Hello, A yearly allocation for the LCRC cluster has been requested with the following updated information: Submitter/PI: Marton Voros Project Name: chargetransport Division: MSD Project title: Charge transport in energy materials from first principles Associated funding: LDRD -- Charge Transport in Nanostructured Materials from ab initio Simulations Other Systems: NERSC, 1.000.000 hours (project on nanoparticle solar cells) Science: The ultimate goal of this project is to understand and engineer charge transfer processes in materials relevant for energy storage and conversion. Following the success of this project in the past years on understanding transport in lithium manganese oxide (LMO) based batteries (one paper in revision in Nature and another being close to submission), this year we will extend our investigation to other cathode materials relevant for lithium-ion batteries. In particular, using our recently developed constrained density functional theory code (J. Chem. Theory Comput. 13, 2581 (2017)), we will model charge transport in layered lithium cobalt oxide (LCO), a paradigmatic material for Li-ion batteries. In the case of favorable results, and if time permits, we will also investigate alloyed lithium transition metal oxides. Since this effect has been ignored in all computational studies to date, we believe our investigation could open new doors in the design of better cathodes . In particular, we will be able to investigate how lithium and electron transfer are coupled and how the movement of the electronic charge affects ion transport. Such electron-ion coupled processes (e.g. proton-coupled electron transfer) are known to be imperative to understand the catalytic behavior of organic or organometallic materials but this effect has been largely ignored in the lithium-ion battery community. In parallel, we will start modeling photo-induced processes using time-dependent density functional theory. In our paper under revision, we have experimentally show that LMO is able to absorb visible light, which in turn gave rise to enhanced charge and ion transport. However, it is not clear what the electronic structure origin of the absorption peaks is so the mechanistic reasons for the enhanced charge and ion transport was only speculated. At the end of this allocation year, we will do the initial steps towards understanding this process and we will start modeling the photo-absorption behavior of LMO and LCO. Project description: We will be using first principles methods based on density functional theory. In particular, for simulating oxide cathode materials we will be using the plane wave code Quantum Espresso (QE). QE can make use of efficient linear algebra (Lapack, BLAS, Scalapack, BLACS) and FFT libraries, if they are available. QE was shown to efficiently scale to a few hundreds of processors in the case of non-hybrid functional calculations. Hybrid functional calculations can scale up to a few thousands of cores, especially because of recent methodological and algorithmic development (J. Chem. Theory Comput., 2012, 2242 (2016)). NERSC staff has restructured the algorithm behind hybrid functionals in Quantum-Espresso and has shown that this method is capable of making use of the new Cori KNL nodes due to its hierarchical, multi-level parallelization strategy that involves using several layers of MPI and OpenMP parallelization (e.g. http://www.nersc.gov/users/computational- systems/cori/application-porting-and-performance/application-case-studies/quantum-espresso-exact-exchange-case-study/). We expect the same or similar benefits when using the KNL nodes of the new Bebop cluster for the hybrid functional calculations, however, we know from our experience that Quantum-Espresso also runs fine the Blues cluster and we expect it can also nicely run on the broadwell nodes. Indeed, our initial tests on small systems during the early user access program showed good performance. We will plan to use 60% of our allocation on non-hybrid functional calculations and we will use the remaining 40% of the time to do more expensive (but also more accurate) hybrid functional calculations on selected systems. A typical non-hybrid functional calculation will take several hours of wall clock time (up to tens of hours) on a few hundred processors. There is an efficient restart mechanism in QE allowing us to run smaller jobs. This is useful if the wall-clock time limit does not allow to finish big calculations in one step. To simulate charge transfer processes, we will first identify sites that localize charges in the form of polaron-like carriers. Then, we will use nudged-elastic band and our own implementation of constrained DFT (CDFT) to simulate charge transfer between the sites. This will allow us to compute the parameters entering semi-empirical charge transport models, such as Marcus-theory. Our own implementation of CDFT is integrated into a private version of the most recent release of Quantum-Espresso and it inherits its performance, although typical CDFT calculations are about one order of magnitude more expensive since they require a double self-consistent cycle instead of the regular single self-consistent cycles. For the optical calculations, we will use time-dependent density functional theory as implemented in the Quantum-Espresso code and first we will be looking lithium cobalt oxide which does not require hybrid functionals of Hubbard corrections to get the correct ground state. Thus, we will expect we will be able to use a low level of theory, which is necessary to make the calculations doable. For the charge transport subproject, we expect to run about 100 regular non-hybrid DFT calculations on LCO, each on ~256 cores (assuming KNL nodes) that would last several hours, possibly in the order of 4-5 hours. This would allow us to have a quick estimate for most relevant charge transfer pathways. Then, after picking the most important configurations (<20), we will run CDFT calculations that would last 10-24 hours on 256 cores. We will then repeat a few calculations using hybrid functionals on ~1024 cores. We estimate we will run about 10 calculations that would last 24 hours. Optical calculations using time-dependent density functional will be run on the perfect bulk material and we estimate its cost about 100 times of a regular non-hybrid ground state. All in all, we request 700.000 core hours for the entire year. Since more expensive calculations will require performing less expensive calculations first, we request proportionally more time for the third and fourth qua rter. Finally, for the reasons discussed above, we believe that it is only our hybrid functional calculations that can really benefit from using the KNL nodes, however, we can definitely use the broadwell nodes efficiently and we had a successful experience in using Quantum-Espresso of Blues. Industry partnership: Project URL: Current FY Hours Used: undetermined amount New FY Requested allocation: 700000 Q1: 125000 Q2: 125000 Q3: 225000 Q4: 225000 Justification: This is detailed in the Project Description: hybrid functional calculations have been shown by NERSC staff to scale up to thousands of cores due to recent numerical and algorithmic advances (J. Chem. Theory Comput., 2012, 2242 (2016), http://www.nersc.gov/users/computational-systems/cori/application-porting-an...), we thus expect that calculations using 40% of our requested allocation can scale up to thousands of cores. Since this part of Quantum-Espresso makes use of multi-level, hierarchical parallelization, if necessary, we will perform benchmarks to find the ideal parallelization layout. Storage requirements: 2 TB. We will need to, at least temporarily, save wave functions and charge densities to enable efficient restarts of QE calculations. Since these quantities are saved on grids and can be several tens of GBs in size, we estimate we will need about 2 TB of storage. Thank You, The LCRC Accounts System