The Impact of Density Functional Theory on Materials Research

www.mrs.org/bulletin

functional. This functional (i.e., a function whose argument is another function) de- scribes the complex kinetic and energetic interactions of an electron with other elec- Toward Computational trons. Although the form of this functional that would make the reformulation of the many-body Schrödinger equation exact is Materials Design: unknown, approximate functionals have proven highly successful in describing many material properties. Efficient algorithms devised for solving The Impact of the Kohn–Sham equations have been implemented in increasingly sophisticated codes, tremendously boosting the application of DFT methods. New doors are opening to in- Density Functional novative research on materials across physics, chemistry, materials science, surface science, and nanotechnology, and extend- ing even to earth sciences and molecular Theory on Materials biology. The impact of this fascinating development has not been restricted to aca- demia, as DFT techniques also find application in many different areas of in- Research dustrial research. The development is so fast that many current applications could Jürgen Hafner, Christopher Wolverton, and not have been realized three years ago and were hardly dreamed of five years ago. Gerbrand Ceder, Guest Editors The articles collected in this issue of MRS Bulletin present a few of these success stories. However, even if the compu- Abstract tational tools necessary for performing The development of modern materials science has led to a growing need to complex quantum-mechanical calculations relevant to real materials problems understand the phenomena determining the properties of materials and processes on are now readily available, designing a an atomistic level. The interactions between atoms and electrons are governed by the series of simulations that meaningfully laws of quantum mechanics; hence, accurate and efficient techniques for solving the represent a physical or chemical property basic quantum-mechanical equations for complex many-atom, many-electron systems or model a process (e.g., a chemical reac- must be developed. Density functional theory (DFT) marks a decisive breakthrough in tion or phase transformation in a mate- these efforts, and in the past decade DFT has had a rapidly growing impact not only on rial), remains a challenging task, requiring fundamental but also industrial research. This article discusses the fundamental a judicious choice of approximations and principles of DFT and the highly efficient computational tools that have been developed an expert handling of these tools. for its application to complex problems in materials science. Also highlighted are state-of- the-art applications in many areas of materials research, such as structural materials, Designing DFT Calculations for catalysis and surface science, nanomaterials, and biomaterials and geophysics. Materials Science The successful application of DFT to a Keywords: computation, density functional theory, modeling, nanoscale, simulation. materials problem involves three distinct steps: (1) translation of the engineering problem to a computable atomistic model, (2) computation of the required physico- Introduction chemical properties, and (3) validation of During the past decade, computer sim- equation for the complex many-atom, the simulation results by confrontation ulations based on a quantum-mechanical many-electron system is not analytically with laboratory experiments. description of the interactions between elec- solvable, and numerical approaches have trons and atomic nuclei have had an in- become invaluable for physics, chemistry, Translation creasingly important impact on materials and materials science. A breakthrough in A theory that relates computable quan- science, not only in fundamental under- these computational efforts was realized tities to macroscopic engineering behavior standing but also with a strong emphasis in 1964 when Walter Kohn and co- is a prerequisite in order for DFT calcula- toward materials design for future tech- workers developed the density functional tions to be relevant. Before any computa- nologies. The simulations are performed theory (DFT), a theory based on electron tion can be started, significant effort often with atomistic detail by solving the density, which is a function of only three must be expended to translate an engi- Schrödinger equation to obtain energies spatial coordinates.1 The Kohn–Sham neering problem into questions and chal- and forces, require only the atomic num- equations of DFT cast the intractable com- lenges that can be addressed with DFT. bers of the constituents as input, and plexity of the electron–electron interactions This initial step requires considerable in- should describe the bonding between the into an effective single-particle potential sight into the materials science and engi- atoms with high accuracy. The Schrödinger determined by the exchange-correlation neering aspects of the problem.

MRS BULLETIN • VOLUME 31 • SEPTEMBER 2006 659 The Impact of Density Functional Theory on Materials Research

The translation step may be illustrated by differ from the computed result because of LDA and GGA are by far the most com- the application of DFT to the design of other factors that are not characterized monly used functionals today. All func- better materials for rechargeable lithium well in the problem translation. However, tionals exist in spin-degenerate and batteries. Some properties, such as the it should be emphasized that one of the spin-polarized versions. Quite generally, electrode potential for a new material, can most important roles of computational GGA calculations represent a systematic be directly related to total energy differ- modeling is to provide information that is improvement, compared with the LDA, ences2 and are therefore easy to compute. not accessible (or accessible only with great although for heavy elements, a certain But another critical property, such as cycle difficulty) by laboratory experiments. tendency of the GGA to overcorrect the life (the loss of capacity after repeated LDA error is observed. In certain cases charge and discharge), is more difficult to Hierarchy of Exchange-Correlation (prominent examples are the structural predict with DFT, because it can be caused Functionals for DFT and magnetic ground state of iron and the by a variety of poorly identified phenom- DFT reformulates the many-body potential-energy landscape for molecular ena, such as reaction of the material with Schrödinger equation in the form of a set dissociation on a metal surface) the differ- the electrolyte to form an insulating layer, of coupled one-electron equations (the ence is not merely a quantitative one: only mechanical degradation with resulting Kohn–Sham equations), where all of the GGA calculations11 predict iron to be bcc loss of electrical contact, and phase trans- many-body interactions are contained in and magnetic, whereas the LDA predicts formations in the material. Each of these the exchange-correlation functional the ground state to be hexagonal and non- factors requires a different computation. magnetic. Only GGA calculations lead to ៬ Exc [n(r)], (1) the correct height and location of the dis- Computation sociation barrier for a diatomic molecule In many cases, solving the Schrödinger where n(r៬) is the electron density distribu- in contact with the surface of a metal.12 equation is not enough. The length and tion in the crystal. (For a modern intro- Climbing the density functional ladder time scales accessible by DFT calculations duction into DFT, see, for example, further to the meta-GGA or hyper-GGA are defined by the maximal system size Dreizler and Gross3 or the recent book by leads to only moderate progress, com- (typically a few hundred atoms; up to a Martin,4 which also provides a thorough pared with the GGA.13 Hybrid functionals few thousand in special cases) and by the introduction to the theory and practice of enjoy enormous popularity in molecular maximum time span that can be covered electronic structure calculations.) However, chemistry, but in materials science they by ab initio molecular dynamics (MD) sim- the exact form of this functional for which are merely at the test stage. Preliminary re- ulations (several tens of picoseconds). A the Kohn–Sham equations would give ex- sults show great promise for insulating strategy of multiscale simulations must be actly the same ground-state answer as the and semiconducting systems, but also used to translate the results of atomistic many-body Schrödinger equation is not show a need to develop new hybrid func- DFT simulations to real-world scales. As known, and approximations must be used. tionals better suited to describing metals.14 an example, the elementary steps in the Perdew5 has referred to the hierarchy of catalytic oxidation of CO on a metallic sur- approximations developed during the last Basis Sets face (adsorption/desorption/diffusion of two decades as the “Jacob’s ladder of DFT.” Practically, the effective one-electron ᭿ CO and O2, surface oxidation, dissociation In the local density approximation Kohn–Sham equations are nonlinear of oxygen, reaction CO + O → CO2, des- (LDA), the local exchange-correlation en- partial differential equations that are itera- orption of CO2, etc.) at fixed chemical ergy density is taken to be the same as in a tively solved by representing the electronic potentials of the reactants may be treated uniform electron gas of the same density. wave functions by a linear combination of by DFT calculations, and the results must Modern forms of the LDA are based on a set of basis functions. Basis sets fall into be fed into kinetic Monte Carlo simula- the total energy of the homogeneous elec- two classes: plane-wave basis sets and tions to develop a comprehensive picture tron gas derived from quantum Monte local basis sets. Basis-set convergence is of the reaction under realistic pressure con- Carlo calculations. crucial for the accuracy of the calcula- ditions and at real time scales (see also ᭿ The generalized gradient approximations—in particular, for the prediction of the article by Nørskov et al. on catalysis in tion (GGA) introduces a dependence of pressure, stresses, and forces. this issue). Exc on the local gradient of the electron density, ͉∇n(r៬)͉. Many different forms of Plane-Wave Basis Sets: Calculating Validation GGA functionals are available in the liter- Accurate Forces and Stresses After DFT computations are performed, ature, but for materials simulations, it is Basis-set convergence is easily con- validation is an important step. This re- good practice to use parameter-free func- trolled in calculations using plane-wave quires going beyond the computation of tionals derived from known expansion basis sets; it is sufficient to monitor the energies, band structures, and atomic con- coefficients and sum rules of many-body evolution of the results with the cut-off en- figurations to the calculation of experi- theory6,7 and to avoid empirical parame- ergy, the highest kinetic energy associated mentally accessible properties such as terizations popular in molecular quantum with a plane wave. However, to achieve mechanical or thermodynamic properties chemistry. convergence at a tractable computational and electronic or vibrational spectra. ᭿ Meta-GGA functionals use the kinetic effort, the plane-wave basis set must be ei- Discrepancies between computation energy density (or the Laplacian of the ther augmented by solutions of the radial and experimental observation can be electron density) as an additional variable.8 Schrödinger equation in a region sur- attributable to inaccuracies in the DFT ap- ᭿ Hyper-GGA uses Kohn–Sham one- rounding the nucleus (a modern approach proach, but more often point to discrep- electron wavefunctions (instead of the following this line is the full-potential lin- ancies between the experimental and many-body wavefunctions) to evaluate earized augmented plane-wave method, computational situations. An experimen- the Hartree–Fock exchange formula; this LAPW) or by using pseudopotentials to tal result may be influenced by impurities is commonly called “exact” exchange. describe the electron–ion interaction. A not accounted for in the computation, it ᭿ Hybrid functionals9,10 mix exact (i.e., particular advantage of plane-wave calcula- may be kinetically constrained, or it may Hartree–Fock) and DFT exchange. tions is that the calculation of forces acting

660 MRS BULLETIN • VOLUME 31 • SEPTEMBER 2006 The Impact of Density Functional Theory on Materials Research

on the atoms and stresses on the unit cell For molecular, insulating, and semicon- by Vanderbilt19 (which have the merit of is straightforward with the Hellmann– ducting systems, linear-scaling, or O(N), making calculations for d- and f-electron Feynman theorem. This opens the route to techniques enable the performance of metals feasible at an acceptable computa- quantum ab initio MD simulations study- atomistic studies for systems containing tion cost). The criterion for the quality of a ing the time development of a system. thousands of atoms. However, this does pseudopotential is not how well the calcu- not hold if the gap between occupied and lation matches experiment, but how well Local Basis Sets: Toward Linear empty states vanishes, as in metals, for it reproduces the result of accurate all- Scaling which it has turned out to be difficult to electron calculations. Mature codes (see Basis-set convergence is much more dif- reconcile linear scaling and high accuracy. Table I) provide extensive databases of ficult to control if local basis functions (for well-tested pseudopotentials. A certain example, atom-centered orbitals in a lin- Potentials and Pseudopotentials drawback of pseudopotential calculations ear combination of atomic orbitals, or Methods using local basis sets or aug- is that because of the nonlinearity of the LCAO) are used. These approaches have mented plane waves may be constructed exchange interaction between valence been actively pursued as, in principle, a as all-electron methods in the sense that and core electrons, elaborate nonlinear much smaller basis set; typically 10 to 20 the effective one-electron potential is cal- core corrections are required in all sys- basis functions per atom should be suffi- culated using the full set of relaxed core tems where the overlap between the cient to achieve the same accuracy as and valence orbitals. In plane-wave calcu- valence- and core-electron densities is not plane-wave calculations using hundreds lations, the electron–ion interactions must completely negligible. The projector- of plane waves per atom. However, very be described by a pseudopotential, elimi- augmented wave (PAW) method pro- recent extended tests have questioned that nating the need for an explicit treatment of posed by Blöchl20 and adapted by Kresse advantage, indicating that to match the ac- the strongly bound and chemically inert and Joubert 21 avoids the necessity of core curacy of fully converged plane-wave cal- core electrons. Pseudopotentials may also corrections and combines the efficiency culations, very large local basis sets are be used in local orbital methods to reduce of pesudopotentials with the accuracy of required.14 A specific advantage of local the computational effort. The theory of full-potential, all-electron calculations. basis sets is that they enable the imple- pseudopotentials is mature, but the prac- mentation of linear-scaling DFT methods tice of constructing transferable and effi- Codes (i.e., algorithms for which the computa- cient pseudopotentials is far from A variety of mature DFT codes are tional effort increases linearly with the straightforward. Methods for generating nowadays available to the materials sci- number of atoms).15–17 These techniques pseudopotentials include the pseudopo- ence community. Some of them are acces- rely on the fact that the wave function tentials by Troullier and Martins18 (conserv- sible through general public licenses, may be localized and have an exponential ing the norm of the wave function) and whereas others are proprietary. The decay leading to a sparse Hamiltonian. the ultrasoft pseudopotentials introduced codes are based on different choices of

Table I: Commercially or Freely Available Density Functional Theory Codes.

Code Name Basis Set Potentials Web Site

Plane Wave Pseudopotential Codes ABINIT Plane wave Pseudo, PAWa www.abinit.org CASTEP Plane wave Pseudo www.tcm.phy.cam.ac.uk/castep/ CPMD Plane wave Pseudo www.cpmd.org/ Dacapo Plane wave Pseudo dcwww.camp.dtu.dk/campos/Dacapo/ FHImd Plane wave Pseudo www.fhi-berlin.mpg.de/th/fhimd/ PWscf Plane wave Pseudo www.pwscf.org/ VASP Plane wave Pseudo, PAW cms.mpi.univie.ac.at/vasp

Pseudopotential Codes with Other Basis Sets Quickstep Gaussian + plane wave Pseudo cp2k.berlios.de/quickstep.html SIESTA Local/numerical Pseudo www.uam.es/departamentos/ciencias/fismateriac/siesta/

All-Electron Codes CRYSTAL Local all-electron www.crystal.unit.it FPLO Local all-electron www.ifw-dresden.de/agtheo/FPLO/ Gaussian 03 Local all-electron www.gaussian.com ADF Local all-electron www.scm.com DMol3 Local/numerical all-electron people.web.psi.ch/delley/dmol3.html FLAIR LAPWb all-electron www.uwm.edu/~weinert/flair.html QMD-FLAPW LAPW all-electron flapw.com WIEN2K LAPW all-electron www.wien2k.at a PAW ϭ projector-augmented wave method. b LAPW ϭ linearized augmented plane-wave method.

MRS BULLETIN • VOLUME 31 • SEPTEMBER 2006 661 The Impact of Density Functional Theory on Materials Research

basis sets, potentials, exchange-correlation properties. Much current code develop- non-collinear structure of frustrated anti- functionals, and algorithms for solving the ment is focused on the development of ferromagnets where exchange interac- Schrödinger equation. A summary of routines for the post-processing of DFT re- tions and lattice topology are in conflict.36 many existing codes is given in Table I. sults (“tools“) to yield a comprehensive 8. Electronic transport. The development There are currently a large number of description of materials and dynamic of tools for the calculation of electronic pseudopotential codes, all of which have modeling of time-dependent processes in transport processes in nanostructured LDA and GGA functionals and enable materials. Examples of such capabilities materials is a very active field. Particular the performance of static electronic struc- include attention has been paid to transport in sys- ture and total energy calculations as well 1. Crystal structure, phase stability, and tems showing giant magnetoresistance37 as ab initio MD simulations, using either phase diagrams. Whereas the calculation and in molecular devices.38 Born–Oppenheimer or Car–Parrinello dy- of structural free-energy differences for 9. Liquids and glasses. Although diffrac- namics. The all-electron PAW method is polytypes of ordered crystalline materials tion experiments yield the full three- implemented in VASP and ABINIT codes. is a standard task within DFT, the combi- dimensional structure for crystalline Exact exchange and hybrid functionals are nation of ab initio total energy calculations materials, these techniques yield only one- implemented in the newest releases of with genetic algorithms22,23 is leading to dimensional projections of the structure of CPMD and VASP. For the pseudopotential the development of a truly predictive liquids and glasses. Ab initio MD leads to codes, an important point is the availabil- computational crystallography. highly accurate structural prediction for ity of thoroughly tested, accurate, and effi- 2. Mechanical properties. Symmetry- liquids and glasses without the use of em- cient pseudopotentials for all elements. general techniques have been developed pirical force fields.39 Many of these codes come with suites of for the calculation of elastic constants.24 Only a few references are given in this sec- pseudopotentials developed and recom- Intense efforts are being made to calculate tion for lack of space; the reader is referred mended by the code developer. tensile and shear strength and to simulate to the Internet home pages of the most im- All-electron codes exist using local or fracture behavior.25 portant ab initio codes (see Table I). augmented plane-wave basis sets. Both 3. Vibrational spectroscopy. The vibra- Gaussian 03 and CRYSTAL can perform tional eigenstates of molecular and crys- Multiscale Modeling: Bridging the DFT, exact exchange, and hybrid calcula- talline systems can now be calculated Gaps in Time and Length Scales tions, and FPLO is a fully relativistic code either via direct force-constant methods26 It is clear that substantial gaps exist be- for solving the Dirac equation. DMol3 uses or a linear response route,27 including tween the dimensions of even the largest a numerical basis set working either in an quantitative predictions of infrared, systems accessible to DFT simulations and all-electron mode or using semilocal Raman, and inelastic neutron scattering between the maximum time span of ab ini- pseudopotentials. WIEN2K, FLAIR, and intensities. tio MD simulations and the time and QMD-FLAPW are all-electron codes using 4. Surfaces and interfaces. Accurate calcu- length scales of laboratory experiments. the full-potential LAPW method. WIEN2K lation of surface and interface energies To bridge these gaps, strategies for multi- offers the ability to add local orbitals to the enables the prediction of the morphology scale simulations are being developed. basis set to speed up convergence. All- of nanoclusters and precipitate shapes. This task is tackled from different sides. electron codes are important for providing Ab initio MD has made it possible to resolve Configurational free energies of extended benchmarks for pseudopotential codes; even very complex surface reconstructions.28 systems, including substitutional disorder this applies in particular to the full- 5. Adsorption, reactions, and catalysis. in multicomponent systems, can be ob- potential LAPW method in which the Whereas theoretical studies previously tained from cluster-expansion methods, augmented plane-wave basis provides an have been restricted to ideal low-index which extract representative lattice mod- efficient control of basis-set convergence. surfaces, recent advances permit the in- els from DFT that can then be used in Today, calculations with tens of atoms vestigation of molecular adsorption and Monte Carlo simulations to obtain such are extremely fast, with pseudopotential reactions at stepped, kinked, and de- quantities as temperature–composition as well as all-electron codes. Calculations fective surfaces29 and the simulation of phase diagrams, short-range order of dis- for unit cells containing a few hundred catalytic reactions under realistic condi- ordered phases, interfacial free energies, atoms are routine with pseudopotential tions of temperature and pressure. Acid- and precipitate shapes.40,41 The special and PAW codes, but are very demanding catalyzed reactions at the internal surfaces quasi-random structure approach42 makes or even out of reach for most all-electron of functionalized microporous and meso- possible the direct calculation of electronic codes. Using the most efficient codes, very porous materials have been studied using and energetic properties of disordered accurate calculations involving a thou- a variety of DFT techniques.30 systems using small-unit-cell structures sand or more atoms are possible (it should 6. Spectroscopy and dielectric properties. amenable to DFT. These models have been be pointed out that the real figure of merit The possibility of calculating the polariz- implemented with significant success and is the number of valence electrons—the ability of a material via a Berry phase lead to true structure predictions, later current limit is somewhere close to 104 approach has led to a new quality of calcu- verified by experiments.43 In a similar electrons). Linear-scaling calculations may lated optical spectra31 and has opened the spirit, DFT computations of the formation be extended to several thousands of atoms. way toward an ab initio treatment of ferro- energy of intermetallic compounds have electricity32 and piezoelectricity.33 Tools for been used as input in the CALPHAD Tools and Capabilities calculating nuclear magnetic resonance models for the thermodynamics of mate- For many applications in materials sci- spectra34 and for scanning tunneling spec- rials.44 One new area that holds consider- ence, the availability of a basic code for troscopy35 have been developed. able promise is the integration of DFT solving the Schrödinger equation is not 7. Magnetic properties. A fully relativistic with phase-field models to study the evo- sufficient; the resulting energies, wave treatment is required to tackle problems lution of microstructures.45 DFT can be functions, and electron and spin densi- related to magnetic anisotropy; a vector- used to compute such quantities as free ties are merely the starting point for the field treatment of magnetization density is energies, diffusion constants, and interfa- calculations of many different materials necessary for the description of the complex cial energies. These first-principles-predicted

662 MRS BULLETIN • VOLUME 31 • SEPTEMBER 2006 The Impact of Density Functional Theory on Materials Research

quantities can then serve as the thermody- types of capabilities and opportunities Catalysis and Surface Science namic and kinetic driving forces for the presently at hand. Electronic-structure calculations have phase-field models of microstructural been used extensively in the fields of sur- evolution. Mechanical Properties and face science and catalysis. Nørskov et al. Another strategy for solving the length Structural Materials review how DFT is being used to eluci- scale problem is embedding. For a small The article by Ikehata et al. from the date atomic-scale mechanisms for reactive center where atomistic resolution Toyota research group of Takashi Saito de- chemical reactions on surfaces and to and high accuracy are important, a calcu- scribes a specific example of the use of quantitatively describe the rates of hetero- lation at a high level of sophistication is DFT in an alloy design problem, namely, geneous catalysis reactions. These authors performed. The boundary conditions for the search for a low-modulus, high- also provide an intriguing discussion of this calculation are determined by a treat- strength titanium alloy. This article pro- the future prospects of the calculations to ment of the area surrounding the center at vides an excellent illustration of the power predict new catalysts and, hence, provide a lower level. Examples for this approach of a combined theoretical and experimen- a rational guide for catalyst development. are the quantum-mechanical/molecular tal approach directed toward the search DFT techniques are also applied to the in- modeling approach used in heteroge- for a material with specified properties. vestigation of acid-based catalysis by mi- neous catalysis;46 the description of crack Applications of DFT to structural mate- croporous materials such as zeolites. An propagation using DFT to simulate the rials also exist in steels, notably, the work interesting innovation is the production crack tip, embedded in an atomistic force- of Olson and co-workers,52 who use vari- of ε-caprolactam (a precursor molecule to field description of the strain field at in- ous computational materials techniques to the polymer nylon-6) by conversion of termediate distances; and a continuum accelerate the design process of new high- cyclohexanone (the so-called Beckmann approach on a macroscopic scale. A partic- strength steels. The Virtual Aluminum rearrangement) in a zeolite or a struc- ularly attractive version of this embed- Castings program at Ford Motor Com- turally related aluminophosphate,53,54 a re- ding scheme is the “learn-on-the-fly” pany involves the use of DFT and its con- action that is environmentally much more strategy,47 where local quantum computa- nection to higher-length-scale modeling benign than the conventional route tions are used to continually tune the pa- tools44,45 directed toward improved through homogeneous catalysis in sulfu- rameters of the force field used for the processing and enhanced properties of ric acid. Figure 2 shows the final step in classical part of the simulations. aluminum cast alloys. the formation of ε-caprolactam, the hy- Time-scale gaps are caused by energy A novel MD approach combining drolysis of an intermediate carbiminium barriers for chemical reactions and first- quantum-mechanical embedding with ion in the pore of a zeolite. order structural phase transitions. Rare classical force-field models has been used events such as activated chemical reactions by Csanyi et al.47 to model crack propaga- Magnetism and Magnetic Materials can be treated within transition-state tion in silicon using a model containing The enormous increase in the recording sampling methods;48 temperature- about 200,000 atoms (see Figure 1). About density of hard disk drives by eight orders accelerated MD methods49,50 or meta- 300 atoms near the crack tip are treated of magnitude during the past 50 years (see dynamic techniques.51 This allows one to quantum-mechanically, and this is essen- the May 2006 issue of MRS Bulletin on identify novel phases under extreme ther- tial for a correct description of the (2 × 1) “Materials for Magnetic Data Storage”) modynamic conditions and to trace out periodicity of the reconstruction of the has been achieved by downsizing the bits transformation pathways for structural crack surface. Although this method has recorded in the magnetic storage layer. phase transitions. Kinetic Monte Carlo currently been applied to Si, future exten- This approach is, however, limited by the simulations can be used to simulate sions of this method to structural materi- onset of superparamagnetism, which oc- growth processes or chemical reactions on als could produce useful insight on crack curs when the grain size in the recording regular lattices under realistic conditions propagation in a variety of materials types. medium is reduced to an extent that the of temperature. magnetic energy stored in a grain be- comes comparable to the thermal energy. Hardware Recently, based on DFT calculations, DFT simulations of materials do not re- tetragonal iron cobalt alloys were pro- quire supercomputer resources. A typical posed as promising materials with the de- hardware base for such simulations sired high uniaxial magnetic anisotropy consists of a PC cluster with fast proc- and high saturation magnetization.55 Sta- essors equipped with sufficient memory bility of the tetragonal FeCo phase can be (Ն1 Gbyte) and a fast interconnect. Mod- achieved in FeCo/Pt superlattices (see ern DFT codes offer efficient simulation Figure 3); again, DFT calculations turned of medium-sized systems on clusters of out to be essential.56 This is a good example about 32 CPUs. of a real computer-aided materials design. Applications of DFT in Materials Oxides and Minerals Research DFT works quite well on oxides of al- This issue of MRS Bulletin collects a di- kali, alkaline-earth, and main group ele- versity of DFT success stories from such ments. In their article, Brodholt and Figure 1. Propagation of a (111)[110] wide-ranging materials research topics as crack in silicon, modeled using the Vocadlo describe the uses of DFT in geol- surface science and catalysis, earth sci- “learn-on-the-fly” scheme mixing ogy and the earth sciences. Calculations ences, nanotechnology, and structural quantum and classical simulations. have the ability to simulate conditions in metals. Of course, any such list of ex- Note the (2 × 1) periodicity of the the Earth’s interior that would otherwise amples will necessarily be incomplete, but reconstruction appearing on the upper be quite problematic to reproduce in the it should give the reader a sense of the opening surface. After Csanyi et al.47 laboratory. Applications of DFT have

MRS BULLETIN • VOLUME 31 • SEPTEMBER 2006 663 The Impact of Density Functional Theory on Materials Research

and dynamics of the retinal chromophore of rhodopsin in different environments. Figure 4 shows the DFT-optimized structure of the chromophore binding site. One of the exploratory studies based on a linear-scaling code presents a total energy calculation of a piece of DNA consisting of more than 2500 atoms.17 High-Throughput Ab Initio Calculations High-throughput (combinatorial) ex- perimentation has played an increasingly Figure 2. (a) Initial, (b) transition, and (c) final states of the hydrolysis of a carbiminium ion to important role in materials science. By form N-protonated ε-caprolactam, catalyzed by an acid zeolite. After Bucko et al.54 studying large numbers of systems, one can screen combinatorial spaces for new systems with desired properties. Modern predicted such fundamental properties as lithium-battery electrode material with DFT codes have now become very robust, 43 the properties of iron and (Mg,Fe)SiO3 higher energy and power density. In this enabling many tasks to be performed au- under the extreme pressures and temper- field, DFT computations are now com- tomatically. Combined with the increasing atures of the Earth’s interior. DFT calcula- monplace in industry and academic labs. speed of computers, this opens the way tions have also recently been shown to put toward high-throughput computation.60 bounds on the possible composition of the Semiconductors and Pioneering applications have been de- Earth’s core and have identified a new Nanotechnology voted to crystal structure predictions61 and 62 phase of (Mg,Fe)SiO3 just above the core. Semiconductors probably form the class the search for stable quaternary alloys. The application of DFT in transition-metal of materials that has seen the most oxides requires considerably more care, as widespread use of DFT. Calculations of Summary and Outlook the spurious self-interaction in DFT overly electronic properties for known semicon- DFT is the most detailed “microscope” delocalizes electrons, leading to excessive ductors, such as bandgaps and defect currently available to glance into the metallic behavior and the washing out of states, are now commonplace and have atomic and electronic details of matter. As mixed-valence states in doped oxides, been performed for virtually all unary and such, it has catalyzed a renewed interest affecting local spin magnetism, conduc- binary semiconductor systems as well as into the fundamental science, chemistry, tivity, and even phase stability. Despite for a wealth of higher-order compounds, and physics of materials. Its impact is these limitations, DFT and enhanced DFT ordered and disordered alloys, dopant likely to continue to grow through the fur- methods can be used to assist in the ra- types, and surface structures. In his contri- ther development of tools that build upon tional design of new materials. A good ex- bution to this issue, Marzari reviews the DFT to model higher-level properties not ample can be found in the search for a applications of DFT to semiconductor directly accessible to DFT because of time- nanotechnology, concentrating on Si-H scale, length-scale, or complexity issues. nanostructures, single-walled carbon nano- Much of this application-driven modeling tubes, and self-assembled monolayers. is likely to be performed by people with The combined investigation, based on ex- field-specific knowledge in cooperation periment and DFT calculations, of the ad- with the DFT practitioners and code devel- sorption of large organic molecules on opers; therefore, educating the materials metallic and semiconducting surfaces, the formation of ordered structures through a self-assembly process,57 and the electronic transport process in these molecular de- vices38 is an essential contribution toward the development of molecular electronics. Biomaterials DFT calculations play an increasingly important role for understanding complex processes in biological materials. For example, the protonation of the retinal chromophore is of crucial importance for Figure 3. Schematic of the tetragonal the visual process. Protonation leads to the “3/7” superlattice of an Fe0.36Co0.64 alloy formation of a highly polarizable π-system Figure 4. Schematic of the DFT and Pt, showing a huge perpendicular required for long-wavelength adsorbance energy-optimized structure of the magnetic anisotropy energy (the lattice and wavelength regulation by the protein chromophore binding site in rhodopsin parameters are C ϭ0.317 nm and of a protonated 11-cis-retinal Schiff FeCo environment, and deprotonation is re- CPt ϭ0.417 nm). This superlattice has base (Lys296) in contact with a been proposed on the basis of DFT quired for the protein to reach the activat- glutamate ion (Glu113), an amino acid calculations and experimentally ing state after light adsorption. Sugihara (Thr94), and a water molecule (Wat2b). realized by sputter deposition. After et al.58 and Röhrig et al.59 have applied HI, HIII, and HVII are regions in Andersson et al.55 DFT techniques to investigate the structure mitochondrial DNA. After Sugihara et al.58

664 MRS BULLETIN • VOLUME 31 • SEPTEMBER 2006 The Impact of Density Functional Theory on Materials Research

scientists of the future through new text- 3. R.M. Dreizler and E.K.U. Gross, Density 36. D. Hobbs, G. Kresse, and J. Hafner, Phys. books, online courses, and tutorials is Functional Theory (Springer, Berlin, 1990). Rev. B 62 (2000) p. 11556. a crucial aspect of increasing the impact 4. R.M. Martin, Electronic Structure Theory: Basic 37. J. Kudrnovsky, V. Drchal, C. Blaas, P. Wein- of DFT. Theory and Practical Methods (Cambridge Uni- berger, I. Truek, and P. Bruno, Phys. Rev. B 62 Despite all of its successes, DFT is not versity Press, Cambridge, UK, 2004). (2000) p. 15084. accurate for all problems. We have already 5. J.P. Perdew and K. Schmidt, in Density Func- 38. K. Stokbro, J. Taylor, M. Brandbyge, and P. tional Theory and its Application to Materials, ed- mentioned the continuing effort to con- Ordejon, Ann. N.Y. Acad. Sci. 1006 (2003) p. 212. ited by V. van Doren, C. van Alsenoy, and P. 39. H.W. Sheng, W.K. Luo, F.M. Alamgir, J.M. struct exchange-correlation functionals Geerlings (American Institute of Physics, Bai, and E. Ma, Nature 439 (2006) p. 419. leading to improved accuracy. Although Melville, NY, 2001) p. 1. 40. D. de Fontaine, Solid State Physics, Vol. 47, DFT does an excellent job on the ground- 6. J.P. Perdew, J.A. Chevary, S.H. Vosko, K.A. edited by H. Ehrenreich and D. Turnbull (Aca- state properties of metals and semicon- Jackson, M.R. Pederson, D.J. Singh, and C. demic Press, New York, 1994) p. 33. ductors, the “bandgap” problem (i.e., the Fiolhais, Phys. Rev. B 46 (1992) p. 6671. 41. J.W. Wang, C. Wolverton, S. Müller, Z.K. difficulty of achieving a quantitatively ac- 7. J.P. Perdew, K. Burke, and M. Ernzerhof, Liu, and L.Q. Chen, Acta Mater. 53 (2005) p. 2759. 77 curate prediction of the gap between the Phys. Rev. Lett. (1996) p. 3865. 42. S.H. Wei, L.G. Ferreira, J.E. Bernard, and A. 8. J. Tao, J.P. Perdew, V.N. Staroverov, and G.E. highest occupied and the lowest empty Zunger, Phys. Rev. B 42 (1990) p. 9622. Scuseria, Phys. Rev. Lett. 91 146401 (2003). 43. G. Ceder, Science 280 (1998) p. 1099. eigenstates) and the inability to describe 9. J.P. Perdew, M. Ernzernhof, and K. Burke, the physics of van der Waals interactions J. Chem. Phys. 105 (1996) p. 9982. 44. C. Wolverton, X.-Y. Yan, R. Vijayaraghavan, are notorious. Also, the properties of 10. J. Heyd, G.E. Scuseria, and M. Ernzerhof, and V. Ozolins, Acta Mater. 50 (2002) p. 2187. strongly correlated systems such as J. Chem. Phys. 118 (2003) p. 8207. 45. V. Vaithyanathan, C. Wolverton, and L.Q. transition-metal oxides are not well repro- 11. E.G. Moroni, G. Kresse, J. Hafner, and Chen, Acta Mater. 52 (2004) p. 2973. duced by common exchange-correlation J. Furthmüller, Phys. Rev. B 56 (1997) p. 15629. 46. M. Sierka and J. Sauer, J. Chem. Phys. 112 functionals. Evidently, explorations of the 12. A. Gross, Surf. Sci. Rep. 32 (1998) p. 291. (2000) p. 6983. many-body aspects of electron correlation 13. V.N. Staroverov, G.E. Scuseria, J. Tao, and 47. G. Csanyi, T. Albaret, M.C. Payne, and A. De Vita, Phys. Rev. Lett. 93 175503 (2004). beyond current DFT methods are needed, J.P. Perdew, Phys. Rev. B 69 075102 (2004). 14. J. Paier, R. Hirschl, M. Marsman, and G. 48. C.D. Dellago, P.G. Bolhuis, F.S. Cajka, and and this is currently a very active field of Kresse, J. Chem. Phys. 122 234102 (2005). D. Chandler, J. Chem. Phys. 108 (1998) p. 1964. research. Many-body perturbation theory 15. J.M. Soler, E. Artacho, J.D. Gale, A. Garcia, 49. K.C. Hass, W.F. Schneider, A. Curioni, and (the GW method) corrects most of the J. Junquera, P. Ordejon, and D. Sanchez- W. Andreoni, Science 282 (1998) p. 265. bandgap problem in DFT, even for diffi- Portal, J. Phys.: Condens. Matter 14 (2002) 50. F. Montalenti, M.R. Sorensen, and A.F. cult cases such as narrow-gap semicon- p. 2745. Voter, Phys. Rev. Lett. 87 126101 (2001). ductors.63 LDA+U approaches combine a 16. F.R. Krajewski and M. Parrinello, Phys. Rev. 51. A. Laio and M. Parrinello, Proc. Natl. Acad. DFT description of extended states with a B 71 233105 (2005). Sci. USA 99 (2002) p. 12562. Hartree–Fock-like treatment of the 17. C. Skylaris, P.D. Haynes, A.A. Mostofi, and 52. S. Hao, W.K. Liu, B. Moran, F. Vernery, and Coulomb repulsion U between states in M.C. Payne, J. Chem. Phys. 122 084119 (2005). G.B. Olson, Computer Methods in Appl. Mechan. 18. N. Troullier and J.L. Martins, Phys. Rev. B 43 narrow d- and f-bands, and lead to a sig- Eng. 193 (2004) p. 1865. (1991) p. 1993. 53. R. Mokaya and M. Poliakoff, Nature 437 nificant improvement in the structural, 19. D. Vanderbilt, Phys. Rev. B 41 (1990) p. 7892; (2005) p. 1243. electronic, magnetic, and optical proper- G. Kresse and J. Hafner, J. Phys.: Condens. Matter 64 54. T. Bucko, J. Hafner, and L. Benco, J. Phys. ties of transition-metal oxides and other 6 (1994) p. 8245. Chem. A 108 (2004) p. 388. systems where electron localization is 20. P.E. Blöchl, Phys. Rev. B 50 (1994) p. 17953. 55. T. Burkert, L. Nordström, O. Eriksson, and strong. Dynamic mean-field theory 21. G. Kresse and D. Joubert, Phys. Rev. B 59 O. Heinonen, Phys. Rev. Lett. 93 (2004) p. 27203. (DMFT) may be considered as a many- (1999) p. 1758. 22. D.M. Deaven and K.M. Ho, Phys. Rev. Lett. 56. G. Andersson, T. Burkert, P. Warnicke, body extension of DFT that attempts to M. Björck, B. Sanyal, C. Chacon, C. Zlotea, L. capture the true many-body physics of the 75 (1995) p. 288. 23. A.R. Oganov, C.W. Glass, and S. Ono, Earth Nordström, P. Nordblad, and O. Eriksson, Phys. electron–electron interaction. For strongly Rev. Lett. 96 037205 (2006). + Planet. Sci. Lett. 241 (2006) p. 95. correlated systems, LDA DMFT agrees 57. M. Böhringer, K. Morgenstern, W.D. + 24. Y. LePage and P. Saxe, Phys. Rev. B 63 with LDA U, but it subsumes the LDA for 174103 (2001). Schneider, R. Berndt, F. Mauri, A. De Vita, and weakly correlated metals and achieves a 25. M. Černy, J. Pokluda, M. Šob, M. Friák, and R. Car, Phys. Rev. Lett. 83 (1999) p. 324. correct description at arbitrary strength of P. Šandera, Phys. Rev. B 67 035116 (2003). 58. M. Sugihara, V. Buss, P. Entel, and J. Hafner, the onsite Coulomb repulsion.65 At this 26. G. Kresse, J. Furthmüller, and J. Hafner, J. Phys. Chem. B 108 (2004) p. 3673. point, no unified “post-DFT” theory ex- Europhys. Lett. 32 (1995) p. 729. 59. U.F. Röhrig, L. Guidoni, and U. Röthlis- ists, and practitioners will need to decide 27. P. Gianozzi, S. de Gironcoli, P. Pavone, and berger, Chem. Phys. Chem. 6 (2005) p. 1836. with care which improvement strategy is S. Baroni, Phys. Rev. B 43 (1991) p. 7231. 60. D. Morgan, G. Ceder, and S. Curtarolo, 28. G. Kresse, M. Schmid, E. Napetschnig, M. best suited for a property that cannot be Meas. Sci. Technol. 16 (2005) p. 296. Shishkin, L. Köhler, and P. Varga, Science 308 61. S. Curtarolo, D. Morgan, K. Persson, accurately computed at the DFT level. Fi- (2005) p. 1440. nally, the recent development of a highly J. Rodgers, and G. Ceder, Phys. Rev. Lett. 91 29. B. Hammer, Top. Catal. 37 (2006) p. 3. 135503 (2003). efficient quantum Monte Carlo approach66 30. J. Hafner, L. Benco, and T. Bucko, Top. Catal. 62. G.H. Johannesson, T. Bligaard, A.V. Ruban, 37 could open the way toward materials sim- (2006) p. 41. H.L. Skriver, K.W. Jacobsen, and J.K. Nørskov, 31. M. Gajdos, K. Hummer, G. Kresse, J. Furth- ulations with an unprecedented level of Phys. Rev. Lett. 88 255506 (2002). müller, and F. Bechstedt, Phys. Rev. B 73 045112 accuracy and sophistication. 63. A. Fleszar and W. Hanke, Phys. Rev. B 71 (2006). 32. D. Vanderbilt, Ferroelectrics 301 (2004) p. 9. 045207 (2005). References 33. D.R. Hamann, K.M. Rabe, and D. Vander- 64. A. Rohrbach, J. Hafner, and G. Kresse, Phys. 1. P. Hohenberg and W. Kohn, Phys. Rev. B 136 bilt, Phys. Rev. B 72 033102 (2005). Rev. B 70 125426 (2004). (1964) p. 864; W. Kohn and L.J. Sham, Phys. Rev. 34. C.J. Pickard and F. Mauri, Phys. Rev. B 63 65. A. Georges, G. Kotliar, W. Krauth, and M.J. A 140 (1964) p. 1133. 245101 (2001). Rozenberg, Rev. Mod. Phys. 68 (1996) p. 13. 2. M.K. Aydinol, A.F. Kohan, and G. Ceder, 35. W.A. Hofer, A.S. Foster, and A.L. Shluger, 66 F.A. Reboredo and A.J. Williamson, Phys. ٗ .(J. Power Sources 68 (1997) p. 664. Rev. Mod. Phys. 75 (2003) p. 1287. Rev. B 71 121105 (2005

MRS BULLETIN • VOLUME 31 • SEPTEMBER 2006 665 The Impact of Density Functional Theory on Materials Research

Jürgen Hafner Christopher Wolverton Gerbrand Ceder John P. Brodholt Hideaki Ikehata

Jürgen Hafner, Materials Science, has co-authored more Institute of Technology, Guest Editor for this Sensengasse 8/12, than 60 peer-reviewed Room 13-5056, 77 Mass- issue of MRS Bulletin, A-1090 Wien, Austria; publications and holds achusetts Ave., Cam- is a professor of Solid- tel. 43-4277-51400 (di- two patents, with bridge, MA 02139, USA; State Sciences at the rect) and 43-4277-51401 several others pending. tel. 617-253-1581, fax University of Vienna (secretary), fax 43-4277- Wolverton can be 617-258-6534, and and director of the 9514, and e-mail juer- reached at Ford Motor e-mail [email protected]. Graduate College for [email protected]. Company, MD Computational 3083/SRL, PO Box 2053, John P. Brodholt is a Materials Science. He Christopher Wolverton, Dearborn, MI 48121, professor of mineral graduated in 1973 from Guest Editor for this USA; tel. 313-390-9552, physics in the Depart- Vienna Technical issue of MRS Bulletin, is fax 313-323-1129, and ment of Earth Sciences University, where he a technical leader at e-mail cwolvert@ at University College also was appointed Ford Research and ford.com. London. The focus of Matthias Scheffler professor of theoretical Innovation Center, his research is computa- physics in 1982 after where he leads the Gerbrand Ceder, tional mineral physics metallic materials and serving for several years Hydrogen Storage and Guest Editor for this applied to the Earth’s theoretical predictions as a research associate at Nanoscale Modeling issue of MRS Bulletin, is interior. Brodholt was of the physical proper- the Max Planck Institute Group. He received his the R.P. Simmons awarded the European ties of metals by first- for Solid-State Sciences. PhD degree in physics Professor of Materials Mineralogical Union principles calculations. In 1998, he accepted a from the University of Science and Engineering Medal for Research Ikehata can be chair at the Faculty California at Berkeley. at the Massachusetts Excellence in 2002 and reached at Toyota of Physics of the After completing his Institute of Technology. was elected as a fellow Central R&D Labs Inc., University of Vienna. PhD degree, Wolverton He holds an engineering of the Mineralogical So- 41-1 Yokomichi, Na- Hafner’s research performed postdoctoral degree from the Univer- ciety of America in 2005. gakute, Aichi 480-1192, concentrates on the work at the National sity of Leuven, Belgium, He has published more Japan; tel. 81-561-63- application of density Renewable Energy and a PhD degree in than 90 papers in 4660, fax 81-561-63-6156, functional methods to a Laboratory (NREL). materials science from his field. and e-mail ikehata@ wide range of problems His research interests the University of Brodholt can be mosk.tytlabs.co.jp. in materials science, include computational California at Berkeley. reached at University from magnetism at the studies of a variety of Ceder’s research College London, De- Shigeru Kuramoto is a nanoscale to surface automotive and interests include first- partment of Earth Sci- researcher in the Toyota science and catalysis. He environmentally principles predictions of ences, Gower Street, Central R&D Labs Inc. has received the Ludwig friendly materials via materials properties and London, WC1E 6BT, (TCRDL). He received a Boltzmann Prize from first-principles atomistic structure, with a focus United Kingdom; tel. DEng degree from the the Austrian Physical calculations and on materials for energy 44-0-20-7679-2622, fax University of Tokyo Society and the Erwin methodologies for generation and storage. 44-0-20-7388-7614, (UT) in 1994, focusing Schrödinger Award of linking atomistic and He has received the and e-mail j.brodholt@ on deformation and the Austrian Academy microstructural scales. Battery Research Award ucl.ac.uk. fracture of aluminum al- of Sciences. Also, Hafner Current topics of inter- from the Electrochemi- loys at low tempera- has authored more than est include the discov- cal Society, the Career Hideaki Ikehata is a re- tures. Prior to joining 500 publications in ery of novel hydrogen Award from the searcher in the Toyota TCRDL in 2002, refereed international storage reactions, phase National Science Foun- Central R&D Labs Inc. Kuramoto worked as a journals. transformations in dation, and the Robert (TCRDL). He received researcher in Furukawa Hafner can be metallic and oxide Lansing Hardy Award his MS degree from Electric Co. Ltd. and as reached at Universität alloys, microstructural from TMS. Kyoto University and a research associate at Wien, Institut für evolution during heat Ceder can be reached at subsequently joined UT. His research Materialphysik and treatment, and the theo- the Department of Ma- TCRDL in 1998. His interests are deforma- Center for retical prediction of new terials Science and Engi- research interests tion and fracture in Computational materials. Wolverton neering, Massachusetts include strengthening of metallic materials.

666 MRS BULLETIN • VOLUME 31 • SEPTEMBER 2006 The Impact of Density Functional Theory on Materials Research

Shigeru Kuramoto Nicola Marzari Naoyuki Nagasako Jens K. Nørskov Takashi Saito

application to materials member of the Royal Haber Institute of the design. Danish Academy of Max Planck Society in Nagasako can be Science and Letters, and Berlin since 1988, and a reached at Toyota Cen- his recent honors in- Distinguished Visiting tral R&D Labs Inc., 41-1 clude the 2005 Innova- Professor at the Yokomichi, Nagakute, tion Prize from DTU University of California Aichi 480-1192, Japan; and an honorary doctor- in Santa Barbara since tel. 81-561-63-4537, fax ate from the Technical 2005. His research areas 81-561-63-5426, and University of Eindhoven include ab initio theories e-mail nagasako@mosk. in 2006. of electronic structure, tytlabs.co.jp. Nørskov can be in particular, density reached at CAMP, De- functional theory, Jens K. Nørskov is a partment of Physics, quantum Monte Carlo Hervé Toulhoat Lidunka Voc`´adlo professor of physics in B. 307, Technical Univer- and self-energy the Department of sity of Denmark, methods, atomistic Kuramoto can be of novel electronic- Physics at the Technical DK-2800 Lyngby, Den- thermodynamics, and reached at Toyota Cen- structure approaches to University of Denmark mark; tel. 45-4525-3175, statistical mechanics. tral R&D Labs Inc., 41-1 characterize and design (DTU). Also, he is direc- fax 45-4593-2399, and Various hybrid Yokomichi, Nagakute, advanced materials and tor of the Center for e-mail norskov@fysik. methods have been Aichi 480-1192, Japan; devices. In 2005, he was Atomic-Scale Materials dtu.dk. developed that tel. 81-561-63-4660, fax the subject editor for Physics, the director of allow the study of 81-561-63-6156, and electronic structure in NanoDTU, and the Takashi Saito is a direc- mesoscopic systems and e-mail kuramoto@mosk. the Handbook of Materials chair of the Danish Cen- tor at the Toyota Central time scales from pi- tytlabs.co.jp. Modeling (Springer). ter for Scientific Com- R&D Labs Inc. (TCRDL). coseconds to seconds. Marzari can be puting. Nørskov He earned a DEng de- Also, Scheffler has Nicola Marzari is reached at the Massa- received his PhD degree gree from Tohoku Uni- investigated semicon- associate professor of chusetts Institute of in physics in 1979 from versity in 1979, focusing ductors, metals, computational materials Technology, Department the University of Aarhus on the calculation of transition-metal oxides, science in the Depart- of Materials Science and in Denmark. Before his phase diagrams and carbon nanostructures, ment of Materials Engineering, Room 13- appointment at DTU in phase transformations. transition-metal Science and Engineering 5066, 77 Massachusetts 1987, he held research After receiving his de- clusters, and at the Massachusetts In- Ave., Cambridge, MA positions at the Univer- gree, he joined TCRDL biomolecules. stitute of Technology. 02139-4307, USA; tel. sity of Aarhus, IBM in in 1980. He was named a He holds a Laurea 617-452-2758, fax 617- Yorktown Heights, the His research interests fellow of the American degree in physics from 258-6534, and e-mail Nordic Institute for include mechanical and Physical Society in 1998 the University of Trieste [email protected]. Theoretical Physics in physical properties of and was honored with in Italy and a PhD de- Copenhagen, and Hal- high-performance the Medard W. Welch gree in physics from the Naoyuki Nagasako is a dor Topsoe. In 1993, he metallic materials. Medal and Prize in University of researcher in the Toyota co-founded DTU’s Saito can be reached 2003 and the Max Born Cambridge. Before join- Central R&D Labs Inc. Center for Atomic-Scale at Toyota Central R&D Medal and Prize ing MIT, he was an NSF (TCRDL). He joined Materials Physics. Labs Inc., 41-1 in 2004. postdoctoral fellow at TCRDL after his receiv- Nørskov’s research Yokomichi, Nagakute, Scheffler can be Rutgers University and ing his MS degree from interests include theoret- Aichi 480-1192, Japan; reached at Fritz-Haber- a research scientist with Hiroshima University in ical surface physics, tel. 81-561-63-3090, fax Institut der Max-Planck- the Naval Research 1997. His research inter- heterogeneous catalysis, 81-561-63-5441, and Gesellschaft, Laboratory and with ests include theoretical nanostructures and ma- e-mail saito@mosk. Faradayweg 4-6, Princeton University. studies on the physical terials properties, tytlabs.co.jp. D-14195 Berlin-Dahlem, Marzari’s research is properties of materials biomolecules, electro- Germany; tel. 49-30- dedicated to the devel- by first-principles chemistry, and hydro- Matthias Scheffler has 8413-4711 and fax 49-30- opment and application calculations and their gen storage. He is a been director of the Fritz 8413-4701.

MRS BULLETIN • VOLUME 31 • SEPTEMBER 2006 667 The Impact of Density Functional Theory on Materials Research

Hervé Toulhoat is Chimie de Paris in 1976. Toulhoat can be reader in mineral Great Britain and deputy scientific direc- Toulhoat is a member of reached at Institut physics in the Ireland. tor of the Institut the Editorial Board of Français du Pétrole, Department of Earth Vo c`´adlo can be Français du Pétrole (IFP) the Journal of Catalysis 1–4 avenue de Sciences at University reached at University and the director of doc- and an associate Bois-Préau, 92852 College London. College London, toral studies at IFP member of the IUPAP Rueil-Malmaison Cedex, Her research interests Department of Earth Sci- include the Earth’s core School in Rueil- Commission on France; tel. 33-1-47-52- ences, Gower Street, Malmaison, France. He Computational Physics. 73-50, fax 33-1-47- and planetary ices. She London, WC1E 6BT, earned a PhD degree in He has authored more 52-70-22, and e-mail was awarded the United Kingdom; chemical engineering than 100 papers and [email protected]. Doornbos Prize for from the Paris School of holds 20 patents, mostly Research Excellence in tel. 44-20-7679-7919, Mines in 1980, and an dealing with catalysis Lidunka Voc`´adlo is a 1998 and the 2004 Max fax 44-20-7679-2685, MSc degree from Ecole and processes in Royal Society University Hey Medal of the and e-mail l.vocadlo@ ٗ .Nationale Supérieure de oil refining. Research Fellow and Mineralogical Society of ucl.ac.uk

CALL FOR PAPERS 2007 MRS SPRING MEETING APRIL 9-13 • SAN FRANCISCO, CA www.mrs.org/spring2007

Legendary Heritage In High Performance Low Mass Materials NON RCF,

Al2O3

Al2O3- SiO2

CaSiO3 Compositions 30 years of good science - an extensive selection of casthouse consumables help Rigid Boards nonferrous metals producers make better Precision- metal - more proﬁ tably. Machined Shapes Blankets Papers Cement ZIRCAR Ceramics, Inc. Temperatures 100 N. Main St. Experienced people who care about your productivity 6000C to bring you a world class range of ceramic ﬁ ber based Florida, New York 10921 USA 0 thermal and electrical insulation solutions. 1900 C Tel: (1) 845 651 6600 www.zircarceramics.com [email protected] Fax: (1) 845 651 0441

For more information, see http://www.mrs.org/bulletin_ads

668 www.mrs.org/bulletin MRS BULLETIN • VOLUME 31 • SEPTEMBER 2006