FET: Investigating Neuromorphic ISING Machines using Single-electron Fowler-Nordheim Annealer Network

Recently, there has been considerable interest in application domains where neuromorphic or neurally inspired architectures can demonstrate significant performance advantages over architectures based on central processing units (CPUs), graphics processing units (GPUs), and tensor processing units (TPUs). Combinatorial optimization and Ising machines represent one such domain, with applications in operations research, artificial intelligence, security, bioinformatics, finance, and materials discovery. This project is investigating neuromorphic Ising machines using a network of single-electron quantum devices that could lead to practical Ising machines with unprecedented energy-to-solution and time-to-solution metrics. The sampling techniques developed in this research will enable scalable optimization methods that can exploit quantum statistics, operate under non-cryogenic conditions, and be implemented using standard silicon fabrication processes. The project also provides training opportunities for undergraduate and graduate students in the areas of artificial intelligence, neuromorphic engineering, and quantum engineering. This project will also support the organization of a Neuromorphic Stochastic Thermodynamics (NeST) workshop focused on Ising machines. The workshop will bring together stakeholders from academia, industry, and government to explore new directions and opportunities at the intersection of neuromorphic engineering and stochastic thermodynamics. The results will be showcased and disseminated at the annual Telluride Neuromorphic Workshop. The technical approach in this project is based on a spiking neuromorphic architecture called NeuroSA, in which the firing thresholds of the neurons are adapted according to the physics of Fowler–Nordheim (FN) quantum mechanical tunneling. In this context, the project will model single-electron transport across the FN-tunneling barrier as surrogates for spikes in NeuroSA. Single-electron transport and sampling processes will be validated using measurements on floating-gate devices implementing the FN tunneling barrier. The project will examine the challenges associated with coupling these barriers, with scaling the proposed Ising machine on a standard semiconductor fabrication process and when operating the machine under non-cryogenic conditions. This will be addressed by investigating circuits with the ability to self-calibrate and self-compensate for mismatch. Both second-order and higher-order Ising architectures will be investigated, and the project will address challenges related to programmability, readout, and interfacing with these machines. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria. NSF Award ID: 2551604 | Program: 01002627DB NSF RESEARCH & RELATED ACTIVIT | Principal Investigator: Shantanu Chakrabartty | Institution: Washington University, SAINT LOUIS, MO | Award Amount: $679,322 View on NSF Award Search: https://www.nsf.gov/awardsearch/show-award/?AWD_ID=2551604 View on Research.gov: https://www.research.gov/awardapi-service/v1/awards/2551604.html