2D structure and/or facilitate exfoliating, these 2D CPs usually exhibit planar structures rather than more complex ones.
Recently, we reported high and tunable CO2RR performances of a series of isoreticular (nbo-a) 3D Cu(I)-based porous frameworks, i.e., [Cu(detz)] (MAF-2, Hdetz = 3,5-diethyl-1,2,4-triazole) and its analogs consisting of different ligand side groups (methyl, ethyl, and propyl).30 With even smaller ligand side groups, [Cu(tz)] (Htz = 1,2,4-triazole)31 and [Cu(atz)] (Hatz = 3-amino-1,2,4-triazole)32 can retain the coordination modes of MAF-2 analogs, i.e., three-coordinated Cu(I) ions and triazolate ligands, as well as planar Cu2(tz)2 units, but they adopt the 2D sql-a network topology (Fig. 1).33 If these 2D CPs can be exfoliated into ultrathin nanosheets, high CO2RR performances similar to those of 3D porous MAF-2 analogs may be achieved. More importantly, because the pores of the sql-a network are too small for amino groups, the 2D networks of [Cu(tz)] and [Cu(atz)] are planar and wavy in their crystal structures, respectively. In other words, the layer structure of [Cu(atz)] exhibits 3D characteristics of molecular complexes, enzymes, and metal–organic frameworks.
Fig. 1 Crystal structures of 2D CPs. (a and b) [Cu(tz)] and (c and d) [Cu(atz)] viewed perpendicularly (a and c) and parallelly (b and d) to the coordination layers. |
In this work, we synthesize ultrathin nanosheets of not only isoreticular [Cu(tz)] and [Cu(atz)] but also isostructural/isomeric solid-solution framework structures containing different concentrations of amino groups and demonstrate that the bending of the layer structure is critical for enhancing the CO2RR and avoiding the HER.
Transmission electron microscopy (TEM) and atomic force microscopy (AFM) showed ultrathin nanosheets with thicknesses of ca. 4.5 and 4.9 nm for p-A0H100 and w-A100H0, respectively (Fig. 2 and S4†). Considering that the thickness of each layer of [Cu(tz)] and [Cu(atz)] is 3.1 and 3.6 Å, respectively (Fig. 1), the ultrathin nanosheets contained ∼14 layers of the 2D coordination networks. X-ray photoelectron spectroscopy (XPS) showed characteristic peaks of Cu(I) at 914.8 eV, 932.6 eV and 952.6 eV, but no satellite peaks of Cu(II) appeared. Since the characteristic peaks of Cu(0) and Cu(I) cannot be distinguished, the 914 eV is further attributed to Cu(I) by the Cu LMM region (Fig. S5†).30,34 Powder X-ray diffraction (PXRD) of p-A0H100 and w-A100H0 showed only one or two of the strongest diffraction peaks of [Cu(tz)] or [Cu(atz)], without any typical inorganic species (Fig. S1 and S2†), respectively, indicating that p-A0H100/w-A100H0 should have a similar coordination structure to those of [Cu(tz)]/[Cu(atz)]. Moreover, these diffraction peaks are significantly broadened, being typical for ultrathin nanosheets.26,35,36 In CO2-saturated KHCO3 solution (0.1 M, electrolyte for the CO2RR), their PXRD patterns can remain unchanged for at least one week, indicating that they are suitable for the CO2RR (Fig. S1 and S2†).
The different affinities toward H2O and CO2 can be explained by Grand Canonical Monte Carlo (GCMC) simulations and periodic density functional theory (PDFT) calculations (Fig. S8 and S9†). The results showed that H2O molecules evenly distribute on the surfaces of both p-A0H100 (C–H⋯O 3.53–3.62 Å) and w-A100H0 (N–H⋯O 3.07 Å and C–H⋯O 3.48–3.73 Å) with similar binding energies of −16.1 and −18.9 kJ mol−1, respectively. On the other hand, CO2 molecules interact very differently with p-A0H100 and w-A100H0. For w-A100H0, CO2 molecules concentrate in the groove sections of the wavy layer surface and simultaneously interact with Cu(I) and –NH2 groups (Cu⋯O 3.13 Å and N–H⋯O 3.18 Å) with a relatively strong binding energy of −24.7 kJ mol−1. For p-A0H100, CO2 evenly distributed on the flat layer surface (C–H⋯O 3.71–3.84 Å) with a weak binding energy of −12.2 kJ mol−1 (Fig. S8†).
Gas chromatography (GC) and 1H nuclear magnetic resonance (NMR) showed that most CO2RR products are gases (i.e., H2, CO, CH4 and C2H4) (Fig. 3b, S10–S15 and Tables S1–S4†). For p-A0H100, the H2 selectivity remained ∼80% over a wide potential range (−1.1 V to −1.5 V), and the total selectivity of CO2RR products was ∼20%. In contrast, w-A100H0 exhibited high CO2RR selectivity (∼76%) and low HER selectivity (∼24%) under the same conditions. Moreover, the C2H4 selectivity (44–52%) is significantly higher than those of other products, such as CO (4–24%) and CH4 (3–24%). It is noteworthy that the C2H4 selectivity of w-A100H0 can reach 51.8 ± 0.6% at −1.3 V, which is 26-fold higher than that of p-A0H100 (2.0 ± 0.2%) under the same conditions and 22-fold higher than the highest value of p-A0H100 (2.4% at −1.1 V) and comparable to the highest values of reported catalysts (Fig. 3c, d and Table S5†).40–45 Isotope labelling experiments confirmed that all the CO2RR products originated from CO2 rather than decomposition of the electrocatalyst (Fig. S16†).
Chronoamperometry tests at −1.3 V showed that the current density and C2H4 selectivity of w-A100H0 and p-A0H100 can remain unchanged for ca. 6 h (Fig. 3e and S17†). After chronoamperometry at −1.3 V for 8 h, PXRD showed that there was no obvious change before and after the CO2RR (Fig. S18†), which revealed that the coordination structure should be maintained; TEM showed that the morphology of the nanosheets was similar to that before catalysis and there was no occurrence of agglomeration or particles (Fig. S4†), which proved that there were no inorganic substances such as Cu or its derivatives present before and after the reaction; XPS combined with X-ray absorption near-edge structure (XANES) showed that the valence state of Cu did not change, and the N 1s region indicated that the intensity of Cu–N did not diminish after the CO2RR (Fig. S19 and S20†), suggesting that the Cu–N bonds were not broken.46,47 Therefore, the degraded catalysis performances after 6 h can be ascribed to the falling-off and/or poisoning effects rather than degradation of the catalysts (Fig. S21†), which have been widely observed in the literature.30,48,49
The operando electrochemical attenuated total reflection Fourier transform infrared (ATR-FTIR) spectra of both p-A0H100 and w-A100H0 showed characteristic peaks (Fig. S22†) of *COOH, *CO, and *CHO (intermediates for CH4) and CO*–*CHO (intermediates for C2H4).43,47 In particular, comparing the ATR-FTIR spectra of p-A0H100 and w-A100H0 at −1.3 V for 800 s, the peaks of p-A0H100 are significantly weaker than those of w-A100H0 (Fig. S23†). Furthermore, the characteristic peaks of C2H4 intermediates in w-A100H0 are significantly stronger than those of CH4, being consistent with the experimental results.
To identify the roles of the amino group and layer bending, p-A0H100 and w-A100H0 were post-synthetically modified by ligand exchange with Htz or Hatz, respectively, i.e., change x in [Cu(tz)x%(atz)100−x%] (Fig. S24, S25 and Table S6†). PXRD patterns of the modified nanosheet products, i.e., p-A12H88, p-A25H75, w-A25H75 and w-A50H50, were similar to those of the parent nanosheets, and the characteristic PXRD peaks did not shift, indicating that the local structures of nanosheets changed little (Fig. S26†). TEM, AFM, and XPS further confirmed the retention of nanosheet morphologies and local coordination structures (Fig. S27–S29†). Note that p-A25H75 and w-A25H75 are supramolecular isomers, which have not been reported for ultrathin nanosheets. PDFT calculations showed that the planar structure of p-A0H100 can be retained when 25% of tz− was replaced by atz−, the wavy structure of w-A100H0 can be retained when 75% of atz− was replaced by tz−, and higher ligand exchanging ratios resulted in irregular layer deformation, possibly implying some unknown kinetic factors (Fig. S25 and 30†).50 These results indicated that the small pores of a planar sql-a network can only accommodate half an amino group, i.e., two pores for one amino group.33
Electrocatalysis tests clearly demonstrated that p-AxH100−x and w-AxH100−x preferred the CO2RR and HER, respectively (Fig. 4a, S31–S36 and Tables S7–S10†). For p-AxH100−x, when x increased from 0 to 25, the HER selectivity slightly decreased (<3.4%) and the CO2RR selectivity slightly increased (<3.7%), indicating that the concentration of amino groups in the planar structures plays a minor role in the selectivity. The isomers p-A25H75 and w-A25H75 offer a straightforward/reliable comparison of the roles of the amino group and the layer shape. Possessing the same chemical compositions and the same concentration of amino groups, w-A25H75 showed 5.3% higher CH4 selectivity, 20% higher C2H4 selectivity, and 25% lower HER selectivity, indicating the critical role of the wavy structure for CO2RR/HER selectivity. For w-AxH100−x, when x increased from 25 to 100, the CO and CH4 selectivities decreased slightly (<4.7%), whereas the C2H4 selectivity increased rapidly (>30%) and H2 selectivity decreased rapidly (>31%) (Fig. 4a). These results indicated that, when the layers are wavy, the amino groups also play important roles in improving not only CO2RR/HER selectivity but also C2H4 selectivity.
Fig. 4 Selectivity and mechanisms for p-AxH100−x and w-AxH100−x. (a) CO2RR/HER selectivity at −1.3 V. (b) Comparison of ΔGmax of the CO2RR/HER. |
To further illustrate the role of wavy structure in CO2RR/HER selectivity, a hypothetical w-A0H100 structure (fixed during simulation, otherwise relaxed to p-A0H100) was constructed to perform PDFT calculations, and the results showed that the ΔGmax values of the CO2RR/HER were calculated to be 1.30/1.21 eV, which were similar to those of p-A0H100 rather than w-A100H0 (Fig. S43 and S44†). This result indicates that a high CO2RR/HER selectivity requires not only the wavy structure but also the amino groups.
For the RDSs, the simulated intermediate structures showed that the amino groups interact weakly with the adsorbed CO and H2O molecules (Fig. 5 and S47†). In addition, the energies of the reactants (i.e., CO*–*CO and H2O*–*) are quite similar for all five simulated structures, indicating that the amino groups play trivial roles on the electronic structure of Cu(I), being consistent with the XPS results (Fig. S29†). Furthermore, the electrostatic potential (ESP) and the Mulliken population analysis of the five structures in both planar and wavy structures are similar, indicating that they possess similar coordination affinity (Fig. S48 and S49†).60,61 Therefore, the differences in ΔGmax values should mainly arise from the product structures (i.e., CO*–*CHO and H*–*OH). Similar to the literature results,55,56,62,63 the free energies of CO*–*CHO and H*–*OH were higher than those of CO*–*CO and H2O*–*, respectively. For CO*–*CHO, the amino groups can form attractive hydrogen-bonding interactions with the CHO fragments of w-A100H0 (N–H⋯O 3.02 Å, H⋯O 2.00 Å, ∠N–H⋯O 166.7°) and w-A25H75 (N–H⋯O 3.12 Å, H⋯O 2.13 Å, ∠N–H⋯O 161.5°), which decrease the system energies to give low CO2RR barriers (Fig. 5a and b). In contrast, for H*–*OH, the amino groups showed obvious steric repulsion with the adsorbed H atoms of w-A100H0 (H⋯H 2.05 Å) and w-A25H75 (H⋯H 2.08 Å), which increased the system energies to give high HER barriers (Fig. 5c and d). For planar structures p-A0H100 and p-A25H75, as well as amino-free structure w-A0H100, such attractive/repulsive effects cannot be formed. Therefore, the inversed CO2RR/HER selectivity from p-A0H100 to w-A100H0 can be explained by the cooperation of the wavy structure and amino groups.
Note that w-A100H0 has larger attractive/repulsive interactions for CO2RR/HER intermediates than w-A25H75, respectively, which can be attributed to the higher curvature degree of the layers (Fig. S50 and S51†). In other words, although experimental techniques such as TEM and AFM cannot distinguish the slightly different curvature degrees of the nanosheet samples, computational simulations indicated that the concentration of amino groups can fine tune the curvature degree and supramolecular microenvironment of the 2D coordination layers, as well as the CO2RR/HER selectivity.
Footnotes |
† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d4ta01982b |
‡ These authors contributed equally to this work. |
This journal is © The Royal Society of Chemistry 2024 |