[Show abstract][Hide abstract]ABSTRACT: We present an implementation of localized atomic-orbital basis sets in the projector augmented wave (PAW) formalism within the density-functional theory. The implementation in the real-space GPAW code provides a complementary basis set to the accurate but computationally more demanding grid representation. The possibility to switch seamlessly between the two representations implies that simulations employing the local basis can be fine tuned at the end of the calculation by switching to the grid, thereby combining the strength of the two representations for optimal performance. The implementation is tested by calculating atomization energies and equilibrium bulk properties of a variety of molecules and solids, comparing to the grid results. Finally, it is demonstrated how a grid-quality structure optimization can be performed with significantly reduced computational effort by switching between the grid and basis representations.
[Show abstract][Hide abstract]ABSTRACT: In heterogeneous catalysis, identifying the active site for key reaction steps is an important contribution for the optimization of industrial synthesis. The structure sensitivity of CO dissociation on a metal catalyst, which is the rate-limiting step for the methanation and the Fischer-Tropsch processes under certain conditions, has been debated for years. Here, scanning tunneling microscopy (STM) and density functional theory (DFT) are used to clarify the role of monatomic steps in the splitting of CO on a stepped Ru(0 1 54) crystal, which displays alternating steps with either 4-fold or 3-fold symmetry. After CO doses at elevated temperatures, the STM images reveal step decorations characteristic of atomic oxygen resulting from CO dissociation on every second step. The comparison of the STM images with the results of DFT calculations shows that the step decoration occurs on the steps displaying the 4-fold symmetry. We conclude that the active sites for CO dissociation on ruthenium are located on the 4-fold symmetry monatomic steps.
Full-text · Article · May 2012 · The Journal of Physical Chemistry C
[Show abstract][Hide abstract]ABSTRACT: Density functional theory has became the workhorse for simulations of
catalytic reactions and computational design of novel catalysis. The
generally applied semi-local exchange-correlation functionals have
successfully predicted catalytic reaction trends over a variety of
surfaces. However, in order to achieve quantitative predictions of
reaction rates for molecule-surface systems, in particular where there
is weak Van der Waals interactions or strong correlation, it is of vital
importance to include non-local correlation effects. The use of random
phase approximation (RPA) to construct the correlation energy, combined
with the exact, self-interaction free exchange energy, offers a
non-empirical way for accurately describe the adsorption energies 
and dispersion forces . We have recently implemented RPA in the GPAW
code [3-4], an electronic structure package using projector augmented
wave method and real space grids. In this talk I will present our
initial results comparing RPA and generalized gradient functionals for
the activation energies and reaction energies for transition metal or
metal oxide surfaces. [4pt]  L. Schimka, et.al, Nature Mat. 9, 741
(2010)  T. Olsen, et.al, Phys. Rev. Lett. 107, 156401 (2011)  J.
Yan, et.al, Phys. Rev. B 83, 245122 (2011).  J. Yan, et.al, Phys.
Rev. Lett. 106, 146803 (2011)
[Show abstract][Hide abstract]ABSTRACT: Electronic structure calculations are a widely used tool in materials science and large consumer of supercomputing resources. Traditionally, the software packages for these kind of simulations have been implemented in compiled languages, where Fortran in its different versions has been the most popular choice. While dynamic, interpreted languages, such as Python, can increase the effciency of programmer, they cannot compete directly with the raw performance of compiled languages. However, by using an interpreted language together with a compiled language, it is possible to have most of the productivity enhancing features together with a good numerical performance. We have used this approach in implementing an electronic structure simulation software GPAW using the combination of Python and C programming languages. While the chosen approach works well in standard workstations and Unix environments, massively parallel supercomputing systems can present some challenges in porting, debugging and profiling the software. In this paper we describe some details of the implementation and discuss the advantages and challenges of the combined Python/C approach. We show that despite the challenges it is possible to obtain good numerical performance and good parallel scalability with Python based software.
Full-text · Article · Dec 2011 · Procedia Computer Science
[Show abstract][Hide abstract]ABSTRACT: We describe the implementation of K-shell core level spectroscopies (X-ray absorption (XAS), X-ray emission (XES), and X-ray photoemission (XPS)) in the real-space-grid-based Projector Augmented Wave (PAW) GPAW code. The implementation for XAS is based on the Haydock recursion method avoiding computation of unoccupied states. The absolute energy scale is computed with the Delta Kohn–Sham method which is possible using specific PAW setups for the core-hole states. We show computed spectra for selected test cases (gas phase H2O and bulk diamond) and discuss the dependence on grid spacing and box size. In the case of diamond we include vibrational effects by sampling spectra from the ground state vibrational distribution and discuss the importance of those effects for the experimentally observed features. We apply the method to XPS, XES and XAS of CO adsorbed on Ni(100) and compare to experimental data where possible.
Full-text · Article · Nov 2011 · Fuel and Energy Abstracts
[Show abstract][Hide abstract]ABSTRACT: We calculate the potential energy surfaces for graphene adsorbed on Cu(111), Ni(111), and Co(0001) using density functional theory and the random phase approximation (RPA). For these adsorption systems covalent and dispersive interactions are equally important and while commonly used approximations for exchange-correlation functionals give inadequate descriptions of either van der Waals or chemical bonds, RPA accounts accurately for both. It is found that the adsorption is a delicate competition between a weak chemisorption minimum close to the surface and a physisorption minimum further from the surface.
Full-text · Article · Oct 2011 · Physical Review Letters
[Show abstract][Hide abstract]ABSTRACT: AbstractWe address the fundamental question of which size a metallic nano-particle needs to have before its surface chemical properties
can be considered to be those of a solid, rather than those of a large molecule. Calculations of adsorption energies for carbon
monoxide and oxygen on a series of gold nanoparticles ranging from 13 to 1,415 atoms, or 0.8–3.7nm, have been made possible
by exploiting massively parallel computing on up to 32,768 cores on the Blue Gene/P computer at Argonne National Laboratory.
We show that bulk surface properties are obtained for clusters larger than ca. 560 atoms (2.7nm). Below that critical size,
finite-size effects can be observed, and we show those to be related to variations in the local atomic structure augmented
by quantum size effects for the smallest clusters.
Full-text · Article · Aug 2011 · Catalysis Letters
[Show abstract][Hide abstract]ABSTRACT: We present an implementation of the linear density response function within
the projector-augmented wave (PAW) method with applications to the linear
optical and dielectric properties of both solids, surfaces, and interfaces. The
response function is represented in plane waves while the single-particle
eigenstates can be expanded on a real space grid or in atomic orbital basis for
increased efficiency. The exchange-correlation kernel is treated at the level
of the adiabatic local density approximation (ALDA) and crystal local field
effects are included. The calculated static and dynamical dielectric functions
of Si, C, SiC, AlP and GaAs compare well with previous calculations. While
optical properties of semiconductors, in particular excitonic effects, are
generally not well described by ALDA, we obtain excellent agreement with
experiments for the surface loss function of the Mg(0001) surface with plasmon
energies deviating by less than 0.2 eV. Finally, we apply the method to study
the influence of substrates on the plasmon excitations in graphene. On
SiC(0001), the long wavelength $\pi$ plasmons are significantly damped although
their energies remain almost unaltered. On Al(111) the $\pi$ plasmon is
completely quenched due to the coupling to the metal surface plasmon.
[Show abstract][Hide abstract]ABSTRACT: Density function theory (DFT) is the most widely employed electronic
structure method due to its favorable scaling with system size and
accuracy for a broad range of molecular and condensed-phase systems. The
advent of massively parallel supercomputers have enhanced the scientific
community's ability to study larger system sizes. Ground state DFT
calculations of systems with O(10^3) valence electrons can be routinely
performed on present-day supercomputers. The performance of these
massively parallel DFT codes at the scale of 1 - 10K execution threads
are not well understood; even experienced DFT users are unaware of
Amdahl's Law and the non-trivial scaling bottlenecks that are present in
standard O(N^3) DFT algorithms. The GPAW code was ported an optimized
for the Blue Gene/P. We present our algorithmic parallelization strategy
and interpret the results for a number of benchmark tests cases. Lastly,
I will describe opportunities for computer allocations at the Argonne
Leadership Computing Facility.
[Show abstract][Hide abstract]ABSTRACT: Electronic structure calculations have become an indispensable tool in many areas of materials science and quantum chemistry. Even though the Kohn-Sham formulation of the density-functional theory (DFT) simplifies the many-body problem significantly, one is still confronted with several numerical challenges. In this article we present the projector augmented-wave (PAW) method as implemented in the GPAW program package (https://wiki.fysik.dtu.dk/gpaw) using a uniform real-space grid representation of the electronic wavefunctions. Compared to more traditional plane wave or localized basis set approaches, real-space grids offer several advantages, most notably good computational scalability and systematic convergence properties. However, as a unique feature GPAW also facilitates a localized atomic-orbital basis set in addition to the grid. The efficient atomic basis set is complementary to the more accurate grid, and the possibility to seamlessly switch between the two representations provides great flexibility. While DFT allows one to study ground state properties, time-dependent density-functional theory (TDDFT) provides access to the excited states. We have implemented the two common formulations of TDDFT, namely the linear-response and the time propagation schemes. Electron transport calculations under finite-bias conditions can be performed with GPAW using non-equilibrium Green functions and the localized basis set. In addition to the basic features of the real-space PAW method, we also describe the implementation of selected exchange-correlation functionals, parallelization schemes, ΔSCF-method, x-ray absorption spectra, and maximally localized Wannier orbitals.
No preview · Article · Jun 2010 · Journal of Physics Condensed Matter
[Show abstract][Hide abstract]ABSTRACT: We use density functional theory (DFT) with a recently developed van der
Waals density functional (vdW-DF) to study the adsorption of graphene on Al,
Cu, Ag, Au, Pt, Pd, Co and Ni(111) surfaces. In constrast to the local density
approximation (LDA) which predicts relatively strong binding for Ni,Co and Pd,
the vdW-DF predicts weak binding for all metals and metal-graphene distances in
the range 3.40-3.72 \AA. At these distances the graphene bandstructure as
calculated with DFT and the many-body G$_0$W$_0$ method is basically unaffected
by the substrate, in particular there is no opening of a band gap at the
[Show abstract][Hide abstract]ABSTRACT: We present a computational screening study of ternary metal borohydrides for reversible hydrogen storage based on density functional theory. We investigate the stability and decomposition of alloys containing 1 alkali metal atom, Li, Na, or K (M(1)); and 1 alkali, alkaline earth or 3d/4d transition metal atom (M(2)) plus two to five (BH(4))(-) groups, i.e., M(1)M(2)(BH(4))(2-5), using a number of model structures with trigonal, tetrahedral, octahedral, and free coordination of the metal borohydride complexes. Of the over 700 investigated structures, about 20 were predicted to form potentially stable alloys with promising decomposition energies. The M(1)(Al/Mn/Fe)(BH(4))(4), (Li/Na)Zn(BH(4))(3), and (Na/K)(Ni/Co)(BH(4))(3) alloys are found to be the most promising, followed by selected M(1)(Nb/Rh)(BH(4))(4) alloys.
Full-text · Article · Aug 2009 · The Journal of Chemical Physics
[Show abstract][Hide abstract]ABSTRACT: Accurate calculations of adsorption energies of cyclic molecules are of key importance in investigations of, e.g., hydrodesulfurization (HDS) catalysis. The present density functional theory (DFT) study of a set of important reactants, products, and inhibitors in HDS catalysis demonstrates that van der Waals interactions are essential for binding energies on MoS(2) surfaces and that DFT with a recently developed exchange-correlation functional (vdW-DF) accurately calculates the van der Waals energy. Values are calculated for the adsorption energies of butadiene, thiophene, benzothiophene, pyridine, quinoline, benzene, and naphthalene on the basal plane of MoS(2), showing good agreement with available experimental data, and the equilibrium geometry is found as flat at a separation of about 3.5 A for all studied molecules. This adsorption is found to be due to mainly van der Waals interactions. Furthermore, the manifold of adsorption-energy values allows trend analyses to be made, and they are found to have a linear correlation with the number of main atoms.
No preview · Article · Apr 2009 · The Journal of Chemical Physics
[Show abstract][Hide abstract]ABSTRACT: We present a practical scheme for performing error estimates for density-functional theory calculations. The approach, which is based on ideas from Bayesian statistics, involves creating an ensemble of exchange-correlation functionals by comparing with an experimental database of binding energies for molecules and solids. Fluctuations within the ensemble can then be used to estimate errors relative to experiment on calculated quantities such as binding energies, bond lengths, and vibrational frequencies. It is demonstrated that the error bars on energy differences may vary by orders of magnitude for different systems in good agreement with existing experience.
Full-text · Article · Dec 2005 · Physical Review Letters
[Show abstract][Hide abstract]ABSTRACT: A grid-based real-space implementation of the Projector Augmented Wave (PAW) method of P. E. Blochl [Phys. Rev. B 50, 17953 (1994)] for Density Functional Theory (DFT) calculations is presented. The use of uniform 3D real-space grids for representing wave functions, densities and potentials allows for flexible boundary conditions, efficient multigrid algorithms for solving Poisson and Kohn-Sham equations, and efficient parallelization using simple real-space domain-decomposition. We use the PAW method to perform all-electron calculations in the frozen core approximation, with smooth valence wave functions that can be represented on relatively coarse grids. We demonstrate the accuracy of the method by calculating the atomization energies of twenty small molecules, and the bulk modulus and lattice constants of bulk aluminum. We show that the approach in terms of computational efficiency is comparable to standard plane-wave methods, but the memory requirements are higher. Comment: 13 pages, 3 figures, accepted for publication in Physical Review B
[Show abstract][Hide abstract]ABSTRACT: The quasicontinuum method is a way of reducing the number of degrees of freedom in an atomistic simulation by removing the majority of the atoms in regions of slowly varying strain fields. Due to the different ways the energy of the atoms is calculated in the coarse-grained regions and the regions where all the atoms are present, unphysical forces called “ghost forces” arise at the interfaces. Corrections may be used to almost remove the ghost forces, but the correction forces are nonconservative, ruining energy conservation in dynamic simulations. We show that it is possible to formulate the quasicontinuum method without these problems by introducing a buffer layer between the two regions of space. The method is applicable to short-ranged potentials in the face-centered cubic, body-centered cubic, and hexagonal close-packed crystal structures.