Our research is focused on theoretical and computational modeling of materials with particular interest in understanding mechanisms that control epitaxial growth and morphological evolution, friction and adhesion, and chemical reactivity of nanostructured surfaces and nanoparticles.  Very recently we have also ventured into developing the tools for understanding and modeling the properties of biomaterials like peptides and proteins. The importance of this field is technological (thin film growth, nanotechnology for drug delivery, novel materials, catalysis, corrosion, lubrication, etc.) and also fundamental.  It raises questions about the nature of the bonding between atoms and molecules in regions of low symmetry and complex local environment and how this bonding is affected by the electronic structure, microscopic geometry, atomic coordination and elemental characteristics of the atoms and molecules.  In addressing these and related questions about the electronic structure, a distinct and important aspect of our work is also to probe the temperature dependencies of the properties, as accurately as feasible, so as to understand the behavior in laboratory environments.  To achieve this our goal is to develop the framework for multiscale modeling of materials in which comprehensive understanding developed at the atomic scale provides input parameters and physical insights for further examination of systems at larger length and time scale (mesoscopic). These studies have considerable predictive power and are expected provide the knowledge base necessary for tailoring functional materials by design.

The field as such has experienced phenomenal development in the past years because of the coupling of experimental techniques like Scanning Tunneling and Atomic Force Microscopy, which probe atoms in the real space, to the existing multitude of powerful spectroscopic and scattering methods which provide information in the momentum space. As a result, intriguing and systematic experimental data on the structure, dynamics, growth, morphological evolution and reactivity of surfaces and nanoparticles continue to become available.  Our theoretical work in this area, essential for the interpretation of some fascinating experimental data, proceeds on several fronts.  At the ground level, many-body, semi-empirical interaction potentials are used to understand both energetics and dynamics of systems of interest.  Such calculations have the advantage that, with the help of modern high-performance computers, they can be applied to systems consisting of several thousands to hundreds of thousands of atoms, thereby making the application feasible for realistic systems i.e. ones with defects, impurities and multi-components. With this approach we have examined the temperature dependent characteristics of surfaces with a wide range of geometries (flat, stepped and kinked) and elemental compositions (pristine metals, alloys), with and without adatoms, vacancies, and their clusters, and of metallic nanocrystals.  These calculations include the full dynamics of the system and have important bearing on thin film growth process, and on the stability and thermodynamics of surfaces and nanoparticles.  They have also provided the guidelines for more sophisticated calculations based on ab initio methods, as discussed below.

Our second approach is based on ab initio quantum mechanical calculations and offers as accurate a technique as is currently feasible for complex systems.  Here the electronic structure of materials and heterogeneous structures, and a limited set of their dynamics, are examined through the usage of density functional theory with reliable approximations for the exchange-correlation terms for treating the valence electrons, while resorting to ab initio pseudopotentials methods for dealing with the core electrons.  For systems requiring a more detailed analysis we resort to all-electron methods within the density functional theory.  Presently such methods are feasible for systems consisting of a few hundred atoms.  The challenge in this area is to make such schemes tractable for complex, multi-component, functional, and active nanosystems.  Such developments are expected to pave the way for comprehending processes like catalysis, corrosion, and a range of novel phenomenon at the nanoscale, from microscopic principles.  Recently, we have developed ab initio electronic structure code which are based on real (coordinate) space and do not require the system to be inherently periodic.  This code is particularly apt for modeling systems with disorder, multi-components and/or reduced symmetry.

By combining ab initio electronic structure calculations of the system energetics with kinetic Monte Carlo simulations, we are carrying out selective studies of the temporal and thermal evolution of chemisorbed metal and oxide surfaces and nanoparticles.  To properly map out the phase diagram for chemical reactions on nanostructured surfaces, for a range of gas pressures, we also calculate the Gibbs free energy and thus incorporate ab initio input into calculations of the system thermodynamics.   A further addition to our computational framework is the development of a “self-learning” kinetic Monte Carlo technique in which the energetics of systems are calculated on the fly as needed with the help of  a pattern recognition scheme used to identify the unique environment of the diffusing entity.  The virtue of this method is that the system evolves in time with atomistic processes of its choice, rather than choosing them from an apriori list that is ordinarily provided as input.  Of course, with enough computational resources the energetics may be calculated from ab initio methods.  This type of multifaceted approach which includes contributions from the system energetics, dynamics and kinetics, we have the ability to carry out accurate studies of a range of phenomena on surfaces and nanostructures which in turn can be tested by direct comparison with experimental data.  The combined usage of methods with varying reliability and feasibility also gives our work versatility.  The ab initio calculations are used to check the validity of the semi-empirical ones and to gain insights into mechanisms underlying phenomena of interest at microscopic levels.  These calculations also provide parameters for the development of robust model potentials for further investigations.

With these theoretical techniques, we have succeeded in isolating the microscopic features in the bonding between atoms that give rise to the observed structural and dynamical characteristics on surfaces of the transition metals like Ag, Cu, and Ni, and s-p bonded metals like Mg, Al and Li.  For example, we have traced the origin of certain experimentally observed vibrational modes on Ni(977), Cu(511) and Cu(211) to changes in the relaxations and force fields at and near the step atoms.  Similarly, with our proposed recipe for the calculation of diffusion coefficients, we have predicted a mechanism for interlayer transport for homo-epitaxial growth on Ag(100) that appears to confirm experimental findings.  In the case of Mg surfaces, on the other hand, we were able to establish the connection between the nature of the surface relaxation (inward or outward) and the surface electronic charge densities brought about in the creation of the surface.  This explains why the top layer atoms on some surfaces of Mg move outward while those on others relax inwards.  Similarly in our studies of chemisorbed overlayers on metal surfaces, we have been able to explain adsorbate induced effects like surface reconstruction, stress, and charge transfer, on the basis of changes in the surface electronic structure.  One example of such work is our mapping of the path to C induced surface reconstruction of Ni(100).   Furthermore, in connection with recent measurements of a striking anisotropy in reflection spectroscopy from two types of atomically-stepped surfaces, our calculations were able to the trace the difference to the local electronic structure of atoms in particular geometries on the surface.  Similarly in a set of recent studies on nanostructured surfaces of Cu and Pd, we were able to show the site selectivity in the adsorption of gases like S and C is guided by the local surface geometry and electronic structure.  In particular we were able to show the conditions under which an impurity like S or C acts as a poison, while alkali adsorption helps enhance the reactivity of these metal surfaces.  On the issue of factors that influence surface reactivity, we have recently provided a rationale for why ammonia prefers to decompose on the steps of Ni surfaces rather than on Ni terraces or Pd steps and terraces.  The above and related applications have helped us establish the role played by the nature of the local bonding in determining the characteristics of systems of interest.

Another aspect of our research has been the isolation of the contribution of atomic vibrations to quantities like the free energy of a system which is ultimately a measure of structural stability of systems.  We find this contribution to be significant and cannot be ignored.  Our work has established that the inclusion of the dynamics of the lattice is critical to understanding microscopic processes that govern the structural stability of surface nanostructures.  In the case of nanocrystals of transition metals also, we find novel properties to arise for lattice dynamical contributions which we find to be distinct from those of bulk material.  The importance of vibrational entropy extends also to bimetallic alloy surfaces and nanoalloys in which we find them to play a critical role in compositional order-disorder transition and in surface segregation.

HIGH PERFORMANCE SCIENTIFIC COMPUTING FACILITY 

Together with the help of several colleagues, I was instrumental in establishing a high performance supercomputing and scientific visualization facility at Kansas State University.  Funding for the project was obtained from the National Science Foundation and from Kansas State University.  The facility became operational in 1995 and served as the computational workhorse for a number of faculty members in the colleges of the Arts and Sciences and of Engineering whose research interests ranged from modeling of materials, to atomic and molecular science to fluid dynamics.  Initially the facility consisted of a powerful Symmetric Multi‑processor consisting of forty-eight processors of the HP/Convex Exemplar S and V class machines.   Clusters of high‑end workstations and PC’s capable of performing high quality graphics and visualization were also acquired.  The establishment of the facility led to the formation of a Center for Scientific Supercomputing, for which I served as the Director from 1997-2000. Seminars, workshops, and symposia on recent developments in computational techniques and their applications were organized.  These activities were particularly beneficial to students and junior scientists who were able to make immediate usage of parallel programming and other computational innovations.   Although financial constrains and ever-evolving trends in computer technology have made this center defunct, its importance in enabling groups like ours to plunge into sophisticated research in the area of computational science cannot be denied.   For our present needs we have recently designed and assembled a Beowulf cluster consisting of 26m nodes of Pentium IV and additional 32 nodes of Xion processors.  Plans are in place for doubling the capacity of the cluster in the near future.