To predict how a material conducts heat or electricity, physicists usually turn to the Boltzmann Transport Equation (BTE). The BTE describes how a distribution of carriers—electrons or phonons—evolves as they are pushed by external fields and scattered by internal imperfections. However, the "scattering" term in this equation is notoriously difficult to solve.
For decades, the standard shortcut has been the Constant Relaxation Time Approximation (CRTA). This approach assumes that every carrier, regardless of its energy or momentum, returns to equilibrium at the same fixed rate, $\tau$. While computationally cheap, a growing body of research shows that this simplification often masks the very physics researchers aim to discover.
The Friction of the Fermi Sea
Think of relaxation time as a "friction" coefficient. In the CRTA, we treat this friction as a single number. But in real materials, friction depends on how fast you are going (energy) and which way you are turning (momentum).
Changwook Jeong and Mark Lundstrom demonstrated in On Landauer versus Boltzmann and full band versus effective mass evaluation of thermoelectric transport coefficients that while simple models can be mathematically massaged to look like complex ones, they fail to capture nuanced interactions like polar optical phonon scattering. When we assume $\tau$ is constant, we lose the "memory" of the electron's state.
This loss of detail leads to significant errors in three main areas:
- Energy Dependence: Carriers with different energies scatter differently. Zelin Jin found that in single-layer $MoS_2$, ignoring this energy dependence leads to inaccurate thermoelectric predictions (A Revisit to High Thermoelectric Performance of Single-layer MoS2).
- Inelasticity: The CRTA assumes scattering is elastic (energy-conserving) and randomizing. Soubhik Chakrabarty argued that this overlooks the complex physics of III-V semiconductors like $AlGaAs_2$, where inelastic processes dominate (Semi-classical electronic transport properties of ternary compound AlGaAs$_2$: Role of different scattering mechanisms).
- Anisotropy: In materials with complex Fermi surfaces, scattering rates vary by direction. Haoyu Guo suggests that "patchwise" evaluations of the Fermi surface are necessary to understand if phonons can truly drive the linear-in-temperature resistivity seen in strange metals (Can electronic quantum criticality drive phonon-induced linear-in-temperature resistivity?).
Resolving Anomalies with First Principles
When the CRTA fails, it often produces "anomalous" results that lead researchers to suspect exotic new physics where none exists. A striking example is the semiconductor $ZrTe_5$. For years, its resistivity peaks and Hall sign reversals were attributed to exotic topological phases.
However, Hanqi Pi and Quansheng Wu showed that these "mysteries" vanish when using a first-principles framework that accurately models how temperature shifts the chemical potential and relaxation time (First-principles methodology for studying magnetotransport in narrow-gap semiconductors: an application to Zirconium Pentatelluride ZrTe5). By moving beyond the isotropic RTA, they proved that the "anomalies" were simply the result of multi-carrier contributions and complex Fermi surface geometry.
Similarly, Shengnan Zhang demonstrated that many "phase transitions" identified in magnetic fields are actually scaling artifacts caused by the failure of simple relaxation time constants (The inadequacy of the $ρ$-T curve for phase transitions in the presence of magnetic fields).
The Computational Frontier
If the CRTA is so flawed, why do we use it? The answer is computational cost. Calculating the full scattering matrix from first principles is orders of magnitude more expensive than assuming a constant $\tau$.
Recent work is bridging this gap through statistical and machine learning accelerations. Ziqi Guo and Xiulin Ruan developed a sampling method using Maximum Likelihood Estimation that predicts phonon scattering rates 1,000 to 10,000 times faster than exhaustive methods (Sampling-accelerated prediction of phonon scattering rates for converged thermal conductivity and radiative properties). This allows for converged thermal conductivity values that finally match experimental data for materials like silicon.
Furthermore, researchers like Léo Laugier are integrating domain-specific physics descriptors into neural networks to predict thermoelectric performance, acknowledging that pure graph-based deep learning still struggles to capture the nuances that the CRTA ignores (Predicting thermoelectric properties from crystal graphs and material descriptors).
Conservation and Consistency
Beyond accuracy, there is the issue of mathematical consistency. Standard relaxation-time models often violate local conservation laws for particle number, momentum, and energy. Gurinder S. Atwal and N. W. Ashcroft addressed this by introducing the "Fully Conserving Dielectric Function" (FCDF), which forces the relaxation-time approximation to obey these laws (Relaxation of an electron system: Conserving approximation). This framework is essential for correctly describing collective oscillations like plasmons, which "soften" due to relaxation effects.
By replacing phenomenological constants with first-principles scattering rates, the field is moving toward a predictive science where the transport properties of new materials—from layered thermoelectrics like $TiNBr$ (Shuofeng Zhang, First-principles study of the layered thermoelectric material TiNBr) to magnetoresistive candidates like $SrPd$ (Xiao-Qin Lu, SrPd, a candidate material with extremely large magnetoresistance)—can be determined with confidence before they are ever synthesized in a lab.