Power Theory

Engineer Chris Wolverton uses computation to improve materials for energy storage and conversion


A network graph — called a “minimum spanning tree” — shows thousands of predicted table compounds from the Open Quantum Materials Database when it launched in 2014. Since this image was completed, the number of compounds predicted has increased to more than 55,000. Chris Wolverton’s research group created the database to take some of the guesswork out of designing new materials.

Machines are changing how engineers think about materials. Using powerful computation, Chris Wolverton, the Jerome B. Cohen Professor at the McCormick School of Engineering, can achieve in hours or days what once may have been impossible in a physical lab. At Northwestern, Wolverton leads a research group that studies the application of computational tools to predict and optimize the properties of materials and help solve energy storage and conversion challenges. Helix spoke to the International Institute for Nanotechnology-affiliated researcher about his materials modelling, how it is used in research, and its larger implications for spurring future innovation.

Helix: What is computational materials science?
CW: We don’t have a laboratory. The idea is that these computational techniques eventually connect a material — a chemical composition and maybe an atomic or crystal structure — with the material’s properties. So given a material (real or hypothetical), a chemical composition, and an arrangement of atoms, we can calculate what the properties of that material should be.

Several years ago we wondered if we could just do calculations for all known inorganic solids and store them in a large database. Depending on your definition, there are between 50,000 and 150,000 known, inorganic solids; so, the question is, could you do that many fundamental physics-based calculations, quantum mechanical calculations? The answer is yes you can. But it takes a long time. We started that process years ago and now have this large-scale archive that we call the Open Quantum Materials Database (OQMD). Because the calculations are fundamental, and thus highly predictive, this means we can calculate the properties of a material and it doesn’t really matter whether or not this material has been synthesized inside a laboratory.

Helix: What kinds of material are you looking for?
CW: We are quite interested in materials problems that have to do with energy and sustainability. A lot of my group’s work tries to find materials for next-generation batteries. Thermoelectric materials are interesting because they can turn heat into electricity or vice versa. There is a lot of interesting work in this area to try and use these materials to capture waste heat. If we could capture a fraction of the world’s waste heat and use it to produce electricity, this would be a fantastic improvement in energy efficiency.

Helix: Can you discuss your collaboration with IIN Director Chad Mirkin?
CW: In Mirkin's group they have fantastic new tools for growing huge libraries of nanoparticles, and are especially interested in what happens when you grow these nanoparticles consisting of a large number of different elements. So not just a single element, not just silver, or platinum or whatever. But what happens when you make a nanoparticle with platinum and silver and copper and cobalt all together? What he’s found is that these nanoparticles tend to segregate into different phases or compounds. Looking in our database, to a large extent we could accurately predict which phases should form in these nanoparticles. At this point, we’ve mostly done these calculations retroactively, where Mirkin’s group grows a particle of a given composition and says “this is what we find.” Then we go to the database and say: “yeah, we could have told you that.”

The idea is that now you have a tool that is actually predictive. So you could start scanning through this database, through the trillions of possible combinations of these elements, and look for the combinations that are likely to form phases that you might find desirable.

This collaboration between Mirkin and myself is going to be a very powerful combination — computing, then experimenting — because my lab can do these computations in a “high-throughput” manner, making tens or even hundreds of thousands of calculations. The Mirkin group can analogously grow a large number — thousands or millions of nanoparticles. Combining these two methods is potentially a game-changer and could dramatically accelerate the rate at which we discover new materials.

Helix: Given this kind of collaboration, what might the future hold?
CW: Different parts of the story are easier or harder to deal with. The OQMD already exists. It’s free, it’s online, it’s open. You could go there now type in six-component chemical compositions and learn within seconds what phases should form. Other parts of the problem are ongoing. Calculating the interfacial properties between phases and the ultimate arrangement of phases in a particle, that’s not done. We don’t have a giant database of interfacial energies yet, so we can’t just look up the answer. We have to do the calculations and those could take a while.

We’ve developed this new computational framework for predicting nanoparticles, now we need data. Once we have data, then we can do these calculations relatively quickly. Part of that problem we can solve right now because of the work with Mirkin. Now it’s a matter of grinding through and doing the calculations. That will take a while, but it will be well worth it.



Add new comment

This question is for testing whether or not you are a human visitor and to prevent automated spam submissions.