Garnets - Rare Earth Elements in Ultramafic and Mafic Rocks and their Minerals - page 64

additional research is needed. However, it is already obvious that this multistage and
different, by PT- parameters, history of garnet crystals growth was the most impor-
tant reason that even within a single manifestation the garnet samples of identical
petrographic composition of the mantle xenoliths often differ on the basis of the REE
pattern configuration. The examples of such geochemical heterogeneity can be the
garnets from diamondiferous Roberts Victor, Udachnaya, Aykhal and Mir kimberlite
pipes. It is important to mention that the sinusoidal REE pattern, identified in the
garnets containing more than 12 wt% Cr 2 O3 and having a higher content of knor-
ringite minal (more than 30%), are currently supposed to be the most important
search criteria for assessing the potential diamond-bearing of kimberlite provinces
[Pokhilenko et al ., 1998].
It is well known that on the periphery of garnet segregations presented in the form
of xenocrystals in kimberlites those were often observed kelyphitic rims of complex
structure, the formation of which is usually associated with exposure of kimberlite
melts and fluids. Special studies showed that this kind of rims by their REE composi-
tion are close to the garnet crystals surrounded by them and differ significantly from
their host kimberlites. These observations suggest that the formation of kelyphitic
rims around the garnet crystals is not causally associated with exposure of the kimber-
lite melts which have passed them to the surface, but is caused by some earlier abyssal
metasomatic processes [Spetsius & Griffin, 1998].
1.4 COEFFICIENTS OF REE DISTRIBUTION BETWEEN
GARNETS AND COEXISTING MELTS
Researchers studying the geochemistry of garnets are constantly paying attention to
the regularity of distribution of REE and other trace elements between this mineral
and coexisting melts, and solid phases [Lesnov, 2001]. In order to obtain estimates
of K d (garnet/melt) there are usually both natural and experimental systems involving
ultramafic, basaltic, andesitic, dacitic, rhyodacitic and rhyolitic melts used. To date,
the biggest number of K d estimates is obtained for Sm, Yb, Lu and Ce, very little data
for Gd, Tb, Ho and almost no data for Pr and Tm (Table 1.13). The diagrams of
change of K d (garnet/melt) estimates are shown in Figure 1.13.
Among the first there were obtained the estimates of K d (garnet/melt) upon the
analysis of REE in garnet phenocrystals and containing them poorly crystallized matrix
of dacites from volcanic complexes in Japan [Schnetzler & Philpotts, 1970]. In this
case, the values of K d (garnet/dacitic melt) for La, Ce and Nd range between 0,3 and
0,4, then, increasing for Sm, have a distinct minimum for Eu, after which the curve
gets a steeper positive slope with increasing K d values in the series from Gd (10.5)
to Yb (26.0) and Lu (24.6) (Figure 1.13, 6). Over time, the results of experimen-
tal studies were mainly used to determine the values of K d (garnet/melt) [Shimizu &
Kushiro, 1975; Irving & Frey, 1978]. The effect of such parameters as the chemical
composition of the garnets, the total REE content in the system, the silica content in
the melts, as well as temperature and pressure during crystallization, is usually taken
into account during these studies [Mysen, 1978; Beattie, 1993; Wim van Westrenen
et al ., 1999]. The data obtained by these experiments confirmed earlier observations
that indicated that the majority of the diagrams of K d (garnet/melt) values have a steep