Pumping fuel made from sunlight into your car, working on an unhackable, lightning-fast computer, installing solar panels that cost pennies and store twice the energy that today’s panels do.
All of these futuristic ideas might become widely available someday if scientists can find the best materials with the most functional properties for each particular task. That’s no easy feat since materials scientists have to consider all kinds of physical and chemical properties across a range of options — metals, polymers, ceramics, gemstones, composites — in order to determine which might be the most optimal one for a specific application.
That’s where scientists at the Midwest Center for Computational Materials (MICCoM), come in. Created in 2015 and headquartered at Argonne National Laboratory, MICCoM is one of five centers across the country that develop open-source advanced software tools to help the scientific community model, simulate, and predict the fundamental properties and behavior of materials in order to create new, cutting-edge technologies. Lately, the team of nine senior scientists, along with about two dozen postdoctoral researchers, have been racking up awards: MICCoM Director Giulia Galli, for instance, just received the prestigious 2022 Aneesur Rahman Prize for Computational Physics.
“This award is a great honor and tells me that the work we do at MICCoM is truly appreciated by our peers. I know my MICCoM colleagues feel similarly about their recent accolades,” said Galli, Liew Family Professor of Molecular Engineering at the Pritzker School of Molecular Engineering (PME) and the Department of Chemistry at the University of Chicago and senior scientist at Argonne National Laboratory.
“In materials science, there is a critical need to build codes and algorithms that then can be applied to important technological and societal problems, in collaboration with experimentalists,” Galli said.
For instance, computational predictions and analysis may help scientists figure out the best photoelectrode (light-absorbing) material to interface with a catalyst and cause water to split into hydrogen and oxygen, thus generating a clean fuel. Scientists want to use sunlight to power the water splitting reaction, and contribute to designing a process known as artificial photosynthesis.
“The ultimate goal,” Galli said, "is to recreate with specific materials what the sun does with leaves, using sunlight to produce fuel.”
Studying the building blocks
The computational simulations that MICCoM researchers generate are based on quantum mechanical theories and serve as a first step in the process of materials development. Using codes that can sift through the properties of hundreds of thousands of materials can be much more efficient than testing each material separately, said Nikita Onizhuk, who was one of four people in the world to receive a 2021 Google PhD Fellowship in quantum computing. The postdoctoral researcher in the Galli group is studying the spin coherent properties of defects in semiconductor materials; spin defects can hold quantum information in a form that physicists call quantum bits, or qubits.
“For the specific problems that I'm focusing on, to run many experiments in the lab would take days and weeks,” he said. “I can run simulations in seconds using a code to compute spin-coherence times that we have developed within MICCoM”
He Ma, another recent award winner, also studied complex defects to realize qubits. Ma worked on numerous MICCoM projects such as developing general methods/algorithms for quantum mechanical simulation of molecules and materials. He recently earned a doctorate in chemistry from the University of Chicago and received the 2021 Elizabeth R. Norton Prize for Excellence in Research in Chemistry.
Some MICCoM scientists such as Ma, Onizhuk and others focus on developing codes to model and understand heterogenous materials — blending materials and exploiting their defects or impurities, to help uncover new properties and functionalities.
To get to that point, MICCoM researchers first write codes that can analyze materials’ properties at the atomistic level, trying to draw structure-function relationships. For instance, Marco Govoni, a MICCoM scientist, develops codes to analyze the properties of electronic excited states. His work should aid in the understanding of spin defects in solid-state materials, complementing Onizhuk’s work. In 2020, Govoni won a highly competitive Department of Energy (DOE) Early Career Research award to further broaden the research conducted within MICCoM.
“Studying spin defects offers an incredible opportunity to solve a materials science challenge,” he said, “and to explore new and exciting paths for computing, communication, and sensing.”
Back and forth collaboration
After Govoni, Galli and others have conducted their computational study, they join forces with other researchers to test their theory and simulations with an experiment on a real material. Experimentalists then provide feedback that, in turn, may help improve the theory on which the codes are built or speed up the codes themselves.
“The ideal feedback loop is the following: We do a simulation, they do an experiment, we do a comparison and then we go back and forth until we solve the problem,” said Galli.
For instance, MICCoM researchers recently published a study describing how computer simulations can predict the atomic-scale structure at the interface between aluminum oxide and water.
“These interfaces are ubiquitous in materials, and those between oxides and water are key to many energy applications,” says Francois Gygi, MICCoM investigator, professor of computer science at UC Davis and main author of the center’s code for simulations of interfacial and other phenomena at finite temperature.
The interface simulations were validated by doing an experiment, using high-resolution X-ray reflectivity (XR) measurements to shed light on an actual aluminum oxide-water interface. The feedback loop between theory and experiment shed light on how to improve the theory and also on the physical properties of the system. For instance, the researchers found the reflectivity is sensitive not only to the atomic positions, but also to the electron distribution surrounding each atom in subtle and more complex ways than previously thought.
These insights will prove beneficial to future experiments on oxide/liquid interfaces in part because MICCoM scientists make all of their codes publicly available. The DOE, which funds MICCoM, requires that they do, and MICCoM’s researchers recognize the importance of sharing their work so that no scientist has to reinvent the wheel each time they are experimenting with new materials.
“Everything we write and use is open source,” said Elizabeth Lee, a MICCoM postdoctoral researcher who, among other things, write codes to understand how heterogenous structures in materials can cause one form of energy to transform into another one. Lee studies processes that are key to creating renewable forms of energy; photovoltaic panels on solar arrays, for instance, transform sunlight into electricity.
“Any researcher can download any MICCoM code and use it,” she said. “Scientific transparency and usability are really important.”
Looking to the future
Although MICCoM has been around for six years, the work being done there is still in its infancy. The scientific directors would like to see an increased focus on computational materials in the scientific community and the validation of results produced by open-source codes. A challenge that MICCoM researchers and their colleagues face is the need to continually reformulate codes because computer architectures are changing — which in principle is a good thing for materials scientists because codes may become faster, but this also requires constant, non-trivial changes in codes and algorithms.
“Because the programs that we have written over time are hundreds of thousands of lines of computer code, it’s really difficult to rewrite them,” said MICCoM Deputy Director Juan de Pablo, Liew Family Professor of Molecular Engineering at PME and a senior scientist at Argonne National Laboratory. “But we have to do it because the new generations of machines are hundreds of times faster. And in this type of research, we're always limited by the size of the calculations that we can do.”
Materials scientists also need to pay attention to the evolution of machine learning and artificial intelligence, something MICCoM researchers are also thinking about.
For example, Professor Jonathan Whitmer, from Notre Dame University, who is part of the MICCoM leadership, develops methods based on machine learning to increase the efficiency of dynamical simulations of rare events. In addition, MICCoM scientists recently employed machine learning in a way that led them to understand in a more efficient manner how atoms and molecules within a material interact with light. Some substances absorb light while others reflect it; analyzing light interaction with matter is often a challenging task. MICCoM scientists figured out how to simplify such computer simulations by re-using information during their simulations at finite temperatures.
Recycling calculations to shorten computing time seems like a fitting goal to pursue for a center that’s dedicated to finding ways to make energy use cleaner and more efficient.
Galli agreed. “Our mission at MICCoM is to accelerate the discovery of new, useful materials,” she said. “Every new discovery gets us closer to that futuristic world where energy is clean and quantum technologies are employed in our everyday life."