CogniGron Seminar by Pavan Nukala: "Neuromorphic materials: electromechanical neuristors and critical networks"

Neuromorphic hardware can be thought of as a bottom-up assembly of individual building blocks, such as neurons and synapses. The memristor crossbars (enabling novel computing strategies) are classic examples of such architectures. However, real biological neural networks have a very complex architectures with ~1015 interacting elements. To emulate such a behaviour, it is worth exploring self-assembled materials networks that behave globally like biological neural networks, and strategies to train these networks.

In this talk, I’ll discuss our recent progress on both these aspects: individual computing elements such as neuristors, and self-assembled network of Ag nodes and filaments hosted in a 2D hBN matrix.

In the first part, I’ll draw analogies between a biological neuron (in a pig gut) that exhibits gel to solid-like phase transition about the body temperature, and correlated Mott insulators (VO2, NdNiO3) that also exhibit a volatile phase transition with current. I’ll show at least three different self-oscillatory behaviors from such systems (tonic spiking, bursting and phasic spiking), and present simple electrothermal models that will model these systems effectively. A traveling action potential in a neuron is also a sound wave, and thus is electromechanical in nature. I’ll clearly show through interferometry experiments that even in VO2, the self-oscillations are both electromechanical and electrooptical in nature, opening up the applications of these neuristors in MEMS and optics.

In the next part, I’ll show that in Ag-hBN (large area, CVD grown) systems, we could stabilize two different networks: (a) a percolative tunnel network of Ag clusters intercalated between vdW BN layers, and (b) a physically percolating filamentary Ag network. Critical exponents of various order parameters will be shown, and I’ll discuss how both these networks exhibit avalanche dynamics of a system poised at criticality. I’ll also discuss how we are attempting to model these systems.

References:

U. Khandelwal, Q. Guo, B. Noheda, P. Nukala, S. Chandorkar, Dynamics of Voltage Driven Self-Sustained Oscillations in NdNiO3 neuristors, ACS Applied Electronic Materials, 2023, 5, pp:2859

U. Khandelwal, …, N.A. Phani, S. Avasthi, S. Chandorkar, P. Nukala et al., Large electro-opto-mechanical coupling in VO2 neuristors, arXiV:2306.14367

A. Rao, S. Raghavan, P. Nukala et al., Realizing avalanche criticality in neuromorphic networks on a 2D-hBN platform, Materials Horizon, 2023.