The ECC project’s goal is to create a computational framework that accelerates discovery and characterization of complex molecular systems. We are targeting gas-phase and coupled heterogeneous/gas-phase reactions and reaction mechanisms with relevance to catalytic conversion of hydrocarbons, oxygenates, and small molecules. We capitalize on recent improvements in theoretical chemistry combined with improved mathematical software for solving complex problems integrating exascale-size supercomputers to develop a uniquely powerful chemical computational toolset for the research community.
We have three major thrusts of the research that we will conduct supported by an array of mathematical, software, and algorithmic tools that enable them. The hierarchy of the parts of the proposal are shown on the right, and we define the three main areas as follows:
Automated reaction path exploration on multidimensional potential energy surfaces. Using advanced machine learning algorithms coupled to quantum chemistry code suitable for heterogeneous exascale architectures we will develop codes that explore reaction pathways for elementary reactions. These algorithms will automatically discover previously unknown elementary reaction classes and produce calculations on a large number of homologous systems to build up minable reaction databases.
Automated reaction mechanism generation for heterogeneous catalysis. We will develop a complete computational infrastructure that generates reaction mechanisms for heterogeneous catalysis. This will store and recall reactions and build them into a mechanism that is able to predict macroscopic observables. Moreover, we will solve the underlying kinetic equations efficiently, so that they allow for advanced mechanism reduction and uncertainty quantification.
Advanced thermochemistry database. We will create a modern, intelligent, user-friendly and accurate thermodynamic database that is self-consistent and can be grown by us and the broader scientific community. For gas-phase systems we will incorporate automatic state-of-the-art energy evaluations and anharmonic corrections and generate thermodynamic properties directly applicable in common kinetics codes, including ours. For heterogeneous systems, we will implement methods beyond DFT to calculate energies, and novel approaches for entropy terms.
Heterogeneous catalysis is a crucial part of the modern economy, used to upgrade heavy fossil fuels, enable the partial reduction of bio-derived feedstocks, or convert small molecules, such as CO, CO2 or methane, into larger and more valuable compounds. To achieve these goals most efficiently and selectively it is necessary to design and fine-tune catalysts and operating conditions through a combination of theory and modeling. Our work will provide the necessary molecular-level understanding and quantitative description of chemical processes at the catalyst surface and in the gas phase above it to create an exascale enabled toolset for the broad research community to help design catalytic systems that can answer challenges of the 21st century, such as energy security and curbing global warming.