Mooij Correlation in Disordered Metals

• Mooij correlation is an empirical trend where the sign of the temperature coefficient of resistivity (dρ/dT) reverses beyond a critical resistivity, signaling a breakdown of semiclassical transport.
• It is analyzed via weak-localization, strong-coupling disorder–phonon theories, and ab initio calculations, each revealing unique scattering mechanisms in disordered metals.
• The correlation serves as a diagnostic tool in experiments and theory, uniting observations from nanoscale control studies to fundamental transport limits like the Mott–Ioffe–Regel scale.

Searching arXiv for recent and foundational work on Mooij correlation in disordered metals. I’m retrieving relevant arXiv records on Mooij correlation, including mechanism papers, weak-localization interpretations, nanoscale control studies, and recent ab initio investigations. Mooij correlation, often termed the Mooij rule, is the empirical correlation in disordered metallic alloys between resistivity and the sign of the temperature coefficient of resistivity (TCR), $d\rho/dT$. In the cited literature, alloys with lower resistivity exhibit $d\rho/dT>0$, whereas sufficiently high-resistivity alloys exhibit $d\rho/dT<0$, with the sign change discussed around $100\div150\,\mu\Omega\cdot\text{cm}$ and in relation to a characteristic Mott–Ioffe–Regel scale $\rho^* \sim 300\,\mu\Omega\cdot\text{cm}$ (Gantmakher, 2011). The phenomenon has been analyzed through weak-localization theory, strong-coupling disorder–phonon theories, ab initio transport calculations, and nanoscale multilayer experiments, and it remains a central probe of the breakdown of semiclassical transport in strongly disordered metals (Ciuchi et al., 2018).

1. Definition and empirical phenomenology

The Mooij correlation is defined in the cited work as a robust empirical trend relating TCR to resistivity in disordered metals and alloys. In one formulation, “the sign of $d\rho/dT$ of an alloy correlates with its resistivity $\rho$,” with

$$d\rho/dT>0 \quad \text{at}\quad \rho<100\div150\,\mu\Omega\cdot\text{cm},$$

and

$$d\rho/dT<0 \quad \text{at}\quad \rho>100\div150\,\mu\Omega\cdot\text{cm},$$

up to $\rho^* \sim 300\,\mu\Omega\cdot\text{cm}$ (Gantmakher, 2011). A closely related formulation describes the Mooij plot as the relation between the TCR, $d\rho/dT>0$0, and resistivity $d\rho/dT>0$1, in which “the TCR value tends to be more negative with increasing $d\rho/dT>0$2” (Kovaleva et al., 2020).

The effect is reported over a broad temperature range. One account states that the rule is valid from $d\rho/dT>0$3–30 K up to room temperature and above in high-resistive metallic alloys (Gantmakher, 2011). Another emphasizes that Mooij correlations are universally observed at elevated temperature spanning ambient conditions, which is important because explanations based on long-distance phase coherence become less compelling under strong dephasing by phonons (Ciuchi et al., 2018).

A standard phenomenological parametrization used in the experimental nanoscale literature is

$d\rho/dT>0$4

with $d\rho/dT>0$5 identified as the TCR extracted from linear fits in a prescribed temperature interval (Kovaleva et al., 2020). In the ab initio V$d\rho/dT>0$6Al$d\rho/dT>0$7 study, the same empirical trend is described as a crossover to negative TCR as residual resistivity increases, with alloys above roughly $d\rho/dT>0$8 tending to exhibit negative TCR (Csire et al., 25 Aug 2025).

2. Transport scales and the breakdown of semiclassical additivity

A central background scale for Mooij correlations is the Mott–Ioffe–Regel limit, written in the cited works as

$d\rho/dT>0$9

and associated with the condition that the electron mean free path is approximately equal to the interatomic distance (Gantmakher, 2011). In the weak-localization treatment, the disorder strength is parameterized by

$d\rho/dT<0$0

where $d\rho/dT<0$1 is the mean time between static scatterings and $d\rho/dT<0$2 is the Fermi temperature, with $d\rho/dT<0$3 corresponding to maximal allowed disorder by the Ioffe–Regel criterion (Gantmakher, 2011).

At low disorder, the cited literature writes the classical resistivity as

$d\rho/dT<0$4

with

$d\rho/dT<0$5

for $d\rho/dT<0$6 (Gantmakher, 2011). This is the regime in which Matthiessen’s rule, or additive independent scattering from defects and phonons, is expected to hold approximately.

The Mooij regime is precisely the regime in which this additive picture fails. One strong-coupling account states that “the scattering events from impurities or thermal excitations can no longer be considered as additive independent processes, as asserted by Matthiessen's rule” (Ciuchi et al., 2018). In that treatment, the total scattering rate is written as

$d\rho/dT<0$7

where the vertex factor $d\rho/dT<0$8 quantifies the correlation between elastic disorder scattering and inelastic phonon scattering (Ciuchi et al., 2018). This places the Mooij correlation in a broader framework: it is not merely an anomaly of the resistivity slope, but a manifestation of correlated scattering in a regime where semiclassical transport ceases to be adequate.

3. Weak-localization interpretation

One influential explanation attributes the Mooij rule to weak localization in highly disordered metals (Gantmakher, 2011). In this approach, constructive interference of time-reversed electronic paths reduces conductivity and produces a quantum correction that becomes important when disorder is strong. The correction is written as

$d\rho/dT<0$9

where $100\div150\,\mu\Omega\cdot\text{cm}$0 is the phase-coherence length (Gantmakher, 2011).

The total conductivity is then expressed as

$100\div150\,\mu\Omega\cdot\text{cm}$1

Within this picture, low-resistivity alloys remain dominated by the classical phonon contribution and therefore exhibit $100\div150\,\mu\Omega\cdot\text{cm}$2, whereas high-resistivity alloys develop sufficiently strong quantum corrections that $100\div150\,\mu\Omega\cdot\text{cm}$3 (Gantmakher, 2011).

The cited analysis gives a condition for the boundary between the two regimes by setting $100\div150\,\mu\Omega\cdot\text{cm}$4: $100\div150\,\mu\Omega\cdot\text{cm}$5 The resulting separating line is stated to lie in the range $100\div150\,\mu\Omega\cdot\text{cm}$6–0.8, corresponding to $100\div150\,\mu\Omega\cdot\text{cm}$7 (Gantmakher, 2011). In this account, the Mooij rule does not imply an Anderson metal–insulator transition in high-carrier-density metallic alloys; rather, it reflects a disorder regime in which weak localization persists while the system remains metallic (Gantmakher, 2011).

4. Strong-coupling disorder–phonon theory and polaronic renormalization

A distinct theoretical account argues that Mooij correlations originate from a strong-coupling local mechanism that is “physically distinct and unrelated to Anderson localization” (Ciuchi et al., 2018). In that framework, the relevant physics is the interplay of strong disorder and lattice deformations in bulk three-dimensional metals at high temperatures. The proposed mechanism is polaronic strong disorder renormalization: the lattice locally responds to the impurity potential, generating an induced potential $100\div150\,\mu\Omega\cdot\text{cm}$8 such that

$100\div150\,\mu\Omega\cdot\text{cm}$9

The key claim is that, in a deformable lattice with finite electron–phonon coupling, local lattice distortions renormalize the effective disorder potential and universally create a dip in the disorder distribution $\rho^* \sim 300\,\mu\Omega\cdot\text{cm}$0 near the Fermi level when disorder is strong (Ciuchi et al., 2018). The cited paper states that $\rho^* \sim 300\,\mu\Omega\cdot\text{cm}$1 at large disorder, thereby pushing site energies away from the Fermi level and depleting $\rho^* \sim 300\,\mu\Omega\cdot\text{cm}$2 near $\rho^* \sim 300\,\mu\Omega\cdot\text{cm}$3.

In this theory, the Mooij regime follows from the correlated nature of elastic and inelastic scattering, encoded in the vertex factor $\rho^* \sim 300\,\mu\Omega\cdot\text{cm}$4. The sign of the TCR is said to hinge on $\rho^* \sim 300\,\mu\Omega\cdot\text{cm}$5, which is governed by the curvature of the disorder distribution at the Fermi level. At sufficient disorder, the polaronic dip suppresses or makes $\rho^* \sim 300\,\mu\Omega\cdot\text{cm}$6 negative, producing $\rho^* \sim 300\,\mu\Omega\cdot\text{cm}$7 (Ciuchi et al., 2018).

Methodologically, this treatment uses dynamical mean-field theory combined with the coherent potential approximation (DMFT-CPA) and evaluates transport with the Kubo-Greenwood formula, explicitly omitting non-local interference and therefore not incorporating Anderson localization (Ciuchi et al., 2018). It also proposes the explicit linear relation

$\rho^* \sim 300\,\mu\Omega\cdot\text{cm}$8

where $\rho^* \sim 300\,\mu\Omega\cdot\text{cm}$9 is the “flat” resistivity where TCR changes sign and $d\rho/dT$0 is a material-specific constant (Ciuchi et al., 2018). This theory therefore interprets the Mooij correlation as a high-temperature consequence of local disorder–lattice correlations rather than low-temperature quantum interference.

5. Ab initio calculations and nanoscale control

Recent work on V$d\rho/dT$1Al$d\rho/dT$2 alloys examines the Mooij correlation with a first-principles transport framework based on KKR-CPA, the Kubo-Greenwood formalism, and a CPA-based alloy analogy model for thermal atomic vibrations (Csire et al., 25 Aug 2025). In that work, the dc conductivity is decomposed as

$d\rho/dT$3

with $d\rho/dT$4 described as a local or on-site term and $d\rho/dT$5 as a nonlocal term including vertex corrections (Csire et al., 25 Aug 2025). Finite-temperature conductivity is obtained from

$d\rho/dT$6

The principal result is that the crossover to negative TCR is reproduced quantitatively in V$d\rho/dT$7Al$d\rho/dT$8, with the calculated TCR crossing zero at $d\rho/dT$9 and experiment at $\rho$0 (Csire et al., 25 Aug 2025). The cited interpretation is that the negative TCR does not arise from quantum coherence effects in the relevant temperature regime, but from a non-Boltzmann local conductivity term, $\rho$1, whose temperature dependence follows the density of states at the Fermi level. As thermal disorder smears sharp density-of-states features, $\rho$2 can increase with temperature and overwhelm the decrease of the Boltzmann-like contribution $\rho$3, thereby yielding negative TCR (Csire et al., 25 Aug 2025).

A complementary experimental direction studies Ta–FeNi nanoscaled multilayered films, where Mooij correlations are actively tuned by FeNi nanoisland coverage across a percolation threshold (Kovaleva et al., 2020). In the discontinuous nanoisland regime, the Ta layer is the primary conductor, the Drude weight is suppressed, and the TCR is strongly negative; the paper reports values as large as $\rho$4cm and $\rho$5 ppm/K for a nanoisland sample (Kovaleva et al., 2020). Upon increasing FeNi thickness across percolation, the FeNi layer becomes conducting, the Drude response increases, resistivity decreases, and the TCR shifts toward positive values. Optical conductivity extracted by spectroscopic ellipsometry is modeled with a Drude-Lorentz dielectric function,

$\rho$6

and low-energy bands in the 1–2 eV range are associated in that work with localized and correlated states (Kovaleva et al., 2020).

6. Interpretation, scope, and unresolved issues

The literature does not present a single settled mechanism for the Mooij correlation. Instead, it presents several non-equivalent frameworks that apply to overlapping but not identical regimes. The weak-localization account emphasizes quantum interference and phase coherence in disordered metals (Gantmakher, 2011). The DMFT-CPA polaronic account argues that the relevant high-temperature phenomenology is unrelated to Anderson localization and is instead driven by local disorder–phonon correlations and the breakdown of Matthiessen’s rule (Ciuchi et al., 2018). The ab initio V–Al study attributes the effect, at intermediate and high temperatures, to a non-Boltzmann local conductivity channel rather than to localization physics (Csire et al., 25 Aug 2025).

A recurring misconception is that negative TCR necessarily signals insulating behavior. In the weak-localization analysis, the observed negative $\rho$7 in high-resistive metallic alloys “does not signal a transition to insulating behavior” (Gantmakher, 2011). Conversely, the Ta–FeNi multilayer study interprets the most extreme regime, in which conductivity falls below the MIR limit by about 60%, as potentially associated with non-ergodicity and purely quantum many-body localization phenomena that “need to be challenged further” (Kovaleva et al., 2020). This suggests that the meaning of negative TCR depends on materials class, temperature window, dimensionality, and the degree to which disorder is accompanied by lattice deformation, orbital structure, or magnetic inhomogeneity.

The overall significance of the Mooij correlation is therefore twofold. Empirically, it provides a compact diagnostic of anomalous transport in highly disordered metals. Theoretically, it marks the point at which independent-scattering and simple Boltzmann pictures become insufficient, forcing recourse to quantum interference, strong local renormalization, non-Boltzmann transport channels, or combinations thereof (Ciuchi et al., 2018). A plausible implication is that the Mooij correlation should be understood less as a single mechanism than as a transport phenomenology that exposes different failure modes of semiclassical metallic conduction across different disordered systems.