Hello, A new project on the LCRC cluster has been requested. Please forward the information on to the LCRC Allocation sub-committee. Applicant's name: John J. Low Applicant's institution: ANL Applicant's division: EGS, CLS, LCF, APS, MSD Project Name: perfect Project title: The Perfect Thermodynamics of Imperfect Materials Associated funding: LDRD Other Systems: ALCF/INCITE Science: We focus on high-k dielectric materials for complementary metal-oxide-semiconductor (CMOS), which are of particular importance for creating Dynamic Memory Allocation (DRAM) devices. Many CMOS properties strongly depend on material defects such as vacancies, interstitials, defect clusters that occur during synthesis, and thermal treatment. Inclusion of other chemical elements (e.g., dopants) in CMOS can significantly change physical properties such as thermal conductivity, electrical conductivity, and magnetism. Far from representing a drawback, the understanding imperfections, including defects and dopant, leads to prediction and control of properties and the design of post-CMOS materials for the next generation of electronic devices. There is, however, a challenge: the thermodynamic stability of CMOS phases, a key element for any application, is difficult to evaluate, especially at high temperatures, close to the phase transformations or the melting point. In this project, we develop the theoretical, experimental, and computational knowledge and tools for predicting the thermodynamic and kinetic stability/metastability of defected CMOS phases at temperatures ranging from 0K to melting. As a proof of principle, we will study the high k dielectric in complementary metal–oxide–semiconductor (CMOS), using as a prototype hafnia, HfO2. Hafnia has unusual electronic and magnetic properties that make it attractive to both science and technology. One of the most important applications is based on hafnia’s high-k dielectric properties, which are of particular importance for creating Dynamic Memory Allocation (DRAM) devices. In 2007, Intel decided to replace silicon oxide with hafnia as gate insulator for field-effect transistors. Current and future technology (sub 14nm) are using “hafnia based” materials (including additional elements). Many CMOS properties strongly depend on materials defects such as vacancies, interstitials, defect clusters (Figure 1), and how the material was synthesized and pretreated. Inclusion of other chemical elements (dopants) in CMOS can significantly change physical properties such as thermal conductivity, electrical conductiv ity, and magnetism. Far from representing a drawback in the integration of CMOS in the next generation of electronic devices, understanding imperfections (defects and dopants) can lead to prediction and control of important magnetic and electronic properties. Project description: Our initial studies will be focused on defects in Monoclinic HfO2. This will allow us to compare our results to the available experimental and theoretical data. We will focus on oxygen vacancies, hydrogen interstitials, nitrogen dopants and grain boundaries. These are common defects in HfO2 high K-dielectrics. Additional work will done modeling the crystallization of amorphous hafnia. Since we will need to study amorphous high temperature phases, the standard method of estimating entropy from phonon bands will not be adequate. We propose to use Bennett methods (Bennett, J. Comp. Phys. 22(1976)245) to estimate thermodynamics of defect formation. This approach involves computing the overlap of two molecular dynamics trajectories to evaluate the free energy. This approach will correctly include the large amplitude motion important at high temperatures near defects and phase transitions. Published reactive force-fields (Phillipot, Phys. Rev. B 81(2010)125328) will be use to establish the length-scale and time-scale required to get converged thermodynamics with Bennett Methods. We estimate that the preliminary force-field studies will require 100,000 core-hours. The SCAN metafunctional [Sun, et al., Nature Chem. 8(2016)831 and ACBN0 pseudo-hybrid functional [Nardelli, et al. Phys. Rev. B 93(2016)0351104] will be evaluated in this work. Both of these recently develop functionals have been shown to yield accurate bond gaps, lattice constants and bulk moduli. All of which is required for accurate predictions of energetics for band gap formation. SCAN has been implemented in VASP and libxc (Abinit and CP2K). ACBN0 has been implemented in Quantum Espresso and will be obtained through a collaboration with Nardelli (Unv. N. Texas). The SCAN functionals are about 30% more expensive than standard LDA and GGA functionals, over an order of magnitude less computationally intense than hybrid functionals and correctly predict that Mott insulators are insulators. ACNB0 is a DFT+U method with no empirical parameters. These functionals will enable the accurate modeling for defects with ab initio molecular dynamics from first principles. We pla n to calculate lattice constants, bulk module, band gaps and energies for defect formation at 0K to validate the SCAN and ACBN0 functionals. This work will be done with 150,000 core-hours of LCRC resources before doing production runs using ALCF resources. Hafnia in semiconductors is actually a thin (10-100 nm) amorphous layer between layers of Si. The structure of this nanolayer is determined by how is was grown. Typically, atomic layer deposition or chemical vapor deposition is used to fabricate these monolayers. We will follow the procedure of Shirazi and Elliot [J. Comp. Chem. 35(2014)244] to use Kinetic Monte Carlo to predict the structure of the layer of amorphous HfO2 grown on Si during ALD or CVD. The predicted reaction rates, needed by KMC, will be derived from density functional calculations. We will be evaluate Eon [Henkelman and Jonsson, JCP 115(2001)9 657] and SPPARKS [Plimpton, et al. JCP 128(2008)205101] to solve the KMC equations. We estimate that 250,000 core-hours will be required for this evaluation. A critical parameter is the crystallization temperature of the amorphous layer of hafnia. Grain boundaries will form between nanocrystals during crystallization. The grain boundaries will carry current which is not desirable in a dielectric. The processing temperature of the DRAM chip must be less than crystallization temperature. Since the amorphous phase is metastable the standard process of estimating free energies from energies and phonon bands to predict phase transitions is not adequate. We propose to use Forward Flux Sampling (FFS) [Allen, et al. J. Phys.: Condens. Matter. 21(2009)463102] to model the crystallization process, predict rates and the crystallization temperature. We will use reactive force-fields in LAMMPS in the FFS simulations. We will develop LAMMPS scripts to conduct FFS through collaboration with Art Voter (LANL). We estimate that this study will require 100,000 core-hours. All data resulting from theory and calculations will be evaluated using Bayesian Analysis (BA) to determine the optimal models (i.e., mathematical expressions) that relate the free energy of the defected oxide phases to temperature and defect concentration (first and second year). To this end, we will use a Non-Naive Bayesian Network machine learning algorithm and software to process that will analyze the training and validation data sets in an unsupervised mode. The data generated by the DFT and kMC computations described above will enrich the diversity of the samples and reduce the final error in our estimations. The computationally intensive calculations wit evaluate the stability and meta-stability of phases with uncertainty intervals, using a methodology developed by the PI and co-authors (M. Stan and B. Reardon., CALPHAD, 27 (2003) 319). We estimate that 100,000 core-hours will be required for these calculations. We expect up to four project members. John Low, Marius Stan, a postdoc and summer student. Industry partnership: Project URL: Requested allocation: 700000 Q1: 175000 Q2: 175000 Q3: 175000 Q4: 175000 Justification: LAMMPS is a high performance code which has been optimized to run on Haswell processors, Nvidia GPUs, and Xeon Phi processors. Molecular dynamics for thousands of atoms will be used to model crystallation which will scale well to thousands or cores. See http://lammps.sandia.gov/bench.html for more details. Quantum Espresso can be built as a hybrid MPI/openMP executable and has been demonstrated good parallel scaling to 1 core for every 2 atoms and fair parallel scaling to 1 core for every atom. Since we are planning to use on the order of hundred atoms for our validation studies, our jobs will scale to about a hundred. We will profile QE with the Intel and Allinea tools to locate potential bottlenecks and opportunities for performance enhancement. See http://www.quantum-espresso.org/benchmarks/ for more details. CP2K is a high performance code which typically scales to one atom per core. It can exploit Haswells, GPUs and the Xeon PHI architecture as well. It should scale to hundred of cores for the models we will be using. See https://www.cp2k.org/performance#h20-64 for more details. Eon and SPPARKS are both designed to run parallel. Eon with run simulataniously run multiple LAMMPS jobs simultaneously. Since multiple jobs run in parallel with be embarrassingly parallel, the Kinetic Monte Carlo has good parallel scaling to a thousand cores or more. See http://henkelmanlab.org/pubs/chill14_055002.pdf for details. Storage requirements: The requester has used undetermined amount hours of their initial startup project. In addition to approving an initial amount, please specify a Category and Subcategory for this project. For a list of the current selection of approved categories, please see: https://wiki.lcrc.anl.gov/wiki/Processes/Categories Once the Allocation committee has approved the project, please go to the Project Management page to create it: https://accounts.lcrc.anl.gov/projects.php Thank You, The LCRC Accounts System