(764a) Grand Canonical Inverse Design of Multicomponent Colloidal Assemblies

Authors: 
Mahynski, N. A., National Institute of Standards and Technology
Pretti, E., Lehigh University
Shen, V. K., National Institute of Standards and Technology
Mittal, J., Lehigh University
We present an inverse design methodology that leverages Machine Learning and a recently developed approach for systematically enumerating crystalline ground states of two-dimensional multicomponent materials. The latter allows us to generate phase diagrams for multicomponent colloidal mixtures which govern their equilibrium self-assembly, while the former enables these diagrams to be optimized to produce a desired target structure. Importantly, this scheme enables inverse design which explicitly accounts for the possibility of phase separation. For k-component systems at fixed temperature and pressure Gibbs' phase rule stipulates there may be as many as k coexisting phases. Conventional inverse design methods typically optimize pairwise interactions in a canonical (NVT) system to yield a Hamiltonian that will best assemble into a desired structure at that same composition. While powerful, the apparent complexity of a chosen target tends to be reflected in the resulting pairwise interactions, making these optimized systems difficult or impossible to realize experimentally. We posit that one way to produce more realizable systems is to rely on mixtures of components with simpler interactions, rather than on a single component with a more complicated interaction. For single component systems only a single (ideally, the target) phase will be thermodynamically stable; however, there is no such guarantee for multicomponent systems. In multicomponent systems containing arbitrarily complex potentials it may be reasonable to expect that the optimized set of interactions will not cause the system to phase separate, however, as simplifications are introduced to make these systems experimentally tractable, this assumption becomes less reasonable. Using our approach, which explicitly accounts for the possibility of phase separation, we optimize mixtures with simple, Lennard-Jones-like potentials to produce complex, open crystals and other lattices with broken symmetries to demonstrate how complex structures can be stabilized in multicomponent systems with very simple pairwise interactions.