289 The Transition-Zone Water Filter Model for Global Material Circulation: Where Do We Stand? Shun-ichiro Karato, David Bercovici, Garrett Leahy, Guillaume Richard and Zhicheng Jing Yale University, Department of Geology and Geophysics, New Haven, CT 06520 Materials circulation in Earth’s mantle will be modified if partial melting occurs in the transition zone. Melting in the transition zone is plausible if a significant amount of incompatible components is present in Earth’s mantle. We review the experimental data on melting and melt density and conclude that melting is likely under a broad range of conditions, although conditions for dense melt are more limited. Current geochemical models of Earth suggest the presence of relatively dense incompatible components such as K 2 O and we conclude that a dense melt is likely formed when the fraction of water is small. Models have been developed to understand the structure of a melt layer and the circulation of melt and vola- tiles. The model suggests a relatively thin melt-rich layer that can be entrained by downwelling current to maintain “steady-state” structure. If deep mantle melting occurs with a small melt fraction, highly incompatible elements including hydro- gen, helium and argon are sequestered without much effect on more compatible elements. This provides a natural explanation for many paradoxes including (i) the apparent discrepancy between whole mantle convection suggested from geophysical observations and the presence of long-lived large reservoirs suggested by geochemical observations, (ii) the helium/heat flow paradox and (iii) the argon paradox. Geophysical observations are reviewed including electrical conductivity and anomalies in seismic wave velocities to test the model and some future direc- tions to refine the model are discussed. 1. INTRODUCTION The most well-known on-going chemical differentiation process on Earth is partial melting beneath mid-ocean ridges, which is primarily due to the adiabatic upwelling ( pressure- release melting) and creates relatively “enriched” oceanic crust and “depleted” residual mantle (e.g., [Hofmann, 1997]). If this is the only chemical differentiation process operating at present and in the recent past, then several immediate consequences follow. For example, if one accepts the model of whole mantle convection, as suggested by seismic tomog- raphy ([Grand, 1994; van der Hilst, et al. , 1997]), then the whole mantle is depleted with only a small volume (~10%) of relatively enriched material continuously replenished by subduction of oceanic crust and sediments. This view is inconsistent with geochemical observations that suggest the entire lower mantle (~70% of the mantle) is relatively enriched [Albarède and van der Hilst, 2002; Allègre, et al. , 1996]). An alternative model assumes that differentiation at mid-ocean ridges involves material only from the upper mantle (e.g., [Allègre, et al. , 1996]). In this model, the deeper mantle is not involved in chemical differentiation at mid- ocean ridges and maintains relatively enriched chemical composition. However, this latter model implies strongly Earth’s Deep Water Cycle Geophysical Monograph Series 168 Copyright 2006 by the American Geophysical Union. 10.1029/168GM22