Energy and food are critical aspects of modern civilization, and the current dependence on fossil fuels and centralized fertilizer production will present increasing geopolitical challenges as the global population continues to grow. The tools and methods developed in the Medford group are applied to address challenges in energy and sustainability. Ultimately, the goal is to develop a computational toolkit that can be applied to arbitrary chemical conversion challenges, allowing chemical engineers to more rapidly develop technology to respond to the energy and food needs of a rapidly changing society.
Fertilizers are critical to agriculture and food production. Most fertilizers are based on “fixed nitrogen” that accelerates the growth of plant tissues. The development of the Haber-Bosch process in the early 1900’s revolutionized humanity’s ability to produce synthetic ammonia, a key ingredient to nitrogen-based fertilizers. This process enables an exponential growth in the global population, and has been called the most important invention of the 20th century. However, the Haber-Bosch process requires tremendous capital investments, is highly centralized due to high temperature/pressure operation, and requires hydrogen that is typically produced from fossil fuels. The Medford group is interested in developing fertilizer production processes based on photochemistry and electrochemistry. These reactions can be carried out at low temperatures and ambient pressures, potentially enabling the small-scale distributed fertilizer production. This will improve access to fertilizer for farmers in remote locations while reducing the energy and carbon footprint of fertilizer production.
Biomass is one of the most abundant and lowest-cost sources of hydrocarbons. However, biomass recovered from non-food sources is difficult to convert to usable chemicals. Significant effort has been devoted to the development of biofuels and biorefineries where biomass can be converted to value-added chemicals such as sweeteners and organic acids. Understanding the fundamental interactions between biomolecules and surfaces will improve the ability to design and optimize catalytic processes for the conversion of cellulosic biomass to high-value renewable chemicals. Research in the group focuses on studying specific biomass conversion model reactions in order to develop insight into the key factors that govern the chemical transformation of biomass feedstocks.
Natural gas conversion
The development of non-fossil energy sources is a critical goal for the long-term sustainability of modern civilization. However, fossil resources are also incredibly energy rich and are likely to remain a key part of the energy portfolio for centuries. Natural gas, composed primarily of methane, is a common byproduct of oil extraction, and is becoming increasingly common with the rise of shale gas globally. One challenge with methane is the capture and transport since it is a gas that must be pressurized in order for transport. This transportation cost is a major factor in the economics of natural gas utilization, and often leads to “flaring” where natural gas is simply combusted because it is too expensive to transport it away from remote drilling locations. Ideally, it would be possible to develop a chemical process for converting natural gas into a liquid product such as ethanol which could be more easily stored and transported. By studying the fundamentals of methane conversion reactions the group seeks to discover economical catalytic processes for producing liquid energy carriers from methane.
Surface catalysis is a complex process that takes place at a complex solid/gas or solid/liquid interface. The phenomenon of catalysis is inherently quantum mechanical as it involves the dissociation of a chemical bond where electrons are interacting. Understanding the energetics of the process requires detailed knowledge of the arrangement of atoms at the catalytic interface. Ultimately these atomic-scale events are coupled with mass and heat transport and control the overall behavior of a large-scale chemical reactor. This inherently multi-scale process is tremendously complex, even for simple model reactions, and most physics-driven methods are not practical for predicting behavior in realistic systems. Method development in the Medford group seeks to combine physics-based models with data-driven approaches in order to tackle the complexity of catalytic reactions.
Density functional theory (DFT) is the leading method for quantum-mechanical simulations of chemical and materials systems of practical size. The framework of DFT is theoretically exact, but despite decades of research the complex exchange-correlation energy of interacting electrons must be approximated. Establishing the accuracy of these approximations is especially challenging for surfaces, where experimental energies are difficult to measure. One goal of the group is to develop machine-learning inspired techniques for assessing and improving the accuracy of exchange correlation approximations for surface and interface systems.
Microkinetic models are critical to understanding catalysis because they connect the atomic-scale phenomenon of bond breaking to the observable conversion of chemicals at the macroscopic scale. These models range considerably in complexity. Relatively simple models require few inputs and typically generalize well, but are prone to missing important details, particularly for systems involving large molecules or selective conversion. On the other hand, more complex models offer more detailed insight into the reaction and allow more accurate predictions, but they are computationally expensive and do not always generalize. The Medford group is interested in developing microkinetic modeling methods that utilize data and uncertainty to balance complexity and accuracy, including approaches for coupling to mass and heat transport.
High-throughput computational techniques have been very successful in the discovery of bulk materials, and the properties of essentially all known three-dimensional crystal structures have been computed and stored in numerous databases. This large amount of data has enabled the discovery of new compounds through the application of data science and machine learning techniques. This high-throughput approach is more challenging for surfaces where each material may have numerous relevant surface chemistries, requiring more calculations and improved methodologies for classifying and digitally representing surfaces, interfaces, and molecular adsorption systems. The group is interested in developing tools and infrastructure for storing, sharing, and analyzing this complex but important class of systems.