Earth and Planetary Science Letters, 114 (1993) 477 489
477
Elsevier Science Publishers B.V., A m s t e r d a m
[CH]
Partial melting of dry peridotites at high pressures: Determination of compositions of melts segregated from peridotite using aggregates of diamond Kei Hirose and Ikuo Kushiro Geological Institute, Unit,ersity of Tokyo, Hongo, Tokyo 113, Japan Received August 25, 1992; revision accepted December 4, 1992
ABSTRACT
The compositions of melts formed by partial melting of two relatively fertile spinel Iherzolites were determined at pressures between 10 and 30 kbar under dry conditions using a layer of diamond aggregates sandwiched between peridotite layers. Partial melts segregate and migrate into the pore space between diamond grains soon after their formation. Overgrowth of minerals at quenching modifies the composition of coexisting melt, but this modification does not extend to the trapped melt in the diamond layer. Microprobe analyses of the trapped melt, therefore, can determine the melt compositions without the quench problem. Melts formed by low degrees of partial melting ( ~ 5%) can be analyzed successfully by this method. The effect of the source composition on partial melting was examined for the two different lherzolites. Both partial melts have nearly the same SiO 2 and MgO contents at the same pressure and temperature conditions regardless of their different M g / ( M g + Fe) ratios, suggesting that SiO 2 and MgO contents in partial melts depend little on the source composition. In contrast, FeO, AI20~, CaO and incompatible elements are controlled by the composition of the source peridotite as well as by the degree of partial melting. In the normative projections, the differences in the lherzolite compositions do not significantly shift the isobaric compositional trends of the partial melts.
1. Introduction
Partial melting of peridotites in the upper mantle is believed to be a process important to our understanding of the origin of basaltic magmas. Numerous high pressure experimental studies have been attempted to determine the compositions of such partial melts in both synthetic and in natural peridotite systems. Yet, there are still problems concerning the method of determination of the compositions of quenched liquid. Melting experiments with peridotite capsules and with the sandwich method were fairly successful in the determination of major elements in equilibrium melts [1-4]. However, the compositions of melts determined by these methods, particularly the abundances of incompatible elements, are affected and modified by the materials added to the host peridotites, and, therefore, they do not
represent the true partial melt compositions. Such modification is especially large when the degree of partial melting is low. In the experiments presented here, analyzable amounts of partial melts can be extracted from the host peridotites at high pressures by a new method using aggregates of diamond grains, which provide pore space for melt extraction [5-7] (Fig. lb). It is anticipated that microprobe analyses of these separated melts would give the true partial melt compositions. Melts formed by low degrees of partial melting (as low as 5%) can also be extracted and determined successfully. Recent studies suggest that small melt fractions can segregate from the host peridotite [e.g. 8-10]. Determination of the compositions of equilibrium partial melts formed by low degrees of partial melting would contribute to modelling magma generation by such fractional melting.
0012-821X/93/$06.00 © 1993 - Elsevier Science Publishers B.V. All rights reserved
478
K. HIROSE AND I. KUSH1RO
The source materials for basalt magmas (upper mantle peridotite) would be heterogeneous and so the effect of different source compositions on partial melts should be investigated. In this paper, we present new experimental results on anhydrous melting of two different, relatively fertile, spinel lherzolites which differ in terms of their M g / ( M g + Fe) ratio and C r / ( C r + A1) ratio in spinel. One is a spinel lherzolite from Salt Lake Crater, Hawaii (HK66101703, hereafter designated HK-66) [1,11]. The other, designated KLB-1, is a less fertile spinel lherzolite from Kilborne Hole Crater in New Mexico [12]. The compositions of partial melts formed near the solidi of these two spinel lherzolites have been determined using the new method presented here, and the effects of pressure, temperature and
composition of the source peridotite on the composition of partial melts are discussed.
2. Experimental and analytical method In the experiments reported here, partial melts were separated from the host peridotite into pore space between diamond grains at high pressures. This experimental technique has already been shown to be efficient in separating partial melts from peridotite [5-7] (Fig. lb). Three different configurations of charges were tested to obtain equilibrium partial melts [6]: (i) an aggregate of diamond powder (40-60 /zm in diameter) is packed on top of tightly packed peridotite powder (the ratio of diamond to peridotite is 1:1-1:2), (ii) a thin layer ( < 0.5 mm thick) of diamond
a.
Diamond
Peridotite Graphite capsule
tp,I ........ (i)
(ii)
(iii) 1 mm
bo
Fig. 1. (a) Three different configurations of charges used in the present experiments described in [5,6]. Melts in equilibrium with the host peridotite can be obtained with configurations (ii) and (iii). (b) Backscattered electron image of a charge obtained at 15 kbar and 1400°C. This shows that melts are segregated and trapped between the diamond grains.
PARTIAL MELTING OF DRY PERIDOTITES AT HIGH PRESSURES
powder is sandwiched between peridotite powder (5-50 txm), and (iii) a chip of loosely sintered diamond powder is embedded in peridotite powder (Fig. la). The sintered diamond was made by heating diamond powder at 10 kbar and 1300°C for 1 h prior to the experiments. In all the configurations, the pore space between diamond grains was preserved at pressures at least to 30 kbar before the run. Although partial melts separate and migrate into the pore space between diamond grains soon after their formation, chemical contact would be maintained with melts in the peridotite layers. Melt filled the pore space between the diamond grains even in a 1 h run. As shown below, equilibrium partial melts can be obtained with configurations (ii) and (iii). The starting materials were two natural peridotites. The charges were loaded in graphite capsules measuring 2 m m in diameter and 3.5 m m in length. To ensure dry conditions, the graphite capsules loaded with the peridotite and diamond were preheated at about 1000°C before sealing in Pt capsules. The runs were made in a temperature range of 1250-1525°C and in a pressure range of 10-30 kbar with a piston-cylinder type apparatus. Calibration for pressure and other experimental procedures with this apparatus have been described elsewhere [13]. T e m p e r a t u r e was measured and controlled by Pt/Pt90Rhl0 thermocouples. After quenching the charges were sectioned longitudinally and polished for microprobe analyses. Polishing was easy because the surface of the diamond grains changes into graphite during the run. Analyses were made with a scanning electron microscope ( J E O L JSM-840) with Link EDS systems. The quenched melts trapped between the diamond grains were 10-50 p~m in size in most cases. Analytical conditions were 15 kV accelerating voltage, 1 nA specimen current and 60-80 s counting time.
3. Attainment of equilibrium It is important to check whether chemical equilibrium is attained between the melts segregated into the pore space in the diamond layer and the residual phases of the peridotite. In our earlier study [6], we conducted experiments on the synthetic system Mg2SiO4-NaA1SiO4-SiO2, the phase relationships (liquidus boundary be-
479
tween forsterite and enstatite solid solution) of which had been determined at high pressures [14]. The results indicated that equilibrium melts can be obtained with configurations (ii) and (iii) within 24 h at least at 10, 15 and 20 kbar [6]. However, with configuration (i) unexpectedly low pressure melt was formed. This is probably duc to the fact that the pore space in the diamond layer was too large to be filled with melt, resulting in the melt pressure being lower than the load pressure. It is important, therefore, that the volume of pore space in the diamond layer should be reduced to less than that of the equilibrium melt. We used sintered diamond chips, the pore space of which is about 50 vol%, as estimated from the SEM images. In most of the runs the sintered diamond chips occupy less than 3 vol% peridotite powder, so that only 1.5% partial melt can fill up the pore space. The pore space was reduced to < 1% by using a smaller sintered diamond chip when the degrees of melting were very low. In the experiments presented here, configuration (iii), which contains a chip of loosely sintered diamond grains, was used in most of the runs; configuration (ii) was also used for a few runs. The time necessary to obtain equilibrium melts was also examined in the same synthetic system and with spinel lherzolite KLB-1. The results from the synthetic system indicated that at least 24 h are necessary to obtain equilibrium melts at 15 kbar and 1400°C [6]. In the experiments presenteh here the time-dependent compositional changes of melts have been examined for KLB-1 with configuration (iii). The results at 15 kbar and 1350°C for three different run durations (6, 24 and 67 h) are shown in Table 1. In the 6 h run, the A120 3 content is relatively high and the MgO content is relatively low. After 24 h, the abundances of major elements reach nearly constant values. Very little compositional zoning in respect of the M g / ( M g + Fe) ratio in the minerals was observed in the 24 and 67 h runs. In contrast, however, weak compositional zoning is observed for A I 2 0 3 and CaO in pyroxenes even in the 24 h run, but this zoning is minimal in the 67 h run. From this point of view, melts that are completely in equilibrium cannot be obtained in 24 h, and the bulk equilibrium is nearly achieved after 67 h. However, the composition of the partial melt changes very little after 24 h. This means that the
480
K. H I R O S E
TABLE 1 Results of runs with different run durations on KLB-1 at 15 kbar and 1350°C
SiO 2 TiO 2 AI20 3 FeO* MnO MgO CaO Na20 K20 Cr20 3 Total Fo (mol%) K d (Fe/Mg)
6h
24 h
67 h
49.53 0.51 16.28 7.54 0.15 12.45 11.67 1.67 0.07 0.13 100.00
49.13 (48.82-49.48) ~' 0.60 (0.43 - 0.69) 15.18 (14.94- 15.32) 7.54 (7.08 - 7.78) 0.14(0.05-0.31) 13.11 (12.72- 13.57) 12.28 (11.96 - 12.89) 1.58 (1.36- 1.70) 0.08 (0.06 - 0.15) 0.36 (0.24-0.47) 100.00
49.44 0.55 15.20 7.53 0.11 13.41 12.00 1.51 0.04 0.21 100.00
90.7 0.302
90.6 0.321
90.6 0.331
* all Fe as FeO ~ The range of analyses of the trapped melt pools
effect of relic pyroxenes on the composition of m e l t s is v e r y s m a l l a f t e r 24 h. T h e m e l t f o r m e d a t 10 k b a r a n d 1250°C f o r 56 h ( r u n 1) is c l o s e in c o m p o s i t i o n t o a c o t e c t i c m e l t a t 10 k b a r a n d 1260°C o b t a i n e d by K i n z l e r a n d G r o v e [15]. The Fe/Mg partitioning between olivine and l i q u i d o b s e r v e d i n t h e p r e s e n t e x p e r i m e n t s is in g o o d a g r e e m e n t w i t h t h a t in t h e p r e v i o u s w o r k [e.g. 1] ( T a b l e 4). A s s h o w n in t h e e a r l i e r e x p e r i m e n t s [5,6], M g 2+ a n d F e 2+ in m e l t s r e a c h c o n stant values over shorter run durations than ions w i t h s m a l l e r d i f f u s i v i t i e s s u c h as Si 4+ a n d A13+. The Fe/Mg ratio, therefore, cannot be used for judging perfect equilibration of partial melting.
A N D 1. K U S H I R O
clinopyroxene and 2% spinel, whereas KLB-1 consists of 58% olivine, 25% orthopyroxene, 15% clinopyroxene and 2% spinel. The Fo contents of
TABLE 2 Compositions of spinel Iherzolites KLB-1 and HK-66 and their constituent minerals
SiO2 TiO 2
AI203 FeO* MnO MgO CaO Na20
K20 P2Os Cr203 NiO Total
KLB- 1
HK-66
Spinel lherzolite xenoliths ~
44.48 0.16 3.59 8.10 0.12 39.22 3.44 0.30 0.02 0.03 0.31 0.25 100.02
48.02 0.22 4.88 9.90 0.14 32.35 2.97 0.66 0.07 0.07 0.25 n.d. 99.53
44.20 0.13 2.05 8.29 0.13 42.21 1.92 0.27 0.06 0.03 0.44 0.28 100.01
KLB-1 b
ol
opx
cpx
sp
SiO 2 TiO z AI203 FeO* MnO MgO CaO Na20
NiO Total Mg/(Mg + Fe)
39.64 0.00 0.03 10.52 0.16 48.25 0.08 0.01 0.00 0.01 0.39 99.09 0.891
54.24 0.11 4.97 6.57 0.16 32.16 0.85 0.12 0.00 0.34 0.11 99.63 0.897
51.13 0.58 7.40 3.11 0.10 14.70 19.54 1.72 0.01 0.78 0.11 99.18 0.894
0.06 0.11 58.48 10.68 n.d. 21.61 0.00 0.00 0.00 7.82 0.43 99.19 0.783
HK-66 ¢
ol
opx
cpx
sp
SiO 2 TiO 2 A120 3 FeO* MnO MgO CaO Na20
38.76 0.00 0.00 14.41 0.17 45.89 0.07 0.00 0.00 0.35 99.65 0.850
53.50 0.11 5.27 8.71 0.13 31.13 0.71 0.13 0.16 n.d. 99.85 0.864
50.93 0.55 7.35 4.30 0.09 14.62 18.75 1.96 0.50 n.d. 99.05 0.858
0.12 0.21 61.31 15.34 0.13 19.53 0.04 0.00 2.66 0.45 99.79 0.694
K20 Cr203
4. Starting materials of natural peridotite Melting experiments were conducted on two different natural spinel lherzolites, HK-66 and KLB-1. Both have relatively fertile compositions, b e i n g e n r i c h e d in A I 2 0 3 a n d C a O a n d d e p l e t e d in M g O a n d C r 2 0 3 c o m p a r e d w i t h t h e w o r l d average composition for spinel lherzolite xenol i t h s o f M a a l o e a n d A o k i [16] ( T a b l e 2). T h e differences between the two lherzolites are obs e r v e d in t h e m o d a l a m o u n t o f o r t h o p y r o x e n e , t h e M g / ( M g + F e ) r a t i o , t h e C r / ( C r + A1) r a t i o of spinel, and abundances of incompatible elements. Lherzolite HK-66 consists of 27% (in terms of weight) olivine, 57% orthopyroxene, 14%
Cr203 NiO Total Mg/(Mg + Fe)
~ Average of 384 spinel lherzolites [16] b Data from [12] c Data from [1]
PARTIAL
MELTING
OF
DRY
PERIDOTITES
1600
AT HIGH
481
PRESSURES
i
i
i
i
i
i
i
/
/
1550
HK-66
/
1500
O/ AI
1450
•
1400 •
D
[]
/ /
g
o
a
10
o
h iJ
/
J I
/
d
/ E"
/
/ 0
/
/0
/ /
1200
0
/
1250
/
/
/
/
1300
q /0
o. 1
1350
1
KLB-
/
I
L
I
15 20 25 Pressure (kbar)
I
30
/I
L
10
I
I
15 20 25 Pressure (kbar)
I
30
35
Fig. 2. Pressure-temperature conditions of the present experiments. Solidi of HK-66 and KLB-1 are from previous studies [2,10]. In some cases, melts were detected below the reported solidus temperature.
the olivine are 85.0 and 89.1 (mol%) for HK-66 and KLB-1 respectively. Takahashi and Kushiro [1] determined the solidus of HK-66 and at-
tempted to determine the equilibrium melt compositions near the solidus by the sandwich method. The melting behavior of lherzolite KLB-1
TABLE 3 Details of runs Run
P (kbar)
T (°C)
Duration (h)
Residue
HK-66 1 2 3 4 5 6 7 8 9 10 11 12 13
Configuration
10 10 10 15 15 15 20 20 20 25 25 30 30
1250 1300 1350 1275 1350 1400 1350 1375 1425 1425 1450 1475 1500
56 24 24 24 23 24 24 24 24 24 24 12 12
ol, opx, ol, opx, ol, opx ol, opx, ol, opx ol, opx ol, opx, ol, opx, ol, opx ol, opx, ol, opx ol, opx ol, opx
KLB-1 14 15 16 17 18 19 20 21 22 23 24 25 26
10 10 10 10 15 15 15 20 20 25 25 30 30
1250 1300 1350 1400 1300 1350 1400 1 375 1425 1 425 1450 1500 1525
56 24 24 24 24 24 44 24 24 24 24 12 12
ol, opx, cpx, sp, pl ol, opx, cpx, sp ol, opx ol, opx ol, opx, cpx, sp ol, opx, cpx ol, opx ol, opx, cpx ol, opx, cpx ol, opx, cpx ol, opx, cpx ol, opx, cpx ol, opx, cpx
cpx, sp, pl cpx cpx, sp
cpx, sp cpx cpx
(iii) (ii) (ii) (ii) (iii) (ii) (iii) (iii) (iii) (iii) (iii) (iii) (iii) (iii) (ii) (ii) (ii) (iii) (ii) (iii) (iii) (iii) (iii) (iii) (iii) (iii)
482
K. HIROSE AND 1. KUSHIRO
TABLE 4 Compositions of the partial melts HK-66
Run P (kbar) T (°C)
1 10 1250
2 10 1300
3 10 1350
4 15 1275
5 15 1350
6 15 1400
SiO 2 TiO 2 AI eO3 FeO * MnO MgO CaO Na20 K20 CreO 3 Total
50.83 1.12 18.57 7.98 0.28 7.77 9.44 3.69 0.32 n.d. 100.00
49.60 0.79 16.49 8.89 0.12 10.35 11.27 2.15 0.15 0.19 100.00
50.37 0.54 14.53 9.44 0.16 13.46 9.55 1.60 0.14 0.21 100.00
49.83 1.07 18.80 8.03 0.08 8.26 7.93 5.41 0.59 n.d. 100.00
48.42 0.85 16.60 9.42 0.11 11.25 10.69 2.36 0.09 0.21 100.00
48.48 0.78 13.06 10.79 0.21 14.26 10.50 1.66 0.10 0.16 100.00
os ab an ne di hy ol il
2.13 20.60 32.07 3.09 19.59
0.89 18.22 34.98
0.83 13.57 32.11
0.53 20.01 34.52
0.59 14.04 27.94
21.01 1.52
17.11 12.71 14.59 1.50
12.41 28.76 11.29 1.03
3.49 27.26 25.28 10.02 11.46 20.47 2.03
15.10 5.23 22.98 1.62
19.65 13.34 22.95 1.48
Fo (mol%) Kd (Fe/Mg) F '~
85.6 0.292 0.179
87.1 0.307 0.307
87.7 0.356 0.412
85.9 0.301 0.122
87.0 0.318 0.279
87.4 0.339 0.398
C1PW norm
ITK-66 Run P (kbar) T (°C)
7 20 1350
8 20 1375
9 20 1425
10 25 1425
11 25 1450
12 30 1475
13 30 1500
SiO 2 TiO 2 AI20 3 FeO* MnO MgO CaO Na20 K20 Cr20 3 Total
47.41 1.44 15.92 10.38 0.29 10.85 9.26 3.91 0.42 0.12 100.00
47.55 1.18 15.61 10.58 0.14 12.01 9.96 2.61 0.25 0.11 100.00
47.12 1.17 13.96 10.47 0.12 14.74 10.23 1.90 0.21 0.08 100.00
46.82 1.16 13.81 11.68 0.21 13.28 10.22 2.34 0.34 0.14 100.00
47.41 0.82 12.29 11.69 0.12 15.54 10.17 1.58 0.19 0.19 100.00
45.89 1.40 12.43 12.91 0.20 14.30 9.89 2.51 0.35 0.12 100.00
45.59 1.18 12.35 12.56 0.30 15.98 9.67 2.02 0.18 0.17 100.00
2.49 19.02 24.70 7.61 17.35
1.48 20.31 30.21 0.95 15.67
1.24 16.09 28.98
2.01 15.94 26.22 2.10 19.99
1.13 13.40 25.93
2.07 12.85 21.68 4.52 22.33
1.07 13.15 24.17 2.11 19.39
26.10 2.74
33.89 2.66
37.88 2.23
85.7 0.311 0.169
86.5 0.308 0.263
87.2 0.333 0.327
CIPW norm
os ab an ne di hy ol il Fo (mol%) K a (Fe/Mg)
F a
29.15 2.24
17.64 2.60 31.23 2.22
31.52 2.21
19.95 7.82 30.23 1.54
86.2 0.324 0.253
86.7 0.385 0.347
86.5 0.316 0.282
87.9 0.326 0.418
a Degree of partial melting calculated from N a 2 0 abundance
PARTIAL MELTING OF DRY PERIDOTITES AT HIGH PRESSURES
483
KLB-1 Run P (kbar) T (°C)
14 10 1250
15 10 1300
16 10 1350
17 10 1400
18 15 1300
19 15 1350
20 15 1400
SiO 2 TiO 2 A120 3 FeO* MnO MgO CaO Na 2° K20 Cr20 3 Total
51.32 1.09 19.09 6.38 0.23 8.14 8.85 4.60 0.27 0.03 100.00
50.49 0.65 17.94 6.69 0.11 10.08 11.37 2.47 0.09 0.11 100.00
50.67 0.42 14.61 7.64 0.13 13.39 11.17 1.50 0.19 0.28 100.00
51.59 0.44 12.58 7.95 0.29 16.41 9.42 0.91 0.05 0.36 100.00
50.71 1.04 19.31 6.37 0.14 8.31 7.75 5.47 0.73 0.17 100.00
49.13 0.60 15.18 7.54 0.14 13.11 12.28 1.58 0.08 0.36 100.00
49.88 0.45 13.78 7.92 0.13 15.74 10.69 1.04 0.04 0.33 100.00
os ab an ne di hy ol il
1.54 33.78 30.69 2.79 10.70
0.36 19.68 38.45
1.13 12.72 32.66
0.30 7.70 30.21
0.47 13.45 34.19
0.24 8.80 32.93
18.42 2.08
14.63 15.41 10.23 1.24
18.43 24.12 10.13 0.80
13.37 43.84 3.75 0.84
4.33 29.61 26.04 9.07 10.06 18.93 1.96
21.67 12.45 16.63 1.14
16.25 28.92 12.01 0.85
Fo (mol%) Kd (Fe/Mg) F a
89.2 0.275 0.065
89.9 0.302 0.121
91.0 0.309 0.200
93.1 0.273 0.331
89.5 0.273 0.055
90.6 0.321 0.189
91.5 0.329 0.289
CIPWnorm
KLB-I Run P (kbar) T (°C)
21 20 1375
22 20 1425
23 25 1425
24 25 1450
25 30 1500
26 30 1525
SiO 2 TiO 2 A120 3 FeO * MnO MgO CaO Na20 K20 Cr20 3 Total
47.47 0.75 15.53 8.51 0.18 13.94 11.11 2.22 0.08 0.21 100.00
48.74 0.51 13.16 8.80 0.24 15.69 11.06 1.37 0.13 0.30 100.00
47.97 0.83 14.88 9.43 0.00 13.36 10.23 2.37 0.82 0.11 100.00
48.37 0.69 13.80 8.47 0.07 15.88 10.93 1.46 0.15 0.18 100.00
45.67 0.99 14.33 9.59 0.17 16.73 10.64 1.80 0.07 0.21 100.00
46.77 0.55 12.87 9.81 0.29 17.82 10.63 0.87 0.09 0.30 100.00
os ab an ne di hy ol il
0.47 17.14 32.25 0.89 18.49
0.77 11.62 29.46
0.89 12.38 30.72
0.42 11.18 30.85 2.22 17.72
0.53 7.38 31.05
29.33 1.42
20.57 14.64 21.97 0.97
4.85 15.87 27.57 2.28 18.77
Fo (mol%) Kd (Fe/Mg) F a
90.2 0.317 0.135
90.8 0.322 0.219
C I P W norm
29.09 1.58
18.91 11.85 23.94 1.31
90.5 0.265 0.126
91.2 0.323 0.206
35.72 1.89
17.58 13.17 29.25 1.0.5
90.1 0.341 0.166
90.8 0.328 0.347
484
K. HIROSE
was originally studied by Takahashi [12], and the degrees of partial melting and the compositions of partial melts were estimated by Takahashi et al. [17] over a pressure range of 1 bar to 65 kbar. 5. E x p e r i m e n t a l
results
Melting experiments were carried out at temperatures just above the solidi of HK-66 and KLB-1 under dry conditions (Fig. 2). Details of the run conditions and the residual assemblages are summarized in Table 3. In some cases, partial melts were detected at temperatures slightly below the solidus reported by Takahashi [12]. The compositions of partial melts (estimated degree of partial melting 5 ~ 40%) determined with the present method are shown in Table 4. The solidus temperature of KLB-1 was reported to be about 50°C higher than that of HK-66 above 10 kbar. The degree of melting of HK-66, therefore, is always higher than that of KLB-1 under the same pressure and temperature conditions. The degree of melting ( F ) was calculated from the mass balance of N a 2 0 and is listed in Table 4. These calculations, however, would give maximum values, because of the presence of small amounts of Na in the clinopyroxene and plagioclase. The homogeneity of melts distributed in the diamond layer was examined in the earlier experiments [6] by analyzing many different quenched melts within the diamond layer. The ranges for the major oxide components are small (Table 1). The quench minerals were scarce in the melt pools in the diamond layer. At 10 kbar, the composition of melts formed by partial melting of KLB-1 just above the solidus is nepheline normative and alkali-olivine basaltic, whereas that of HK-66 is olivine tholeiitic and within the compositional range of high alumina basalt (A120 3 > 18%) [18]. At higher temperatures melts are also olivine tholeiitic, but they become depleted in the albite component and enriched in the hypersthene component. Above 15 kbar, melts formed near the solidi are alkali basaltic or alkali picritic. A melt with nephelinite composition (normative nepheline > 10%) was formed from HK-66 at 15 kbar. With increasing pressure, S i O 2 c o n t e n t s decrease from about 50 wt% at 10 kbar to about 46 wt% at 30 kbar. MgO and FeO contents are higher at higher pressures,
AND
I. KUSHIRO
and melts formed near the solidi are picritic at 30 kbar.
5.1 Systematic changes in major elements in melts under varying pressure and temperature conditions The SiO 2 contents in the partial melts are strongly dependent on pressure (Fig. 3): they decrease with increasing pressure from 50-52 wt% at 10 kbar to 45-47 wt% at 30 kbar. At a given pressure, the SiO 2 contents are nearly constant regardless of the degree of partial melting in the present pressure-temperature range. They differ little (~< 1 wt%) between melts of HK-66 and KLB-1 under the same pressure and temperature conditions, despite the SiO 2 contents of these lherzolites being significantly different. MgO contents in the melts are, on the other hand, dependent on temperature, but they are relatively insensitive to pressure (Fig. 4): at a given pressure, they increase by 1 wt% with an increase in temperature of about 20°C. The MgO values are similar in the partial melts of the two peridotites at the same temperatures despite significantly different M g / ( M g + Fe) ratios. The F e O contents in the melts of HK-66 and KLB-1 appear to depend on pressure (Fig. 5a) and in contrast to MgO they are considerably different in the two lherzolites. At a given pressure the FeO contents in the melts increase with increasing degree of partial melting at pressures below 20 kbar, but above 25 kbar they are nearly constant or decrease (although there are no data 52
i
[]
51
i
i
i
• HK-66 Ez KLB-1
[]
50 []
49
o
48
[]
~
•
47 46 45
20 25 30 35 ~ Pressure(kbar) Fig. 3. SiO2 contents (wt%) in melts formed by partial melting of KLB-1 and HK-66. 5
l0
1~5
PARTIAL
MELTING
OF
20
,
DRY
PERIDOTITES
,
,
AT HIGH
,
485
PRESSURES
,
a.
,
20
i
,5
18
•
HK-66 [] K L B - I
I
[]
I
16
[]
o
~ ~ ~
•
14
[] •
12
•
i
i
i
,0 _
18 .,d
•
16
© <
10
14 12
8
~
•
3O
25
[] 10
6 I I I I I I 1200 1250 1300 1350 1400 1450 1500 1550
I
0
I
0.1
I
0.2
0.3
I
0.4
0.5
F
Temperature(°C) Fig. 4. M g O c o n t e n t s (wt%) in melts from the two different
b.
13
.
.
.
.
15
lherzolites. 12 10
for low degrees of partial melting ( < 10%) above 25 kbar (Fig. 5b)). The A1203 contents in melts become depleted with increasing degrees of partial melting (Fig. 6a). The AI203 contents in a.
i
i
i
i
[] KLB-1
9
; H--U~-66I -1
I
0
0.1
I
I
0.2
0.3
I
0.4
0.5
F
I
.~ 10
Fig. 6. (a) A I 2 0 3 contents (wt%) versus F (degree of partial
melting) for melts from the two lherzolites. (b) C a • contents
I
•
•
9
•
8
g
7
5
['t
[]
I:1
( w t % ) versus F (degree of partial melting) for melts from the
[]
two lherzolites.
D
g
6
o
I
I
I
I
I
10
15
20
25
30
35
Pressure(kbar) 14
i
13
• tJ
i
i
HK-66 ] KLB-I
~
I
12
L
30 •
•
25
11 25
9
10
2
8
6
L
0
0.1
0.2
0.3
I
0.4
0.5
F Fig. 5. F e O * c o n t e n t s (wt%) in melts from KLB-1 and HK-66. F (calculated degree of partial melting) is calculated from
Na20
25
7
~" ii
~"
10
i
I..-61
12
b.
i
15
8
14 13
[.l.
10
11
abundances in melt and host peridotite. The values indicate pressures.
melts of HK-66 are higher than those of KLB-1 at the same degree of partial melting and thus reflect the bulk A1203 content in the source lherzolites. Fujii and Scarfe [3] showed that the Cr/(Cr + A1) ratio of spinel correlates well with AI203 in melts. The Cr/(Cr + A1) ratio of the spinel in HK-66 is lower than that in KLB-1, which is consistent with Fujii and Scarfe's observation. With increasing pressure, the A1203 contents in melts become depleted at the same degree of partial melting; this is well observed in melts from HK-66 which are enriched in orthopyroxene. The behavior of C a • in the melts is complex: with increasing temperature, it initially increases after the depletion or elimination of spinel; it then decreases after clinopyroxene disappears (Fig. 6b). The C a • contents in the melts are thus strongly dependent on the composition of the source spinel lherzolites.
486
K. H I R O S E
51
5.2 Comparison with sandwich method Determination of the equilibrium melt compositions by the sandwich method is accompanied by the problems caused by addition of basalt to peridotite: (i) The bulk composition is changed. Enrichment in the basaltic component modifies the equilibrium melt composition, especially when the degree of melting is low. The effect of changing bulk composition on partial melt composition is significant for A1203, CaO and incompatible elements, as shown in the previous section. For example, at 20 kbar and 1375°C (run 8) the abundances of these elements determined by the sandwich method [1] are significantly different from those obtained by the present method; the K 2 0 content is 0.25 wt% in our results, compared to 1.0 wt% by the sandwich method (Table 5). In the latter experiments, alkali basalt with 1.7 wt% K20 w a s sandwiched. CaO contents in melts determined by the sandwich method [1] were also considerably affected by the added basalts. (ii) There is the possibilty that the sandwiched
melt may not be homogenized with partial melts formed apart from the sandwiched layer. Until they are homogenized, melt in the sandwiched layer would be in only local equilibrium with the surrounding peridotite. It should be also emphasized that the local bulk composition could be remarkably different from the original peridotite composition. The homogenization scale of SiO 2 in melt is about 0.3 mm in 24 h, using an interdif-
A N D I. K U S H I R O
g-
5O
o ~
49
o
o
6 ~ o
! 48
0
[
o •
T&K[1 ] This work
I
o o
o ©
©
47 ©
46 45 5
10
15
I I 310 2O 25 Pressure (kbar)
I 35
40
Fig. 7. Comparison of SiO 2 contents in melts obtained by the present experiments and those obtained by the sandwich method [1].
fusion coefficient D of ~ 10 -s cm2/s at 1300°C [19]; this scale is much smaller than the size of the capsule. The elements of low diffusivity in melts, such as SiO 2 and A1203, could not be equilibrated (Fig. 7). To minimize the effect of the bulk compositional change Falloon and Green [20] carried out sandwich experiments using basalts of compositions which were expected to be close to those of the partial melts formed from the adjacent peridotite. Their results are consistent with the present study, suggesting that the equilibrium melt composition is not modified significantly if the added basalt is chosen carefully.
5.3 Compositional trends of melts during isobaric partial melting of lherzolites TABLE 5 Comparison of a melt composition obtained by the present experiments at 20 kbar and 1375°C with that obtained by the sandwich method [1]
SiO~ TiO 2 AI20 3 FeO* MnO MgO CaO Na20 KzO Total
Sandwich method [1]
This work
47.80 1.55 15.60 9.91 0.19 11.20 9.71 3.33 1.00 100.29
47.55 1.18 15.63 10.59 0.19 12.02 9.98 2.61 0.25 100.00
The compositions of partial melts determined in this study are plotted on an ol-ne-Q ternary projection of the basalt tetrahedron (Fig. 8) and on Walker's ol-pl-qz and ol-di-qz ternary projections (Fig. 9) [21,22]. Isobaric compositional trends are recognized as single lines in these projections. As shown in these figures, the melt composition moves toward the hypersthene apex with increasing temperature at constant pressure. It is also clear that the isobaric trends shift toward the olivine apex with increasing pressure. The melts become progressively poorer in normative quartz as the pressure increases to 30 kbar. In these respects, the present results are in good
487
PARTIAL MELTING OF DRY PERIDOTITES AT HIGH PRESSURES
KLB-1 I0 15 20 25 30
kb kb kb kb kb
o *x o * o
HK-66
he*
In Walker's ol-di-qz ternary projection the results presented here do not show the marked pressure dependence of the diopside component reported by Takahashi and Kushiro [1] (Fig. 9b). The pressure dependence of the olivine component should be observed in this projection at higher degrees of partial melting after clinopyroxene disappears.
,, • • * •
6. Discussion
ol*
opx
It is very important to evaluate the effects of the source peridotite compositions on the partial melt compositions. The experiments in this paper show the following behavior of major elements in melts formed by partial melting of the two relatively fertile spinel lherzolites: (1) SiO 2 and MgO contents depend little on the compositions of the source lherzolites (Figs. 3 and 4); (2) S i O 2 c o n tents depend little on the degree of partial melting, but they strongly depend on pressure; (3) MgO contents are controlled mainly by temperature; and (4) in contrast, the abundances of FeO, A1203, CaO and incompatible elements are significantly affected by the degree of partial melting and by the compositions of the source peridotites. Fujii and Scarfe [3] have already made these observations by changing bulk compositions at 10 kbar in their experiments. It should be noted that the FeO content in the melts could be
Q*
Fig. 8. Normative compositions of melts formed by partial melting of the two different lherzolites. After the projection scheme described by Irvine and Baragar [21]: n e * = ne + 0.6ab, Q* = Q +0.4ab +0.25opx, ol* = ol +0.75opx.
agreement with those of Takahashi and Kushiro [1]. The compositions of melts formed in both HK66 and KLB-1 lie approximately along the same isobaric line at the same pressure. This result means that the above-mentioned compositional differences between the two different lherzolites do not significantly affect the isobaric trends in the normative projections. However, if the source peridotite is much more enriched in incompatible elements, such as K 2 0 , the isobaric lines should move significantly [23].
a
KLB-I HK-66
15
~
)1
\ o
\ ~
3O
if 2 10 k b
o
15 kb
zx
25 kb 30 kb
* o
•
/
ol
o )x
b
qz
ol
di
: "° opx
qz
Fig. 9. Normative compositions of melts and the isobaric compositional trends defined by the present experiments in Walker's ternary projections [22]. (a) Projection from diopside onto the plane o l - p l - q z . (b) Projection from plagioclase onto the plane o l - d i - q z . Melts projected far from the normative triangle were omitted in both projections.
488
K. HIROSE AND 1. KUSH1RO
used as a depth indicator if the M g / ( M g + Fe) ratio in the source peridotite is known [24]. These results imply that when the composition of the primary magma is known, the pressure and temperature conditions of its generation from a fertile lherzolite can be approximately estimated from SiO 2 and MgO contents. In the normative projections, the isobaric compositional trends of the melts do not shift significantly with variation in the source peridotite composition, as shown in the present study and in previous melting experiments on several fertile spinel lherzolites (Figs. 8 and 9) [1,3,4,20]. Theoretical, experimental and geochemical studies [e.g. 8-10] suggest that melt can segregate from host peridotite even when the melt fraction is very small. Models of fractional melting have recently been applied to magma genesis at midoceanic ridges [e.g. 24,25]. Yet, some magnesian MORBs have compositions that are similar to melts formed by batch partial melting at about 10 kbar, as shown in some melting experiments [3,26-28]. The present results also indicate that the partial melt formed at 10 kbar (run 2) is close in composition to one of the magnesian MORBs (519-4-2) from the FAMOUS area (Table 6) [29]. The sandwich experiments showed that compatible major elements of added basalts reach equilibrium with the surrounding peridotite within 24 h [1,3,4]. As the grain size in the mantle would be much larger than that used in the experiments, the time for equilibration in these experiments cannot be applied directly. However, the time for
TABLE 6 Compositions of magnesian M O R B 519-4-2 [29] and a partial melt formed at 10 kbar
SiO~ TiO 2 AI20 3 FeO * MnO MgO CaO Na20 K20 Cr20 3 Total
519-4-2
Partial melt (run 2)
49.38 0.74 16.32 8.99 0.16 10.44 l 1.69 2.17 0.07 0.04 100.00
49.59 0.79 16.49 8.89 0.12 10.35 11.27 2.15 0.15 0.19 100.00
equilibration should also depend on the flow rate of melt in the mantle. The time required for melts to ascend through mantle peridotite must be much longer than 24 h. This suggests that, during their ascent through the mantle peridotite, compatible elements could interact between the melts and peridotites and re-equilibrate at shallower depths [4]. The compositions of ascending melts, therefore, would approach those of melts formed by batch partial melting at shallow depths (i.e. about 30 kin).
Acknowledgements The authors thank Drs. K.T. Johnson, E. Takahashi, T. Kawamoto and S. Yamashita for discussion and comments. Drs. T.J. Falloon and S.A. Morse and an anonymous reviewer provided critical and constructive comments. The authors learned from Dr. E. Stolper that his group at the California Institute of Technology independently developed an experimental method very similar to that discussed in the present paper, and we thank Dr. M.B. Baker for sending us an abstract of their paper.
References 1 E. Takahashi and I. Kushiro, Melting of a dry peridotite at high pressures and basalt m a g m a genesis, Am. Mineral. 68 859 879, 1983. 2 E. Stolper, A phase diagram for mid-ocean ridge basalts: preliminary results and implications for petrogenesis, Contrib. Mineral. Petrol. 74, 13-27, 1980. 3 T. Fujii and C.M. Scarfe, Composition of liquid coexisting with spinel lherzolite at 10 kbar and the genesis of MORBs, Contrib. Mineral. Petrol. 90, 18-28, 1985. 4 T.J. Falloon, D.H. Green, C.J. Hatton and K.L. Harris, Anhydrous partial melting of a fertile and depleted peridotite from 2 to 30 kb and application to basalt petrogenesis, J. Petrol. 29, 1257-1282, 1988. 5 K.T. Johnson and I. Kushiro, Segregation of high pressure partial melts from peridotite using aggregates of diamond: a new experimental approach, Geophys. Res. Lett. 19, 1703 1706, 1992. 6 I. Kushiro and K. Hirosc, Experimental determination of composition of melt formed by equilibrium partial melting of peridotite at high pressures using aggregates of diamond grains, Proc. Jpn. Acad. 68, 63-68, 1992. 7 M.B. Baker, S. Newman, J.R. Beckett and E. Stolper, Separating liquid from crystals in high-pressure melting experiments using diamond aggregates, Geol. Soc. Am. Abstr., 1992.
PARTIAL MELTING OF DRY PERIDOTITES AT HIGH PRESSURES
489
8 G.N. Riley Jr. and D.L. Kohlstedt, Kinetics of melt migration in upper mantle-type rocks, Earth Planet. Sci. Lett. 105,500-521, 1991. 9 D. Mckenzie, 23°yh-Z3Su disequilibrium and the melting processes beneath ridge axes, Earth Planet. Sci. Lett. 72, 149-157, 1985. 10 K.T. Johnson, H.J. Dick and N. Shimizu, Melting in the oceanic upper mantle: an ion microprobe study of diopsides in abyssal peridotites, J. Geophys. Res. 95, 26612678, 1990. 11 I. Kushiro, Y. Syono and S. Akimoto, Melting of a peridotite nodule at high pressures and high water pressures, J. Geophys. Res. 73, 6023-6029, 1968. 12 E. Takahashi, Melting of dry peridotite KLB-1 up to 14 GPa: implications for the origin of peridotitic upper mantle, J. Geophy. Res. 91, 9367-9382, 1986. 13 I. Kushiro, A new furnace assembly with a small temperature gradient in solid-media, high-pressure apparatus, Carnegie Inst. Washington Yearb. 75, 832-833, 1976. 14 1. Kushiro, Compositions of magmas formed by partial zone melting of the earth's upper mantle, J. Geophys. Res. 73, 619-634, 1968. 15 R.J. Kinzler and T.L. Grove, Primary magmas of mid-ocean ridge basalts 1. Experiments and methods, J. Geophys. Res. 97, 6885-6906, 1992. 16 S. Maal0e and K. Aoki, The major element composition of the upper mantle estimated from the compositions of lherzolites, Contrib. Mineral. Petrol. 63, 161 173, 1977. 17 E. Takahashi, T. Shimazaki, Y. Tsuzaki and H. Yoshida, Melting study of peridotite KLB-I to 6.5 GPa and origin of basaltic magmas, Philos. Trans. R. Soc. London, in press, 1992. 18 H. Kuno, High-alumina basalt, J. Petrol. 1, 121-145, 1960. 19 T. Koyaguchi, Chemical gradient at diffusive interfaces in
magma chambers, Contrib. Mineral. Petrol. 103, 143-152, 1989. T.J. Falloon and D.H. Green, Anhydrous partial melting of MORB pyrolite and other peridotite compositions at 10 kbar: Implications for the origin of primitive MORB glasses, Mineral. Petrol. 37, 181-219, 1987. T.N. Irvine and W.R. Baragar, A guide to the chemical classification of the common volcanic rocks, Can. J. Earth. Sci. 8, 523-548, 1971. D. Walker, T. Shibata and S.E. DeLong, Abyssal tholeiites from the Oceanographer Fracture Zone II: phase equilibria and mixing, Contrib. Mineral. Petrol. 70, 111-125, 1979. R.J. Sweeney, T.J. Falloon, D.H. Green and Y. Tatsumi, The mantle origins of Karoo picrites, Earth Planet. Sci. Lett., 107, 256-271, 1991. E.M. Klein and C.H. Langmuir, Global correlations of ocean ridge basalt chemistry with axial depth and crustal thickness, J. Geophys. Res. 92, 8089-8115, 1987. D. Mckenzie and M.J. Bickle, The volume and composition of melt generated by extension of lithosphere, J. Petrol. 29, 625-679, 1988. I. Kushiro and R.N. Thompson, Origin of some abyssal tholeiites from the Mid-Atlantic Ridge, Carnegie Inst. Washington Yearb. 71,403-406, 1972. T. Fujii and H. Bougault, Melting relations of a magnesian abyssal tholeiite and the origin of MORBs, Earth Planet. Sci. Lett. 62, 283-295, 1983. T. Fujii, Genesis of mid-ocean ridge basalts, in: Magmatism in the Ocean Basins, A.D. Saunders and M.J. Norry, eds., Geol. Soc. London Spec. Publ. 42, 137-146, 1989. W.B. Bryan and J.G. Moore, Compositional variations of young basalts in the Mid-Atlantic Ridge rift valley near lat 36°49'N, Geol. Soc. Am. Bull. 88, 556-570, 1977.
20
21
22
23
24
25
26
27
28
29