Sustainable Materials Design using Machine Learning



Materials informatics draws on the fields and materials science and machine learning to identify high performing compounds for applications in electronics, energy, environment, infrastructure, manufacturing and health.   The goal is to find reliable structure/property relationships that allow researchers to make informed decisions about what to focus on.  The ultimate decision about what to make, however, is rarely based solely on the “desirable properties”, and necessarily includes pragmatic considerations such as the cost, availability and safety of materials as well. Materials informatics must be coupled with materials economatics to predict aspects of supply, as well as demand, and identify compounds that could actually be made in practice. 

In this project you will use machine learning methods to predict and rank the properties of some industrially important materials, based on the physicochemical features.  You will then develop entirely new features based on microeconomics and apply the machine learning methods to predict and rank their economic viability and market potential.  By comparing these results, you will identify the most sustainable materials for the given application domain.  Methods will include interpretable classification and regression models and artificial neural networks.

This is a 24cp project. Data sets will be provided.



To design a framework for predicting materials that are both physically optimal and financially viable.


python programming, data science and machine learning and microeconomics


machine learning, economics, materials, industry 4.0

Updated:  10 August 2021/Responsible Officer:  Dean, CECS/Page Contact:  CECS Marketing