A scalable kinetic Monte Carlo platform enabling comprehensive simulations of charge transport dynamics in polymer-based memristive systems

TL;DR

This paper presents a scalable, GPU-accelerated kinetic Monte Carlo simulation platform for modeling ion transport in polymer-based memristive systems, enabling detailed analysis of charge dynamics relevant to energy storage and neuromorphic computing.

Contribution

The authors develop a flexible, efficient simulation platform that integrates experimental parameters and captures complex ion transport behaviors in solid-state devices.

Findings

Successfully validated with polymer memristive devices

Reproduces key phenomena like hysteresis and plasticity

Offers high computational efficiency and scalability

Abstract

Polymer-assisted ion transport underpins both energy storage technologies and emerging neuromorphic computing devices. Efficient modeling of ion migration is essential for understanding the performance of batteries and memristors, but it remains challenging because of the interplay of drift, diffusion, and electrostatic interactions, as well as the limitations of continuum and molecular dynamics approaches. Addressing these challenges is particularly relevant in the context of the climate and energy crisis, where high-performance, low-carbon technologies require optimized ion-conducting materials and devices. Here, we introduce a scalable and flexible stochastic simulation platform that uses Markov chain Monte Carlo methodology to model ion migration in solid-state systems. The platform employs a vectorized, rail-based representation of device geometry, enabling rapid simulation of lateral ion transport and space-charge effects while preserving the stochastic nature of hopping events. It accommodates a wide range of material systems and can integrate experimental input parameters without code modification. We also provide an implementation of the model that takes advantage of highly energy-efficient GPUs, improving the performance and reducing the carbon footprint of the simulations. Validation using polymer-based memristive devices demonstrates the simulator's ability to capture key behaviors, including relaxation decay, current-voltage hysteresis, spike-timing-dependent plasticity, and learning/forgetting rates. By balancing computational efficiency with mesoscale physical considerations, the platform provides a versatile tool for exploring ion-driven phenomena in energy storage and neuromorphic devices, supporting exploratory research.