Diffusion in a d-dimensional rough potential

arXiv:2508.04674 · cond-mat.mtrl-sci, cond-mat.stat-mech · Submitted 2025-08-06 · Read on arXiv

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: We're looking at the paper: "Diffusion in a d-dimensional rough potential".

Mira: Here's the gist of the paper. The prediction of diffusion in solids is necessary to understand the microstructure evolution in materials out of equilibrium This work develops an analytical model for diffusivity in a noisy solid solution using mean first passage time analysis The analytical results show that roughness always decreases the diffusivity, aligning with sluggish diffusion theories in concentrated alloys Fundamental Modeling of Transport Accurately modeling diffusion in solids is paramount to fundamentally understand the microstructure evolution in materials out of equilibrium In a homogeneous pure solid, transport coefficients are often modeled using harmonic transition state theory (HTST) The hopping rate over an energy barrier is estimated by HTST as R = ν exp (−β∆E).

Kai: For multi-component solid solutions, the activation energy for a diffusive hop depends on local chemistry, varying across the system.

Mira: Dimensionality Dependence and Analytical Expression The authors develop an analytical expression for diffusivity in d dimensions on a lattice with normally distributed energy barriers consistent with Zwanzig’s expression for d ≤ 5 This expression is valid for d = 2, but not d > 2.

Kai: The model assumes that the saddle point and initial energies can be modeled as independently distributed random variables with normal distributions The hopping rates are then modeled by a lognormal distribution L(−βµ, 2β2σ2).

Paper summary: Kai: To recap what we just heard, we talked about the paper Diffusion in a d-dimensional rough potential and how they use mean first passage time analysis to model noise effects on diffusivity.

Mira: They claim that roughness always decreases the diffusivity, which is consistent with sluggish diffusion theories in concentrated alloys. The authors develop an analytical expression for this diffusivity in an arbitrary number of dimensions d using mean first passage time analysis.

Lev: The core of their method relies on modeling the saddle and initial energies as independently distributed normal variables, which then leads to a lognormal distribution for the hopping rates.

Kai: They then use the harmonic mean of these rates to find a characteristic rate, reff, which is essentially nu exp minus mu with two times sigma squared.

Mira: And they show that this work provides an analytical expression for diffusivity in d dimensions, but they explicitly state that this expression is only valid for d equals two and not for d greater than two <ref:2508.04674#pg1>.

Conclusion: Kai: So, to wrap up the discussion on Diffusion in a d-dimensional rough potential, the authors developed a mean first passage time analysis to get an analytical model for diffusivity in noisy solid solutions.

Mira: The main implication we see is that roughness always decreases the diffusivity under these conditions, which aligns with sluggish diffusion theories. They also found that increasing dimensionality enhances the diffusivity due to extra percolation pathways.

Lev: From a practical standpoint, their work gives researchers a way to get faster estimates for diffusivities compared to running expensive atomistic simulations like kinetic Monte Carlo when you are dealing with systems in two dimensions.

Kai: And the paper's limitation is that this analytical expression specifically fails for three or more dimensions, so you have to be careful about applying their formula outside of two dimensions.

Jacob Jeffries, Emilio Mendoza Reyes, Fadi Abdeljawad, Murray Daw, Enrique Martinez

Department of Materials Science and Engineering, Clemson University · School of Computing, Clemson University · Department of Materials Science and Engineering, Lehigh University · Department of Physics and Astronomy, Clemson University · Department of Mechanical Engineering, Clemson University

cond-mat.mtrl-sci, cond-mat.stat-mech

Submitted: 2025-08-06

Updated: 2026-06-17

Comments: 12 pages, 3 figures

DOI: 10.1073/pnas.2601219123

License: http://creativecommons.org/licenses/by-nc-nd/4.0/

Importance score: 81/100

The gist: The prediction of diffusion in solids is necessary to understand the microstructure evolution in materials out of equilibrium This work develops an analytical model for diffusivity in a noisy solid…

Key concepts

Fundamental Modeling of Transport
Accurately modeling diffusion is crucial for understanding microstructure changes in materials out of equilibrium, such as radiation-induced segregation or sintering. In pure solids, transport is often modeled using harmonic transition state theory (HTST), estimating the hopping rate based on an energy barrier.
Dimensionality Dependence and Analytical Expression
The authors create an analytical expression for diffusivity in $d$ dimensions on a lattice with normally distributed energy barriers, consistent with Zwanzig's work for $d ext{ } ext{ } ext{ } ext{ } ext{ } heta ext{ } d heta = 2$. This model is valid for $d=2$ but not for $d>2$, assuming saddle point and initial energies are independent normal random variables.
Mean First Passage Time Analysis
In the limit of infinite dimensions ($N ightarrow ext{ } ext{ } ext{ } ext{ } heta$), the characteristic rate along a path is the harmonic mean of hopping rates. This leads to an effective barrier ($eta_{ ext{eff}}$) and an effective hopping rate ($ ilde{ u}$) that describe the overall transport behavior.

Terminology

Summary

The prediction of diffusion in solids is necessary to understand the microstructure evolution in materials out of equilibrium This work develops an analytical model for diffusivity in a noisy solid solution using mean first passage time analysis The analytical results show that roughness always decreases the diffusivity, aligning with sluggish diffusion theories in concentrated alloys

Fundamental Modeling of Transport

Accurately modeling diffusion in solids is paramount to fundamentally understand the microstructure evolution in materials out of equilibrium In a homogeneous pure solid, transport coefficients are often modeled using harmonic transition state theory (HTST) The hopping rate over an energy barrier is estimated by HTST as R = ν exp (−β∆E). For multi-component solid solutions, the activation energy for a diffusive hop depends on local chemistry, varying across the system. p. 0

Where the paper says it

In a homogeneous pure solid, transport coefficients, such as the diffusivity D, are often modeled using harmonic transition state theory (HTST)[10] : D = D0 exp ( − ∆E kBT ) (1) where D0 is the rate prefactor, which can be estimated from the vibrational modes at the initial and transition states[11][12] and a hopping length, kB is the Boltzmann constant, T the absolute temperature, and ∆E the activation energy of diffusion, which can be estimated using the nudged elastic band (NEB) method[13][14][15][16][17] coupled with atomistic simulation softwares[18][19][20][21][22][23] . However, for multi-component solid solutions, the activation energy for a diffusive hop depends on local chemistry, varying across the system.

Dimensionality Dependence and Analytical Expression

The authors develop an analytical expression for diffusivity in d dimensions on a lattice with normally distributed energy barriers consistent with Zwanzig’s expression for d ≤ 5 This expression is valid for d = 2, but not d > 2. The model assumes that the saddle point and initial energies can be modeled as independently distributed random variables with normal distributions The hopping rates are then modeled by a lognormal distribution L(−βµ, 2β2σ2). p. 2

Where the paper says it

This reduction to a single barrier is an intentional coarse-graining of the full reactioncoordinate landscape, whose detailed shape can itself influence mean first-passage times when intermediate barriers or structured potential profiles are present[42] . We then assume that E† and E◦ can be modeled as independently distributed random variables: E ◦ ∼ N ( µ ◦ , σ2 ) E † ∼ N ( µ † , σ2 ) (5) where N (µ, σ2 ) denotes a normal distribution with mean µ and variance σ 2 .

Mean First Passage Time Analysis

The characteristic rate along the path in the N → ∞ limit is exactly the harmonic mean of λ This characteristic rate is reff = ⟨λ −1⟩−1 = ν exp(−βµ − β2σ2). The effective barrier for one jump is described by βµeff = βµ + β2σ2, yielding an effective hopping rate λ′ ∼ νL(−βµeff, 2β2σ2). p. 2

Where the paper says it

that the diffusivity in a noisy potential in one dimension is D ∝ exp(−βµ − β 2σ 2 ) [1] . Using this characteristic rate, we describe diffusion paths in the noisy potential as single hops with an effective barrier µeff: βµeff = βµ + β 2σ 2 (10) which yields an effective hopping rate λ ′ for one jump: λ ′ ∼ νL(−βµeff, 2β 2σ 2 ) = ν exp ( −β 2σ 2 ) L(−βµ, 2β 2σ 2 ) (11) From our coarse-grained description, we return to the d-dimensional problem.

Validation and Conclusions

The results are compared with kinetic Monte Carlo simulations, which show good quantitative agreement for d ≤ 5 and excellent quantitative agreement for d ≤ 3 The diffusivity scales inversely with the first reaction time, D/D∗ = exp (−2β2σ2n + O(β4σ4)). The work concludes that the noise always reduces the diffusivity for a given mean activation energy. p. 7

Where the paper says it

We then performed kinetic Monte Carlo simulations to calculate the diffusivity as a function of the solid solution noise, which are in excellent agreement with the analytical expression for d ≤ 3. We observe that the noise always decreases the diffusivity for a given mean activation energy, i.

Furthermore, an increase in dimensionality increases the diffusivity, relating to extra percolation pathways that enhance walker migration

Pruning redundant reasoning tokens

The prediction of diffusion in solids is necessary to understand the microstructure evolution in materials out of equilibrium The calculation of transport coefficients in solids is necessary for modeling radiation-induced segregation, grain boundary diffusion, solid-state sintering, pitting corrosion, and more.

How the training signal is built

The hopping rate over an energy barrier ∆E is estimated by HTST: R = ν exp (−β∆E). In a homogeneous single-component solid, there is a singular lattice point energy E◦ and a single saddle point energy E†, yielding an energy barrier ∆E = E† − E◦. p. 1

Where the paper says it

The hopping rate over an energy barrier ∆E is estimated by HTST: R = ν exp (−β∆E) (4) where ν is the rate prefactor and β = 1/kBT. In a homogeneous single-component solid, there is a singular lattice point energy E◦ and a single saddle point energy E† , yielding an energy barrier ∆E = E† − E◦ .

p. 1
Where the paper says it

The hopping rate over an energy barrier ∆E is estimated by HTST: R = ν exp (−β∆E) (4) where ν is the rate prefactor and β = 1/kBT. In a homogeneous single-component solid, there is a singular lattice point energy E◦ and a single saddle point energy E† , yielding an energy barrier ∆E = E† − E◦ .

Where the gains come from

The hopping rates in the solution are then: λ = ν exp (−β∆E) ∼ νL(−βµ, 2β2σ2). The walker sees n transitions, all with transition rates drawn from the distribution of λ. p. 2

Where the paper says it

The hopping rates in the solution are then: λ = ν exp (−β∆E) ∼ νL ( −βµ, 2β 2σ 2 ) (7) where L denotes a lognormal distribution, and ν is implicitly assumed to be constant throughout the solution, and independent of temperature. In a single lattice site in a solid solution, a given walker will see n transitions, all with transition rates drawn from the distribution of λ.

p. 2
Where the paper says it

The hopping rates in the solution are then: λ = ν exp (−β∆E) ∼ νL ( −βµ, 2β 2σ 2 ) (7) where L denotes a lognormal distribution, and ν is implicitly assumed to be constant throughout the solution, and independent of temperature. In a single lattice site in a solid solution, a given walker will see n transitions, all with transition rates drawn from the distribution of λ.

The hopping rate over an energy barrier ∆E is estimated by HTST: R = ν exp (−β∆E). In a homogeneous single-component solid, there is a singular lattice point energy E◦ and a single saddle point energy E†, yielding an energy barrier ∆E = E† − E◦.

The hopping rates in the solution are then: λ = ν exp (−β∆E) ∼ νL(−βµ, 2β2σ2). The walker sees n transitions, all with transition rates drawn from the distribution of λ.

The characteristic rate along the path in the N → ∞ limit is exactly the harmonic mean of λ This characteristic rate is reff = ⟨λ −1⟩−1 = ν exp(−βµ − β2σ2).

The characteristic rate along the path in the N → ∞ limit is exactly the harmonic mean of λ This characteristic rate is reff = ⟨λ −1⟩−1 = ν exp(−βµ − β2σ2).

The effective barrier for one jump is described by βµeff = βµ + β2σ2, yielding an effective hopping rate λ′ ∼ νL(−βµeff, 2β2σ2).

The escape time from this coarse-grained point is determined by the standard KMC total-rate rule We can then find T′ via the law of total expectation: T′ = Eλ1,···,λn [T′ λ1, · · ·, λn] = exp(β2σ2) (∑n i=1 λi)−1.

The escape time from this coarse-grained point is determined by the standard KMC total-rate rule We can then find T′ via the law of total expectation: T′ = Eλ1,···,λn [T′ λ1, · · ·, λn] = exp(β2σ2) (∑n i=1 λi)−1.

We can then find T′ via the law of total expectation: T′ = Eλ1,···,λn [T′ λ1, · · ·, λn] = exp(β2σ2) (∑n i=1 λi)−1.

We are then concerned with computing the mean first passage time E[T′] From this expectation, we can then evaluate the ratio of the MFPT and the noiseless MFPT: ϕ = E[T′] / limβ2σ2→0 E[T′] = nν exp (−βµ + β2σ2) E[T′].

We are then concerned with computing the mean first passage time E[T′] From this expectation, we can then evaluate the ratio of the MFPT and the noiseless MFPT: ϕ = E[T′] / limβ2σ2→0 E[T′] = nν exp (−βµ + β2σ2) E[T′].

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

  1. Improved analytical models for diffusion in solids can predict diffusivity for arbitrary dimensions d using a mean first passage time analysis, which is useful for understanding microstructure evolution in materials out of equilibrium.

  2. The improved AI system can accurately model transport coefficients in solids, such as radiation-induced segregation and grain boundary diffusion, by using an analytical expression derived from the mean first passage time analysis that is consistent with Zwanzig's result for d less than or equal to 5.

  3. The system can perform faster estimates of diffusivities compared to computationally expensive atomistic methods like kinetic Monte Carlo simulations, especially for dimensions d less than or equal to 3 where excellent quantitative agreement is achieved.

Abstract

The prediction of diffusion in solids is necessary to understand the microstructure evolution in materials out of equilibrium. Although one can reasonably predict diffusive transport coefficients using atomistic methods, these approaches can be very computationally expensive. In this work, we develop an analytical model for the diffusivity in a noisy solid solution in an arbitrary number of dimensions (d) using a mean first passage time analysis. We observe that roughness always decreases the diffusivity, aligning with sluggish diffusion theories in concentrated alloys, finding that an increase in diffusivity induced by alloying elements must be due to a decrease in the average activation energy, not to the noise. These analytical results are then compared with kinetic Monte Carlo simulations, which are in good quantitative agreement with the simulation data for d at most 5, and excellent quantitative agreement for d at most 3. This generalization to arbitrary dimensions has been elusive to the community since Zwanzig [PNAS, 85, 2029 (1988)] published his seminal work on 1-dimensional systems.

Sources

Related papers