Hello,
A yearly allocation for the LCRC cluster has been requested with the
following updated information:
Submitter/PI: Zhenpeng Yao
Project Name: CEES-II-Wolverton
Division: Computing, Environment, and Life Sciences
Project title: DFT Computation of Hybrid and Li-ion Battery Materials
Associated funding: CEES-II EFRC
Other Systems:
Science: Lithium ion battery (LIB) has been the most prominent electrochemical energy storage technology over the past decades and enabled the wireless evolution of portable electronic devices. Improving the power density, safety, and cycle life of batteries is now one of the key challenges in increasing the use of both electric vehicles and renewable energy sources, and therefore in curbing greenhouse gas emissions. Therefore, the development of new high energy electrode materials for rechargeable lithium ion batteries is becoming more essential. State-of the-art electrodes like LiCoO2 and LiMn2O4 function via the intercalation mechanism during Li accommodation and extraction with the preserved crystal structure, accompanied by redox reactions of transition metal cations. The specific capacity of the electrode is then limited by the safe amount of Li removal without impairing the structure backbone and the number of electrons per transition metal cation that can participate
in the redox reaction. Electrodes work with conversion mechanism are then suggested which can achieve a significantly larger capacity by overcoming the inherent structural limitation. Yet the detailed phase evolution and electrochemistry during the cycling of conversion type electrodes are usually unclear. Meanwhile, the exclusive transition metal redox is challenged by the recent discovery of anionic redox reactivity in Li-excess materials, which can be taken advantage of to greatly boost the capacity of anionic redox active electrodes. Understanding the origin of the oxygen redox, therefore, has become essential. Furthermore, solid state batteries lithium batteries, where the liquid electrolyte in standard lithium-ion batteries is replaced by a solid conductor, offer a path forward in improving their safety. Understanding the role of interfaces, both grain boundaries and, perhaps more importantly, heterointerfaces between electrodes and electrolytes, is essential in study
ing and engineering solid-state batteries. Electrode/electrolyte interfaces pose extremely different chemical environments for lithium, making drastic atomic reconstruction at even stable interfaces likely. Density functional theory (DFT) has been widely used as effective tools to study battery materials to explore new, high-performance electrode materials, understand in detail the underlying mechanisms during electrochemical reactions. Using LCRC, we were able to provide new insights to the conversion mechanism of transition metal oxides and sulfides, the origin of higher capacities through anionic redox activities, as well as explore the stability and mass transport through the interfaces. We wish to continue using such high-performance computing resource to understand more fundamentals of the battery dynamics and design advanced electrodes for the next generation lithium ion batteries.
Project description: Using computational resources from LCRC, we succeeded to reveal the conversion mechanisms of transition metal oxides and sulfides such as Co3O4, (Cu,Co)3O4, NiO, and CuS by the nonequilibrium intermediate phase search method as developed. Based on what we have learned from these systems, we plan to continue our research on the conversion type electrodes and apply our method to other conversion material systems in order to obtain a more general understanding of the detailed mechanism for conversion materials during charging and discharging. Our study could help future experiments to overcome the current limitations of the conversion-type electrode materials and promote the development of more advanced LIBs.
Meanwhile, with the assistance of LCRC, we were able to clarify the interplay between cationic and anionic redox during the delithiation of Li5FeO4 and Li4Mn2O5 based electrode materials. Considering the anionic redox activity is of great interest to boost the energy density of electrodes, it is essential to explore the anionic chemistry in more lithium-rich transition metal oxides to build a thorough predictive theory. This extensive new understanding of oxygen redox in different coordination environments will enable exploration of new pathways to next-generation, high-energy cathode materials based on simultaneous anionic and cationic redox chemistry.
Nascent efforts in the Wolverton group have begun in adapting the Minima Hopping Method (MHM), used in crystal structure prediction, to study interfacial reconstruction. Crystal structure prediction presents a notoriously difficult problem: find the minimum energy arrangement of a set of atoms, with no other information. MHM uses consecutive, short molecular dynamics steps to overcome kinetic barriers between phases, and the search is biased towards previously unexplored minima by the use of atomic fingerprinting. Solving for crystalline interfaces presents as a further constraint the orientations of the two crystals forming the interface. Thus, we bias the search away from the global minima, and instead towards the local minima which conserve these orientations, by the use of an atomic disorder parameter. We are currently using interfacial minima hopping to study the canonical example of grain boundaries in SrTiO3. From this point, we intend to study grain boundary reconstru
ction in two electrode materials: LiCoO2 and LiMn2O4, and heterointerface reconstruction in Li3N/Li interfaces. There are multiple steps in this process – a) Use Minima Hopping to find low energy interfacial structures for multiple orientations, using empirical potentials. b) Relax these structures using Density Functional Theory. c) Calculate barriers to lithium migration across interfaces using Nudged Elastic Band.
Industry partnership:
Project URL:
Current FY Hours Used: undetermined amount
New FY Requested allocation: 4085504
Q1: 1021376
Q2: 1021376
Q3: 1021376
Q4: 1021376
Justification: The intermediate phase search method used for studying the conversion reactions is effective yet computational resource consuming. For each material system, the method is initiated with total energy sampling of the whole set of geometrically different configurations (~100,000) using coarse setting DFT calculations. For each configuration, very limited CPU hours around 4 are needed. Afterward, we planned to work on a series of stoichiometries (~16) along the lithiation/delithiation process will be considered. For each of these stoichiometries, we rank all the configurations with their total energies and select lowest ones (~5) to be relaxed in DFT. The relaxation for each of them usually needs 72 cores for 48 hours on average. We will choose several typical conversion materials (~2) to study which will spend: 2*(100000*4+16*5*72*48) = ~ 1,352,960 CPU hours in total.
The anionic redox chemistry related calculations require the usage of hybrid functional because of the need to catch the oxygen electron configuration accurately which are usually pretty expensive. For each system, around 10 stoichiometries with 6 configurations are needed and for each of them, 144 cores for 36 hours are necessary. Kinetic simulations like the nudged elastic band theory calculation will be necessary for study the possible transition metal migrations. For each system, migration path number around 5 is reasonable with each path requires 144 core for 72 hours. The total CPU hours needed are around: 2*(10*144*36*6+5*144*72) = 72,5760 hours.
The first step of finding minima using pair potentials is very low cost. Assuming we are interested in the 3 systems described above, 8 crystal orientation pairs, 4 stoichiometries, running on a single core for a week, this step consumes about 16,128 hours. Since these structures are found using empirical potentials, a selection of them must be relaxed using DFT. Assuming for our 3 systems, we choose the top 4 structures, for 8 orientations, 4 stoichiometries, using 96 cores for 36 hours, this step consumes 1,327,104 hours. Finally, to calculate ionic migration barriers using nudged elastic band, we would consider 3 systems, 1 structure, 4 stoichiometries, and 4 distinct migration paths, using 96 cores for 144 hours, totaling 663,552 hours. The interface-related calculations need an amount of 2,006,784 CPU hours in the next fiscal year.
To summarize, we request a total amount of 4,085,504 CPU hours in the next fiscal year (~1,021,376/Qtr.).
Storage requirements:
Thank You,
The LCRC Accounts System
