by Allison Kleber

The conventional electronic device design process utilizes a well-established approach: materials with known properties, tested device structures, and designer experience, have led to the manufacturing of a variety of semiconductor devices critical to our daily life. Under this conventional process, new devices must fit the desired application into these established parameters. With the help of advanced, physics-informed machine learning (PIML) techniques, however, BU ECE researchers are setting out to transform the status-quo and develop a new paradigm in which novel semiconductor materials and device structure are designed to satisfy specific requirements, rather than the opposite. In this context, ultra-wide-bandgap semiconductors (UWBGs) are ideal test candidates.

Professor Enrico Bellotti’s (ECE, MSE) collaborative team includes researchers from the Universities of Utah and Notre Dame, as well as BU ECE colleague Professor Luca Dal Negro (ECE, MSE) . With the support of a $2.5M grant from the Army Research Office, they plan to demonstrate that their AI/Machine Learning-augmented theoretical approach and experimental validation methodology, under the title AI/ML Augmented Materials And Device Exploration (AI/ML-MADE), can deliver an impressive result: ultra-wide-bandgap semiconductor devices capable of operating in the  millimeter-wave frequency spectrum (necessary to achieve the high-capacity transmissions which fuel 5G communications, among other applications) with 10 times the power density currently available via conventionally-designed devices.

To accomplish this, the team will tackle two problems simultaneously: the development of novel materials from which to build their devices, and designing new device architectures to utilize them. Machine learning approaches will be applied to the task of identifying material combinations that will yield the desired attributes. For example, enhancing channel charge density by controlling piezoelectric characteristics and increasing operating voltages by modifying the high-field carrier transport properties, will lead to a higher power generated for a given device size.

In addition to focusing on millimeter-wave devices with such enhanced performance, Professor Bellotti and Dal Negro’s project will build a foundation for ongoing research towards novel, more powerful and efficient devices that will affect applications well beyond radio frequency electronics, such as optoelectronics and photonics. Furthermore, the research program will be enhanced by the involvement of faculty, post-doctoral researchers and students from different institutions across the country and national laboratories.

Professor Bellotti headshotProfessor Enrico Bellotti is a prolific researcher with a focus on computational electronics, parallel computing, and semiconductor materials and device simulations. He earned his PhD from the Georgia Institute of Technology in 1999 and is a member of the IEEE and life-members of International Society for Optics and Photonics (SPIE). Prof. Bellotti is also the lead-PI of the Center for Semiconductor Materials and Device Modelling (CSM) and the director of the industry consortium associated to the center.

Professor Dal Negro headshotProfessor Luca Dal Negro received his PhD from the University of Trento in Italy in 2003, and is an OSA fellow. His research interests include nano-optics and metamaterials, nonlinear and quantum optics, and advanced imaging and multifunctional optical devices.