Electroreduction of Carbon Dioxide by Heterogenized Cofacial Porphyrins

Great attention has been paid to cofacial porphyrins due to their many unique advantages over their monomeric analogs. However, their synthesis is usually complicated. In this work, a facile impregnation method for preparing heterogenized, cofacially stacked porphyrins is proposed. An anionic porphyrin is introduced as an underlayer for immobilization of cationic cobalt porphyrin via electrostatic force. The metal center of the underlying molecule contributes to the electronic structure of the upper cationic cobalt porphyrin. Screening reveals the anionic iron porphyrin to be the most efficient underlayer molecule, lowering the activation energy barrier of CO2 electroreduction, with an improved turnover frequency by 74% to 8.0 s−1 at − 0.6 V versus RHE.


Introduction
Metal porphyrins have received great attention as catalysts because of the clear molecular structures, which is beneficial for understanding the structure-activity relationships. Besides, the electronic structure of the catalytically active metal center, coordinated with four nitrogen atoms, could be effectively tuned by the surrounding ligands [1][2][3][4]. So far, there have been a variety of porphyrin-based molecular catalysts reported active for oxygen reduction reaction, oxygen evolution reaction, hydrogen evolution reaction, and CO 2 electroreduction (CO 2 RR) [5][6][7][8][9].
Motivated by natural enzymes with dual metal centers, such as cytochrome c oxidase and carbon monoxide dehydrogenase, much effort has been devoted to designing cofacial systems with two porphyrin units [10][11][12][13][14], probably because the presence of two metal centers can distribute the demands for redox and coordination numbers across two sites to favor reactions involving the multielectron transfer and a cofacial configuration could optimize the transition state energy level and therefore result in a more enhanced activity than their monomeric analogs [15]. In the past decade, the synthesis of numerous cofacial porphyrin catalysts has been reported for different reactions [16][17][18][19][20][21]. Taking the CO 2 RR, an essential reaction to mitigate greenhouse CO 2 emissions, as an example [22,23]. Mohamed et al. [24] reported a cofacial iron tetraphenyl porphyrin (FeTPP) dimer for electrochemical CO 2 -to-CO reduction in DMF/H 2 O solution, achieving high Faradaic efficiency (FE) of 95% and outstanding turnover frequency (TOF) number of 4300 s −1 at a moderate overpotential of 660 mV. They further assembled the cofacial molecule onto a fluorine-doped tin oxide electrode with a phosphonic acid anchor, showing high activity and stability for CO 2 RR in aqueous electrolytes [10]. Zahran et al. [25] demonstrated that the substituents on cofacial porphyrin rings greatly affected the catalysis process, and electron-donating groups such as mesityl could lead to increased TOF values. The concept of cofacial porphyrins has also been extended to fabricating the porous organic cage composed of six iron porphyrin units. The supramolecular catalyst produced CO 1 3 with near-quantitative Faradaic efficiency, with activities reaching 55250 turnovers [26].
However, the synthesis of cofacial porphyrins usually involves a series of complicated procedures preventing their large-scale production. Herein, the neutral tetraphenyl porphyrin molecules are functionalized with cationic methylanilinium (-N(CH 3 ) 3 + ) and anionic sulfonate groups (-SO 3 − ) (Scheme 1), with the electrostatic interaction between these two substituents driving the self-assembly of the cofacial porphyrin system. Characterization measurements, electrochemical analysis, and density functional theory (DFT) calculations reveal a strong electronic interaction between the metal centers, significantly decreasing the reaction barrier and resulting in a more improved CO 2 RR performance.

Catalyst Synthesis
Reduced graphene oxide (rGO, surface area 500 m 2 /g) was purchased from Tokyo Chemical Industry Co., Ltd. and used as received. Cobalt tetra-(4-N,N,N-trimethylanilinium) porphyrin (CoTMAP), cobalt tetrasulfonatophenyl porphyrin tetrasodium (CoTSPP), iron tetrasulfonatophenyl porphyrin tetrasodium (FeTSPP), and nickel tetrasulfonatophenyl porphyrin tetrasodium (NiTSPP) were synthesized using a reported procedure [27]. Cofacial porphyrins supported on rGO were prepared by two-step incipient wetness impregnation (Fig. 1a). Briefly, CoTSPP (1.5 mg) was dissolved in deionized water and impregnated on rGO (36.5 mg). The  b TEM image, STEM image, and the corresponding EDX maps for C, Co, and N of the as-prepared Co/Co-Por catalyst. c XPS spectrum showing the Co 2p 3/2 signals and d UV-Vis spectra for CoTMAP, CoTSPP, and Co/Co-Por mixture was aged for 12 h followed by vacuum drying at 80 ℃ overnight. A second impregnation step was then conducted to further impregnate CoTMAP (2 mg). Again, a 12-h aging step was performed to allow a sufficient interaction between the CoTMAP and CoTSPP molecules. The resulting catalyst, denoted as Co/Co-Por, was eventually dried in a vacuum oven at 80 ℃ overnight. The same procedures were also done to prepare the Ni/Co-Por and Fe/Co-Por catalysts, but using NiTSPP and FeTSPP, respectively, in the first impregnation step. Inductively coupled plasma-atomic emission spectroscopy (ICP-OES) was carried out to calculate the actual Co contents in Co/Co-Por, Ni/Co-Por, and Fe/ Co-Por of 0.51%, 0.24%, and 0.26%, respectively, which are close to their nominal values.

Catalyst Characterizations
ICP measurements were conducted using an Agilent 725 ICP-OES. Catalyst morphology was analyzed using a JEOL JEM-2100F field emission source transmission electron microscope (TEM) operating at 200 kV. X-ray photoelectron spectroscopy (XPS) analysis was obtained using a Thermo ESCALAB 250Xi spectrometer equipped with a monochromatic AlK radiation source (1486.6 eV, pass energy 20.0 eV). The diffuse reflectance UV-Vis spectrum was recorded in the 200-800 nm range using a UV-Vis spectrophotometer (Lambda 750, PerkinElmer, Inc.).

Electrochemical Measurements
To prepare the working electrode, the catalyst powder (5 mg) and Nafion solution (13 μL, 5 wt%, Sigma-Aldrich) were dispersed in ethanol (1 mL, Aladdin, ACS, ≥ 99.5%) and sonicated for 1 h to form a uniform catalyst ink. Then, the ink was drop cast onto carbon paper (Toray, TGP-H-060) to reach a catalyst loading of 0.1 mg/cm 2 . Its electrochemical performance in CO 2 reduction was tested in a customized H-cell separated by a Nafion 117 membrane. Platinum foil and a leak-free Ag/AgCl electrode (LF-2, Innovative Instrument Inc.) were used as the counter and reference electrodes, respectively. Before the experiments, sodium bicarbonate (1.75 mL, 0.5 M, Sigma-Aldrich, ≥ 99.5%) solution was injected into the working and counter compartments. CO 2 was purged into the chambers for 10 min before electrochemical polarization. Electrochemical responses were recorded by a CHI650E potentiostat. Electrolyte resistance between the reference and working electrodes was measured by potential electrochemical impedance spectroscopy and manually compensated. During the electrolysis, gas products were analyzed using an online gas chromatograph equipped with a Hayesep-D column, TCD detector, and FID detector.
The Faradaic efficiencies of gas products were calculated using the following equation: where i represents CO or H 2 ; x is the concentration of CO or H 2 ; ν is the feeding rate of CO 2 ; n is the electron transfer number (2, for both CO and H 2 ); F is the Faradaic constant (96,485 C/mol); and I is the average current during electrolysis.

Results and Discussion
Incipient wetness impregnation can allow anionic porphyrins to uniformly immobilize onto the rGO support through π-π interactions and sufficient interaction between the cationic porphyrins and anionic ones through electrostatic force. Figure 1b shows the TEM image of the as-prepared Co/Co-Por catalyst with a clean two-dimensional morphology. The energy-dispersive X-ray spectroscopy (EDS) maps also show the uniform distribution of C, N, and Co and no apparent aggregates. To verify, we also prepared the hybrid catalyst at a high loading by impregnating 15 mg of CoTSPP on 35 mg of rGO. The resulting catalyst shows many porphyrin aggregates with an estimated average size of 5-10 nm (Fig.  S1). From this perspective, a low porphyrin loading, which requires a dilute porphyrin-water solution for impregnation, is key to the successful preparation of the highly dispersed hybrid catalysts.
XPS and UV-Vis spectroscopy were performed to study the electronic structure of Co/Co-Por. Figure 1c shows the XPS Co 2p spectrum showing Co 2p 3/2 peaks at 781.7 and 780.6 eV for CoTMAP and CoTSPP, respectively. The different binding energies of the Co 2p 3/2 band are due to the distinct electronegativity features of the peripheral functionalities. The electron-withdrawing methylanilinium group decreases the electron density around the Co center, while the electron-donating sulfonate group increases the electron density around cobalt [27]. The two-porphyrin cofacial Co/Co-Por reveals a neutralized binding energy value of 781.3 eV for the Co 2p 3/2 signal, indicating the electrons within the stacked molecule system to be significantly redistributed, as confirmed by the shifted Soret band for Co/Co-Por (419.8 nm) as compared to CoTMAP (426.3 nm) or CoT-SPP (413.4 nm) (Fig. 1d). The Q bands for all three samples could hardly be distinguished from the strong background of rGO absorbance. The above results reveal the intimate contact between the CoTMAP and CoTSPP molecules. Ni/Co-Por and Fe/Co-Por were also prepared by introducing NiTSPP and FeTSPP as the sublayer porphyrins, respectively, for systematically investigating the structure-activity relationship of cofacial porphyrins during the electrochemical CO 2 reduction. For comparison, monomeric CoTMAP, CoTSPP, FeTSPP, and NiTSPP catalysts on rGO were also prepared at a content of 5 wt%. The EDS maps in Fig. S2 confirm the highly disperse NiTSPP and CoTMAP in Ni/ Co-Por and uniform distribution of FeTSPP and CoTMAP in Fe/Co-Por. In addition, the binding energies of the Co 2p 3/2 signal are also found to shift in these two catalysts due to the electronic interactions (Fig. S3). The steady-state electroreduction of CO 2 was conducted in a customized H-cell in a CO 2 -saturated 0.5 M NaHCO 3 electrolyte. Cyclic voltammetry measurements show prominent reduction currents at potentials more negative than approximately − 0.4 V versus RHE (Fig. S4). CoTMAP exhibits a stable current at a potential of − 0.6 V versus RHE with an average current density of 4.6 mA/cm 2 during 1 h of electrolysis (Fig. 2a, Fig.  S5). Moreover, the activities for cofacial porphyrins follow the order Fe/Co-Por (8.0 mA/cm 2 ) > Co/Co-Por (6.5 mA/ cm 2 ) > Ni/Co-Por (5.7 mA/cm 2 ). Current densities are much lower in the sulfonate group functionalized metal porphyrin than in the methylanilinium group functionalized CoTMAP catalyst. CoTSPP, NiTSPP, and FeTSPP have overall current densities of 1.4, 0.7, and 1.1 mA/cm 2 , respectively, at − 0.6 V versus RHE (Fig. S6), in agreement with our previous study that CO 2 reduction favors cationic substituents over neutral or anionic ones [27]. Excellent selectivities toward CO production are shown by all tested catalysts, with FE CO values over 90% (Fig. 2b, Fig. S7).
The resistances of the catalysts were investigated by AC impedance analysis at an open-circuit voltage (OCV). The Nyquist plots demonstrate that the four samples have similar interfacial charge transfer resistance (R ct ), possibly due to the identical and low catalyst loading. TOF values were calculated for investigating the intrinsic reactivity of each catalyst based on the number of CoTMAP molecules on the top layer. It could be argued that the underlying MTSPP may not be completely stacked (i.e., covered) and may still contribute to the reductive currents. Therefore, a set of H/Co-Por samples were prepared as a control with various H 2 TMAP loadings for the second impregnation step. As described above, CoTSPP alone delivers a j CO of 1.3 mA/cm 2 (Fig. 2c). The introduction of H 2 TMAP, which is inactive for CO production, suppresses the current densities. This serves as direct proof that the cationic porphyrins preferentially graft on the pre-deposited anionic porphyrins by electrostatic force and block the underlying metal centers from accessing the reactants. The partial current density of CO becomes negligible  (Fig. 2d), similar to that of Ref. [27]. Fabricating the cofacial porphyrins using MTSPP leads to improved intrinsic activities with the order of Fe/Co-Por (7.6 s −1 ) > Co/Co-Por (6.1 s −1 ) > Ni/Co-Por (5.4 s −1 ).
DFT calculations were performed for analyzing the thermochemical reaction energetics of CO 2 RR on CoTMAP, Co/ Co-Por, Ni/Co-Por, and Fe/Co-Por. The molecular structures were first optimized for the cofacial porphyrins according to the H-aggregation configuration. The introduced H 2 TMAP blocked the underlying metal sites (Fig. 2c), which implies a high overlapping degree (Fig. S8). CO 2 -to-CO electroreduction by molecular catalysts involves *COOH and *CO intermediates [32][33][34]. *COOH formation is usually considered the rate-determining step (RDS), and the *CO species are believed to desorb promptly to form the CO products. CoTMAP, Co/Co-Por, Ni/Co-Por, and Fe/Co-Por have Tafel slopes of 114.9, 123.8, 118.4, and 133.7 mV/dec (Fig. S9). Furthermore, a CO 2 reaction order of approximately 1 is determined for all the catalysts, indicating the CO 2 -to-*COOH transformation to be the rate-limiting step (Fig. S10) [9]. *COOH formation is the most energy-demanding, as suggested by the DFT results, associated with the highest energy barrier of + 0.6 eV compared to other reaction steps (Fig. 3a). Figure 3b shows the free energy diagrams for all CoTMAP-based catalysts. While all the cofacial porphyrins show the same RDS for CO 2 -to-CO conversion as the monomeric CoTMAP, their reaction energy barriers appear to be lower (Fig. S11). Then, the calculated theoretical reaction overpotentials over the tested catalysts follow the order Fe/Co-Por (0.58 V) < Co/Co-Por (0.60 V) < Ni/Co-Por (0.65 V) < CoTMAP (0.68 V). The above results show that, in addition to peripheral substituents, metal atoms from the underlying catalytically inactive porphyrin molecules also play an important role in tuning the electronic structure of the catalytically active porphyrin molecules on the top layer. Correspondingly, the log(TOF) values appear to be linearly correlated with the theoretical overpotential values.

Conclusions
In summary, a two-step impregnation method is proposed to prepare cofacially stacked porphyrins by utilizing the electrostatic force between the cationic and anionic porphyrin molecules, which not only avoids the common tedious chemical synthesis route but also effectively tunes the electronic structure of the molecular catalysts, eventually leading to a lower reaction energy barrier and a significantly improved reaction rate for CO 2 electroreduction. Our method should theoretically be adaptable to other carbonaceous supports such as carbon black and carbon nanotubes. In addition, the facile approach has the potential to guide the rational design of other cofacial porphyrin catalysts. Fig. 3 a Optimized structures for Co/Co-Por with adsorbed *COOH and *CO as well as free-standing CO (g) . b Free energy diagrams for CO production on CoTMAP, Co/Co-Por, Ni/Co-Por, and Fe/Co-Por. c The plot of log(TOF) against the theoretical overpotential values showing a linear correlation