Prof. Dr. Blazej Grabowski
University of Stuttgart
“Atomic-scale understanding of mesoporous metallosilicates via ab initio-based machine learning potentials”
Mesoporous silica is a workhorse for industrial catalysis. It is likewise a crucial support material to produce well-defined geometrical confinement utilizing the true liquid crystal templating (TLCT) approach for the efforts in the CRC1333. Despite the significant role it plays, a detailed understanding of the electronic and atomic structure is not yet available.
This applies, in particular, to the current developments of the TLCT approach where small amounts of metal-atom additions are envisioned to systematically promote active catalytic sites on the pore surfaces. The challenge to simulate such mesoporous metallosilicates is tremendous. The amorphous (non-periodic) nature of the silica matrix on the sub-nanoscale needs to be coupled with the long-range periodicity of the mesopore arrangement on the nanoscale. Explicit electronic effects become crucial when the metal atoms get incorporated into the silica matrix. Accurate density-functional-theory (DFT) calculations are, however, too costly to explicitly model the required large system sizes.
In this talk I will present our strategy to tackle the challenge. We develop machine learning potentials, specifically moment tensor potentials (MTPs), that enable the transfer of the DFT accuracy to larger length scales. Thereby, accurate simulations (molecular dynamics and Monte Carlo) of mesoporous metallosilicates with pore dimensions of relevance to the CRC become possible. Since the machine learning potentials provide only access to energies and forces but not to the electronic structure, we propose a hybrid approach in which approximate electronic-structure methods, specifically density-functional tight-binding, are used to access the electronic information for the large-scale models.
With our approach, we are able to understand the details of metal-atom insertion (current focus on Al) and its interaction with charge-compensating protons. The critical question of metal atom distribution—mesopore surface versus silica bulk—is addressed. We study water diffusion through the mesopores and water interaction with the surface silanol groups. To generate a broader understanding, the obtained simulation insights are put in connection with experimental results.
Prof. Blazej Grabowski is a full professor at the University of Stuttgart. He is head of
the Materials Design Department at the Institute of Materials Science. Previously, Prof. Grabowski was group leader at the Max-Planck-Institut für Eisenforschung in Düsseldorf and senior scientist at the Livermore National Laboratory, USA. His research interests are finite temperature ab initio simulations, machine learning potentials, and large-scale simulations. Prof. Grabowski was awarded twice the prestigious ERC grant.