Advancing sustainability: Biodegradable electronics and materials discovery through artificial intelligence

Abstract

The pressing need for sustainable materials and devices stems from growing environmental concerns and the imperative to mitigate climate change. Traditional materials and devices often rely on non-renewable resources and generate significant waste and pollution throughout their lifecycle. By prioritizing sustainability in material and device design, we can foster innovation, promote circular economies, and build a greener future for generations to come. Artificial intelligence (AI) and machine learning (ML) can analyze vast datasets to identify novel materials with desirable properties by reducing the experimental workload. In this paper, we explore the synergistic relationship between sustainable materials discovery and ML models. By leveraging advanced algorithms, researchers can efficiently explore vast chemical spaces to identify environmentally friendly materials with tailored properties. ML techniques, including predictive modeling and generative models, facilitate the rapid discovery and optimization of sustainable materials for various applications, ranging from renewable energy technologies to eco-friendly consumer products. We present a landscape view of the field with a focus on the most recent developments, focusing mainly on transitory materials such as metals, polymers, and semiconducting materials. Furthermore, classification and regression techniques to model the degradation behavior of polymers have been addressed, pointing to key challenges and proposing solutions for enhanced ML applications. The paper discusses the challenges of scaling up data-driven technologies from small molecules to polymers, underscoring AI’s role in discovering new molecular designs and optimizing existing ones for novel applications. It emphasizes the importance of defining and standardizing polymer systems to enable ML models to create a unified data collection system for AI and automation enhancements. Furthermore, it stresses the necessity of refining ML methods to harness the benefits of data-driven polymer chemistry fully, emphasizing the importance of reliable and diverse datasets for predictive models in polymer synthesis.