2D Nanosystems: Applications of 2D Semiconductors for In-Memory Computing

Abstract: This talk explores the potential of using two-dimensional (2D) materials, specifically monolayer MoS2, in in-memory processors. Traditional computing architectures face challenges in efficiently computing matrix-vector multiplications, crucial for machine learning algorithms. In-memory computing, where operations occur in the memory's physical layer, offers energy benefits. While various materials have been considered, 2D materials remain unexplored. The first part of the talk demonstrates the fabrication of monolayer MoS2-based floating-gate memories, showcasing their potential as scalable analog memories and memristors. The second part focuses on designing logic-in-memory and analog computing systems, revealing the promising energy efficiency of MoS2 memories compared to silicon counterparts. The final part explores large-scale integration, culminating in a 1024-device in-memory computing core, affirming MoS2 as a promising avenue for advancing in-memory processors.

Bio: Guilherme Migliato Marega is a Postdoctoral researcher at EPFL in the Laboratory of Nanoscale Electronics and Structures. He holds a Ph.D. in Microsystems and Microelectronics from EPFL under the supervision of Professor Andras Kis, Switzerland. During his Ph.D., he worked on developing nanosystems based on two-dimensional semiconductors and their applications to in-memory computing. He earned a Master's in General Engineering from Ecole Centrale Lyon, France, and graduated first in his class in Electrical & Electronics Engineering at the University of São Paulo, Brazil. For his Ph.D. thesis, Dr. Marega received the Doctoral Thesis Award in Microsystems & Microelectronics at EPFL 2023. His current interests are in exploring emerging device technologies to craft heterogeneous circuits with the potential to pioneer unprecedented applications.

---

Hosted by Peter McMahon (AEP). Please email kicnano@cornell.edu for the Zoom link.