Calculating Properties of Materials from First Principles—A Tool for Faster Advanced Materials Development
Once the domain of guesswork and intuition, the field of developing new materials for advanced batteries and other applications is taking a turn towards a more systematic and predictive approach. Predicting the properties of new materials from "first principles" has become a scientific reality, thanks to the growth in computing power, a deeper understanding of how materials work, and databases of materials properties.
This will mean faster development of materials for high-energy batteries for electric vehicle applications, as well as better materials for many other applications, such as fuel cells and solar panels, high-strength materials, and catalysts.
Kristin Persson, a scientist in the Environmental Energy Technologies Division of Lawrence Berkeley National Laboratory, is moving forward on several fronts to predict the behavior of materials from first principles. In 2011, she and Gerbrand Ceder at the Massachusetts Institute of Technology (MIT) launched the Materials Project—a materials design gateway that allows users to browse existing materials (more than 20,000 currently) and their properties, modify them, and predict new materials using data-mining algorithms. [Read more about the project here.]
The U.S. Department of Energy (DOE) is supporting the hunt for new materials through computation. Earlier this year, it announced that it would fund a new DOE Center for Functional Electronic Materials Design at $11 million over five years. Persson will be the Center's director, and Ceder, its associate director. The Center's purpose will be to conduct large-scale data generation, data-mining, and benchmarking for new materials. Scientists at Berkeley Lab; the National Energy Research Scientific Computing Center (NERSC); the University of California (UC), Berkeley; MIT; Duke; UC San Diego; and elsewhere will participate in the research.
Computation Can Accelerate Materials Innovation
Using the computational approach, researchers can apply first-principle calculations and great computing power to many materials at a time, to search out one whose properties may meet the needs of the application they are developing. "It takes 15 to 18 years to develop a material the traditional way, from the laboratory to commercial application," Persson says. "The lack of organized, comprehensive information about materials can cause delay during the scale-up to manufacturing."
As a result, materials innovation has been dependent on single investigators and intuition. Now, with high throughput first principles calculations on the properties of many materials, the Materials Project team has developed an organized, searchable database of materials properties. "This is assisting researchers in predicting the properties of new compounds, and it creates a new materials design environment," she adds.
Using the Materials Project database and tools, a researcher can ask a question such as, "are there any fluoride materials out there that would work as a cathode in a lithium ion battery?" The Materials Project database could screen "nature and beyond," as Persson puts it, for materials that have the desirable properties such as voltage, capacity, diffusivity, and stability, among other things.
As of July 2012, the Materials Project includes more than 20,000 compounds, as well as a materials explorer, a reaction calculator, a phase diagram application, a lithium battery explorer, a crystal toolkit, and a structure predictor. It has more than 2,500 registered users, has predicted more than 8,000 new structures using the structure predictor, and has generated more than 10,000 phase diagrams for its users.
Applying First Principle Calculations to Basic Research
Persson also uses these methods to solve basic scientific problems, focusing on one material at a time. The aim of the research is to develop better materials for higher-power or more-stable advanced batteries. For example, lithium-ion batteries, destined for such applications as plug-in hybrid and all-electric vehicles, need to have higher energy densities for greater range, lower cost, longer lifetimes, and a higher safety factor before they will be economical to use in vehicle applications.
"When I started, you couldn't do more than calculate very basic properties of materials," she says. "In the last 15 years, there has been an explosion of computing power, and advances in analytical methods. You can now submit a quantum mechanical calculation to a supercomputer with much less tuning and manual labor."
Persson has used first-principle calculations to answers questions about why certain experimental battery materials thought to have the ability to solve some of these problems have not yet lived up to their potential.
Graphitic carbons, for example, form a class of carbons that are most commonly found as the anode (the negative electrode) in a lithium-ion (Li-ion) battery. The material has been used since the commercialization of the first Li-ion battery by SONY, and its properties are thought to be relatively well understood. For example, it is well-known that the rate capability of graphitic carbons deteriorate significantly at lower temperatures, causing the battery to degrade and lose capacity to store energy. Loss of rate capability is usually tied to lowered lithium diffusivity (its ability to move through graphite), but measurements of lithium's diffusivity varies widely from one experiment to the next, making the root cause of the performance loss difficult to identify.
Calculating Graphite's Properties
In 2010, Persson and her colleagues Robert Kostecki at Berkeley Lab and Gerbrand Ceder of MIT teamed up to elucidate the problem. Persson and Ceder calculated the inherent diffusivity of lithium ions in carbon using first principles and found that the lithium's diffusivity was extremely high, suggesting that the diffusivity of the graphite was not the problem.
"We can take the pure graphite and calculate how fast the lithium moves around in there. This is inherent diffusivity. We were able to show that the graphite is really fast, she says. "This tells us that the graphite in itself is an extremely fast material." In fact, lithium ions could enter and leave a micron-sized graphite particle in less than 0.2 microseconds.
At a molecular level, graphite is made of carbon atoms in stacked sheets (see figure). The solution is to engineer the graphite material so that lithium ions can travel parallel to the sheets of graphite, instead of at grain boundaries.
With the proper materials engineering, it should be possible for the graphite to live up to its potential. "We would like to engineer the material so that the Li can travel along the fastest route," she says.
In parallel, a group led by Robert Kostecki at Berkeley Lab was able to measure the speed of lithium ions diffusing parallel to graphite's sheets of carbon, and perpendicular to it. The experimental results independently demonstrated that lithium ions diffuse rapidly through the graphite when traveling in between the graphene sheets, and slowly when traveling along the boundaries of graphite domains.
This concerted work pointed to a way of engineering a graphite anode to exploit the fastest pathway of diffusivity for lithium ions: creating graphite particles which are aligned radially, so that the planes of graphite's molecular structure are parallel to each other.
Other Research Directions
Persson has studied other battery materials as well, looking for such properties such as faster ionic diffusivity, and greater stability and energy storage capacity. For example, replacing five to ten percent of the cobalt (another candidate for cathodes in lithium batteries) in the layered materials with aluminum leads to a material with higher voltage, greater thermal stability, and lower electronic conductivity, yet it has a higher rate capability. Why? Persson and her colleagues found, using first principle calculations, that the aluminum lowers barriers to the migration of lithium ions when concentrations of lithium are low. Because cobalt is an expensive material, substituting cheaper aluminum would not only improve performance, but also lower material costs.
Persson's research continues to address materials for improved lithium ion batteries, even as she maps out other areas for future research. "We would love to look more at the surface properties of materials," she says. "First principle calculations can give you a very good handle on their bulk properties, which is the starting point. But surface properties are more difficult because they are more amorphous."
Persson, Kristin et al. "Thermodynamic and kinetic properties of the Li-graphite system from first-principles calculations." Physical Review B 82.12 (2010): c2010 The American Physical Society.
Persson, Kristin et al. "Lithium Diffusion in Graphitic Carbon." Journal of Physical Chemistry Letters. 1 (8) 11767-1180 (2010).
This research is funded by the U.S. Department of Energy's Office of Energy Efficiency and Renewable Energy.