HPCNano09

Workshop Technical Program

8:30AM-8:40AM
Opening
Dr. Almadena Y. Chtchelkanova, Program Director, NSF

8:35AM-8:40 AM
Introduction to HPCNano Series
Dr. Jun Ni, College of Medicine, University of Iowa, USA
Dr. Andrew Canning, Lawrence Berkeley National Laboratory, DOE, USA
Dr. Lin-Wang Wang, Lawrence Berkeley National Laboratory, DOE, USA

9:10AM-9:40AM

Large scale calculations of electronic transport properties in quantum dot structures

Gil Speyer¹ and Richard Akis²
¹High Performance Computing Initiative, Ira A. Fulton School of Engineering, Arizona State University, Tempe AZ, 85287-5206
²Department of Electrical Engineering and Center for Solid State Electronics Research, Arizona State University, Tempe, Arizona 85287-5706

Abstract. A fundamental issue in quantum mechanics concerns the manner in which the discrete level spectrum of an isolated system is modified when it is coupled to some external, macroscopic measuring environment. From both experimental and theoretical perspectives, an ideal system for the study of this issue is provided by semiconductor quantum dots, which are quasi-zero dimensional semiconductor structures in which the flow of electrical current is confined on length scales comparable to the size of the electron itself. Recent progress investigating such systems has revealed that studying fine structure (whether it be in physical space or energy space) to be considerable importance in understanding their physics. Moreover, increasingly complex devices, involving multiple dots connected together, are of now of considerably interest for quantum computing applications. Unfortunately, in the effort to investigate these developments, transport simulations have been hamstrung by the limitations of using a single processor. This work will show results from a parallelized simulation program, which uses a cascading scattering matrix method in order to determine current flow through such devices. We concentrate specifically on the case of a coupled dot system, where parallelization may lead to a much greater capability in be able to understanding the physics and potential device applications of such systems.

9:40AM-10:10AM

Virtual experiments for exploring photovoltatic phenomenon and photo-induced deformation of nanomaterials using parallel computational code FPSEID

Yoshiyuki Miyamoto, Nano Electronics Res. Labs., NEC

Abstract. In this talk, I will show application of the time-dependent density functional theory (TDDFT) scheme to simulate photovoltatic phenomena in polar crystal and laser-pulse fabrication of nano-carbons. The light illumination is mimicked by alternating electric-field (E-field), and real-time propagation of electron wave functions are computed followed by classical molecular dynamics using a code FPSEID. Usual photovoltatic devices are consisted of /pn-/junctions of semiconductor to split excited electrons and hole apart. The splitting should also be induced by internal E-field in polar crystallographic direction of compound semiconductors. To confirm this idea, polar (001) surface of silicon carbide under illumination was examined and the increase of polarization of valence electrons with some selected energy of light was found. Interestingly, the increased polarization can sustain even after the light was off suggesting feasibility of photovoltatic phenomena in polar crystal with neither /p/-type nor /n/-type doping. When the intensity of light is increased in the order of volt per angstrom and the pulse width is short in the order of few tens femto-seconds or less, the light turns to be a non-thermal tool for deformation of materials. Here I demonstrate possibility of designing the shape and intensity of the laser-pulse for demanded structural change by performing the TDDFT simulations prior to experiments. Controlled pealing-off of a graphene sheet from a graphite surface and deformation/destruction of carbon nanotubes will be demonstrated.

10:10AM-10:20AM Coffee Break

10:20AM-1050AM

Quantum simulation of materials at micron scale and beyond

Gang Lu, Department of Physics and Astronomy, California State University Northridge, Northridge, California, USA

Abstract. We present a multiscale modeling approach that can simulate multimillion atoms effectively via density functional theory. The method is based on the framework of the quasicontinuum (QC) approach with the density-functional theory (DFT) as its sole energetics formulation. The local QC part is formulated by the Cauchy-Born hypothesis with DFT calculations for strain energy and stress. The nonlocal QC part is treated by a DFT-based embedding approach, which couples DFT nonlocal atoms to the local region atoms (vertices of the finite-element mesh at the atomic dimension). The method—QCDFT—is applied to a nano-indentation study of an Al thin film in the presence and absence of Mg impurities. The system contains over 60 million atoms. The results suggest that QCDFT represents a new direction for the quantum simulation of materials at length scales that are relevant to experiments. If time allows, an application of QCDFT to magnetism-induced dislocation mobility and cross-slip in NiAl alloys will also be presented.

10:50AM-11:20AM

A Method of Simulating Electron Conduction on Nanotube through extended Tight-Binding Molecular Dynamics at Finite Temperature

Satoshi Nakamura, Syogo Tejima, Mikio Iizuka, Yoshihisa Shizawa, Kazuo Minami, and Hisashi Nakamura Research Organization for Information Science and Technology 2-2-54, Nakameguro, Meguro-ku, Tokyo, 153-0061, Japan David Tomanek Department of Physics and Astronomy, Michigan State University East Lansing, MI 48824-1116, USA

Abstract. The tight-binding molecular dynamics method is extended to cope with electron conduction in carbon nanotube or nano-carbon- structure at finite temperature, incorporating the time-dependent electromagnetic field into its scheme, and taking into account the retardation of electron response to lattice motion, within the linear response theory. The non-equilibrium Green’s function method is used to study the electron-conduction in carbon nanotube that is connected to the electrodes at finite bias voltages and temperature. In our scheme, the large-dimensional matrix inversion to calculate the Green’s function in the scattering region becomes the most heavy part of the model computationally. A embedding potential algorithm is implem ented to obtain the equilibrium and non- equilibrium Green’s functions. The Hartree potential is also computed through a substitute charge method combi ned with least-squares analysis. Our method allows us to impose the Dirichlet boundary conditions on three-dimensional surfaces with arbitrary shape and to treat even problems with incomplete boundary conditions.

11:20AM-11:50AM

Toward Accuarte Calculations of Large-Molecular Ssystems

Mark S. Gordon
Iowa State University and Ames Lab

The most accurate methods available to treat large molecular systems, such as liquids, enzymes, and polymers, are well correlated ab initio approaches, such as many body perturbation theory and coupled cluster theory. However, these methods typically scale so poorly with system size that they are only applicable to species with tens of atoms. This gap between desired and feasible length scales is being reduced by the devleopment of novel methods, such as the fragment molecular orbital method. The essence of this method will be briefly discussed and examples will be presented.

Lunch Break

1:10PM-1:40PM

Fast algorithms for electronic structure calculations for metallic system based on selected inversion.

Jianfeng Lu, The Courant Institute of Mathematical Sciences, New York University, USA

Abstract. We present a fast algorithms for Kohn-Sham DFT calculations for metallic system. The main components of the algorithms are efficient pole-based representations of the Fermi operator and fast selected inversion of the shifted Hamiltonian matrix. The algorithm is highly accurate and has complexity scales as O(N^{3/2}) for 2D and quasi-2D systems and O(N^2) for 3D system. Application to quantum dot system shows the great potential of this algorithm for system with large number of electrons. This is a joint work with Roberto Car, Weinan E, Lin Lin, Chao Yang, Lexing Ying.

1:40PM-2:10PM

Designing Hydrogen Storage Materials Using High-Performance Computing
Vidvuds Ozolinš, Department of Materials Science & Engineering, University of California, Los Angeles

Abstract. General adoption of hydrogen as a vehicular fuel depends on the ability to store hydrogen at high volumetric and gravimetric densities, as well as on the ability to extract it at sufficiently rapid rates. Practical requirements for on-board hydrogen storage systems are very challenging and cannot be met with existing technologies, and significant effort has been invested in searching for new materials that could efficiently store hydrogen at near ambient temperatures and pressures. Among the various hydrogen storage options (such as compressed gas, sorbents and metal-organic frameworks, chemical liquid hydrides), solid-state storage in multicomponent complex hydrides has attracted particular attention due to the very high storage densities and favorable thermodynamics that can be achieved in these systems. We will show how first-principles theoretical calculations based on the fundamental physics theories of quantum mechanics and statistical mechanics can be used as a valuable tool for understanding and predicting novel hydrogen storage materials. Recent studies in our group have used density-functional theory (DFT) calculations to (i) predict crystal structures of new solid-state hydrides, (ii) determine phase diagrams and thermodynamically favored reaction pathways in multinary hydrides, and (iii) study microscopic kinetics of diffusion, phase transformations, and hydrogen release.

2:10PM-2:40PM
Determining the Nonlocal Optical Properties of Metallic Nanostructures via High—Performance Computing
Jeffrey M. McMahon , George C. Schatz Research Group, Department of Chemistry, Northwestern University

Abstract. Interest in metallic structures with features on the order of 10 nm or less has significantly increased as experimental techniques for their fabrication have become possible. Even if the features involve many hundreds of atoms or more so that a continuum level of description is adequate, their optical response can be difficult to correctly model due to spatially nonlocal dielectric effects (which arise from the quantum mechanical momentum dependence in the dielectric function). Until recently, the treatment of such effects has been limited to spherical structures and planar surfaces. In this talk, I will discuss a computational approach that we have developed to model such effects. Our approach is based on finite—difference approximations to the relevant partial differential equations. Because nanostructures often contain very detailed features, such simulations can require many millions or billions of grid points. Performing such calculations thus requires the use of high—performance computers. Recent results from such calculations will also be presented and discussed, including the optical response of nanowires and nanoparticles, transmission through thin films, and electromagnetic enhancements from junction structures.

2:40PM-3:10PM
Structure, Dynamics and Charge Transport at Metal-organic Interface
Hai-Ping Cheng, Department of Physics and the Quantum Theory Project, University of Florida
Gainesville, FL 32611

Abstract. In this talk, I use a few example problems to illustrate the role of computational methods in studies of metal-organic interfaces. Issues including phase transition of alkanethiol monolayer films on gold surface, diffusion of metal atoms and clusters on the
surfaces, and electron transport through metal-organic-metal junctions will be discussed. These problems are of fundamental and application importance in molecular electronics, new composite materials, and energy research.

3:10PM-3:40PM

Computer Simulation Studies of Metallic Liquids and Glasses
C. Z. Wang, Ames Laboratory – USDOE & Department of Physics, Iowa State University, Ames, IA 50011

Abstract. In order to develop new materials with novel properties including, for instance, metallic glasses and crystalline/amorphous composites, it is necessary to understand how atomic scale behavior during rapid cooling manifests in the emergence of amorphous structure or metastable phases. Atomistic simulations play an important role in elucidating the connections between atomic scale structures exhibited in the liquid phase, as a sample is brought close to and below the glass transition, and the resultant structure after solidification. However, there is a large disparity between the time scales accessed in the microscopic simulations and realistic experimental time scales. On the other hand, there exists a well-developed theoretical framework for describing solidification phenoemena using a continuum description. Phase-field simulations have been successful in describing the dynamical non-equilibrium processes in the evolution of morphology in a cooling liquid/solid system. Our studies seek to establish a bridge between these two simulation methods at very different length scales and time scales. We have developed a number of theoretical tools that can analyze the results of atomistic simulations to elucidate the short- and medium-range chemical and structural ordering that is developing in the system. We have also developed a framework that can encapsulate the thermodynamic and structural information from the atomistic simulations into succinct thermodynamical descriptions to capture the essential structural fluctuations and ordering that is developing in the undercooled liquid system. Most of our current work is focused on binary Zr-X alloys with X being Cu, and Ni. Binary Zr-Cu metallic systems have attracted considerable interest because of the relatively high glass forming ability in such a simple metallic system, which offers a good opportunity for analyzing the structural features to understand dynamical and mechanical properties in glass-forming systems.

Work done in Collaboration with K. M. Ho, S. G. Hao, M. Z. Li, L. Huang, X. W. Fang, M. J. Kramer, M. I. Mendelev, and R. Napolitano

3:40PM-3:50PM
Coffee Break

3:50PM-4:20PM

Time-domain Ab Initio Dynamics of Wet-Electrons on TiO2

Oleg Prezhdo and Sean Fischer, Department of Chemistry, University of Washington, Seattle, WA 98195-1700

Abstract. The electron transfer (ET) dynamics of wet-electrons on a TiO2 surface was investigated using state-of-the-art ab initio nonadiabatic molecular dynamics. Delocalized over both water and TiO2, wet-electrons are supported by a new type of state that is created at the interface due to the strong water? TiO2 interaction and that cannot exist separately in either material. Our simulations indicate that ET is sub-10fs and driven manly by low frequency modes. Similar states are present in a number of other systems with strong interfacial coupling, including certain dye-sensitized semiconductors. The ET dynamics involving such interfacial states share many universal features, such as an ultrashort time scale and weak-dependence on temperature, surface defects, and other system details. Time-resolved studies of wet-electrons provide a fundamental understanding of the role of the solvent in dye-sensitized semiconductor solar cells. While pristine wet-electron states are not directly involved in the photovoltaic process, solvents do play critical roles by saturating the dangling bonds of the TiO2 surface and strongly shifting the energy levels of the charged species generated during ET.

4:20PM-4:50PM

Magnetism in the Diluted Magnetic Semiconductor within the Dynamical Cluster Approximation

Unjong Yu1, Abdol-Madjid Nili1, Karlis Mikelsons1, 2, Brian Moritz3, Juana Moreno1 and Mark Jarrel1
1Department of Physics and Astronomy & Center for Computation and Technology,
Louisiana State University, Baton Rouge, LA 70803, USA; 2Department of Physics, University of Cincinnati, Cincinnati, Ohio 45221, USA; 3Stanford Institute for Materials and Energy Science; SLAC National Accelerator Laboratory and Stanford University, Stanford, CA 94305, USA

Abstract. The discovery of high temperature ferromagnetism in diluted magnetic semiconductors (DMS) has stimulated a great deal of attention. The interest in these materials is due to possible applications in spintronics as the source of a spin polarized current or as the base material for a chip that can simultaneously store and process data. However, the ferro-magnetic transition temperature (TC) of DMS should be increased at least above the room temperature for practical uses. A complete theoretical understanding of the ferromagnetism of these systems is essential to help guide the search for a room temperature DMS, but it is inhibited by the need to include strong electronic correlations, strong disorders, the spin-orbit coupling, and non-local effects. We implement the dynamical cluster approximation (DCA) with the k-dot-p method to treat all these four effects, on an equal footing. The DCA is a cluster extension of the dynamical mean-field approximation, and systematically incorporates non-local dynamical corrections by mapping the lattice system onto a cluster embedded into an effective medium. By increasing the cluster size, longer range of correlations are included while the calculation remains in the thermodynamic limit. Since all the possible configurations of magnetic ions within clusters are considered effectively in this method, we require high performance com- putting with high level of parallelization. The algorithm and the parallelization scheme are presented. We apply this method to study the magnetic properties of a prototypical DMS Ga 1-x MnxAs. We show that non-local effects are essential for explaining the experimentally observed TC and saturation magnetization. We demonstrate that the reduction of TC and saturation magnetization is mainly due to rotational frustration and propose DMS nano-structures that eliminate the frustration and increase TC. We also explain spin reorientation observed in this system by the cluster anisotropy.

4:50PM-5:20PM

A Multi-scale Modeling and large-scale Computations of Biotransports in Nanothermo/Chemoradiation Therapies
Jun Ni, Department of Radiology, Carver College of Medicine, the University of Iowa

Abstract. A thermo-therapeutic processing can be applied to either regional tumor (local hyperthermia) or the whole body (whole-body hyperthermia), depending on the tumor position, the cancer stage, and the health status of cancer patients. The major challenge in hyperthermia is to properly introduce a heat source for increasing tumor temperature during therapeutic process. Thermotherapy is often with chemo-radiation therapies. Such treatment brings not only promising but also challenges, since the tissue biotransport is very complex and the thermal-chemical dose profiles are coupled each other. Recently-developed nanothermotherapy brings a high promise to cancer and disease treatments. Induction of ferromagnetic or superparamagnetism (SPM) nanoparticles into a tumor region by injection or by blood-perfusion help terminate tumor cell lives. By applying an external electromagnetic field, nanoparticles help generate heat and reorient the direction to targeted tumor. The nanothermotherapy, on the other hand, yields challenges. It is urgent to develop a general model for simulate its clinical uses. Such demand requires an intensive study on the interactions of hyperthermia with tumor biological metabolism, its environments, the effects of angiogenesis and vasculature, blood perfusion, therapeutic gain of heat, etc. The general model of biotransport (heat, mass, electromagnetism etc) must be developed based on multi-scales (nano-/ micro-/macroscopic). The simulation must be full-field. The thermal and mass conservations should be coupled. The thermo non-equilibrium which accounts for temperature differences should be considered. The thermal heat should be microscopically managed and monitored. The biotransport simulations use large-scale peta-scale high performance computing facility. The research results will be publicly useful in designing future Computer-Aided Planning (CAP) systems for clinical trials. This presentation focuses on how we develop a simulator (model, simulation, and software) that contains a volume-averaged, generalized, multi-scale, multi-medium mathematical model of bio-electronic-magnetic (EM) -biotransport phenomena in a therapeutic process. The model accounts for nanoscale particulate motion, size and shape effects, microscopic cell and vasculatures, micro-fluid and heat transfer vessel perfusion, tumor tissue microstructure and morphology, energy and mass (dose and other biological compositions) conservation with proper physics-based constitutions, and interfacial heat/mass transfers for blood perfusion expressions and other micro-biological structures. The model equations are implemented numerically using the our computing facility and then NSF national petascale supercomputing facility at NCSA. Many geometric effects of nanoparticles, external alternate electromagnetic fields, nanoparticle materials, and biological properties of target tissue and vascular structures are studied. Presentation also reports the current status of developing the state-of-the-art open software for clinical CAP systems for future nanothermo- and chemo-radiation therapy.

5:20PM
Workshop Closure (see you at HPCNano10)
Dr. Jun Ni, College of Medicine, University of Iowa, USA
Dr. Andrew Canning, Lawrence Berkeley National Laboratory, DOE, USA
Dr. Lin-Wang Wang, Lawrence Berkeley National Laboratory, DOE, USA