Controlling curvature of self-assembling surfaces via patchy particle design

TL;DR

This paper presents an inverse design method using automatic differentiation to control the Gaussian curvature of self-assembling patchy particle surfaces, enabling precise tuning of their geometry.

Contribution

It introduces an optimization scheme that effectively designs patch patterns for targeted curvature in self-assembling systems, overcoming gradient explosion issues.

Findings

Successfully designed patch patterns for various curvature radii

Demonstrated consistent control over surface geometry

Optimized patch distributions guide self-assembly into desired shapes

Abstract

Curved structures in soft matter and biological systems commonly emerge as a result of self-assembly processes where building blocks aggregate in a controlled manner, giving rise to specific system structure and properties. Learning how to precisely tune the curved geometry of these assemblies can in turn elucidate new ways of controlling their functionality. We discuss how one can target self-assembly into surfaces with specified Gaussian curvature in a one-component system of model patchy particles. Given the vast design space of potential patch distributions, we address the problem using an inverse design approach based on automatic differentiation and develop an optimization scheme which solves the exploding gradients problem that arises when we differentiate through long molecular dynamics trajectories. We discuss the model requirements for successful optimization, determine the significant hyperparameter choices influencing algorithm performance and, finally, we demonstrate that we can consistently design patch patterns for assembly into clusters with different target curvature radii.