Converting abundant natural resources cleanly and efficiently into a useable form of energy is one of the single most important challenges of the 21st century. This is a materials problem. In fact, this may be the biggest materials problem of our lifetime. The understanding and discovery of new materials will be the key to realizing the massive challenge of providing clean, abundant energy on a global scale. We must discover new materials to generate energy that are abundant, inexpensive, efficient, and scalable. The materials in use today cannot addressall of these needs simultaneously, and will therefore simply not do if we are to genuinely counter the deleterious environmental and political impacts of our long-standing reliance on crude oil.  Innovation, understanding, enormous amounts of creativity and hard work, and genuine collective efforts between the scientific and engineering disciplines will be essential.

Our research program places substantial effort in applying computational materials science approaches to the challenge of tailoring new materials for applications in energy conversion. Our current predominant focus is on understanding and predicting important properties in photovoltaic and thermoelectric materials, although we also work on other energy conversion challenges, from efficient solar fuels to cement chemistry. In each of these types of energy conversion, new materials have been discovered very recently that show enormous promise for dramatic improvements in conversion efficiencies at substantially reduced costs.  Yet, many of the central mechanisms that govern the conversion efficiencies in these materials, such as sunlight-to-electricity and heat-to-electricity as shown in these figures, remain poorly understood and therefore difficult to control.  The role of computation for energy conversion is therefore paramount. Computational modeling holds high promise to accelerate the key discoveries in energy conversion: it is now possible to predict many properties of materials without any experimental input so that one can probe a given material through "virtual synthesis" before the real synthesis in the laboratory.

The primary energy used in the U.S. is approximately 100 quads, of which 80% involves thermal energy in the form of heating, cooling, or waste heat.  Most practical thermal energy systems that involve transport, storage, and/or conversion of heat operate far from their limits. Furthermore, heat dissipation and thermal management are key issues in energy systems such as solid-state lighting, photovoltaics, and ultra-low power electronics. By understanding the behavior of important mechanical phenomena, such as the nature of heat dissipation, thermal transport and scattering, and mechanical coupling between nanoscale objects, our aim in this component of my research is to predict new materials and possible devices with improved or novel thermal and mechanical properties, particularly as they relate to applications in energy.

In this work, we apply primarily classical molecular dynamics simulations to understand and predict thermal transport and dissipation in and across materials, with applications ranging from energy to chemical sensing. For example, nanoscale morphologies can be used to drastically reduce the thermal conduction of a material, thus providing an attractive route to improving the efficiency of thermoelectric materials (as discussed above).  On the other hand, and in sharp contrast to this effect, the quasi one-dimensional structures of carbon nanotubes can lead to ballistic thermal conduction in which there is no phonon scattering, and thus carbon nanotubes could be used in materials to dramatically improve thermal conduction.  Carbon nanotubes and other near one-dimensional structures such as nanowires are also used to make ultra high frequency resonators with potential applications in wireless communications, signal processing, mass balances, and chemical detection systems.  However, these applications are at present severely limited by the fairly low quality factors and so a detailed understanding of the origin and limits of dissipation within these nanoscale systems is critical.

Crude oil has excellent energy storage characteristics, such as a very high energy density of 45 mega-joules per kilogram as well as the ability to be moved from one place to another safely and conveniently by almost any means of transportation. Our challenge lies in understanding and developing new mechanisms and materials for storing and transporting energy safely. At present, our primary work on energy storage focuses on two directions: materials for hydrogen storage and materials for direct sunlight-to-chemical storage. 

Ideally hydrogen will be stored in a lightweight and compact manner for mobile applications. Apart from the often-cited weight percent of hydrogen that a material can store, a number of other criteria are important. One of these is the amount of energy required to get the hydrogen out of the storage material. Our effort in this area is aimed at understanding and optimizing both the storage weight as well as the desorption temperature general, which is quite challenging as these two properties are often inversely related in bulk materials.

Our work in the area of solar fuels is focused on a class of molecules that can convert sunlight directly into "stored heat" in the form of chemical bonds.  These molecules undergo a reaction upon exposure to light that is reversible with either a catalyst or heat. In some cases a considerable amount of energy can be stored. While there are many examples of molecules that can do this once or perhaps several times, only one case has been shown to date to be able to do this reversibly many times, with no degradation: a di-ruthenium fulvalene complex. In this case, the mechanism shown in the figure, the Ru-Ru bond and the C-C bond are broken upon light exposure, and the molecule effectively “flips”. This stored chemical energy is highly stable, with a large back-reaction barrier, and can then be released in a very straightforward manner. In addition to understanding and predicting the details of both the forward photoisomerization step as well as the back-reaction heat release step, we are more broadly interested in identifying new classes of molecules (based on cheaper, more abundant materials) that can undergo the same kind of reaction as robustly.

One of the greatest challenges of materials design lies in the ability to controllably make the desired material. In fact, today often the biggest bottleneck is not in predicting a new material with improved or interesting properties, it is in the synthesis of the material.  Yet, the direct prediction of synthesis processes by computational materials science methods is extremely challenging   due to the vast chemical and environmental complexity, the enormous number of variables and large length and time scales, to name only a few.  With this in mind, we tackle the problems of synthesis via multiple computational approaches, from Monte Carloto molecular dynamics to hybrid methods, in order to understand the most important aspects of a given synthesis route, predict the resulting structural evolution, and correlate the resulting structures to key properties.

For example, in our work on self-asembly, our goal is to predict new techniques for efficiently bringing nanscale objects together in an ordered, controlled manner. We employ a combination of ab initio and classical molecular dynamics to investigate the possible self assembly of combinations of organic and inorganic nanoparticles, ranging from silicon nanowires to polymers (example shown in the figure) and carbon nanotubes. We attempt to predict the effects of varying chemical functionalizations and the subtle interplay of energetic contributions, including competition with solvation effects. In some cases, high-accuracy quantum Monte Carlo simulations are performed to gauge the accuracy of the DFT energetics, and to fit classical potentials more accurately for the larger molecular dynamics runs.

To address the challenging problems we are interested in, we apply an “arsenal” of advanced computational approaches, including quantum Monte Carlo, density functional theory, post-Hartree-Fock methods, plane-wave-based ab initio molecular dynamics, Gaussian-based quantum chemistry, tight binding and classical Monte and molecular dynamics, empirical methods, and genetic algorithms. In our work, we do not necessarily rely on a single method to solve a given problem nor do we favor one approach over another; rather, we use the best method(s) available and those most suitable to tackle the challenge at hand. In some cases, this means developing new methods or improving upon the efficiency of existing ones.

Examples of our recent algorithm development include new approaches to optimize structures using quantum Monte Carlo, the coupling of quantum Monte Carlo with molecular dynamics, methods for generating amorphous structures, and a new technique for thermostatting phonons in a molecular dynamics simulation. In addition we are involved in making computational tools more accessible to the experimental researchers and for learning theoretical methodologies through a collaboration with the Network for Computational Nanoscience. The mission of the NCN is to connect theory, experiment, and computation in a way that makes a difference to the future of nanotechnology. As part of the NSF's infrastructure for the National Nanotechnology Initiative, the NCN engages the community through workshops and seminars and novel educational resources.

The periodic monotone of bulk electronic and structural properties comes to an end at the surface of a material. A rather abrupt change in the topology and the crystal potential at the surface produces many interesting physical and chemical properties comprising surface phenomena. For instance, while atoms in the bulk of a solid are happily bonded in all directions to their neighbors, bonds at the surface often remain unsaturated leading to reconstructions, enhanced chemical reactivity, and interesting electronic properties. Since most of our interactions with materials occur via surfaces, the study of these also forms one of the most fascinating topics of current research and one of great practical interest as well. 

Surface chemistry and physics become increasingly more important for understanding materials properties as the material size becomes smaller. At the nano scale, the surfaces can even dominantly determine the materials properties, playing the critical role together with the confinement effects. It is therefore important to understand both structural and electronic roles of the material surfaces for designing/investigating the nano-materials.

Our work in this area includes the prediction of relative surface stabilities, dissolution processes, charge effects, chemical reactions that take place on a surface, the effects of solvent-surface interactions, and mechanical deformations of surfaces. Our interests in surface phenomena are quite broad, ranging from carbon nanotubes (every atom being a surface atom), to understanding the dissolution of cement (shown in the figure), to predicting the strain response of nanomechanical cantilvers.