Oak Ridge National Laboratory (ORNL) in cooperation with the U.S. Department of Energy conducts a broad range of theoretical and computational research in materials sciences. This work is tightly integrated with experimental programs and is committed to making effective use of modern theory and advanced computation to progress core science and technology. Efforts include a full range of theory activities, ranging from basic science aimed at providing the fundamental understanding and basis for long term solutions to our energy problems, to near term work addressing our nation's most pressing energy and security needs.
Energy Frontier Research Centers (EFRC)
ORNL hosts two Energy Frontier Research Centers (EFRCs), the Fluid Interface Reactions, Structures, and Transport (FIRST) EFRC (http://www.ornl.gov/sci/first/index.shtml) and the Center for Defect Physics (CDP) EFRC. The aim of the former is to address the fundamental gaps in our current understanding of interfacial systems of high importance to future energy technologies, including electrical energy storage (batteries, supercapacitors) and heterogeneous catalysis for solar energy and solar fuels production. The latter is focused on structural materials, with an aim to bring a radically new level of rigor and insight to the discussion of defect structure, interactions, and dynamics in metals and alloys. Both EFRCs employ a combination of state of the art experimental and computational techniques.
Within the FIRST EFRC, we are working to develop improved molecular dynamics algorithms and implementations, specifically aimed to improve time to solution for first principles molecular dynamics calculations. Careful optimization and tuning of simulation codes has enabled, e.g., comprehensive studies of the of the liquid electrolytes used in lithium ion batteries with a time to solution acceptable to researchers.
Within the CDP EFRC, we are developing and applying state of the art electronic structure techniques to enable prediction of materials properties to unprecidented accuracy and confidence, using the Quantum Monte Carlo (QMC) technique. We recently performed extensive studies of the properties of aluminum metal and its defects (http://prb.aps.org/abstract/PRB/v85/i13/e134109), finding large (3000 Kelvin) errors in the predictions of the more commonly applied density functional methods for the formation energies of typical defects. This is significant because it is the defect properties that govern materials performance. The QMC calculations required several million CPU hours and are able to fully exploit the Leadership Computing machines at Oak Ridge. Due to the improvements in QMC algorithms and the improvements in available computer power, we expect to predict phase diagrams and defect properties of a wide range of materials within the next few years.
Molecular dynamics (MD) simulations essentially consist of numerically integrating Hamilton's, Newton's or Lagrange's equations of motion using small integration time steps. Although these equations are valid for any set of conjugate positions and momenta, the use of Cartesian coordinates greatly simplifies the kinetic energy term. The resulting coupled first-order, ordinary differential equations can be solved using numerical integrators. Despite the simplicity of MD methods, the simulation of millions of atoms over long time scales like seconds is not possible without substantial reduction of the number of equations-of-motion, a feat typically achieved by course graining groups of atoms into one element. This approach not only reduces the number of equations, but if proper separation of the time scales for motion can be achieved, it also allows the use of considerably larger integration time steps and thus helps bridge the large gap between simulation and experimental time scales.
Molecular mechanics (MM) methods apply the laws of classical physics to predict structures and properties of molecules by optimizing the positions of atoms based on the energy derived from an empirical force field describing the interactions between all nuclei (electrons are not treated explicitly). As such, molecular mechanics can determine the equilibrium geometry in a much more computationally efficient manner than ab initio quantum chemistry methods, yet the results for many systems are often comparable, at least qualitatively. However, since molecular mechanics treats molecular systems as an array of atoms governed by a set of potential energy functions, the model nature of this approach should always be noted.
Course Grained Models
Course grained models (CGM) consist of replacing an atomistic description of a molecule with a lower-resolution model that averages away some of the fine details of the interactions. Numerous coarse-grained models (differing in the metric used to obtain the course-grained potential, such as the fitting to the forces or to structural features, but also in the definition of the course grain) have been developed in order to investigate the longer time- and length-scale dynamics that are critical to many long-chain molecular processes, such as polymer, lipid membranes, and proteins. Coarse graining can also refer to the removal of certain degrees of freedom (e.g., vibrational modes between two atoms) by freezing the bonds, bends, or torsional degrees of freedom, but more typically it implies that two or more atoms are collapsed into a single particle representation (the so-called united atom model was one of the first popular coarse grained models). Fundamentally the level to which a system may be coarse grained is bound by the accuracy in the dynamics and structural properties desired from a simulation. We refer to several review articles that provide explicit details.
Nanomaterials Theory Institute
The NTI develops and applies modern computational and mathematical capabilities for the understanding, prediction and control of chemical and physical processes ranging from the molecular to the nanoscale using a multidisplinary approach that integrates chemistry, physics, and materials science. Work is focused toward using theory and multiscale simulations and modeling for providing interpretive and predictive frameworks for virtual design and understanding of novel nanoscale materials with specific and/or emergent properties. The work is fully integrated with experimental studies and pushes the boundaries of computationally directed materials design.
Cleaning Oil Spills with Nanotube Sponges - Carbon Nanotubes
Scientists at the US Department of Energy's (DOE) Oak Ridge National Laboratory (ORNL) have managed to use computational simulations developed in-house to create a new type of sponge that is perfect for removing oil residues from water.
The material is based on carbon nanotubes (CNT), extremely small wires that come together to form a special type of mesh. These thin filaments can be obtained from single-atom-thick sheets of graphene, which are rolled and then welded into cylinders.
CNT are among the strongest materials out there today, and exhibit a great potential for high electrical conductivity. They are also extremely lightweight, which makes them ideal for a wider variety of practical applications.
Since constructing nanotubes is a very complex scientific process (plagued by numerous limitations), researchers first had to develop a way of growing large clumps of the thin wires. ORNL investigator Bobby Sumpter was part of a larger team that managed to achieve this objective.
CNT can now be grown in large clumps by creating a pure carbon lattice, and then selectively substituting boron atoms within this structure. For this study, Sumpter worked closely with expert Vincent Meunier, who is now based at the Rensselaer Polytechnic Institute (RPI).
In order to conduct the necessary simulations, the team used a number of supercomputers, including one at the ORNL Leadership Computing Facility. These models were meant to reveal how the addition of boron atoms would influence the structure of the nanotubes.
"Any time you put a different atom inside the hexagonal carbon lattice, which is a chicken wire-like network, you disrupt that network because those atoms don't necessarily want to be part of the chicken wire structure," Sumpter explains.
"Boron has a different number of valence electrons, which results in curvature changes that trigger a different type of growth," he adds. Details of the new investigation were published in the latest issue of the esteemed journal Nature Scientific Reports.
"Instead of a forest of straight tubes, you create an interconnected, woven sponge-like material. Because it is interconnected, it becomes three-dimensionally strong, instead of only one-dimensionally strong along the tube axis," the ORNL expert concludes.
Polymer-based Multicomponent Materials
Multicomponent polymeric materials are widely used in various modern technologies and will have even broader applications in future technologies, from lightweight materials, to solar cells and electrical energy storage to biomedical technologies. Yet, our fundamental understanding of the processes and interactions that control macroscopic properties in these materials remains limited. The overarching goal of the research is to develop a fundamental understanding of how interfacial properties and interactions affect structure, morphology, dynamics, and macroscopic properties of multicomponent polymeric systems, in both the liquid and solid states. The research focuses on two themes. The first seeks to correlate structure-property relationships in polymer-nanoparticle mixtures to the nanoparticle structure and interfacial interactions, while the second involves the correlation of molecular architecture, electrostatic interactions and external fields to the morphology of multiblock copolymer materials, including both neat block copolymers and those containing discrete nanoparticles. To fully understand the underlying processes and mechanisms, we will pursue a comprehensive interdisciplinary approach lead by advanced theory and simulations, precise synthesis with nano-scale control and state-of-the-artcharacterization (with special emphasis on neutron scattering). The fundamental knowledge developed in this program will contribute to the scientific foundation for the rational design of multicomponent polymer based materials with superior properties and function that can address many DOE challenges such as organic photovoltaics, fuel cell membranes, and stronger light-weight materials that result in energy savings.
This project uses ab-initio many-body electronic structure calculations to unravel outstanding problems in the prediction of materialsproperties of interest to DOE. In particular, we are developing an understanding of metal oxides that have wide application including energy storage, catalysis, and energy production, and metals that are widely used as structural materials. This is achieved by using a state of the art electronic structure method, Quantum Monte Carlo (QMC), implemented in QMCPACK. To our knowledge, this is the only QMC code fully optimized and proven to run on the hybrid GPU-CPU Titan architecture at OLCF. The package utilizes modern CUDA, C++, XML, HDF5, MPI and OpenMP and implements state of the art algorithms. The package is fully open source. Significant developments in the algorithms and code have been recently implemented and the code has been fully “GPUized” with a large number of custom CUDA kernels written and tested on “titandev”. We aim to both solve scientific problems in these materials and to identify the fundamental problems limiting accuracy with density functional theory approaches. Thus, this project targets significant scientific impact and a longer lasting impact in the materials modeling community.Diagonalization Solvers for Electronic Collective Phenomena in NanoscienceThe objective of this project is to study collective phenomena at the electronic level, using the DMRG algorithm, and to develop and make available the corresponding computational codes. The proposed research will be carried out for realistic models of relevance for practical applications of strongly correlated electronic materials. We aim to understand three aspects: the real time evolution in electron transport, the temperature dependence of electronic properties in nanostructures, and the multiscale nature of the emerging orders in strongly correlated systems. The main approach of these theoretical studies is the use of the density matrix renormalization group algorithm to obtain information from the models. We use also auxiliary approaches, suchas the Lanczos and Davidson algorithm, and for testing purposes either series expansions, or the non-interacting case when available and meaningful.
DCA++ is a computer code for the simulation of properties of high-temperature superconductors and other strongly correlated electronic systems. The code is composed of two parts, a self-consistent dynamical cluster approximation (DCA), and a quantum Monte Carlo ``kernel'' acting as an impurity solver inside. The simulation code is written in C++ with a generic and extensible approach and is tuned to perform well at scale. The message passing interface (MPI) is used for parallelization of the code. The bottleneck of the DCA algorithm (70% or more of the CPU time) is in dense linear algebra operations, mainly matrix-vector multiplications. These are delegated to the BLAS library, and its high-performance and parallel implementations. Signiﬁcant algorithmic improvements have been made to make effective use of current supercomputing architectures. By implementing delayed Monte Carlo updates and a mixed single-/double precision mode, we are able to dramatically increase the efﬁciency of the code. Notably, research work that used DCA++ was awarded the Gordon Bell Prize 2008 (team award) for first petaflop calculation. DCA++ was used to simulate disorder effects in high temperature superconductors.
Computational Nanoscience Endstation (CNE)
DCA++, DMRG++, and QMC are among a suite of codes and simulation capabilities that comprise the computational nanoscience end-station (CNE) developed in collaboration between CNMS and CSMD. In analogy to experimental end-stations at large experimental facilities, the CNE provides users with the leading edge scientific instrumentation (i.e., modeling software) and expertise to perform scientific research at scale on leadership computing facilities. In addition to DCA++, DMRG++, QMC and a toolkit to support atomistic simulations of magnetic nanosystems, the CNE currently supports large-scale electronic structure codes that allow direct ab-initio simulations of nanoscale systems. The CNE has been an important driver of the CNMS user program.