Where heat flows and where to change it: gauge-free sensitivity maps of thermal conductivity in disordered solids
Listen
Radio episode about this paper
Transcript
Introduction to the show: ident: Condensed Matter. The day in condensed matter: materials, phases and the measurements under them.
Kai: I'm Kai.
Mira: And I'm Mira, here with Lev.
Kai: We're looking at the paper: "Where heat flows and where to change it".
Mira: Here's the gist of the paper. Where heat flows and where to change it: gauge-free sensitivity maps of thermal conductivity in disordered solids is a study developing two local descriptions of thermal conductivity in amorphous silicon and testing them on amorphous silicon.
Kai: The exact gradient of the AF conductivity with respect to the dynamical matrix, contracted into mass and bond sensitivity maps, decomposes the mass map exactly into three terms, one of which is the SPTC itself.
Mira: Gauge Invariance and Map Robustness The two local descriptions of thermal conductivity are anti-correlated in amorphous and paracrystalline silicon.
Kai: The explanation for this anti-correlation is local softness, where soft regions, which carry little heat, act through the mixing of vibrational modes about 1–2 meV apart.
Title and authors: Kai: Let's talk about the actual title and who wrote this paper, "Where heat flows and where to change it: gauge-free sensitivity maps of thermal conductivity in disordered solids." It’s a very descriptive title that tells you immediately what the paper is about.
Mira: It’s not just measuring conductivity; they are creating these maps to show where heat flows and where the material is most sensitive to changes, like adding a substituent or changing a bond. That moves it from just knowing a number to knowing *where* the transport bottleneck is.
Lev: The authors are Drabold and Ugwumadu at Los Alamos National Laboratory, working on theoretical physics, but they are clearly focused on applying this to real material science problems. I wonder if their approach is robust enough for something like designing new electronic materials.
Kai: They introduce two local descriptions of thermal conductivity in amorphous silicon and then test them directly against each other using the exact gradient of the thermal conductivity with respect to the dynamical matrix, which they contract into mass and bond sensitivity maps.
Mira: The real strength here is that they show this decomposition works exactly. They find that one term in this decomposition is actually the specific thermal transport coefficient, or SPTC, which confirms their framework is mathematically sound.
Lev: So it’s not just a guess about where heat goes; they have a precise mathematical tool to decompose the response into known physical components, like the SPTC itself and some sensitivity terms.
Kai: It sounds like they are trying to untangle the different physical effects that contribute to how heat moves in these complex materials. That’s what I find really compelling when you’re looking at disordered systems.
Mira: And they introduce this concept of local softness as a mediator between the two maps, which is key because it explains why the heat-flow map and the mass sensitivity map are anti-correlated in amorphous silicon.
The paper's summary: Kai: So summarizing what they found, this paper shows that local softness is what connects the heat flow map to the mass sensitivity map. They suggest that soft regions, which carry little heat, are actually where the material would respond most strongly to changes in its structure or composition.
Mira: It means if you want to change thermal transport significantly in amorphous silicon, you should look at those soft regions because they are both where heat is low and where it can be changed most effectively. That’s a very useful insight for materials design.
Lev: I like the idea of soft regions acting as the mediator; it implies that the way vibrational modes mix around one to two meV apart is what controls this sensitivity, not just some random disorder <ref:2611.06958#pg1>.
Kai: And they show that the site projection, which tells you where heat flows on a scale of a few angstroms, isn't quite right for individual atoms because it doesn't capture this softness. They introduce a gauge-invariant generalization called kappa q, which goes to zero as the wave vector q goes to zero.
Mira: That gauge invariance is important because it means the physical result isn't dependent on how you mathematically describe the system locally, which makes the physics more reliable across different theoretical setups.
Lev: From a computational standpoint, if this works, it suggests that we can use these sensitivity maps to screen large models in simulations to find where the critical transport channels or rattlers are located.
Kai: So they’ve moved beyond just saying "heat flows here" to saying "heat flows here, and this is the best place to make a change if you want a big effect." That’s a big step for material science.
The paper's improvements: Mira: The authors suggest that this sensitivity map approach can guide material design by telling you where to place dispersed substituents or where the transport is limited across an amorphous/crystalline interface. It moves the research from just predicting static properties to suggesting functional designs.
Lev: That application is what I’m most excited about for hardware, because if we know exactly which atom or bond to perturb, we can guide synthesis toward a better electronic or thermal conductor.
Kai: The exact tests they run show that isotope changes and modest bond changes have nearly linear effects over regions spanning up to forty-six atoms, which is about six angstrom radius. For heavier substituents, the map becomes a useful ranking tool when those substituents are dispersed across the material.
Mira: But for compact clusters, it seems like only the number of substituted atoms really matters in that context. It gives you a different kind of guidance based on the geometry of what you’re trying to modify.
Lev: It’s also useful for identifying transport along an interface or a filament because the sensitivity map pinpoints exactly where local perturbations will have the biggest impact on cross-plane conductivity.
Kai: So they show that even with complex disorder, you can use this exact response as a first-order tool to locate bottlenecks and plan your material modifications. It’s about being prescriptive rather than just descriptive.
Conclusion: Mira: So to wrap up, the paper "Where heat flows and where to change it: gauge-free sensitivity maps of thermal conductivity in disordered solids" shows that local softness mediates the anti-correlation between where heat flows and how sensitive the material is to changes.
Kai: They prove that using a gauge-invariant approach allows us to decompose this response into three exact terms, specifically showing how the rotation term is linked to those pairs of vibrational modes about one or two meV apart.
Lev: From an engineering standpoint, what they really demonstrate is that the site projection tells you where heat flows, but the sensitivity map tells you what decision you need to make about changing something in that spot.
Mira: I think the main implication for us is shifting our focus from just asking where something is to asking how it responds when we alter it. That’s a constructive answer to the gauge question that has shadowed local descriptions of heat transport.
Kai: So, if you’re working on designing thermal management systems or new disordered materials, use both tools: use the site projection first to see where heat flows, and then use the sensitivity map when you need to decide what specific change will have the most effect.
Lev: I just want to add that for real hardware implementation, we need to be careful about how many atoms we’re considering; their results hold up over regions of up to forty-six atoms with a six angstrom radius.
Mira: That's right, and it sets a clear direction for future work in this area. We'll be looking at how these maps apply to more complex, larger systems next.
Ohio University · Los Alamos National Laboratory
cond-mat.mtrl-sci
Submitted: 2026-11-03
Updated: 2026-11-03
Comments: 16 pages, 9 figures, 3 tables
License: http://creativecommons.org/licenses/by/4.0/
Importance score: 87/100
The gist: Where heat flows and where to change it: gauge-free sensitivity maps of thermal conductivity in disordered solids is a study developing two local descriptions of thermal conductivity in amorphous…
Key concepts
- Gauge-free Sensitivity Maps
- These maps are derived from the exact gradient of thermal conductivity with respect to the dynamical matrix. They provide a rigorous way to map out how local changes in atomic properties affect heat flow, independent of arbitrary mathematical choices.
- Local Softness
- Soft regions in a material carry little heat. This is because these regions allow vibrational modes separated by 1–2 meV to mix easily. This mixing explains why soft areas are anti-correlated with the heat-flow map, as they don't efficiently transport energy.
- Sensitivity Map Decomposition
- The mass sensitivity map ($\kappa_m$) is exactly decomposed into three terms: a term related to the explicit gradient ($\zeta$), an eigenvalue-shift term (seig), and a rotation term (srot). This decomposition helps identify which physical mechanism drives the relationship between heat flow and atomic mass changes.
Terminology
Summary
Where heat flows and where to change it: gauge-free sensitivity maps of thermal conductivity in disordered solids is a study developing two local descriptions of thermal conductivity in amorphous silicon and testing them on amorphous silicon. p. 0 Where heat flows and where to change it: gauge-free sensitivity maps of thermal
conductivity in disordered solids
D. Drabold1, ∗ and C. Since the AF expression for κ
involves no division of the energy among atoms, neither
do its derivatives (Sec. We derive the exact gradient of the AF conductivity with respect to the dynamical matrix, contract it into mass and bond sensitivity
maps, and decompose the mass map exactly into three
terms, one of which is the SPTC itself.Where the paper says it
Where the paper says it
Gauge Invariance and Map Robustness
The two local descriptions of thermal conductivity are anti-correlated in amorphous and paracrystalline silicon. The explanation for this anti-correlation is local softness, where soft regions, which carry little heat, act through the mixing of vibrational modes about 1–2 meV apart. p. 0 The harmonic vibrational amplitude of each atom predicts both maps, and at room
temperature the anti-correlation largely disappears at fixed amplitude, in a 32 768-atom replica as
well; below about 100 K a component remains that softness does not capture. Thus soft regions,
which carry little heat, act through the mixing of vibrational modes about 1–2 meV apart. 995 at 10 ˚A), and for a map of
where heat flows that is the claim that matters. Thus
the site projection is a well-defined description of where
heat flows at the scale of a few ˚A, but not of individual
atoms (Figs. Thus
the site projection is a well-defined description of where
heat flows at the scale of a few ˚A, but not of individual
atoms (Figs. The gauge-invariant generalization of a local conductivity is the wave-vector-dependent
κ(q), whose gauge dependence vanishes as q → 0 [21].Where the paper says it
Where the paper says it
Where the paper says it
Sensitivity Map Decomposition
The exact three-term decomposition of the mass sensitivity map is given by κm = 4ζ + 2seig + 2srot. The first term arises because the explicit part of the gradient contracted with D is exactly twice the AF gauge ζ. p. 4 (7)
gives an exact three-term decomposition,
κm(ℓ) = 4ζ(ℓ) + 2seig(ℓ) + 2srot(ℓ), (9)
seig(ℓ) = X
i
∂κ
∂λi
λipi(ℓ), srot(ℓ) = X
i<k
Nik
λi + λk
λi − λk
cik(ℓ),
(10)
where pi(ℓ) = P
a∈ℓ E2
ai is the weight of mode i on atom
ℓ and cik(ℓ) = P
a∈ℓ EaiEak the local overlap of two
modes. The first term arises because the explicit part of
the gradient contracted with D is exactly twice the AFgauge ζ. The eigenvalue-shift term seig is a sensitivityweighted local density of states. The rotation term srot
sums to zero over sites, because P
ℓ
cik(ℓ) = δik: it only
redistributes sensitivity in space. The eigenvalue-shift term seig is a sensitivityweighted local density of states. The rotation term srot
sums to zero over sites, because P
ℓ
cik(ℓ) = δik: it only
redistributes sensitivity in space.Where the paper says it
Where the paper says it
Where the paper says it
Local Softness as the Mediator
Local softness mediates the anti-correlation between the heat-flow map and the mass sensitivity map. The heat-flow map is strongly anti-correlated with the local vibrational amplitude δR. p. 7 Mechanism: local softness and mode mixing
Local softness mediates the anti-correlation. The
heat-flow map is strongly anti-correlated with the local
vibrational amplitude δR [Table II, Fig. Darker colours and thicker bonds indicate larger field values. In the soft,
large-amplitude regions, which carry little heat, κm is positive and large: a heavier atom there lowers κ. above 45 meV contribute nothing. The window is set
mainly by the network, not by the broadening: at
η = 2 K (0.Where the paper says it
Where the paper says it
Where the paper says it
Application to Material Design
The sensitivity map is a first-order response, and it can guide the placement of dispersed substituents and locate the region that limits transport across an amorphous/crystalline interface. The exact tests show that isotopes and modest bond changes are nearly linear over regions of up to 46 atoms (6 ˚A radius). For heavy substituents, the map is a useful ranking tool when the substituents are dispersed, while for compact clusters only the number of substituted atoms matters. p. 0 In
paracrystalline silicon the grains conduct more heat per atom, while the disordered matrix controls
the response. Exact recalculations show that the response map guides the placement of dispersed
substituents and locates the region that limits transport across an amorphous/crystalline interface.Where the paper says it
Interface Transport and Conclusions
The sensitivity map identifies the region where local perturbations most affect cross-plane conductivity. The site projection is the tool for asking where heat flows, screening large models, finding channels and rattlers, and transport along an interface or a filament. p. 1 A single
local quantity, the harmonic vibrational amplitude δR(ℓ)
of each atom, mediates the relation, and at fixed δR the
anti-correlation between the two maps largely vanishes. Soft, large-amplitude sites (“rattlers”) contribute little
to transport, but a change there mixes the vibrational
modes that do. A single
local quantity, the harmonic vibrational amplitude δR(ℓ)
of each atom, mediates the relation, and at fixed δR the
anti-correlation between the two maps largely vanishes. Soft, large-amplitude sites (“rattlers”) contribute little
to transport, but a change there mixes the vibrational
modes that do. Thus “where the response is large” and “where heat flows
little” coincide only on average; what the heat flow sets
reliably is the sign of the response. The picture is plain
in three dimensions, where the two maps occupy largely
different parts of the network: the 12% of the volume
where the smoothed ζ is largest and the 12% where κm
is largest share 0. 995 at 10 ˚A), and for a map of
where heat flows that is the claim that matters. Thus
the site projection is a well-defined description of where
heat flows at the scale of a few ˚A, but not of individual
atoms (Figs. Each
delivers a single number, the conductivity κ. Machinelearned interatomic potentials now give force constants
of near first-principles quality for 104–106 atoms [9], so
local descriptions of heat transport are within reach, and
they are what design needs: where in a material does
heat flow, and where would a change matter?Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Where the paper says it
Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.
Improvements for AI systems
-
Incorporate gauge-free sensitivity maps to determine where heat flows and where changes would matter in disordered solids. This allows AI systems to identify material regions based on transport properties rather than arbitrary energy partitions.
-
Develop an AI capability to predict the anti-correlation between the heat flow map and the mass sensitivity map using local softness as a mediator. This enables models to understand why
soft regions, which carry little heat, act through the mixing of vibrational modes about 1–2 meV apart.
-
Implement a mechanism for designing materials by using sensitivity maps to guide the placement of dispersed substituents and locate regions that limit transport across amorphous/crystalline interfaces. This moves AI from predicting static properties to suggesting functional material designs.
-
Create an AI system capable of distinguishing between different types of local modes—diffusons, propagons, and locons—by analyzing the energy band where they reside (e g 10–30 meV). This allows for a more accurate description of transport regimes in different frequency domains.
-
Use the exact decomposition formula κm = 4ζ + 2seig + 2srot to understand which term carries the anti-correlation between heat flow and mass sensitivity, specifically identifying that the rotation term is mediated by pairs of modes about 1–2 meV apart.
Abstract
Designing materials for thermal management requires knowing where heat flows, not only how much. Within harmonic (Allen -- Feldman) theory we develop two local descriptions and test them on amorphous silicon. The first, a projection of the conductivity κ onto atoms, depends on how the potential energy is divided among atoms (a "gauge"); we find the dependence moderate, 23% of the map's spatial standard deviation at single sites and 9.5% over regions of radius 0.4 nm, apart from a uniform offset of 0.8% of κ. The second avoids the division altogether. It is the exact response of κ to the mass of each atom and the stiffness of each bond, it obeys an Euler sum rule, and it contains the heat-flow map as one of three exact terms. In amorphous and paracrystalline silicon, with ACE, GAP-18 and Keating force constants, the two maps are anti-correlated: where little heat flows a heavier atom lowers κ, and where much heat flows it tends to raise it. The explanation is local softness. The harmonic vibrational amplitude of each atom predicts both maps, and at room temperature the anti-correlation largely disappears at fixed amplitude, in a 32,768-atom replica as well; below about 100 K a component remains that softness does not capture. Thus soft regions, which carry little heat, act through the mixing of vibrational modes about 1 -- 2 meV apart. In paracrystalline silicon the grains conduct more heat per atom, while the disordered matrix controls the response. Exact recalculations show that the response map guides the placement of dispersed substituents and locates the region that limits transport across an amorphous/crystalline interface.
Sources
- Site-projected Thermal Conductivity: Application to defects, interfaces, and homogeneously disordered materials
- Spatially Local Estimates of the Thermal Conductivity of Materials
- Atomic cluster expansion potential for the Si-H system
- A Defect-Free Model of Amorphous Silicon with Pristine Electronic Structure
Related papers
- Structural contribution to light-induced gap suppression in Ta 2 NiSe 5
- Anatomy of fast current-induced skyrmion motion in synthetic antiferromagnets
- Magnetic control of an exciton-polariton condensate in a van der Waals magnet
- High-pressure elastic properties of GeO2 polymorphs up to 120 GPa
- Decoupling Strain-Rate Sensitivity and Deformation Length Scale Effects in Neutron-Irradiated Tungsten: A Coupled Nano-Indentation, HR-EBSD and Crystal Plasticity Study
- Magnetism modulation in Co3Sn2S2 by current-assisted domain wall motion