This award is part of the Cyber-enabled Discovery and Innovation CDI-Type I and the recipients are Michael Falk of Johns Hopkins University and Krishnakumar Garikipati of University of Michigan Ann Arbor.
Materials are increasingly being applied in devices that derive their function from small-scale, and even nanometer-scale, features. Examples include advanced battery materials and quantum wires proposed for use in next generation opto-electronics. Features on these scales are subject to profound changes over time due to the thermally induced motion of atoms. The focus of this project is to develop the computational tools that are required for predicting the way these features evolve with time by developing computer programs referred to as "meta-codes." These meta-codes allow researchers to consider these problems by integrating knowledge obtained from the smallest scales, on which electrons control atomic interactions, to the largest scales, on which elastic interactions drive features to form or dissolve with time. The expected result is an automated, top-down working paradigm that naturally integrates tools from quantum mechanics, statistical physics and continuum elasticity.
The intellectual impact comes both from an increased understanding of how to develop meta- codes to undertake multi-scale modeling and from a deeper understanding of the energy storage and electronic materials studied with these novel tools. The broader impact of this work arise from the development of an approach that integrates chemistry, physics and mechanics, and that can be used to refine our understanding of how a wide variety of materials systems evolve when there exist large spatial variations in composition and stress. The project includes active participation of students at the undergraduate, graduate and postdoctoral level, and incorporates a series of outreach activities that leverage those ongoing in the involved institutions.
One of the grand challenges of materials simulation is how to muster all the computational techniques that have been developed to understand how a material behaves including information from the quantum mechanical to the macroscopic scale. This is necessary to understand the shortcomings of current materials and to design new materials that overcome these limitations. One important example is the limitation of materials that are currently used in batteries to store and deliver electricity when and where needed. Understanding the electrochemistry of a battery requires understanding how electrode materials store and release ions. In many batteries this is accomplished using crystalline oxides. But in the process of charging and discharging these materials often undergo phase transitions, altering their crystal structure. This has the benefit that the voltage remains constant during such a process, but since phase transitions also typically involve changes in the shape of the crystal, as one phase is replaced by another stresses can build up in the crystal. This can limit our ability to repeatedly charge and discharge the battery, leading to breakdown and failure. In this research project simulation methods were deployed to characterize two oxide electrode materials, a lithium vanadium oxide and a lithium titanium oxide, used in lithium ion batteries. Quantum mechanical methods were used to characterize the structural states in which the material resides as it incorporates different amounts of lithium during charging and discharging. Since it is not possible to calculate every possible atomic configuration, cluster expansions were constructed that predict the energies of uncalculated states from the database of calculated states. These cluster expansions were then used in statistical mechanics calculations to determine which phases are stable at different compositions as well as the mobility of the lithium ions and the interfaces between phases. This information was, in turn, fed to larger scale simulations that are able to predict the stresses that arise during the charge/discharge cycle. In the course of this investigation we identified the range of composition over which the lithium vanadium oxide material would undergo a phase transition, and the sequence of structures likely to define the pathway by which lithium incorporates itself into the crystal. The energetic barriers to lithium motion within the crystal were also measured in the calculations. However, the complications arising due to correlated electron effects in the material resulting from the presence of the transition metal vanadium prevented us from developing a local cluster expansion to fully characterize the diffusion process. In the lithium titanium oxide material we were able to fully characterize the thermodynamics of the phase transition and the kinetics of lithium transport. This required the development of unique algorithms to handle the unstable configurations of atoms that arise, particularly near the phase boundary. Using this method the orientation dependent interfacial energy and interfacial mobility were extracted. These were communicated to colleagues who undertook a continuum mechanics analysis of the stresses that arise during the charge/discharge process. The intellectual merit of this project arose from a deeper understanding of the materials studied and the way their structure and phase behavior limit cylcleability. The broader impacts came from the development of improved methods for integrating computational tools for modeling kinetic processes in important materials including, but not limited to, energy storage applications.
