University of Illinois at Urbana-Champaign
Like many scientists, Morgan Hammer has always wanted to know how things work. In his case, that curiosity is rooted in the basics.
Hammer, a Department of Energy Computational Science Graduate Fellowship (DOE CSGF) recipient, prefers a bottom-up perspective. In theoretical chemistry, his chosen subject, “we look at the fundamental building blocks and how properties at those small scales build up to the properties of things as we experience them.”
Computation facilitates that approach. “You can start from nothing and add in the atoms and molecules and physics to your model and see what happens,” says Hammer, a doctoral candidate at the University of Illinois at Urbana-Champaign. That meshes with the top-down tactics of experimental scientists. “What we model on the computer and what they measure in the laboratory hopefully meet in the middle,” validating simulations and providing insights into experiments. “This iterative process is quite fun.”
For a bottom-up approach, few things are as fundamental as solvation – how molecules and ions connect in solution. The most familiar case is dissolving salt in water. The attraction that holds sodium and chlorine atoms together weakens, and “it becomes more favorable for the ions to pair off with a group of water molecules,” Hammer says. Solvation is “the act of the water molecules grouping around the ions in that more favorable interaction.”
With advisor Yang Zhang, Hammer is examining solvation structure – “how the interaction of the ion with its solvating water affects the dynamics and structure of the liquid – both water’s impact on the ions and the ions’ impact on the water.”
Computational chemists typically have two options to calculate such reactions, Hammer says. The first, classical molecular dynamics (MD) (used interchangeably with molecular mechanics, or MM), treats all particles as effective charges, disregarding explicit electronic interactions. This works well in most cases and doesn’t demand extraordinary computational resources, enabling simulations of up to hundreds of thousands of atoms. “That also gives you a large range of concentrations you can look at because the concentration is dictated by how many water molecules for every ion you put in your simulation.”
But researchers sometimes require details that MM can’t provide. In those cases, they tap quantum mechanics (QM) electronic structure methods that treat electrons explicitly. The downside is that such models demand far more computer time and power, limiting simulation size.
Hammer combines the approaches to get the best of both. He uses the nanoscale molecular dynamics (NAMD) code, a longstanding MD code developed at Illinois that researchers recently integrated with QM programs. It uses the more precise method to calculate how ions interact with the nearest water molecules. “Then you can switch to purely classical interactions with the idea that there’s a much weaker coupling” between ions and more distant molecules, Hammer says.
Hammer plans to compare his models with experiments he and his fellow researchers conducted at Oak Ridge National Laboratory’s Spallation Neutron Source. The scientists bounced neutrons off biologically relevant salts such as sodium chloride, potassium chloride, magnesium chloride, calcium chloride and others, each dissolved in water. How the neutrons scattered will provide information on how ions are distributed throughout the water. What Hammer learns could help illuminate how different concentrations of ions in diverse locations can affect water’s structure and behavior – information useful in biology and other arenas.
In his 2016 Los Alamos National Laboratory practicum, Hammer focused less on using molecular dynamics codes and more on programming them. He worked on an interface between two lab-developed codes, coupled wavepackets and non-adiabatic molecular dynamics. Combining them will let researchers make more precise calculations of interactions between electrons and between nuclei in electronically excited states, an important topic in photochemistry research.
Hammer’s advisor, Sergei Tretiak, and his colleagues have applied the coupled wavepackets technique to only model systems. With software that allows the codes to communicate, they could try it on actual molecular systems.
The practicum “was a very useful experience for looking under the hood at how these programs are built, the rationale behind how they’re designed,” even if there wasn’t time to complete the task, Hammer says. What he’s learned has helped his thesis research and given him insights into improving the codes’ efficiency.
Hammer hopes to get a postdoctoral research post after graduating in 2020 and then move into academia to focus on teaching. Exposure to computation as an undergraduate helped spark his interest in the field, and he’d “enjoy being able to help foster that curiosity for the next generation of computational chemists.”