Machine Learning Guided Multiscale Design of DNA-functionalized Nanoparticles for Targeted Self-Assembly of the Double Gyroid

Abstract

In soft matter science, it is often the goal to design new materials with targeted properties. These materials can be used in many applications, each requiring specific features to be optimized for maximum fitness. The use of self-assembly processes, where a target structure is attained from suitably designed building blocks, is a powerful tool for this optimization. However, it is still an open question how to generally design realizable building blocks that lead to the desired phases. In this work, the double gyroid is chosen as target structure and DNA functionalized nanoparticles are used as our building blocks. Using existing pair potentials as inspiration, a large design space is defined for exploration of the target structure. An effective search strategy is implemented, where free energy calculations are first used to coarse grain our originally fine-grained model of the building blocks and then quickly evaluate the fitness of each design. These data are then fed into a machine learning algorithm that allows obtaining predictions for all candidates in our design space through an active learning loop. Successful coarse-grained designs are identified and evaluated again through interfacial pinning calculations with the fine-grained model. This work leads to the development of specific, experimentally relevant designs of DNA functionalized nanoparticles that self-assembly into the target phase. The methodology used can be extended to other types of building blocks and target structures.