---
title: 'BoltzTraP: First-Principles Transport Coefficients'
url: https://www.emergentmind.com/topics/boltztrap
type: topic
---

# BoltzTraP: First-Principles Transport Coefficients

BoltzTraP is a computational code for deriving semi-classical transport coefficients from first-principles electronic structures. In the literature represented here, it is used primarily to calculate the Seebeck coefficient, electrical conductivity, electronic thermal conductivity, power factor, and, in some workflows, the thermoelectric figure of merit as functions of temperature and chemical potential. Its standard use is grounded in semiclassical Boltzmann transport theory and, in most studies, the constant relaxation time approximation (CRTA), while later developments such as BoltzTraP2 extend the framework through smoothed Fourier interpolation, a command-line interface and Python module, and support for non-constant scattering rates [1712.07946, 2509.00959].

## 1. Definition and place in first-principles transport workflows

BoltzTraP occupies the post-processing stage of a typical electronic-structure workflow. A DFT code first supplies the band energies on a sufficiently dense \(k\)-point mesh; those eigenvalues are then passed to BoltzTraP, which evaluates transport coefficients within a linearized Boltzmann framework. This pattern appears across multiple implementations: ABINIT was used for PbSe, WIEN2k for FeRuTiSi, LaFeO\(_3\), LaNiO\(_3\), GaN, FeMnScGa, and several oxide and half-Heusler systems, VASP for GeTe-based materials, and Quantum-Espresso for intercalated twisted bilayer graphene [2408.03786, 1912.03708, 1709.10327, 2108.09969, 2008.11387, 1912.03709, 2301.13193, 2109.00815].

The code is routinely used with dense Brillouin-zone sampling because transport integrals are sensitive to band curvature near the Fermi level. Representative meshes in the cited studies include \(30\times30\times30\), \(40\times40\times40\), \(50\times50\times50\), \(50\times50\times45\), and, in another formulation, 5000 \(k\)-points [1709.10327, 2108.09969, 2408.03786, 2008.11387, 2008.06384]. This recurring choice reflects the requirement that fine details of the band dispersion, especially near band extrema or spin-split manifolds, be resolved before transport properties are inferred.

A central feature of these workflows is that BoltzTraP connects microscopic band-structure information to macroscopic observables. In PbSe, for example, DFT band structures with and without spin-orbit coupling (SOC) were propagated into transport calculations, and the inclusion of SOC led to better agreement with experiment [2408.03786]. In LaNiO\(_3\) and FeMnScGa, spin-resolved band structures were propagated into transport coefficients through spin-channel analysis and the two-current model [2108.09969, 1912.03709]. This suggests that BoltzTraP is most informative when the preceding electronic-structure model already captures the relevant gap, spin splitting, correlation correction, or relativistic effect.

## 2. Transport formalism and characteristic outputs

In the studies surveyed here, the principal quantities computed with BoltzTraP are the Seebeck coefficient \(S\), electrical conductivity \(\sigma\), electronic thermal conductivity \(\kappa_e\), and power factor \(PF = S^2 \sigma\). Some works also report Hall coefficient and figure of merit \(ZT\) or \(ZT_e\), depending on how thermal conductivity is treated [2408.03786, 1912.03708, 2202.05600, 2109.00815].

The Seebeck coefficient is commonly expressed in the Boltzmann-transport form
\[
S_{\alpha\beta}(T,\mu)=\frac{1}{eT\Omega \sigma_{\alpha\beta}(T,\mu)} \int \sigma_{\alpha\beta}(\epsilon)(\epsilon-\mu)\left[-\frac{\partial f_0(T,\epsilon,\mu)}{\partial \epsilon}\right] d\epsilon,
\]
or equivalently through moment ratios of the transport distribution [2202.05600, 2008.11387, 2410.19252]. Electrical conductivity is computed from band velocities and Fermi-window weighting, typically reported by BoltzTraP as \(\sigma/\tau\), while \(\kappa_e\) is similarly reported as \(\kappa_e/\tau\) unless a specific relaxation time is supplied or assumed [2408.03786, 1912.03708, 2512.15822].

Several studies emphasize a practical asymmetry among outputs: under CRTA, \(S\) is independent of \(\tau\), whereas \(\sigma\) and \(\kappa_e\) require a choice of relaxation time for conversion to absolute values [1709.10327, 2008.11387, 2410.19252]. This is why Seebeck coefficients are often compared directly to experiment, while conductivity and electronic thermal conductivity are either scaled by an assumed \(\tau\) or discussed as trends.

For thermoelectric performance, the derived quantities are usually written as
\[
PF = S^2 \sigma
\]
and
\[
ZT = \frac{S^2 \sigma T}{\kappa}.
\]
When only the electronic thermal conductivity is included, some authors explicitly use
\[
ZT_e = \frac{S^2 \sigma T}{k_e},
\]
which is a restricted quantity rather than the full thermoelectric figure of merit [1912.03708, 2202.05600]. In spin-polarized systems, the total Seebeck coefficient may be constructed from channel-resolved values through a two-current form such as
\[
S = \frac{\sigma^{\uparrow} S^{\uparrow} + \sigma^{\downarrow} S^{\downarrow}}{\sigma^{\uparrow} + \sigma^{\downarrow}},
\]
as done for LaNiO\(_3\) and FeMnScGa [2108.09969, 1912.03709].

## 3. Parameterization, calibration, and comparison with experiment

A recurrent use of BoltzTraP is quantitative comparison to measured transport data after selecting chemical potential, carrier concentration, or relaxation time. The PbSe study provides a particularly explicit example: after ABINIT calculations with and without SOC, the authors used a dense \(50\times50\times50\) mesh, identified the chemical potential matching the experimental Seebeck coefficient at 300 K, and then rescaled \(\sigma/\tau\) and \(\kappa_e/\tau\) using \(\tau = 1.57 \times 10^{-16}\) s rather than the default \(10^{-14}\) s. The calculated \(S\), \(\sigma\), \(\kappa_e\), and \(PF\) then tracked experiment well over 300–500 K, with the SOC case giving the better match [2408.03786].

A second calibration strategy is to impose experimentally measured carrier concentrations. In n-type GaN, Hall-measured carrier concentrations from \(2\times10^{16}\) to \(8\times10^{17}\,\mathrm{cm}^{-3}\) were used to tune the chemical potential in BoltzTraP. Under CRTA and with TB-mBJ electronic structure, the calculated Seebeck coefficients agreed very well with experiment from 260 to 625 K, while deviations at lower temperature were attributed to the limitations of the computational tool and the omission of scattering-parameter-dependent calculations [2008.11387].

A third pattern is correlation-sensitive model selection before BoltzTraP post-processing. In LaNiO\(_3\), several DFT+\(U\) and double-counting choices were tested, and the best agreement with the measured negative Seebeck coefficient in the 300–620 K range was obtained for spin-polarized DFT+\(U\) with \(U=1\) eV and self-interaction correction [2108.09969]. In La\(_{0.82}\)Ba\(_{0.18}\)CoO\(_3\), BoltzTraP outputs were combined with temperature-dependent, spin-dependent relaxation times in a two-current model, yielding good agreement with measured \(\alpha\) and \(\sigma\) from 300 to 600 K [1703.06196].

These case studies show that BoltzTraP is not usually deployed as a parameter-free predictor of absolute transport coefficients. Rather, it is commonly used as a controlled mapping from a chosen band structure to transport observables, with calibration entering through \(\mu\), \(n\), \(\tau\), or spin-channel weighting [2408.03786, 2008.11387, 1703.06196].

## 4. Material classes and research uses

The range of applications represented in the cited arXiv literature is broad. BoltzTraP appears in narrow-gap thermoelectrics such as PbSe and GeTe-based alloys, in correlated oxides such as LaFeO\(_3\), LaNiO\(_3\), and La\(_{0.82}\)Ba\(_{0.18}\)CoO\(_3\), in Heusler and half-Heusler compounds such as FeRuTiSi, FeMnScGa, CaZnC, CaZnSi, and CoHfSi, in perovskite-related oxides and halides such as Te-doped BaTiO\(_3\) and Cs\(_2\)AgSbX\(_6\), and in two-dimensional intercalated twisted bilayer graphene [2408.03786, 2301.13193, 1709.10327, 2108.09969, 1703.06196, 1912.03708, 1912.03709, 2512.15822, 2410.19252, 2202.05600, 2008.06384, 2109.00815].

| Material system | Transport use of BoltzTraP | Representative feature |
|---|---|---|
| PbSe | \(S\), \(\sigma\), \(\kappa_e\), \(PF\) vs. \(T,\mu\) | SOC-dependent agreement with experiment |
| FeRuTiSi | \(S\), \(\sigma/\tau\), \(k_e/\tau\), \(ZT_e\) | n-type doping at \(1\times10^{19}\,\mathrm{cm}^{-3}\) |
| LaNiO\(_3\) | \(S(T)\), \(PF/\tau\) vs. \(\mu\) | spin-polarized two-current analysis |
| Cs\(_2\)AgSbX\(_6\) | \(S\), \(PF\), \(ZT\) vs. \(T\) | p-type Seebeck behavior |
| C\(_8\)MC\(_8\) graphene | \(S\), \(R_H\), \(\sigma\), \(\kappa\) | twist-angle tuning |

Across these classes, BoltzTraP is used not only to calculate coefficients but also to interpret them. Flat bands near a band edge are linked to large effective masses and larger \(|S|\) in FeRuTiSi and LaFeO\(_3\) [1912.03708, 1709.10327]. Mixed heavy and light bands near the valence-band maximum are associated with favorable p-type power factors in \(\mathrm{M}_2\mathrm{Zn}_5\mathrm{As}_4\) [1306.0648]. In Te-doped BaTiO\(_3\) and Ti-Bi co-doped GeTe, BoltzTraP is used to translate gap reduction, direct-gap formation, or band convergence into changes in conductivity, Seebeck coefficient, and figure of merit [2202.05600, 2301.13193].

This breadth of usage indicates that BoltzTraP functions less as a material-specific code than as a general bridge between electronic dispersion and transport behavior. A plausible implication is that its value depends as much on the fidelity of the underlying band structure as on the transport formalism itself.

## 5. Methodological evolution: from CRTA practice to BoltzTraP2 and beyond

The dominant approximation in the literature is CRTA. Many studies explicitly assume \(\tau = 10^{-14}\) s, while others fit \(\tau\) to experiment or vary it with temperature after the BoltzTraP calculation [1912.03708, 1912.03709, 2408.03786, 1703.06196]. This approximation is computationally efficient and often adequate for Seebeck analysis, but several papers identify its limits.

BoltzTraP2 formalizes a broader framework. It computes a smoothed Fourier expression of periodic functions and the Onsager transport coefficients using the linearized Boltzmann transport equation; it can use only band and \(k\)-dependent quasi-particle energies, as well as intra-band optical matrix elements and scattering rates, as input; and it can be used via a command-line interface and as a Python module [1712.07946]. A key methodological advance is the option to include derivatives in the interpolation, so that both the energies and their derivatives match exactly at DFT-calculated points [1712.07946]. The same work states that the positive Seebeck coefficient of lithium is reproduced in an example of going beyond the constant relaxation time approximation [1712.07946].

Related efforts extend the original BoltzTraP workflow toward more realistic scattering physics. In p-type SnSe, the code was extended by incorporating realistic \(k\)-dependent relaxation-time models for screened polar and nonpolar optical phonons, acoustic phonons, and ionized impurities with screening, yielding quantitative agreement with experiment for anisotropic thermoelectric coefficients as functions of temperature and chemical potential [2006.05506]. In CoHfSi, BoltzTraP outputs were combined with a temperature-dependent relaxation time \(\tau = A T^{-3/2}\), and the study explicitly contrasted the resulting \(zT\) values with those obtained under constant-\(\tau\) practice [2410.19252].

Taken together, these developments indicate a methodological transition. The original CRTA-based BoltzTraP workflow remains the standard screening tool, but later work increasingly treats the scattering model, interpolation fidelity, and derivative information as first-order determinants of transport accuracy rather than secondary details [1712.07946, 2006.05506, 2410.19252].

## 6. Limitations, recurring errors, and interpretive discipline

The most systematic critique in the present corpus concerns incorrect use of BoltzTraP outputs in thermoelectric figure-of-merit calculations. One analysis identifies three error modes: using the electronic thermal conductivity at zero electric field, \(\kappa_0\), in place of the electronic thermal conductivity at zero electric current, \(\kappa_e\); computing \(zT\) by combining a constant relaxation time of unity while keeping the lattice thermal conductivity \(\kappa_\ell\) in standard units; and doing both at once [2509.00959]. The physically relevant relation is
\[
\kappa_e = \kappa_0 - S^2 \sigma T,
\]
and the correct thermoelectric figure of merit is
\[
zT = \frac{S^2 \sigma T}{\kappa_e + \kappa_\ell},
\]
not the variants obtained by directly inserting \(\kappa_0\) or by mixing per-\(\tau\) quantities with SI-valued \(\kappa_\ell\) [2509.00959].

The same study derives analytic limits for a single parabolic band and shows that certain faulty results can appear superficially reasonable. In the near-degenerate regime, the erroneous use of \(\kappa_0\) and omission of consistent \(\tau\) treatment can yield \((zT)_0 \approx 0.757\), while another erroneous construction gives \((zT)_e \approx 3.11\) [2509.00959]. This is not a minor bookkeeping issue; it changes whether a predicted material performance is physically meaningful.

Other papers document limitations that are less categorical but equally important. The GaN study reported very good agreement above 260 K but larger discrepancies below 260 K, attributing them to the CRTA and to scattering processes not included in the calculations [2008.11387]. The perovskite study on Cs\(_2\)AgSbX\(_6\) states that CRTA may not fully represent actual carrier scattering mechanisms and that phonon scattering, microstructural effects, and grain-boundary impacts are not explicitly included [2008.06384]. Work on CaZnC and CaZnSi explicitly separates lattice thermal conductivity, estimated via Slack’s model, from the electronic quantities directly obtained from BoltzTraP [2512.15822].

These critiques establish an interpretive rule that applies across the literature: BoltzTraP results are most robust when reported as \(S\), \(\sigma/\tau\), \(\kappa_e/\tau\), or \(PF/\tau\) unless \(\tau\) and \(\kappa_\ell\) are treated consistently. This suggests that the reliability of a BoltzTraP-based study is determined not only by the DFT band structure and \(k\)-mesh convergence, but also by whether the thermodynamic and transport observables are assembled with the correct physical constraints.

Source: https://www.emergentmind.com/topics/boltztrap