Metal single-atom coordinated graphitic carbon nitride as an efficient catalyst for CO oxidation

Abstract

Single-atom catalysts (SACs) often present outstanding activity due to their high ratio of low-coordinated metal atoms and can be applied to the activation of strong chemical bonds such as C≡O. Herein, we investigate the potential usage of a single-atom catalyst, in which isolated cobalt atoms are supported on porous graphitic carbon nitride (Co/g-C₃N₄), for CO oxidation. Based on the adsorption/co-adsorption energies of O₂, CO, 2O₂, CO + O₂ and 2CO, the screening criteria and the reaction mechanisms of CO oxidation, including the Eley–Rideal, New Eley–Rideal, Langmuir–Hinshelwood, and termolecular Eley–Rideal mechanisms, are established and compared. In particular, the energy barriers of the rate-limiting steps for the CO oxidation process by all possible reaction pathways are in a range from 0.21 to 0.59 eV, suggesting that the Co/g-C₃N₄ catalyst can boost CO oxidation at low temperature. Moreover, the preparation of the SAC (Co/g-C₃N₄) by using CoCl₂ as an appropriate metal precursor and the stability (up to 600 K) are evaluated by ab initio molecular dynamics simulations. The high stability and excellent activity of the Co/g-C₃N₄ SAC for CO oxidation offer a high possibility of clean energy production.

---

1. Introduction

The catalytic oxidation of carbon monoxide (CO) has attracted great interest due to its importance in various industrial^1,2 and environmental applications,^3,4 both from a fundamental point of view and for practical applications, e.g., removal of CO contamination from H₂-rich fuel gases for ammonia synthesis and low temperature fuel cell applications or reducing CO emission in exhaust gases from catalytic converters for cars, and serving as key prototype reactions for the design of heterogeneous catalysts.⁵

Over the past few years, SACs have become a hot topic due to their distinct advantages of maximized atomic utilization, uniform active sites with excellent stability and selectivity, and much improved catalytic activity over conventional metal nanoparticles.^6–10 Generally, metals are dispersed finely on a support with a high surface area for the efficient use of catalytically active components, which maximizes the metal atom efficiency. To date, various SACs^6,7,11,12 have been successfully synthesized with high activity for catalytic CO oxidation reactions. However, the synthesis of SACs with high stability and cost-efficiency remains a huge challenge. In 2011, Zhang and coworkers successfully uniformly deposited single Pt atoms on an iron oxide surface, which showed excellent stability and activity for CO oxidation with experimental and theoretical elucidation.⁶ Inspired by such an important study, various types of single metal atoms supported on Al₂O₃,¹³ CeO₂,^14–16 MgAl₂O₄,¹⁷ and AuTi₂O_3–6 ¹⁸ catalysts have been reported to be active for low-temperature CO oxidation. Additionally, it is not limited to successful stabilization of metal single atoms on a metal oxide surface, but can be developed further and made applicable to 2D materials based SACs. For instance, Zhang et al.¹⁹ demonstrated that various metal single atoms anchored on Ti₂CO₂ were oxidized by CO via Eley–Rideal (ER) and Langmuir–Hinshelwood (LH) mechanisms, and titanium atoms showed the best performance. Du et al.²⁰ reported MoS₂ as an excellent substrate for fabricating SACs, which can effectively oxidize CO with an energy barrier of the rate determining state of only 0.40 eV.

In recent years, porous 2D materials have been used as potential substrates to anchor metal single atoms in catalysis.^21–44 For example, SACs supported on graphdiyne, β₁₂ boron monolayers, h-BN and C₂N could serve as low-cost but highly efficient catalysts for CO₂ reduction,^21–25 hydrogen evolution reduction,^26–30 N₂ reduction,^27,31–35 CO oxidation,^36–39 and oxygen reduction/evolution reactions^29,40–43 from previous experimental and theoretical studies. Recently, highly active and stable single-atom Cu supported by a 2D metal–organic framework (Cu/UiO-66) catalyst has also been reported, which showed an excellent selectivity of about 100% for CO oxidation.⁴⁴

In particular, the excellent wide-bandgap semiconductor electro-/photo-catalyst 2D porous material of g-C₃N₄ exhibits high chemical stability and an appealing catalytic capability for splitting of water into hydrogen and oxygen.^45–50 Moreover, large periodic holes of g-C₃N₄ provide sufficient space as an excellent substrate for supporting catalysts. Thus, the introduction of SACs can effectively adjust the electronic structures, and the synergistic effect between SACs and g-C₃N₄ plays a crucial role in improving the catalytic activity of g-C₃N₄. Previous literature reports have confirmed that g-C₃N₄ can serve as an excellent support catalyst, and the coordination of single-atom Co and g-C₃N₄ shows a remarkable performance toward both the oxygen reduction and evolution reactions, which is even better than those of benchmark Pt and other carbon nitride-based electrocatalysts.^51–53 Inspired by the exciting experimental discovery, it is natural to ask: (I) whether Co/g-C₃N₄ also faces CO poisoning as found in Pt-based or other electrocatalysts?⁵⁴ In other words, whether the CO oxidation process on Co/g-C₃N₄ can be realized at low temperature? (II) If so, what is its mechanism? In this work, we systematically evaluated that the CO oxidation reaction on Co/g-C₃N₄ occurred through the ER, NER, TER and LH mechanisms, and found that the CO oxidation process can be realized with the energy barriers of the rate-limiting steps in a range from 0.21 to 0.59 eV, avoiding the CO poisoning of the catalyst. Thus, with the excellent oxygen reduction/evolution reactions and CO oxidation properties, Co/g-C₃N₄ can be further developed as a multi-functional catalyst for fuel cells.

2. Computational methods

Spin-polarized calculations were performed by using plane-wave density functional theory (DFT) as implemented in the Vienna Ab Initio Simulation Package (VASP) code.^55,56 The electronic exchange–correlation energy was calculated using the Perdew–Burke–Ernzerhof (PBE) functional within the generalized gradient approximation (GGA).^57,58 The projector augmented wave (PAW) method was used to describe the ionic cores.⁵⁹ The kinetic energy cutoff for the plane-wave basis set was chosen to be 450 eV. The vacuum space was 20 Å, which should be sufficient to avoid interactions between periodic images.^51,60,61 Supercells consisting of 2 × 2 × 1 g-C₃N₄ unit cells were used and the Brillouin zones were sampled using a Monkhorst–Pack k-point mesh with a 3 × 3 × 1 k-point grid.^51,62 The convergence criteria were set to 10⁻⁵ eV and −0.01 eV Å⁻¹ for the energy and the forces on each atom, respectively.⁶³

To assess the thermodynamic stability of Co/g-C₃N₄, ab initio molecular dynamics (AIMD) simulations were performed. The atomic charge transfer in different systems was evaluated by Bader's charge analysis.⁶⁴ To investigate the kinetic processes of O₂ activation and CO oxidation, the climbing image nudged elastic band method⁶⁵ (CI-NEB) was employed to identify the minimum energy pathways, and each transition state structure was further confirmed by vibrational analysis. Further calculation details of the binding energies and the activation energy barriers are given in the ESI.†

3. Results and discussion

3.1. The stability and electronic structures of the Co/g-C₃N₄ catalyst

To find the most favorable deposition sites for the single-atom catalysts, three possible Co locations on a 2 × 2 × 1 g-C₃N₄ supercell were considered, as shown in Fig. 1a. After geometry optimization, Co connected with the two adjacent pyridinic-N atoms from two separate triazine units, forming a CoN₃C₂ ring (Fig. 1b). The Co atom exhibits a short bond length with N of 1.95 Å and the strongest bonding of 3.25 eV, which are much larger than those of Pd and Pt embedded in g-C₃N₄ ⁶⁰ (2.17 and 2.95 eV, respectively). Moreover, the mobility of the Co atom needs to overcome a rather big energy barrier of 1.98 eV, as shown in Fig. S1.† The large diffusion barrier indicates that it is difficult for the isolated Co atom to form clusters on g-C₃N₄. Moreover, the relative energy (ΔE) of two cobalt atoms scattered at two sites (Fig. S2a†) is lower than that of cobalt dimers adsorbed at one site (Fig. S2b and c†) of g-C₃N₄, suggesting that isolated distribution is more preferred compared with forming clusters. Furthermore, we calculated the energy barrier of a Co dimer dissociated into two separated Co monomers at two pore sites of g-C₃N₄, which is energetically favorable, as shown in Fig. S3.† A moderate energy barrier of 0.58 eV implies that the relocation of the Co dimer into two pore sites of g-C₃N₄ readily occurs under ambient conditions.

Fig. 1 (a) Schematic of adsorption of Co atoms on hexagonal holes, bridge sites, and pentagonal holes of g-C₃N₄ and (b) the most stable adsorption structure is illustrated in the right panel where the Co atom is connected to two adjacent pyridinic-N atoms from two separate triazine units, forming a CoN₃C₂ ring on g-C₃N_4. (c) The density of states (DOS) and (d) variations of temperature and energy fluctuations with time progress for AIMD simulations of Co/g-C₃N₄. The simulation is run at 600 K for 10 ps with a time step of 2 fs. Color scheme: gray, blue and magenta represent C, N and Co atoms, respectively.

The electronic structure and bonding features are analyzed by means of density of states (DOS), which can provide a fundamental understanding of the interaction between the single Co atom and the substrate as well as the activity. As shown in Fig. 1c, strong coupling between the d-orbitals of Co and the p-orbitals of N is observed in the energy range from −7.5 to 2.5 eV, indicating the strong interaction between Co and the g-C₃N₄ substrate. The DOS of Co/g-C₃N₄ shows a higher peak at the Fermi energy, which is attributed to the change in the electronic structure of the outermost Co modified by g-C₃N₄, indicating high activity of supported Co on g-C₃N₄. In addition, significant amounts of electrons (0.84 e⁻ based on Bader charge analysis) are transferred from the Co atom to the g-C₃N₄ sheet. The stability of Co/g-C₃N₄ was further evaluated by performing AIMD simulations, in which the time step was set as 2 fs for a total period of 10 ps. As shown in Fig. S4,† the energy fluctuations with time evolution oscillates near the equilibrium state and the snapshot of the atomic configuration remains very well at 400 K. Even when the temperature increases to 600 K (Fig. 1d), there is still no significant distortion of the geometric structure (Fig. S5†), indicating the high thermodynamic stability of Co/g-C₃N₄.

3.2 Adsorption of reaction species on the Co/g-C₃N₄ catalyst

It is important to investigate the interactions of adsorbates and the catalyst and gain fundamental insights into the catalytic performance. Thus, we first study the adsorption/co-adsorption of O₂ and CO on Co/C₃N₄, and then explore their corresponding reaction mechanisms. As shown in Fig. 2a, O₂ prefers to adsorb at the Co-Top site with the O–O bond parallel to the g-C₃N₄ sheet forming two Co–O bonds of 1.80 Å. In contrast to the O–O bond length (d_O–O) of free O₂ (1.23 Å), the d_O–O of the adsorbed O₂ molecule is elongated by about 0.15 Å on Co/g-C₃N₄ (1.38 Å). The adsorption energy of 2.12 eV is accompanied by a large amount of charge transfer (0.32 electrons) from Co/g-C₃N₄ to O₂. Moreover, the 1π and 5σ orbitals of O₂ at the energy region of −12.5 eV to the Fermi level well overlap with Co 3d and O₂-2π* with a higher peak appearing around the Fermi level (Fig. 2b), suggesting the strong interaction between the Co atom and O₂ molecule. In summary, the enlarged O–O bond, the charge transfer and the strong hybridization of Co 3d and O₂ 2p states demonstrate that the O₂ molecule is activated on Co/g-C₃N₄.

Fig. 2 Adsorption structures and energies of (a) O₂, (c) CO, (e) CO₂, (f) CO + O₂, (g) 2CO and (h) 2O₂ on Co/g-C₃N₄. DOS of (b) O₂ and (c) CO adsorbed on Co/g-C₃N₄ are also shown in the figure. Color scheme: red represents O atoms.

The most stable CO adsorption configuration and corresponding DOS are presented in Fig. 2c and d, respectively. It is found that the CO molecule prefers to adsorb vertically at the Co-Top site by forming Co–C bonds of 1.76 Å and the C–O bond is elongated as compared with the free CO molecule (1.21 vs. 1.14 Å). The adsorption energy is 1.74 eV, which is smaller than that of O₂ adsorption, suggesting that the adsorption of O₂ is slightly more preferable than that of CO. The Bader charge analysis shows that the adsorbed CO transferred 0.16 electrons to Co/g-C₃N₄, which is in line with the result that the 4σ*, 5σ and 1π orbitals of CO in the energy range of −10 to −5 eV are hybridized with the Co 3d states. Upon CO adsorption, CO-2π* orbitals are also partially occupied because some electrons transfer from Co to CO-2π*. The charge transfer and the strong hybridization of the Co atom and CO molecule lead to the elongation of the C–O bond from 1.14 to 1.21 Å.

As the final product of CO oxidation, CO₂ is physisorbed on Co/g-C₃N₄ with a rather small adsorption energy of 0.03 eV, as shown in Fig. 2e. The long distance of the Co–O bond (∼3.00 Å) and the weak interaction together with the negligible charge transfer between CO₂ and the support indicate that the formed CO₂ species would be spontaneously released.

The most stable adsorption structure of atomic O on Co/g-C₃N₄ is presented in Fig. S6,† in which atomic O prefers to adsorb on the top site of Co. Compared with that of O₂ adsorption (Fig. 2a), the adsorption energy of atomic O is reduced to 1.37 eV and the Co–O bond is elongated to 1.84 Å. The adsorbed O atom is negatively charged by 0.23 electrons, indicating the formation of the O˙⁻ species. The O˙⁻ species is believed to be more reactive than O₂ and would present great activity to oxidize CO.

In order to investigate the reaction mechanism by which CO oxidation proceeds on the catalyst surface, it is also necessary to study the capture performance (co-adsorption of the reactants). Note that the co-adsorption energy of CO + O₂ (Fig. 2f) and 2CO (Fig. 2g) is significantly larger than the individual E_ads of CO or O₂ (3.26 and 3.22 vs. 1.74 or 2.12 eV), and larger than that of 2O₂ (Fig. 2h) as well (2.49 eV). The large adsorption energies and the small adsorption energy difference between CO and O₂ may facilitate the catalytic reaction of CO oxidation and decrease the reaction barriers accordingly.

According to the adsorption/co-adsorption energies of O₂, CO, 2O₂, CO + O₂ and 2CO, the screening criteria and reaction mechanisms of CO oxidation based on SACs are thus proposed. As shown in Scheme 1: (I) the adsorption of O₂ is slightly more preferable than that of CO, in which the ER mechanism (i.e., the reactant CO molecule approaching an already activated O₂) would take place. (II) The co-adsorption of O₂ and CO is stronger than the isolate O₂ adsorption (low CO partial pressure), and the LH mechanism is preferred. (III) As two CO molecules are pre-adsorbed on the catalyst surface (high CO partial pressure), a TER mechanism, in which a free O₂ molecule can be activated by two co-adsorbed CO molecules, can dominate the CO oxidation process.^66,67

Scheme 1 The screening criteria, reaction mechanism and reaction route of CO oxidation.

3.3 CO oxidation on the Co/g-C₃N₄ catalyst

3.3.1 ER mechanism. As the adsorption of CO is weaker than that of O₂ (1.74 vs. 2.12 eV) on Co/g-C₃N₄, the CO oxidation process probably follows the ER mechanism: CO + O₂* → CO₂ + O* and CO + O* → CO₂. In Fig. 3a, the incoming CO molecule directly attacks the pre-adsorbed O₂ species and consequently produces one linear CO₂ molecule with an exothermic energy of 2.53 eV. In this process of CO + O₂* → CO₂ + O*, CO physisorbed around pre-adsorbed O₂ with a distance of 3.45 Å (Fig. 3a-IS). Then CO approaches activated O₂ by overcoming a small energy barrier of 0.25 eV, which is accompanied by the breaking of the O–O bond and the first CO₂ molecule can be readily emitted. Thus, the formation of the first CO₂ molecule is both thermodynamically and kinetically possible. After that, one O atom is still left from the dissociated O₂ molecule, which can be further readily captured by the second approaching CO molecule via the ER mechanism of CO + O* → CO₂. The removal of O by the CO molecule requires a barrier of 0.40 eV to produce the second CO₂ molecule on Co/g-C₃N₄, which is smaller than those on V/g-C₃N₄ (0.90 eV) and Cr/g-C₃N₄ (0.52 eV).⁶⁸ From the thermodynamics, this reaction process is found to release 1.94 eV of heat, suggesting that this step is feasible in the experiment.

Fig. 3 Minimum energy profiles and the configurations of different states including the IS, TS, MS and FS for CO oxidation on Co/g-C₃N₄. (a) ER mechanism: CO + O₂* → CO₂ + O* and CO + O* → CO₂, (b) NER mechanism: 2CO + O₂* → OOCCOO* → CO₂, (c) LH mechanism: CO* + O₂* → CO₂ + O* and CO* + O* → OCO* → CO₂, and (d) TER mechanism: O₂ + 2CO* → OCOOCO* → 2CO₂. Here, * denotes a catalytically active site.

3.3.2 NER mechanism. Recently, theoretical studies^39,69 have shown that two CO molecules could assist in O₂ scission and promote CO oxidation, and the CO oxidation process probably follows the NER mechanism: 2CO + O₂* → OOCCOO* → 2CO₂. As shown in Fig. 3b, the O₂ molecule is activated by two physisorbed CO molecules thereby forming two CO₂ molecules simultaneously. The configuration of the two physisorbed CO molecules suspended above the pre-adsorbed O₂ molecules was selected as the IS, where 2CO is close to the O₂ molecule with an intermolecular distance of ∼3.53 Å. As the CO molecule approaches O₂, the O–O bond would gradually stretch resulting in bond cleavage and the formation of two new C–O bonds, and then a pentagonal ring OOCCOO intermediate (MS) is formed. The reaction process of 2CO + O₂* → OOCCOO* would easily occur with a relatively low energy barrier of 0.23 eV and an exothermic reaction energy of 4.29 eV. This reaction pathway confirms that two CO molecules play a key role in promoting O–O bond scission. Upon C–C bond cleavage, the OOCCOO intermediate dissociates to form two CO₂ molecules, in which the reaction process of OOCCOO* → 2CO₂ would be exothermic by 0.43 eV and needs to pass an energy barrier of 0.35 eV. The barriers of OOCCOO formation and dissociation are slightly lower than CO oxidation reactions of CO + O₂* → CO₂ + O* and CO + O* → CO₂ on Co/g-C₃N₄via the ER mechanism (Fig. 3a). Recently, Li et al.⁶⁹ reported a new NER mechanism in which two CO molecules approach the pre-adsorbed O₂ at the same time, and yield two CO₂ molecules simultaneously. As a comparison, we also considered the new NER mechanism of 2CO + O₂* → 2CO₂ (Fig. S7†), in which this process releasing 4.42 eV of heat needs to pass an energy barrier of 0.46 eV.

3.3.3 LH mechanism. Considering the co-adsorption of CO and O₂, the LH mechanism (Fig. 3c) is then investigated. Different from the ER mechanism (Fig. 3a), once CO and O₂ are co-adsorbed on Co/g-C₃N₄, the C atom of CO would approach one O atom of O₂. When the transitional state (TS) passes, the first CO₂ molecule would be released spontaneously from the catalyst with the cleavage of the O–O bond, leaving an unreacted O atom of O₂ adsorbed on the Co site. This step of O₂* + CO* → CO₂ + O* releases 1.65 eV of heat and needs to overcome only a low energy barrier of 0.10 eV. Subsequently, a second CO molecule is adsorbed on the Co site as IS1, which is exothermic of 1.35 eV (the step of O* + CO* → OCO*). Then, the attached CO molecule approaches O with an extremely low energy barrier of 0.08 eV and releases 1.05 eV of heat, forming the intermediate OCO (MS). Finally, after TS2 (the step of OCO* → CO₂), the Co/g-C₃N₄ catalyst releases the second CO₂ molecule via the LH mechanism, which is exothermic by 0.23 eV with an energy barrier of 0.59 eV.

Compared with the reaction process of CO + O₂* → CO₂ + O* in the ER mechanism, the first CO₂ molecule is easier to be released via the LH mechanism of O₂* + CO* → CO₂ + O* (0.25 vs. 0.10 eV). However, CO could readily react with an unreacted O atom for releasing the second CO₂ molecule via the ER mechanism (CO + O* → CO₂), which is superior to the LH mechanism (O* + CO* → OCO* → CO₂) on the Co/g-C₃N₄ catalyst (0.40 vs. 0.59 eV). After releasing two CO₂ molecules, the Co/g-C₃N₄ catalyst could be recovered for a new cycle of CO oxidation.

3.3.4 TER mechanism. Alternatively, the stronger co-adsorption structure of two CO molecules on Co/g-C₃N₄ can be used as a potential mechanism for CO oxidation as well. Fig. 3d shows the reaction profiles of CO oxidation via the TER mechanism. The most stable adsorption configuration of an O₂ molecule suspended above the two co-adsorbed CO molecules is selected as the IS. With the gradual decrease of the distance between the C atoms from CO and O₂ molecules, the O–O bond gradually stretches, forming an intermediate state of OCOOCO (MS). In this process of O₂ + 2CO* → OCOOCO*, the O–O bond is elongated from 1.24 to 1.48 Å and two new C–O bonds are formed, which would easily occur with a low-energy barrier of 0.17 eV and release 0.37 eV of heat. After TS1, the O–O bond is broken with an energy barrier of 0.21 eV, and then the OCOOCO intermediate dissociates to form two adsorbed CO₂ molecules. The small reaction barriers and a large exothermic energy of 3.24 eV show that the release of the CO₂ molecule and the decomposition of MS would readily take place.

Compared with the ER, NER and LH mechanisms, the TER mechanism is preferable on Co/g-C₃N₄. The high spin density of the single Co atom during the CO oxidation process via the TER mechanism (Fig. S8†) is regarded as a key factor for the higher activity of Co/g-C₃N₄ SACs.

Moreover, the rate-limiting-steps of CO oxidation of Co/g-C₃N₄ are compared with those of metal single-atom supported catalysts (Table 1). Overall, the energy barriers with TER mechanisms on Co/g-C₃N₄ are lower than those on other substrates such as C₂N,³⁷h-BN,³⁸ graphene,⁷⁰ Mo₂CO₂,⁷¹ Ti₂CO₂,¹⁹ SnO₂,⁷² and FeO_x,⁶ suggesting that Co/g-C₃N₄ catalysts have better tolerance against CO poisoning.

Table 1 The rate-limiting-steps and the corresponding energy barriers of CO oxidation on various substrates via the corresponding mechanisms

Substrate	Barrier^mechanism (eV)	Substrate	Barrier^mechanism (eV)
Co/g-C3N4 (this work)	0.21^TER, 0.35^NER, 0.40^ER, 0.59^LH	Ti/Ti2CO2 ^19	0.25^ER, 0.35^LH
Cr/g-C2N^37	0.59^ER	Pt/Mo2CO2 ^71	1.39^LH, 0.49^TER
Mn/g-C2N^37	0.64^ER	Pt/graphene^70	0.77^ER, 0.59^LH
V/g-C2N^37	1.01^ER	Pt/SnO2 ^72	0.51^ER
g-C3N4/Pt(111)^75	1.31^ER, 0.77^LH	Pt/FeOX ^6	0.79^LH
Co/h-BN^38	0.52^ER	Co/FeOX ^76	0.45^LH

---

Based on the formula of the Sabatier rate⁷³ of CO₂ formation described in the computational details, there are six parameters that determine the kinetics: E_ads (CO), E_ads (O₂), E_ads (CO + O₂), E_ads (CO + CO), E_IS and E_TS. In order to find the independent variables characterizing the SAC in the microkinetic model, it is necessary to associate the adsorption energies of reactants with relative energies of the different states along the MEPs. Therefore, we focus on finding the Brønsted–Evans–Polanyi (BEP) linear correlation between E_IS and E_TS on Co/g-C₃N₄ from the adsorption energy of CO, O₂, O, CO + O₂, O + CO, and CO + CO. The E_TS as a function of the E_IS is shown in Fig. 4, with the data (Table S1†) fitted by using linear regression. E_TS can be correlated with E_IS using the following formula

E_TS = 0.99 × E_IS − 0.24 (R² = 0.98).

Fig. 4 BEP linear correlations between the initial-state energies (E_IS) and transition-state energies (E_TS) for CO oxidation on Co/g-C₃N₄ catalysts via different mechanisms of ER, NER, LH and TER. E_IS corresponding to the adsorption energy of CO, O₂, O, CO + O₂, O + CO and 2CO.

Our results indicate the BEP relationships that have been established for different SACs,⁷⁴ which can also describe CO oxidation via different mechanisms of ER, NER, LH and TER on Co/g-C₃N₄. Thus, the idea of Sabatier analysis may be applied to identify activity trends with different CO oxidation mechanisms on SACs.

3.4 The feasibility of Co/g-C₃N₄ in experiment

A wet chemical method is a suitable method to achieve high dispersion of single atoms, which can solve the problem of the possibility of experimental synthesis of SACs. Using CoCl₂ as an appropriate metal precursor, the feasibility of experimental synthesis of Co/g-C₃N₄ is verified by calculating the energy profile of the synthetic route (Fig. 5a) and simulating the synthesis process by AIMD at 350 k (Fig. 5c). The formation of Co/g-C₃N₄ was simulated by using the following synthetic route (Fig. 5a).

CoCl₂ + * → *CoCl₂ (step 1)

*CoCl₂ + 2H₃O + 2e⁻ → *Co(HCl)₂ + 2H₂O (step 2)

*Co(HCl)₂ → *Co + 2HCl (step 3)

where * denotes the g-C₃N₄ sheet.

Fig. 5 (a) Designed synthetic route for Co/g-C₃N₄. Here * denotes g-C₃N₄. (b) Energy profile of the proposed synthetic route. (c) Snapshots of atomic configurations of AIMD simulations for the synthetic process of Co/g-C₃N₄. The upper panel shows the initial structure, and the lower panel shows the final structure. The model consists of a 2 × 2 × 1 supercell of g-C₃N₄, two CoCl₂ molecules, and 25H₂O molecules. The simulation is run at 350 K for 2 ps with a time step of 1 fs. Color scheme: white and pale green represent H and Cl atoms, respectively.

Adsorption of CoCl₂ on a g-C₃N₄ monolayer is the first step with an adsorption energy of 1.12 eV. In the second step, H⁺ ions in the H₃O⁺ groups approach Cl⁻ ions in the adsorbed CoCl₂*, and then H₃O⁺ groups dissociate to form H–Cl bonds and produce H₂O molecules; such a process is spontaneous and barrierless. Finally (in step 3), the HCl groups can be easily released from the surface with a binding energy as low as 0.11 eV, resulting in the formation of Co/g-C₃N₄. All the reaction steps can easily occur since they are spontaneous (Fig. 5b). In our AIMD simulations (Fig. 5c), CoCl₂ in the solution is observed to adsorb on g-C₃N₄, and afterward the Cl⁻ ions are desorbed, and the ultimate formation of the Co/g-C₃N₄ system occurs within 2 ps at 350 K. Thus, the highly stable and efficient Co/g-C₃N₄ catalysts may be synthesized by using CoCl₂ as an appropriate metal precursor.

4. Conclusions

In summary, we have demonstrated that a g-C₃N₄ supported metal single-atom catalyst (Co/g-C₃N₄) exhibits excellent stability and catalytic activity for CO oxidation computationally. The stability of the SAC is guaranteed by the strong binding strength between metals and the substrate, the high diffusion barrier of the metals, and excellent thermal stability. The superior catalytic activity of Co/g-C₃N₄ is verified by its considerably low kinetic energy barriers for all possible reaction mechanisms, ranging from 0.21 to 0.59 eV. In addition, the results also show that Co supported on g-C₃N₄ shows higher activity for CO oxidation than on other substrates such as h-BN and graphene, and other kinds of carbon nitrides, MXenes and metal oxides. Based on the adsorption and co-adsorption energies of CO, O₂, 2O₂, CO + O₂ and 2CO, the BEP relationships have been established for different reaction mechanisms on the Co/g-C₃N₄ catalyst, which can quantitatively assess the catalytic activity trends and provide guidelines for designing new SACs. Therefore, this work not only spans the non-noble metal single atom catalysts, but also provides useful insights and guidance for the design and development of novel low-cost and efficient SACs for many other important reactions.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

This work is supported by the Natural Science Foundation of China (21525311, 21773027 and 21703032). We thank the Big Data Center of Southeast University for providing the facility support for the numerical calculations in this paper.

References

1. J. Kua and W. A. Goddard, J. Am. Chem. Soc., 1999, 121, 10928–10941.
2. J. Saavedra, T. Whittaker, Z. Chen, C. J. Pursell, R. M. Rioux and B. D. Chandler, Nat. Chem., 2016, 8, 584–589.
3. M. V. Twigg, Appl. Catal., B, 2007, 70, 2–15.
4. M. Shelef and R. W. McCabe, Catal. Today, 2000, 62, 35–50.
5. W. Lubitz and W. Tumas, Chem. Rev., 2007, 107, 3900–3903.
6. B. Qiao, A. Wang, X. Yang, L. F. Allard, Z. Jiang, Y. Cui, J. Liu, J. Li and T. Zhang, Nat. Chem., 2011, 3, 634.
7. X.-F. Yang, A. Wang, B. Qiao, J. Li, J. Liu and T. Zhang, Acc. Chem. Res., 2013, 46, 1740–1748.
8. J. Jones, H. Xiong, A. T. DeLaRiva, E. J. Peterson, H. Pham, S. R. Challa, G. Qi, S. Oh, M. H. Wiebenga, X. I. Pereira Hernández, Y. Wang and A. K. Datye, Science, 2016, 353, 150.
9. B. Zhang, H. Asakura, J. Zhang, J. Zhang, S. De and N. Yan, Angew. Chem., Int. Ed., 2016, 55, 8319–8323.
10. A. Wang, J. Li and T. Zhang, Nat. Rev. Chem., 2018, 2, 65–81.
11. W. Zhang and W. Zheng, Adv. Funct. Mater., 2016, 26, 2988–2993.
12. J. Liu, ACS Catal., 2017, 7, 34–59.
13. M. Moses-DeBusk, M. Yoon, L. F. Allard, D. R. Mullins, Z. Wu, X. Yang, G. Veith, G. M. Stocks and C. K. Narula, J. Am. Chem. Soc., 2013, 135, 12634–12645.
14. B. Qiao, J. Liu, Y.-G. Wang, Q. Lin, X. Liu, A. Wang, J. Li, T. Zhang and J. Liu, ACS Catal., 2015, 5, 6249–6254.
15. L. Nie, D. Mei, H. Xiong, B. Peng, Z. Ren, X. I. P. Hernandez, A. DeLaRiva, M. Wang, M. H. Engelhard, L. Kovarik, A. K. Datye and Y. Wang, Science, 2017, 358, 1419–1423.
16. C. Wang, X.-K. Gu, H. Yan, Y. Lin, J. Li, D. Liu, W.-X. Li and J. Lu, ACS Catal., 2016, 7, 887–891.
17. Y. Lu, J. Wang, L. Yu, L. Kovarik, X. Zhang, A. S. Hoffman, A. Gallo, S. R. Bare, D. Sokaras, T. Kroll, V. Dagle, H. Xin and A. M. Karim, Nat. Catal., 2018, 2, 149–156.
18. J.-J. Chen, X.-N. Li, Q. Chen, Q.-Y. Liu, L.-X. Jiang and S.-G. He, J. Am. Chem. Soc., 2019, 141, 2027–2034.
19. X. Zhang, J. Lei, D. Wu, X. Zhao, Y. Jing and Z. Zhou, J. Mater. Chem. A, 2016, 4, 4871–4876.
20. C. Du, H. Lin, B. Lin, Z. Ma, T. Hou, J. Tang and Y. Li, J. Mater. Chem. A, 2015, 3, 23113–23119.
21. C. Yan, H. Li, Y. Ye, H. Wu, F. Cai, R. Si, J. Xiao, S. Miao, S. Xie, F. Yang, Y. Li, G. Wang and X. Bao, Energy Environ. Sci., 2018, 11, 1204–1210.
22. W. Bi, X. Li, R. You, M. Chen, R. Yuan, W. Huang, X. Wu, W. Chu, C. Wu and Y. Xie, Adv. Mater., 2018, 30, 1706617.
23. C. Zhang, S. Yang, J. Wu, M. Liu, S. Yazdi, M. Ren, J. Sha, J. Zhong, K. Nie, A. S. Jalilov, Z. Li, H. Li, B. I. Yakobson, Q. Wu, E. Ringe, H. Xu, P. M. Ajayan and J. M. Tour, Adv. Energy Mater., 2018, 8, 1703487.
24. C. Ling, Q. Li, A. Du and J. Wang, ACS Appl. Mater. Interfaces, 2018, 10, 36866–36872.
25. X. Cui, W. An, X. Liu, H. Wang, Y. Men and J. Wang, Nanoscale, 2018, 10, 15262–15272.
26. J. Mahmood, S.-M. Jung, S.-J. Kim, J. Park, J.-W. Yoo and J.-B. Baek, Chem. Mater., 2015, 27, 4860–4864.
27. L. Hui, Y. Xue, H. Yu, Y. Liu, Y. Fang, C. Xing, B. Huang and Y. Li, J. Am. Chem. Soc., 2019, 141, 10677–10683.
28. Y. Xue, B. Huang, Y. Yi, Y. Guo, Z. Zuo, Y. Li, Z. Jia, H. Liu and Y. Li, Nat. Commun., 2018, 9, 1460.
29. C. Ling, L. Shi, Y. Ouyang, X. C. Zeng and J. Wang, Nano Lett., 2017, 17, 5133–5139.
30. C. Hu, H. Liu, Y. Liu, J.-F. Chen, Y. Li and L. Dai, Nano Energy, 2019, 103874.
31. C. Choi, S. Back, N.-Y. Kim, J. Lim, Y.-H. Kim and Y. Jung, ACS Catal., 2018, 8, 7517–7525.
32. X. F. Li, Q. K. Li, J. Cheng, L. Liu, Q. Yan, Y. Wu, X. H. Zhang, Z. Y. Wang, Q. Qiu and Y. Luo, J. Am. Chem. Soc., 2016, 138, 8706–8709.
33. C. Ling, Y. Ouyang, Q. Li, X. Bai, X. Mao, A. Du and J. Wang, Small Methods, 2018, 1800376.
34. J. Zhao and Z. Chen, J. Am. Chem. Soc., 2017, 139, 12480–12487.
35. H.-R. Zhu, Y.-L. Hu, S.-H. Wei and D.-Y. Hua, J. Phys. Chem. C, 2019, 123, 4274–4281.
36. D. W. Ma, T. Li, Q. Wang, G. Yang, C. He, B. Ma and Z. Lu, Carbon, 2015, 95, 756–765.
37. D. W. Ma, Q. Wang, X. Yan, X. Zhang, C. He, D. Zhou, Y. Tang, Z. Lu and Z. Yang, Carbon, 2016, 105, 463–473.
38. S. Lin, X. Ye, R. S. Johnson and H. Guo, J. Phys. Chem. C, 2013, 117, 17319–17326.
39. G. Xu, R. Wang, Y. Ding, Z. Lu, D. Ma and Z. Yang, J. Phys. Chem. C, 2018, 122, 23481–23492.
40. H. Fei, J. Dong, Y. Feng, C. S. Allen, C. Wan, B. Volosskiy, M. Li, Z. Zhao, Y. Wang, H. Sun, P. An, W. Chen, Z. Guo, C. Lee, D. Chen, I. Shakir, M. Liu, T. Hu, Y. Li, A. I. Kirkland, X. Duan and Y. Huang, Nat. Catal., 2018, 1, 63–72.
41. H. Xu, D. Cheng, D. Cao and X. C. Zeng, Nat. Catal., 2018, 1, 339–348.
42. T. He, S. K. Matta, G. Will and A. Du, Small Methods, 2019, 1800419.
43. X. Zhang, A. Chen, Z. Zhang, M. Jiao and Z. Zhou, J. Mater. Chem. A, 2018, 6, 11446–11452.
44. A. M. Abdel-Mageed, B. Rungtaweevoranit, M. Parlinska-Wojtan, X. Pei, O. M. Yaghi and R. J. Behm, J. Am. Chem. Soc., 2019, 141, 5201–5210.
45. X. Wang, X. Chen, A. Thomas, X. Fu and M. Antonietti, Adv. Mater., 2009, 21, 1609–1612.
46. X. Wang, K. Maeda, X. Chen, K. Takanabe, K. Domen, Y. Hou, X. Fu and M. Antonietti, J. Am. Chem. Soc., 2009, 131, 1680–1681.
47. J. Zhang, X. Chen, K. Takanabe, K. Maeda, K. Domen, J. D. Epping, X. Fu, M. Antonietti and X. Wang, Angew. Chem., Int. Ed., 2010, 49, 441–444.
48. Y. Wang, X. Wang and M. Antonietti, Angew. Chem., Int. Ed., 2012, 51, 68–89.
49. X. Wang, K. Maeda, A. Thomas, K. Takanabe, G. Xin, J. M. Carlsson, K. Domen and M. Antonietti, Nat. Mater., 2009, 8, 76–80.
50. X. Chen, J. Zhang, X. Fu, M. Antonietti and X. Wang, J. Am. Chem. Soc., 2009, 131, 11658–11659.
51. Y. Zheng, Y. Jiao, Y. Zhu, Q. Cai, A. Vasileff, L. H. Li, Y. Han, Y. Chen and S. Z. Qiao, J. Am. Chem. Soc., 2017, 139, 3336–3339.
52. W. Niu, K. Marcus, L. Zhou, Z. Li, L. Shi, K. Liang and Y. Yang, ACS Catal., 2018, 8, 1926–1931.
53. Y. Wu, C. Li, W. Liu, H. Li, Y. Gong, L. Niu, X. Liu, C. Sun and S. Xu, Nanoscale, 2019, 11, 5064–5071.
54. M. Montano, K. Bratlie, M. Salmeron and G. A. Somorjai, J. Am. Chem. Soc., 2006, 128, 13229–13234.
55. G. Kresse and J. Furthmüller, Phys. Rev. B: Condens. Matter Mater. Phys., 1996, 54, 11169–11186.
56. G. Kresse and D. Joubert, Phys. Rev. B: Condens. Matter Mater. Phys., 1999, 59, 1758–1775.
57. J. P. Perdew, J. A. Chevary, S. H. Vosko, K. A. Jackson, M. R. Pederson, D. J. Singh and C. Fiolhais, Phys. Rev. B: Condens. Matter Mater. Phys., 1992, 46, 6671–6687.
58. J. P. Perdew and Y. Wang, Phys. Rev. B: Condens. Matter Mater. Phys., 1992, 45, 13244–13249.
59. P. E. Blöchl, Phys. Rev. B: Condens. Matter Mater. Phys., 1994, 50, 17953–17979.
60. G. Gao, Y. Jiao, E. R. Waclawik and A. Du, J. Am. Chem. Soc., 2016, 138, 6292–6297.
61. Z. Chen, J. Zhao, C. R. Cabrera and Z. Chen, Small Methods, 2018, 1800368.
62. C. Ling, X. Niu, Q. Li, A. Du and J. Wang, J. Am. Chem. Soc., 2018, 140, 14161–14168.
63. C. Liu, Q. Li, C. Wu, J. Zhang, Y. Jin, D. R. MacFarlane and C. Sun, J. Am. Chem. Soc., 2019, 141, 2884–2888.
64. G. Henkelman, A. Arnaldsson and H. Jónsson, Comput. Mater. Sci., 2006, 36, 354–360.
65. G. Henkelman, B. P. Uberuaga and H. Jónsson, J. Chem. Phys., 2000, 113, 9901–9904.
66. C. Liu, Y. Tan, S. Lin, H. Li, X. Wu, L. Li, Y. Pei and X. C. Zeng, J. Am. Chem. Soc., 2013, 135, 2583–2595.
67. K. Mao, L. Li, W. Zhang, Y. Pei, X. C. Zeng, X. Wu and J. Yang, Sci. Rep., 2014, 4, 5441.
68. S. L. Li, H. Yin, X. Kan, L. Y. Gan, U. Schwingenschlogl and Y. Zhao, Phys. Chem. Chem. Phys., 2017, 19, 30069–30077.
69. F. Li and Z. Chen, Nanoscale, 2018, 10, 15696–15705.
70. Y. Tang, Z. Yang and X. Dai, Phys. Chem. Chem. Phys., 2012, 14, 16566–16572.
71. C. Cheng, X. Zhang, M. Wang, S. Wang and Z. Yang, Phys. Chem. Chem. Phys., 2018, 20, 3504–3513.
72. S. Li, Z. Lu, Z. Yang and X. Chu, Sens. Actuators, B, 2014, 202, 83–92.
73. T. Bligaard, J. K. Nørskov, S. Dahl, J. Matthiesen, C. H. Christensen and J. Sehested, J. Catal., 2004, 224, 206–217.
74. H. Xu, C.-Q. Xu, D. Cheng and J. Li, Catal. Sci. Technol., 2017, 7, 5860–5871.
75. S. Wang, Y. Feng, M. Yu, Q. Wan and S. Lin, ACS Appl. Mater. Interfaces, 2017, 9, 33267–33273.
76. F. Li, Y. Li, X. C. Zeng and Z. Chen, ACS Catal., 2014, 5, 544–552.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c9nr07726j

---