To demonstrate the advantage of compositing MOF on the carbonaceous scaffold, we prepared HKUST-1 nanoparticles inside a porous activated carbon. HKUST-1 is a well-known Cu-based MOF constructed of dimeric metal units connected by benzene-1,3,5-tricarboxylic acid (BTC) linkers. The MOF/carbon composite was prepared according to the method described previously.26 The precursors of the MOF, namely copper chloride and BTC, were each separately introduced into the nanopores of activated carbon by continuously stirring the activated carbon particles in the BTC solution, followed by a copper chloride solution. A subsequent hydrothermal reaction resulted in MOF nanoparticles (Fig. 1).
X-ray diffraction (XRD) spectroscopy was employed to characterize the crystal structure of the composite (Fig. 2a). Notably, the carbon matrix did not shield the XRD signals of the encapsulated MOF. Moreover, HKUST-1 is known for its morphological sensitivity upon synthesis conditions;27 the porous carbon composition does not adversely affect the MOF crystallinity. The XRD pattern of the composite perfectly matches the pattern of MOF synthesized out of the carbon, together with the broad peak centered at ⁓25⁰, corresponding to (002) plans of graphitic segments in the carbon. In the proper MOF/carbon ratio of 1:3 (w/w), almost all the MOF has located inside the carbon's pores, as observed by SEM (Figure S1). Due to the MOF's organic component's strong interactions with the aromatic rings of the host carbon, the originally non-conductive MOF gains excellent lateral conductive properties (up to 17.4 S/m), an 85-fold increase compared to 0.2 S/m of the non-composited HKUST-1 (Table 1). A simple mixture of the MOF and the carbon in the same mass ratio leads to an 8.4 times conductance improvement (1.69 S/m). Importantly, the composite maintains an extremely high surface area (940 m2 g− 1). The composite's surface area and pore size distribution are a combination of the two materials considering the partial blockage of the mesopores of the activated carbon host by the MOF nanoparticles. Additionally, the size of the nanoparticles perfectly matches the size of the pores in which they have been grown. A large and accessible surface area is of great importance for the electrochemical reactions to enable good accessibility of the catalyzed compounds (Fig. 2b, for isotherms, see Figure S2).
Thermal Gravimetric analysis (TGA, Figure S3) demonstrates the good physical interactions between the MOF and the carbon. The excellent adhesion of the MOF onto the carbon scaffold increases its thermal stability. Un-composited HKUST-1 loses approximately half of its original mass at ⁓400⁰ C, whereas this loss is obtained at 50⁰ C higher when incorporated into the carbon. Furthermore, the electric interaction between the MOF and the carbon was previously proved by Electron Paramagnetic Resonance (EPR) spectroscopy measurements. The g-value of the Cu2+ unpaired electron of the composite was downshifted compared to that of the bare HKUST-1 (Figure S4). This shift toward free-electron values (g-value of 2.0023) indicates that the Cu2+ unpaired electron in the composite has gained a higher level of mobility (less localized to the Cu atoms). A higher level of freedom for electrons on the metal center proves the electric conductivity contributed from the carbon surrounding. The change in electronic structure imposed by the composite structure may affect the catalytic capabilities of HKUST-1, including yield and onset potential and different catalytic mechanisms resulting in various products. Moreover, by analyzing the different g-values recorded in the composite, we could simulate the average size of the MOF nanoparticles (considering HKUST-1 unit-cell of a diameter of 0.9 nm). Interestingly, the mean diameter of MOF nanoparticles is 6 nm and perfectly matches the average pore size of the carbon host, measured by gas adsorption measurement (Fig. 2b).26
Table 1
The lateral electric conductivity of HKUST-1/carbon composites with different MOF/carbon ratios presents the significant advantages of our composition method compared to the simple mixing of MOF with carbon.
Sample #
|
HKUST-1 content in sample
|
Composited with carbon (S m− 1)
|
Mixed with carbon (S m− 1)
|
1
|
Carbon
|
65
|
65
|
2
|
16%
|
17.2
|
1.69
|
3
|
23%
|
7.6
|
0.95
|
4
|
37%
|
0.2
|
0.09
|
5
|
HKUST-1
|
0.0003
|
0.0003
|
A thorough insight into the chemical impact of the composition on the electronic state of the MOF is revealed by X-ray photoelectron spectroscopy (XPS) measurements. XPS analyzes the binding energy of valance electrons to the surface atom and is therefore sensitive to the oxidation state of the elements. The full survey spectra, including peaks for C 1s, O 1s, and Cu 2p, are presented in Figure S5. The spectra of the activated carbon consist of 96 atomic percentages of carbon with no detected copper (Table S2). On the other hand, the spectra of the MOF comprise 68.4% carbon, 29.8% oxygen, and 2.3% copper, similar to the atomic ratio in HKUST-1. Predictably, the spectrum of the composite is an average of its two precursors. It contains one-quarter of the Cu compared to the MOF, which is in good agreement with the MOF/carbon ratio in the composition. The spectra were normalized according to the first fitting C 1s peak at 284.8 eV. Deconvolution of the C1s peak reveals the chemical interactions between the atoms. The C 1s spectra of the activated carbon consist mainly of carbon-carbon atoms (79%). On the other hand, the MOF spectra comprised 60% oxygen-bonded carbon atoms, typical of the carboxyl groups of the BTC linkers (Fig. 3a and Table S3). Interestingly, the binding energies of the corresponding peaks shift in each spectrum. While the binding energy of the C-O peak in HKUST-1 is 287.3 eV, the binding energy of the composite is shifted toward higher binding energy up to 288.27 eV. This is closer to the peak found in the activated carbon at 288.65 eV. The upshifting of the C 1s reveals higher oxidation of the carbon atoms in the composite compared to the original MOF. The O 1s deconvoluted peak follows the same trend, where the second peak, related to O-C atoms, is shifted from 532.6 eV in HKUST-1 to 533.7 eV in the composite (Fig. 3b). The most significant shift is found in the Cu 2p spectra (Fig. 3c). Importantly, this peak is absent in the activated carbon spectra and enables direct tracking of the impact of the composition on the electronic state of the MOF. The fitted XPS spectra of Cu confirm the existence of both Cu+ and Cu2+ with peaks at binding energies 933.9 and 953.3 eV corresponding to Cu(I) and peaks at 936.2 and 956.0 eV associated with Cu(II) of Cu 2p3/2 and Cu 2p1/2, respectively.28–31 All copper peaks shift to higher binding energies, attributed to the strong interactions between HKUST-1 and activated carbon. Here, the peak related to Cu2+ is upshifted from 936.2 eV in the original MOF to 938.0 eV in the composite. The upshifting of the peaks resulting from the electronic interactions of the MOF with the hosting carbon supports the interpretation of the EPR observations that were reconfirmed by electrochemical activity.22 When encapsulating the MOF guest, the more oxidized activated carbon acts as an electron withdrawer and applies a partial positive charge on the MOF, both at the chelating oxygen atoms and the copper centers. The electronic interactions ultimately lead to higher electron mobility and electric conductivity. The comparison of atomic ratios obtained from XPS spectra is shown in a histogram (Fig. 3d). Further analysis of the XPS spectra can be found in the supporting information.
To evaluate the electrochemical and electrocatalysis performances of the HKUST-1/C, the composite was operated as a cathode in a three-electrode cell. A conductive ink comprising HKUST-1/C, carbon black (CB) as conducting agent, and Nafion as an ion-conducting binder was prepared and drop-cast on carbon cloth as electrodes. The cyclic voltammetry (CV) of the electrode in a CO2 saturated KHCO3 solution shows a widely spaced oxidation/reduction couple corresponding to Cu+/Cu2+ at 0.76 and 0.45 V vs. RHE, respectively (Fig. 4a). An important parameter for high surface composited catalysts is the amount of electroactive coverage, i.e., the number of active sites available for electrochemical reaction in the material. The reduction wave in the CV at -0.05 V vs. Ag/AgCl can be used to measure the concentration of electroactive copper centers (Figure S5). Integrating the reduction wave of the CV at a scan rate of 10 mV s− 1 reveals that 0.015 Coulombs participated in the electrochemical reaction (Figure S5). Considering that this reaction involves only a single electron transfer, approximately 155 nmol of Cu2+ sites are exposed to the electrolyte solution and actively participate in the electrochemical reduction reaction. This is approximately 18% of the Cu atoms in the total electrode. The high electroactive coverage indicated exceptional high exposure of the metal centers to the electrolyte, among the highest ever reported for MOF-based cathodes (Table 2 and Table S1).
Linear Sweep Voltammetry (LSV) was employed to study the cathode performances upon the CO2 electroreduction reaction. When polarized to negative potentials, HKUST-1/carbon demonstrates an exceptionally high potential increase in negative current density, starting at an onset potential of -0.31 V (Fig. 4b, black line). Moreover, at -1.0 V vs. RHE, the LSV of the MOF/carbon reaches − 18 mA cm− 2 in the presence of CO2. This result is considerably higher than usually reported MOF-based beyond CO electrocatalysis (Table 2 and Table S1). The excellent performance could be realized due to extremely high electrochemical coverage to the particular structure of the MOF and the electrical activity originated by high amalgamation with the carbon host. To ensure that the cathodic current is originating from CO2 reduction (CO2 + 2H+ +2e− →CO + H2O) and not from the hydrogen evolution reaction (HER, 2H+ +2e− →H2), the electrodes were tested in a similar apparatus where the only difference was using NaCl as the supporting electrolyte. In this CO2-free electrolyte solution, the onset potential is shifted down to -0.52 V. Presumably, in the absence of carbonate ions, the reduction is associated only with HER, proving the good selectivity of the catalyst. (Fig. 4b, blue dashed line). LSV at lower scan rates (2 mVs− 1) with identical results confirms the source of this current is Faradaic and not capacitive (Figure S7).
The stability of the catalysts upon prolonged operation is a major challenge in electrocatalysis processes.24 To evaluate the stability of the electrocatalysis, we applied a constant potential (E = − 0.9 V vs. RHE) on a MOF/carbon composite electrode in a CO2-saturated solution (Fig. 4c). An average current density of -10.56 mA cm− 2 was measured at the initial stage of the reaction and − 11.85 mA cm− 2 after 12 hours. The sustained current density points out excellent chemical stability. The remarkable stability should be attributed to the stability effect projected by the carbon host. Durability was also tested by repeated LSV and CV experiments after 12 hours of chronoamperometry (Fig. 4d, a). Both measurements reveal an almost identical response for the fresh electrodes and the electrodes after 12 hours of constant potential. The XRD measurements were done before and after the chronoamperometry test (Figure S8). The results show the MOF/carbon composite retains its stable structure. These results are promising and encourage further study of MOF/carbon composites as catalysts for CO2RR in terms of both activity and selectivity.
To make CO2 reduction economically favorable, the products of the process must be valuable chemicals. It is worth noting that many studies measure the CO production and calculate the Faradaic efficiency according to the production of CO, which by itself is not a commodity chemical. In this study, we examined the formation of more highly reduced (hydrogenated) species; therefore, the identification of the products focused on NMR spectroscopy.
Table 2
Performance of different MOF-based cathodes for CO2 electroreduction.
Cathode
|
J,
mA cm− 2
|
E
V vs. SHE
|
Electrochemical coverage
nmoles cm− 2
|
Main product
|
Reference
|
MOF-545Fe/CB
|
1.2
|
-0.6
|
3
|
CO
|
32
|
Al2(OH)2TCCP-Co
|
1
|
-1.7
|
18
|
CO
|
15
|
COF-367-Co
|
3.3
|
-0.67
|
2
|
CO
|
33
|
MOF-1992/CB
|
16.6
|
-0.63
|
270
|
CO
|
25
|
CuPc
|
2.8
|
-1.6
|
N.A. c
|
C2H4
|
14
|
Fe-TPP
|
3
|
-1.0
|
0.1
|
CO
|
16
|
HKUST-1/carbon
|
18
|
-1.0
|
155
|
HCOOˉ
|
This work
|
c - could not be extracted from paper. |
To identify the products of the reaction, the electrolyte solutions of the CO2RR were collected from the reactor after 0.5-hour, 2 hour, 3 hour, 5 hour, and 10 hours. The products were analyzed in H-NMR spectroscopy and quantified using maleic acid as the internal standard. Formic acid was identified as the main product from the H-NMR spectra (Fig. 5). Importantly, products were obtained even in the first sampling after 30 minutes. The amount of each product varies during the 10 hours of reaction. The peak at δ = 8.3 ppm was attributed to formic acid. Two additional peaks at δ = 3.0 are formed from the initial stage of the reaction. These could be possibly attributed to central ethylene (CH2) in HOC = OCH2C = OOH. We couldn't be able to confirm the peak at δ = 3.0. We tried manually adding the possible product, such as malonic acid, into the NMR tube in the same vial after CO2 reduction (Figure S9). The malonic acid peak has a new peak which tells the formed species doesn't belong to malonic acid. Notably, the concentrations of formic acid product increase in the first three hours and decrease when the reaction continues for 12 hours (Figure S10). Interestingly, the decline in peak after 3 hours of reaction, implies a second electrochemical reaction that further reduced these initial products. It should be noted that when products are left in the reactor cell subjected to low potentials, further reduced products may result,34 and studies have identified situations where products may be seemingly lost from the reaction solutions, exacerbated by the reactions being carried out at low concentrations.35 Moreover, different potentials result in various products (Figure S11).[35] The faradaic efficiency of formic acid and hydrogen gas produced with different potentials is displayed in Figure S11. The formation of formic acid at lower potential is higher and reduces with an increase in potential because of the hydrogen evolution reaction (Figure S11). This result shows that CO2 electroreduction predominantly occurs at c V vs. RHE.
According to the products identified in NMR, we can hypostasize a reduction mechanism, resulting in formic acid as the main product. This mechanism is based on multiple recent publications using a similar copper-based catalyst and results in similar products. [36–39] The plausible mechanism of the HKUST-1/C composite is proposed in Fig. 7. In the initial reaction stage, the electrolyte containing CO2 gets adsorbed on the copper center and, upon negative potential, is radicalized to CO2.− on the HKUST-1/C surface. The adsorbed CO2.− could bind with the active sites on the catalyst surface and react with proton-electron pairs to form carboxyl (.COOH) intermediate, resulting in formic acid as the product.
Figure 6. A proposed mechanism for the CO2 electroreduction on HKUST-1/C composite.