Measurement of elastic properties of silicates at realistic mantle pressures

Measurements of seismic wave travel times through the Earth’s interior provide one of the few sources of information on the properties of deep mantle rocks. These travel times can be interpreted in terms of their implications for the mineralogy, chemistry and temperature of the mantle if they can be compared with models determined from laboratory measurements for the elastic properties of mantle minerals at high pressures and temperatures. In this study MHz ultrasonic measurements have been used to determine the velocities of P and S acoustic waves in mantle minerals at mantle conditions. These results are used to constrain the properties of the mantle through comparison with seismic data. Subduction zones exhibit faster seismic wave velocities compared to the surrounding mantle due to the recycling of relatively cold oceanic lithosphere. In certain subduction zones, however, a 5-10 km thick low velocity layer (LVL) has been inferred to exist along the top surface of the subducting slab at depths of up to 250 km. Shear-wave velocities, in particular, within these layers have been estimated as up to 10% slower than in the surrounding mantle. High-pressure ultrasonic interferometric measurements were performed to gain insight into the elastic properties of lawsonite [CaAl2(Si2O7)(OH)2.H2O], a hydrous mineral phase stabilized under cold subduction zone conditions. It was found that lawsonite has an unusually low shear modulus at high pressure and its formation in subducted oceanic crust can explain some seismic evidence for LVL at depths exceeding 100 km. To approach estimated LVL velocities requires lawsonite to form in the subducting crust as a result of a fluid influx due to the breakdown of other hydrous minerals such as serpentine. The formation of lawsonite additionally lowers seismic velocities because it forms at the expense of garnet, a mineral with relatively fast seismic velocities. LVL observations may therefore be used to place constraints on the amount of H2O subducted into the deep mantle. Chemical heterogeneities in the transition zone are potentially developed through the subduction of basaltic oceanic crust and lithospheric ultramafic mantle. The mineralogy of these lithologies in the transition zone will mainly differ in terms of the proportion and chemistry of garnet. Garnets formed from originally basaltic and ultramafic compositions will differ in the proportions of Ca, Fe and majoritic component, with the former being richer in all components except majorite. Most models use partial derivatives of elastic properties with respect to composition to determine the elastic properties of complex garnet compositions in the mantle. In this study the effect of chemical variations on seismic wave velocities of garnets up to transition zone conditions have been intensively studied by examining the effects of Ca, Fe and majoritic substitutions on the elasticity of pyrope. In addition, complex multicomponent garnet compositions, expected to form from both subducted basaltic material and typical ultramafic mantle at transition zone conditions, have been studied. The results indicate that elastic properties of multicomponent garnets can be quite reliably interpolated from the end member properties, with an accuracy that is generally within the experimental uncertainties. No evidence was found that large excess properties exist that cause non-linear contributions to multicomponent elastic properties for the mineral garnet. Deviation between the two calculated mantle models, pyrolitic and MORB, is about 3% for both VP and VS, with the modelled slab composition being slower. Within the uncertainties of the seismic data therefore a pyrolite mantle composition is quite consistent with velocities at the base of the transition zone. Relatively low temperatures and olivine-rich nature of the stagnant slab support that there would be few grounds to argue that significant accumulation of MORB composition material may occur at the base of the transition zone. Ultrasonic interferometry measurements in conjunction with in situ X-ray techniques have been used to measure compressional and shear wave velocities and densities of MgSiO3 perovskite (Mg-Pv) and perovskite ((Mg, Fe)-Pv) in the multianvil at pressures up to 25 GPa and temperatures to 1200 K. Data for Mg-Pv are consistent with previous studies and the (Mg, Fe)-Pv sample has almost identical shear properties to Mg-Pv. The adiabatic bulk modulus, Ks, for (Mg, Fe)-Pv, however, is found to be substantially lower than Mg-Pv, with a refined value of 236 GPa and a pressure derivative of 4.7. It is proposed that this low KS value results from a change in the elasticity of Fe-bearing perovskite at low pressures <30 GPa. High temperatures measurements of VP and VS of and perovskite samples lead to the conclusion that the substitution of Mg by Fe seems to have a consistent effect on Ks, strongly lowering it in both Al and Al-free perovskites while it does not have any noticeable effect on the shear elastic properties. By substituting Al into the Si octahedral site, the rigidity of this framework is being reduced which clearly has an influence on the shear modulus G. G of the Al-bearing perovskites is obviously controlled by the Al content in the structure. Seismic observations indicate low visibility of underside PP reflections from the 660 km discontinuity. Seismologists have used this in the past to argue that the P velocity contrast at the discontinuity is much lower than would be expected from seismic reference models. The data collected on Fe-Al perovskite in this study provide excellent support for this argument as model calculations based on these data show very little contrast in Vp at 660 km depth. Instead, the model derived in this study indicates that much of the change in P velocity between the transition zone and lower mantle occurs over a much broader depth, 200km, interval and is mainly attributed to the transformation of garnet to perovskite. A major conclusion therefore is that a sharp 660 km discontinuity reflection should be visible in S but not P wave seismic data. The model calculation for VS is in a good agreement with the seismic reference models up to depths as great as 1071 km, while VP remains 3 to 4% lower at lower mantle conditions. The extrapolations of pressure and temperature dependences significantly out of the ranges over that they were measured may well explain this discrepancy however.

Level thesis:degree_level

thesis.doctoral

Grantor dc:publisher

Universität Bayreuth

Year

2013

Author dc:creator

• Chantel, Julien

Contributors dc:contributor

• Frost, Daniel

Repository record source_url

https://epub.uni-bayreuth.de/id/eprint/83/

OAI identifier oai:identifier

oai:epub.uni-bayreuth.de:83