Research
Drying-induced nanoparticle self-assembly
Nanoparticle-based materials are important for numerous technologies, ranging from consumer products such as paints to advanced materials used for catalysis or photonics. Often these materials are made by dispersing nanoparticles in a solvent at low concentrations, then removing the solvent by evaporation or drying to concentrate them. As the solvent is being removed, the nanoparticles self-assemble into their final solid structure due to their physical interactions and the changing concentration. The self-assembled structure, and its corresponding material properties, can depend sensitively on how the nanoparticle dispersion is processed. This is especially true when the dispersion contains multiple types of nanoparticles.
We are interested in using solvent drying to cheaply and efficiently create nanoparticle coatings or bigger “supraparticles” that have controlled composition gradients. However, it is currently hard to know what combination of parameters, such as drying rate or nanoparticle chemistry, will produced the desired structure after the solvent is removed. We are using different simulation methods, including both particle-based and continuum-level modeling, to understand how thermodynamic and transport considerations influence drying-induced self-assembly. Our ultimate goal is to computationally design conditions that produce a targeted self-assembled structure.
This material is based upon work supported by the National Science Foundation under Award No. 2223084. Any opinions, findings and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation.
Multiscale inverse design of nanocrystal superlattices
Due to their small sizes, nanocrystals have remarkable properties that make them promising components of advanced sensors and photocatalytic devices. The properties of individual nanocrystals can be enhanced and controlled by assembling them into larger ordered superlattices; however, there are limited scalable strategies for fabricating superlattices with low defect rates.
We are computationally exploring a new solvent-based strategy to address this problem. Our computational approach couples particle-based simulations with optimization methods in an inverse design strategy that aims to determine physicochemical properties of the solvent and nanocrystals that cause the nanocrystals to self-assemble into a target superlattice. We aim to make computational predictions that can be directly tested in the laboratory, so we are actively developing inverse-design strategies to enable this connection, including:
- Integrating atomistic and coarse-grained simulations in a single inverse design loop
- Using data-driven (surrogate) models to accelerate optimization
We have also developed a new software package, relentless, to carry out our inverse design calculations. We anticipate relentless will be broadly useful beyond our own application!
Ultra-coarse-grained simulations of protein self-assembly
![Methodology for ultra coarse-graining [[source]](https://pdb101.rcsb.org/motm/109).](/images/research-ultra-cg.jpg "Methodology for ultra coarse-graining [[source]](https://pdb101.rcsb.org/motm/109).")
Capsids are container-like structures made from proteins that form a protective shell around a virus’s genetic material. Self-assembly of the capsid plays a critical role in the virus life cycle, and understanding the dynamics of this process may provide an avenue for designing therapeutics. However, characterizing the dynamic pathway of capsid self-assembly is challenging to for experiments, but the length and time scales of this process exceed current capabilities of atomistic molecular dynamics.
To address this challenge, we are working with Dr. Chris Kieslich to develop a computationally efficient ultra-coarse-grained model, representing each capsid protein as a single coarse-grained particle, so that we can simulate capsid assembly from protein solutions at unprecedented scales. Our approach systematically integrates surrogate modeling, sparse sampling, and statistical mechanics for coarse graining to efficiently compute an accurate approximation of the pairwise potential of mean force and torque.
This work is supported by Auburn University’s Research Support Program.