Reaching the Full Potential of Machine Learning in Mitigating Environmental Impacts of Functional Materials

In conventional ways of functional material design, the performance of synthesized materials is the focal point, whereas the toxicity of and environmental problems caused by synthesized materials are neglected to a large extent. Only with a balanced consideration of all the above-mentioned factors can we ensure the development of eco-friendly functional materials. In recent years, with big data generated by experiments and computing technology becoming increasingly accessible, data-driven solutions, especially machine learning methods have opened a new window for the discovery and rational design of eco-friendly functional materials. In this review, we first presented a brief introduction of functional materials, the most commonly used machine learning models and relevant processes. The applications of ML-based approaches and computational methods in functional prediction and material design were then summarized. More importantly, by combining machine learning methods with the toxicity prediction of functional materials, we proposed a framework for sustainable functional material design to achieve better functionality and eco-friendliness. Such a framework will ensure both the practicability and effectiveness of functional materials, balance their functionality and environmental sustainability, and eventually pave the path toward the Sustainable Development Goals set by the United Nations.