CO organization at ambient pressure on stepped Pt surfaces: first principles modeling accelerated by neural networks

Step and kink sites at Pt surfaces have crucial importance in catalysis. We employ a high dimensional neural network potential (HDNNP) trained using first-principles calculations to determine the adsorption structure of CO under ambient conditions (T = 300 K and P = 1 atm) on these surfaces. To thoroughly explore the potential energy surface (PES), we use a modified basin hopping method. We utilize the explored PES to identify the adsorbate structures and show that under the considered conditions several low free energy structures exist. Under the considered temperature and pressure conditions, the step edge (or kink) is totally occupied by on-top CO molecules. We show that the step structure and the structure of CO molecules on the step dictate the arrangement of CO molecules on the lower terrace. On surfaces with (111) steps, like Pt(553), CO forms quasi-hexagonal structures on the terrace with the top site preferred, with on average two top site CO for one multiply bonded CO, while in contrast surfaces with (100) steps, like Pt(557), present a majority of multiply bonded CO on their terrace. Short terraced surfaces, like Pt(643), with square (100) steps that are broken by kink sites constrain the CO arrangement parallel to the step edge. Overall, this effort provides detailed analysis on the influence of the step edge structure, kink sites, and terrace width on the organization of CO molecules on non-reconstructed stepped surfaces, yielding initial structures for understanding restructuring events driven by CO at high coverages and ambient pressure.


Clustering mutation algorithm
The following steps are used in implementing this modified version of random atomic displacement.   1. Create a polygon (parallelogram) that maps the surface of the unit cell 2. Randomly generate n points within the polygon which act as the centroids of the Voronoi tessellation (like a power diagram). Using the polygon boundary and centroids, we can define the edges of the Voronoi tessellation (intersections of half-spaces).
3. Within the each obtained cell ("cluster"), we identify the adsorbate positions and all CO molecules within the cell are displaced in same direction randomly generated. This is implemented using a pythonic code (rand_clustering.py) using scipy and shapely packages and added to the github repository. We use an iterative process (as shown in in Fig. S3) for developing the HDNNP:

HDNNP Training procedure
• Initialization -reference dataset utilized to generate a preliminary HDNNP.
• Data Generation-new structures generated using the developed HDNNP and Basin Hopping Monte Carlo simulations.
•  Figure 2 in the manuscript. Table S1: Pt(553) LEME structures data: Free energy per unit area (G/A), Coverage of CO on the terrace top site (θ t (T)), bridge site (θ t (B)), hollow site (θ t (H)) and on the step edge top site (θ e (T)), bridge site (θ e (B)), hollow site (θ e (H)), total coverage of CO on the terrace (θ t ) and on the step edge (θ e ) and the surface area of the unit cell (A)  Figure S5: CO orientation on Pt(553) at θ =0.65 in the LEME structure

Pt(557)
Data used to generate Figure 4 in the manuscript. Table S2: Pt(557) LEME structures data: Free energy per unit area (G/A), Coverage of CO on the terrace top site (θ t (T)), bridge site (θ t (B)), hollow site (θ t (H)) and on the step edge top site (θ e (T)), bridge site (θ e (B)), hollow site (θ e (H)), total coverage of CO on the terrace (θ t ) and on the step edge (θ e ) and the surface area of the unit cell (A)

Pt(643)
Data used to generate Figure 6 in the manuscript. Table S3: Pt(643) LEME structures data: Free energy per unit area (G/A), Coverage of CO on the terrace top site (θ t (T)), bridge site (θ t (B)), hollow site (θ t (H)) and on the step edge top site (θ e (T)), bridge site (θ e (B)), hollow site (θ e (H)), total coverage of CO on the terrace (θ t ) and on the step edge (θ e ) and the surface area of the unit cell (A)

CO-Surface vs CO-CO lateral interaction
4.1 CO-Surface Interaction Figure S6: Configurations used to compare adsorption energy of CO on step edge and the terrace. Table S4: Comparison the adsorption energy of CO on the step edge and the terrace.

Low Coordination adsorption sites
Figure S11: Parity plot comparing the reference DFT energies and forces with the neural network estimates for structures with low coordination adsorption sites