Skip To Main Content
Student Examines Material
Find answers to common questions about undergraduate research and resources to match the opportunity that's right for you.

Dr. Raymundo Arroyave

Most significant advances in energy technologies have been enabled by the discovery of new materials, which have increased the overall efficiency of energy usage (e.g., new high temperature alloys33-34), expanded the renewable energy inventory (e.g., more efficient photovoltaics35-37), facilitated the storage of energy (e.g., battery materials38-39), and allowed the recovery of energy that otherwise would be wasted (e.g. thermoelectrics40-41). Novel multifunctional materials enable the conversion of various energy forms (magnetic, mechanical, thermal and electrical) through couplings between different kinds of energy forms and have the potential to significantly outperform conventional conversion technologies that will contribute to our energy independence. When faced with the problem of developing a material to solve a specific need, a materials scientist conducts a series of time-consuming and costly physical and/or computational experiments to navigate the complex materials design space. Fortunately, recent advances in materials genomics42, Artificial Intelligence43-44(AI) and data science45-46make it possible to potentially discover relationships between materials features and performance through data-enabled approaches that can be further leveraged to accelerate the discovery of materials with optimal functionality (Fig. 3).

Research Plan:
The Reu students will be involved in the development of machine learning (ML) approaches to build predictive models connecting materials features and performance. Over the 10 weeks of the program, students will be trained in some of the basic physical aspects of particular multi-functional materials system, such as ferrocaloric Heusler alloys and thermoelectrics (Weeks 1, 2). They will also be trained in data visualization, data analysis and ML workflows through the Orange platform47-48(Week 2, 3). Students will be capturing data sets of the relevant materials to be investigated and will be mentored by the graduate student in the curation of materials data, which will be stored in the Materials Data Curation System, MDCS49(Weeks 4, 5). At the same time, they will be trained in the use of scripting programming languages (such as Python50) to automate many of the pre- and post-processing of files necessary to run calculations. using Orange under the mentorship of graduate students, the REU students will explore the use of different supervised and unsupervised ML models and evaluate the models through rigorous informatics evaluation techniques (Weeks 6-10).