Controlling Colloidal Assembly & Reconfiguration

Assembly of colloidal nano- and micro- particles into ordered configurations is often suggested as a scalable process to manufacture microstructured materials with multifunctional properties. To solve this engineering challenge, it is essential to understand how Brownian motion, colloidal interactions, collective dynamics, and external stimuli can be controlled to assemble colloidal components into functional materials. A key scientific challenge is understanding how building block interactions determine the time evolution of stochastic assembly processes along dynamic pathways via transition and metastable states toward target states.

Our approach to this problem is to implement open and closed loop control over the assembly of different shaped building blocks into a variety of useful microstructures (glassy, liquid crystalline, crystalline). Key elements that enable formal control of colloidal assembly, are developed using machine learning methods, which include: (1) the ability to quantify microstructures and morphology (sense states), (2) the capability to tune colloidal interactions (actuate state changes), (3) information about non-equilibrium microstructure and morphology evolution after tuning colloidal interactions (stochastic dynamic models), and (4) determining rules for changing colloidal interactions (control policies) based on current and target states (objectives). We demonstrate real-time control of colloidal assembly and reconfiguration for a variety of colloidal particle shapes and states including hierarchical microstructures. Our approach demonstrates formal control over non-equilibrium dynamic processes in colloidal systems, which we have shown can be extended to diverse materials and control objectives involving non-equilibrium target states, particle navigation, and colloidal machines.