Lam Research Group

Our group combines predictive simulation, data analytics and informed experiments to accelerate the development of materials in nuclear and other energy applications. We are interested using nanoscale insights to provide understanding of material properties, applying new computational tools for exploring materials chemistry, predicting material performance in service, and screening for desired properties in new and extreme energy conditions. To achieve these objectives, we combine electronic structure calculation, molecular dynamics, coupled multiphysics simulation, machine learning, materials characterization, irradiation testing and more. The ultimate goal of our research is to advance next generation clean energy technologies by overcoming critical challenges in areas of materials science and chemistry.

Advanced Nuclear Reactors

The last decade has seen significant interest in advanced nuclear reactors that could dramatically increase safety and reduce the cost of nuclear energy. However, there remains significant challenges in developing materials that can sustain high temperatures, high neutron fluxes, and chemically harsh environments that are found in many of these systems. Further, qualifying a material for a new nuclear application can take decades due to the extensive experimentation required to understand material behavior in reactor operation. Thus, in order for advanced reactors to make a timely impact on reducing global carbon emissions, the process of understanding material behavior must be significantly accelerated and improved. This group supports the development of advanced systems such as high-temperature gas-cooled reactors (HTGRs) or molten salt fission and fusion reactors by addressing key materials challenges. We focus on performing physics-informed simulations that provide fundamental insight into material screening and discovery.

Predicting Molten Salt Properties

Molten salts are proposed as heat-transfer fluids due to their high heat capacity and are useful in fission, fusion and solar energy storage applications. However, their chemical and thermophysical properties are not understood over a wide range of compositions and potential impurities. This is especially true in nuclear systems where neutron transmutation will alter the composition throughout the fuel cycle. This work uses first principles atomistic simulation and molecular dynamics to 1) predict and optimize properties of molten salts, 2) understand the structure, dynamics, and properties of molten salts, 3) elucidate fission and activation product chemistry, and 4) determine materials compatibility and understand phenomena at the interface of molten salt and structural materials. Simulations are validated by comparing to experimental data such as diffraction patterns from x-ray and neutron diffraction that can be used to calculate the average local structures of different isotopes.

Neural Network Interatomic Potentials for Complex Systems

This work involves developing neural network (and other machine learning) interatomic potentials for simulating high-dimensional systems. We have recently shown that NNIPs can simulate multi-component systems orders of magnitude faster than ab initio (DFT), while maintaining a similar level of accuracy. This allows studying extended systems at much larger scale and for longer time periods, enabling the exploration of many more material properties, and better sampling of rare events. However, there remain many challenges in using neural network potentials such as the large number of DFT calculations that are required to train the neural networks, and estimating error in neural network predictions. Here, we develop methods for systematically training robust and reliable potentials in order to optimize the training and calculation time, and to assess accuracy during simulation.

Multiphysics Simulation of Chemical Processes

Advanced reactors, fuels, and coolants require chemical processes that differ significantly than those found in traditional reactors. Further, this can result in different waste forms for which there is no well-defined fuel cycle. To design suitable process technologies (e.g. adsorption of radioactive gases and recovery of valuable actinides), the simulation of multiple simultaneous physics such as fluid flow, heat transfer, mass transfer and reaction chemistry is often required to understand behavior and predict performance at the system level. Further, multiphysics simulations can be useful in interpreting experimental results. This group combines atomistic insights with component level multiphysics simulation to support the development of new energy systems.