(6l) Condition and Support Dependent Development of Computational Methods for the Engineering of Materials

My Ph.D. work under the supervision of Dr. William F. Schneider is an example of a system where I integrated density functional theory calculations with my own Monte Carlo code to accurately predict the distribution of active sites in a metal-exchanged zeolite post-synthesis. I then used a combination of ab initio molecular dynamics and analytical statistical mechanical models to describe the dynamic response of the active sites to varying gas exposure conditions. These findings brought the insight of solvated and mobile active sites to this system, a more malleable view of these active sites than had previously been assumed in the literature and more closely resembled a homogenous catalyst under catalytic conditions than a heterogeneous one which was critical to describing the catalytic mechanism. The thermal motion of these active sites also had direct consequences to their electronic structure, which required the integration of time-dependent density functional theory and ab initio molecular dynamics to accurately model. These â??operandoâ? type computational models were validated through in situ and operando spectroscopic experiments that I played a major role in while a visiting student at Purdue University for one semester under the supervision of Dr. Fabio H. Ribeiro, Dr. W. Nicholas Delgass, and Dr. Rajamani Gounder, and at Argonne National Laboratory under the supervision of Dr. Jeffrey T. Miller.

Research Interests: 

Development of computational models that account for the dynamic response of active sites to their contacting support and fluid phase is a critical step for being able to predict new materials that are not only active, but thermally and chemically stable and also capable of being synthesized. These are concepts often neglected in current computational models where the focus has historically been activity, in large part due to the difficulties involved with taking quantum calculations at 0K performed on small systems and scaling these results up to macroscopically significant length and time scales. Computational methods integrating interactions between active sites, supports, and the contacting environment are also important for providing a direct integration between theory and the bleeding edge of experimental methodology in catalysis which is highly focused on in situ and operando characterization of catalysts.

I provide several examples below for models informed by quantum chemical calculations and ab initio molecular simulations that can be coarse-grained to macroscopic length and time scales and a wide range of catalyst exposure conditions through the use of analytical and modern numerical statistical mechanics based theories using novel approaches that either combine computational methods that work efficiently on different energy scales, or those that work akin to classical molecular simulations but integrate quantum based energetics and structures into the model which is a new approach in this field:

I) The restructuring of supported nanoparticles at high surface coverages and temperatures is important for many catalysts, including those used for hydrocarbon chemistry and electrocatalysis, and can lead to dynamical behavior such as oscillatory rates or deactivation. A major challenge is the coupling of nanoparticle reconstruction, which involves high energy barriers and longer time scales, with surface coverage, which will influence the nanoparticle structure and is tied to the chemical potential at the interface and that in the fluid phase and must be equated. Iterative numerical schemes coupling a genetic algorithm with density functional theory computed molecular binding energies allow for rapid convergence of structure and coverage, creating accurate condition-dependent based models of these materials that naturally lend themselves to parallel computing environments such as GPU computing. Accurately modeling these systems informs support engineering at the atomistic scale and allows for mediation of stability challenges through approaches such as 3D printing of supports and chemical tethering of nanoparticles.

II) In microporous media modeling of percolation informs both the birth of the catalyst in support synthesis, metal exchange, or nanoparticle encapsulation, and its death in deactivation, from coking for example. Quantum chemical calculations describe how and where blockages occur, which are then integrated into statistical mechanical models that simulate the zoning off of portions of the support at macroscopic length scales and these geometries can be easily transferred into computational fluid dynamics software. Once these models are validated, new support geometries, for instance including mesoporous and microporous domains, can then be simulated to predict materials with advantageous synthesis properties or enhanced resistance to deactivation.

III) Computational models that incorporate fluid phase and support effects have use beyond just catalysis in the design of materials where thermodynamic properties are critical. Examples of this include supports designed for storage of gaseous NOx or COx species through chemisorption, and Lithium batteries where the formation of dendrites and ultimately shortage is obviously undesirable. Nucleation phenomena like dendrite formation is propagated by the exposure conditions of the contacting fluid and solid interface and take place over long timescales. The dendrite propagation occurs in a way that can be described by analytical fractal models once initiated, but only quantum chemical calculations can describe the initial nucleation accurately, which creates a separation of time and length scales. This can be resolved by tying these analytical models to condition-dependent quantum ones that accurately describe initiation and then use the analytical models to extend to deactivation timescales that will be macroscopically observable.

All of these processes involve systems where quantum-based methods are critical to describe the chemisorption of molecules or covalent interactions between active sites and a support, but must be integrated with analytical and numerical modeling to extend to phenomena that occur on the length and time scales of catalyst synthesis, restructuring, and deactivation.

Teaching Interests: 

I have a strong background in analytical and numerical mathematical methods including:

I)analytical: solution to different classes of ODEs, PDEs, asymptotic methods, integral transforms, perturbation theory, and non-linear dynamics

II)numerical: classes of numerical PDE solvers and stability, finite element and volume methods, numerical solutions to the Navier-Stokes equations and higher order expansions of them

While many of these methods are not appropriate for an undergraduate curriculum, what my exposure to a diversity of analytical and numerical methods has taught me is that typically one methodology works well when the other does not and due to this complementary nature both analytical and numerical techniques should be emphasized in an undergraduate curriculum. In particular a strong theoretical foundation is critical to any undergraduate education since an experiment or computational simulation performed without an underlying theory behind it is worthless.

My experience performing fundamental experiments both as an undergraduate at Notre Dame and as a Graduate Student visiting at Purdue also provides me with the background necessary to teach any undergraduate chemical engineering course, including those involving experiments such as unit operations, phase equilibria, process control, and catalytic systems.

A priority of mine as an educator will be on exposing undergraduates to programming languages, with emphasis on python/matlab which are very similar and have a small learning curve, as well as interfacing with unix based operating systems. Many of the best catalysis experimental laboratories currently incorporate automated reactor design, "big" data analysis that relies on codes to facilitate the interpretation of large amounts of collected data, and working with unix based systems that rely on terminal commands, macros, and skilled use of on-site software when performing experiments at government funded laboratories. Contemporary approaches to the incorporation of coding into classical chemical engineering courses will be easy (for instance, automation of steam tables or UNIFAC) to integrate naturally into the core chemical engineering curriciulum. In the modern era some computational knowledge is key not only for theoretical models and simulation, but also for the design and analysis of experiments.