rocholl@classic.min.uni-heidelberg.de
Mineral-melt partition coefficients (D) may vary drastically with changing chemical and physical conditions (e.g. Green, 1994). However, for many petrogenetic models of
the evolution of mantle derived melts the availability of appropriate D's (i.e., obtained from experiments which most closely approach the natural P-T-X conditions of the system studied) is essential. For example, many basanites are believed to be derived from a garnet-bearing mantle source (e.g. Class et al., 1994). However, respective experimental D's are restricted to only a few elements and to pressures below 3 GPa (Adam and Green, 1994), and, to our knowledge, D's (garnet/melt) do not exist for basanitic compositions.
We have carried out a series of experiments at 1.5-5.0 GPa and 1200-1640°C. Starting material was a basanite (BAS 914 from Kenya; Class et al., 1994), doped with 0.5 wt % Sc, Y, and Lu, or, alternatively, with 0.5 wt % La, Sm, and Gd. These elements were measured by electron microprobe in order to study the general partitioning behavior of trace elements through a wide range of experimental conditions (Seitz and Altherr, 1996). In addition, trace elements at natural levels were measured by ion probe for three selected experiments (Table 1; see over).
At first approximation, results for the two different experimental conditions (829, VD-15) are quite similar (Table 1). In detail, they differ in that D-values for 3+ to 5+ ions are always higher by a factor of 1.3-3.2 for experiment VD-15. (The effect on Eu is even higher, but this probably reflects lower fO2 during run 829.) Partitioning of trace elements is expected to vary with P, T, and phase compositions. Seitz and Altherr, (1996) observe good correlations between D's for elements doped in our experiments with Al(IV) contents in cpx, but no systematic variations with P, T, or the site occupancies of M1 and M2 in cpx: The incorporation of Ti4+ and the doped trivalent elements appears to be most easily modeled by coupled substitution involving Al(IV). Our data suggest that this interpretation is valid for all (except the divalent) trace elements, since Al(IV) is higher by a factor of × 2 in VD-15 cpx (relative to run 829; Table 1). Moreover, according to the respective Al(IV) contents in cpx, our data lie within the ranges of published D's for basanites (Table 1; Adam and Green, 1994), while those obtained for alkali basaltic compositions (Hart and Dunn, 1993) are intermediate between the data of runs 829 and VD-15.
Many D's obtained from experiments 829 and 832 are identical within errors. However, with increasing compatibility of the element, results for run 829 are consistently higher by about 20-80%. More experimental data are required to interprete this observation in context of crystal-chemistry or melt composition.
Adam, J. & Green, T.H., Chem. Geol. 117, 219-233 (1994).
Class, C., Altherr, R., Volker, F., Eberz, G. & McCulloch, M.T., Chem. Geol. 113, 1-22 (1994)
Green, T.H., Chem. Geol. 117, 1-36 (1994).
Hart, S.R. & Dunn, T., Contrib. Mineral. Petrol. 113, 1-8 (1993).
Seitz, H.-M. & Altherr, R., J. Conf. Abs. 1, 555 (1996).
Table 1.
Run 829 832 829 VD-15 A&G (a) H&D (b)
T [° C]
P [GaP]
Al(IV) cpx
Run prod.
Ba
Nb
La
Ce
Sr
Nd
Zr
Sm
Eu
Ti
Gd
Dy
Y
Er
Yb
Lu
Mn
Sc
V
Cr
1580
5
-
Cpx, Grt, Gl(q)
D(Grt/M)
0,0007
0,0064
0,0067
0,0146
0,0048
0,0854
0,257
0,308
0,170
0,234
0,734
1,57
2,57
2,32
1,06
5,28
1,66
4,42
2,73
9,01
1640
5
-
Grt, Glass
D(Grt/M)
0,0007
0,0068
0,0043
0,0143
0,0053
0,0823
0,240
0,231
0,155
0,184
0,545
1,23
1,94
1,95
1,03
4,30
1,01
2,87
1,76
4,81
1580
5
0.143
Cpx, Grt, Gl(q)
D(Cpx/M)
0,0012
0,0032
0,0524
0,0786
0,1706
0,1317
0,062
0,202
0,098
0,1868
0,262
0,254
0,304
0,268
0,244
0,273
0,424
1,10
2,31
3,73
1280
1.5
0.256
Cpx, Gl
D(Cpx/M)
-
0,0086
0,0605
0,1246
0,0889
0,3124
0,195
0,521
0,580
0,506
0,457
0,728
0,538
0,639
0,513
0,498
0,642
2,15
3,45
9,06
1025-1350
0.5-3.0
0.10-0.22
2 Cpx, Gl, a.o.
D(Cpx/M)
-
-
0.04-0.10
-
-
-
-
0.31-0.54
-
0.29-0.51
-
-
-
-
-
0.28-0.66
-
-
-
-
1380
3
0.2145
2 Cpx, Gl
D(Cpx/M)
0.00068
0.0077
0.0536
0.0858
0.1283
0.1873
0.1234
0.291
-
0.384
-
0.442
0.467
0.387
0.43
0.433
-
1.31
3.1
3.8
a: Adam and Green (1994); b: Hart and Dunn (1994); M: melt; Gl: glass; q: quenched; Grt: garnet; Cpx: clinopyroxene.
For all phases analyzed, trace element concentrations were standardized using NIST glass SRM 610.