Dresden 2020 – scientific programme

The DPG Spring Meeting in Dresden had to be cancelled! Read more ...

Parts | Days | Selection | Search | Updates | Downloads | Help

CPP: Fachverband Chemische Physik und Polymerphysik

CPP 14: Topical Session: Data Driven Materials Science - Materials Design I (joint session MM/CPP)

CPP 14.4: Talk

Monday, March 16, 2020, 11:15–11:30, BAR 205

Towards Building New Zeolites with Machine Learning — •Benjamin A. Helfrecht¹, Rocio Semino^{1, 2}, Giovanni Pireddu^{1, 3}, Scott M. Auerbach⁴, and Michele Ceriotti¹ — ¹École Polytechnique Fédérale de Lausanne, Lausanne, Switzerland — ²Université de Montpellier, Montpellier, France — ³Università degli Studi di Sassari, Sassari, Italy — ⁴University of Massachusetts Amherst, Amherst, Massachusetts USA

Synthesizing new zeolites, which are useful for applications like gas separation and catalysis, with specific properties is an ongoing challenge in the zeolite community. Ideally, one would like to select a handful of compatible “building blocks” from which a new zeolite with desired properties can be synthesized. In this work, we make progress toward this goal by constructing an “atlas” of local atomic environments comprising several thousand all-silica zeolites from the Deem SLC PCOD database [1] using machine learning techniques. We evaluate the utility of this atlas by examining correlations between the locations of the atomic environments in the atlas and their energy and volume contributions to their parent frameworks.

[1] R. Pophale, P. A. Cheeseman, M. W. Deem, A database of new zeolite-like materials, Phys. Chem. Chem. Phys 13(27):12407-12412, 2011.