Nanostellar's Computational Nanoscience

Nanostellar uses a variety of computational methods as part of the RCD process. The most fundamental methods involve ab initio quantum chemistry and density functional theory, which provide atomic-level data on structure and reactivity. While such methods are highly accurate, they are also computationally expensive, and can only treat systems that are smaller than the real nanoclusters required for optimum performance in many industrial applications. Recognizing the need for novel computational approaches, we thus developed new methods to treat metal systems that provide reasonable accuracy with a significant computational cost reduction. With these methods, we can routinely study realistically-sized nanoclusters and nanoclusters on the oxide supports that they are routinely deposited on for use in many catalytic applications. We can also study the time evolution of these particles under realistic pressures and temperatures. We then take the information we obtain from our atomic level methods and feed it into more macroscopic methods of simulation, such as Kinetic Monte Carlo, to study reactions over longer time scales.

The figure below shows the possible length and time scales achievable by our computational methods. With this hierarchy of methods we proceed from microscopic features (atomic-level quantum chemistry effects) to macroscopic features (device scale modeling). Our RCD methodology includes all of these simulation methods and the necessary computational "bridges" that link the methods together.

A very important aspect of Nanostellar's RCD approach is the tight coupling of computational methods with experimental approaches. While our computational methods can provide a great deal of data, we require experimental data to compliment and verify our theoretical work. A main goal of the computational program is to help to prioritize our experimental work toward the most promising new materials.

The following picture is a snapshot from a dynamic simulation of a platinum nano-particle on an alumina support. Such simulations give us information on how particles behave over time at different temperatures. They also show how the particle interacts with its support. These are the kind of systems Nanostellar can routinely treat with its simulation methods.

The following two movies give examples of the kind of results we obtain by simulation. The first movie is of an atomic-level Kinetic Monte Carlo (KMC) simulation of CO oxidation of the (111) surface of platinum. You first see disordered adsorption of CO on the surface. Over time, the CO configuration evolves into a regular pattern. Finally, the CO on the surface combines with oxygen to form CO2 molecules, which desorb from the surface. These KMC simulations require binding energies, reaction barriers, and rate constants for all the species and processes that occur on the surface. We obtain these data by quantum mechanical calculations.

CO oxidation on Pt(111) video clip

The second movie also shows the results of a KMC simulation. In this case we are simulating the sintering of metal alloy particles on an alumina surface under realistic temperature. As the nanoparticles age (evolve over time) small particles tend to combine to form larger particles. This process is seen in the simulation. We design our products to resist the sintering process, which adversely effects catalyst performance.

Alloy AB Nanoparticle Sintering video clip