🔬 materials science

The viscoelastic rheology of transient diffusion creep

This paper presents a simple finite element model of transient diffusion creep in polycrystalline materials, demonstrating that its linear viscoelastic behavior can be effectively described by an extended Burgers model where the high-frequency Andrade exponent is determined by grain junction geometry, though the model provides only a lower bound for attenuation observed in laboratory experiments due to unaccounted dissipative processes.

Original authors: John F. Rudge

Published 2026-01-26

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: John F. Rudge

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine a block of rock not as a solid, unyielding brick, but as a giant, three-dimensional puzzle made of millions of tiny, interlocking grains. When you push or pull on this rock, it doesn't just snap or flow like water; it behaves like a strange, stretchy memory foam that remembers how hard you pushed it, but also forgets over time. This paper by John F. Rudge is a recipe for understanding exactly how that "memory foam" works when the rock is made of many tiny crystals.

Here is the story of the paper, broken down into simple concepts:

1. The Puzzle of the Rock Grains

Think of a rock as a crowd of people (the grains) standing shoulder-to-shoulder. If you try to push the whole crowd, the people in the middle can't move easily because they are stuck. But the people on the edges (the grain boundaries) can slide past each other.

In this model, the "people" inside the crowd are stiff and springy (elastic). However, the "edges" between them are special. They allow tiny particles (atoms and empty spaces called vacancies) to walk along the lines where the grains meet. This walking is called diffusion.

2. The Two Ways the Rock Moves

The paper looks at how the rock reacts when you shake it at different speeds (frequencies).

The Slow Shake (Low Frequency): Imagine pushing the crowd very slowly. The particles on the edges have plenty of time to walk around, find a spot, and let the grains slide past each other. The rock flows like thick honey. This is called steady-state creep.
The Fast Shake (High Frequency): Now, imagine shaking the crowd very fast. The particles on the edges don't have time to walk far. They get stuck near the corners where three grains meet (called triple junctions). The rock acts more like a stiff spring, but it still wiggles a little bit.

3. The "Traffic Jam" at the Corners

The most interesting part of the paper happens at the corners where three grains meet.

In a perfect, slow world, the stress (pressure) is spread out evenly.
In a fast world, the stress piles up at these corners like a traffic jam. The paper calculates exactly how bad this "jam" gets based on the angles of the corners.
The Analogy: Think of a triple junction like a three-way intersection. If cars (stress) try to turn quickly, they get bunched up. The paper found that the shape of this bunched-up traffic follows a specific mathematical rule (a power law) that depends only on the angle of the intersection.

4. The "Goldilocks" Model

The author built a computer simulation (using shapes like hexagons in 2D and a 14-sided shape called a tetrakaidecahedron in 3D) to see how this works. He then tried to describe the results using simple mathematical "models" that scientists use to describe squishy materials.

He found that the rock's behavior is best described by a hybrid model called the Extended Burgers Model.

The Maxwell Part: This describes the slow, honey-like flow.
The Andrade Part: This describes the fast, wiggly behavior. It's named after a scientist who noticed that materials don't just snap back instantly; they have a "creep" that follows a specific curve.

The paper shows that the rock behaves like a Maxwell fluid when you are slow, and like an Andrade solid when you are fast. The transition between the two is smooth and predictable.

5. Comparing to the Real World

The author took his computer model and compared it to real experiments done in labs with rocks and rock-like materials (like borneol, a waxy substance).

The Good News: The model matches the lab experiments surprisingly well for certain materials. It predicts that the "wiggly" behavior (attenuation) is about one-third of a power law.
The Bad News: The model predicts less energy loss (damping) than what is seen in some real, hot rocks deep inside the Earth.
The Conclusion: The model is a "lower bound." It tells us the minimum amount of squishiness we should expect from grain sliding. If real rocks are squishier than the model, it means there are other secret mechanisms at play—perhaps melting at the edges of the grains or impurities—that the simple model doesn't see yet.

Summary

In short, this paper builds a simple, clear map of how tiny grains in a rock slide past each other when atoms diffuse along their edges. It proves that the shape of the grains and the angles where they meet dictate exactly how the rock absorbs energy. While the model explains a lot, it also hints that the real Earth is even more complex and "squishy" than our simplest models can currently explain.

Technical Summary: The Viscoelastic Rheology of Transient Diffusion Creep

Problem Statement
Polycrystalline materials exhibit viscoelastic rheology where strain responses to stress are dependent on the deformation timescale. While energy is stored elastically within grain interiors, it is dissipated through various mechanisms, notably diffusionally-accommodated (or assisted) grain boundary sliding, also termed transient diffusion creep. A fundamental challenge in Earth sciences and material science is extrapolating laboratory results—constrained by small length and time scales—to planetary conditions. Current understanding relies heavily on laboratory experiments, but safe extrapolation requires models grounded in the fundamental physics of grain-scale processes. This paper addresses the need for a simple, grain-scale model of rock viscoelasticity to serve as a basis for interpreting and extrapolating experimental data.

Methodology
The study employs a homogenisation approach based on finite element (FE) calculations to determine effective upscaled viscoelastic tensors. The model geometry consists of infinite tilings of identical grains:

2D Geometry: A regular hexagonal tiling.
3D Geometry: A tetrakaidecahedral (truncated octahedron) tiling.

The physical assumptions include:

Grain Interiors: Deform linear elastically with intrinsic shear modulus $\mu$ and Poisson's ratio $\nu$ .
Grain Boundaries: Allow free sliding (zero shear stress).
Diffusion: Vacancies and atoms diffuse along grain boundaries driven by normal stress gradients, leading to material plating or removal.

The governing equations couple elasticity within grains with diffusion along boundaries. The linear viscoelastic response is characterized by the complex shear compliance $J^*(\omega)$ as a function of frequency $\omega$ . The FE models solve for the displacement field and normal stresses on boundaries under time-harmonic loading. Results are non-dimensionalized using the Maxwell time $\tau_M = \eta/G_0$ , where $\eta$ is the steady-state Coble creep viscosity and $G_0$ is the unrelaxed shear modulus.

Key Contributions and Results

Viscoelastic Characterization: The FE results for both hexagonal and tetrakaidecahedral grains are well-described by a parameterized extended Burgers model. This model behaves as a Maxwell model at low frequencies (steady-state diffusion creep) and as an Andrade model at high frequencies.
Andrade Exponent ( $\alpha$ ): The study establishes that the high-frequency power-law exponent $\alpha$ $α$ is determined solely by the geometry of triple junctions. It is analytically linked to the stress singularity exponent $\lambda$ $λ$ of purely elastic models (without diffusion) via $\alpha = \frac{2}{3}(1-\lambda)$ $α = \frac{2}{3} (1 - λ)$ .
- For 2D hexagons (120° junctions), $\lambda \approx 0.449$ , yielding $\alpha \approx 0.367$ .
- For 3D tetrakaidecahedrons (junctions of 125.26° and 109.47°), $\lambda = 0.5$ , yielding $\alpha = 1/3$ .
Relaxation Parameters: The models provide specific values for the relaxation strength $\Delta$ $Δ$ , the Andrade time $\tau_A$ $τ_{A}$ , and their ratios to the Maxwell time.
- For hexagons ( $\nu=0.3$ ): $\Delta \approx 0.33$ , $\tau_A/\tau_M \approx 1.3$ .
- For tetrakaidecahedrons ( $\nu=0.3$ ): $\Delta \approx 0.55$ , $\tau_A/\tau_M \approx 1.7$ .
Comparison with Previous Models: The results differ from previous bicrystal models (e.g., Lee et al. [7]) primarily due to geometry. The bicrystal models, which involve two sliding surfaces meeting at a corner, yield significantly different $\tau_A/\tau_M$ ratios and lower attenuation compared to the triple-junction models presented here.
Comparison with Laboratory Data:
- The model parameters show remarkable agreement with laboratory experiments on borneol (a rock analogue), particularly in the values of $\alpha$ and $\tau_A/\tau_M$ .
- However, the simple models provide a lower bound on the attenuation observed in synthetic rock aggregates (e.g., olivine, dunite, MgO) at high homologous temperatures.
- The models predict higher Andrade exponents ( $\sim 0.33-0.37$ ) than some laboratory fits to synthetic rocks (which suggest $\alpha$ as low as 0.2). The paper suggests this discrepancy may arise from additional dissipation mechanisms not captured by the simple diffusion model, such as premelting, impurities, or elastically-accommodated sliding.

Significance and Claims
The paper claims that the presented simple grain-scale models successfully describe the effective viscoelastic behavior arising from diffusionally-accommodated grain boundary sliding. The close agreement with borneol experiments suggests the model is robust for describing polycrystalline materials in regimes dominated by this mechanism.

Crucially, the study highlights that while the model captures the fundamental physics of transient diffusion creep, it does not fully account for the attenuation levels observed in high-temperature rock experiments or seismic observations of the Earth's asthenosphere. The paper posits that additional dissipative processes—potentially associated with premelting, grain boundary viscosity, or dislocations—are required to explain the higher attenuation and lower Andrade exponents seen in some experimental datasets. The work provides a rigorous baseline (a lower bound) against which these additional mechanisms can be quantified.