Temperature, grain size, and chemical controls on polycrystal anelasticity over a broad frequency range extending into the seismic range

Abstract

Recent experimental studies have shown that anelasticity of polycrystalline materials is subject to the Maxwell frequency (f M ) scaling:However, the applicability of this scaling to the seismic waves has not been guaranteed because experimental frequencies normalized to f M of the laboratory samples are usually much lower than the seismic frequencies normalized to f M in the upper mantle (10 6 ≤ f ∕f M ≤ 10 9 ). In this study, by using polycrystalline organic borneol as an analogue to mantle rock, we measured anelasticity up to f ∕f M ≃ 10 8 and found that the Maxwell frequency scaling is not fully applicable at f ∕f M > 10 4 . A closer examination showed that each of the relaxation spectra obtained under various temperature, grain size, and chemical composition can be represented by the superposition of a background dissipation subject to the Maxwell frequency scaling and a peak dissipation centered at f ∕f M ≃ 10 3 . Significant increases of the peak amplitude and width with increasing temperature, grain size, and impurity content result in failure of the Maxwell frequency scaling at f ∕f M > 10 4 , where the peak dissipation dominates over the background dissipation. The peak is significantly broadened near the solidus temperature (T∕T melt = 0.93), producing an absorption band toward the normalized seismic frequencies. The absorption band suggested by the present data is characterized by variable amplitude and width, indicating a nonlinear reduction of seismic velocity near the solidus.

Polycrystal anelasticity at near‐solidus temperatures

“…However, the data obtained so far are not enough to understand the underlying mechanism which is necessary to extrapolate the experimental data to the mantle conditions. To address this lack, forced oscillation tests on the rock analogue (polycrystalline aggregates of organic “borneol”) have been performed [ McCarthy et al , ; McCarthy and Takei , ; Takei et al , ]. The attenuation spectra obtained from the rock and rock analogue generally consist of a monotonic part ( , α = 0.2–0.3) observed at low frequencies and a plateau or flattening observed at high frequencies.…”

“…The Maxwell frequency f M is calculated from the unrelaxed elastic modulus M U and diffusion creep viscosity η as f M = M U / η . Using the normalized frequency f / f M as a single master variable, the high‐temperature background can be expressed as [ Takei et al , ]. This result shows that the underlying mechanism of the high‐temperature background is “diffusionally accommodated grain boundary sliding”, which is rate limited by matter diffusion in the same manner as the diffusion creep [e.g., Raj , ].…”

“…Generally, the high‐frequency peak is small for melt‐free rocks [ Jackson and Faul , ] and large for melt‐bearing rocks [ Jackson et al , ; Sundberg and Cooper , ]. Recent data obtained from the rock analogue showed that the amplitude and width of the peak increase with increasing temperature and grain size [ Takei et al , ]. Takei et al [] considered this result as an important clue to understand the mechanism for the large high‐frequency peak in the partially molten rocks.…”

“…Recent data obtained from the rock analogue showed that the amplitude and width of the peak increase with increasing temperature and grain size [ Takei et al , ]. Takei et al [] considered this result as an important clue to understand the mechanism for the large high‐frequency peak in the partially molten rocks. However, their data were limited to the subsolidus temperatures ( T / T m ≤0.93, T m : solidus temperature).…”

“…Although the anelasticity of the same material at the subsolidus and supersolidus temperatures was reported by McCarthy and Takei [], the modulus data of the forced oscillation test in this previous study were not accurate enough to quantitatively estimate the total relaxation strength of the high‐frequency peak. The accuracy of the modulus data was greatly improved by the new forced oscillation apparatus and pore‐free samples developed by Takei et al []. Using these improved methods, we obtained a new data set which enables a detailed examination of the high‐frequency peak at near‐solidus temperatures.…”