Just some empty space here

Phase transitions, dynamics, and thermal structure of the core-mantle boundary region The recently discovered post-perovskite phase transition in Earth's lowermost mantle offers an unprecedented opportunity to probe variations in temperature near the core-mantle boundary, where heat is transferred from the core into the mantle. Additionally, this new tool may shed light on a number of other important issues in Earth Science, such as: Does subducted oceanic lithosphere sink all the way to the bottom of Earth's mantle? Do large scale "primordial" chemical reservoirs exist in the lower mantle? Is the thermal boundary layer in D" convectively unstable, and thus able to nucleate deep-seated mantle plumes that trace out volcanic chains ("hot spots") at Earth's surface? Below you will find additional information regarding our recent interpretation of the manifestation of post-perovskite in Earth's lowermost mantle (the "double-crossing") addressing various comments I've received since our paper went to press: Hernlund, J., C. Thomas, and P.J. Tackley, A doubling of the post-perovskite phase boundary and structure of the Earth's lowermost mantle, Nature, 434, 882-886, 2005. (PDF Reprint) Issue #1: How might variations in mantle chemistry affect your double-crossing model? This is an important question for determining how post-perovskite manifests itself in a more realistic multi-phase system. The MgSiO3-perovskite phase is known to have a finite solubility for other chemical components, including: FeO, which forms a limited solid solution with MgO in the binary system MgSiO3-FeSiO3 perovskite, with maximum FeSiO3 mole fractions of less than 1/2. The solubility of FeO in post-perovskite relative to co-existing Fe-periclase in a pyrolitic composition is the subject of debate, as Mao et al. [2004] have reported a higher FeO while Murakami et al. [2005] have reported less FeO. In either case, the smaller size of Fe+2 (relative to Mg+2) is expected to lower the pressure of the phase transition, creating a two-phase region in the MgSiO3-FeSiO3 phase diagram for the post-perovskite transition. Fe2O3, which by itself forms a structure similar to post-perovskite at lower pressures than pure MgSiO3, and is therefore expected to cause a lowering of the transition pressure if it is significantly soluble in post-perovskite. Al2O3, which is soluble in MgSiO3-perovskite, though it may be strongly coupled with Fe2O3. New ab initio calculations suggest a higher transformation pressure for Al2O3 in assuming a structure similar to post-perovskite, thus this component would likely increase the phase transition pressure if it were significantly soluble in post-perovskite. (Note: Trace elements that are generally incompatible in the upper mantle might be more likely to enter the CaSiO3-perovskite phase because Ca has a large ionic radius). More work obviously needs to be done in order to assess the affects of variable chemistry on the post-perovskite transition. However, dynamical considerations can give us some insight into how chemical variations might affect post-perovskite distribution within a convecting mantle. For example, the large chemically distinct dense FeO-rich "piles" postulated to exist beneath the Pacific and Africa will have become relatively hot while sitting on top of the core-mantle boundary and are therefore less likely to have any post-perovskite within them (post-perovskite prefers cooler temperatures) unless the increased FeO has a large enough effect on the phase boundary. In particular, a shallowing of the phase boundary might permit warmer material to transform to post-perovskite, though in the hottest portions of piles this might be difficult to achieve. In colder regions of the lowermost mantle, smaller scale variations in chemical heterogeneity might give rise to complexity in the post-perovskite phase boundaries. A shallowing of the phase boundary by the increased presence of Fe (in any form) will be convectively unstable, because the elevated phase transition in addition to the heavier Fe will complement one another in creating a negative buoyancy anomaly that will tend to drive Fe-rich material downward relative to Fe-poor material. Therefore any smaller-scale variations along the post-perovskite phase boundary due to chemistry will probably arise from chemical heterogeneities that are presently being advected into cooler parts of D", and may be fundamentally transient features. A likely candidate of this type may occur when subducted oceanic lithosphere encounters the D" layer, with layers of basaltic composition crust contrasting with the depleted former mantle lithosphere. A particular example of this scenario might be what is happening beneath the Cocos plate at present, where extremely rapid variations in discontinuity topography might not be compatible with the smoother profiles generated by temperature variations alone. The observational consequences of dissolved Fe oxides or other components upon the discontinuity created by the post-perovskite phase transition are more straightforward to address, and were mentioned in the original paper (final paragraph). The primary effect is to broaden the transition into a gradient, with the top of the two-phase region containing more iron-rich post-perovskite. According to seismic studies, a gradient of up to 75 km is possible for the discontinuity, which translates to about 4 GPa interval in pressure. Recent experimental results (e.g. Murakami et al., 2005) yield numbers in this range (but slightly higher). It is important to note, however, that the seismically observed gradient can be smaller than the actual width of the two-phase region generated by solid solutions, and therefore the value of 4 GPa might be taken as a lower bound. Issue #2: How reliable is the seismic interpretation that two (rather than just one) discontinuities appear in D"? This is an ongoing subject of research, and one which the seismology community will clearly be addressing from a number of different approaches in the near future. Our goal in the original paper was simply to show that a double-crossing interpretation was consistent with previous data analysis and interpretations for which two discontinuities were found beneath Eurasia and the Cocos plate. The upper discontinuity is well-established, but the lower discontinuity exhibits a velocity decrease and these types of features are notoriously difficult to probe with seismic waves. Several groups are using a variety of other techniques to look for the double discontinuities in these and other regions, such as stacking much larger data sets. The preliminary results of these studies seem to confirm the predictions of a double-crossing model, so stay tuned on this topic! Issue #3: The core-mantle boundary heat flow implied by the double-crossing model seems very large, how can this be reconciled with credible core thermal histories? This is also an ongoing research effort, which is obviously of significant importance for understanding the thermal and dynamical history of the Earth, and also carries implications for the distribution of radiogenic elements (e.g. 44 TW observed at Earth's surface minus 9-13 TW at the CMB requires 31-35 TW of radiogenic heat production in the mantle), perhaps requiring some sort of internal heating in the core (e.g., Potassium). For now, it suffices to say that the figure of 9-13 TW given in the paper is a gross extrapolation, and does not account for possibly significant variations from place to place. In particular, the heat flow beneath the so-called "superplume" structures beneath the Pacific and Africa could be substantially smaller (i.e., by a factor of two or more) than in regions where slabs can exist, and these should not be included in an areal extrapolation. However, there remains the additional factor that 9-13 TW is a lower bound on heat flow in the gross extrapolation. Thus the removal of some regions from the extrapolation would lower the estimate of CMB heat flow, while the fact that it is a lower bound in the presence of subducted slabs increases the estimate of the actual heat flow. It may be that these two effectively cancel one another out, but further research is required to find better constraints. Another element of uncertainty implicit in the estimate of heat flow is the appropriate value for the thermal conductivity, which is uncertain by at least a factor of two. In particular, the contribution of radiative transfer is not well known (especially in the composite aggregate of phases in D"), and may be important. An exciting possibility is that high density sampling of discontinuities might enable an inference of thermal diffusivity from the length-scales of heterogeneity observed, although this is complicated by potential chemistry effects.