Skip Navigation | ANU Home | Search ANU | Directories

RSES SITE SEARCH

Seismic wave attenuation in polycrystalline olivine: the influence of
grain size and partial melting

Ian Jackson, John Fitz Gerald, Uli Faul, Harri Kokkonen and Joshua Carr

---

Much of the variation of seismic wavespeeds and attenuation in the Earth's upper mantle is attributable to high-temperature viscoelastic relaxation in rocks composed mainly of the mineral olivine. Quantitative interpretation requires rheological data obtained under controlled laboratory conditions on well-characterised materials. In order to build a mechanistic understanding of high-temperature viscoelastic relaxation in such ultramafic materials, we are committed to an ongoing programme involving fabrication and characterisation of a suite of synthetic polycrystalline olivine aggregates and measurement of their mechanical properties at high temperature through a combination of torsional forced oscillation and microcreep tests. Last year we reported on the frequency, temperature and grain-size sensitivity of the strain energy dissipation and modulus dispersion for melt-free olivine polycrystals. These departures from ideal elastic behaviour are attributed to diffusional processes occurring predominantly at grain boundaries - namely elastically and diffusionally accommodated grain-boundary sliding (Figure 5). Further analysis this year led to the realisation that, for any given mean grain size, the strength of the grainsize sensitivity must vary with distance in temperature-period space from the threshold for essentially elastic behaviour (plotted in the lower right of Figure 6). It follows that the mild grain size sensitivity measured for dominantly anelastic behaviour under conditions not too far removed from the elastic threshold provides the most robust extrapolation to the much larger grain sizes of the upper mantle.

Figure 6. A schematic deformation mechanism map for the linear viscoelastic behaviour encountered at very low stress and strain levels in fine-grained polycrystalline materials, illustrating the expected frequency, temperature and grainsize sensitivity of the strain energy dissipation. The parallel lines trending from lower left to upper right (of slope E/R) are contours of constant Q-1 associated with constant values of X µ To exp(-E/RT). As X increases from its value at the elastic threshold (i.e. moving towards the upper left of the figure by increasing temperature T or period To) the frequency and grain-size sensitivity of the dissipation steadily strengthen.

During 2001 the focus has shifted increasingly to the behaviour of partially molten ultramafic materials. To this end pellets have been cold-pressed from mixtures of Fo₉₀ olivine ([Mg_0.9Fe_0.1]₂SiO₄ - either natural or synthetic) and synthetic basaltic glass powders containing either 2 or 4% of the basaltic component. These pellets have been converted into dense polycrystalline aggregates by hot-isostatic pressing within Ni₇₀Fe₃₀ foil-lined mild steel sleeves in an internally heated gas-medium apparatus typically for 25 hr at temperatures of 1200-1300°C and pressures of 200-300 MPa. Firing of some of the precursor powders and in some cases also the hot-pressed specimens has resulted in substantial variation of the water content (~20-250 wt ppm H₂O) of the resulting specimens.

Both shear modulus and strain energy dissipation Q^-1 have been inferred from torsional forced oscillation measurements performed during slow staged cooling to room temperature following a protracted annealing period at the highest temperature Ð identical in most cases to that of the prior hot-pressing experiment. Exploratory experiments have revealed a variation of dissipation with oscillation period and temperature that is qualitatively different from the monotonic variation characteristic of the melt-free materials. For the melt-bearing specimens a well-defined Q^-1 plateau at temperatures of 1400-1500 K separates more markedly frequency and temperature dependent dissipation at both lower and higher temperatures (Figure 7(a)). That this perturbation is associated with a melt-related dissipation peak superimposed upon the monotonic background characteristic of melt-free materials is seen most clearly in the plot of log Q^-1 versus 1/T (Figure 7(b)). The peak enters the observational window from periods longer than 100 s contributing to a steepened dependence of Q^-1 upon period at 1270-1320 K. Its systematic displacement to shorter period with increasing temperature results in reduced frequency sensitivity first at long periods and ultimately across the entire observational window. The nearly frequency-independent behaviour at 1420 and 1470 K evidently results from the near cancellation of the contrasting frequency dependences associated with the background and the long-period side of the dissipation peak. At sufficiently high temperatures, the melt-related dissipation peak moves to periods significantly shorter than 1 s and the monotonic frequency and temperature dependence associated with the background dissipation is progressively restored.

Figure 7 The variation of strain energy dissipation with oscillation period and temperature for specimen #6384 (sol-gel Fo90 + 4% basaltic melt). Data are indicated by the plotting symbols, whereas the curves represent a fit to the data involving superposition of a Gaussian dissipation peak upon the monotonic background characteristic of melt-free materials.

In work in progress, the methods of light microscopy and scanning and transmission electron microscopy are being combined to determine the temperature-dependent melt fraction in such specimens and the nature of the grain-scale melt distribution (Figure 8). This additional information is expected to help identify the cause at the microscopic scale of the observed melt-related dissipation. The stress induced local redistribution of melt ('melt squirt') is a serious contender.

Figure 8 A suitable micrograph showing melt distribution -possible from one of the recent cook-quench-look experiments on AT6384. The unusual frequency dependence of dissipation seen in these exploratory experiments, if confirmed more generally, might prove to be a seismologically observable characteristic of partially molten regions of the Earth's upper mantle.