p-State of surface oxygen for mediating the s-band center of a single-atomic Ag catalyst for enhanced catalytic property for the oxygen reduction reaction†

Fu-li Sun‡ , Qiao-jun Fang‡ , Wei Zhang , Cun-biao Lin , Wen-xian Chen and Gui-lin Zhuang *
H-PSI Computational Chemistry Lab, Institute of Industrial Catalysis, State Key Laboratory Breeding Base of Green-Chemical Synthesis Technology, College of Chemical Engineering, Zhejiang University of Technology, Hangzhou 310032, P. R. China. E-mail: glzhuang@zjut.edu.cn

First published on 16th October 2023

---

---

Introduction

H₂O₂ is extensively utilized as a green oxidant in various industries and sectors, including papermaking, textile manufacturing, chemical synthesis, military applications, electronics, food production, medicine, cosmetics, environmental protection, and metallurgy, among others.¹ Currently, the predominant method for the industrial production of over 95% of H₂O₂ is the anthraquinone process,^2,3 which relies on the costly and environmentally impactful Riedl–Pfleiderer process, involving the oxidation and hydrogenation of an alkyl anthraquinone.⁴ Hence, discovering a green, sustainable, and on-site producible alternative method is crucial.^5,6 In recent years, there has been growing interest in the electrochemical 2e⁻ ORR method for the direct generation of H₂O₂ from O₂ and H₂.^7,8 This method allows the on-site production of the required amount of H₂O₂, thereby eliminating transportation and storage concerns. Moreover, the only by-product generated is H₂O, making it environmentally friendly and pollution-free. However, the rational design of high-performance (especially for high selectivity) electrocatalysts for 2e⁻ ORR poses significant challenges.

Substantially, H₂O₂ is primarily produced through a 2e⁻ ORR process on the cathode, while other side reactions, such as the four-electron O₂-reduction reaction (4e⁻ ORR) and hydrogen-evolution reaction (HER), also occur concurrently. The stability of the intermediates (e.g., *O₂, *OOH), dependent on O₂ activation, plays a key role in achieving high activity and selectivity for the 2e⁻ ORR process.⁹ Experimentally, the electrocatalytic reaction occurs under a complex micro-environment, consisting of the electrolytes and electrodes. Acidic electrolytes, besides acting as pH regulators,¹⁰ impact the electronic properties of the surface and interface on heterogeneous catalysts, affecting the catalytic performance for the ORR.¹¹ Understanding the influence of proton adsorption on the electronic properties, such as the d-band center of active sites, is vital for designing effective electrocatalysts. Moreover, single-atomic catalysts (SACs),^12,13 exhibiting maximum atom utilization and flexible tunability, have attracted significant attention because of the huge potential applications in industrial catalysis and meanwhile as they can serve as crucial theoretical models for exploring the intricate relationship between catalytic properties and structural functionalities.¹⁴ Despite the remarkable developments in transition metal (TM) catalysts, the competitive dominance of both the d-band center and s-band center in ds-block TMs on electrocatalytic performance has remained largely unexplored until now.

Herein, combining DFT calculations and AIMD simulations, we systematically studied the 2e⁻ ORR properties of SACs on Ti₂C MXene under acid electrolytes. The results demonstrated that Ag@Ti₂C displayed exceptional intrinsic performance in facilitating the 2e⁻ ORR. Specifically, we analyzed how Ag@Ti₂C behaves in acidic electrolytes and observed its preferential interaction with both O-containing and H species, leading to the formation of Ag@Ti₂CO₂H_x. To gain deeper insights into the electrocatalytic properties for the ORR, we conducted a comprehensive series of electronic-structure calculations. Our primary focus was to understand the electronic effects of the O-containing and H species on Ag@Ti₂CO₂H_x. These calculations revealed the correlation between the oxygen 2p-states and the s-band or d-band of the Ag sites within Ag@Ti₂CO₂H_x, providing valuable insights into the catalyst's catalytic performance in the ORR.

Computational details

All DFT calculations were performed using the first-principle VASP software (VASP 6.3.1).¹⁵ Following the generalized gradient approximation¹⁶ (GGA), the Perdew–Burke–Ernzerhof¹⁷ (PBE) functional was used to deal with the exchange–correlation. The interaction between ions and electrons was described by the projector-augmented wave (PAW) method.¹⁸ The cutoff energy of the plane-wave basis set was set to 450 eV.¹⁹ The convergence threshold was set to 10⁻⁵ eV for energy and 10⁻² eV Å⁻¹ for force for the geometric optimization and self-consistent field calculations. DFT-D3²⁰ was chosen to correct the van der Waals interaction. The k-point grid²¹ sampling in the Brillouin zone was set to 4 × 4 × 1, which was validated by the convergence test. AIMD simulations used the NVT ensemble with a constant temperature of 500 K. Equilibration took 10 ps, and a time step of 1 fs was used for integrating equations of motion. Temperature control was achieved using a Nosé–Hoover thermostat.²² In addition, the VASPsol module was used to add an implicit aqueous electrolyte to the catalytic system to simulate the implicit solvation model.^23,24

The binding energy of the TM single atom in the support was typically evaluated as follows:²⁵

Eb = ETM@MXene − (ETM-single + EMXene)	(1)

where E_TM@MXene is the total energy after loading the TM single atom on MXene, E_TM-single is the energy of the TM single atom, and E_MXene is the energy of pure MXene. Also, the cohesive energy (E_c) was typically evaluated based on the following formula eqn (2).

Ec = ETMn/n − ETM-single	(2)

where E_TMn is the energy of the most stable phase of the TM, n is the atomic number in the unit cell of the most stable phase, and E_TM-single is the energy of the single TM atom.

According to computational hydrogen electrode (CHE) strategy from Nørskov et al.,²⁶ the Gibbs free energy profiles were estimated. Under standard reaction conditions, the chemical potential of the proton and electron pair (μ(H⁺ + e⁻)) is equal to half that of gaseous hydrogen (μ(H₂)). For each primitive step, the Gibbs free energy, ΔG, was calculated as per the following eqn (3).

ΔG = ΔEDFT + ΔZPE + ΔH − TΔS + ΔGpH + ΔGU	(3)

where E is the total energy calculated by VASP, ΔZPE and TΔS are the zero-point energy and entropy contributions calculated by VASPKIT,²⁷T is the temperature (298.15 K), and ΔG_pH and ΔG_U are the effects of the pH and electrode potential U, respectively, with these latter parameters determined by eqn (4) and (5).

GU = −neU	(4)

where n is the number of transferred electrons.

GpH = −kBTln([H+]) = pH × kBTln10	(5)

where k_B is Boltzmann's constant. For all the calculations presented here, the pH was set to zero.

The adsorption energy (E_ads) of different adsorption intermediates could be calculated as follows:

Eads = Etotal − (Ecatal + Emol)	(6)

where E_total is the total energy of the whole after adsorption, E_catal is the energy of the catalyst, and E_mol is the energy of the adsorbate (such as O₂, *OOH, *O, or *OH).

For an ideal ORR electrocatalyst, the ΔG at each elementary step is very close to the equilibrium potential (in acidic solutions, for the 4e⁻ pathway: U_equilibrium = 4.92/4 = 1.23 V; for the 2e⁻ pathway: U_equilibrium = 1.40/2 = 0.70 V). Therefore, the overpotential (η) is usually defined by the following formula eqn (7).

η = Uequilibrium − UL	(7)

U_L is calculated by the following formula:

	(8)

ΔG_PDS is the largest ΔG in each pathway.

The selectivity of the 2e⁻ ORR can be simply estimated from the Boltzmann distribution:²⁸

	(9)

If the possibility of the H₂O₂ formation is set as 1, δG_{4e− ORR} = ΔG_PDS(4e⁻ ORR) − ΔG_PDS(2e⁻ ORR), and δG_HER = ΔG_PDS(HER) − ΔG_PDS(2e⁻ ORR).

Results and discussion

Screening of TM@Ti₂C

First, Ti₂C MXene with the 3 × 3 × 1 supercell, including the three-layer structure of Ti–C–Ti, was selected as the substrate (see Fig. 1a), and the supporting metal involved the 3d, 4d, and 5d TMs (TM@Ti₂C). Furthermore, the geometrical optimization of TM@Ti₂C was carried out (see Fig. S1†). There were two hollow sites on the surface of Ti₂C, denoted as H1 and H2. At these two hollow sites, the single atom of the TM can be attached to three Ti atoms on the surface in a tetrahedral structure, which makes the single atom more stably loaded on the Ti₂C surface. The calculation results for the binding energy at the adsorption sites are shown in Fig. 1b. It was apparent that the binding energies of the 29 selected TM single atoms were all negative, which means that these single TM atoms were thermodynamically stable²⁹ when loaded on the Ti₂C surfaces.

Adsorption of O₂

The adsorption of O₂ is an important prerequisite for the occurrence of the whole ORR. Therefore, the adsorption energies of O₂ on the above 29 SACs were first calculated. The optimized adsorption configurations are displayed in Fig. S2.† It can be seen from the figure that no matter whether the TM was 3d, 4d, or 5d, O₂ was directly dissociated on the pre-TM SACs and further bound to the Ti atoms on the surface, while it could maintain a robust adsorption configuration on the post-TM SACs. Such differences may be due to the electronegativity of the post-TM being stronger than that of the pre-TM, resulting in less electron transfer between the post-TM and O₂. The lesser electron-transferring ability effectively reduces the activation degree of O₂ and maintains the existence of the O–O bond. When considering the intended product of H₂O₂, it is crucial to ensure that the O–O bond within the O₂ remains intact. Following the O₂ adsorption configuration analysis, a total of 14 TM SACs were carefully chosen for further investigation of their performance in the ORR. These catalysts included the TMs: Co, Ni, Cu, Zn, Ru, Rh, Pd, Ag, Cd, Os, Ir, Pt, Au, and Hg.

Electrocatalytic 2e⁻ ORR process

Essentially, the ORR mainly involves two primary pathways: the 2e⁻ ORR and the 4e⁻ ORR. In the 2e⁻ ORR pathway, the initial step involves the hydrogenation of O₂ to form *OOH, which subsequently undergoes further hydrogenation to yield H₂O₂. On the other hand, the 4e⁻ ORR pathway involves the hydrogenation of O₂ to generate *OOH, followed by its additional hydrogenation to produce *O and an H₂O molecule. This is succeeded by two successive hydrogenation steps of *O, resulting in the formation of a second H₂O molecule. The specific reaction pathway can be summarized by the following eqn (10)–(15):

2e⁻ ORR pathway:

O2 + H+ + e− → *OOH	(10)

*OOH + H+ + e− → * + H2O2	(11)

4e⁻ ORR pathway:

O2 + H+ + e− → *OOH	(12)

*OOH + H+ + e− → *O + H2O	(13)

*O + H+ + e− → *OH	(14)

*OH + H+ + e− → * + H2O	(15)

After the formation of *OOH, an important intermediate in the ORR, the distinction between the 2e⁻ ORR and 4e⁻ ORR becomes evident in the reaction pathway. This signifies the significance of *OOH as a crucial intermediate in the process. As a result, the free energy of generation of *OOH, i.e., ΔG_*OOH, was calculated for the 14 catalysts screened above that were capable of stable O₂ adsorption. As illustrated in Fig. 2a, the O₂ adsorption energy is positively correlated with ΔG_*OOH (R² = 0.85). This means that the more negative the O₂ adsorption energy, the more negative the ΔG_*OOH. This phenomenon is due to the stronger activation of O₂ by the catalyst when the O₂ adsorption energy is negative, which makes O₂ susceptible to H⁺/e⁻ attack and leads to a more negative ΔG_*OOH. To further clarify the ORR performance of the catalysts, the specific pathways of the 2e⁻ ORR and 4e⁻ ORR were investigated. As seen in Fig. 2b, the potential energy-determining step (PDS) for both the 2e⁻ ORR and 4e⁻ ORR was the last desorption step for all 14 catalysts, except for Ag@Ti₂C, which was *OOH + H⁺ + e⁻ → * + H₂O₂ for the 2e⁻ ORR and *OH + H⁺ + e⁻ → * + H₂O for the 4e⁻ ORR. This implies that the TM–O bonds on these catalysts (apart from Ag@Ti₂C) were stronger, so that the weaker TM–O bonds would facilitate the desorption and thus the synthesis of the target product. The overpotentials of the 2e⁻ ORR and 4e⁻ ORR were obtained from the ΔG_PDS of these catalysts and are compared in Fig. 2c. It can be seen that for the 14 catalysts mentioned above, the overpotential of the 2e⁻ ORR was lower than that of the 4e⁻ ORR, confirming that the 2e⁻ ORR process was more likely to occur on these catalysts at lower applied voltages.

To investigate the impact of the metal sites on the catalytic performance of the mentioned catalysts, an analysis was conducted to determine the d-band center for each of the 14 catalysts. The calculated values are summarized and presented in Fig. 2d. This analysis aimed to provide insights into how the position of the d-band center is related to the overall catalytic activity of the catalysts. It could be noted from the graph that ΔG_*OOH and the d-band center of the catalyst featured a volcano-shaped curve with the apex at around −4.00 eV. Remarkably, both Ag@Ti₂C and Au@Ti₂C demonstrated exceptional performance in the 2e⁻ ORR pathway, exhibiting overpotentials of 0.06 V for Ag@Ti₂C and 0.15 V for Au@Ti₂C, respectively, in close proximity to the optimal point. This indicates their superior catalytic activity and highlights their potential as highly efficient catalysts for the specific reaction under investigation.

Based on the results of the overpotential calculations, Ag@Ti₂C exhibited a particularly outstanding 2e⁻ ORR performance. Therefore, the catalyst was investigated in more detail and the specific 2e⁻ ORR and 4e⁻ ORR pathways are shown in Fig. 3a and b. In the 2e⁻ ORR with Ag@Ti₂C, the PDS was *OOH + H⁺ + e⁻ → * + H₂O₂ with an overpotential of 0.06 V. In the 4e⁻ ORR, the PDS was *O₂ + H⁺ + e⁻ → *OOH with an overpotential of 0.47 V. As is well known, the 2e⁻ ORR reaction has two competing reactions (the 4e⁻ ORR and the HER), resulting in a low H₂O₂ selectivity. Apart from the 4e⁻ ORR, the competing HER also plays a role in determining the 2e⁻ ORR performance. HER uses protons and electrons directly during electrocatalysis, decreasing the Faraday efficiency of the 2e⁻ ORR. Energetically, the weak *H adsorption (E_ads = 0.10 eV) on Ag@Ti₂C compared to O₂ adsorption (E_ads = −0.11 eV) highlights O₂'s preference. Additionally, the ΔG(*H) at 0.95 eV (Fig. 3c) indicated a poorer HER performance than the 2e⁻ ORR, leading to a 99.9% 2e⁻ ORR selectivity on Ag@Ti₂C. Most 2e⁻ ORR electrocatalysts previously reported could achieve a H₂O₂ selectivity of ca. 93%.^30,31 In other words, the Ag@Ti₂C catalyst featured better 2e⁻ ORR performance than most the other reported catalysts. Moreover, the AIMD simulations demonstrated that the total energy of Ag@Ti₂C fluctuated within a small range at 500 K for 10 ps (see Fig. 3d), and the fluctuation became diminished, illustrating that the catalyst was able to maintain its original structural configuration and had high stability. Thermodynamically, the negative energetic difference (ΔE = −0.91 eV) between the binding energy (E_b = −3.42 eV) of Ag on Ti₂C and the cohesive energy (E_c = −2.51 eV) of the Ag metal revealed that the single-atomic Ag on Ti₂C was robust. Therefore, the strong metal–support interaction effectively prevents the single Ag atom from agglomerating into clusters (Ag_n@Ti₂CO₂) both thermodynamically and kinetically.

Effect of the surface H coverage on Ag@Ti₂C

Experimentally, the electrocatalytic process typically occurs within a solvent environment. To assess the stability of the single-atomic Ag in a liquid environment, AIMD simulations of Ag@Ti₂C in aqueous solution were performed and showed that, within 5 ps, O atoms and OH functional groups predominantly bind with Ti atoms (see Fig. S4†). Such a phenomenon demonstrated that Ti₂C surfaces were commonly coated with functional groups, e.g., O and OH, due to the abundance of unsaturated bare Ti atoms, which accords with previously reported results.³² Meanwhile, the stability of the single-atomic Ag sites on the Ti₂C support remained evident. As a result, our investigation primarily centered on the 2e⁻ ORR electrocatalytic performance of O-binding Ag@Ti₂C (Ag@Ti₂CO₂).

As seen in Fig. 4a and b, the PDS for the 2e⁻ ORR on Ag@Ti₂CO₂ was *OOH + H⁺ + e⁻ → * + H₂O₂ with an overpotential of 0.23 V. The PDS for the 4e⁻ ORR was *OOH + H⁺ + e⁻ → *O + H₂O with an overpotential of 0.41 V. Comparing Ag@Ti₂C with and without O functional groups on the surface, the introduction of O functional groups in Ag@Ti₂CO₂ resulted in an increased overpotential for the 2e⁻ ORR (from 0.06 V to 0.23 V) and a decreased overpotential for the 4e⁻ ORR (from 0.47 V to 0.41 V). These findings suggest that the presence of O functional groups on the Ag@Ti₂C surface adversely affected its 2e⁻ ORR catalytic performance. To identify the intrinsic reasons for this, calculations for the electronic structures were performed, including the charge density difference, electron localization function (ELF), and crystal orbital Hamilton population (COHP). The calculations are depicted in Fig. 4c–e. As the active center in Ag@Ti₂C, the Ag atom is directly connected to Ti atoms, which causes the transfer of electrons from the Ti atoms to the Ag atom due to the stronger electronegativity of Ag (1.93) compared to Ti (1.54). This results in a higher electron localization around the Ag atom, leading it to be less able to adsorb O₂ (E_ads(O₂) = −0.11 eV). Yet the number of electrons around the Ag atom is higher, which in turn enables O₂ to receive more electrons (0.42|e|) and leading to a longer O–O bond (1.30 Å). In Ag@Ti₂CO₂, the Ag atom is bonded to O atoms on the surface as Ag⁺ ions by means of spⁿ hybridization, and the lone pair of electrons of the O atoms are distributed in the spⁿ orbitals of Ag⁺, resulting in an increased degree of electron delocalization around the Ag–O bonds. This contributes to an enhanced interaction with O₂ and stronger O₂ adsorption with an E_ads(O₂) of −0.49 eV. In contrast to Ag@Ti₂C, however, O₂ receives fewer electrons (0.28|e|) and the O–O bond is relatively shorter (1.27 Å) owing to the lower number of electrons in the Ag atom. As O₂ is strongly adsorbed on the Ag@Ti₂CO₂ surface, there is strong bonding between the O₂ and Ag atoms, which raises the energy barrier required for the final desorption step. As a result, the 2e⁻ ORR overpotential of Ag@Ti₂CO₂ was higher than that of Ag@Ti₂C. In the case of the 4e⁻ ORR, since the O–O bond length of the adsorbed O₂ state on the Ag@Ti₂CO₂ surface was relatively short, causing difficulty in breaking the O–O bond, the PDS for the 4e⁻ ORR was the second hydrogenation step (*OOH + H⁺ + e⁻ → *O + H₂O), and in general there was a slight decrease in the overpotential of the 4e⁻ ORR.

After exposing the Ag@Ti₂CO₂ surface to an acid electrolyte, the O end of the surface readily binds with H protons, leading to Ag@Ti₂CO₂H_x. Thus, we further assessed the impact of the H coverage on the ORR performance. First, the adsorption distribution of H on Ag@Ti₂CO₂ was investigated. The adsorption energy of the H atom on the Ti₂CO₂ surface was first considered. As shown in Fig. 5a, as the H coverage increased, the H adsorption gradually weakened according to the relation: y = −0.0046x + 0.0003x² − 0.3416, R² = 0.88. At a H coverage of 0.0%, the E_ads(H) was −0.35 eV. At a H coverage of 33.3%, the H adsorption energy was −0.07 eV, which was much weaker than that at 27.8% (E_ads(H) = −0.28 eV). This phenomenon may be due to the influence of steric hindrance and the electronic properties (surface charge, p-band center of O) with the increase in H coverage, which makes the adsorption of H difficult. At a H coverage of 44.4%, the adsorption energy of H was extremely low, measuring only −0.02 eV, which was very close to 0 eV. Moreover, by analyzing the data presented in Fig. 5a, it becomes evident that the p-band center (ε_p) of the oxygen atoms could be accurately determined for different H coverages. It was observed that ε_p followed a parabolic trend in relation to the H coverage. This trend indicated that as the H coverage increased, the adsorption of H atoms by the surface O atoms gradually diminished. Additionally, it was revealed that O atoms with higher ε_p values exhibited stronger affinities for binding hydrogen. The parabolic relationship approached its peak when the H coverage reached 44.4%, which corresponded to the minimum value of ε_p (y = −0.0355x + 0.0004x² − 3.7093, R² = 0.90). This indicates that subsequent hydrogenation would become exceptionally challenging due to the minimal adsorption energy. Therefore, this study considered a range of H coverages, specifically from 0% to 44.4%. Furthermore, based on the analysis of the optimal adsorption sites of H atoms at different coverage levels, it was observed that H atoms exhibited a tendency to bind with O atoms beneath the surface of unoccupied Ag atoms when the H coverage ranged from 5.6% to 16.7%. This phenomenon could be attributed to the fact that the surface O atoms of unoccupied Ag atoms possessed a lower electron density compared to the upper surface (see Fig. S6†). Consequently, they had a higher affinity for bonding with H atoms, which contributes electrons to the system. Contrarily, when the H coverage exceeded or equaled 22.2%, the preferred adsorption sites for H atoms shifted to the upper surface due to considerations of site resistance and charge. To calculate the effect of H coverage on the ORR performance, ΔG_*OOH was first calculated as an important descriptor of the ORR performance (R² = 0.89). ΔG_*OOH is the outcome of the calculation illustrated in Fig. 5b (where all the H atoms are from the H₂ in the air). The calculations demonstrated that when the H coverage was equal to or below 16.7%, the impact on the active site Ag was minimal since the H atoms primarily adsorbed on the under surface. With 22.2% H coverage, O₂ had a 1.29 Å bond length and acquired 0.39|e| electrons (Fig. S7†), surpassing the 0.28|e| without H coverage. The initially adsorbed H atom on the upper surface enhanced the available H atom source for O₂ hydrogenation, facilitating the formation of *OOH. In the other case, ΔG_*OOH decreased with increasing the H coverage for a H coverage ≥ 22.2%. The Bader charge data revealed a correlation between the number of H atoms on the upper surface of Ti₂CO₂ and the electron count carried by the material (see Fig. S8†). With the increasing number of H atoms, the electron count around the active center Ag atom also rose. Consequently, there was an increase in the number of electrons available for the adsorption of O₂ on the Ag atom. The greater the number of O₂ electrons, the easier it becomes for O₂ to bind with H⁺ and e⁻, resulting in a more negative ΔG_*OOH. In this scenario, the ΔG_*OOH at a H coverage of 38.9% was very similar to that at 33.3%. This could be attributed to the tendency of H atoms to preferentially adsorb on the lower surface as well, which reinforces the earlier conclusion drawn for H coverages equal to or below 16.7%.

As mentioned above, the ORR reaction intermediates could capture the H adsorbed on the surface O functional group when the H coverage was ≥22.2%. Consequently, this study delved into the specific pathways of the ORR by separately investigating the ORR performance using different hydrogen sources. When the H coverage was ≥22.2%, the process of capturing H on the surface O atom necessitated overcoming the energy barrier associated with O–H bond breaking. As a result, ΔG_*OOH (activation energy) for the formation of *OOH, which involves the binding of O₂ to H from H₂, became more negative. This resulted in a higher ΔG (free energy change) for the second hydrogenation step of the 2e⁻ ORR, leading to an increased overpotential. In contrast, the direct capture of surface H by O₂ reduced the surface H coverage of Ti₂CO₂ and O atoms could contribute more electrons to the spⁿ hybrid orbital of Ag⁺, making the electrons localized between Ag and O atoms, reducing the number of electrons transferred from Ag atoms to *OOH. This weakened the O–Ag bond in *OOH, reducing the potential required for the second-step hydrogenation and ultimately resulting in a lower overpotential. These conclusions were confirmed by the results of the ICOHP calculations. At 22.2% H coverage, the ICOHP value for the O–Ag bond in *OOH generated by the direct capture of surface H by O₂ was −1.13, which was higher than the ICOHP value (−1.15) for the O–Ag bond in *OOH generated by binding in H₂. This reveals that the direct capture of surface-adsorbed H atoms by adsorbed O₂ was more favorable to the 2e⁻ ORR process when the H coverage was ≥22.2%. In order to compare the overpotentials of the two pathways more visually, the pathway with the lowest overpotential was selected based on the above results (see Fig. 5c). The overpotential of the 2e⁻ ORR was clearly lower than that of the 4e⁻ ORR for a H coverage ≤ 33.3%. At H coverages of 22.2% and 27.8%, the overpotential for the 2e⁻ ORR was notably lower compared to the other coverage degrees. Specifically, the overpotential measured 0.10 V for 22.2% H coverage and 0.08 V for 27.8% H coverage. Whereas the overpotential of the 4e⁻ ORR was lower than that of the 2e⁻ ORR under a H coverage of more than 38.9%. This suggests that the 2e⁻ ORR is more likely to occur at low potentials than the 4e⁻ ORR when the H coverage (≤33.3%) is low.

Meanwhile, the competitive HER of the ORR performance at each H coverage was also considered. As shown in Fig. 5d, the value of ΔG_*H decreased as the H coverage increased. The best HER performance was achieved when ΔG_*H = 0 eV. Consequently, as the H coverage increased, the performance of the HER performance became better and then worse, and the HER performance was best at a H coverage of 27.8% (ΔG_*H = 0.08 eV). Considering the performance of the two competing reactions 4e⁻ ORR (ΔG_PDS = 0.62 eV) and HER (ΔG_PDS = 0.31 eV) together, the 2e⁻ ORR (ΔG_PDS = 0.10 eV) had the best selectivity (99.9%) under a H coverage of 22.2%.

Correlation between the p-band of the surface O and the s-band center of the Ag site

To investigate the impact of different H coverages on the catalytic performance of the ORR, we conducted an in-depth analysis. We focused on calculating the p-band center of the oxygen (O) atom, as well as the d-band center and s-band center of the silver (Ag) atom, on the surface of the Ag@Ti₂CO₂ catalyst under varying H coverages (see Fig. S9†). Upon examining Fig. S9a,† it is apparent that the change in the d-band center of the Ag atom was not significantly influenced by the H coverage (y = −0.0111x + 0.0002x² − 3.5784, R² = 0.35). This observation could be attributed to the Ag atom's location in the ds region, characterized by an outer electron configuration of 4d¹⁰5s. In the case of Ag atoms with electron-filled 4d orbitals, the role of the d-band in the electron-transfer process is weaker compared to that of the s-band. Interestingly, both the p-band center of the O atom (y = −0.0355x + 0.0004x² − 3.7093, R² = 0.87) and the s-band center of the Ag atom (y = −0.0322x + 0.0004x² − 0.1605, R² = 0.79) exhibited a clear parabolic relationship with the H coverage (see Fig. S9c and d†). Furthermore, the p-band center of the O atom demonstrated a linear relationship with the s-band center of the Ag atom (y = 0.83x + 2.88, R² = 0.76). Consequently, it was evident that the H coverage directly modified the electron distribution of the 2p-state of the O atom, which, in turn, indirectly regulated the electron distribution of the Ag atom. This, in consequence, affected the adsorption energy of the O₂ and the overall performance of the ORR. Charge analysis revealed that increasing the H coverage prompts the H atoms to transfer some charge to O atoms, resulting in a more negative p-band center for the O atoms. This observation aligns with the inference that adsorbed H atoms atop O atoms donate charge.^33,34 The more H atoms adsorbed by the surface O atom, the greater the charge the surface O atom accumulates. This leads to a rise in the electrons contributed by O atoms to the spⁿ hybrid orbitals of Ag⁺. As a result, more diffuse electrons are manifested on Ag atom surfaces, causing a more negative s-band center. The upshifting of the s-state center of the metal single atoms would be important in decreasing the adsorption strength of the ORR intermediates,³⁵ and conversely the fall in the s-state center enhances the adsorption strength of the intermediates. With a weaker O₂ and Ag atom interaction, O₂ activation is challenging, hampering O₂'s transformation into the OOH intermediate via H addition. In contrast, a stronger O₂ and Ag atom interaction increases O₂ dissociation, facilitating the 4e⁻ ORR. Sabatier's principle highlights that an appropriate interaction between O₂ and the catalyst is advantageous for facilitating the 2e⁻ ORR. From this, it can be inferred that the optimal H coverage for the 2e⁻ ORR performance lies in the region above the two parabolas in Fig. S10.† With the Ag atom's s-band center suitably adjusted away from the Fermi energy level, a stable adsorption structure forms between O₂ and the single-atomic Ag catalyst. This alignment results in a favorable ΔG(*OOH), leading to a reduced 2e⁻ ORR overpotential. At 22.2% H coverage, the adsorption energy of O₂ and ΔG(*OOH) were closest to this optimal range, thereby leading to the best performance for the 2e⁻ ORR.

Effects of the solvent environment

Since electrocatalytic reactions proceed under aqua solution, we further identified the mixed effect of aqueous media and H coverage on the 2e⁻ ORR performance. Under an aqueous medium, the number of electrons lost from the active site Ag atom increased, but the number of electrons gained by O₂ adsorption increased. This resulted in a slight difference in the trend of ΔG_*OOH generated by the subsequent *OOH compared to that in vacuum. In the presence of an aqueous medium, the H atoms still made ΔG_*OOH decrease (R² = 0.94), even when H atoms were adsorbed on the lower surface of Ti₂CO₂ (see Fig. 6a), which could also be considered to show that the presence of an aqueous medium expands the effect of the surface H coverage.

In addition, the difference between the two pathways of the ORR in the presence of aqueous media with different H coverages and in vacuum was investigated, and the ORR performance with different H sources was also considered. The results of the calculations showed that, similar to in the vacuum, the 2e⁻ ORR process was facilitated by the direct capture of surface-adsorbed H atoms by the adsorbed O₂ at a H coverage ≥ 22.2% in aqueous media. The presence of the aqueous media decreased the overpotential of the 2e⁻ ORR and increased the overpotential of the 4e⁻ ORR when the H coverage was <22.2%, as can be seen in Fig. 6b and c, indicating that the aqueous media accelerated the selectivity of the 2e⁻ ORR reaction. However, at higher H coverage, the presence of the aqueous medium greatly increased the overpotential of the 2e⁻ ORR, which reduced the activity of the 2e⁻ ORR at lower applied voltages. Therefore, controlling the H coverage of the Ti₂CO₂ surface is an effective strategy to improve the catalytic performance of the 2e⁻ ORR in real experimental solution environments.

A comprehensive comparison of the overpotentials for the three competing reactions (2e⁻ ORR, 4e⁻ ORR, and HER) in the presence of aqueous media revealed that the observed trends were consistent with those observed in a vacuum environment (as depicted in Fig. 6d). At low degrees of coverage, the overpotential of the 2e⁻ ORR was the lowest of the three reactions, with the smallest overpotential of 0.08 V for the 2e⁻ ORR at a H coverage of 22.2%. As the coverage increased, the HER became the reaction with the lowest overpotential. The specific pathways of the 2e⁻ ORR and 4e⁻ ORR for Ag@Ti₂CO₂ at the optimum H coverage (22.2%) are shown in Fig. 6e and f. The capture of the O–Ag bond in the adsorbed state H atoms on the surface to form *OOH (ICOHP = −1.13) was weaker than the binding of H to H in H₂ to form the O–Ag bond in *OOH (ICOHP = −1.16), which facilitated the desorption of the product H₂O₂. Since the formation of *OOH involves the breaking of the O–H between the surface H and O functional group, this makes the first hydrogenation step as the PDS for the 2e⁻ ORR and 4e⁻ ORR, with corresponding overpotentials of 0.08 V and 0.61 V, respectively. In comparison to the vacuum conditions (where the overpotentials for the 2e⁻ ORR and 4e⁻ ORR at a 22.2% H coverage were 0.10 V and 0.63 V, respectively), there was a slight reduction in the overpotential for both pathways in the presence of aqueous media. The whole catalytic selectivity of the 2e⁻ ORR was mostly superior to the 4e⁻ ORR and HER.

Conclusion

In summary, we developed a computational framework to systematically investigate the 2e⁻ ORR catalytic performance of SACs on Ti₂C by combining DFT methods and AIMD simulations and screened an ideal 2e⁻ ORR catalyst Ag@Ti₂C with an overpotential of only 0.06 V. In the presence of real acidic electrolytes, Ag@Ti₂C tended to selectively bind with surface O functional groups and H species, forming Ag@Ti₂CO₂H_x. Notably, the p-band center of the surface O atoms followed a parabolic trend in relation to the H coverage. When the H coverage on the surface of Ag@Ti₂CO₂ reached 44.4%, the p-band center corresponded to the minimum value. Furthermore, the presence of O functional groups increased the degree of electron delocalization of Ag atoms in the active center and thereby enhanced the O₂–Ag interaction. Under the H coverage, the adsorbed O₂ can take over the H bound on the surface O, resulting in a lower overpotential of 0.10 V for the 2e⁻ ORR (at a H coverage of 22.2%). Therefore, it was concluded that the p-state of the surface O can mediate the s-band center of the single-atomic Ag catalyst, thereby effectively affecting its catalytic property for the ORR. The best 2e⁻ ORR performance on Ag@Ti₂CO₂H_x was achieved at a H coverage of 22.2%, with an overpotential of 0.08 V and high selectivity of ca. 99.9%.

Author contributions

Fu-li Sun: methodology, validation, investigation, data curation, writing-original draft, writing-review & editing; Qiao-jun Fang: methodology, validation, writing-original draft; Wei Zhang, Cun-biao Lin, Wen-xian Chen: validation, writing-review & editing; Gui-lin Zhuang: conceptualization, software, resources, writing-review & editing, supervision.

Conflicts of interest

The authors declare no conflict of interest.

Acknowledgements

This work was supported by the National Natural Science Foundation of China (Grant No. 22022108, 22072135, 21671172), Zhejiang Provincial Natural Science Foundation of China (Grant No. LR19B010001, LTGY23B010001), Natural Science Foundation of Shandong Province (ZR2020ZD35).

References

1. G. Gao, Y. Tian, X. Gong, Z. Pan, K. Yang and B. Zong, Chin. J. Catal., 2020, 41, 1039–1047 CrossRef CAS .
2. P. C. Foller and R. T. Bombard, J. Appl. Electrochem., 1995, 25, 613–627 CrossRef CAS .
3. Y. Guo, C. Dai, Z. Lei, B. Chen and X. Fang, Catal. Today, 2016, 276, 36–45 CrossRef CAS .
4. R. J. Lewis and G. J. Hutchings, ChemCatChem, 2019, 11, 298–308 CrossRef CAS .
5. J. M. Campos-Martin, G. Blanco-Brieva and J. L. G. Fierro, Angew. Chem., Int. Ed., 2006, 45, 6962–6984 CrossRef CAS PubMed .
6. S. Siahrostami, S. J. Villegas, A. H. B. Mostaghimi, S. Back, A. B. Farimani, H. Wang, K. A. Persson and J. Montoya, ACS Catal., 2020, 10, 7495–7511 CrossRef CAS .
7. I. Yamanaka and T. Murayama, Angew. Chem., Int. Ed., 2008, 47, 1900–1902 CrossRef CAS PubMed .
8. C. Tang, L. Chen, H. Li, L. Li, Y. Jiao, Y. Zheng, H. Xu, K. Davey and S.-Z. Qiao, J. Am. Chem. Soc., 2021, 143, 7819–7827 CrossRef CAS PubMed .
9. R. Ma, G. Lin, Y. Zhou, Q. Liu, T. Zhang, G. Shan, M. Yang and J. Wang, npj Comput. Mater., 2019, 5, 78 CrossRef .
10. X. Zhang, X. Zhao, P. Zhu, Z. Adler, Z.-Y. Wu, Y. Liu and H. Wang, Nat. Commun., 2022, 13, 2880 CrossRef CAS PubMed .
11. Y. Wang, G. I. N. Waterhouse, L. Shang and T. Zhang, Adv. Energy Mater., 2021, 11, 2003323 CrossRef CAS .
12. J. Su, L. Zhuang, S. Zhang, Q. Liu, L. Zhang and G. Hu, Chin. Chem. Lett., 2021, 32, 2947–2962 CrossRef CAS .
13. C. Ye, N. Zhang, D. Wang and Y. Li, Chem. Commun., 2020, 56, 7687–7697 RSC .
14. Y. Mu, T. Wang, J. Zhang, C. Meng, Y. Zhang and Z. Kou, Electrochem. Energy Rev., 2022, 5, 145–186 CrossRef CAS .
15. J. Hafner, J. Comput. Chem., 2008, 29, 2044–2078 CrossRef CAS PubMed .
16. J. P. Perdew, K. Burke and M. Ernzerhof, Phys. Rev. Lett., 1996, 77, 3865–3868 CrossRef CAS PubMed .
17. J. P. Perdew, A. Ruzsinszky, G. I. Csonka, O. A. Vydrov, G. E. Scuseria, L. A. Constantin, X. Zhou and K. Burke, Phys. Rev. Lett., 2008, 100, 136406 CrossRef PubMed .
18. G. Kresse and D. Joubert, Phys. Rev. B: Condens. Matter Mater. Phys., 1999, 59, 1758–1775 CrossRef CAS .
19. M. Ernzerhof and G. E. Scuseria, J. Chem. Phys., 1999, 110, 5029–5036 CrossRef CAS .
20. T. Bučko, J. Hafner, S. Lebègue and J. G. Ángyán, J. Phys. Chem. A, 2010, 114, 11814–11824 CrossRef PubMed .
21. A. Maniopoulou, E. R. M. Davidson, R. Grau-Crespo, A. Walsh, I. J. Bush, C. R. A. Catlow and S. M. Woodley, Comput. Phys. Commun., 2012, 183, 1696–1701 CrossRef CAS .
22. D. J. Evans and B. L. Holian, J. Chem. Phys., 1985, 83, 4069–4074 CrossRef CAS .
23. K. Mathew, R. Sundararaman, K. Letchworth-Weaver, T. A. Arias and R. G. Hennig, J. Chem. Phys., 2014, 140, 084106 CrossRef PubMed .
24. K. Mathew, V. S. C. Kolluru, S. Mula, S. N. Steinmann and R. G. Hennig, J. Chem. Phys., 2019, 151, 234101 CrossRef PubMed .
25. Y. Cao, C. Zhao, Q. Fang, X. Zhong, G. Zhuang, S. Deng, Z. Wei, Z. Yao and J. Wang, J. Mater. Chem. A, 2020, 8, 2672–2683 RSC .
26. J. K. Nørskov, J. Rossmeisl, A. Logadottir, L. Lindqvist, J. R. Kitchin, T. Bligaard and H. Jónsson, J. Phys. Chem. B, 2004, 108, 17886–17892 CrossRef .
27. V. Wang, N. Xu, J.-C. Liu, G. Tang and W.-T. Geng, Comput. Phys. Commun., 2021, 267, 108033 CrossRef CAS .
28. W. Zhao, L. Zhang, Q. Luo, Z. Hu, W. Zhang, S. Smith and J. Yang, ACS Catal., 2019, 9, 3419–3425 CrossRef CAS .
29. X. Guo, S. Lin, J. Gu, S. Zhang, Z. Chen and S. Huang, ACS Catal., 2019, 9, 11042–11054 CrossRef CAS .
30. Z. Chen, G. Liu, W. Cao, L. Yang, L. Zhang, S. Zhang, J. Zou, R. Song, W. Fan, S. Luo and D. D. Dionysiou, Appl. Catal., B, 2023, 334, 122825 CrossRef CAS .
31. C. Jiang, Y.-F. Fei, W. Xu, Z. Bao, Y. Shao, S. Zhang, Z.-T. Hu and J. Wang, Appl. Catal., B, 2023, 334, 122867 CrossRef CAS .
32. H. Hussain, G. Tocci, T. Woolcot, X. Torrelles, C. L. Pang, D. S. Humphrey, C. M. Yim, D. C. Grinter, G. Cabailh, O. Bikondoa, R. Lindsay, J. Zegenhagen, A. Michaelides and G. Thornton, Nat. Mater., 2017, 16, 461–466 CrossRef CAS PubMed .
33. M. M. Balmeo, J. S. C. Dizon, M. J. F. Empizo, E. J. C. D. Solibet, V. C. Agulto, A. A. Salvador, N. Sarukura, H. Nakanishi, H. Kasai and A. A. B. Padama, Surf. Sci., 2021, 703, 121726 CrossRef CAS .
34. Y. Meng, X.-Y. Liu, M.-M. Bai, J. Chen, Y.-J. Ma and X.-D. Wen, Appl. Surf. Sci., 2020, 502, 144097 CrossRef CAS .
35. Y. Chen, F. Sun and Q. Tang, Phys. Chem. Chem. Phys., 2022, 24, 27302–27311 RSC .

---

Footnotes

† Electronic supplementary information (ESI) available. See DOI: https://doi.org/10.1039/d3ta04453j

‡ These authors contributed equally to this work.

This journal is © The Royal Society of Chemistry 2023