For thousands of years we have relied on experimental correlations between cause and effect to understand the world around us. With the rapid acceleration of computing power in recent years, we are at the cusp of complementing the experimental work with the ability to probe concepts previously inaccessible to physical experiments. Computational approaches often use abstract mathematical expressions to digitally simulate experiments that are too complex or expensive to first perform in the lab. However, it is vital to bridge the understanding from this basic computational level to physical realizations. Enter the interdisciplinary work of the Energy Frontier Research Centers.
Knowing your sample size. The catch with computational research is scaling. Even with the immense processing speed available to the scientific community today, simulations based on the fundamental physical theory without any empirical factors are still limited to a small number of interactions between hundreds atoms together in a solid. To put that in perspective, that's like picking up a handful of sand at the beach and trying to make a conclusion about the entire East Coast. You didn't happen to pick up a seashell, but that doesn't mean there aren't any seashells or crabs or driftwood on the beach. Similarly, you can't just look at the beach to know if there are rocks or pebbles below the surface.
William Green, Combustion Energy Frontier Research Center steering committee member and Massachusetts Institute of Technology chemical engineering professor, says, "Computational research can only tell you information based on what information you tell it. We need to impose restrictions in the form of initial conditions that may or may not perfectly represent the real world. Often, these parameters are not trivially related."
Sometimes these restrictions can lead to computational solutions without much physical grounding. In some solvent chemistry models, for example, intensive computational requirements mean researchers have to ignore the effects of solvent-interface interactions. Unfortunately, this solid-liquid interface is responsible for many optical and electronic properties of the material.
Fortunately, researchers at the Energy Materials Center at Cornell have developed a model for density-functional study of these nanocrystalline surfaces by combining quantum-mechanical descriptions of the solute particles with the continuum representation of the solvent. Remarkably, the model was able to predict experimentally validated solvation energies for different molecules in water. Essentially, the model allows them to accurately predict how the solvent interacts with the solid-particle surface by describing a smooth mathematical transition from solvent to solid.
Their model has had an immediate impact in the field as over 70 users from institutions throughout the world have downloaded their freely available code and documentation in an effort to continually advance the field!
Passing a message. One of the essential outcomes of computational research is developing an understanding of what is important when you move from a cluster of atoms to a larger experimental scale. Take, for example, the fuel combustion in your car. Knowing what intermediate reactions take place at the molecular level is important when trying to replace petroleum-based gasoline with fuels from renewable sources. If you know what chemicals are produced during combustion, you can anticipate compounds and reactions that may be detrimental to the engine's combustion process. Researchers at the Combustion Energy Frontier Research Center are modeling the reaction pathways of iso-butanol, an alternative fuel that could be compatible with existing engines.
Nils Hansen, a chemist with the Combustion Research Facility at Sandia National Laboratories, describes the challenge, "In this fuel combustion, there are over 10,000 reactions leading to over 30,000 unknown parameters that need to be accounted for before factoring in any additional thermodynamic or kinetic relationships. That's simply too many parameters to consider solely for computation programs."
Realistically, experimental techniques to measure chemical compounds in flames are limited to detection of millimeter-size flames, which is roughly a thousand times larger than the length scale of the models. Even with cutting-edge experimental techniques, the researchers can identify only a few hundred of the reaction products as they evolve over time. The key for experimentalists is to know what to look for.
Theory does just that.
Through complex models, quantum chemical computations integrated into kinetic models, the researchers predicted the amount of different reaction products throughout the combustion process within the uncertainty of their experimental techniques!
Finding a needle in a haystack in the middle of a monsoon. When dealing with a complex research problem, often there are too many variables, and interpreting the results becomes far from trivial. Furthermore, there is a limit to what we can detect experimentally. For example, transmission electron microscopes can give us atomic-resolution images of materials, but the imaging is typically limited to high vacuum conditions and very thin specimens, which can change sample properties or limit the type of samples we can measure.
An experimentalist, therefore, needs to choose very carefully what to measure and how to measure it, and having access to computation can make a huge difference.
Let's say we want to find the element that has the highest lithium capacity for use in lithium-ion batteries. If we rely only on experiments, we would need to test all of the existing elements to find the best one. But, theoretical calculation can predict the list of most probable candidates much more easily, saving us a lot of time and resources.
It is obvious that for this process to work, experimentalists and computational scientists should collaborate closely. Being in a collaborative environment such as the Energy Frontier Research Centers is a great advantage.
Aside from predicting new results, computation can be coalesced with experiments as a complementary technique, working together to provide the whole picture. For example, researchers at the Center for Electrical Energy Storage (CEES) have studied single-walled carbon nanotubes (CNTs) for application in lithium-ion batteries.
These nanotubes are rolled-up sheets of carbon atoms. Depending on the angle at which these sheets are rolled up, CNTs can act as a metal or as a semiconductor. Because these tubes are very small--a few thousand times smaller than red blood cells--they have a very high surface-area-to-weight ratio. If added to lithium-ion battery electrodes, these high-surface-area particles will facilitate electrolyte access to the electrode resulting in batteries that can be charged and discharged much faster than existing lithium-ion batteries.
But, metallic- and semiconducting-type CNTs act differently in lithium-ion batteries.
Experimentally, the scientists at CEES have shown that electrodes made of metallic CNTs have much higher capacities than semiconducting ones. On the other hand, if the semiconducting tubes are treated with acid, their capacity improves by tenfold. They explained these results by computational studies of the interaction between lithium ions and metallic and semiconducting CNTs. Theoretical calculations showed that the interaction of lithium ions with a CNT bundle depends on the nanotube-to-nanotube distance; semiconducting CNTs are much more sensitive to this distance than metallic ones.
Hakim Iddir from CEES and Argonne National Laboratory, who performed the computational studies in this work, finds the collaboration between experiments and modeling advantageous. "One of the major benefits I found was the increase of focus and a better understanding of the structure-properties relationships, resulting from the mutual influence and dynamic feedback between theory and experiments," said Iddir.
Conclusion. Thanks to the collaboration in these Energy Frontier Research Centers and others, we have a glimpse into the future of materials research that maximizes efficiency through fundamental computation and cutting-edge experimental work. As computing and characterization techniques continue to improve, so too will the ability to merge these complementary features for a more thorough understanding of the world around us.
