Electrocatalytic Oxygen Reduction to Hydrogen Peroxide on Graphdiyne-Based Single-Atom Catalysts: First-Principles Studies

Abstract

The electrocatalytic oxygen reduction reaction (2e⁻ ORR) via a two-electron process is a promising pathway for the production of hydrogen peroxide (H₂O₂). Here, we systematically investigated the 2e⁻ ORR process on graphdiyne (GDY) supported single transition metal atoms (TM₁@GDY) using density functional theory (DFT) calculations. Among the 23 TM₁@GDY catalysts, Pt₁@GDY showed the best performance for the H₂O₂ product with an overpotential as low as 0.15 V. The electronic structure analysis, on the one hand, elucidates that the electron transfer between Pt₁@GDY and the adsorbed O₂ facilitates the activation of O₂, and, on the other hand, reveals that the high 2e⁻ ORR activity of Pt₁@GDY lies in the transfer of electrons from the filled Pt-3d orbitals to the 2p antibonding orbitals of OOH*, which effectively activates the O–O bond. This work provides insights to design efficient electrocatalysts for H₂O₂ generation.

1. Introduction

Hydrogen peroxide (H₂O₂) is an important green chemical with various applications in industrial processes such as paper bleaching, textiles, water treatment [1,2,3,4] and chemical synthesis [5]. More importantly, COVID-19, which is currently sweeping the world, has intensified the demand for H₂O₂ for medical purposes. Global production of H₂O₂ was close to 4.5 million tons in 2020, and the market demand is expected to reach approximately 5.7 million tons by 2027 [6]. However, current industrial H₂O₂ production relies mainly on the anthraquinone oxidation (AO) process, which is costly and highly polluting [7]. In addition, transporting, storing and handling high concentrations of H₂O₂ is dangerous and expensive. Hence, it has become urgent to develop a simple, low-cost and environmentally friendly method for H₂O₂ production. In recent years, electrocatalytic reduction of O₂ has been considered as a promising method for the production of H₂O₂ [8,9,10]. In this case, H₂O₂ can be produced on-site, which is both safe and convenient. The electrocatalytic O₂ reduction reaction (ORR) involves multiple electron steps, where O₂ can be reduced to H₂O₂ via the 2e⁻ pathway or to H₂O via the 4e⁻ route. Therefore, the main challenge in H₂O₂ production with the electrocatalytic ORR is how to control the selectivity of the reaction.

An increasing number of catalysts for the electrocatalytic synthesis of H₂O₂ have been reported recently, including noble metal [11,12] and their alloy catalysts [9,13,14], single-atom catalysts and carbon-based catalysts [15,16,17]. Among these, carbon-based catalysts are of great interest due to their abundant sources and easy tunability. In particular, graphene-based single-atom catalysts (SACs) have been designed and applied for the preparation of H₂O₂. For example, previous studies have reported that metal–nitrogen–carbon SACs can promote the selectivity of the electrocatalytic 2e⁻ ORR while maintaining high activity [18,19,20]. As a member of the carbon material family, graphdiyne (GDY) has recently been reported in various fields [21]. Importantly, single metal atoms can be stably anchored on the GDY surface [22,23] due to the in-plane conjugated network of alkyl and aryl groups and the inhomogeneous charge distribution of GDY. More importantly, the pristine GDY monolayer (Figure 1a) has an excellent electron mobility, reaching 10⁵ cm²/(Vs) at 300 K [24]. These advantages of GDY make it of great potential as a substrate for electrocatalysts.

In this work, we have investigated the activity of a series of single transition metal atoms (TM₁) supported on GDY (TM₁@GDY) for the 2e⁻ ORR using density functional theory (DFT) calculations under acidic conditions (pH = 0). Of all the TM₁@GDYs, calculations showed that Pt₁@GDY has the best activity and selectivity in the 2e⁻ ORR. Encouragingly, the predicted activity of Pt₁@GDY is comparable to that of known commercial catalysts. Electronic structure analysis further revealed the origin of the high activity of Pt₁@GDY towards the 2e⁻ ORR. This work provides insight into the design of efficient electrocatalysts for the production of H₂O₂.

2. Results and Discussion

2.1. Catalyst Structure and Stability

As the stability of a catalyst is a prerequisite for its application, we first assessed the thermal stability of TM₁@GDY by calculating the binding energy (E_b) of TM₁ on GDY. As shown in Figure 1b, TM₁ prefers to adsorb on corners near the six-membered carbon ring and is bonded to four carbon atoms. TM₁ involves all 3d, 4d and 5d transition metal elements except for radioactive technetium (Tc), lanthanides (La~Lu) and the liquid metal mercury (Hg), as shown in Figure 1c. The E_b of TM₁ was calculated as follows:

$${\mathrm{E}}_{\mathrm{b}}{=\mathrm{E}}_{{\mathrm{TM}}_1@\mathrm{GDY}}-{ \mathrm{E}}_{{\mathrm{TM}}_1 }-{ \mathrm{E}}_{\mathrm{GDY}}$$

(1)

where ${\mathrm{E}}_{{\mathrm{TM}}_1@\mathrm{GDY}}$ and E_GDY represent the total energy of TM₁@GDY and GDY, respectively, and${ \mathrm{E}}_{{\mathrm{TM}}_1 }$ represents the energy of the single transition metal atom. The E_b of all TM₁@GDY are shown in Figure 1d and Table S1 from Supplementary Materials, and the E_b < 0 indicated that the single metal atom can be stably supported on GDY. It is found that the E_b of single atoms of zinc (Zn), cadmium (Cd), argentum (Ag) and aurum (Au) are less than 0 eV but greater than −1 eV. Their optimized structures are shown in Figure S1 from Supplementary Materials. However, it is found that the Zn and Cd on GDY are not at the corners of the acetylene ring but at the center of the ring. They are far from the GDY plane at distances of 2.32 and 3.03 Å, respectively, indicating that Cd₁@GDY and Zn₁@GDY are unstable. The E_b of both Ag₁ and Au₁@GDY are very close to 0 eV, indicating that they are also unstable. Meanwhile, the E_b of Cu₁@GDY was calculated to be −2.00 eV. Notably, Li et al. synthesized Cu single atoms anchored to GDY and demonstrated that the supported Cu₁ is very stable [25]. Therefore, we consider the supported single atoms with E_b more negative than −2.00 eV to be stable by using Cu₁@GDY as a criterion. Accordingly, Zn, Cd, Ag and Au single atoms are relatively unstable because their E_b is more positive than −2.00 eV.

2.2. Catalyst Performance

The ORR performance of TM₁@GDY under acidic conditions (pH = 0) was investigated. The scheme of the reaction process is shown in Figure 2a. Firstly, the adsorbed O₂ on TM₁@GDY obtained one (electron (e⁻) and proton (H⁺)) pair from the aqueous solution, producing an OOH* intermediate. Followed by the further reduction of OOH* by (H⁺ + e⁻), this step can generate either H₂O₂ (2e⁻ ORR) or H₂O (4e⁻ ORR), and the product depends on if the O–O bond is broken. The steps of the ORR are as follows, the * in the following steps represents the adsorption site on the catalyst [26]:

The 2e⁻ ORR pathway:

$${\mathrm{O}}_2+*+{\mathrm{H}}^{+}+{\mathrm{e}}^{-}\to \mathrm{OOH}*$$

(2)

$$\mathrm{OOH}*+{\mathrm{H}}^{+}+{\mathrm{e}}^{-} \to { \mathrm{H}}_2{\mathrm{O}}_2+*$$

(3)

The complete reaction can be expressed as:

$${\mathrm{O}}_2+{2\mathrm{H}}^{+}+{2\mathrm{e}}^{-} \to { \mathrm{H}}_2{\mathrm{O}}_2 (0.7 \mathrm{V} \mathrm{vs}. \mathrm{SHE})$$

(4)

The 4e⁻ ORR pathway:

$${\mathrm{O}}_2+*+{\mathrm{H}}^{+}+{\mathrm{e}}^{-}\to \mathrm{OOH}*$$

(5)

$$\mathrm{OOH}*+{\mathrm{H}}^{+}+{\mathrm{e}}^{-}\to { \mathrm{H}}_2\mathrm{O}+\mathrm{O}*$$

(6)

$$\mathrm{O}*+{\mathrm{H}}^{+}+{\mathrm{e}}^{-} \to \mathrm{OH}*$$

(7)

$$\mathrm{OH}*+{\mathrm{H}}^{+}+{\mathrm{e}}^{-}\to { \mathrm{H}}_2\mathrm{O}+*$$

(8)

The complete reaction can be expressed as:

$${\mathrm{O}}_2+{4\mathrm{H}}^{+}+4{\mathrm{e}}^{-} \to { 2\mathrm{H}}_2\mathrm{O} (1.23 \mathrm{V} \mathrm{vs}. \mathrm{SHE})$$

(9)

From the above ORR reaction pathway, it is clear that the 4e⁻ ORR and 2e⁻ ORR are two competing reactions during the electrochemical synthesis of H₂O₂, which depends on the type of O₂ adsorption [26]. The adsorption of O₂ on the catalyst surface can be generally classified into two types: end-on and side-on modes [27]. Obviously, in both the 4e⁻ and 2e⁻ pathways, the intermediate OOH* is formed in the first step of oxygen adsorption and hydrogenation, so we first calculated the free energy of OOH* (G_OOH*) (Table S2 from Supplementary Materials). As shown in Equations (5) and (6), the ΔG of each electron step in the 2e⁻ ORR under ideal conditions is 0.70 eV (G_OOH* = 4.22 eV), so the change in free energy of the whole reaction is 1.40 eV (${\mathrm{G}}_{{\mathrm{H}}_2{\mathrm{O}}_2}$= 3.52 eV).

In general, the adsorption conformations of OOH* on these TM₁@GDYs can be divided into four types, as shown in Figure 2b–e. The first one (TM = Re, W, Nb, Ta and Mo) is shown in Figure 2b, where the O–O bond of OOH* undergoes spontaneous cleavage due to strong interactions between OOH* and TM₁@GDY, indicating the poor selectivity of these TM₁@GDYs (TM = Re, W, Nb, Ta and Mo). For the second one shown in Figure 2c (TM = Sc, Y and Hf), the G_OOH* of OOH* are all less than 3.52 eV (Table S1 from Supplementary Materials), which indicates that Sc₁, Y₁ and Hf₁@GDY have strong adsorption of OOH. In addition, after OOH adsorption, a facile migration of the single atom to the C–C triple bond site (located between C2 and C2′) can be observed, indicating that the single atom is unstable during the catalysis. In the third one (TM = Ti, V, Mn, Zr, Rh and Ir), OOH* is in the side-on mode (Figure 2d); it can be seen that the G_OOH* of the laterally adsorbed OOH* are all less than 3.52 eV (Table S1), which means that ΔG_OOH* are greater than 1.40 eV, which is not favorable for the 2e⁻ ORR. For the last one (TM = Pt, Pd, Cu, Co, Ni, Ru, Os, Cr and Fe, Figure 2e), OOH* is following the end-on pattern. The G_OOH* are all greater than 3.52 eV, which means that these TM₁@GDYs have strong adsorption of OOH, and is, therefore, favorable for the 2e⁻ ORR pathway. Therefore, the structures in Figure 2e were used for the subsequent studies.

Next, we compared the electrocatalytic ORR performance of TM₁@GDYs (TM = Pt, Pd, Cu, Co, Ni, Ru, Os, Cr and Fe). The free energy curves, as well as the corresponding η and potential limiting steps, are given in Figure 3 and Figure S2 from Supplementary Materials to further clarify the catalytic activity and selectivity. We can see that the G_OOH* on Cr₁@GDY, Fe₁@GDY, Co₁@GDY, Cu₁@GDY, Ru₁@GDY and Os₁@GDY are 2.53 eV, 3.15 eV, 3.44 eV, 3.50 eV, 2.41 eV and 2.55 eV, respectively (Figure S2 from Supplementary Materials and Figure 3a,c), which are all lower than 3.52 eV, implying that they all have a strong adsorption for OOH*. Therefore, this strong adsorption of OOH* leads to the second elementary step of the 2e⁻ ORR being uphill and becoming PDS. In contrast, the G_OOH* of the other three catalysts (Ni₁@GDY, Pd₁@GDY and Pt₁@GDY) are 3.76 eV, 4.41 eV and 4.07 eV, respectively (Figure 3b,d–f), which are all above 3.52 eV. In particular, the G_OOH* of Pd₁@GDY and Pt₁@GDY (4.41 eV and 4.07 eV) are very close to 4.22 eV, which indicates that the adsorption strength of OOH is moderate on both catalysts with the first and second elementary step as PDS, respectively. In contrast, the G_OOH* of Ni₁@GDY is lower than 4.22 eV, which indicates a stronger adsorption of OOH*, leading to the second step as PDS and thus unfavorable to the 2e⁻ ORR pathway.

In addition, we can see the 4e⁻ ORR pathway on each TM₁@GDY in Figure 3 and Figure S2 from Supplementary Materials. The PDSs of Cr₁@GDY, Fe₁@GDY, Ru₁@GDY and Os₁@GDY are the fourth step (Figure S2 from Supplementary Materials), which are all uphill in energy with ΔG equal to 0.90 eV, 0.23 eV, 0.72 eV and 0.94 eV, respectively. This suggests that the 4e⁻ ORR process cannot be spontaneous. As can be seen from Figure 3a–c, the PDS of Co₁@GDY, Ni₁@GDY and Cu₁@GDY is the last reaction step with ΔG of −0.63 eV, −0.77 eV and −0.25 eV, respectively, while the PDS of the 4e⁻ ORR on Pd₁@GDY and Pt₁@GDY (Figure 3d–e) is the first reaction step with ΔG of −0.51 eV and −0.85 eV, respectively.

For the 2e⁻ ORR pathway, it is well known that a good catalyst for the electrochemical generation of H₂O₂ should have a U_L close to U_equilibrium. Therefore, U_L can be used as a descriptor to evaluate its 2e⁻ ORR activity. To better understand the ORR ability of catalysts, we calculated the U_L of the 2e⁻ and 4e⁻ ORR for these nine catalysts (Cr₁@GDY, Fe₁@GDY, Co₁@GDY, Ni₁@GDY, Cu₁@GDY, Ru₁@GDY, Pd₁@GDY, Os₁@GDY and Pt₁@GDY) by using the U_L formula and plotted Figure 3f and Figure S3 from Supplementary Materials with the red dashed lines representing the standard U_equilibrium, which are 0.7 V and 1.23 V, respectively. Theoretically, the closer the U_L is to U_equilibrium, the closer the G_OOH* is to 4.22 eV, implying a higher 2e⁻ ORR activity of the catalyst. The reason is that G_OOH* is lower than 3.52 eV when U_L < 0, i.e., the adsorption of OOH is too strong. In addition, η can be used more intuitively to evaluate the activity because η = U_equilibrium − U_L. It can be seen from Figure 3f that the catalysts with U_L > 0 are favorable catalysts for the 2e⁻ ORR. The results show that the U_L values of Ni₁@GDY, Pd₁@GDY and Pt₁@GDY are 0.24 V, 0.51V, and 0.55 V, corresponding to η of 0.46 V, 0.19 V, and 0.15 V, respectively. As a result, Pt₁@GDY and Pd₁@GDY have the smallest and second smallest η, which are comparable to those on Au(100) and Au(111) [28] and PtHg4 electrocatalysts [29], indicating that they have good 2e⁻ ORR activity.

In the ORR process, selectivity is a key indicator to evaluate the catalytic performance of the catalyst for H₂O₂ synthesis. Figure 4 shows the Gibbs free energy diagrams of the 2e⁻ and 4e⁻ ORRs on Pt₁@GDY. At U = 0 V (black curve), the reaction steps in both the 2e⁻ and 4e⁻ ORR proceed spontaneously, as their ΔG are negative. The results also show that ΔG_PDS is 0.55 eV for the 2e⁻ ORR and 0.85 eV for the 4e⁻ ORR. Thus, for the 2e⁻ ORR and 4e⁻ ORR, the free energy of the PDS changes to zero when the applied potential is equal to 0.55 V (blue curve) and 0.85 V (red curve), respectively, while the other steps remain downhill in energy. The PDS steps of both the 2e⁻ and 4e⁻ ORR are uphill when the applied potential is equal to U_equilibrium. The results also show that η is 0.15 V and 0.38 V for the 2e⁻ and 4e⁻ ORR, respectively, indicating that the 2e⁻ ORR exhibits higher selectivity over the 4e⁻ ORR on Pt₁@GDY. In conclusion, it is predicted that Pt₁@GDY is an ideal catalyst for the catalytic synthesis of H₂O₂.

2.3. Origin of the 2e⁻ ORR Activity

Essentially, catalytic properties, such as activity, stability and selectivity, are often determined by the electronic structure of the catalyst [30]. Previously, we discussed in detail the effect of catalyst structure on the activity and selectivity of the 2e⁻ ORR. In order to better design catalysts for the 2e⁻ ORR, we further discussed the origin of catalytic activity using electronic structure analysis. First, we calculated the Bader charge and charge differential density (CDD). As shown in Figure 5a, the CDD results show that there is significant electron transfer between Pt₁ and C atoms of Pt₁@GDY. From the Bader charge analysis, we find that about 0.32 e⁻ are transferred from Pt₁ to the C atom of GDY. This indicates that the interaction between Pt₁ and the C atoms allows Pt₁ to be stabilized on the GDY support. In addition, the CDD plot (Figure 5b) shows a significant charge transfer between Pt1 and OOH when OOH species are adsorbed on Pt₁@GDY. This leads to an elongation of the O–O bond length of OOH from 1.21 Å to 1.45 Å (Figure S4 from Supplementary Materials). Moreover, the Bader charge analysis further confirms that there is about 0.39 e⁻ transferred from Pt₁ to the adsorbed OOH. This implies that the Pt single atom effectively activates the O–O bond, which facilitates the subsequent hydrogenation of OOH* to generate the final H₂O₂ product.

Finally, we determined the activation mechanism of the O₂ molecule from the perspective of molecular orbital theory. First, as shown in Figure 5c, the obvious orbital overlap between Pt₁ 3d and C 2p indicates a strong interaction between the anchored Pt1 and the surrounding C atoms, which further demonstrates the stability of Pt₁@GDY. Second, we investigated spin-polarized PDOSs to elucidate the interactions between Pt₁ and O atoms in O₂* and OOH* species. It is seen from Figure 5c that the 2p antibonding orbital π* and the bonding orbital π of the free O₂ are near the Fermi level, which indicates that they are the most active orbitals of O₂. When the free O₂ is adsorbed onto Pt₁@GDY, the π* antibonding orbitals of O₂ orbitals shift to the lower energy region and hybridize with the d orbitals of Pt₁. On the other hand, the empty π* of O₂* is partially occupied, which indicates that the d electrons of Pt₁ are partially transferred to the empty π* of O₂, allowing O₂ to be activated. As shown in the PDOS diagram for OOH*, the 3d orbital of Pt₁ couples simultaneously with the bonding and antibonding orbitals of 2p of O, which leads to a moderate activation of the O–O bond in OOH. Thus, the electronic structure analysis suggests that the ORR process on Pt₁@GDY tends to follow the 2e⁻ pathway.

3. Materials and Methods

Computational Details

All spin-polarized DFT calculations were performed using the Vienna Ab-initio Simulation Packages (VASP.5.4.4) [31], and the electron–ion interaction was described with the projector-augmented wave (PAW) method [32,33]. The exchange-correlation potential was treated by the Perdew–Burke–Ernzerhof (PBE) version of the generalized gradient approximation (GGA) [34]. The valence electrons were expanded in a plane-wave basis set with an energy cutoff of 450 eV. The optimized monolayer GDY has a lattice constant of 18.74 Å (Figure 1a). The 2 × 2 supercells of GDY were adopted for the subsequent calculations. A 15 Å vacuum space was adopted along the z-direction to avoid interactions between periodic slabs. K-points mesh was set to 3 × 3 × 1 for Brillouin zone sampling in structural optimization and electronic structure calculations. The DFT-D3 method with zero-damping [35] was introduced to describe van der Waals weakly dispersive interactions (vdW). All the structures were relaxed until the forces on each ion were less than 0.05 eV/Å, and the convergence criteria for the energy was set as 10⁻⁴ eV.

Based on Nørskov’s computational hydrogen electrode model [36,37], we calculated the Gibbs free energy change (ΔG) for each elementary step as follows:

$$\Delta \mathrm{G}=\Delta \mathrm{E}+\Delta {\mathrm{E}}_{\mathrm{zpe}}- \mathrm{T}\Delta \mathrm{S}+\mathrm{eU}$$

(10)

where ΔE refers to the energy difference directly calculated with DFT before and after each elementary reaction step; ΔE_zpe and TΔS are the differences in the zero-point energies and entropies at 298.15 K, respectively; and the value of eU was determined using the applied potential (U).

In addition, we defined a descriptor of limiting potential (U_L) to describe the activity of the electrocatalytic ORR with the free energy change of the potential-determining step (PDS):

$${\mathrm{U}}_{\mathrm{L}}=-\Delta {\mathrm{G}}_{\mathrm{PDS}}/\mathrm{e}$$

(11)

where the ΔG_PDS is the Gibbs free energy change of PDS.

The overpotential (η) was calculated as:

$${\mathsf{\eta }=\mathrm{U}}_{\mathrm{equilibrium}}-{ \mathrm{U}}_{\mathrm{L}}$$

(12)

where the equilibrium potentials (U_equilibrium) are 1.23 V and 0.70 V for the 4e⁻ ORR and 2e⁻ ORR, respectively.

4. Conclusions

In summary: the 2e⁻ ORR catalyzed by the GDY-supported single transition metal atom catalysts in an acidic environment was investigated using first-principles DFT calculations. The results show that Pt₁@GDY has good stability and the best activity for the 2e⁻ ORR, as it exhibits the lowest thermodynamic η (0.15 V) among the 23 catalysts studied. In addition, the electronic structure analysis of the interactions between O₂*, OOH* and Pt₁@GDY provides an in-depth understanding of the 2e⁻ ORR pathway to produce H₂O₂. The high activity of Pt₁@GDY is attributed to the electron transfer between the 3d orbital of Pt₁ and the antibonding orbital π* of the 2p orbital of O in OOH*, allowing the efficient activation of the O–O bond. This work provides theoretical insights into the design of efficient electrocatalysts for the generation of H₂O₂.