---
title: Negative Differential Thermal Resistance
url: https://www.emergentmind.com/topics/negative-differential-thermal-resistance-ndtr
type: topic
---

# Negative Differential Thermal Resistance

Negative Differential Thermal Resistance (NDTR) is the nonlinear transport phenomenon in which the steady heat current decreases when the applied thermal bias increases, i.e. over some interval the differential response satisfies $\partial J/\partial(\Delta T)<0$. It is the thermal analogue of negative differential resistance in electronics and is a central ingredient in proposals for thermal transistors, switches, logic gates, amplifiers, memories, and current limiters [2211.06872]. Across the literature, NDTR has been reported in nonlinear lattices, spin chains, graded crystals, graphene nanoribbons, classical fluids, photonic heat circuits, macroscopic heterojunctions, and, more recently, as a point of contrast between classical and quantum low-temperature transport [1407.7087]. The phenomenon is not governed by a single universal microscopic mechanism; rather, the common structure is a competition between increasing thermal drive and a stronger bias-induced suppression of effective transport.

## 1. Definition, terminology, and operational criteria

In its standard form, NDTR denotes the regime in which the heat current $J$ decreases as the temperature difference is increased, so that
\[
\frac{\partial J}{\partial(\Delta T)}<0.
\]
This operational definition is used across lattice, spin, fluid, and low-temperature transport models [2509.14027]. Several works use the closely related term negative differential thermal conductance (NDTC); in the macroscopic heterojunction formulation, the differential thermal conductance is defined as $k=\mathrm{d}J/\mathrm{d}T_{\rm R}$ and the differential thermal resistance as $r=1/k$, so NDTR corresponds to $k<0$ and $r<0$ [2302.14065].

The way the bias is varied is not incidental. Many studies hold one bath fixed and vary the other, so increasing $\Delta T$ also changes the average temperature $\overline{T}$. This appears explicitly in graphene nanoribbons, where $T_L=600\ \mathrm{K}$ is fixed and increasing $\Delta T$ means lowering $T_R$, and in the square-well transport problem, where the thermal bias is effectively increased by lowering the cold bath temperature while the hot bath is fixed [1407.7087]. A broader scaling analysis formalizes this dependence by writing
\[
j=\frac{\kappa_e(\overline{T},\Delta T)\Delta T}{N},
\]
with $\overline{T}=(T_+ + T_-)/2$, and derives the general NDTR criterion
\[
n_1 n_2 < -(1+n_3),
\]
where $n_1$ encodes the chosen path in the $(\overline{T},\Delta T)$ plane, and $n_2,n_3$ encode the dependence of the effective conductivity on $\overline{T}$ and $\Delta T$ [1401.0178]. This formulation is important because it shows that NDTR is not only a material property; it also depends on how the external driving protocol is imposed.

## 2. Transport formalisms used to analyze NDTR

NDTR has been analyzed within several transport frameworks, each emphasizing a different bottleneck. In nonlinear lattice models, one common route is a Landauer-type description with temperature-dependent transmission. For two weakly coupled nonlinear lattices, the current is written in classical form as
\[
j=\frac{k_B(T_+-T_-)}{2\pi}\int_{\omega_{\min}}^{\omega_{\max}}\mathcal{T}(\omega)\,d\omega,
\]
with nonlinearity incorporated through self-consistent phonon theory, so that the transmission depends on temperature via renormalized phonon spectra [1001.3782]. In the ballistic Frenkel–Kontorova treatment, the low-temperature current is likewise interpreted through a Landauer equation, with the turnover attributed to the temperature dependence of phonon occupation factors in the ballistic regime [0905.3792].

In diffusive settings, the reference framework is Fourier transport. A finite one-dimensional system with local conductivity $\kappa(x,T)$ obeys
\[
q(x)=-\kappa(x,T)\frac{dT}{dx},
\]
and, in the absence of abrupt junctions, a rigorous result shows that NDTC cannot occur when one end temperature is fixed; with a single junction, NDTC requires temperature-dependent thermal contact resistance, and the paper derives a necessary and sufficient condition in terms of that junction response [1201.3054]. A complementary macroscopic heterojunction model makes this dependence explicit by introducing an interfacial thermal resistance
\[
R_{\rm i}(T_{\rm mL},T_{\rm mR})=C(T_{\rm mL}+T_{\rm mR})^\alpha,
\]
which leads, under constant-conductivity approximation, to
\[
J=\frac{A(T_{\rm R}-T_{\rm L})}{2+AC(T_{\rm R}+T_{\rm L})^\alpha}.
\]
Within this model, NDTR requires $\alpha>1$, because the interface resistance must increase strongly enough with temperature to overpower the direct increase in thermal driving [2302.14065].

Other platforms adopt domain-specific formalisms. Classical and quantum transport in the one-dimensional infinite square well are compared by using Maxwell or MCMC Maxwell baths in the classical problem and a Lindblad master equation in the quantum problem,
\[
\frac{d\hat{\rho}}{dt}=-\frac{i}{\hbar}[\hat{H},\hat{\rho}]+\mathcal{D}_L(\hat{\rho})+\mathcal{D}_R(\hat{\rho}),
\]
with the steady-state current extracted from the energy continuity equation [2509.14027]. In photonic heat circuits, the current is instead determined by impedance-controlled electromagnetic transmission,
\[
\dot Q_{\gamma}(T_S,T_D)=\frac{1}{h}\int_0^\infty \omega\,\tau(\omega,T_S,T_D)\,[n(\omega,T_S)-n(\omega,T_D)]\,d\omega,
\]
so the differential anomaly is tied to abrupt temperature-induced impedance mismatch rather than phonon or particle transport [2112.12627].

## 3. Microscopic mechanisms across major model classes

A large fraction of the NDTR literature can be organized around the statement that the thermal bias increases the numerator of the transport problem while some temperature-dependent bottleneck increases the denominator more rapidly. In weakly coupled nonlinear lattices, this bottleneck is often the thermal boundary conductance. Defining
\[
j=\Delta T\,\sigma,
\]
with $\sigma$ the effective boundary conductance, NDTR occurs when the decrease in $\sigma$ with bias dominates the increase in $\Delta T$ [1001.3782]. In the two-segment Frenkel–Kontorova chain, the same logic is reframed as a crossover from diffusive to ballistic transport: at low temperature, the ballistic occupation factor becomes strongly temperature dependent, so decreasing the hot-side temperature can reduce current even as the bias grows [0905.3792].

In momentum-conserving and spin systems, the suppressing mechanism is often described as a nonlinear bottleneck in the bulk or at an interface. In the two-segment Fermi–Pasta–Ulam lattice, NDTR is attributed to a competition between the positive effect of linear coupling and the negative effect induced by nonlinearity; it is absent in the harmonic limit, promoted by stronger quartic coupling, suppressed by stronger intersegment coupling, and anomalously appears at higher temperatures rather than lower temperatures [1110.4942]. In the two-segment classical Heisenberg chain, the crucial ingredient is a mechanism that impedes current in the bulk or across the interface, produced by smaller interface coupling, stronger local magnetic field, or interface anisotropy [1307.5421]. In the Heisenberg chain with a spatially varying magnetic field, the field gradient itself acts as the bottleneck: stronger local fields stiffen spin motion, and once magnetic pinning dominates over thermal agitation, the current can saturate and then decrease with further bias [1411.5200].

In fluids, the dominant mechanism is bath-induced suppression of bath-to-bath exchange. For hard-point gases and higher-dimensional MPC fluids, lowering the cold-bath temperature slows particles emitted from the cold side, thereby reducing their collision frequency with the hot bath; when the reduction in collision frequency outweighs the increased temperature difference, NDTR appears [2211.06872]. Gravity amplifies this mechanism in MPC fluids by making it harder for particles from the colder side to traverse the system against the field, thereby lowering the bias threshold for NDTR and extending the effect into more strongly interacting regimes [2510.08909].

In photonic circuits, the suppressing mechanism is neither phonon scattering nor particle slowdown but transmission mismatch. The photon transmission coefficient is maximized at impedance matching, and if increasing the source temperature drives the source impedance abruptly away from the matched condition, the radiative heat flux decreases even though the source is hotter [2112.12627]. In macroscopic heterojunctions, the same structure appears at the interface scale: increasing $T_{\rm R}$ both increases the driving force and raises the interfacial thermal resistance, and NDTR emerges when the latter wins [2302.14065].

## 4. Dependence on dimensionality, size, interactions, and driving protocol

Dimensionality and geometry can create or destroy NDTR, but not in a uniform way. In graphene nanoribbons, NDTR is absent in the effectively one-dimensional limit, appears in an intermediate quasi-two-dimensional regime, and disappears again as the system becomes more three-dimensional through increased width or layer number [1407.7087]. The reported interpretation is that intermediate-width single-layer ribbons support edge-localized phonon modes whose effect on conductivity becomes sufficiently strong under increasing bias, whereas in very narrow ribbons the behavior resembles a one-dimensional chain without on-site potential and in wider or multilayer systems the relevant edge contribution is diluted or altered [1407.7087].

Finite-size dependence is equally model specific. In several lattice models, NDTR weakens or vanishes in the thermodynamic limit. The Frenkel–Kontorova ballistic study reports that increasing system size drives a crossover to diffusive transport and removes NDTR [0905.3792]. The deformable FK lattice similarly shows NDTR for smaller systems such as $N=32$ and $N=128$ but not for $N=1024$, identifying it as a small-size effect in that model [1102.0109]. Double-NDTR in homogeneous lattices with nonlinear on-site potentials also shrinks with increasing $N$ and eventually vanishes [1208.3008]. By contrast, the anomalous two-segment FPU lattice reports an opposite trend relative to FK and $\phi_4$ systems: NDTR is absent for small systems and the NDTR region appears and shifts from large to small temperature differences as system size increases [1110.4942]. This suggests that size dependence is mechanism specific rather than universal.

Interaction strength can either preserve or erase NDTR. In three-dimensional MPC fluids, bath-induced NDTR survives over a wide range of weakly interacting fluids, roughly for $\tau>0.1$, but disappears when interactions become too strong because collisions boost the speed of slow cold-side particles and restore more conventional Fourier-like transport [2211.06872]. Gravity shifts that boundary: with gravity aligned with the thermodynamic force, the same mechanism can operate even at stronger interactions, for example at $\tau=0.1$ when $g=0.2$ [2510.08909]. Mixed composition need not remove the effect; both higher-dimensional MPC fluids and gravity-modulated MPC fluids report robustness in binary mixtures over large mass ratios and composition ranges [2211.06872].

A recurring misconception is that NDTR should be expected whenever transport is nonlinear. The scaling analysis shows instead that its appearance can depend strongly on the constraint used to vary temperatures, because the condition
\[
n_1 n_2 < -(1+n_3)
\]
contains the protocol exponent $n_1$ as an external ingredient [1401.0178]. Another misconception is that any diffusive one-dimensional conductor can exhibit NDTR under strong enough bias. The Fourier-law analysis rules this out for a finite 1D system with no abrupt junctions when one boundary temperature is fixed; in that class of systems, a temperature-dependent thermal contact resistance is required [1201.3054].

## 5. Classical–quantum divergence at low temperatures

The low-temperature square-well study provides a particularly sharp statement about the limits of classical intuition. The model consists of a single particle in a one-dimensional infinite square well of length $L$, with Hamiltonian
\[
H=\frac{p^2}{2m}+V(x),\qquad
V(x)=
\begin{cases}
0, & 0<x<L,\\
\infty, & \text{otherwise},
\end{cases}
\]
and thermal contacts at the left and right boundaries [2509.14027]. In the classical setting, simulations reveal NDTR: lowering the cold bath temperature can reduce the steady-state current because the cold reservoir becomes less able to thermalize the particle, the return time grows strongly, and the net throughput is suppressed [2509.14027]. The same qualitative effect persists under an MCMC Maxwell bath, where incomplete relaxation at the cold boundary shrinks the effective energy difference transported between the two sides [2509.14027].

The quantum treatment of the same model, however, exhibits no NDTR. With baths represented by boundary-localized Lindblad dissipators and Bose occupation factors
\[
n_\alpha(\omega)=\left[\exp\!\left(\frac{\hbar\omega}{k_B T_\alpha}\right)-1\right]^{-1},
\]
the steady-state heat current increases monotonically with thermal bias [2509.14027]. The paper attributes the divergence to the fact that the classical suppression mechanism relies on a vanishing relaxation rate or effective freezing at the cold boundary, whereas the Lindblad dynamics describes finite-rate Markovian transitions and continuous exchange with the reservoirs [2509.14027].

A second low-temperature quantum ingredient is the discrete spectrum of the square well,
\[
\epsilon_n=\frac{n^2\pi^2\hbar^2}{2mL^2}.
\]
When $k_B T \ll \epsilon_2-\epsilon_1$, transport is strongly suppressed by the spectral gap, but the suppression remains monotonic rather than turning over into NDTR [2509.14027]. This establishes a concrete classical–quantum split: classically, low-temperature slowing near the cold bath can generate NDTR; quantum mechanically, low-temperature spectral suppression reduces transport without producing a negative differential branch. A plausible implication is that device designs extrapolated from classical low-temperature models can fail qualitatively once the relevant transport channel becomes quantum.

## 6. Device relevance, design principles, and unresolved issues

NDTR is repeatedly identified as a foundational nonlinear ingredient for thermal transistors, thermal switches, thermal logic gates, thermal amplifiers, thermal memories, and heat-current regulators [2509.14027]. In photonic circuits, the existence of an NDTC window around the superconductor-to-resistive transition enables transistor-like behavior: increasing the input power can produce a reduction in the drain temperature over a finite power interval [2112.12627]. In macroscopic heterojunctions, analytic expressions for $J$, $k$, and $r$ provide direct design rules in terms of bulk conductance scale $A$ and interface parameters $C,\alpha$, with the decisive criterion $\alpha>1$ for the temperature dependence of interfacial thermal resistance [2302.14065]. In fluidic settings, the robustness of bath-induced NDTR in weakly interacting, mixed, and gravity-modulated MPC fluids points toward fluidic thermal transistors in micro- and nanosystems [2211.06872].

The literature also defines several boundary conditions on what NDTR is not. It is not restricted to momentum-nonconserving lattices with on-site substrate potentials, because it has been reported in momentum-conserving FPU systems [1110.4942]. It is not limited to specially joined two-segment structures, because graded anharmonic crystals and Heisenberg chains with spatially varying magnetic fields can exhibit it without a specially engineered two-segment junction [1101.4589]. It is likewise not confined to phononic conduction, since photonic heat circuits realize the same differential anomaly through impedance-controlled electromagnetic transport [2112.12627].

At the same time, the literature does not support a universal recipe. Some models exhibit NDTR only in finite-size or low-dimensional windows, as in graphene nanoribbons and deformable FK lattices [1407.7087]. Some require sufficiently strong nonlinearity, as in the weakly coupled nonlinear-lattice analysis and the double-NDTR homogeneous lattices [1001.3782]. Some require explicitly temperature-dependent interface resistance, as in diffusive heterojunction theories [1201.3054]. And the low-temperature square-well comparison shows that even when a classical model predicts NDTR, the quantum version may not [2509.14027]. Taken together, these results suggest that NDTR is best understood not as a single effect with a single cause, but as a family of negative differential responses produced by distinct transport bottlenecks whose validity depends on transport regime, dimensionality, system size, interaction strength, boundary protocol, and, at low temperatures, quantization.

Source: https://www.emergentmind.com/topics/negative-differential-thermal-resistance-ndtr