HR: 13:55h
AN: V52D-02 [PDF]
TI: Trace Element Composition of the Depleted Upper Mantle
AU: * Workman, R K
EM: rworkman@whoi.edu
AF: Woods Hole Oceanographic Institution, 266 Woods Hole Rd., Woods Hole, MA 02543 United States
AU: Hart, S R
EM: shart@whoi.edu
AF: Woods Hole Oceanographic Institution, 266 Woods Hole Rd., Woods Hole, MA 02543 United States
AB:
The trace element composition of the depleted upper mantle (DMM) is a crucial parameter in modelling the generation of
mid-ocean ridge basalts (MORBs), calculating the crust-mantle mass balance, and establishing the chemical and thermal
evolution of the stratified Earth. Although MORBs have some degree of heterogeneity in radiogenic isotope ratios
(Sr-Nd-Pb-Hf), they are focused within a small range of values relative to ocean island basalts and are, with few exceptions,
depleted from bulk earth values in $^{87}$Sr/$^{86}$Sr, $^{143}$Nd/$^{144}$Nd and $^{176}$Hf/$^{177}$Hf. We interpret DMM to
be a reservoir residual to variable degrees of melt depletion from a primitive mantle, and offer a method of estimating the
trace element composition of a depleted mantle given any degree of depletion at any time. This method is based on the
following: (1) Trace element compositions of clinopyroxenes from a global suite of abyssal peridotites with known modal
abundances (compiled from the literature), (2) Parent/daughter ratios needed to evolve to today's average isotopic ratios of
DMM based on segregation from bulk silicate Earth (BSE) at 1.8 Ga and (3) Canonical trace element ratios of MORBs.
Trace element compositions (Sr, Zr, Ce, Nd, Sm, Eu, Dy, Er and Yb) of cpx grains in abyssal peridotites have been converted
into bulk rock compositions using modal abundances and mineral/mineral partition coefficients. We have excluded samples
containing more than 1% modal plagioclase. The relationship between the concentrations of two elements, A and B, in the
residue of fractional melting can be linearized as follows:
\[ \textrm{ln}(C^{A}_{s}) = R \textrm{ln}(C^{B}_{s}) + \textrm{ln} \left(\frac{C^{A}_{o}}{ (C^{B}_{o})^{R}} \right) \]
Where R is a function of the bulk partition coefficients:
\[ R = \frac{D_{B}(1-D_{A})}{D_{A}(1-D_{B})} \]
Calculated bulk rock compositions were plotted as ln([A]) vs. ln([B]) and fit with a two-error regression. Most regressions
involving two REE intersect the primitive upper mantle (PUM) coordinates of McDonough and Sun (1995) within error of the
regression, reinforcing the assumption that the starting composition for DMM was PUM, and suggesting that bulk partition
coefficients have been very nearly the same throughout the melt depletion process.
Since this regression scheme only provides relationships between elements, the absolute concentration of one or more elements
must be determined from an additional constraint. For this, the time-integrated Sm/Nd ratio is determined for DMM based on
its isotopic evolution from BSE. The Sm/Nd ratio has a unique intersection with the Sm-Nd abyssal peridotite regression
line, defining unique Sm and Nd concentrations. This intersection sets the absolute Sm and Nd concentrations of DMM equal to
0.27 and 0.64 ppm, respectively. With these absolute concentrations of Sm and Nd, estimates of the absolute concentration of
all other measured trace elements are made. The same is done in turn with each of the other elements. With the abyssal
peridotite regressions, constraints from radiogenic isotopes, interpolations between the REE, and canonical trace element
ratios (i.e. U/Nb, Ce/Pb, Nb/Ta, Ba/Rb), a complete trace element pattern has been calculated for DMM and is shown at
http://www.whoi.edu/pclift/whAGU03. This pattern represents ~3% melt depletion with calculated partition coefficients
matching those of spinel-field peridotite melting. Depletions relative to PUM are 0.04 (Rb), 0.16 (Th), 0.21 (U), 0.28 (La),
0.28 (Sr), 0.70 (Eu), and 0.85 (Yb).
UR: http://www/whoi.edu/pclift/whAGU03
DE: 1025 Composition of the mantle
DE: 1040 Isotopic composition/chemistry
DE: 1065 Trace elements (3670)
DE: 3035 Midocean ridge processes
SC: Volcanology, Geochemistry, Petrology [V]
MN: 2003 Fall Meeting