AI-driven random walk simulations of viscophoresis and visco-diffusiophoretic particle trapping

TL;DR

This paper introduces a novel simulation approach for viscophoresis and visco-diffusiophoresis, demonstrating their potential for nanoparticle sorting in microfluidic systems, with the aid of AI-generated modeling tools.

Contribution

It presents a new random walk simulation model for viscophoresis and diffusiophoresis, validated against experimental data, and highlights the use of AI tools in accelerating physical modeling.

Findings

Simulation aligns with experimental data

Predicts size-based nanoparticle separation

Demonstrates AI-assisted model development

Abstract

Viscophoresis refers to the transport of suspended nanoparticles driven by a steep viscosity gradient. This work investigates this new transport effect using a random walk simulation. By modelling position-dependent Brownian motion, viscophoresis, and diffusiophoresis in a one-dimensional geometry, the simulation yields results that align well with experimental data, demonstrating viscophoresis as a new phoretic transport mechanism. Additionally, the simulation predicts the efficient separation of nanoparticles based on size, suggesting potential applications for sorting in microfluidic systems. The Python script for the simulation was generated using ChatGPT o1, significantly accelerating model development and providing accurate physical insights and efficient equations. However, caution is advised, as ChatGPT may generate non-physical results; iterative testing and validation is important.