Discovery of New Anode SEI Forming Additives Using an in silico Evolutionary Approach
High throughput quantum chemistry screening of large libraries of structures is an approach that is becoming more popular with the ever decreasing cost of CPU cycles, however this can be a time consuming and highly curated procedure. The automated evolution of a set of input structures and defining the direction in which chemical properties should be automatically adjusted requires less user management and enables knowledge creation rather than knowledge implementation. As the expense of computing resources decreases a monitored genetic algorithm to improve the chemistry of an initial set of structures can be a faster method requiring less user intervention than traditional high throughput chemistry.
In this presentation we show how we reduce the chemical hardness and increase the reduction potential of a set of potential SEI additives or structures having potential SEI additive traits automatically while additionally suggesting optional structural characteristics that the molecules should have. An improvement of these characteristics of a set of input structures is shown using our genetic optimization through supervised learning.