Magmas: windows into the mantle - New Theory of the Earth - page 180

relatively depleted in strontium and europium,
elements that have been removed from the
KREEP source region by plagioclase fractiona-
tion. Kimberlite is depleted in sodium, HREE,
hafnium, zirconium and yttrium, elements that
are removed by garnet-plus-clinopyroxene frac-
tionation. It appears that the differences between
KIMB and KREEP are due to difference in pressure
between the Earth and the Moon; garnet is not
stable at pressures occurring in the lunar man-
tle. The plagioclase and garnet signatures show
through such effects as the different volatile-
refractory ratios of the two bodies and expected
differences in degrees of partial melting, extent
of fractional crystallization and other features.
The similarity in composition extends to metals
of varying geochemical properties and volatilities
such as iron, chromium, manganese as well as
phosphorus. KREEP is depleted in other metals (V,
Co, Ni, Cu and Zn), the greater depletions occur-
ring in the more volatile metals and the met-
als that are partitioned strongly into olivine. This
suggests that olivine has been more important in
the evolution of KREEP than in the evolution of
kimberlite, or that cobalt, copper and nickel are
more effectively partitioned into MgO-rich fluid
such as kimberlite.
It appears from Figure 14.3 that KREEP and
kimberlite differ from chondritic abundances in
similar ways. Both are depleted in volatiles rel-
ative to refractories, presumably due to preac-
cretional processes. Strontium is less enriched
than the other refractories, although this is
much more pronounced for KREEP. The pro-
nounced europium and strontium anomalies
for KREEP are consistent with extensive plagio-
clase removal. The HREE, yttrium, zirconium and
hafnium are relatively depleted in kimberlite,
suggesting eclogite fractionation or a garnet-rich
source region for kimberlite. The depletion of
scandium simply indicates that olivine, pyroxene
or garnet have been in contact with both KREEP
and KIMB. The depletion of sodium in KIMB is
also consistent with the involvement of eclogite
in its history. The depletion of sodium in KREEP,
relative to the other alkalis, is presumably due
to the removal of feldspar, and the greater rela-
tive depletion of sodium in kimberlite therefore

10

eclogite residue
melt

D eclogite =

1

MORB
KIMB

0.1

0.01

MORB + 50 % Ol
KIMB

0.001
U K Rb Sr Y Ba La Nd Sm Yb Sc

Fig. 14.2 Solid line is ratio of concentrations in MORB and
kimberlite (KIMB). Vertical bars are solid/liquid partition
coefficients for garnet plus clinopyroxene. The dashed line
gives the ratio for MORB plus 50 percent olivine, a possible
picrite parent magma for MORB. If the MORB or picrite
source region is the crystalline residue remaining after
removal of a kimberlitic fluid, the ratio of concentrations,
MORB/KIMB or picrite/KIMB, should equal the solid--liquid
partition coefficient, which depends on the crystalline (xl)
phases in the residue.

and low-degree melts. If diapirs initiate in a deep
depleted layer, they must traverse the shallow
mantle during ascent, and cross-contamination
seems unavoidable. The more usual model is that
the whole upper is depleted and enriched mag-
mas originate as plumes in the deep mantle.

The Kimberlite--KREEP Relation
KREEP is a lunar material having very high con-
centrations of incompatible elements (K, REE, P,
U, Th). It is thought to represent the residual
liquid of a crystallizing magma ocean .
Given proposals of a similar origin for kim-
berlite, it is of interest to compare the com-
position of these two materials. An element
by element comparison gives the remarkable
result that for many elements (K, Cs, P, S, Fe,
Ca, Ti, Nb, Ta, Th, U, Ba and the LREE) kim-
berlite is almost identical (within 40%) to the
composition of KREEP (Figures 14.3 and 14.4).
This list includes compatible and incompati-
ble elements, major, minor and trace elements,
and volatiles as well as refractories. KREEP is