U21B-0399
The Allegresque History of Chemical Geodynamics: the Tenth Anniversary Revision
Chemical geodynamics is a field of studies of the dynamic workings of the earth using chemical and isotopic tracers. It is a forum where geophysicists and geochemists exchange thoughts and ideas, and Prof. Allegre's contributions to the field go much beyond the term he coined. At the Allegre Symposium in 1997 in Paris, a tapestry depicting the history of chemical geodynamics was unveiled. The evolution of the field, creatures and denizens in it, and some historical defining moments of the field were pictorially woven into it, covering a period of 1960 - 1997 CE (i.e., 23 - 60 ACA, or Anno Claude Allegre). For the occasion of the second Allegre Symposium, a new segment of the tapestry has been woven, depicting significant advances made in the field over the past ten years. Highlights emphasize expanding collaboration of dynamicists, geochemists and mineral physicists, and include chemical structure of mantle plumes, convective stirring and scale lengths of mantle heterogeneity, early earth, and deep earth. Days of Enlightenment are near?
U21B-0400
Mantle mixing revisited: The influence of viscosity and implications for chemical geodynamics
The introduction of the term chemical geodynamics was accompanied by an interest in the processes by which mantle convection disperses heterogeneities. The rate and efficacy of mantle mixing constrains the origin and fate of the isotopic heterogeneity seen in mid-ocean ridges and oceanic islands. Heterogeneity is introduced into the mantle by processes associated with melting, alteration, and subduction, while heterogeneity is destroyed by the stirring action of convection, the stretching and folding common to all kinematic mixing processes. Mantle mixing is affected by a variety of factors, including time-varying flow, plate motion, viscosity variations, and phase transitions. Chaotic mixing can be observed in calculations of time-varying 2-D flows; this is confirmed by calculating the deformation of strain markers to obtain an estimate of the Lyapunov exponent. Chaotic mixing can be countered to some extent by increases in mantle viscosity at depth and by high-pressure solid-state phase transitions that influence the pattern of convection. Stirring may be rapid on a regional scale (resulting in fairly uniform mid-ocean ridge basalts on length-scales up to thousands of km) while heterogeneities at the global scale of the Dupal anomaly are retained for billions of years because of isolation across long-wavelength cells. In numerical calculations of the paths of passive tracers in convection, extreme changes of viscosity and thermal conductivity in the lower mantle slow the rate of flow in the lower mantle, but this does not much inhibit the overall rate of dispersal of heterogeneities. Essentially, the regions that exhibit high rates of stretching and thinning have the most important influence on mixing, and packets of material that are stirred rapidly in regions of high strain rate are carried wholesale into regions of more sluggish convection. The analysis of mixing is complicated by the fact that structures created by passive tracers in 3D exhibit different characteristics than structures observed in 2D models. Nevertheless, it remains difficult to account for mantle isotope systematics and for the Earth's overall heat flow budget without invoking some barriers to flow in the mantle's interior, such as would be provided by mantle layering.
U21B-0401
Marble Cake Perspectives
Since the original suggestions by Hanson (Geol. Soc. London, 1977) and Allegre and Turcotte (Nature, 1986), the concept of a "veined" or "marble cake" mantle has gained wide acceptance as a paradigm for mantle composition and components. The "veined mantle" was conceived thinking of the mantle as an ultramafic migmatite with many types of veins, but emphasized metasomatic components contained in hydrous phases as an explanation for alkali basalts. The "marble cake" mantle emphasized recycled oceanic lithosphere. Both types of veins are inevitable consequences of mantle convection. Oceanic lithosphere is recycled and stretched; low melting components of the mantle are inevitably melted in ascending mantle flow, even beneath thick lithsosphere. Both vein types have been widely invoked to explain incompatible element enriched basalts from the mantle. Most recently, disequilibrium melting of veined mantle sources by various mechanisms have become a popular suggestion to explain diverse aspects of mantle geochemistry (e.g. Sobolev et al., Nature, 2005; Phipps Morgan et al., EPSL, 1999). The physical mechanisms that would allow disequilibrium melting of fine scale veins, however, remain to be demonstrated. Average upper mantle composition is residual to continents and requires removal of low F melts to generate the depleted MORB source, and enrichment by low F melts to create the enriched source. Such a process is also necessary in the Sobolev et al model for Hawaii, which generates the equivalent of a low F melt by two stages of larger degree melting. Enriched sources are not restricted to ocean islands, and the name "OIB source" is a misnomer. Enriched basalts occur on normal ridges, in back-arc basins, behind subduction zones, in continental rifts and in isolated volcanic cones. Most of these are not mantle plumes. Enriched components have been ascribed to recycled ocean lithosphere, but recycled ocean crust is depleted, not enriched. Therefore the isotopic signature of crustal components are needed, but the trace element chemistry requires the influence of low degree (low F) melts. These observations can be reconciled by the recognition that low degree melting of recycled eclogite (as well as hydrous peridotite) is inevitable. One likely location is in subduction zones, particularly in light of the recent evidence for hot mantle wedges. At depth in subduction zones, where rutile is not residual, eclogite would melt and metasomatize overlying peridotite. Various flavors of this material are suitable for enriched sources. This environment also provides a much more physically plausible mechanism for creating large volumes of enriched pyroxenite lithologies and separating them from their eclogite source. If eclogite melts at subduction zones, it is even less suitable as an ultimate source of enriched basalts. Tests of this mechanism of mantle heterogeneity come from the occurrence of enriched basalts in arc back-arc environments. Enriched basalts in back-arcs are very common, more abundant than on normal ocean ridges, and often have an isotopic composition that requires very recent low degree melt addition. The recycled eclogite, depleted by low degree melt, then provides fertile major elements and depleted trace elements that when added to depleted peridotite is suitable for the depleted MORB source.
U21B-0402
Plumes, Superplumes, and Dupal
The Dupal anomaly is a prominent low-order spatial pattern in the geochemistry of oceanic basalts. In 1988 Castillo noted an apparent correlation between the Dupal spatial pattern and the large-scale structure of seismic velocity anomalies in the deep lower mantle. Here we explore a simple physical mechanism that can explain this correlation. First, long-time subduction around the rim of the Pacific would drive flow in D" toward the center of the 'rim of fire' and the center of the African superplume. This continuous supply of D" would promote the concentration of plumes into the superplume regions. This is relatively independent of migration of the subduction zones as the Pacific closes in pace with the opening of the Atlantic and Indian oceans. Second, longterm outward conduction of heat from the plume pipes would heat the mantle between the pipes (it would heat outward ~100 km in 100 m.y.; ~200 km in 400 m.y.). This heating of the 'background' mantle to plume-rim temperatures would change the geotherm of the superswell region compared to the mantle beneath 'normal' suboceanic asthenosphere. The result would be that a larger fraction of ambient mantle would be heated enough for its lowest-solidus plums to begin to partially melt, and to become buoyant enough for this non-plume mantle to rise into and become part of the overlying layer of hot plume-fed asthenosphere – i.e. only in superplume regions is non-plume material hot and buoyant enough to upwell in significant amounts into the asthenosphere. We predict that Dupal material differs in its mean age from the age of the D'' source of other plume-fed asthenosphere, and also differs in that it has been (on-average) cooler and less-melted than ‘average' plume-fed asthenosphere. This leads to its distinctive isotopic differences. Global asthenosphere flow then preferentially transports Dupal material towards the southern ocean spreading centers (Yamamoto et al., 2007). The patchiness of superswell volcanism reflects the pulsing of higher-speed upwelling and melting within the background large-scale upwelling of superplume material.
U21B-0403
The role of sulfur in the Earth's formation
When sulfur is considered in the Earth's formation it is mostly because of its possible role in lowering the density of the Core. This, however, looks less and less likely as experimental studies of its effects on other physical properties of the Core progress. On the contrary it could have important effects on the relative concentrations of other elements, like the refractory lithophile elements (RLE). « Reference » Earth models have not basically changed for the last thirty years and they more or less rely on two basic assumptions : 1- One can, by a way or another, retrieve a « primitive » mantle composition from the composition of accessible peridotite samples. 2- In that primitive composition the ratios of RLE concentrations are equal to those in the CI chondrites (that is essentially one chondrite, Orgueil). These two hypotheses, when looking carefully at the sets of data from which they derive are somewhat conjectural and the best balanced discussion of their possible use still remains that of Hart and Zindler (1), more than 20 years ago. It appears more and more that the reducing phase through which the Earth-forming materials had to go to render possible the formation of the Core was characterized, among other things, by high sulfur fugacities. This does not mean necessarily that the final sulfur concentration of the Earth is high, because these high sulfur fugacities were probably labile at the million years' scale. But these sulfur fugacities induced a partitioning of elements between the sulfide phases, the metal and the silicates. Since the accretion of the Earth involved a somewhat random mixing of these three phases' categories, the partitioning of elements, and, for example the so called Refractory Lithophile Elements (RLE), depended largely on the relative stability of their sulfides. This is true for example of the Ca/Al ratio. Ca is very concentrated in the sulfides, whereas Al resides mainly in the silicates. Other elements will be similarly fractionated, wether they are refractory (eg U, some REE) or volatile (K vs Na) Hence, if part of the very limited and very biased sampling of the Earth's Mantle can somehow retrieve some of the chondritic RLE ratios, this is probably much more because of correcting effects of silicate partial melting during accretion, than by a magic stability of these ratios. (1) Hart S.R. & Zindler Chem. Geol. 57 247-267 1986
U21B-0404
Isotopic Heterogeneity and Mantle Stirring: Importance of Early Versus Late Processes
Mid-ocean ridge basalts (MORBs) that make up the bulk of the oceanic crust (OC) exhibit a relatively uniform isotopic composition. In particular, Sm-Nd isotopic results, suggests that their mantle source had been depleted (the depleted mantle or DM) by melt extraction to form the continental crust (CC) over geologic time with a mean age of extraction of 1.8 Ga. With reservoir models that take into account mantle heterogeneities by keeping track of all sub-reservoirs in DM we have established that the most frequent MORB isotopic compositions faithfully record the average isotopic composition of DM. Mass balance considerations based on Nd and Sr isotopic thus still suggest that mass of the DM is only about 30 % of the mantle. The lower mantle could thus in principle have a composition close to the bulk silicate Earth. We have a good model for sampling the DM but need a better understanding of its location/distribution in the mantle. It is now well established that 146Sm was live in the early Solar System and that samples from Isua, W. Greenland (~ 3.8 Ga old) have high 142Nd/144Nd values when compared to normal terrestrial Nd. This demonstrates that the earliest mantle differentiation (likely in a magma ocean) happened within 100 Myr of Solar System formation and has opened up possibilities for better understanding the earliest part of mantle evolution. Basalts from ocean islands show a much wider range in isotopic composition than MORBs. Their (and MORB) Nd, Sr and Pb isotope data can be described rather well by four isotopic components: DMM, HIMU, EMI, EMII. Many of the OIB data arrays point toward an isotopic component called FOZO or C. Our current analysis shows that these apparent isotopic components are non-existent or fictitious components/reservoirs. The real components are never seen because of the averaging that happens when basaltic melts sample the mantle. In particular, the isotopic composition of FOZO corresponds closely to the average composition of the matrix in the DM, which consists of small length- scale sub-reservoirs (<15km). Part of OIB or plume isotopic signatures are likely to originate in the lower mantle or the D" layer. The isotopic evidence suggests that the lower mantle was completely molten during part of the early magma ocean stage. Recent suggestions that the D" layer formed by foundering of a very early enriched crust is incompatible with the fact that the mean age of differentiation of the DM-CC system is on the order of 2 Ga (as shown by long-lived chronometers). Long-lived chronometers are of course much more reliable in giving us the long-term evolution of major earth reservoirs. However, extinct nuclides can give us a view into early processes that are not at all or only poorly recorded by the long-lived chronometers. We need better models for how plumes sample the mantle to make good use of the isotopic data on plume basalts.
U21B-0405 INVITED
Modeling The Role of Subduction in the Production and Evolution of Thermal and Chemical Heterogeneity in the Mantle
One cornerstone of multidisciplinary chemical geodynamics is modeling. Both numerical and laboratory experiments have been employed in modeling studies of mantle dynamics. The different techniques bring different strengths to the field. There are many examples in the scientific literature where laboratory methods are used to develop empirical relationships between the response of a fluid system and simple combinations of the physical parameters that define the system. Laboratory methods are also useful in modeling fine scale thermal- chemical structures in three dimensional flow morphologies and for testing first-order features of observationally based, dynamic process models. The focus of this talk is subduction, which introduces heterogeneity deep into the mantle and acts to stir the mantle. We summarize results from laboratory experiments on the relationship between subduction and both the deep production of thermal-chemical upwellings and how mantle heterogeneity is recycled to the surface. In one set of experiments chemically laminated slabs are subducted into a deep thermal boundary layer. Different styles of buoyant upwelling evolve from the slab pile through time, which are related to the isothermal densities of the slab laminates. Observed length scales of chemical heterogeneity ranging from passive 1 km features in tendrils embedded within upwellings to active 300km wide features within plume heads have implications for melt production models. In a second set of experiments, laboratory models consider the interaction between rollback subduction and back-arc extension, three-dimensional mantle circulation and the shallow dispersion of buoyant upwellings. The models are developed to represent basic features of the Cascades subduction system. Results suggest that first order features of the Cascades system (e.g., gross spatial and temporal patterns in melt production) may be consistent with a model in which an upwelling is deformed by large scale, subduction driven shear flow in the mantle wedge. Flow induced by slab rollback efficiently draws plume material into the wedge and produces spatial offsets (~300 km) between the surfacing point of the plume head and the subsequent surface expression of the plume tail. Continued deformation of the ponded head and the tail leads to the evolution of two linear, time-progressive thermal/chemical features beneath the overriding plate which trend in opposite directions. We attempt to place basic features recorded in both laboratory models within the context of a road map for chemical geodynamics.
U21B-0406
Geodynamic Regimes of the Mantle - Constraints From a He-Inclusive Global Data Compilation
27 years after the introduction of Chemical Geodynamics by Allègre (1982), fundamental issues about the geodynamic evolution of the Earth's mantle remain to be resolved. Geophysical evidence for whole mantle convection combined with the difficulty to preserve mantle heterogeneity since the early Earth contrast with geochemical evidence for longterm isolation of parts of the Earth's mantle. Noble gases, in particular helium, remain central to this discussion with three contrasting models considered here: (1) The layered mantle by Allègre et al. (1983) with the lower mantle preserving high 3He/4He ratios since early Earth; (2) The D" layer model by Tolstikhin & Hofmann (1995) with D" preserving high 3He/4He ratios since early Earth; (3) The incompletely degassing mantle model of Class and Goldstein (1995) with depleted mantle domains in the lower mantle preserving high 3He/4He ratios since 1-2 Ga. Here we present new constraints from our updated He-inclusive global data compilation. Ocean island basalts show a correlation between 3He/4He and Th contents, consistent with He isotope ratios reflecting the radiogenic production rate of the variably enriched OIB mantle sources. A high 3He/4He – low Th component is required to explain the global systematics. Constraints on the Th-U content in the three models place tight constraints on the absolute and relative contribution from the high 3He/4He reservoir to OIB mantle sources. Both the layered mantle model and the D" model requires small and extremely uniform contributions of a few percent or permil respectively; contributions outside of this small range, even if slightly higher or lower, are inconsistent with the global OIB data. Only the incompletely degassing mantle model allows for variability in the scale of contributions from old depleted mantle preserving higher 3He/4He ratios and from sources having a variety of ages. This model is also consistent with the requirement of dominantly peridotitic mantle sources for OIB. Allègre, C.J. (1982) Tectonophysics 81, 109-132. Allègre, C.J. et al. (1983) Nature 303, 762-766. Tolstikhin, I.N. & Hofmann, A.W. (2005) Phys. Earth Planet. Int. 148, 109-130. Class, C. & Goldstein, S.L. (2005) Nature 436, 1107-1112.
U21B-0407
Scale Length of Mantle Heterogeneities: Helium Diffusion Constraints
While Earth's mantle is unequivocally heterogeneous, the size, formation and distribution of these geochemical heterogeneities remain enigmatic. Following the veined mantle proposals of Hanson (1977) and Wood (1979), various postulates of mesoscale lithologic heterogeneities (veins, pods, layers, plums) have been advanced. However, the issue remains contentious, and no smoking gun has survived scrutiny. Do the heterogeneities reflect large scale (tens of km) chemical variability in a lithologically homogeneous (peridotitic) mantle, or smaller mesoscale (less than a few kms) mafic layers or veins embedded in a peridotitic matrix (or both)? We argue that the high diffusion rates of helium preclude survival of He isotope heterogeneities on scales smaller than a few hundred meters, especially if they represent long term in-growth of 4He in the convecting mantle. Using a coupled He diffusion-production model, 1.5 Gy residence times, and a diffusion coefficient of 10-10 m2/sec (olivine, mid-upper mantle), 1 km slabs or 2.3 km cylinders will lose >80% of in-grown 4He. However, substantial 3/4He signatures may persist in slabs or be induced in adjacent mantle, depending on initial He, U and Th contents. We have modeled three cases of 1.5 Gy 3/4He equilibration between mantle domains: an ocean crust (OC) slab in depleted upper mantle (DMM) or in enriched mantle (BSE), and a BSE slab in DMM. For a 1 km OC slab in DMM (8 Ra today), the slab today will have 3/4He of only 3 Ra, and will have infected the surrounding mantle with 4He for >5 km on each side. The average 3/4He of this mixed zone will be <6 Ra even when sampled by melts over a total width of 20 km. For the case of a 1 km OC slab in BSE (50 Ra today), the slab will be 47 Ra today, almost fully equilibrated with ambient mantle. For the case of a 1 km BSE slab in DMM (8 Ra today), the slab will be 37 Ra today, and will have infected a mantle domain >16 km wide. Even with a 50 km melt sampling width, the average 3/4He will be >20 Ra. In essence, slabs may lose their He signature by diffusion, but it will remain recorded in the surrounding mantle; i.e. veins may run but they can't hide! For both enriched and depleted upper mantle slabs, sampled along a spreading ridge, the 3/4He variability on 10-20 km scale lengths would be easily observed; even massive along-axis melt mixing (50-100 km) would not hide these signatures. We have evaluated 3 extant ridge-crest data sets in this context (MAR 0-47S; EPR 19-23S; SWIR 16- 24E), with a view to defining scale-lengths of He isotope variability. The average 3/4He variability for these 3 areas is 0.47, 0.19 and 0.21 Ra (±1 sigma); a well-sampled sub-area on the MAR (25.7-26.5S) is 0.13 Ra. There is a monotonic variation along the SWIR, from 6.6 to 7.3 Ra; variability about a best fit line is 0.09 Ra (maximum deviation is only 0.20 Ra). At the smallest scale, a single 20 km EPR flow field shows similar variability (0.29 Ra) to the above examples. Since these ridges range from slow to very fast-spreading, the variability in size of along- axis magma chambers will lead inevitably to various scales of melt averaging. We conclude that these ridge areas are not sampling mantle that contains enriched veins or recycled oceanic crust slabs of any significant size. This is especially clear for the 500 km domain on the SWIR, where very small He variability is observed, superimposed on a large scale He gradient. In particular, the view of the upper mantle as a ubiquitous mixture of veins and depleted matrix, with MORB always representing an averaging of this mixture, appears untenable.
U21B-0408
What Comes Around Goes Around: Mantle Convection and the Meaning of Mantle Isochrons
The perceived significance of isotopic data arrays for oceanic basalts has long occupied a middle ground between the endmember interpretations of mixing and age. Brooks et al. [1] were the first to attach age significance to the correlations between parent-daughter ratios and daughter isotope ratios that are a regular feature of the geochemistry of OIB and MORB [e.g. 2,3]. Pseudo-isochrons derived from mixed mantle can still have age significance if the various packets of source material have been physically juxtaposed for long periods of time, yet the current paradigm has generally been to focus on daughter ratios alone, and to interpret their variations in terms of multi-component mixing. Numerous high-quality geochemical data sets now exist, and continute to be generated, for specific regions of OIB and MORB volcanism. In order to take the next step in a more accurate interpretation of this data, a forward model is needed that delimits the bounds of chemical variability and isotopic correlations expected to arise from the major processes operative during terrestrial mantle convection. In this talk, we will present the results of cylindrical 2D convection models with force-balanced plates [4] and examine specifically the roles of subduction and convective mixing of oceanic crust, with extraction of continental crust superimposed. In these models, melting occurs at divergent plate boundaries and geochemical evolution is recorded by millions of passive (harzburgite) and active (basalt) tracers that record the times, extents of melting, and extents of degassing and continent extraction at every melting event, allowing the geochemical evolution of any isotope system to be easily calculated (and recalculated) in a post-processing algorithm that operates on the tracer data independently from the dynamic calculations. We will explore the range of isotopic variability in these models as functions of partition coefficients, chemical density of basalt tracers, convective vigor, and history of continental crust extraction. In particular, these models reveal relationships between mantle isochron "ages" and true tracer ages that is not obtainable from statistical box- model calculations incorporating idealized mixing scenarios [e.g. 3-5]. References: [1] Brooks, C., Hart S.R., Hofmann A.W. & James D.E. (1976) EPSL 32, 51-61. [2] Albarede F. (2001) EPSL 189, 59-73. [3] Donnelly, K.E., Goldstein, S.L., Langmuir C.H. & Spiegelman, M. (2004) EPSL 226, 347-366. [4] Brandenburg, J.P. et al (2007) in review. [5] Rudge J.F. (2006) EPSL 249, 494-513.
U21B-0409
Temporal-Compositional Variation Over 100 - 102 yr in Primitive Basaltic Single Eruptions: Covarying Mixing and Melting
Compositional variation within primitive basaltic single eruptions is useful to isolate and describe short length- and time-scale phenomena in the mantle. In this study, we focus on multivariate data models, with the added constraint of temporal control, of the systematic temporal-compositional variation within primitive monogenetic single eruption sequences that erupt over a short 100-102 yr time scale. We use whole rock major element, trace element, and isotopic data from intraplate monogenetic eruption sequences in the Big Pine Volcanic Field, CA. To easily compare species with different variance and to make errors equivalent, we transform data to log mean centered concentrations. For the REEs, 78% and 9% of the total variance are described by the first and second principal components (PC), respectively. The first PC varies monotonically with time suggesting one dominant univariant reaction, and reflects the large magnitude decreases in LREEs during the eruption. The second PC is non-monotonic and reflects increasing HREE near the middle of the eruption. Pre-treating the data in the above or similar ways also allows us to relate PC scores to differentiation and homogenization equations in differential form. This enables comparison of the direction and magnitude of melt variation from a particular process to our data without having to invoke arbitrary source compositions. These models show (without appealing to isotopic variation) that melting (both dynamic and batch) and crystal fractionation cannot account for the variance structure of the single eruption temporal compositional trends. This is confirmed by the systematic isotopic depletion seen during the course of an eruption (87Sr/86Sr: 0.7063 to 0.7055; εNd: -3.4 to -1.1), which requires systematic mixing between two sources. We also see that the principal vector loading ratios for element pairs are proportional to their distribution coefficients, suggesting a melting relationship. Thus a coupled model is required in which F (melting) and X (mixing) co-vary. To further distinguish the cause of primitive monogenetic temporal compositional trends, we examine, in a similar manner, two main hypotheses of coupled melting and mixing: melt-rock interaction and melting of a lithologically heterogeneous source.
U21B-0410 INVITED
Noble Gases and Siderophile Elements in the Mantle: Unconventional Experimental Results and Their Implications for Chemical Geodynamics
Recent and ongoing experimental studies reveal unexpected behavior of noble gases and siderophile elements that may affect future geodynamic interpretations. Ar-uptake experiments on mantle minerals provide insight into Ar compatibility (solubility) and diffusivity in these phases. Contrary to expectation, solubility results suggest that Ar behaves as a compatible element during mantle melting, favoring residence in point defects in minerals over ‘escape' to the melt. In terms of diffusion, Ar is sufficiently mobile in olivine and orthopyroxene to ensure effective dispersal from relatively K-rich minerals on a geodynamically short time scale. Solid-liquid equilibration during melting of the MORB source is also likely; however, Ar diffusion is slow enough for disequilibrium to arise during phenocryst growth in a pre-eruption setting. The main implications of these results are that: 1) Ar degassing of the mantle through MORB volcanism may be an inefficient process; and 2) the existence of deeply-sequestered Ar (and other noble gases?) is plausible. New data on the behavior of siderophile elements (Os, Ir, Ru, Rh, Pt, Re, Au, W, Mo, Co, Cu and C) in polycrystalline MgO and synthetic peridotite reveal that these elements are highly mobile in the grain boundaries of mantle assemblages and analogs. W, Au and C have the highest diffusivities (10-8 to 10-7 m2/s); Ir, Ru, Re and Os are at the low end of the range (~10-12 to 10-11 m2/s). There is no discernible correlation between grain-boundary diffusivity and any particular property of the atoms or elements. These new data imply that Earth's outer core may ‘leak' siderophile elements into the lower mantle over distances exceeding 100 km in 4 GYr for W, Au and C. Although not significant in itself as a whole-mantle transport process, grain boundary diffusion appears capable of ‘contaminating' the lower mantle over a sufficient distance to enable entrainment of a core signature into plume- or convective mantle flow. If the outer core contains C, our data suggest that there may be a significant core-to-mantle flux of this element.
U21B-0411
Diamonds From Dachine Komatiites (French Guiana) : Obvious Link to Enigmatic Carbonados and a Possible Primordial Heterogeneity Below the Guiana Shield.
We studied 150 diamonds from the Dachine area (French Guiana). These diamonds are the only samples recovered so far from a volcaniclastic komatiite [1]. The samples are monocrystalline (i.e. non polycristalline) macrodiamonds (> 500 microns) mostly (>95%) irregular in shape with resobred octahedron and cubes being occasionnally recognised. No syngenetic inclusion has been identified yet. Most (>90%) samples are nitrogen free (Type II) and the few samples containing nitrogen (Type I) are characterised by low aggregation states. Their carbon isotope compositions range from -31.9 to +0.15‰ with an average of - 24.4°‰ The observed general 13C-depletion among diamonds from Dachine including both octahedron and cuboids (and their respective resorbed forms), among both Type II and Type I diamonds demonstrates that an entire diamond population with 13C-depletion can originate from the mantle. Furthermore, the nitrogen speciation (that reflects the time-temperature history of diamond) is little advanced (Type Ib-IaA diamonds). This suggests that Dachine diamonds spent virtually no time in the mantle and thus most likely precipitated from their host komatiite. Our results have two implications, the long-standing argument that carbonados could not form in the mantle as inferred from their distinct carbon-isotopes compositions is not valid anymore. The similarties between carbonado and Dachine diamonds being striking from a carbon-isotope but also from a nitrogen aggregation point of view. The second implication concerns the origin of such a carbon isotope distribution. So far we favor a primordal heterogeneity below the Guiana Shield rather than the contribution of subducted material. A result that can be checked further using radiogenic isotope geochemistry. [1] Capdevila et al., Nature 399, 456-458 (1999)
U21B-0412
Volcanic eruptions, global change and evolution of species
Our group proposed in 1986 that the Deccan traps of India had been co-eval with the Cretaceous-Tertiary mass extinction, had lasted less than 800 kyears, straddling the KT. In 1995, Bhandari and colleagues showed that the iridium level marking the Chicxulub impact could be found in the Deccan and was sandwiched between flows, demonstrating that they were co-eval events but that volcanism had started first and could not be a consequence of impact. Since then, many groups have contributed to dating more precisely continental flood basalts around the world and found that there was almost a one to one correspondance between flood basalts and mass extinctions (but not impacts, except for the KT) : for a recent review see Courtillot and Renne (2003). In recent years, a number of significant advances have been made. It has been shown in the case of the historically large but geologically very small Icelandic Laki eruption of 1783 that such eruptions could inject large amounts of sulfate aerosols all the way to the stratosphere and have a global impact on climate : therefore, effusive basaltic volcanism on a large scale could alter climate (our group with Frédéric Fluteau and Anne-Lise Chenet and Steve Self with Thor Thordarson). Petrologic, volcanological, paleomagnetic, paleontological and geochronologic studies of the entire 3000m thick Deccan pile have been resumed (mainly our group, that of Self with Mike Widdowson and Anne Jay, and that of Gerta Keller). The result is that the thick lava pile actually erupted in a relatively small number of gigantic pulses (with mega-flows up to 10000 km3 in volume having erupted in decades !). Field evidence has been given that flows could extend over almost 1000km, and paleontological and K-Ar dating now reveals a history of mainly two mega-pulses having occurred, one just prior to the KT the second somewhat afterwards explaining the long delay of recovery of species from the catastrophe. There is little doubt now that the KT catastrophe would have occurred even if the impact (which did have a significant additionnal effect) had not struck, and that the impact could not have generated a mass extinction if volcanism had not already been going on. Finally, global change during these periods of anomalous volcanism may provide interesting benchmarks for modelling of more current potential global change, and geologists be a major help to climate modelers.
U21B-0413
Chemical Geodynamics: a perspective from the Continental Crust
Chemical geodynamics has been applied both to the differentiation and evolution of the continental crust, and to its recycling back into the Earth's mantle. Estimates of the residence time of material in different reservoirs within the continental crust constrain models for the evolution of the crust and the volumes of material recycled. The present data indicate that the generation of new crust in the Archaean and Early Proterozoic may have occurred in restricted periods of rapid crust generation. If correct, the rates at which new crust is reworked into the sedimentary reservoir are much slower than had been previously envisaged. More widely, if the average residence times in the upper crust are typically of the order of 2 Ga, those in the lower crust are ~0.4-0.6 Ga. The relative volumes of recycled upper and lower crustal material are therefore better constrained, and the effects of the recycling of both upper and lower crust compositions on the trace element and radiogenic isotope ratios of mantle sampled by oceanic basalts can be further explored. For example, the differentiation of basaltic magma to the composition of the bulk continental crust yields large volumes of residue with distinctive Rb/Sr, Sr/Nd, Sm/Nd, Lu/Hf and U/Pb ratios delaminated into the mantle after relatively short periods in the crust. Current issues that can now be investigated in more detail include the extent that the pulses of crust generation are global, rather than of slightly different ages in different continental blocks, the Lu/Hf ratio, and hence by inference the composition of early crust from in situ Pb and Hf isotope data on old zircons, and the separate evolution of the sedimentary and igneous portions of the continental crust.
U21B-0414 INVITED
The role of the lower continental crust in chemical geodynamics
Recycling of continental crust into the mantle is generally conceptualized in terms of subduction of terrigenous sedimentary rocks produced by weathering and erosion of the upper continental crust. However, another form of crustal recycling is increasingly recognized as adding chemical diversity to the convecting mantle: recycling of lower continental crust. Lower crustal recycling can occur by several processes, including density foundering and subduction erosion. Given the compositional diversity of the continental lower crust, the chemical signature of such recycled material can be highly variable. For example, the radiogenic isotopic signature will depend on the age of the lower crust and its tectono-magmatic history. Newly formed, high density crystal cumulates foundered from colliding arcs may have Nd, Hf, Sr and Pb isotopic signatures similar to mantle-derived arc melts. But if these packages of recycled lower crust remain intact in the convecting mantle, they may evolve to radiogenic Nd and Hf (due to igneous cumulate or restitic garnet), radiogenic Os (due to high Re/Os) and unradiogenic Pb isotope compositions. In contrast, recycling of ancient lower crust should be characterized by evolved Nd, Hf and Os isotopic compositions, but may have variable Sr (depending on whether and when Rb was depleted) and Pb isotopic compositions (depending on the timing of the last orogenic event that homogenized Pb isotopes). Examples of intraplate magmas that derive from foundered ancient lower continental crust are found in eastern China, where thinning and destabilization of the Archean craton occurred coincident with a magmatic flare-up in the Mesozoic. Some of these Mesozoic magmas contain evidence for lower crustal foundering. For example, high Mg# andesites derive from melts of foundered lower crustal eclogite that interacted with mantle peridotite, whereas some high Mg basalts and alkaline picrites derive from a mantle source that was hybridized by silicic melts derived from foundered lower crustal eclogite. This, and other examples of lower crustal recycling point to an additional source of mantle heterogeneity derived from the continental crust.
U21B-0415
Carbonate-rich Sediments in the Hawaiian Plume: Evidence from Mahukona Volcano
Hawaiian shield lavas exhibit considerable heterogeneity in Sr-Nd-Hf-Pb isotope space, as well as in their major and trace element abundances. Some of these geochemical features are attributed to recycled oceanic crust, including sediments. Here we report major and trace element concentrations, and Sr-Nd-Hf-Pb isotopic compositions for shield lavas from Mahukona Seamount, Hawaii. Using these data, we identify a carbonate-rich sedimentary source component in the Mahukona lavas. Specifically, within Mahukona lavas, Nd-Hf-Pb isotopic ratios are highly correlated with each other, while 87Sr/86Sr is decoupled. Since the isotopic compositions were obtained on strongly acid-leached residues, the discordance of 87Sr/86Sr is not inferred to be alteration-related. Moreover, 87Sr/86Sr is positively correlated with Sr/Ce and negatively correlated with Rb/Sr and Ba/Sr. Because the Mahukona lavas are tholeiitic to transitional basalts, their Sr/Ce, Rb/Sr and Ba/Sr reflect their source values. We infer that one of the Mahukona source components has high Sr/Ce (>16) and 87Sr/86Sr (>0.7038), and low Rb/Sr (<0.015) and Ba/Sr (<0.15). Such a composition cannot be generated through mantle metasomatism, because high 87Sr/86Sr requires high Rb/Sr. We rather propose recycled ancient carbonate-rich sediment as a possible component in the Mahukona source. Carbonate-rich sediments, in general, have high Sr/Ce, and low Rb/Sr and Ba/Sr. Furthermore, ancient carbonate-rich sediments would have inherited their 87Sr/86Sr from seawater, which had 87Sr/86Sr greater than 0.7040 at 2 Ga. Since a carbonate-rich sedimentary component has very high Sr/X (X=Nd, Hf, Pb), sampling it would affect only 87Sr/86Sr and leave the Nd-Hf-Pb isotopic ratios unchanged.
U21B-0416
Revised estimate of radioactive energy in the Earth
The radioactive elements, K, Th and U, play a paramount role in chemical geodynamics by providing a significant proportion of the energy that drives mantle convection. A major unknown is the absolute abundance and distribution of these elements in the mantle. For example, a mantle with a barren upper mantle and a radioactively hot core-mantle boundary (CMB) implies a significantly different mode of mantle convection than typically invoked for a more internally-heated mantle. The canonical K/U ratio of the silicate Earth (12,700), which is based on N-MORB samples, has been challenged in recent years by alternative models suggesting lower values for the bulk silicate Earth. Support for a lower K/U value for the silicate Earth comes from reconciling the missing Ar-40 problem, a re-examination of the continental crust composition, and evaluating the relative composition and contributions of different components in the mantle. In addition, a commonly held view is that the abundances of the most-depleted elements in the Depleted Mantle (MORB-source), the highly incompatible elements, are kinetically controlled by fluxes from sediment recycling and/or plume additions. We have analyzed >50 MORB, spanning a range from N- (normal) to (enriched) E-types from the Indian, Pacific and Atlantic oceans, and report an average K/U of 18,500 and Th/U of 3.1. In addition, we also observe a weak correlation of decreasing K/U with higher 6/4Pb isotope ratios, with the E-type MORB (K/Ti>0.14) having, on average, lower K/U values. Our findings are inconsistent with these incompatible elements being controlled by additions from either a continental or plume flux, both of which are believed to have lower K/U and higher Th/U values, and indicate that the Depleted Mantle component has an even higher K/U value. We will present the implications of these data with (1) revised compositional model (and uncertainties) of the DM and the silicate Earth, (2) a model for the distribution of heat producing elements in the mantle and (3) constraints on the proportion of recycled crustal and plume components in the DM.
U21B-0417
Global Variations in Abyssal Peridotite Compositions II: What Determines the Major Element MORB Composition
Large scale global variability of abyssal basalts and peridotites was discovered over 20 years ago, with increasingly depleted peridotites correlating to increasingly enriched MORB that carries a major element signature of higher degrees of mantle melting (Dick et al., 1984; Klein & Langmuir, 1987). While time and further analysis shows that these correlations do not hold for the Pacific, they remain largely valid for the Atlantic, Arctic and Indian Oceans. Given the large variability in abyssal peridotite compositions at local and regional scales this is remarkable. Nearly the full range of mantle major, trace and isotopic composition can be found in peridotites dredged from a single fracture zone or ridge segment. Detailed analysis of large peridotite suites from the Gakkel Ridge (Dick, Hellebrand & Snow, unpub. data), and in the central Atlantic from 14° to 16°N also show large magma starved regions with highly depleted mantle compositions that must be inherited from prior melting events. Our recent investigations of SW Indian Ridge Peridotites also lead us to believe that much of the local peridotite variability at the scale of a single dredge could be inherited from an early melting event. All this leaves one wondering what, if anything the "global" (actually regional) variability of MORB and abyssal peridotites defined by Dick et al. (1984) and Klein and Langmuir (1987) means. For abyssal peridotites the correlations only show up when the mineral and modal compositions of large numbers of samples are averaged by locality. Moreover, the sample source on which the correlations are based was almost exclusively fracture zones. Thus, the samples come from the same geodynamic environment, with the same P-T-Melt Flux history, and therefore do not reflect the full variability produced beneath a ridge segment, but only the melting history at the distal end of a magmatic segment. We would contend, therefore, that the correlations found by the cited authors remain valid and show that MORB aggregates from large regions and therefore the melt aggregation process averages out the variability seen in the source at segment scales. As the previously analyzed peridotites are from fracture zone walls 0.5 to 14 m.y. old, and the ‘spatially associated basalts' are largely from the modern ridge axis, this argues for a long-term stability in magma composition and therefore mantle composition as well. There remains the question, however, to what degree the lateral variability of MORB major element compositions reflect varying degrees of mantle melting of a uniform source, and to what degree this is a function of varying mantle composition. It is certainly the case that trace and isotopic composition vary, and we suggest that it is equally likely that mantle major element composition varies significantly as well.
U21B-0418
Recycled MORB in the Convecting Upper Mantle? Evidence From Os, Sr and Nd Isotopes in Orogenic Peridotites and Pyroxenites From the Totalp Massif, Eastern Switzerland
The presence of recycled mafic material in the MORB source region remains a subject of debate. Some orogenic peridotite massifs include materials from the deep lithosphere and may thus provide information on the asthenospheric sources of melts. The Totalp ultramafic massif in eastern Switzerland, which formed part of the Jurassic ocean floor in the ultraslow spreading Piedmont-Liguria ocean basin, is a site, where the deep lithosphere may be studied. Serpentinized fertile spinel lherzolites are interlayered with folded and stretched websterites and modally layered pyroxenites. In addition, igneous textures and chemical evidence suggest widespread influx of presumably asthenosphere-derived melts into the peridotite via websteritic layers and limited temperature contrasts between melts and host rock. Bulk rock rare earth element patterns in both pyroxenites and lherzolites are depleted in the LREE, suggesting a MORB-like composition of the precurser melt for the residual pyroxenites. Measured Sr and Nd isotopic compositions in clinopyroxene separates from lherzolites and pyroxenites fall into the depleted sector of modern MORB compositions with 87Sr/86Sr between 0.7022 and 0.7036, and εNd ranges between +8.3 and +12.7. Initial isotopic compositions at 160 Ma yield a similar picture. In contrast, measured and initial 187Os/188Os in the lherzolites range from subchondritic to suprachondritic, with radiogenic lherzolite compositions that are clearly affected by melt influx from the pyroxenites, as the latter have very radiogenic measured and initial 187Os/188Os. No correlation exists between Os isotopes and the lithophile radiogenic isotopes. In current models advocating recycled materials in the mantle, the combination of radiogenic Os, enhanced Re contents, but MORB-like Sr and Nd isotopic compositions can only be reconciled with a component from recycled mafic oceanic crust. The enhanced Re and Pd contents of some Totalp pyroxenites, however, indicate that the pyroxenites cannot be simple residues of molten recycled MORB. More likely, the pyroxenites may be precipitates from percolating melts derived from the sublithospheric mantle. Thus, the required recycled crust component may have resided in the convecting mantle.
U21B-0419
Strongly foliated garnetiferous amphibolite clasts in ophiolitic melanges, Yarlung Zangbo Suture Zone, Tibet; Early Cretaceous disruption of a back-arc basin?
Metre to decameter-size clasts of amphibolite are found embedded in ophiolitic melanges underlying the Yarlung Zangbo Suture Zone Ophiolites, South Tibet, China. These ophiolites and melanges occur at the limit between Indian and Tibetan-derived rocks and represent remnants of an Early Cretaceous intraoceanic supra-subduction zone domain, the Neo-Tethys. In the Saga-Dazuka segment (500 km along-strike), we discovered new occurrences of strongly foliated amphibolites found as clasts in the ophiolitic melange. In garnet-free samples, hornblende is green-blue magnesio-hornblende and cpx is low-Al diopside. In garnet- bearing samples, garnet is almandine with a strong pyrope component (up to 30 mol%) whereas coexisting hornblende is brown Ti-rich tschermakite and clinopyroxene is Al-diopside. Plagioclase composition was ubiquitously shifted to albite during a late metasomatic event. Geochemistry of these rocks indicates that their igneous protoliths crystallized from a slightly differentiated tholeiitic basaltic liquid that did not undergo major fractionation. Trace element patterns reveal geochemical characteristics identical to those of the overlying ophiolitic crust. These are 1) trace element abundances similar to that of N-MORBs or BABBs, 2) a slight depletion of LREE and 3) a moderate to strong Ta-Nb negative anomaly and a slight Ti anomaly. Such characteristics suggest genesis over a spreading center close to a subduction zone, possibly a back-arc basin. Step-heating Ar/Ar plateau ages were obtained from hornblende separates. All ages fall in the range of 123-128 Ma, overlapping the crystallization ages from the overlying ophiolite (126-131 Ma). Pseudosections were built with the THERMOCALC software in the system NCFMASH. Results indicate that the observed assemblage Hb+Pl+Gt+Cpx is stable over a wide range of P-T conditions, between 10-18 kbars and at more than 800°C. Measured mineral modes and solid solution compositions were successfully modeled, indicating equilibrium between 11-13 kbars and 825-850°C, corresponding to high-P granulite facies conditions. In a general way, the geochemistry of the strongly foliated amphibolite clasts suggests that their igneous protolith probably crystallized within the same supra-subduction zone as the crustal rocks from the overlying ophiolite. Then some of these rocks were entrained to mantle depth and were rapidly exhumed, most likely along a lithospheric scale thrust fault underneath the ophiolite. This event corresponds with the end of magmatic activity within the ophiolitic crust and mantle and could be regarded as the inception of a subduction plane at the spreading ridge of a back-arc basin. The whole package was later on obducted over the Indian passive margin, at about 70 Ma. Such a model suggests that closure of the oceanic domain separating India from Eurasia implied disruption of at least one arc-back-arc system, thus requiring at least one early intraoceanic collision or major plate movement reorganization prior to the Late Cretaceous obduction.
U21B-0420
Geology and Geochemistry of the Saga and Sangsang Ophiolite, Yarlung Zangbo Suture Zone, Southern Tibet
The Saga and Sangsang ophiolites belong to the E–W-trending Early Cretaceous ophiolite belt of the Yarlung Zangbo Suture Zone (YZSZ), resulting from the closure of the Neo-Tethys Ocean during the collision between the Indian and Eurasian plates. These ophiolitic massifs crop out west of Xigaze and presumably represent a western extention of the central portion of the YZSZ ophiolite belt. The Saga massif comprises fresh coarse- grained mantle tectonite (lherzolite), an ophiolite mélange (harzburgite ± dunite ± garnet- clinopyroxene -bearing amphibolite), a stripe of metamorphosed upper crust rocks and a sequence of uppermost crust rocks, such as chert, basaltic lavas and diabase. Sangsang ophiolite consists of harzburgitic ophiolite mélange and fine-grained mantle tectonites. Spinel chemistry suggests that Saga massif peridotites have abyssal peridotite affinities and have undergone only apparent low degrees of partial melting (5-12%) while Sangsang peridotites have both abyssal peridotite and fore-arc peridotite affinities and have underwent higher apparent degrees of partial melting (17-30%). The Ol-Sp equilibrum for Saga and Sangsang peridotites therefore suggest that Saga peridotites have a deepest mantle provenance (>20 kbar) than Sangsang peridotites (<15 kbar) and that the composition of Saga peridotites is similar to the composition of pre-oceanic peridotites while peridotites from the Sangsang massif are alike abyssal and subduction margin peridotites. Peridotites from both massifs show variable degrees of serpentinization (LOI:0.7-13.6%). Their Mg# vary between 0.89 and 0.91. Saga massif peridotites show distinctive flat REE-patterns with La/YbN ratios close to 1. Thus, the abundances of REE elements in these rocks suggest that their composition is similar to a chondritic mantle (or a refertilized mantle with chondritic liquids). The uppermost crust and metamorphosed upper crust rocks compositions correspond to basalt and basaltic andesite. They are slightly depleted in highly incompatible elements (LREE) with respect to the less incompatible elements (HREE), with La/YbN between 0.5 and 0.8 and with a slight negative Ta anomaly suggesting the presence of a subduction component. The abundances of incompatible elements in these mafic rocks are therefore quite similar to N-MORB or BABB. On the basis of along-strike correlation, age relationships and geochemical characteristics, we suggest that the Saga and Sangsang ophiolites belong to the same intraoceanic suprasubduction zone segment as others ophiolitic massifs to the east (Xigaze ophiolites). However, the geochemistry of mantle rocks from these two massifs is quite different compared to each other and compared to other eastern Yarlung Zangbo ophiolites suggesting contrasted petrogenetic histories. The Saga and Sangsang massifs may thus represent different portions of an arc – back-arc system configuration.
U21B-0421
Influence of major element disequilibrium on mantle structure
At mid-ocean ridges, partial melting of a presumably homogeneous and equilibrated pyrolitic source generates a basaltic crust and leaves behind its depleted complement, harzburgite. The oceanic lithosphere that subducts into the mantle is thus physically and chemically layered. During subduction we expect basalt to separate from harzburgite and the rest of the oceanic lithosphere. Hence, over convective and diffusion time-scales, do basalt and harzburgite re-equilibrate chemically as pyrolite? Most mineralogical models, upon which comparisons to seismology are based, view the mantle as homogeneous and pyrolitic or chemically stratified with homogeneous and equilibrated compositions in each layer. Petrological experiments have shown that a homogeneous pyrolite source region explains MORB and the seismic velocity profile of upper mantle and transition zone to first order. However, this view appears to violate the dynamical constraints given the low chemical diffusivity for mantle materials (10-14-10-16 cm2s-1) in the solid-state, ignoring the effects of fluids and partial melting. Allègre and Turcotte (1986) suggested a mechanically mixed mantle, a marble cake structure in which subducted oceanic lithosphere is deformed into pervasive, narrow pyroxenite veins. Computer simulations suggest a heterogeneous mantle made of a mechanical mixture of basalt and harzburgite, in which pools of basalt may accumulate at the bottom. A stirring time of the mantle between 250 and 750 Myr limits the mixing, stretching and folding of heterogeneity in the mantle. Thus, it seems implausible to equilibrate basalt and harzburgite into pyrolite with a fine (0.1–10 m) stratification given typical chemical diffusivities. We demonstrate that, even with identical bulk compositions, an equilibrium assemblage (EA) along the basalt- harzburgite join and a mechanical mixture of basalt and harzburgite in perfect disequilibrium (MM) have different phase equilibria and therefore different seismic velocities. We compute the seismic velocities of EA and MM using a previously developed self-consistent thermodynamic model and explore the effects of bulk composition (in terms of basalt depletion) and temperature. For MM, VS in the transition zone is higher and increases more rapidly with depth and is virtually insensitive to basalt fraction, while VS decreases for EA. For both EA and MM increasing potential temperature from 1400K to 1800K yields a deeper 410-km and a shallower 660-km discontinuity. The radial gradient of the velocity between discontinuities decreases with increasing potential temperature. We find that the magnitude of the 520- km discontinuity depends strongly on temperature, which may explain lateral variations in its seismic detection. Both MM and EA feature "double-step" discontinuities in the range of 660-750 km due to the ringwoodite- perovskite transition and the gradual dissolution of garnet into perovskite between 665 km and ~725 km depth. Both MM and EA have lower velocities than published radial seismological models, a discrepancy that increases with increasing depth from 400 to 740 km depth. This suggests the presence of a radial gradient in bulk composition in the mantle, a sub-adiabatic geotherm, or both. We explore the dynamical consequences for Earth's gravitational field and plate motions by constructing a velocity-density scaling that takes into account compositional and thermal effects as well the potential for phase transformations to induce lateral heterogeneity in seismic velocities.
U21B-0422
Allegre's Lead Paradox Revisited
Allegre (1969), using a generalized Concordia plot, was first to note that in a 4.55 Ga old Earth, Pb has been removed from the mantle in preference to U, even though magmatism tends to do the opposite. He noted that this "contradiction" might require a separate reservoir where the missing lead is stored. This "contradiction", more commonly expressed by the conventional Holmes-Houtermans diagram, has become known as the "lead paradox." Several models have been proposed to resolve it, ranging from late "core pumping" of Pb (as suggested by Allègre) to Pb storage in the mantle transition zone (Murphy et al. 2003) or in ancient parts of the lower continental crust. The idea of late core pumping has recently been revived by Wood & Halliday (2005) who suggested that Pb was sequestered through late sulfide segregation into the core. Here we propose that crystallization of Ca-perovskite, accompanied by segregation of a dense silicate melt toward the core-mantle boundary, can account for the apparently elevated U/Pb ratio of the accessible silicate Earth, particularly if the partition coefficient for U and Th in Ca-perovskite is as high as 400 as suggested by Corgne and Wood (2005). Such a dense liquid is the inferred consequence of the measured crossover of melting temperatures of silicate perovskite and ferro-periclase at about 1200 km depth, and the predicted Fe-rich eutectic and low melting temperatures in the lowermost mantle (Boehler, 2000). In addition, lower-mantle melt segregation with residual Ca-perovskite will cause a decrease in Nb/Ta from the primitive (chondritic) value in the accessible mantle, another, more recently discovered puzzle of mantle geochemistry. Downward segregation of a dense melt fraction and final solidification of the lowermost mantle may have been a slow process requiring more than 100 Ma, and involving a substantial fraction of the mantle. We suggest that this process served to stabilize the D'' reservoir storing solar noble gases (Tolstikhin & Hofmann, 2005) and subchondritic 142Nd/144Nd ratios (Boyet & Carlson, 2005), which these authors have ascribed to subduction of a primordial crust at an even earlier stage. Allegre, C.A. (1969) Earth Planet. Sci. Lett. 5, 261-269. Boehler, R. (2000) Reviews Geophys. 38, 221-235. Murphy, D.T. et al. (2003) J. Petrol. 44, 39-53. Tolstikhin, I.N. & Hofmann, A.W. (2005) Phys. Earth Planet. Int. 148, 109-130. Boyet, M. & Carlson, R.W. (2005) Science 309, 576-581. Corgne, A & Wood, B.J. (2005) Contrib. Min. Pet. 149, 85-97.
U21B-0423
Putting the Dynamics in Chemical Geodynamics
An outstanding goal for both geochemists and geophysicists is to understand how to use the wide range of proxy geochemical (and geophysical) data to make useful inferences about the current and past dynamics of the planet. To relate data to dynamics, however, requires models that include the fundamental processes that affect chemical variability: i.e. source heterogeneity, chemical fractionation (melting/reactions), chemical transport and mixing. In particular, most models of chemical evolution do not include explicit fluid or magma transport and questions remain as to how much observed chemical variability can be attributed to magma dynamics. We discuss recent developments and models that suggest that at least some of the observed variability arises from transport processes. Driven by field observations, experiments and computational models, there is an emerging picture of partially molten regions as highly localized, channelized plumbing systems. Computations suggest that melt localization can arise from both chemical/physical and purely mechanical instabilities and can provide non-trivial mixing pathways through the mantle. Questions remain as to how the different instabilities interact and which may be dominant in the mantle. Regardless, a highly localized melt transport system can lead to significant trace element (and U-series) variability and fractionation even for a homogeneous source. Recent work extends these results to consider the interaction of a channelized melt system with a heterogeneous source and suggests that small scale spatial variations in partitioning can lead to significant scatter in the ratios of highly incompatible elements (Fang, Spiegelman & Kelemen). Current work is extending these approaches to try to understand the variability of major elements and reaction in open systems (Collier, Kelemen & Spiegelman) Looking forward, the integration of magma dynamics and small-scale localization into global mantle dynamics presents a major scientific and computational challenge. A key addition to magma dynamics models will be the consistent coupling of fluid/solid mechanics with thermodynamics to resolve both melting and crystallization of magmas (e.g. see Katz U06, this meeting). These models will also require advanced methods for the efficient solution of highly multi-scale problems. The CIG Magma dynamics project is actively pursuing these computational objectives with the long-term goal of developing true dynamic geochemistry models whose output can be directly compared to observations.
U21B-0424
Hydrogen and Mineral Elasticity: Modelling Hydration in the Earth's Interior
Hydrogen is perhaps the most poorly constrained compositional variable in the Earth's interior. Hydrogen can enter all of the nominally anhydrous phases of the interior in trace, minor, or even major amounts. In order to include the effects of hydration in seismic and geodynamic models of the Earth's mantle it may be useful to identify a relatively simple relationship between elastic properties, bulk composition including H content, and crystal structure. We have identified a relation that predicts bulk modulus based on Birch's law for both hydrous and anhydrous silicate and oxide phases. The relation also holds for hydration of nominally anhydrous phases provided the volume of hydration is known. The resulting regression equation is: K = 2747 (ρ/M) – 308 (GPa) with a correlation coefficient of 98.5% , where ρ is density and M is mean atomic weight excluding hydrogen. A similar empirical relation predicts shear moduli with a correlation coefficient of 90% allowing rough modelling of the effects of hydration on Vp and Vs.
U21B-0425
Geochemical Potamology
The surface of our planet has been reworked by erosion and weathering processes throughout its geological history. Erosion processes are so fast that only 10's of millions years are necessary to remove major reliefs formed by plate tectonics. The material of the surface of the Earth has been recycled many times and the only visible remains of the ancient continental crust are in the sediments, in the atmosphere, or dissolved in the ocean. The chemical action of water on rocks and the associated physical gravity-driven processes are the key features of the « weathering engine ». Rivers are messengers from the surface of the Earth and integrate weathering processes at various scales of time and space. They provide a unique opportunity to have average information about the chemical interactions between surface waters and rocks, about their kinetics and about the major geochemical fractionations created by this major geological process. The study of rivers has been focused on small- and large-scale approaches, both giving different information. Although the sampling of small river basin allows us to simplify nature and separate the potential controlling parameters by fixing them (climate, vegetation, type of rock, slopes…), the study of large rivers give global information about the composition of sediments, oceanic budgets, the significance of isotopic proxies. A geochemical potamology has developed over the last decades and we now more fully understand the critical questions that have to be solved. Numerous isotopic and elemental proxies have been developed to shed light on weathering processes, deciphering their control and calculate rates and timescales of erosion based on river products. A coupling between physical and chemical processes of erosion is now clear at different scales but probably the most important factor in controlling weathering style and rates are the protons. Low pH solutions are much more aggressive than neutral waters, as confirmed by experimental studies. On Earth there are a few proton donors (acids) that can contribute to feed the weathering engine : carbonic acid (the form of dissolution of atmospheric CO2), sulfuric acid , nitric and chlorhydric acids. Only carbonic and sulfuric acids have climatic implications by changing the long-term composition of the atmosphere and of the ocean. Because plants essentially control the production of CO2, the influence of ecosystem dynamics on chemical weathering is probably very important. In parallel, an international network of drainage basin is developing to select a panel of "node" river sites on which isotopic and geochemical tools will be applied. In a changing world, rivers are acting as sensors of environmental changes and their long-term monitoring appears to be essential.
U21B-0426
Subadiabatic geotherm and compositional gradient in the mantle
We further explore the properties of a mantle model consisting of a mechanical mixture of basalt and harzburgite with varying basalt fraction. We have found that such a mechanical mixture is faster and better explains 1-D seismic profiles than an equilibrated pyrolite of the same bulk composition, particularly in the transition zone [Xu et al., 2007]. We now invert for the geotherm that would best fit seismic profiles, given our bulk compositional models. We find that we are not able simultaneously to fit a range of seismological 1D VS and VP models by varying the temperature alone. Hence we explore the simultaneous optimal solution for temperature and basalt fraction throughout the whole mantle. Our results indicate a radial gradient in bulk composition throughout the mantle: basalt depletion in the upper mantle and enrichment in the lower mantle. The transition zone is insensitive to basalt fraction for the mechanical mixture, which gives us an independent thermometer for this region. The comparison between model and observed seismic structure suggests significant subadiabaticity. For the lower mantle, the basalt fraction monotonically increases with depth: from 20% below the 660 km discontinuity and 80% near the core-mantle boundary. Remarkably, this is in agreement with the basalt enrichment expected in the dynamical calculations of Nakagawa and Buffett [2005]. A remaining uncertainty is the role on the phase equilibria and physical properties of high-pressure basaltic phases stabilized by Na, such as calcium ferrite. We test the dynamical consequences of our results by constructing a velocity-density scaling, which we use to predict the Earth's geoid and plate motions.
U21B-0427
In Search of the Continental Mantle End Member
The advent of Chemical Geodynamics some 25 years ago spawned a rush of radiogenic isotope tracer studies on a wide variety of direct mantle samples from the lithosphere beneath continents. In particular, Rb-Sr, Sm-Nd, Lu-Hf, and Re-Os isotope studies have been carried out on diverse mantle rocks and minerals including peridotite and eclogite xenoliths, macrocrysts, and silicate and sulfide inclusions in diamonds from kimberlites erupted through Archean cratons. These studies have contributed significantly to current understanding of old thick cold lithospheric mantle keels coupled to and translating with the overlying continental crust. The general consensus is that old continental mantle keels are characterized by unradiogenic Os and Nd, and highly radiogenic Sr isotope signatures as a consequence of early melting and subsequent metasomatism of residual harzburgitic compositions. However, the availability of such material for delamination and eventual incorporation into the source of ocean island volcanics remains controversial. Inclusion bearing diamonds have come to be appreciated as the ultimate time capsules from old continental mantle by virtue of negligible diffusion rates of the relevant elements through diamond stored in the keel on a billion-year timescale. Our current work is focused on harzburgitic diamonds and their potential precursors from the 0.52 Ga Venetia kimberlite, Limpopo Belt, Kaapvaal-Zimbabwe craton. Peridotite xenoliths from this locality include the least radiogenic Os isotope composition of any known terrestrial rock, in particular a garnet lherzolite with an Os model (TRD) age of 3.7 Ga, effectively a minimum age for the keel to this continental nucleus. Yet, Venetia harzburgitic garnet bearing diamonds yield a diamond crystallization (or recrystallization) age of around 2.0 Ga though with an unradiogenic initial Nd (and radiogenic initial Sr) that requires a >3 Ga lithospheric mantle precursor. A complementary Re-Os study of available sulfide inclusions from Venetia diamonds is currently in progress. These features are reminiscent of the characteristics of lherzolitic diamonds from the 1.2 Ga Premier kimberlite, that mark the major Proterozoic modification of the central Kaapvaal craton in association with emplacement of the massive 2.05 Ga Bushveld complex. They also resemble those of harzburgitic diamonds from the 0.37 Ga Udachnaya kimberlite, Siberian craton, where both ca. 3.5 and 2.0 Ga generations of harzburgitic to lherzolitic diamonds have been identified. A consistent picture is emerging of episodic diamond formation associated with Archean craton formation, stabilization and modification that highlights the continental mantle keel as a potential reservoir of unradiogenic Nd and Os, and radiogenic Sr that would be available for recycling into the convecting mantle if it were to detach from the continent.