Science1 publisherNot yet confirmed elsewhere2 min readPublished
DFT+U calculations trace electron mobility in GaAs, InAs and InSb to the effective mass
Dehghani and colleagues tuned Hubbard parameters in DFT+U calculations and found effective mass governs phonon-limited electron mobility in GaAs, InAs and InSb. That gives modellers a route to mobility estimates that avoids computationally demanding GW calculations.
The Scientist · Science desk
What happened
- The method combines DFT+U electronic-structure calculations with an iterative solution of the linearized Boltzmann transport equation, using Wannier interpolation.
- The team computed both drift and Hall mobilities, and the effective mass came out as a key parameter for each of them.
- The paper, by M. Dehghani, D. Waldhoer, A. Gentles and colleagues, appears in npj Computational Materials.
Compiled by The ScientistSomething wrong?How this is made
Why it matters
- decision Screening candidate semiconductors on effective mass alone can misrank those with strongly nonparabolic conduction bands, so band shape has to be part of the comparison.
- constraint Because the results cover phonon scattering only, any comparison with a measured device mobility has to account separately for every other source of scattering.
- capability Groups designing high-mobility semiconductors can compare phonon-limited mobility trends across candidate compounds in a first-principles calculation before growing any of them.
The authors varied the Hubbard parameters on purpose. They used them to shift the conduction band of each compound and recomputed transport at each setting [3]. The same framework lets the electronic bands and the phonon dispersion be refined step by step, and the authors report that this gives a consistent description of phonon-limited transport [7]. The abstract says the tuning altered the conduction band's "characteristics" [3], plural, so each step may have moved more than the mass alone.
Nonparabolicity means the conduction band's curvature changes with energy instead of holding a single value. The authors find that it pulls mobility away from the conventional power-law dependence on effective mass [5].
On accuracy, the authors wrote: "The resulting mobilities are consistent with available experimental trends and with previously reported theoretical values obtained either from computationally demanding GW calculations or from established semi-empirical methods." [6] The abstract does not give mobility values or a computing-time comparison with GW. I think matching trends across three compounds is a looser test than matching each measured value within a stated error. That is enough to take the effective-mass finding seriously within this model, but not yet enough to say DFT+U can stand in for GW in transport work.
Prediction is harder. The authors present the method as a practical framework for analysing and predicting phonon-limited mobility trends, with implications for designing high-mobility semiconductors [8]. For a compound nobody has measured, the Hubbard parameters would have to be set with no GW or experimental value to check them against.
The Austrian Research Promotion Agency FFG funded the work under project number 895289, and the Christian Doppler Research Association, the National Foundation for Research, Technology and Development and the Austrian Federal Ministry for Digital and Economic Affairs also contributed support [10]. The calculations ran on Austrian Scientific Computing infrastructure, and the authors declare no competing interests [11].
What to watch
- Whether the full paper's tables put calculated mobilities for GaAs, InAs and InSb within a stated margin of measured values.
- Whether the approach reproduces mobility in a semiconductor outside these three when the Hubbard parameters are set without a reference value.
- A published comparison of computing time between this DFT+U workflow and a GW-based mobility calculation for the same material.