Project

Molecular Machine Learning for Chemical Reaction Engineering

Code
DOCT/002337
Duration
10 July 2020 → 19 May 2025 (Defended)
Doctoral researcher
Research disciplines
  • Natural sciences
    • Organic chemical synthesis
    • Physical organic chemistry
    • Cheminformatics
  • Engineering and technology
    • Chemical process design
    • Modelling and simulation
Keywords
artificial intelligence chemical reaction engineering Computer aided synthesis planning cheminformatics
 
Project description

The production of new chemical products, such as pharmaceuticals, is costly, time-consuming, and environmentally harmful. The use of continuous reactors is a technique that made the petrochemical industry much more efficient decades ago and is gradually beginning to spread into other parts of the chemical industry. However, there are major contradictions between these industries. The petrochemical industry produces a limited number of products on a massive scale, whereas the pharmaceutical industry aims to produce a wide variety of products on a much smaller scale. These contradictions necessitate the development of new methods to ensure the continuous production of medicines. This thesis describes research into two crucial elements: predicting the properties of new chemical substances and devising ways to manufacture them. Specifically, artificial intelligence (AI) has been used to develop models that are reliable, interpretable, and universally applicable for chemical experts. These models have led to the development of a chemical search engine that enables researchers to discover faster methods for producing new chemical substances in continuous reactors. In this way, this research contributes to making the industry more sustainable and accelerates the search for new medicines.