Research Description
Our research introduces a computational framework that treats information-processing capacity as a designable property of materials. Information-processing capacity refers to a material’s ability to acquire, transform, or store information through its physical dynamics. It can be quantified through quantum Fisher information for quantum sensing, and through computational capacity for bio-inspired computing. These quantities depend on underlying material properties such as nonlinear dynamics, relaxation and dissipation processes, which are set by material structure and can be engineered. We use density-functional theory and quantum chemistry to design atomic-scale structure, molecular dynamics to model the resulting dynamics, and quantum transport simulations to predict device-level function.
Theme 1: Quantum Sensing Through Materials Design
In defect-based quantum sensors, sensitivity is ultimately limited by the physics of the defect-host system. T₁, T₂, and C are set by spin–phonon coupling, spin–bath interactions, defect geometry, strain, and the local crystalline environment, including surface terminations. These same mechanisms influence bandwidth, spatial resolution, and frequency resolution, meaning that enhancing sensitivity requires navigating multi-parameter trade-offs quantitatively. We currently focus on spin defects in semiconductors, with future extension to molecular spin systems.
Our approach is to identify the dominant decoherence channels for a given defect-host combination, model how material modifications such as strain, surface termination, and crystalline environment alter them, and develop experimentally testable design strategies. Sensitivity remains the critical bottleneck for applications in biomedicine, semiconductor metrology, and navigation, and narrowing the gap between current performance and what the material physics can in principle support is the challenge this work addresses.
Theme 2: Bio-Inspired Computing Through Materials Design
In biological computation, stochastic dynamics enable exploration of large state spaces. Collective interactions among many degrees of freedom produce emergent function. Self-adaptation allows the system to modify its own behavior in response to changing inputs. Realizing these principles in material structure requires designing dynamics that are nonlinear, multistable, and reconfigurable. Our work on the context-aware neuron demonstrates this approach: a single device whose oscillatory dynamics give rise to context-awareness, cross-frequency coupling, and feature binding through the material’s own physics.
Materials that compute through their own dynamics could solve optimization problems by physically relaxing through energy landscapes rather than executing algorithms. They could also adapt to changing conditions without retraining. Processing signals at the physical interface before digitization would reduce energy and latency. Distributing function across the material rather than concentrating it in individual components could allow graceful degradation under partial failure, a property essential for deployment in uncontrolled physical environments.