Effects of microstructural heterogeneity on the macroscopic spectrum of elastically accommodated grain-boundary sliding

This study demonstrates that while microstructural heterogeneity in grain geometry has a modest effect, a broad distribution of grain-boundary viscosities can suppress and broaden the characteristic Debye-like peak of elastically accommodated grain-boundary sliding into a weak background, thereby explaining the absence of a pronounced peak in dry olivine experiments without negating the mechanism's relevance to upper-mantle seismic attenuation.

The Gist

The Big Picture: The Missing "Hum" in the Earth

Imagine the Earth's upper mantle (the layer just below the crust) as a giant, slow-moving block of rock made of tiny mineral grains, mostly a mineral called olivine.

When seismic waves (earthquake energy) travel through this rock, they lose a little bit of energy and slow down. Scientists call this attenuation and dispersion.

For a long time, a popular theory suggested that this energy loss happens because the tiny grains slide slightly past one another, like a deck of cards shifting. This is called Elastically Accommodated Grain-Boundary Sliding (EAGBS).

The Problem:
According to the classic math for this theory, if you slide these grains, the rock should act like a radio tuned to a specific station: it should produce a sharp, loud "peak" of energy loss at one specific frequency.

• In metals and ice: Scientists see this sharp peak clearly.
• In dry olivine (the main rock of the upper mantle): Scientists look for this peak, but it's barely there. It's like a radio that's supposed to be loud but is whispering.

This paper asks: Why is the "peak" missing in dry rock? Is the sliding mechanism broken, or is the signal just hidden?

---

The Experiment: Building Digital Rock

The authors built a computer simulation of rock made of thousands of tiny, polygon-shaped grains. They tested two things to see if they could hide the "peak":

1. Changing the Shape and Size of the Grains: (Geometric Heterogeneity)
2. Changing how "sticky" or "slippery" the boundaries between grains are: (Viscosity Heterogeneity)

Finding #1: Irregular Shapes Don't Hide the Signal

First, they looked at the shapes. Real rocks have grains of all different sizes and weird shapes, unlike the perfect hexagons used in old theories.

• The Analogy: Imagine a crowd of people trying to shuffle through a doorway.
  • Old Theory: Everyone is the same height and walks in a perfect line.
  • New Test: People are different heights and walk in a jumbled mess.
• The Result: While the jumbled crowd moves slightly differently (the baseline changes), they still all shuffle at roughly the same speed. The "peak" in energy loss just shifts a little bit; it doesn't disappear or get fuzzy.
• Conclusion: Just having grains of different sizes cannot explain why the peak is missing in dry olivine.

Finding #2: Different "Stickiness" Hides the Signal

Next, they looked at the boundaries between the grains. In real olivine, the boundary between two grains can be very different depending on how the crystals are oriented. Some boundaries are very "sticky" (high viscosity), while others are very "slippery" (low viscosity).

• The Analogy: Imagine a relay race with 100 runners.
  • Scenario A (Uniform): All 100 runners are identical. They all run at the exact same speed. If you time them, you get one sharp, clear peak on the stopwatch.
  • Scenario B (Heterogeneous): Now, imagine the runners have vastly different speeds. Some are sprinters, some are joggers, and some are walking.
  • The Result: If you try to time the whole group, you don't get one sharp peak. Instead, you get a long, flat, messy line. The fast runners finish early, the slow ones finish late, and the "peak" is smeared out into a broad background.
• The Result: When the authors gave the grain boundaries a wide range of "stickiness," the sharp peak completely disappeared. It was smeared out into a weak, broad background.
• Conclusion: The missing peak in dry olivine isn't because the sliding mechanism is broken. It's because the rock has such a huge variety of "stickiness" at the grain boundaries that the signal gets smeared out.

What This Means for the Earth

The paper suggests that EAGBS is still happening in the Earth's upper mantle, even though we don't see the sharp peak in experiments.

• Dry Rock: Because the boundaries are so diverse, the energy loss is spread out over a wide range of frequencies. It looks like a weak background hum rather than a sharp note. This explains why dry olivine experiments look "boring" (no peak).
• Wet Rock: The paper notes that when olivine has water in it, the peak becomes visible again. The authors suggest that water might make the grain boundaries more uniform (like turning all the runners into identical sprinters), which brings the sharp peak back.

The Bottom Line

The "missing" energy loss peak in dry rocks isn't a mystery of a broken mechanism. It's a case of statistical smearing.

If you have a billion tiny grain boundaries, and they all have slightly different speeds of sliding, their individual "peaks" overlap and cancel each other out, leaving a broad, flat background. This broad background is actually strong enough to explain the energy loss and speed changes we see in the Earth's upper mantle, even without a sharp peak.

In short: The rock isn't silent; it's just singing a chord instead of a single note.

Technical Summary

Technical Summary: Effects of Microstructural Heterogeneity on the Macroscopic Spectrum of Elastically Accommodated Grain-Boundary Sliding

Problem Statement
Elastically accommodated grain-boundary sliding (EAGBS) is a primary candidate mechanism for explaining seismic attenuation and velocity dispersion in the Earth's upper mantle. Classical theory, based on the Raj-Ashby framework, predicts that EAGBS should produce a distinct, localized Debye-like peak in the attenuation spectrum. While such peaks are observed in metals, ice, and wet olivine, they are notably absent or extremely weak in dry olivine experiments, where attenuation appears as a broad, featureless background often attributed to transient diffusion creep. This discrepancy between theoretical predictions and experimental observations in dry olivine remains unresolved. The central question addressed is whether microstructural heterogeneity—specifically variations in grain geometry and grain-boundary viscosity—can reconcile the classical prediction of a sharp peak with the observed broad, weak attenuation in dry olivine.

Methodology
The authors employ 2-D finite-element simulations on periodic polycrystalline aggregates generated via Voronoi tessellations. The model treats grain interiors as isotropic linear elastic solids and grain boundaries as infinitesimally thin viscous interfaces obeying a Newtonian sliding law. Under time-harmonic macroscopic shear deformation, the effective complex shear modulus ($G^* = G' - iG''$) is computed across a range of frequencies.

The study isolates two distinct sources of microstructural heterogeneity:

1. Geometric Heterogeneity: The authors vary the grain-size distribution (using a log-normal distribution with standard deviation $\sigma_a$) while maintaining uniform grain-boundary viscosity. They compare irregular polygonal grains against a benchmark of perfectly regular hexagonal tessellations.
2. Rheological Heterogeneity: The authors assign a log-normal distribution of grain-boundary viscosities ($\eta_{gb}$) to the boundaries within fixed geometric microstructures, varying the distribution width ($\sigma_\eta$) from near-uniform to highly heterogeneous.

To interpret the numerical results, the authors develop a reduced-order 0-D rheological model. This model approximates the macroscopic response as a weighted superposition of localized Debye-like relaxation elements (standard linear solids), where the weighting is determined by the length-weighted probability density function of the grain-boundary viscosities.

Key Results

• Geometric Heterogeneity: Irregular grain shapes (relative to the regular hexagonal benchmark) significantly alter the baseline EAGBS response, increasing the relaxation strength ($\Delta$) from ~0.18 to ~0.28 and shifting the peak frequency. However, increasing the variance in grain size ($\sigma_a$) alone produces only modest changes in the storage modulus and peak height. Crucially, grain-size variance does not cause significant spectral broadening; the attenuation peak remains narrow and Debye-like even with high grain-size variance.
• Viscosity Heterogeneity: In contrast, a broad distribution of grain-boundary viscosities fundamentally transforms the spectrum. As the width of the viscosity distribution ($\sigma_\eta$) increases, the localized Debye peak is progressively suppressed in amplitude and broadened into a weak, distributed attenuation background spanning a wide frequency interval.
• Mechanism of Broadening: Decomposition of the total dissipation into discrete viscosity bins reveals that the broad macroscopic spectrum arises from the linear superposition of many localized relaxation processes, each with a distinct characteristic timescale. The integrated dissipation for a given viscosity class scales approximately linearly with its total boundary length, validating the additive nature of the 0-D model.
• Dry vs. Wet Olivine: The results suggest that the weak peak in dry olivine experiments is consistent with a scenario where grain boundaries sample a broad viscosity distribution (spanning several orders of magnitude), effectively smearing the EAGBS signal into the background. Conversely, the more distinct peaks observed in wet olivine may result from hydration narrowing the effective viscosity distribution, perhaps by creating a more uniform, liquid-like interfacial state.
• Upper Mantle Implications: When applied to upper-mantle conditions using a reduced-order model fitted to surface-wave data, a distributed-viscosity EAGBS mechanism (with a fitted viscosity distribution width $\sigma_\eta \approx 3.92$) successfully reproduces both the nonlinear high-temperature velocity reduction and the attenuation levels ($Q^{-1} \sim 10^{-2}$) observed in the asthenosphere. Narrow distributions fail to produce sufficient attenuation.

Significance and Claims
The paper argues that the absence of a pronounced EAGBS peak in dry olivine does not necessarily imply the absence of the EAGBS mechanism itself. Instead, the macroscopic signature of EAGBS is highly sensitive to the distribution of grain-boundary viscosities. If the viscosity distribution is sufficiently broad, the mechanism manifests experimentally only as a broad attenuation background rather than a distinct peak.

The authors claim that grain-boundary viscosity heterogeneity is a fundamental control on the macroscopic expression of EAGBS in polycrystalline mantle rocks. This finding suggests that EAGBS remains a plausible contributor to upper-mantle seismic attenuation and dispersion, provided that the grain-boundary network samples a sufficiently wide range of rheological properties. The study provides a physical basis for the "missing" peak in dry olivine and offers a mechanism to explain the broad anelastic behavior observed in the asthenosphere without relying solely on transient diffusion creep or partial melting.

Limitations
The authors note that the calculations are restricted to 2-D isotropic elastic aggregates and do not explicitly include crystallographic elastic anisotropy, correlated boundary properties, or coupling to transient diffusion creep. Consequently, the seismic comparison is presented as a plausibility test rather than a unique inversion for mantle grain-boundary properties. Future work is required to extend the framework to 3-D and better constrain the physically realistic distribution of grain-boundary viscosities in olivine.