2024 Volume 37 Issue 5
Article Contents

Chongchong Li, Li Sheng, Yanan Zhou, Wenhua Zhang. Enhanced Oxygen Evolution Reaction Performance by Dynamic Adsorption of Intermediates on C2N-Supported Single Atom Catalysts[J]. Chinese Journal of Chemical Physics, 2024, 37(5): 662-670. doi: 10.1063/1674-0068/cjcp2305048
Citation: Chongchong Li, Li Sheng, Yanan Zhou, Wenhua Zhang. Enhanced Oxygen Evolution Reaction Performance by Dynamic Adsorption of Intermediates on C2N-Supported Single Atom Catalysts[J]. Chinese Journal of Chemical Physics, 2024, 37(5): 662-670. doi: 10.1063/1674-0068/cjcp2305048

Enhanced Oxygen Evolution Reaction Performance by Dynamic Adsorption of Intermediates on C2N-Supported Single Atom Catalysts

通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Figures(6)

Article Metrics

Article views(263) PDF downloads(0) Cited by(0)

Access History

Enhanced Oxygen Evolution Reaction Performance by Dynamic Adsorption of Intermediates on C2N-Supported Single Atom Catalysts

Abstract: The dynamic adsorption of possible intermediates on single-atom catalysts (SACs) under working condition plays a key role in the electrocatalytic performance by the oxygen evolution reaction (OER), and therefore the performance of the dynamic adsorption should be fully considered in the theoretical screening of potential SACs. Based on density functional theory calculations, the OER performance of 27 types of C2N-supported single transition metal atoms (TM@C2N) is systematically investigated without and with considering the dynamic adsorption of possible intermediates. Without considering dynamic adsorption, only Rh@C2N and Ni@C2N are screened out as good catalysts. However, by further considering the dynamic adsorption configurations of possible intermediates, more promising TM@C2N SACs including Fe(Co, Ni, Ru, Rh, Ir)@C2N toward the OER are screened out. The presence of the intermediates (*HO, *O) on SACs could shift their d band center toward lower energy level, which makes the interaction between the adsorbate and SACs moderate and thus enhances their OER performance. The present work is instructive for further screening and designing of efficient single-atom catalysts for the oxygen evolution reaction.

    • Since the 21st century, the environmental and energy crisis has become increasingly serious. And it is expected that the development of new and clean energy sources can solve the environmental and energy problems [1, 2]. Hydrogen is one of the most promising energy carriers for the high energy density and clean combustion product water [3, 4]. The production of hydrogen by electrolyzing water using the electricity generated by renewable energies such as solar energy, wind energy, etc., is considered as the most efficient and sustainable strategy [5, 6]. However, the wide application of electrolyzing water is blocked by the slow kinetics of the oxygen evolution reaction (OER) at the anode [7, 8]. Traditionally, Ir or Ru based materials are the most active OER catalysts [9, 10]. However, due to their high cost and low abundance, they are not widely available. Therefore, there is an urgent need to find some alternative, efficient and sustainable OER catalysts.

      Different strategies have been proposed to design efficient OER catalysts, mainly by reducing the content of precious metals or developing non-precious alternatives [11-14]. With the development of synthetic technologies, the large-scale preparation of single-atom catalysts (SACs) provide a bright way to reduce the utilizing amount of noble metals and even improve the activity of non-precious metal atoms as OER catalysts [15-17]. Two dimensional (2D) materials such as graphene, boron nitride, mexene, covalent-organic framework (COF), metal-organic framework (MOF) and so on [18-21], are widely used as the support of single metal atoms for their large surface areas. Recently, a lot of theoretical works have been performed to design efficient SACs supported by 2D materials for OER by evaluating the stability of SACs and comparing the calculated overpotential of OER (ηOER) on SACs. For example, Jiang et al. [15] reported that a quadruple-coordinated Co is a promising candidate for the OER with a low computed ηOER of 0.39 V, which is comparable to those of noble metal catalysts. Huang et al. [22] reported the Ir/pyrrolic-N4-G can perform as a promising electrocatalyst for OER with ηOER of 0.32 V. Zhou et al. [23] also theoretically predicted that RhN1O3−HTP is a promising catalyst for OER with ηOER of 0.28 V. Clearly, a number of different supports can be used to construct efficient SAC catalysts, and the selection of suitable supports can play a crucial role in screening catalysts for high OER catalytic activity.

      In 2015, a novel two-dimensional layered material, C2N two-dimensional crystal with uniformly distributed holes, was successfully prepared [24]. Due to its natural pore-like structure, C2N can be used as a support for metal clusters or even SACs, which has been used as substrates of catalyst for water splitting [25]. Ying et al. [26] theoretically predicted Rh@C2N to be a potential OER SAC with ηOER of 0.37 V. It is noticed that during the theoretical design of OER SACs, only bare SACs are considered to bind one *O or *OH or *OOH. However, it has been suggested that the dynamic adsorption of intermediates such as *OH, *O, *OOH or even *O2 may play an important role in the OER performance [27]. Cao et al. [28] suggested that under working condition the dynamic adsorption of *O on atomically dispersed Ru1-N4 on g-C3N4 greatly improved the OER activity. As suggested by Ying et al. [26], some of the TM@C2N *OOH is ready to dissociate to co-adsorbed *OH and *O rather than intact *OOH, indicating the tendency of regulation of the active sites during OER.

      In this work, we design a series of transition metal (TM= Sc–Zn, Y–Cd, La, Ta, W, Os, Ir, Pt, and Au) atoms anchored in C2N monolayers of pores and explore their stability as well as their catalytic activity towards OER. Considering the bare TM@C2N, Rh@C2N and Ni@C2N are screened out as potential catalysts for OER. Based on the construction of surface phase diagrams, the dynamic structures of active sites are determined and the OER performance is re-evaluated. With the modulating effect of pre-adsorbed intermediates, Fe(Co, Ni, Ru, Rh, Ir) are screened out as potential electrocatalysts for OER. Analysis from the d-band center level indicates that the adsorption of intermediates weakens the adsorption of other intermediates, and the multiple intermediate adsorptions can place it at the optimal value.

    II.   COMPUTATIONAL DETAILS
    • All the spin-polarized density functional theory (DFT) calculations are performed using the Vienna ab initio simulation package (VASP) [29], employing the projector augmented wave (PAW) [30] pseudopotential to describe the interaction between nuclei and electrons, and the Perdew-Burke-Ernzerhof (PBE) [31] functional to describe exchange-correlation effects. The DFT-D3 [32] method is used to consider weak interaction corrections. We employ a (2×2×1) C2N supercell to simulate the C2N two-dimensional material with a vacuum layer of 15 Å. A 3×3×1 k-point grid is used to sample the reciprocal space for all the calculations. The solvent effect is considered via implicit solvent model with a dielectric constant of 80 [33].

      The reaction free energies (ΔG) of each elementary reaction are calculated using the computational hydrogen electrode (CHE) method developed by Nørskov et al. [34] as:

      where, ΔE is the energy difference directly obtained from DFT calculations, ΔZPE is the change of zero-point energy, T represents the temperature (T = 298.15 K), ΔS is the change in entropy, and ΔGU is the free energy contribution related to the electrode potential U. ΔGpH is the correction of free energy of H+ with respect to concentration, which can be determined as ΔGpH = 2.303 × kBT × pH (or 0.059 × pH), where kB is the Boltzmann constant.

    III.   RESULTS AND DISCUSSION
    • The most stable geometric configuration of two-dimensional material C2N is shown in FIG. 1(a), with its large pores and high specific surface area that can embed single atoms. The adsorptions of 27 types of single transition metal (TM) atoms on g-C2N (TM@C2N) are carefully investigated. After geometric optimization, the adsorption configurations of 27 TM@C2N can be divided into three classes: (i) TM (TM=Ti–Zn, Zr–Pd, W, Os–Pt) atom binding with two N atoms (FIG. 1(b)); (ii) Sc (Ta) atom binding with three N atoms (FIG. 1(c)), and (iii) Y (Ag, Cd, La, Au) atom located at the center of the pore (FIG. 1(d)).

      The binding energies Eb defined as Eb = $E_{\mathrm{TM} @ \mathrm{C}_{2} \mathrm{N}} $–ETM–$E_{{\rm{C}}_{2}{\rm{N}}}$, where $E_{\mathrm{TM} @ \mathrm{C}_{2} \mathrm{N}}$, ETM, and $E_{{\rm{C}}_{2}{\rm{N}}} $ represent the calculated energies of TM@C2N, isolated single TM atom, and C2N monolayer, respectively. By calculating Eb, we can investigate the thermodynamic stability of TM@C2N by comparing the binding ability of single TM atoms on C2N, if Eb < 0, the binding of single TM atoms on C2N is stable. The results listed in Table S2 (Supplementary materials, SM) show that the 27 TM@C2N SACs thermodynamic stability of TM@C2N can be confirmed.

      The OER proceeds via four-electron transfer steps H2O(l)→*OH→*O→*OOH→O2(g). The relative free energies of each intermediate ΔG*O, ΔG*OH, ΔG*OOH are calculated relative to H2O(l) as listed in Table SIII (SM), while $\Delta G_{^*{\rm{O}}_2}$ is set as 4.92 eV. In order to obtain the OER catalytic performance of TM@C2N, the reaction free energy of each elementary step (S2a–S2d) (ΔG1, ΔG2, ΔG3, ΔG4) is calculated and the overpotential is calculated as ηOER=max[ΔG1, ΔG2, ΔG3, ΔG4]/e –1.23 V as listed in Table SIV (SM) with the potential-limiting step on each bare SACs. On Sc(V, Cr, Mn, Co, Ni, Tc, Ru, Rh, Os)@C2N, the potential-limiting step (PLS) is the evolution of *O to *OOH; while on Cu(Zn, Y, Pd, Ag, Cd, La, Au)@C2N, the potential-limiting step is formation of *O from *HO; and the formation of gas phase oxygen is the potential-limiting step on Fe(Ir, Pt)@C2N. As listed in Table SIV (SM), Rh@C2N and Ni@C2N are expected to be potential electrocatalysts for OER with overpotential of 0.43 V and 0.49 V, respectively, and Gibbs free energy diagram for OER is shown in FIG. S1 (SM). The calculated results show that Rh@C2N exhibits the best OER performance among all the TM@C2N, which is in consistence with the results of previous work [26].

      As listed in Table SIV (SM), on all of the SACs except Ag@C2N, ΔG*OH is less than 1.23 eV or even negative, indicating a strong interaction between *HO and SACs. Furthermore, it should be noted that on Ti(Zr, Mo, Ta, Nb, W)@C2N, during geometric optimization *OOH is ready to break as co-adsorbed *O and *HO as shown in FIG. S2 (SM), which indicates that multi-adsorption of possible intermediates should be further considered.

    • The Gibbs free energy of different adsorbed species including *OH, *O, *OOH, and various combinations at different U are calculated to construct surface phase diagram of each TM@C2N with an expected working potential range of 1.23–1.73 V (ηOER<0.5 V), allowing to understand the stable intermediate adsorption structures at different U values and determine the most stable configurations of TM@C2N under working condition. It is known Rh, Ru, Pd, Os, Ir, Pt are traditional noble-metal used as OER catalysts and Fe, Co, Ni are the most potential non-noble metal to regulate as OER catalysts [27]. Thus, in the following work, we focus on the investigation of Fe(Co, Ni, Rh, Ru, Pd, Os, Ir, Pt)@C2N and the free energy diagrams of each possible dynamic adsorption configuration relative to the applied potential are shown in FIG. 2. And the results for other TM@C2N are shown in FIG. S3–S5.

      As shown in FIG 2(a,b), with the increasing of applied U, the surface species on Fe(Co)@C2N evolve via *OH→2*OH→3*OH→2*OH+*O→2*OH +*OOH. It is noticed that the free energy of *O is higher than 2*OH and 3*OH between 0.0 V and 3.0 V, which indicates that when *OH is present on Fe(Co)@C2N, 2*OH and 3*OH are more free energy favorable to be formed rather than the evolution from *OH to *O. Thus, the simple consideration of OER via H2O(l)→*OH→*O →*OOH→O2(g) cannot give an exact description of the OER activity and further investigation with real active site should be considered. As for Ni@C2N, the surface species change from *OH→2*OH →*OH + *O→*OH + *OOH (FIG. 2(c)). Thus, the dynamic active site on Fe(Co)@C2N can be considered as 2*OH-Fe(Co)@C2N while on Ni@C2N the active site under working condition is expected as *OH-Ni@C2N. On Ru(Rh, Ir)@C2N (FIG. 2(d, e, h)), the surface species changes as 2*OH →*O+*OH→2*O→2*O+ *OH →3*O→2*O+*OOH with the increasing of applied bias up to 1.73 V, which indicates that under working condition the active site is dynamically evolved as 2*O-Ru(Rh, Ir)@C2N. As for Os@C2N, similar to the case of Ru@C2N, but as shown in FIG. 2(g), the surface species on Os@C2N evolve from 2*O→2*O + *OH at 1.57 V, where 3*O is ready to be formed. Due to the sequential nature of species formation in the evolutionary pathway, the 2*O adsorption structure persists until 1.57 V, followed by the formation of 3*O and 2*O+*OOH. As shown in FIG. 2(f), bare Pd@C2N is the active site and the formation of *O requires more positive potential than the formation of *OOH. On Pt@C2N as shown in FIG. 2(i), the surface species vary via *OH→*O→*O+*OH→*OH+*OOH, which indicates a synergetic effect of two oxygen containing species and the active site can be considered as *OH-Pt@C2N.

    • The Gibbs free energy changes (ΔG1, ΔG2, ΔG3, ΔG4) of each step of the OER process on a series of TM@C2N with and without the dynamic adsorption of intermediates are summarized in Table SVIII (SM) and the potential energy diagrams are shown in FIG. 3. On Fe(Co)@C2N, with the presence of pre-adsorption 2*OH, as shown in FIG. S6(a) (SM), the overpotential decreases from 0.57 V (0.79 V) to 0.39 V (0.50 V) as shown in FIG. 3 (a) and (b). On Fe@C2N, the PLS is the formation of O2 from *OOH, while the formation of *OOH is the PLS on Co@C2N. On Ni@C2N, the dynamic adsorption of *OH under working condition shifts the PLS from the formation of *OOH to the formation of *O (FIG. S6(b)) and the overpotential is lowered from 0.49 V to 0.39 V as shown in FIG. 3(c). On Ru@C2N, the PLS is the formation of *OOH with the presence of two dynamic *O, the overpotential decreases from 1.06 V to 0.38 V as shown in FIG. 3(d). On Rh@C2N, the PLS shifts from the formation of O2 to the formation of *OOH with the presence of two extra *O and the overpotential is lowered from 0.43 V to 0.33 V as shown in FIG. 3(e) and FIG. S6(c) (SM). As shown in FIG. 3(f), OER prefers to proceed on bare Pd@C2N without dynamic adsorption of intermediates with overpotential of 0.54 V to form *O. On Os@C2N, the formation of *OOH is the PLS and the required overpotential decreases from 1.29 V to 0.86 V as shown in FIG. 3(g). While on Ir@C2N, the PLS shifts from the formation of O2 to the formation of *OOH with overpotential lowering from 0.85 V to 0.42 V as shown in FIG. 3(h). On bare Pt@C2N, the formation of O2 is the PLS with overpotential of 0.60 V. While, the real OER pathway on Pt@C2N is interesting, which proceeds via *OH→*O→*O+*OH→*OH+*OOH, and the overpotential is lowered to 0.45 V as shown in FIG. 3(i) and FIG. S6(d) (SM).

      Thus the overpotentials on Fe(Co, Ni, Ru, Rh, Ir, Pt)@C2N are all lower than 0.5 V, which indicates that if the dynamic active sites are considered, more potential electrocatalysts for OER can be screened out. What should be mentioned is that Fe(Co, Ni)@C2N are non-noble metal based SACs for OER.

      To clearly illustrate the variation of the limiting potential of Fe(Co, Ni, Ru, Rh, Pd, Os, Ir, Pt)@C2N catalysts with and without dynamic adsorption of intermediates, the corresponding overpotential of the OER are summarized as shown in FIG. 4. The OER performance of Fe(Co, Ni, Ru, Rh, Os, Ir, Pt)@C2N is improved under the regulation of pre-adsorption. However, bare Pd@C2N achieves lower overpotential than dynamic adsorption of intermediates, which agrees well with the free energy diagrams shown in FIG. 2(e). Additionally, to evaluate the stability of SACs under the dynamic adsorption of potential intermediates, ab initio molecular dynamics (AIMD) simulations were performed to evaluate the stability of the 2OH-Fe@C2N, 2OH-Co@C2N, OH-Ni@C2N, 2O-Ru@C2N, 2O-Rh@C2N, 2O-Ir@C2N, OH-Pt@C2N. All the AIMD simulations are performed using an NVT ensemble at 300 K with time step of 1 fs and each simulation lasts 10 ps as shown in FIG. S8 (SM). The snapshots of the last step of each simulation are shown in FIG. S9 (SM), which indicates the stability of dynamic intermediates liganded Fe(Co, Ni, Rh, Ru, Ir, Pt)@C2N.

      The d orbital of TM atom and p-orbital of accumulated *O or *OH on Fe(Co, Ni, Ru, Rh, Pd Os, Ir, Pt)@C2N are shown in FIG. 5 and FIG. S7 (SM). The d-orbital centers of Rh and Ru shift far away from Fermi level due to strong coupling between TM atom and adsorbed *O or *OH. Generally, the lower energy levels of d-band centers weaken the interaction between metal sites and adsorbate species. As for OER, the presence of *OH or *O weakens the adsorption of *OH, *O, and *OOH intermediates and apparently makes ΔG*OH more positive, which may shift the PLS to earlier elementary steps. For example, on bare Rh@C2N, the ΔG*OH value is 0.40 eV; and when one *O and two *O species are adsorbed, the ΔG*OH values are shifted to 0.54 eV and 1.53 eV, respectively. Furthermore, the ηOER on Rh@C2N gradually decreases from 0.43 V to 0.37 V to 0.33 V with the increasing of *O numbers as listed in Table SVIII (SM). On Fe@C2N, the ΔG*OH value shifts from −0.09 V to 0.04 V to 0.56 V without and with one and two *OH, respectively. The ηOER on Fe@C2N is gradually lowed from 0.57 V to 0.49 V to 0.39 V with the presence of 0, 1, 2 *OH as listed in Table SVIII (SM).

    • In previous studies [15, 23, 25], the general trends and relationships between the adsorption free energies of species have been used for catalyst design or investigation, and scaling relationships between the adsorption free energies of *O and *OOH and *OH have been found in the context of considering different intermediate pre-adsorption for OER. As shown in FIG. 6(a, b), with the presence of dynamic adsorption of possible intermediates on each TM@C2N, there is a good linear relationship between ΔG*OOH, ΔG*O, and ΔG*OH, with ΔG*OOH = 0.83ΔG*OH + 3.23, R2 = 0.94 and ΔG*O = 1.36ΔG*OH + 0.90, R2 = 0.87, which indicates that the linear relationship of adsorption free energies still applies under dynamic adsorption conditions. Moreover, with ΔG*OH and ΔG*O−ΔG*OH as independent variables, the ηOER value is plotted on the colored contour map in FIG. 6(c), with the blue region in the two-dimensional contour map volcano plot being the area of highest OER activity. Furthermore, as shown in FIG. 6(d), the limiting potential is usually determined by the formation of *OH or *OOH. For OER, as shown in FIG 6(d), it is expected that when ΔG*OH is 1.52 eV, the OER overpotential can reach its minimum of 0.29 V.

    IV.   CONCLUSION
    • In summary, by employing first-principles calculations combined with multiple adsorption methods, we systematically investigate a series of transition metal atoms doped on C2N as the OER active site. Dynamic active site configurations on each TM@C2N under the expected working potential are determined by surface phase diagrams calculations, and potential electrocatalysts for OER are screened out using Gibbs free energy calculations. Without considering dynamic evolution of SACs under working condition, only Rh@C2N and Ni@C2N are selected as potential OER SACs. While, with the formation of dynamic active site configuration, more TM@C2N SACs with excellent OER performance are screened out including Fe(Co, Ni, Ru, Rh, Ir, Pt)@C2N. The linear relationships between ΔG*OH and ΔG*O, ΔG*OOH still exist with the presence of dynamic *OH or *O; and it is expected the highest OER performance can be achieved when ΔG*OH shifts towards the optimal value. Considering the pre-adsorption of dynamic intermediates, this work provides new insights for the accurate and rational design of single-atom electrocatalysts.

      Supplementary materials: The binding energy, the DFT-calculated adsorption energies, free energy change, and limiting potentials, comparison of the stability of TM@C2N pre-adsorption (Tables SI−SVIII) are shown. The Gibbs free energy diagrams for OER on bare Rh(Ni)@C2N, optimized structures of *OOH adsorbed on Rh(Mo)@C2N, 18 SACs potential energy diagram of OER, OER pathway diagrams of TM@C2N, PDOS, total energy and temperature variations during AIMD simulations, and the snapshots of the last steps of each AIMD simulation (FIG. S1–S9) are available.

    Figure (6)  Reference (34)

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return