Transient Localization in Quantum Systems

• Transient Localization is a transport regime where quantum carriers exhibit temporary localization due to internal, slow, and thermally induced randomness.
• This phenomenon reconciles non-Drude optical conductivity with low carrier mobilities observed in systems like organic semiconductors and quantum magnets.
• Numerical techniques such as finite-temperature Lanczos reveal that the balance between localization timescales and environmental fluctuations governs transport behavior.

Searching arXiv for relevant papers on transient localization. Transient localization (TL) denotes a regime in which quantum carriers or excitations undergo localization over finite time or frequency windows because they move in an effectively random environment that is generated internally and evolves slowly, rather than by externally imposed static disorder. In the electron–boson problem, numerically exact finite-temperature results for the quantum Holstein model show that thermally populated bosons create emergent randomness that produces Anderson-like localization at transient times before diffusion can set in, despite the absence of extrinsic disorder (Rammal et al., 2023). Earlier work on crystalline organic semiconductors formulated TL as the quantum localization of the electronic wavefunction at timescales shorter than the typical molecular motions, thereby reconciling non-Drude optical response, low mobilities, and metallic-like temperature dependence in systems with large dynamical lattice disorder [(Fratini et al., 2015); (Ciuchi et al., 2010)]. Subsequent studies have generalized the same logic to dissipative electron–phonon models in molecular semiconductors (Janković, 20 Aug 2025), to electrons coupled to two-level systems (Rammal et al., 24 Sep 2025), and to gapless quantum magnets where fractionalization produces slow visons and mobile spinons (Feng et al., 8 Sep 2025). In these settings, TL is not full asymptotic localization: it is a finite-time manifestation of quantum interference in a self-generated, quasi-static landscape whose dynamics eventually restore transport.

1. Definition and conceptual scope

Transient localization is a transport regime intermediate between fully Anderson-localized motion in a static random potential and conventional diffusive or semiclassical transport in a rapidly fluctuating environment. In the organic-semiconductor formulation, carriers become localized at timescales shorter than the molecular motion timescale, while the evolution of the dynamical disorder later allows them to escape and diffuse [(Fratini et al., 2015); (Ciuchi et al., 2010)]. In the electron–boson problem, the same structure appears when abundant thermal bosons generate random local energy shifts and the boson dynamics are slow compared with the electronic localization time (Rammal et al., 2023).

A central distinction is therefore between extrinsic disorder and emergent randomness. The 2023 Holstein study explicitly demonstrates quantum localization processes in the absence of extrinsic disorder, with localization caused by the emergent randomness resulting from a large thermal boson population (Rammal et al., 2023). The 2025 two-level-system study reaches an analogous conclusion for inelastic scatterers: sufficiently slow TLS fluctuations generate a quasi-static, dynamically fluctuating landscape that transiently localizes the electron, whereas fast inelastic scatterers suppress localization and restore a more conventional regime governed by independent scattering events (Rammal et al., 24 Sep 2025). In the Kitaev-ladder setting, slow visons act as quasi-static disorder for light spinons in a translation-invariant model and at zero temperature, again without any quenched disorder (Feng et al., 8 Sep 2025).

This suggests a broad operational definition: TL arises when internal degrees of freedom produce a random environment that is effectively static on the timescale required for quantum interference to localize motion, but not static on asymptotically long timescales. A plausible implication is that the phenomenon should be sought whenever a system combines mobile carriers with slow, thermally populated, fractionalized, or otherwise sluggish auxiliary modes.

2. Microscopic mechanisms

The mechanism emphasized in the Holstein problem is local electron coupling to quantum bosons,

$$H = -t \sum_{\langle ij \rangle} c_i^\dagger c_j + \omega_0 \sum_i a_i^\dagger a_i + g \sum_i c_i^\dagger c_i (a_i^\dagger + a_i),$$

with calculations performed using the finite-temperature Lanczos method (FTLM) (Rammal et al., 2023). At finite temperature, abundant thermal bosons create random local energy shifts; in the static boson limit, the problem reduces to an Anderson localization scenario with site-energy randomness of variance

$\Delta^2 = 2\lambda\omega_0 / \tanh(\omega_0/2T),$

which increases with temperature and electron–boson coupling (Rammal et al., 2023). When $\omega_0 \ll t$ and $T \gtrsim \omega_0$, electrons effectively perceive a quasi-static random potential and undergo Anderson-like localization over transient timescales (Rammal et al., 2023).

In crystalline organic semiconductors, the microscopic origin is dynamical disorder generated by large thermal intermolecular motions that modulate transfer integrals. A commonly studied model is the Peierls or Su–Schrieffer–Heeger form

$$H= - J\sum_i [1-\alpha (u_i-u_{i+1})] \; (c^+_i c_{i+1} + c^+_{i+1} c_{i}) + H_{vib},$$

with

$$H_{vib}=\sum_i \frac{M \omega_0^2}{2} u_i^2+\sum_i \frac{p_i^2}{2M},$$

and dimensionless coupling

$$\lambda = \alpha^2 \left(\frac{\hbar}{2M\omega_0}\right) \frac{J}{\hbar\omega_0}$$

(Fratini et al., 2015). Here the fluctuating transfer integrals act as dynamical disorder: if frozen, all states would be Anderson localized, but because the disorder itself moves, the localization is only transient (Fratini et al., 2015).

The same slowness criterion appears in other models. For electrons coupled locally to two-level systems,

$$H = -t \sum_{\langle i j \rangle} c_{i}^{\dagger} c_{j} + \frac{\omega_0}{2} \sum_i \sigma^z_i - g \sum_i c_{i}^{\dagger} c_{i} \sigma^x_i,$$

slow TLS fluctuations generate an effectively static random potential over short times, mapping to a binary disorder model in the $\omega_0 \to 0$ limit (Rammal et al., 24 Sep 2025). In gapless quantum magnets, fractionalization into slow visons and mobile spinons yields an intrinsic, quasi-static random background for the heat-carrying spinons (Feng et al., 8 Sep 2025). Across these models, the common ingredient is not the statistics of the slow sector but its temporal slowness relative to the localization dynamics (Rammal et al., 24 Sep 2025).

3. Timescales, localization length, and transport formulas

TL is organized by a competition between a localization timescale and an environmental fluctuation timescale. In the electron–boson study, the localization time is

$\tau_L \sim 1/\omega_L,$

while the boson fluctuation time is

$\Delta^2 = 2\lambda\omega_0 / \tanh(\omega_0/2T),$0

Transient localization exists when

$\Delta^2 = 2\lambda\omega_0 / \tanh(\omega_0/2T),$1

equivalently when $\Delta^2 = 2\lambda\omega_0 / \tanh(\omega_0/2T),$2, so that electrons can interfere before boson motion erases the disorder (Rammal et al., 2023). The localization length $\Delta^2 = 2\lambda\omega_0 / \tanh(\omega_0/2T),$3 is extracted from the displaced Drude peak position via

$\Delta^2 = 2\lambda\omega_0 / \tanh(\omega_0/2T),$4

This gives

$\Delta^2 = 2\lambda\omega_0 / \tanh(\omega_0/2T),$5

a relation used as a quantitative bridge between spectroscopy and real-space localization (Rammal et al., 2023).

The organic-semiconductor TL scenario packages the same physics into a relaxation time approximation (RTA). Starting from the velocity autocorrelation function $\Delta^2 = 2\lambda\omega_0 / \tanh(\omega_0/2T),$6 of the frozen-disorder reference problem, dynamical disorder is incorporated through

$\Delta^2 = 2\lambda\omega_0 / \tanh(\omega_0/2T),$7

where $\Delta^2 = 2\lambda\omega_0 / \tanh(\omega_0/2T),$8 is the inelastic timescale associated with molecular motion [(Fratini et al., 2015); (Ciuchi et al., 2010)]. The resulting diffusion constant and mobility are

$\Delta^2 = 2\lambda\omega_0 / \tanh(\omega_0/2T),$9

and

$\omega_0 \ll t$0

respectively (Fratini et al., 2015). The 2023 Holstein work adopts the same Thouless-style structure for the TL mobility,

$\omega_0 \ll t$1

with $\omega_0 \ll t$2 obtained from $\omega_0 \ll t$3 and $\omega_0 \ll t$4 (Rammal et al., 2023).

These formulas express the defining balance of TL: transport is finite because the environment eventually evolves, but suppressed because the carrier explores only a finite localization length before the environment re-randomizes. In the static limit, $\omega_0 \ll t$5 or $\omega_0 \ll t$6, the construction reduces to Anderson localization with vanishing mobility (Fratini et al., 2015, Rammal et al., 2023). In the fast-fluctuation limit, the environment changes too rapidly to sustain interference, and conventional diffusive transport is recovered (Rammal et al., 2023, Rammal et al., 24 Sep 2025).

4. Optical and dynamical signatures

The most characteristic spectroscopic signature of TL is the displaced Drude peak (DDP): a finite-frequency peak in the optical conductivity, replacing or coexisting with a conventional zero-frequency Drude peak. In the Holstein model, numerically exact spectra show that at low temperature only inelastic boson emission is visible, whereas at higher $\omega_0 \ll t$7 and dense thermal boson population the absorption shifts and broadens into a finite-frequency DDP (Rammal et al., 2023). The DDP matches the static-disorder result at frequencies $\omega_0 \ll t$8, confirming Anderson-like localization on those time and frequency scales (Rammal et al., 2023). Its presence correlates with the amount of self-generated bosonic disorder, temperature, and $\omega_0 \ll t$9, and it disappears once boson dynamics become too fast, $T \gtrsim \omega_0$0 (Rammal et al., 2023).

The organic-semiconductor literature had already identified a finite-frequency absorption peak as a direct consequence of TL rather than free-carrier absorption. A key formal result links optical conductivity to the time-dependent quantum spread,

$T \gtrsim \omega_0$1

with inverse transform

$T \gtrsim \omega_0$2

so that sub-gap optical measurements can be used to reconstruct real-time transport dynamics, including localization and subsequent diffusion (Ciuchi et al., 2010). This relation underpins the use of spectroscopy as a probe of TL.

Later work refined the interpretation of low-frequency features. In a dissipative Peierls-type electron–phonon model treated with the dissipaton equations of motion (DEOM), increasing vibrational damping suppresses the superdiffusive transient and removes very-low-frequency optical-conductivity upturns, while preserving the DDP at $T \gtrsim \omega_0$3 (Janković, 20 Aug 2025). That study concludes that low-frequency enhancements are artifacts of undamped vibrations and that realistic damping makes optical-conductivity profiles increasingly similar to TL predictions (Janković, 20 Aug 2025). This clarifies a potential misconception: not every deviation from a Drude peak signals TL in the same manner; the robust hallmark in this body of work is the displaced finite-frequency peak together with suppressed low-frequency conductivity.

In the Kitaev ladder, the dynamical conductivity is strongly suppressed at intermediate and finite but low frequencies, with restoration only at ultralow frequencies when vison motion becomes relevant (Feng et al., 8 Sep 2025). Although the phenomenology is expressed in thermal rather than charge transport, the same structure appears: a suppressed response over experimentally relevant scales and delayed recovery at asymptotically long times or tiny frequencies (Feng et al., 8 Sep 2025).

5. Numerical frameworks and the role of quantum interference

A recurring conclusion is that TL is a genuinely quantum interference effect and is therefore missed by approaches that average over the fluctuating environment too early or neglect nonlocal vertex corrections. The Holstein study uses FTLM to provide unbiased, exact results on finite-size chains while controlling finite-size and boson-number effects, and emphasizes that semiclassical or mean-field approaches such as DMFT omit the interference processes responsible for localization (Rammal et al., 2023). The DDP is absent in DMFT and other self-averaging approaches, which is presented as direct evidence that the effect is beyond local self-energy physics (Rammal et al., 2023).

The two-level-system study sharpens this point by directly comparing FTLM and DMFT. FTLM captures enhanced resistivity and DDPs for slow TLSs, while DMFT fails in precisely that regime and converges to FTLM only when the scatterers become fast and localization corrections become negligible (Rammal et al., 24 Sep 2025). The study identifies TL as arising from the nonlocal vertex corrections neglected in DMFT, and therefore as a genuine quantum effect rather than a renormalized scattering rate (Rammal et al., 24 Sep 2025).

In molecular semiconductors, DEOM serves as a “numerically exact” real-time method for a dissipative nonlocal carrier–phonon model with Brownian-oscillator spectral density,

$T \gtrsim \omega_0$4

The method computes the finite-temperature current autocorrelation function

$T \gtrsim \omega_0$5

and hence the dynamical mobility

$T \gtrsim \omega_0$6

including full quantum correlations and equilibrium entanglement (Janković, 20 Aug 2025). The comparison between DEOM and TL phenomenology shows that the phenomenological picture becomes quantitatively and qualitatively accurate once vibrational damping is sufficiently strong (Janković, 20 Aug 2025).

A broader inference is that TL is best regarded not as a particular approximation but as a dynamical regime that can be detected in exact or controlled calculations and then summarized phenomenologically by an RTA-type envelope or a localization-length extraction procedure.

6. Material realizations and broader generalizations

The earliest major application of TL was to crystalline organic semiconductors, where weak van der Waals intermolecular interactions, narrow bandwidths, and large thermal molecular motions place transport outside both conventional Bloch–Boltzmann theory and activated small-polaron pictures [(Fratini et al., 2015); (Ciuchi et al., 2010)]. Experimental motivations include low mobilities, non-Drude optical conductivity, and evidence for finite spatial extent of carrier wavefunctions despite metallic-like temperature dependence (Fratini et al., 2015). In this setting TL explains why mobility can fall below the Mott–Ioffe–Regel limit while remaining nonactivated and why optical spectra develop finite-frequency peaks (Fratini et al., 2015).

The Holstein-model results extend the relevance of TL beyond semiclassical organic-transport models by placing it on a fully quantum footing in clean interacting electron–boson systems with quantized bosons (Rammal et al., 2023). That study further suggests applicability to quantum materials with slow bosonic modes, including cases near quantum criticality, and proposes a unified explanation for displaced Drude peaks in various bad metals and possibly strange metals (Rammal et al., 2023). Because this is presented as a suggestion rather than a catalog of confirmed materials, it is best interpreted as an extension of the organizing principle rather than a complete materials classification.

In dissipative molecular-semiconductor models, DEOM calculations indicate that for parameters representative of room-temperature hole transport in single-crystal rubrene, the TL phenomenology is already established in the underdamped-oscillator regime and remains consistent with experiment under reasonable variations of the damping constant (Janković, 20 Aug 2025). The result is significant because it reconciles numerically exact transport dynamics with the empirical success of Drude–Anderson-type phenomenology in terahertz spectroscopy (Janković, 20 Aug 2025).

The 2025 Kitaev-ladder study generalizes TL from charge transport to heat transport in fractionalized quantum matter. There, a gapless quantum spin liquid can show linear-in-$T \gtrsim \omega_0$7 specific heat from gapless spinons together with vanishingly small thermal conductivity over accessible frequencies because slow visons transiently localize the spinons (Feng et al., 8 Sep 2025). This reframes suppressed heat transport not as evidence against spin-liquid physics but as a possible intrinsic consequence of fractionalization (Feng et al., 8 Sep 2025).

The TLS study, finally, shows that the essential prerequisite is the slowness of inelastic scatterers rather than their bosonic character (Rammal et al., 24 Sep 2025). That observation broadens TL into a general transport concept for clean interacting systems with internal degrees of freedom that fluctuate slowly enough to mimic disorder over mesoscopic time windows.

7. Related usages and terminological cautions

The term “transient localization” also appears in nonlinear lattice dynamics, where it refers to spontaneous, temporary bursts of localized energy in a diatomic nonlinear Lennard–Jones chain (Lepri et al., 2021). In that context the mechanisms are thermal excitation of discrete breathers in the phonon gap and nonlinear coupling of fast light particles to slow heavy-particle vibrations, described by a multiple-scale approximation and a stochastic single-particle model (Lepri et al., 2021). Although this shares the generic idea of temporary spatial confinement generated by internal dynamics and timescale separation, it is conceptually distinct from Anderson-like quantum localization of carriers in a quasi-static random landscape.

A separate terminological overlap occurs in machine-learning work where “TL” is used as shorthand for “Task Learning” within in-context learning, identified through Task Subspace Logit Attribution and attention-head analysis (Yang et al., 29 Sep 2025). That usage is unrelated to localization in condensed-matter or transport theory.

Within quantum transport itself, several misconceptions are directly addressed by the cited literature. First, TL does not require extrinsic or quenched disorder: the Holstein, TLS, and Kitaev-ladder studies all emphasize self-generated randomness in clean systems (Rammal et al., 2023, Rammal et al., 24 Sep 2025, Feng et al., 8 Sep 2025). Second, TL is not equivalent to ordinary inelastic scattering, because it depends on quantum interference and therefore disappears in methods that neglect vertex corrections or in regimes where the fluctuating environment is too fast (Rammal et al., 2023, Rammal et al., 24 Sep 2025). Third, TL does not imply permanent insulation; transport is restored on sufficiently long timescales once the environment evolves, though those scales may be experimentally inaccessible (Fratini et al., 2015, Rammal et al., 2023, Feng et al., 8 Sep 2025).

Taken together, these results establish TL as a unifying framework for anomalous transport in dynamically disordered or internally fluctuating systems. Its defining content is the emergence of Anderson-like localization over transient windows set by the relative slowness of environmental modes, with direct fingerprints in optical or dynamical response and with quantitative characterization through localization lengths, autocorrelation functions, and fluctuation times extracted from exact or controlled calculations.