Dr. Brian Kolb
BKolb @ mit . edu


  • PhD: Physics – Wake Forest University (2012)
  • BS : Physics, Chemistry, and Astronomy -- Northern Arizona University (2004)

Research Interests

My work can be categorized into three components:

  1. understanding the basic physics of molecular and condensed systems
  2. engineering materials for applications
  3. method development—especially involving machine learning

The first and second go together, as one needs to understand what governs material properties in order to engineer them. At the moment, I am mainly interested in alternative energy applications, particularly ferroelectric/polar photovoltaics. I have gained an interest in the dynamical aspects of material properties. That is, how do properties such as bandgap, polarization, etc. change when the system is vibrating at finite temperature. I also have a spirited interest in method development, especially machine-learned functionals for improving DFT. This includes more than just new exchange-correlation functionals. During my graduate work I studied van der Waals interactions and methods to include them in first-principles calculations, and I am always on the lookout for new systems where there effects are important.

I use a number of methods to conduct my work, including DFT, the GW method, quantum chemistry methods, and classical MD approaches. I also use various tertiary methods in my work including Monte Carlo, genetic algorithms, neural networks, and support vector machines.

Personal Interests

As a post-doc, the phrase “free time” is somewhat of an oxymoron. However, I enjoy spending time with my wife and dogs, particularly in the outdoors. I occasionally get involved with music projects, whenever I can find like-minded people to play with (I attended Musician’s Institute in Hollywood, CA before beginning college). My preferred style is progressive metal, though I enjoy jamming on almost anything. My favorite activity is programming, and I have several pet projects I occasionally toy with including a first-person shooter game and a new platform for creating scientific talks with increased capabilities (e.g. the ability to embed 3D interactive visualizations within the talk itself).

Current Project

Photovoltaics, Method Development
  • Ferroelectric photovoltaics
    • Ferroelectric PV materials hold many potential advantages over other absorbers in that they can produce above bandgap open-circuit voltages and their lack of centro-symmetry helps drive exciton separation and charge extraction. Currently, however, they suffer from poor photocurrent, largely due to excessively large bandgaps and poor hole conductivity. I am working with a recently discovered ferroelectric, ZnSnO3, to bring the bandgap down and increase carrier mobilities. By substituting sulfur for oxygen atoms, I have been able to lower the bandgap from about 3 eV to 1.3 eV, a substantial improvement. This is accomplished without lowering the remnant polarization of the material, although the nature of this polarization changes (an effect I am looking into further).
  • Machine learning of a bandgap density functionals
    • Many groups are performing high-throughput searches for potential new photovoltaic absorber materials, which requires the ability to accurately and efficiently calculate the bandgap. Unfortunately, inexpensive DFT calculations do not yield good values and the more accurate GW method is prohibitively expensive for high-throughput searches. I am attempting to construct, using machine learning, a density functional capable of approximating the GW bandgap at the cost of DFT. This would allow high-throughput screening to identify potential absorber candidates, and require expensive GW calculations on only a select few, predicted good performing materials.
  • Phase transitions and other characteristics of cubic perovskites
    • The cubic perovskites (formula ABO3, where A and B are any of a number of different species) have shown an immense amount of promise for use in a host of applications. There is a wide range of chemistries that this structure supports, each behaving differently. I am interested in the similarities and differences between the different chemistries. For example, some materials show a phase transition from cubic to tetragonal phases (e.g. SrTiO3) while others (eg. CaTiO3) possess tetragonal symmetry up to their melting point. I believe I have identified the source of this difference and it leads to surprising conclusions about the cubic-tetragonal phase transitions in those materials that exhibit them. There are also interesting differences in the rate of change of the bandgap with respect to epitaxial strain in these materials. Understanding the root of the properties and the similarities nd differences between the chemistries will help us better engineer desired properties for a particular application.