Convergent Nanosystems for Information Processing: From Atoms to Function

Rapid advances in artificial intelligence are driving the need for fundamentally new information processing paradigms that exceed the limits of existing nanoscale technologies in energy, performance, and functionality. Biological systems offer a compelling model, tightly coupling sensing, computation, and memory within distributed architectures that achieve remarkable energy efficiency and adaptability. Such performance is enabled by diverse, interacting atomic and molecular processes. To address this gap, this seminar introduces an atomically informed framework for designing nanoscale nonlinear processes within materials and translating them into functional devices and systems for computing, memory, and sensing. We first present new platforms that leverage principles of additive manufacturing to overcome the resolution and compatibility limits of current fabrication approaches, enabling precise integration of atomic-scale building blocks into functional systems. Building on these capabilities, we introduce a computing paradigm based on molecular building blocks that embed computation and memory directly within material properties, demonstrated through energy-efficient mechanical neuromorphic information processing. Finally, we apply this platform to materials-informed sensors and transducers, including for ultrabroadband rectification and gas sensing, enabling enhanced performance in bandwidth, energy efficiency, sensitivity, speed, and functionality.

Speaker: Farnaz Naroui, Massachusetts Institute of Technology