Ab initio study of anomalous temperature dependence of resistivity in V-Al alloys

V$_{1-x}$Al$x$ is a representative example of highly resistive metallic alloys exhibiting a crossover to a negative temperature coefficient of resistivity (TCR), known as the Mooij correlation. Despite numerous proposals to explain this anomalous behavior,none have provided a satisfactory quantitative explanation thus far. In this work, we calculate the electrical conductivity using an ab initio methodology that combines the Kubo-Greenwood formalism with the coherent potential approximation (CPA). The temperature dependence of the conductivity is obtained within a CPA-based model of thermal atomic vibrations. Using this approach, we observe the crossover to the negative TCR behavior in V${1-x}$Al$_x$, with the temperature coefficient following the Mooij correlation, which matches experimental observations in the intermediate-to-high temperature range. Analysis of the results allows us to clearly identify a non-Boltzmann contribution responsible for this behavior and describe it as a function of temperature and composition.

💡 Research Summary

This paper presents a first‑principles investigation of the anomalous temperature dependence of resistivity in the binary alloy V₁₋ₓAlₓ, a prototypical system that exhibits the Mooij correlation—i.e., a crossover from positive to negative temperature coefficient of resistivity (TCR) as the residual resistivity exceeds a certain threshold. The authors combine the Kubo‑Greenwood linear‑response formalism with the Coherent Potential Approximation (CPA) within the Korringa‑Kohn‑Rostoker (KKR) electronic‑structure framework to compute the electrical conductivity of disordered V‑Al solid solutions. Thermal lattice vibrations are incorporated via the Alloy Analogy Model (AAM), which treats atomic displacements as an isotropic mixture of shifted atomic positions and evaluates their effect on the electronic scattering within CPA.

The conductivity expression naturally separates into two contributions: σ₀ and σ₁. σ₁ corresponds to the conventional Boltzmann term; it decreases monotonically with temperature because phonon‑induced scattering grows, and it is largely insensitive to the exact position of the Fermi level. σ₀, by contrast, is a non‑Boltzmann term arising from on‑site current matrix elements averaged over alloy configurations. The authors demonstrate that σ₀ is strongly correlated with the electronic density of states (DOS) at the energy under consideration. In particular, σ₀ grows with temperature when the Fermi level lies near a local minimum of the DOS—a situation that occurs for Al concentrations between roughly 30 at.% and 38 at.%. This temperature‑induced increase of σ₀ can outweigh the decreasing σ₁, leading to an overall increase of conductivity (or, equivalently, a decrease of resistivity) with temperature, i.e., a negative TCR.

Calculations reproduce the experimentally observed three‑regime behavior of resistivity versus temperature: (i) a low‑temperature regime (<200 K) where additional mechanisms (e.g., spin fluctuations, weak localization) dominate and are not captured by the present CPA+AAM approach; (ii) an intermediate regime (200–600 K) where the computed resistivity matches measurements very well; and (iii) a high‑temperature regime (>600 K) where experimental samples undergo a phase transition (precipitation of V₃Al) that produces a cusp not present in the calculations. The crossover concentration at which TCR changes sign is predicted at x_c≈0.30, close to the experimental value x_c≈0.35, confirming that the essential physics is captured.

A systematic analysis of σ₀ and σ₁ as functions of Al concentration at 0 K shows that σ₁ follows the classic Nordheim law, while σ₀ displays a nearly linear dependence on the DOS and crosses σ₁ near 30 at.% Al. This crossing point corresponds to the condition σ₀≈σ₁, which the authors identify as a necessary (though not sufficient) condition for the TCR to vanish or become negative. Moreover, plotting σ₀ versus DOS for several compositions and temperatures reveals an almost perfect linear relationship at 0 K, persisting albeit with slight non‑linearity at 1000 K. This reinforces the interpretation of σ₀ as a DOS‑driven, non‑Boltzmann transport channel.

The authors also examine the Mooij correlation directly by plotting TCR against residual resistivity for a range of Fermi‑level shifts (±0.5 eV). Both the true Fermi level and the energy‑resolved data follow the experimentally observed inverse relationship, indicating that the CPA‑based methodology naturally reproduces the Mooij trend without invoking quantum coherence effects such as weak localization.

In the discussion, the paper argues that previous explanations based on weak localization or spin‑fluctuation mechanisms are insufficient because they rely on quantum coherence persisting up to several hundred kelvin, which is unlikely. Instead, the present work shows that a purely semiclassical treatment, augmented by a non‑Boltzmann σ₀ term that is sensitive to the electronic structure, can account for the negative TCR. The authors acknowledge that low‑temperature discrepancies likely stem from omitted effects (e.g., inelastic scattering, magnetic fluctuations) and suggest that incorporating such mechanisms could extend the quantitative agreement across the full temperature range.

Overall, the study provides a robust, ab‑initio framework that simultaneously treats alloy disorder and thermal vibrations, identifies the non‑Boltzmann σ₀ contribution as the key driver of the Mooij correlation in V‑Al alloys, and offers a clear pathway for applying the same methodology to other high‑resistivity metallic systems exhibiting anomalous temperature coefficients.