Global Optimization of Adsorbate–Surface Structures While Preserving Molecular Identity
As the complexity of atomistic simulations in catalysis and surface science increases, the challenge of manually finding the lowest-energy adsorbate–surface geometries grows significantly. In the current work, a global optimization approach that preserves adsorbate identity is applied to enable the automated search for optimized binding geometries. This technique is based on the minima hopping method developed by Goedecker, but is modified to preserve the molecular identity of adsorbates by the application of a new class of Hookean constraints. These constraints are completely inactive when the adsorbate identity is preserved, but act to restore the adsorbate structure via a Hookean force when the bond length exceeds a threshold distance. Additionally, a related Hookean constraint has been developed to prevent adsorbates (particularly such adsorbates as CO and CH2O that have stable gas-phase forms) from volatilizing during the molecular dynamics portion of the minima hopping technique. This combination, referred to herein as the constrained minima hopping method, was tested for its suitability in finding the minimum-energy binding configuration for a set of 17 C x H y O z adsorbates on a stepped Cu fcc(211) surface and in all cases found the global minima in comparable or fewer steps than the previous brute force methodologies. It is expected that methods such as this will be crucial to finding low-energy states in more complex systems, such as those with high coverages of adsorbed species or in the presence of explicit solvent molecules.
KeywordsConstrained minima hopping Global optimization Adsorption Electronic structure theory Constrained optimization
The author appreciates valuable technical discussions and feedback from Jens Nørskov (Stanford, SLAC), Lars Grabow (University of Houston), A.J. Medford (Stanford), and Sandip De (University of Basel); and wishes Professor Nørskov continued success as he celebrates his 60th birthday. The author thanks Yin-Jia Zhang (Brown University) for help in enumerating the possible binding configurations on an fcc(211) surface. This work was supported by the Young Investigator Award from the Office of Naval Research under award N00014-12-1-0851. High-performance computational resources were employed at the Center for Computation and Visualization, Brown University.
- 17.Molina LM, Lee S, Sell K, Barcaro G, Fortunelli A, Lee B, Seifert S, Winans RE, Elam JW, Pellin MJ, Barke I, von Oeynhausen V, Lei Y, Meyer RJ, Alonso JA, Rodrguez AF, Kleibert A, Giorgio S, Henry CR, Meiwes-Broer K-H, Vajda S (2011) Size-dependent selectivity and activity of silver nanoclusters in the partial oxidation of propylene to propylene oxide and acrolein: a joint experimental and theoretical study. Catal Today 160:116CrossRefGoogle Scholar
- 45.Pedersen A, Berthet J-C, Jónsson H (2012) Simulated annealing with coarse graining and distributed computing. In: Jónasson K (eds) Applied parallel and scientific computing, volume 7134 of Lecture Notes in Computer Science. Springer, Berlin, pp 34–44Google Scholar
- 65.Wang S, Petzold V, Tripkovic V, Kleis J, Howalt JG, Skúlason E, Fernández EM, Hvolbæk B, Jones G, Toftelund A, Falsig H, Björketun M, Studt F, Abild-Pedersen F, Rossmeisl J, Nørskov JK, Bligaard T (2011) Universal transition state scaling relations for (de)hydrogenation over transition metals. Phys Chem Chem Phys 13:20760CrossRefGoogle Scholar