Atomistic Simulation Incorporated with Science Pedagogy to Advance Machine Learning-Assisted Materials Research

The current study aims to use a guided inquiry-based instruction framework and other evidence-based science strategies to advance materials research in coordination with computational modeling and machine learning approach. For the delivery of our program, guided inquiry-based instruction was used where students generate inquiry from their own experiences into authentic questions and are given opportunities to explore and discover materials processes such as computational synthesis of functional metal nanoparticles and 2D materials. Dynamic representations and manipulation of abstract and unobservable phenomena (e.g., chemical reactions at molecular and atomic levels) will be made available to student scholars where they build molecular structures, run atomistic simulations, and visualize and analyze simulation results via machine learning assisted characterization. We measure students’ perception and motivation toward learning materials science and engineering; and machine learning assisted computational design skills using the existing tools and researcher-developed evaluation criteria. Results indicated positive impacts of our pedagogical approached simulation training. Implications and future directions will be discussed. We believe our approach will strengthen a diversity in the community of machine learning-based materials research.