(4fv) Applications of Nonequilibrium Thermodynamics & Simulation

I propose an ambitious research program to develop and use nonequilibrium simulations to study the important processes of life and guide development of artificial, autonomous molecular systems. With undergraduate and graduate degrees in chemical engineering and a breadth of research experience in theory and simulation, I am well-suited to lead such a program. Because I recognize the importance of undergraduate education and the need for computational skills, I propose to integrate computer programming concepts in my courses, giving students a competitive edge in the modern job market.

Research Interests

If, as the saying goes, “equilibrium is death,” then its less morbid counterpart is that “nonequilibrium is life.” At the molecular level, the biological machines that carry out life processes are buffeted by thermal fluctuations and, in equilibrium, are as likely to move in the forward direction as the backward direction, rendering them useless. Only with a nonequilibrium supply of free energy can these machines produce work and directed motion. Poetically, the study of these machines and their mechanisms is the study of the building blocks of life, and practically, it could lead to the development of better artificial molecular motors and a better understanding of molecular physiology.

Simulation is an incredibly powerful tool for studying molecular systems, but its current domain is largely limited to equilibrium situations. Equilibrium simulation is successful for determining molecular structures, calculating static properties, and illuminating transitions that occur in thermodynamic equilibrium. But many important processes—most notably those of life itself—occur away from equilibrium. This nonequilibrium realm is much wider, and molecular simulation has left it largely uncharted. For a system to be in a nonequilibrium state, it must have net energy or mass flows, and these flows can come from many sources: influxes or outfluxes of mass, heat baths at different temperatures, moving boundaries, imposed external torques or forces, hydrodynamic flows, electric or magnetic coupling—the list is endless. Any imaginable chemical, mechanical, or heat-flow process could place a system into a nonequilibrium state, and no general recipe for simulation can be applied that would address all of these processes. Nonequilibrium conditions need to be treated carefully, at an individual level, specific to each process.

My proposed research program has three main goals: (1) develop nonequilibrium molecular simulation methodology, (2) apply nonequilibrium simulation to study important biological and chemical systems, and (3) elucidate design principles of artificial autonomous systems through simulation and theory. Success with this program would significantly extend the uses of molecular simulation, increase our understanding of the mechanisms at the heart of life, and contribute to the design of artificial molecular motors and self-assembly processes.

Teaching Interests

I am most excited to teach core chemical engineering courses in transport, thermodynamics, and kinetics. I could also be called upon to teach controls, separations, or mass and energy balances. With undergraduate and graduate degrees in chemical engineering, I have completed extensive coursework in these subjects. Additionally, I have teaching experience in chemical engineering classes, and these subjects have large overlap with my research expertise. Because my proposed research program focuses on statistical mechanics and simulations, I would also be well-suited to instruct graduate courses or electives on statistical thermodynamics or numerical methods.

Because most chemical engineering careers today involve computer work of some kind—data analysis, computer-aided design, process simulation, or general programming—a chemical engineering curriculum should develop students’ programming skills beyond the rudimentary programming learned in the general engineering curriculum. To this end, I would include programming concepts in coursework. For example, I would introduce numerical methods for solving differential equations into courses in kinetics and controls because these concepts fit naturally into these classes. Other core classes also offer numerous opportunities to include programming problems or projects: students can write programs to iteratively solve simple feedback systems in material and energy balances, use scripts to easily visualize how flow lines change with different physical parameters in fluid mechanics, and program different equations of state to easily see how they affect answers for problems in thermodynamics, to name a few.

I would introduce these problems, with appropriate background and supplementary material, gradually throughout the semester. The first problems will consist of mostly complete code where students would fill in a few lines and visualize and analyze results. By the end of a semester, students will be able to develop programs from the ground up. Hands-on work with computer programming will give students concrete examples to discuss in employment interviews and career fairs and keep them competitive in the job market.

Research Experience

Simulating Chemically-Fueled Molecular Motors, Department of Chemistry, Northwestern University (Todd Gingrich Group)

Representative Publication: A. Albaugh, T. R. Gingrich. “Simulating a Chemically-Fueled Molecular Motor with Nonequilibrium Molecular Dynamics”, submitted (arXiv:2102.06298), (2021).

Inspired by recent developments of artificial, autonomous molecular motors, I have developed nonequilibrium simulation methodology and motor models as part of my postdoctoral work. With these methods and models, I can simulate the complete dynamics of an information ratchet over many motor cycles. This information ratchet gates the natural diffusion of a small ring interlocked with a larger ring in a preferred direction by coupling to the decomposition of a fuel molecule. While much simpler and less efficient than biological molecular motors, these studies capture similar mechanisms. By simulating complete dynamics at a particle level, I can compute the complete thermodynamic cost of the motor as well as its accuracy, efficiency, and velocity. In turn, by leveraging thermodynamic uncertainty relationships, I can determine how close the motors operate to theoretical limits. By easily changing configurations and interactions in silico, I can use the simulations as an inexpensive design tool to inform the design of future artificial motors.

Designing Chemical Systems to Target Dynamics, Department of Chemistry, Northwestern University (Todd Gingrich Group)

Representative Publication: A. Albaugh, T. R. Gingrich. “Estimating Reciprocal Partition Functions to Enable Design Space Sampling”, Journal of Chemical Physics, 153 (20), 204102, (2020).

Designing a system that undergoes a desired transition quickly is a difficult task. Given a vast chemical design space of possible intermolecular interactions and configurations, searching this space for fast rates becomes expensive. As part of my postdoctoral work, I developed a computational procedure to efficiently explore this space by combining transition path sampling with joint sampling of the given chemical design space. This necessitated a technical innovation in generating unbiased, but inexpensive, estimates of reciprocal partition functions for the acceptance criterion of the sampling procedure. In the end, the sampling procedure samples the design space in proportion to the targeted rate at each point in design space. Simply, designs that generate fast rates will be sampled more often. This bias towards faster rates can even be made arbitrarily strong, but at the cost of greater expense.

Improved Methods for Polarizable Molecular Simulations, Department of Chemical Engineering, University of California, Berkeley (Teresa Head-Gordon Group)

Representative Publication: A. Albaugh, A. M. N. Niklasson, T. Head-Gordon. “Accurate Classical Polarization Solution with No Self-Consistent Field Iterations”, Journal of Physical Chemistry Letters, 8 (8), 1714-1723, (2017).

Polarization, the ability of a molecule’s electron density to mutually affect and be affected by its environment, can be a critical physical component for accurate simulation of condensed phase systems. Classically, the effect can be modeled with inducible dipoles, fluctuating charges, or Drude particles. The additional cost of these approaches comes in the form of iterative calculations to converge the polarization. In a series of breakthroughs in my doctoral work, I developed methods that reduced the required number of self-consistent iterations from 5-10 (depending on the system) to 1, effectively making them iteration-free. The hybrid approach used dynamically driven auxiliary degrees of freedom in conjunction with the mathematical framework of iteration. This can be seen as an improvement over pure extended Lagrangian (Car-Parrinello) dynamics. While initially developed for a polarizable dipole model, I was also generalized it to fluctuating charge and Drude models and demonstrated its performance over a variety of systems—aqueous, ionic, small molecule, and protein.

Teaching Experience

• Transport Processes (UC Berkeley Chemical Engineering), Graduate Student Instructor (2017)
• Process Dynamics and Control (UC Berkeley Chemical Engineering), Graduate Student Instructor (2015)
• Chemical Kinetics and Reactor Design (UC Berkeley Chemical Engineering), Graduate Student Instructor (2013)
• Introduction to Chemical Engineering (UC Berkeley Chemical Engineering), Graduate Student Instructor (2012)