Abstract
In this post, the transition state search for a elementary reaction in carbon dioxide reduction on Cu(111) surface is performed using CASTEP, and the activation barrier for this reaction is calculated.
Introduction
Electrochemical reduction of CO_2 to valuable chemicals and fuels such as hydrocarbon and alcohols is of great interest. However, experimental results show that most metal electrodes except for copper could only reduce CO_2 mainly to CO. Previous computational work[1][2] showed that the transition state energy of reduction of CO is a good descriptor that could help us to understand the trends of catalytic activity of transition metal surfaces. To understand the process that adsorbed CO reduced to COH on Cu(111) surface, the transition state geometry and energy of *CO reduction to *COH on Cu(111) surface will be calculated and compared to reported results, here * signs are used to show the species are absorbed on the surface.
Method
In this work, plane-wave based DFT code CASTEP is used. The exchange-correlation functional is described by a GGA-PBE functional. As for pseudopotential, “on the fly” generated ultrasoft pseudopotential for Cu is used with a core radius of 2.2 Bohr(1.16Å) and was generated with a 3d10 4s1 valence electronic configuration . The SCF tolerance is set to be 2e-6 eV per atom for all the optimization and single point energy calculations.
A 2×2 four-layer slab model is used to simulate Cu(111) surface structure with 10Å of vacuum. The optimized bulk with lattice parameter of 3.63Å is used to build the surface model. When the optimization and transition state search are performed, bottom two layers are kept fixed and remaining atoms are relaxed to simulate bulk behaviors. A optimized 5x5x1 k points grid and energy cutoff of 500 eV are kept consistent for all calculations.
Results
k points grid and energy cutoff optimization
With an energy cutoff of 500 eV, single point energies of 2×2 slab models are calculated. In figure 1, the energies of different k points gird relative to 8x8x1 k points grid is shown.
The energy cutoff convergence optimization is carried out with a k point grid of 8x8x1. In figure 2 energies of different energy cutoff relative to 550 eV are shown.
According to those two convergence test results, a 5x5x1 k points grid with an energy cutoff of 500 eV is reasonable for calculation of a 2×2 slab model of Cu(111) surface. Thus these settings are kept consistent for subsequent optimization and search for transition state.
Cu(111) surface optimization
In order to build a reasonable model of reaction happens on Cu(111) surface, the optimization of surface structure is carried out first. A 2×2 Cu(111) surface slab model is used with bottom 2 layers fixed to represent bulk effects on metal surface. BFGS algorithm is used to perform the optimization with energy convergence of 2e-5 ev/atom and force convergence of 0.05 eV/A.
Transition state search
According to previous work[2], co-adsorption of CO and H on Cu(111) surface will be neighboring hcp sites. In figure 3, co-adsorbed H and CO on neighboring Cu(111) surface is set as reactant geometry. In figure 4, adsorbed COH on Cu(111) hcp sites is set as product geometry. The structures of reactant and product are optimized using the same convergence setting for surface optimization as shown in figure 3 and 4.
![](https://sites.psu.edu/dftap/files/2020/04/post3-figure5-300x253.png)
Figure 5. Transition state structure. The adsorbed species is no longer perpendicular to Cu(111) surface.
The transition state search is performed using a complete LST/QST method implemented in the CASTEP module of Material Studio. The RMS(Root Mean Square) convergence is set to 0.24 eV/A. Followed by a transition state confirmation of higher precision with energy convergence of 2.0e-5 eV/atom, maximum force of 0.1 eV/Å and maximum displacement of 0.005Å. In figure 5, the transition state geometry is shown. Although in both reactant and product structure, adsorbed species are perpendicular to Cu surface, the adsorbed species in transition state structure is rotated towards the initial position of adsorbed H atom.
Figure 6 is the relative energy diagram. Here energy values are reported relative to that of initial reactant state. As shown in the figure, activation barrier Consequently the activation energy from reactant is found to be 2.48 eV. Comparing to reported activation energy in [2-4], which is less than 1.0 eV, this activation energy is significantly larger. Since in their models, they include explicit water molecules to model what happens in solution. The structure of reactant and product used in this work actually modeled the reaction happens in gas phase. Meanwhile, they also argued that possible proton and electron transfer between adsorbed CO molecules and solvent molecules, which won’t induce surface atoms rearrangement as much as in gas phase.
Conclusion
The transition state geometry of *CO to *COH is found with a reaction barrier from reactant of 2.08 eV. Comparing to the reported electrode model, gas phase reduction of *CO to *COH is significantly more difficult. If given more computational time and resources, similar procedure could be used to address more complex surface reaction network starting from *CO and include explicit solvent molecules to model more realistic reaction conditions.
Reference
[1] Liu, X.; Xiao, J.; Peng, H.; Hong, X.; Chan, K.; Nørskov, J. K. Understanding Trends in Electrochemical Carbon Dioxide Reduction Rates. Nat. Commun. 2017, 8 (1), 15438. https://doi.org/10.1038/ncomms15438.
[2] Hussain, J.; Jónsson, H.; Skúlason, E. Calculations of Product Selectivity in Electrochemical CO2 Reduction. ACS Catal. 2018, 8 (6), 5240–5249. https://doi.org/10.1021/acscatal.7b03308.
[3] Nie, X.; Esopi, M. R.; Janik, M. J.; Asthagiri, A. Selectivity of CO2 Reduction on Copper Electrodes: The Role of the Kinetics of Elementary Steps. Angew. Chemie – Int. Ed. 2013, 52 (9), 2459–2462. https://doi.org/10.1002/anie.201208320.
[4] Nie, X.; Luo, W.; Janik, M. J.; Asthagiri, A. Reaction Mechanisms of CO2 Electrochemical Reduction on Cu(1 1 1) Determined with Density Functional Theory. J. Catal. 2014, 312, 108–122. https://doi.org/10.1016/j.jcat.2014.01.013.