WEB High Throughput Screening for Electrochemical Nitrogen Reduction for the Ammonia Synthesis via Machine LearningWednesday (05.08.2020) 08:47 - 08:47 Poster Room Part of:
Ammonia (NH3) as a fundamental chemical material is essential to industrial areas for fertilizer, hydrogen storage, and fuels. Generally, the production of 180 million tons of NH3 per year depends on the Haber-Bosch process. This energy-intensive process under high temperature (400~ oC) and high pressures (150~ bar) consumes nature gas with huge CO2 emissions. Electrochemical nitrogen reduction for the ammonia synthesis has emerged as a sustainable approach to overcoming the limits of the Haber-Bosch process. So, we want to find a new electrochemical NRR catalyst through high throughput screening using machine learning. High throughput screening of electrocatalysts for nitrogen reduction needs many calculations in the search space, making the computational cost for predicting eligible electrocatalyst. To overcome drawback, we used an artificial neural network (ANN). This neural network consists of Long short-term memory model (LSTM) and light gradient boosting machine model (LGBM), and it predicts for eligible electrocatalysts through supervised learning using peraday efficiency and yield. Various parameters used in deep learning models include simplified molecular input line entry system(smiles), loading mass, conductor, electrolytes and temperature, which are important factors in finding new NRR catalysts. We anticipate the study of high throughput screening using machine learning to help discover NRR catalysts and to be the starting point for expansion to other catalysts research.
|Category||Short file description||File description||File Size|
|Extended Abstract||This is a short manuscript to the abstract||71 KB||Download|