Public Abstract

DE-SC0026261: CMOS+NVM Izhikevich Primitive Circuit Prototyping for Biologically Realistic Neuromorphic Computing

Award Status: Active

Institution: The University of Texas at Dallas, Richardson, TX
UEI: EJCVPNN1WFS5
DUNS: 800188161

Most Recent Award Date: 09/25/2025
Number of Support Periods: 1
PM: Pino, Robinson

Current Budget Period: 09/01/2025 - 08/31/2026
Current Project Period: 09/01/2025 - 08/31/2027
PI: Friedman, Joseph

Supplement Budget Period: N/A

Public Abstract

CMOS+NVM Izhikevich Primitive Circuit Prototyping for Biologically Realistic Neuromorphic Computing Joseph S. Friedman The University of Texas at Dallas This project will design, model, fabricate, and demonstrate prototype neuromorphic computing circuit primitives that are biologically realistic. Given its ability to exhibit the firing patterns of all known types of cortical neurons despite its simplicity, the prototyping will focus on the Izhikevich neuron and synapse models. Furthermore, given the analog nature of neurobiological circuitry as well as the energy efficiency that can be achieved with analog microelectronics, the prototyping will focus specifically on analog implementations of these circuit primitives and their integration into large-scale cortical networks. Analog neuron circuit primitives will be designed and fabricated with standard complementary metal-oxide-semiconductor (CMOS) transistors in concert with MOS capacitors that provide the hysteretic behavior necessary to exhibit firing patterns with complex time dynamics, while analog synapse devices will be designed and fabricated using NVM devices controlled by CMOS transistors that perform STDP on the NVM. The behavior of these analog neuron circuit primitives will be modeled across the complete range of firing patterns exhibited by the Izhikevich neuron model, and the four Izhikevich neuron model parameters will be mapped to circuit parameters for experimental demonstrations. Similarly, the analog synapse circuit primitives will be modeled across a range of STDP parameters and their circuit realizations. The prototype Izhikevich neuron and synapse primitive circuits will be interconnected in small structures analogous to those found in the human cortex, and the behavior of these structures will be shown to emulate biological cortical behavior. Altogether, this project will produce biologically realistic circuit primitives that can be interconnected in biologically realistic circuit structures that advance neuroscience and enable large-scale AI hardware with the energy efficiency of biological intelligent systems.

CMOS+NVM Izhikevich Primitive Circuit Prototyping for Biologically Realistic Neuromorphic Computing
Joseph S. Friedman
The University of Texas at Dallas

This project will design, model, fabricate, and demonstrate prototype neuromorphic computing circuit primitives that are biologically realistic. Given its ability to exhibit the firing patterns of all known types of cortical neurons despite its simplicity, the prototyping will focus on the Izhikevich neuron and synapse models. Furthermore, given the analog nature of neurobiological circuitry as well as the energy efficiency that can be achieved with analog microelectronics, the prototyping will focus specifically on analog implementations of these circuit primitives and their integration into large-scale cortical networks.
Analog neuron circuit primitives will be designed and fabricated with standard complementary metal-oxide-semiconductor (CMOS) transistors in concert with MOS capacitors that provide the hysteretic behavior necessary to exhibit firing patterns with complex time dynamics, while analog synapse devices will be designed and fabricated using NVM devices controlled by CMOS transistors that perform STDP on the NVM. The behavior of these analog neuron circuit primitives will be modeled across the complete range of firing patterns exhibited by the Izhikevich neuron model, and the four Izhikevich neuron model parameters will be mapped to circuit parameters for experimental demonstrations. Similarly, the analog synapse circuit primitives will be modeled across a range of STDP parameters and their circuit realizations. The prototype Izhikevich neuron and synapse primitive circuits will be interconnected in small structures analogous to those found in the human cortex, and the behavior of these structures will be shown to emulate biological cortical behavior. Altogether, this project will produce biologically realistic circuit primitives that can be interconnected in biologically realistic circuit structures that advance neuroscience and enable large-scale AI hardware with the energy efficiency of biological intelligent systems.