Enhancing Electrocatalytic CO₂ Reduction of Nickel Porphyrin via Sulfur-Coordination and Through-Space Effects

Abstract

Metalloporphyrin-based complexes featuring stable tetrapyrrole structures and tunable coordination environments are promising electrocatalysts for CO₂ reduction. Modifying the first-coordination atoms that bonded with the central metal or changing the outer-sphere functional groups was found to affect the catalytic performance. However, the combined effect of these two strategies remains unexplored. Herein, using nickel tetraphenyl porphyrin (NiN₄-TPP, 1) as the original catalyst, we first investigated the effect of first-coordination modification by replacing one N atom with a S atom, yielding NiN₃S-TPP (2). Density functional theory calculations revealed that, as an electrocatalyst, NiN₃S-TPP outperforms NiN₄-TPP in converting CO₂ to CO. Notably, the S-substituted catalyst 2 exhibits a positively shifted reduction potentiala trend consistent with experimental observations. This is attributed to the lowered LUMO energy level after introducing the S atom. Moreover, the S-substitution stabilizes the [*–COOH] intermediate and shifts the rate-determining step from the protonation of [1–CO₂]⁻ to form [1–COOH]⁰ to the absorption of CO₂ on [2]⁻, reducing the barrier. Charge analysis reveals that the S atom donates electrons to the metal center and porphyrin ligands of the catalyst. Then, upon binding to COOH, these reserved electrons are transferred to COOH, thus facilitating the interaction and stabilizing the intermediate. Moreover, peripheral substituent modifications with the cationic –N­(Me)₃ ⁺ functional group yield great enhancement in the catalytic performance. It shifts the reduction potential to be more positive and further stabilizes the [*–COOH] intermediate via through-space electrostatic interaction. This work demonstrates how first- and peripheral-coordination modifications synergistically enhance CO₂ reduction catalysis, offering a strategy for designing efficient porphyrin-based electrocatalysts through core coordination environment regulation and through-space interactions.

1. Introduction

Electrochemical conversion of carbon dioxide (CO₂) into value-added hydrocarbons and chemical feedstocks is promising to alleviate greenhouse gas emission and energy crisis, and the utilization of renewable energy sources makes electrocatalytic CO₂ reduction an environmentally sustainable and eco-friendly approach. However, the inertness of CO₂, whose CO bond energy is ca. 750 kJ/mol, and its negative reduction potential (CO₂ + e^– → CO₂ ^•–, E _red = −1.9 V vs SHE), poses great challenges to the direct reduction of CO₂. Hence, an efficient electrocatalyst is required to facilitate the activation of CO₂ and promote its conversion to value-added products.

Metalloporphyrins have a stable tetrapyrrole-conjugated macrocyclic structure and an M–N₄ coordination environment. Here, M refers to the central metal, which is usually the active center. Metalloporphyrins serve as active centers in many biological enzymes, playing a central role in various life processes such as light-harvesting, oxygen transport, and catalysis. − Therefore, the development of catalysts based on metalloporphyrins has attracted considerable research interest. Metalloporphyrins, with M equals Fe, Co, or Ni, have been demonstrated as effective homogeneous catalysts for the reduction of CO₂. − Efforts have been made to improve their catalytic performance by regulating either the first or outer coordination spheres. − For instance, Choi et al. reported that by replacing the N₄-coordinated nickel tetraphenylporphyrin (NiN₄-TPP) to 21-oxatetraphenylporphyrin (NiN₃O-TPP), the performance of electrocatalytic CO₂ reduction to CO has been greatly enhanced, and the broken ligand-field symmetry was reported to be the key. He and co-workers adopted the same strategy and synthesized both NiN₃O- and NiN₃S-coordinated nickel porphyrin complexes and achieved improved performance in the photocatalytic reduction of CO₂ to CO. This strategy has also been successfully applied to many carbon-based M–N–C single-atom catalysts for CO₂ reduction reactions (CO₂RRs). − In addition, modifying the outer coordination layer by introducing peripheral substituents of the porphyrin rings was found to be an effective strategy. Savéant and co-workers introduced –OH, –OCH₃, –F, and –N⁺Me₃ substituents into iron tetraphenylporphyrins and enhanced the electrocatalytic performance for CO₂ reduction. ,− Manthiram et al. found that the CO₂RR performance of cobalt tetraphenylporphyrins can be benefited from electron-donating and cationic peripheral substituents. It has also been demonstrated that introducing urea or pendant pyridine groups into iron porphyrins can stabilize *CO₂ or *COOH intermediates through hydrogen-bonding interactions, thereby promoting CO₂ reduction. Nevertheless, to the best of our knowledge, the study of peripheral substituents is limited to the original, i.e., N₄-coordinated, metalloporphyrins.

Inspired by previous works, herein, we investigated how modifying both the first and outer coordination spheres affects the CO₂-to-CO catalytic performance of a nickel-base porphyrin. As shown in Scheme , we first targeted on the nickel tetraphenylporphyrin (NiN₄-TPP, 1) and N₃S-coordinated nickel tetraphenylporphyrin (NiN₃S­(-Cl)-TPP, i.e. 2–Cl), and both have been experimentally synthesized and have good stability under electrochemical conditions. − Through density functional theory computations, we examined their electrocatalytic performance of reducing CO₂ to CO and unveiled the underlying mechanism that why the N₃S-coordinated system outperforms the NiN₄ system. To our knowledge, in most previous computational studies of NiN₃X–porphyrin systems, the CO₂RR was assumed to follow concerted proton–electron transfer (CPET) and ignored the possibility of separated proton–electron transfer (SPET) mechanism. ,,,, The selection of CPET and SPET pathways depends on the relative rate of proton and electron transfer steps, which are closely related to the nature of the catalyst, as well as the local pH and applied potential under practical working conditions. Typically, for molecular catalysts, decoupled ET and PT transfer steps are expected. However, CPET was also reported for systems like cobalt porphyrin systems under acidic conditions, where the affirmation of such mechanism was the detection of key intermediates in experiments. , Nevertheless, there is no simple and accurate computational methods available to address the competition between SPET and CPET, partly due to the difficulty in treating the effects of pH, applied potential, and solvents on both the thermodynamics and kinetics of these steps. Hence, in this work, the electron transfer and protonation steps were treated separately, and the possible role of CPET will be discussed. Then, on top of the NiN₃S–TPP complex, we examined the peripheral substituent effects by introducing multiple functional groups. Calculations predict that the cationic –N­(Me)₃ ⁺ group can further improve the catalytic performance by shifting the reduction potential and lowering the overall barrier, in which the through-space effect plays a role. This work exemplified that a good combination of first and outer coordination sphere modifications can jointly enhance the catalytic capability of the metalloporphyrins.

1. Schematic Structures of Nickel Tetraphenylporphyrin (NiN₄-TPP, 1) and N₃S-Coordinated Nickel Tetraphenylporphyrin (NiN₃S-TPP, 2)­ .

^a R represents the substituent at the para-position of the phenyl group.

2. Computational Methods

2.1. Density Functional Theory Calculations

The ωB97X-D functional was used throughout this work. The computations were performed using Gaussian 16 software. Structures were optimized with a combined basis set (denoted as BS1) in an aqueous solution using the SMD implicit solvent model. BS1 refers to the 6-31+G­(d,p) basis set for H, C, N, O, S, and Cl elements and the LANL2DZ basis set and pseudopotential for Ni. Vibrational frequencies were computed to confirm the nature of stationary points, where the minima have no imaginary frequency and transition states (TSs) have one imaginary frequency. On top of the optimized structures, single-point calculations were performed with a larger combined basis set (denoted as BS2) in an aqueous solution. BS2 refers to 6-311++G­(d,p) for Cl, C, N, S, and H and LANL2DZ for Ni. The choice of this method is based on the good agreement between the calculated reduction potential of NiN₄-TPP and NiN₃S-TPP with available experiments (Table S1). Specifically, the calculated reduction potentials for NiN₄-TPP and NiN₃S-TPP are −1.14 V and −0.75 V, respectively, which matches well with the experimental values of −1.08 V and −0.58 V, and correctly reflects the trend of change. ,

The Gibbs free energy of a species in the aqueous phase, G _aq, is calculated as

$$G_{\mathrm{a}\mathrm{q}}=E_{\mathrm{a}\mathrm{q},\mathrm{B}\mathrm{S}2}+\mathrm{\Delta }{G}_{\mathrm{c}\mathrm{o}\mathrm{r}\mathrm{r}\mathrm{e}\mathrm{c}\mathrm{t},\mathrm{a}\mathrm{q},\mathrm{B}\mathrm{S}1}+1.89\mathrm{k}\mathrm{c}\mathrm{a}\mathrm{l}/\mathrm{m}\mathrm{o}\mathrm{l}$$ 1

where E _aq,BS2 is the single point energy calculated in the aqueous phase using BS2 on top of the BS1-optimized structures in the aqueous phase; ΔG _{correct,aq,BS1} is thermochemical correction to the Gibbs free energy as computed in the aqueous phase under 298.15 K using BS1. A concentration correction of 1.89 kcal/mol, i.e., RT ln­(nRT/p), is added to all species except gaseous molecules (CO₂, CO, H₂) at 298.15 K. This correction accounts for the change in free energy from 1 atm (24.5 L/mol in the ideal gas phase) to 1 M (1 mol/L in an aqueous solution). The concentration of H₂O is further corrected by 4.27 kcal/mol (standard concentration is 55.56 mol/L). In addition, to reduce the error in calculating the entropy effect at low frequencies, the free energy was corrected, and all vibrational frequencies below 50 cm^–1 were replaced with values of 50 cm^–1. Natural population analysis (NPA) was performed for key intermediates using the BS2 basis set.

2.2. Reduction Potential and pK _a

The reduction potentials were calculated by using the standard hydrogen electrode (SHE, experimental value of 4.281 V) as the reference, corresponding to an electron affinity of 98.72 kcal/mol. The reduction potential E _red vs SHE of a species is calculated as

$$E_{\mathrm{r}\mathrm{e}\mathrm{d}}=-\mathrm{\Delta }{G}_{\mathrm{r}\mathrm{e}\mathrm{d}}/(nF)-4.281\mathrm{V}$$ 2

where ΔG _red is the Gibbs free-energy change from the oxidation state to reduction state as calculated in an aqueous solution; n is the number of transferred electrons; F is Faraday’s constant, 96,485 C/mol. The free-energy change under applied potential U is computed as −nF·(E _red – U). To match the experimental work of CO₂ reduction by NiN₄- and NiN₃O-TPP systems, U is selected as −1.03 V vs SHE.

The pK _a value is calculated to determine the state of the species at a given pH. If the pK _a of the species is greater than the pH, the species will exist in the protonated form. The pK _a is calculated from the Gibbs free-energy change of the deprotonation step of the species. For protonated species AH⁺, the reaction of the deprotonation step is

$${\mathrm{A}\mathrm{H}}^{+}(\mathrm{s}\mathrm{o}\mathrm{l})={\mathrm{H}}^{+}(\mathrm{s}\mathrm{o}\mathrm{l})+\mathrm{A}(\mathrm{s}\mathrm{o}\mathrm{l})$$ 3

The Gibbs free-energy change for the deprotonation of AH⁺, ΔG _aq, is calculated as

$$\mathrm{\Delta }{G}_{\mathrm{a}\mathrm{q}}=G_{\mathrm{a}\mathrm{q}}({\mathrm{H}}^{+})+G_{\mathrm{a}\mathrm{q}}(\mathrm{A})-G_{\mathrm{a}\mathrm{q}}({\mathrm{A}\mathrm{H}}^{+})$$ 4

The pK _a is calculated as

$$\mathrm{p}{K}_{\mathrm{a}}({\mathrm{A}\mathrm{H}}^{+})=\mathrm{\Delta }{G}_{\mathrm{a}\mathrm{q}}/RT\ln 10$$ 5

where G _aq(A) and G _aq(AH⁺) are the Gibbs energies in the aqueous phase of A and AH⁺, respectively. G _aq(H⁺) is the Gibbs free energy of the proton in the aqueous phase at pH = 0. The Gibbs free energy of the proton in the gas phase is −6.27 kcal/mol, and the experimental absolute solvation energy of the proton in the aqueous phase is −264.0 kcal/mol. So, the G _aq(H⁺) is −270.27 kcal/mol. Considering the effect of pH, the Gibbs free energy of the proton in the aqueous phase is (−270.27 – 1.364 pH) kcal/mol. In this work, we set pH = 7.2 to match the experimental work, so the total Gibbs free energy of the proton is −280.1 kcal/mol.

2.3. Energy Decomposition Analysis (EDA)

EDA was performed with the Amsterdam density functional (ADF) module of the AMS software package ωB97X functional with the TZ2P basis set on top of the geometries optimized using ωB97X-D/6-31+G­(d,p) as described in Section . The binding energy of two fragments, ΔE _b, is defined as

$$\mathrm{\Delta }{E}_{\mathrm{b}}=\mathrm{\Delta }{E}_{\mathrm{p}\mathrm{r}\mathrm{e}\mathrm{p}}+\mathrm{\Delta }{E}_{\mathrm{int}}$$ 6

where the preparation energy ΔE _prep is the amount of energy required to deform the fragments from their equilibrium structures to the geometry that they acquire in the overall complex. The interaction energy ΔE _int corresponds to the energy change when the geometrically deformed fragments are combined to form the overall complex. The interaction ΔE _int is further decomposed into the electrostatic attraction, the Pauli repulsion, and the orbital interactions.

$$\mathrm{\Delta }{E}_{\mathrm{int}}=\mathrm{\Delta }{E}_{\mathrm{e}\mathrm{l}\mathrm{a}\mathrm{t}\mathrm{a}\mathrm{t}}+\mathrm{\Delta }{E}_{\mathrm{P}\mathrm{a}\mathrm{u}\mathrm{l}\mathrm{i}}+\mathrm{\Delta }{E}_{\mathrm{o}\mathrm{i}}$$ 7

where the electrostatic interaction energy ΔE _elatat is usually attractive, and it is the energy between the unperturbed charge distributions of the prepared fragments. The Pauli repulsion energy ΔE _Pauli is responsible for steric repulsion, and it consists of the destabilizing interactions between occupied orbitals of the fragments. The orbital interaction energy ΔE _oi accounts for the charge transfer and polarization.

2.4. Nomenclature

For the sake of clarity, we adopted the following naming convention. The catalyst nickel tetraphenylporphyrin, NiN₄-TPP, is labeled as 1 and N₃S-coordinated nickel tetraphenylporphyrin, NiN₃S-TPP, is labeled as 2. A species is labeled in the format of ^2S+1[1 ^X –L] ^Q , where (2S + 1) gives spin multiplicity, Q is the charge, X is the assigned oxidation state of central metal Ni, and L denotes the species that was added to the catalyst. For example, ¹[1 ^II–CO₂]⁰ describes the neutral singlet species with which CO₂ binds to NiN₄-TPP, and the oxidation state of Ni is assigned to be 2.

3. Results and Discussion

We started from the nickel tetraphenylporphyrin (NiN₄-TPP, 1) and N₃S-coordinated nickel tetraphenylporphyrin with a Cl ligand (NiN₃S­(-Cl)-TPP, i.e., 2–Cl), and both have been synthesized and maintain good stability under electrochemical conditions. − Calculations show that NiN₄-TPP has a singlet ground state ¹[1]⁰, where the triplet and quintet are 2.5 and 30.5 kcal/mol higher in energy, respectively. As shown in Figure a, ¹[1]⁰ has a planar NiN₄ motif.

1.

Optimized structures of the ground state of (a) 1 and (b) 2–Cl. Relative Gibbs free energies (in kcal/mol) are given for each species. Bond distances (in Å) are labeled in black. Spin populations on Ni and the porphyrin ligand (Por) are in red.

Replacing one N atom of the NiN₄ motif by a S atom, along with the addition of a Cl ligand to maintain the oxidation state of Ni­(II), distorts the planar geometry. Due to its large radius, the S atom lies outside the NiN₃ plane of ³[2 ^II–Cl]⁰ (Figure b). The ground state of ³[2 ^II–Cl]⁰ is a triplet, where the singlet and quintet are 19.3 and 22.7 kcal/mol higher in energy, respectively. The calculated ground-state structures of ¹[1 ^II]⁰ and ³[2 ^II–Cl]⁰ agree well with experimental data (Tables S2–S5).

The net reaction of reducing CO₂ to CO by a catalyst requires two protons and two electrons. Depending on the order of protonation, reduction, and CO₂ adsorption, the reaction may proceed through multiple pathways (Scheme ). In the following, we examined each possibility to reveal the most feasible path for catalysts 1 and 2. The energetics of all the computed species and reaction steps are present in Tables S6–S11.

2. Possible 2H⁺/2e^– Paths for Reducing CO₂ to CO by Catalyst 1 or 2 .

^a Note that the bold Ni in the scheme represents either NiN₄- or NiN₃S-coordinated TPP. Dashed lines refer to CPET. The initial species for NiN₄-TPP is [Ni]⁰, while for NiN₃S-TPP, it is [Ni–Cl]⁰.

3.1. Reduction of CO₂ to CO by Catalyst 1

Figure depicts the Gibbs free-energy diagram of the CO₂-to-CO pathway on catalyst NiN₄-TPP (1). The optimized structures are shown in Figure S1. The adsorption of CO₂ is a prerequisite for its reduction. The adsorption of CO₂ to form ¹[1···CO₂]⁰ is 9.0 kcal/mol uphill, where the CO₂ maintains linear configuration and the Ni–C­(O₂) distance is 3.59 Å (Figure S5a). The protonation of ¹[1···CO₂]⁰ to form ³[1–COOH]⁺ is endothermic by 72.1 kcal/mol at pH = 7.2, and the calculated pK _a of ¹[1–COOH]⁺ is −19.5. This suggests that ¹[1–COOH]⁺ is difficult to form through direct protonation of ¹[1···CO₂]⁰. Alternatively, directly reducing [1]⁰ to [1]⁻ has a reduction potential of −1.14 V, agreeing fairly well with the experimental value of −1.08 V; and this step is endothermic by 2.5 kcal/mol under U = −1.03 V. The ground state of [1]⁻ is a doublet, whereas the quartet and the sextet are 31.7 and 30.7 kcal/mol higher in energy, respectively. For ²[1]⁻, the spin population is 0.88 on Ni and 0.12 on the porphyrin ligand (Figure S2a), suggesting Ni was reduced, consistent with the experiment. Successive reducing [1]⁻ to [1]^2– is unlikely to happen as the calculated reduction potential is −1.45 V. The adsorption of CO₂ on ²[1 ^I]⁻ to form ²[1–CO₂]⁻ is 7.0 kcal/mol uphill, and the CO₂ moiety has a bent configuration. Attempts to locate the transition state were failed, and we consider this adsorption of CO₂ to be a diffusion-controlled step (Figure S3), similar to what has been reported for the Fe porphyrin system by Liao’s group. Accordingly, the first step of catalyst 1 is more likely to be a one-electron reduction.

2.

Gibbs free-energy diagram (in kcal/mol) of electroreduction of the CO₂-to-CO pathway on catalyst 1 (NiN₄-TPP) under an applied potential of −1.03 V and pH = 7.2 in an aqueous solution. Black line denotes the most favored pathway. Red arrows in the structures represent the unpaired electrons, either on Ni or the porphyrin ligand.

Starting from ²[1–CO₂]⁻, its reduction to form [1–CO₂]^2– has a reduction potential of −1.70 V; hence, it is unlikely to happen. Meanwhile, the protonation step to form [1–COOH]⁰ is endothermic by 13.0 kcal/mol at pH = 7.2, and this is the rate-determining step. The calculated pK _a of ²[1–COOH]⁰ is −2.3, implying that ²[1–COOH]⁰ hardly exists at a pH of 7.2. This supports the experimental observation that CO production by catalyst 1 is almost not detected (FE_CO <2%) at a working potential of −1.03 V. Suppose ²[1–COOH]⁰ is formed, it can undergo one-electron reduction to form [1–COOH]⁻, with an E _red of 0.11 V corresponding to an exothermicity of 26.3 kcal/mol under U = −1.03 V. Then, ³[1–COOH]⁻ receives a proton and releases a H₂O molecule to form ³[1–CO]⁰. This step is 26.3 kcal/mol downhill. Finally, the desorption of CO is also exothermic by 10.2 kcal/mol, making it feasible to regenerate catalyst 1.

In brief, when SPET steps are concerned, the favored CO₂RR pathway for catalyst 1 (Figure , black path) is the adsorption and reduction of CO₂, followed by protonation, reduction, protonation, and release of H₂O and CO in sequence. The rate-determining step (RDS) is the first protonation step that forms ²[1–COOH]⁰, which is 13.0 kcal/mol uphill under pH = 7.2, and the overall barrier of this path is 22.5 kcal/mol. It is worth mentioning that ²[1–COOH]⁰ has a low pK _a value of −2.3, so it can hardly exist at a pH of 7.2. Therefore, catalyst 1 cannot effectively reduce CO₂ to CO by this stepwise mechanism.

Meanwhile, we also note that if CPET is concerned, the step of ²[1–CO₂]⁻ + H⁺ + e^– → [1–COOH]⁻ is 13.3 kcal/mol downhill. This CPET step circumvents the high energy barrier required for the formation of ²[1–COOH]⁰, the resultant RDS becomes the adsorption of CO₂ on the reduced [1]⁻, and the overall barrier is 9.5 kcal/mol. Given such a low energy barrier, the experimental lack of CO formation during CO₂ electroreduction on catalyst 1 is unexplained, implying the PT and ET steps may not be concerted. Nevertheless, the computed barrier of catalyst 2 is lower than this value, as we will discuss in the next section.

3.2. Reduction of CO₂ to CO by Catalyst 2

As for the S-substituted catalyst, ³[2 ^II–Cl]⁰, calculations indicate it first undergoes reduction and then dechlorination to form intermediate ²[2]⁰ (Figure S4). The calculated reduction potential of ³[2–Cl]⁰ to ⁴[2–Cl]⁻ was −0.48 V, giving an exothermicity of 12.7 kcal/mol under U = −1.03 V. The dissociation of Cl^– to form ²[2]⁰ was 6.7 kcal/mol uphill in free energy. The ground state of [2]⁰ is a doublet, and the quartet state and the sextet state are 39.2 and 68.3 kcal/mol higher in energy, respectively. For ²[2]⁰, the spin population is 0.82 on Ni and 0.18 on the porphyrin ligand, confirming that Ni^II in ³[2–Cl]⁰ was reduced to Ni^I in ²[2]⁰ (Figure S2e).

Starting from 2, the adsorption of CO₂ to form [2···CO₂]⁰ is 5.2 kcal/mol uphill. This is similar to the case of [1···CO₂]⁰, where CO₂ maintains a linear structure on catalyst 2, with a Ni–C­(O₂) distance of 3.49 Å (Figure S5c). Subsequently, one-electron reduction enables the activation of CO₂, forming [2–CO₂]⁻. The reduction potential of this process is calculated to be −0.97 V, and this step is exothermic by 1.4 kcal/mol given an applied potential U of −1.03 V vs SHE. Alternatively, the reduction of [2]⁰ to [2]⁻, with the calculated reduction potential of −0.75 V, is exothermic by 6.6 kcal/mol under U = −1.03 V. Calculation suggested that changing the coordination environment from N₄ to N₃S leads to a significant positive shift in the reduction potential of the catalysts. This agrees with what was reported in the experimental work by He et al., in which the synthesized [NiN₃S-TPP–(COOH)₄] catalyst has a less negative E _red (−0.66 V) than its NiN₄-parent catalyst (−1.02 V).

The ground state of [2]⁻ is a triplet, and the corresponding singlet and quintet states are 6.0 and 49.4 kcal/mol higher in energy, respectively. For ³[2]⁻, the spin population is 0.87 on Ni and 1.13 on the porphyrin ligand (Figure S2f), implying that an electron is added to the porphyrin ligand, while Ni remains in the Ni^I state. Successively reducing [2]⁻ to [2]^2– is unlikely to happen as the calculated reduction potential is −1.64 V. The nucleophilic attack of CO₂ by [2]⁻ to form ³[2–CO₂]⁻ is 10.4 kcal/mol uphill and with a mild barrier of 10.5 kcal/mol ([2–TS1]⁻, Figure S6). Hence, in comparison, for catalyst 2, it is more likely to absorb CO₂ first and then undergo a reduction step to form ³[2–CO₂]⁻. Meanwhile, reverse order is also possible.

Thereafter, ³[2–CO₂]⁻ can either undergo one-electron reduction or protonation. The reduction of ³[2–CO₂]⁻ to ²[2–CO₂]^2– has a reduction potential of −1.13 V and the corresponding free energy change is 2.3 kcal/mol uphill. The followed protonation of ²[2–CO₂]^2– to form ⁴[2–COOH]⁻ is exoergic by 14.8 kcal/mol at pH = 7.2. Alternatively, protonation of [2–CO₂]⁻ to form [2–COOH]⁰ is 1.0 kcal/mol downhill, and the followed electron reduction to ⁴[2–COOH]⁻, with a reduction potential of −0.53 V, is 11.5 kcal/mol downhill. The pK _a values of ³[2–COOH]⁰ and ⁴[2–COOH]⁻ are 8.0 and 18.1, respectively, implying they are stable at pH = 7.2. Then, similar to the case of ³[1–COOH]⁻, the followed protonation and desorption of H₂O and CO are exothermic, and finally the catalyst ²[2]⁰ is regenerated, taking into account the CPET for the [2–CO₂]⁻ to [2–COOH]⁻ does not change the pathway.

In comparison, the favored CO₂RR pathway of catalyst 2 (Figure , black lines) proceeds in the sequence of CO₂ adsorption, reduction, protonation, reduction, protonation and desorption of H₂O, and the desorption of CO. The RDS is the adsorption of CO₂ on [2]⁰ with a barrier of 5.2 kcal/mol. If the first two steps are reversed, then the barrier is 10.5 kcal/mol. Both are much lower than the SPET barrier of catalyst 1, i.e., 22.5 kcal/mol, and the former is lower than the CPET barrier of catalyst 1, i.e., 9.5 kcal/mol.

3.

Gibbs free-energy diagram (in kcal/mol) of electroreduction of the CO₂-to-CO pathway on 2 (NiN₃S-TPP) under an applied potential of −1.03 V and pH of 7.2 in an aqueous solution. Black line denotes the most favored pathway. Red arrows in the structures represent the unpaired electrons, either on Ni or the porphyrin ligand.

The above mechanism treats the metal center as the active site. Here, we also considered that N atoms in the porphyrin ligand act as the active site. As shown in Scheme , N atoms are at either para- or ortho-position of the S atom; hence, either p-N or o-N of ³[2]⁻ can be protonated (Figure S7a–f). Upon protonation, ³[2–(o-N)­H]⁰ is 5.2 kcal/mol more stable than ³[2–(p-N)­H]⁰, so the following discussion focuses on ³[2–(o-N)­H]⁰. Starting from it, the adsorption of CO₂ to form ³[2–(o-N)­H···CO₂]⁰ is 5.1 kcal/mol uphill, where the CO₂ maintains linear configuration and the Ni–C­(O₂) distance is 3.42 Å (Figure S7g). Thereafter, the proton on N can directly transfer to the O atom of the CO₂ part to form ³[2–COOH]⁰. This proton transfer process is 4.2 kcal/mol uphill but has a high barrier of 26.4 kcal/mol (Figure S8). This high barrier prohibits this path. Alternatively, ³[2–(o-N)­H–CO₂]⁰ can be reduced to ²[2–(o-N)­H–CO₂]⁻, which is endoergic by 3.9 kcal/mol under U = −1.03 V. Figure displays this pathway in red. Subsequently, the intramolecular proton transfer process of proton on N transferring to the O atom of CO₂ is exothermic by 11.2 kcal/mol with a barrier of 4.3 kcal/mol ([2–TS2]⁻), culminating in the formation of ²[2–COOH]⁻. Thereafter, the following protonation and desorption of H₂O and CO are exothermic, and finally, the catalyst [2]⁰ is regenerated. In comparison, this N-mediated pathway (with an overall barrier of 13.4 kcal/mol) is less favored than the metal-mediated pathway (with an overall barrier of 5.2 kcal/mol).

3.3. Comparison between Catalysts 1 and 2

Taking together, for catalyst 1, the most feasible CO₂-to-CO reaction mechanism is following the order of reduction → CO₂ adsorption → protonation → reduction → protonation → CO release. This mechanism is also feasible for catalyst 2, but reversing the order of the first two steps will be more favorable (Figure a). Impressively, substituting N₄ to N₃S led the reduction potential to shift positively from −1.14 V to −0.75 V and significantly lowered the overall barrier from 22.5 to 5.2 kcal/mol, suggesting catalyst 2 is a better catalyst for CO₂RR under the studied condition.

4.

(a) Proposed CO₂RR mechanisms on catalysts 1 and 2. (b) DFT-calculated Gibbs free-energy profiles on catalyst 1 and 2 when a potential of −1.03 V vs SHE is applied. RDS refers to the rate-determining step. Arrows above *COOH intermediates highlight that the improved performance of catalyst 2 arises from the greater stabilization of 2–COOH as compared that of catalyst 1.

The N₃S-coordinated catalyst 2 exhibits a notably more positive reduction potential, and this can be elucidated by analyzing the frontier molecular orbitals (FMOs) as illustrated in Figure a. For ¹[1]⁰, the lowest unoccupied molecular orbital (LUMO) is a σ* orbital formed by the Ni d _{x ²–y ²} orbital and N 2p orbitals, implying that, upon reduction, the electron will be added to the Ni center. For ²[2]⁰, the β-LUMO is a ligand−π* orbital, indicating that the electron is added into the ligand. That is to say, the porphyrin ligand in 2 is noninnocent during the reduction step, in contrast to the case of 1. The incorporation of a S atom, which differs from N in electronegativity and atomic orbital characteristics, modifies the FMO composition and energy. Specifically, the introduction of S leads to a lowering of the LUMO energy level, it changes from −1.26 eV for LUMO of 1 to −1.56 eV for β-LUMO of 2. A lower LUMO energy signifies a higher electron affinity, meaning that the molecule can accept electrons more readily, which accounts for a more positive reduction potential. This more positive reduction potential makes catalyst 2 a better electrocatalyst.

5.

(a) LUMOs of 1 and 2. (b) Optimized structures of ²[1–COOH]⁰ and ³[2–COOH]⁰. NPA charges on the COOH group are shown. Arrows represent the direction of electron transfer. (c) EDA energies between the COOH group and bare catalyst in ²[1–COOH] (left) and ³[2–COOH] (right). Energies are given in kcal/mol. (d) HOMOs of ²[1–COOH]⁰ and ³[2–COOH]⁰.

For the reaction energy barrier, when the coordination environment changes from N₄ to N₃S, the rate-determining step (RDS) changes (Figure b). The RDS for catalyst 1 is protonation of [1–CO₂]⁻ to [1–COOH], whereas the RDS for catalyst 2 is CO₂ adsorption. This change arises from stabilization of the neutral [*COOH] species by 2. We added an arrow to the top of [*COOH] in Figure b to highlight its important role. To understand the origin of the extra stabilization by the NiN₃S core in 2 relative to NiN₄ in 1, we investigated the properties of [1–COOH]⁰ and [2–COOH]⁰.

We first compared the binding energy E _b between the bare catalyst and the COOH-group. The value of E _b is more negative for 2 (−33.8 kcal/mol) than that for 1 (−16.3 kcal/mol), indicating a stronger interaction between 2 and the COOH group. This is also supported by the natural population analysis (NPA). The NPA charge on the COOH moiety is more negative for ³[2–COOH]⁰ (−0.48 |e|) than that for ²[1–COOH]⁰ (0.12 |e|) (Table ), implying that more electrons were transferred to COOH within ³[2–COOH]⁰, thus making it more stable. To determine the source of additional transferred electrons, we tracked the NPA charges from the bare catalysts to [1/2–COOH]⁰ complexes. As shown in Table , the NPA charge values of Ni atom, N atoms, and the porphyrin ligand that exclude N or S atoms are smaller in 2 than those in 1. For instance, the average NPA charge of N atoms in 2 is −0.63 |e|, which is 0.03 |e| smaller than that in 1, −0.60 |e|. This is because the S atom donates electrons to its surrounding atoms, and the NPA charge of the S atom in 2 is 0.54 |e|. Note this is a positive value, and this makes a significant difference as it binds to COOH. Upon COOH binding to catalyst 1, the COOH group transfers ∼0.12 |e| to the N atom of the catalyst, and the NPA charges of other parts barely change. In contrast, upon binding of COOH to catalyst 2, a considerable amount of electron is transferred from Ni (0.21 |e|) and the porphyrin ligand (∼0.29 |e|) to the COOH group. Hence, we concluded that strengthened interaction between COOH and catalyst 2 is because the substituted S atom donates electrons to the central metal and porphyrin ligands of the catalyst; then upon binding to COOH, these reserved electrons transfer to COOH and facilitate the interaction.

1. NPA Charges of ²[1–COOH]⁰ and ³[2–COOH]⁰ .

	•COOH (|e|)	Ni (|e|)	S (|e|)	N1 (|e|)	N2 (|e|)	N3 (|e|)	N4 (|e|)	Por-Ligand (|e|)
1[1]0	/	0.71	/	–0.60	–0.60	–0.60	–0.60	1.69
2[2]0	/	0.37	0.54	–0.63	–0.64	–0.62		0.98
2[1–COOH]0	0.12	0.75	/	–0.69	–0.63	–0.63	–0.64	1.72
3[2–COOH]0	–0.48	0.58	0.48	–0.63	–0.62	–0.60		1.27

^aThe N and S atoms were excluded from the Por-Ligands.

Furthermore, we also performed energy decomposition analysis (EDA) to gain more insights into the interaction. As described in the Computational Methods section, the interaction energy between COOH and the catalyst is decomposed to electrostatic attraction (ΔE _elatat), the Pauli repulsion (ΔE _Pauli), and the orbital interactions (ΔE _oi). In comparison (Figure c), among the three energy compositions, the most prominent differences occur in E _oi and the E _Pauli. The more negative E _oi between 2 and COOH (−71.3 kcal/mol) than that in ²[1–COOH]⁰ (−40.9 kcal/mol) is the primary contribution to the more negative binding energy of ³[2–COOH]⁰. This enhanced orbital interaction is primarily attributed to the increased orbital overlap and the consequent strengthened electron transfer between 2 and COOH. The molecular orbitals (Figure d) show a higher electron density on the C 2s orbitals and O 2p _y orbitals of the COOH group in ³[2–COOH]⁰ than that in ²[1–COOH]⁰. This is consistent with the NPA charge results. In addition, the weaker E _Pauli between 2 and COOH can be attributed to two factors. On the one hand, the S atom in 2 is electron deficient as compared to the N1 atom counterparts in 1, so it experiences less repulsion as it interacts with the electrons on the O atom of COOH. On the one hand, the S···O distance (3.67 Å) in ³[2–COOH]⁰ is longer than the corresponding N···O distance (3.03 Å) in ²[1–COOH]⁰. This can be attributed to the out-of-plane feature of the S atom, where the N₃S framework is more flexible than the N₄ framework. Taken together, the introduction of the S atom into the first coordination sphere redistributes the electron and increases the distance between the repelling atoms within the critical [*COOH]⁰ intermediate. Consequently, the increased orbital interaction and reduced Pauli repulsion stabilizes [*COOH]⁰ and thus promotes the CO₂ reduction performance of the catalyst.

Tracking the location of the spin densities indicates that the porphyrin ligands in catalyst 2 are noninnocent during the catalytic cycle, whereas they are innocent in catalyst 1. The spin density on the porphyrin ligand of ²[2]⁰ is 0.18, which increases to 1.13 in ³[2]⁻, 1.30 in ³[2–CO₂]⁻, and 1.24 in ⁴[2–COOH]⁻ (Table S14). The corresponding values in 1 barely change within 0.3 (Table S13). We conclude that one consequence of changing the coordination from N₄ to N₃S is that the porphyrin ligands become noninnocent participants in the electrocatalytic CO₂ reduction. The nature of the noninnocent ligand is rooted in the redistribution of electrons and the modification of the molecular orbital levels of the catalysts as we discussed in this section.

3.4. Peripheral Substituents on Catalyst 2

So far, the substitution of N to S is at the first-coordination sphere, and in the following, we examine whether modifying the peripheral substituents , of the porphyrin ring can further improve the catalytic performance. On top of catalyst 2, we substituted the H atom at the para-position of each phenyl ring with three types of functional groups, including an electron-withdrawing group, –F; an electron-donating group, –NH₂; and a cationic group, –N⁺Me₃. ,,

Calculations indicate that these substituents affect the reduction potential E _red of the catalyst and the formation energy of the *COOH intermediate. As shown in Figure a horizontal coordinate, F– and cationic substituent –N⁺Me₃ make E _red more positive, meaning it is more easy to be reduced. The respective E _red values are −0.72 and −0.64 V vs SHE for 2–p-F₄ and 2–(p-N⁺Me₃)₄. In contrast, NH₂ substitution makes the E _red more negative, where the value is −0.81 V for 2–p-(NH₂)₄. The change trends of E _red’s are consistent with previous experimental works on the modified Fe porphyrins, where –F, –OCH₃, and –N⁺Me₃ functional groups were investigated, albeit that the substituent sites include ortho-, meta-, and para-positions. , Analyzing the frontier molecular orbital energy levels of these modified catalysts with catalyst 2 shows that the lowest unoccupied molecular orbitals (LUMOs) of 2–p-F₄ and 2–(p-N⁺Me₃)₄ are lowered, while the LUMO of 2–p-(NH₂)₄ is raised (Table S12, Figures S9–S11). A lower LUMO means the species can accept electron, i.e., being reduced, more easily; therefore, the E _red values become more positive.

6.

(a) Binding free energy of COOH (ΔG _b) versus the DFT-calculated reduction potential (E _red) of catalyst 2 and its modified forms. (b) Optimized structure of [2–p-(NMe₃)₄–COOH]⁴⁺. (c) Gibbs free-energy diagram of the critical steps for CO₂-to-CO reduction on catalyst 2 and its modified forms under an applied potential of −1.03 V and pH = 7.2 in an aqueous solution.

As shown in Figure a vertical coordinate, F-substituent slightly increases the binding free energy of COOH on the catalyst (denoted as ΔG _b,COOH) to −29.6 kcal/mol, whereas NH₂ and N⁺Me₃ substituents decrease ΔG _b,COOH to −44.4 and −40.1 kcal/mol, respectively. Inspiringly, the p-N⁺Me₃ substituents not only make the catalyst easier to be reduced but also decrease the ΔG _b,COOH. It is known that neutral substituents (like –F and –NH₂) influence the electronic structure of the catalyst via through-bond inductive effect, i.e., electrons are transferred through chemical bonds. Cationic substituents can influence the catalyst via nonbonding interactions, i.e., electrostatic interactions in this case (Figure b), which are named as through-space effect. This approach has been adopted and was found to improve the catalytic performance of Fe-/Co-porphyrin-based electrocatalyst for CO₂ reduction. , Herein, we validated its usefulness in a Ni-porphyrin system with a N₃S core, in which the through-space electrostatic interaction between N⁺Me₃ substituents and the COOH group stabilizes the *COOH species. Moreover, we computed the reaction potential energy profiles of these catalysts and confirmed that the modified 2–(p-N⁺Me₃)₄ catalyst has the lowest barrier (Figure c). The −N⁺Me₃ substituent significantly lowers the barrier in both the CO₂ adsorption-first and reduction-first pathways. Under an applied potential of −1.03 V, the barriers of the RDS for the reduction-first pathway, i.e., the adsorption of CO₂ are 10.3, 11.2, and 5.2 kcal/mol for 2–p-F₄, 2–p-(NH₂)₄, and 2–(p-N⁺Me₃)₄, respectively. For the CO₂ adsorption-first adsorption pathway, adsorption of CO₂ remains the RDS, and the corresponding barrier for 2–(p-N⁺Me₃)₄ is further reduced to 1.1 kcal/mol.

In brief, modifying the peripheral substituents of 2 by the cationic N⁺Me₃ functional group can achieve a better CO₂ reduction performance, owing to its capability to lower the LUMO levels and to facilitate through-space interactions.

4. Conclusions

In conclusion, this computational study demonstrates that rational modifications to both the first and outer coordination spheres of nickel-porphyrins can synergistically enhance their catalytic performance for the electrochemical reduction of CO₂ to CO. Using DFT calculations, we identified that the introduction of sulfur into the coordination environment of the Ni-porphyrin, along with cationic peripheral substituents, significantly lowers the energy barriers associated with CO₂ reduction. These improvements arise from the positive shift of reduction potential, stabilization of key intermediates *COOH through increase orbital interaction, reduced Pauli repulsion and enhanced electron donation, as well as through-space electrostatic interactions facilitated by the cationic groups.

Beyond the specific findings, this work provides a broader framework for the design of efficient electrocatalysts for the reduction of CO₂ and other challenging transformations. The insights gained heresuch as the importance of heteroatom coordination and through-space electronic effectscan be extended to other metallo-porphyrin systems and even nonporphyrin-based catalysts. For instance, the principles of stabilizing reactive intermediates through tailored coordination environments and substituent effects could inspire the development of catalysts for the reduction of other small molecules.

Looking ahead, several promising directions emerge from this study. These include exploring more substituent positions of the outer sphere, a wider range of heteroatoms in the first coordination sphere, ,− integrating designed metallo-porphyrins into covalent organic frameworks or other porous materials , for enhanced stability and accessibility, and investigating multimetallic porphyrin systems for cooperative catalysis. Such advancements could further bridge the gap between molecular-level understanding and practical applications, contributing to the global effort toward carbon neutrality and sustainable energy solutions. By combining computational guidance with experimental validation, the rational design of catalysts can be accelerated, paving the way for transformative breakthroughs in energy and environmental science.

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/prechem.5c00429.

• Optimized structures, spin density plots, potential energy scans and Gibbs energy diagrams, frontier orbital analyses, tables of calculated reduction potentials, structural parameters, spin populations and energetics (PDF)

• Coordinates (XYZ)

The authors declare no competing financial interest.