Journal of Petrology Advance Access originally published online on February 25, 2005
Journal of Petrology 2005 46(6):1283-1308; doi:10.1093/petrology/egi017
© The Author 2005. Published by Oxford University Press. All rights reserved. For Permissions, please email: journals.permissions@oupjournals.org
Using In Situ Trace-Element Determinations to Monitor Partial-Melting Processes in Metabasites
A. C. STORKEY1,
J. HERMANN2,
M. HAND3 and
I. S. BUICK4,*
1 DEPARTMENT OF EARTH SCIENCES, LA TROBE UNIVERSITY, BUNDOORA, VIC 3086, AUSTRALIA
2 RESEARCH SCHOOL OF EARTH SCIENCES, AUSTRALIAN NATIONAL UNIVERSITY, CANBERRA, ACT 0200, AUSTRALIA
3 DEPARTMENT OF GEOLOGY AND GEOPHYSICS, UNIVERSITY OF ADELAIDE, ADELAIDE, SA 5005, AUSTRALIA
4 SCHOOL OF GEOSCIENCES, MONASH UNIVERSITY, CLAYTON, VIC 3800, AUSTRALIA
RECEIVED
JANUARY 25, 2004;
ACCEPTED
JANUARY 20, 2005
 |
ABSTRACT
|
|---|
Peak metamorphism (
800–850°C, 8–10 kbar) in
the Harts Range Meta-Igneous Complex (Harts Range, central Australia)
was associated with localized partial melting by the reaction
hornblende + plagioclase + quartz + H
2O = garnet + clinopyroxene
+ titanite + melt.
In situ trace-element determinations of prograde,
peak and retrograde minerals in migmatitic metabasites and associated
tonalitic melts using laser-ablation ICP–MS has allowed
monitoring of a range of partial-melting processes (melting,
melt segregation and back-reaction between crystallizing melt
and restitic minerals). Mass balance calculations indicate that
titanite is a major carrier of trace elements such as Ti, Nb,
Ta, Sm, U and Th, and therefore may be an important accessory
phase to control the redistribution of these elements during
the partial melting of amphibolites. Titanite preferentially
incorporates Ta over Nb and, hence, residual titanite might
assist in the formation of melts with high Nb/Ta. The fact that
single minerals record different rare earth element (REE) patterns,
from prograde to peak to retrograde conditions, demonstrates
that REE diffusion is not significant up to
800°C. Therefore,
trace-element analysis in minerals can be a powerful tool to
investigate high-grade metamorphic processes beyond the limits
given by major elements.
KEY WORDS: Harts Range; laser-ablation ICP–MS; metabasites; partial melting; trace elements
|
INTRODUCTION
|
|---|
It is believed that a significant portion of the early continental
crust was formed by partial melting of amphibolites or eclogites
to generate tonalite–trondhjemite–granodiorite (TTG)
gneisses (Kröner, 1985
; Jahn
et al., 1988
; Rapp
et al.,
1991
; Martin, 1994
; Foley
et al., 2002
). Studies have also shown
that lower and middle crustal amphibolites may be potential
sources of felsic melts, particularly tonalites and trondhjemites,
with this process providing a mechanism for producing very dense
garnet-bearing granulites as restites (Percival, 1983
; Rapp
et al., 1991
; Wolf & Wyllie, 1993
, 1994
).
During the partial melting of mafic rocks, amphibole is generally regarded to control the distribution of trace elements such as Ti, Nb, Ta, Zr, Sm, Th and U (Foley et al., 2002). The role of accessory minerals such as titanite or zircon, which also have high concentrations of these important trace elements, is generally not considered. However, Green & Pearson (1987) noted that Nb and Ta readily partition into titanite and rutile, and the presence of only accessory amounts of these minerals will have a significant effect on the Nb, Ta and Nb/Ta ratio of the rock and any derivative magmas.
It is ineffective to use major elements in minerals to monitor partial-melting processes in high-grade, upper amphibolite- to granulite-facies metabasic rocks because they may re-equilibrate via volume diffusion under peak and retrograde P–T conditions, therefore destroying compositional evidence of the prograde melting reaction (Tracy, 1982; Barnes & Carlson, 2001). However, recent studies of trace-element distributions in granulite-facies garnets have shown that REE display strong zoning and hence can be used to infer high-grade metamorphic processes such as partial melting (Ottamendi et al., 2002; Hermann & Rubatto, 2003). This is in agreement with experimental studies indicating that diffusion of trivalent REE is at least two orders of magnitude slower than that of divalent cations such as Fe and Mg (Van Orman et al., 2002).
To test the effectiveness of trace elements as monitors of melt-related processes, we have used laser ablation ICP–MS to determine trace-element concentrations for minerals in situ for a suite of amphibolites, their migmatitic equivalents and associated tonalitic melts from Mt Ruby in the Harts Range of central Australia (Fig. 1). The Mt Ruby metabasites experienced high-grade metamorphism and partial melting at 800°C and 10 kbar. They represent an excellent natural laboratory to study partial melting of amphibolites at the base of the continental crust or in shallow hot subduction zones, which is important for the better understanding of the formation of the early continental crust. Based on textural relationships and mineral trace-element determinations of the Mt Ruby metabasites, prograde, peak and retrograde minerals can be identified. The major-element concentrations of these minerals show little variation, but changes in the trace-element concentrations have enabled us to monitor partial-melting processes. In addition, mass balance calculations highlight the importance of accessory phases as trace-element carriers during partial melting and melt crystallization.
View larger version (100K):
[in this window]
[in a new window]
|
Fig. 1. Map of the regional geology, showing the location of the Harts Range within the Arunta Inlier (inset), the location of Mt Ruby (study area) in the southern Harts Range, and the location of the Ruby Mine (Sivell & Foden, 1985) and Mt Mabel (Sivell, 1988). Abbreviations in inset are as follows: AB, Amadeus Basin; AS, Alice Springs; CZ, Central Zone; GB, Georgina Basin; NB, Ngalia Basin; NZ, Northern Zone; SZ, Southern Zone; TCB, Tennant Creek Block; WB, Wiso Basin.
|
|
|
GEOLOGICAL SETTING
|
|---|
The Harts Range (
Fig. 1) occurs in the Arunta Inlier of central
Australia and comprises three lithostratigraphic units: the
Neoproterozoic–Cambrian (Buick
et al., 2001
) Harts Range
Metamorphic Complex (HRMC), the structurally underlying Palaeoproterozoic
Entia Gneiss Complex (EGC), and the Palaeoproterozoic Bruna
Granitic Gneiss, which separates the HRMC from the EGC. The
HRMC can be divided into the Irindina Supracrustal Association
(ISA) and the Harts Range Meta-Igneous Complex. The ISA contains
four metasedimentary divisions and mostly comprises metapelite,
calc-silicate, marble and quartzofeldspathic gneiss.
The Harts Range Meta-Igneous Complex is the focus of this study, and consists mainly of medium- to coarse-grained granoblastic amphibolites interlayered with anorthositic gneisses and ultramafic rocks (Sivell & Foden, 1985; Sivell, 1988). These occur within laterally extensive interlayered quartzose, calcareous and metapelitic schists and gneisses of the ISA (Sivell, 1988). The completely recrystallized nature of the amphibolites, and the local occurrence of concordant calc-silicate bands in the Ruby Mine area and at Mt Mabel (Fig. 1), indicate that these rocks originated as fine-grained lava flows or sills (Sivell & Foden, 1985; Sivell, 1988). This study looks in detail at the Harts Range Meta-Igneous Complex at Mt Ruby in the southern Harts Range (Fig. 1). Geochemical work on a large set of whole-rock samples from Mt Ruby has shown that the protoliths of the amphibolites were tholeiitic basalts that probably formed during strong crustal thinning at 520 Ma (Storkey, 2004; Storkey et al., in preparation).
Peak metamorphism in the HRMC during the Ordovician (480–460 Ma; Hand et al., 1999; Mawby et al., 1999; Buick et al., 2001) reached granulite-facies conditions (800–850°C, 8–10 kbar; Miller et al., 1997; Hand et al., 1999; Mawby et al., 1999). Thermobarometry of samples from Mt Ruby yield P–T estimates of 780–830°C and 10–11 kbar (Mawby et al., 1999; Mawby, 2000). This granulite-facies metamorphism led to partial melting of the amphibolites and produced garnet–clinopyroxene-rich segregations that are associated with a tonalite melt. Detailed field observations, combined with geochemistry (Storkey, 2004; Storkey et al., in preparation), indicate that highly localized partial melting took place under fluid-present conditions according to the following generalized melting reaction:
 | (1) |
The presence of unmelted amphibolites
in direct contact with partial melt zones containing large garnet
and clinopyroxene grains, as well as tonalite dykelets, makes
the Mt Ruby area ideal for studying how trace elements are redistributed
and how mineral trace-element compositions respond to partial
melting of amphibolites.
|
SAMPLE DESCRIPTION
|
|---|
Field relations
This study looks in detail at a 200 m across-strike creek section
of the Harts Range Meta-Igneous Complex at Mt Ruby (
Fig. 2).
Five representative samples were chosen for detailed petrographic
and geochemical analysis to provide information on the prograde,
peak and initial retrograde processes of partial melting in
the Mt Ruby amphibolites. The main rock type found at Mt Ruby
is a strongly foliated (S
1) amphibolite that displays some compositional
layering parallel to the foliation. Along the investigated creek
section, the foliated amphibolite grades over 2 m into zones
that host patchy, locally discordant, coarse-grained segregations,
rich in garnet and clinopyroxene, and associated with a leucocratic
quartz and plagioclase-rich matrix. Garnet in these segregations
is idiomorphic and reaches up to 15 cm in diameter, whereas
clinpopyroxene always forms smaller grains. The big garnet and
smaller clinopyroxene crystals represent products of the peritectic
melt production [
reaction (1)], whereas the plagioclase- and
quartz-rich matrix represents remnants of trapped tonalitic
melt (Mawby, 2000
; Storkey, 2004
). In places where the peak
metamorphic garnet + clinopyroxene-bearing assemblages have
been preserved, the majority of melt is thought to have migrated
away from the system, leaving a largely anhydrous or restitic
assemblage that was unable to undergo significant back-reaction
(
Fig. 3a and b). In places where some partial melt stayed in
the system, the peak garnet + clinopyroxene-bearing assemblages
have been partially to extensively replaced by hornblende during
retrograde back-reaction with the melt (
Fig. 3c). Leucocratic
layers (up to 1 m wide) of broadly tonalitic composition interleaved
with unmigmatized amphibolites are also found at Mt Ruby (
Fig. 3d).
Locally, these tonalites crosscut the S
1 foliation or are
found in necks of interfoliar boudins in the amphibolites.
View larger version (53K):
[in this window]
[in a new window]
|
Fig. 2. Map of the creek section on the southern flank of Mt Ruby, showing the schematic structural context and the location of samples discussed in the text. Tonalite sample 02-180 was collected from a small creek 500 m to the east. Grey shaded areas represent extent of outcrop. Numbers 1–7 indicate zones: (1) metabasalt containing sparse peak-metamorphic garnet–clinopyroxene-bearing restitic segregations (black fill); (2) unmigmatized metabasalt; (3) interlayered tonalitic segregations (diagonal fill) and unmigmatized metabasalt; (4) high-strain zone containing abundant peak-metamorphic garnet–clinopyroxene-bearing restitic segregations within a foliated hornblende–plagioclase–quartz matrix; (5) unmigmatized metabasalt with sparse restitic segregations near the base, grading upwards into rehydrated amphibole–garnet segregations near the top of the unit; (6) unmigmatized metabasalt; (7) interlayered tonalitic melt and unmigmatized metabasalt. Grid numbers refer to the Australian Map Grid, Zone 53, and are taken from the Quartz 1:100 000 scale geological map, Bureau of Mineral Resources (1990).
|
|
View larger version (143K):
[in this window]
[in a new window]
|
Fig. 3. (a) Peak-metamorphic garnet + clinopyroxene-bearing segregation parallel to the S1 foliation in a high-strain zone. (b) Peak-metamorphic garnet + clinopyroxene-bearing segregation truncating the S1 foliation. (c) Migmatitic segregation containing hornblende partially replacing the peak garnet + clinopyroxene assemblage. (d) Tonalitic melt layer interfingered with the host metabasite. All mineral abbreviations after Kretz (1983). Width of lens cap is 5 cm.
|
|
Petrography
The unmigmatized amphibolite (02-169) contains quartz, hornblende,
clinopyroxene, plagioclase, titanite, ilmenite and apatite in
a granoblastic polygonal mosaic (
Fig. 4a). This sample is used
to characterize the trace-element distribution prior to partial
melting.
View larger version (155K):
[in this window]
[in a new window]
|
Fig. 4. (a) Unmigmatized metabasite (02-169); (b) coarse-grained peak garnet + clinopyroxene-bearing migmatitic segregation (01-292); (c) large replacement hornblende with relic garnet from the rehydrated amphibole–garnet segregation (02-174); (d) plagioclase-rich tonalitic segregation (02-180). All mineral abbreviations after Kretz (1983).
|
|
Samples 01-292 and 02-132 are peak metamorphic garnet-rich segregations.
They contain a coarse-grained garnet + clinopyroxene + plagioclase
+ quartz + hornblende assemblage, with minor titanite, ilmenite,
apatite and zircon (
Fig. 4b), that is interpreted to have formed
by partial-melting
reaction (1). The cores of some large garnets
contain abundant small inclusions of clinopyroxene, as well
as numerous larger inclusions of hornblende, clinopyroxene,
plagioclase, quartz, titanite, ilmenite, apatite and zircon.
Rims of these garnets are relatively inclusion-free; they contain
some larger inclusions of hornblende, clinopyroxene, titanite
and ilmenite, but lack the numerous small inclusions seen in
garnet cores. Plagioclase occurs as inclusions in garnet cores
but not rims. Titanite and ilmenite occur both as inclusions
in garnet as well as within the matrix, indicating that they
were stable through the major part of the metamorphic evolution.
These two samples best document the processes related to peak
metamorphic partial melting of the amphibolite.
The amphibole–garnet segregation (sample 02-174) contains coarse-grained hornblende that extensively replaces a peak metamorphic assemblage of garnet + clinopyroxene. The assemblage is now hornblende + clinopyroxene + plagioclase + quartz, with minor titanite, ilmenite, apatite and zircon, and small relict garnet present in larger hornblende grains because of the incomplete reversal of melting reaction (1) (Fig. 4c). Relict garnet contains inclusions of clinopyroxene, hornblende, plagioclase, quartz and titanite. As well as relict garnet, the replacement hornblende contains inclusions of plagioclase and titanite. This sample is mainly used to constrain processes following peak metamorphism related to the consumption of garnet.
Sample 02-180 is a tonalite containing a mineral assemblage of hornblende + plagioclase + quartz, with minor titanite + ilmenite + apatite + zircon (Fig. 4d). Hornblende in the tonalite contains inclusions of plagioclase, quartz, titanite, apatite and zircon. This sample provides information on the crystallization of melts at or after peak metamorphic conditions.
Figure 5 shows a summary of the occurrence of the different minerals in relation to the observed metamorphic stages. The wide stability of titanite and amphibole during prograde, peak and retrograde metamorphism makes them suitable to investigate whether or not there are distinct trace-element signatures in these minerals as a function of varying coexistent phases and metamorphic conditions.
View larger version (10K):
[in this window]
[in a new window]
|
Fig. 5. Paragenesis table of mineral growth during the prograde, peak and retrograde stages of the metamorphic history of the Mt Ruby metabasites. All mineral abbreviations after Kretz (1983).
|
|
|
ANALYTICAL PROCEDURES
|
|---|
Major-element analysis of minerals in polished thin section
(
30 µm thick) was carried out at the University of Melbourne
on a Cameca SX50 electron microprobe equipped with four wavelength-dispersive
spectrometers (WDS). Analytical conditions were 15 kV and 25
nA. Using the same polished thin sections, trace-element analyses
of the same minerals were performed
in situ on the laser-ablation
ICP–MS at the Research School of Earth Sciences (RSES)
at the Australian National University (ANU) in Canberra. A pulsed
193 nm ArF Excimer laser at 100 mJ energy and a repetition rate
of 5 Hz coupled to an Agilent 7500 quadrupole ICP–MS was
used for ablation. During the time-resolved analysis of minerals,
the contamination from inclusions, fractures and zones of different
composition was detected by monitoring several elements and
integrating only the ‘clean’ part of the signal.
The relative 1
standard deviation of multiple mineral analyses
is generally
5–20%. Spot size was 86 µm for most
minerals and 24 µm for smaller inclusions. A NIST-612
glass was used as standard material and values were taken from
Pearce
et al. (1997)
. Si and Ca determined by electron microprobe
(minerals) and XRF (bulk-rock data) were used as an internal
standard for all minerals except ilmenite, for which Ti was
used.
Powders for bulk-rock analyses were obtained by grinding the sample in a tungsten carbide mill to a grain size of <25 µm. The powders were then made into fused glass discs using a La2O3-doped lithium borate flux (0·84 g of sample to 4·5 g of flux), and major-element determinations were carried out on the Siemens SRS303AS XRF spectrometer (XRFS) with a Rh end-window X-ray tube at La Trobe University. The same rock powders were used for trace-element whole-rock determinations. These powders were made into fused glass discs using a REE-free lithium borate (Sigma® 12:22 X-Ray flux; sample:flux = 1:2) at the University of Stellenbosch, South Africa. They were then analysed for trace elements in the Laser-Ablation ICP–MS facility at the RSES at ANU using a 142 µm spot size, with the resulting data being an average of five ablation spots.
|
BULK-ROCK COMPOSITIONS
|
|---|
Detailed descriptions of bulk rocks have been given by Storkey
(2004)
. Here, we briefly discuss the bulk-rock compositions
of the five samples on which detailed mineral trace-element
work was done (
Fig. 6;
Table 1).
View larger version (22K):
[in this window]
[in a new window]
|
Fig. 6. Bulk-rock chondrite-normalized (Sun & McDonough, 1989) REE patterns for the five samples from this study.
|
|
The unmigmatized amphibolite has a relatively flat REE pattern
at
20–30 times chondrite, with a slight LREE enrichment
and a small negative Eu anomaly (Eu/Eu
* = 0·86). Such
a pattern is in agreement with tholeiitic basalts as the protolith
for the amphibolites.
The garnet-rich segregations are characterized by increasing HREE contents with increasing atomic number. Sample 01-292 is significantly more enriched in HREE (Yb 
90 times chondrite) than 02-132 (Yb 50 times chondrite). The HREE enrichment of these segregations with respect to the unmigmatized rock confirms field observations that these rocks are strongly restitic and must have lost a significant melt component. The garnet-rich segregations have slightly LREE-enriched patterns (LaN/SmN = 1·2–1·4), and no appreciable Eu anomalies (Eu/Eu* = 0·89–1·02). The LREE enrichment combined with the presence of large ion lithophile elements (LILE) suggest that a minor amount of trapped melt is present, in agreement with major-element analyses of these rocks (Storkey, 2004; Storkey et al., in preparation). The amphibole–garnet segregation has a similar LREE pattern to the restites. The REE pattern shows a more pronounced negative Eu anomaly (Eu/Eu* = 0·75), and HREE enrichment, with Yb 50 times chondrite, similar to sample 02-132. This pattern indicates that the amphibole–garnet segregation represents a garnet-rich restite that trapped a significant amount of melt. Field observations might suggest that the amphibole–garnet segregations represent partially molten amphibolites, where the melt was not able to escape. However, the significant difference in the bulk-rock composition when compared with the unmigmatized amphibolite demonstrates that this is not the case.
The tonalite has a slightly LREE-enriched pattern, no Eu anomaly (Eu/Eu* = 0·96) and a mostly flat M–HREE pattern that lies at 10 times chondrite. The tonalite has markedly lower concentrations of REE than the unmigmatized rock, and the two types of segregations. It is also worth noting that the concentration of ‘melt mobile’ elements such as Rb, Ba and U are lower in the tonalite than in the amphibolite.
|
MINERAL MAJOR- AND TRACE-ELEMENT COMPOSITIONS
|
|---|
Unmigmatized rock (02-169)
Major elements
Representative major-element compositions for minerals from
the unmigmatized rock are presented in
Table 2. Amphibole from
the unmigmatized sample is ferroan pargasitic hornblende [6·3–6·5
Si cations per 23 oxygen formula unit (p.f.u.)], according to
the nomenclature of Leake (1978)
. Hornblende grains show little
zoning, with cores having only slightly higher
XMg [Mg/(Mg +
Fe) = 0·56] than rims (0·55) and indistinguishable
Ti contents. Clinopyroxene shows no internal zoning of major
elements and a
XMg of 0·67. Plagioclase is andesine in
composition (An
41–44) and some plagioclase grains show
minor zoning with a decrease in anorthite content from core
to rim. Titanite and ilmenite are unzoned in major elements.
View this table:
[in this window]
[in a new window]
|
Table 2: Representative major- (electron microprobe, wt %) and trace-element (laser-ablation ICP–MS, ppm) concentrations of minerals from the unmigmatized rock (02-169)
|
|
Trace elements
Representative trace-element compositions for minerals from
the unmigmatized rock are given in
Table 2, and chondrite-normalized
REE patterns are presented in
Fig. 7. Hornblende from the unmigmatized
rock has a slightly depleted LREE pattern, a small negative
Eu anomaly and a flat HREE pattern (
Fig. 4a). Clinopyroxene
REE patterns show variable Eu anomalies and slight HREE enrichment
(
Fig. 4b). Titanite has by far the highest REE contents and
displays a slight decrease from LREE to HREE, with variable
concentrations of LREE and a small negative Eu anomaly (
Fig. 4c).
The variation of LREE might indicate that titanite grew
at different stages during the prograde metamorphism. Additionally,
titanite hosts significant amounts of high field strength elements
(HFSE), and Th and U. Plagioclase from the unmigmatized rock
is LREE enriched with a positive Eu anomaly (
Fig. 4d) and contains
high amounts of Sr and Ba. HREE concentrations were below the
detection limits. All of the analysed minerals display equilibrium
textures and are chemically unzoned, indicating that they are
in also in chemical equilibrium. We were not able to analyse
apatite in this sample because it ‘exploded’ under
the laser beam. In addition, ilmenite was too small to obtain
reasonable analyses. Ilmenite analyses are reported from the
peak-metamorphic samples and the tonalite, and apatite could
be analysed only in the tonalite.
View larger version (23K):
[in this window]
[in a new window]
|
Fig. 7. Chondrite-normalized (Sun & McDonough, 1989) REE patterns for mineral cores (filled circles) and rims (open circles) from the unmigmatized rock (02-169): (a) hornblende; (b) clinopyroxene; (c) titanite; (d) plagioclase. Dashed lines in (d) indicate the inferred pattern when trace-element concentrations were below the detection limit.
|
|
Peak garnet-rich segregations (01-292 and 02-132)
Major elements
Representative major-element compositions for minerals from
the peak-metamorphic garnet + clinopyroxene-bearing restites
are presented in
Table 3. Garnet has an almandine–grossular-rich
composition (Alm
46–57Gr
23–36 Prp
15–21Sps
2–5).
Core-to-rim analyses reveal minor zoning of the major elements
in most garnet grains (
Fig. 8a). Large garnets (up to 10 mm)
show relatively flat mol % end-member profiles in the core (over
a distance of
4–5 mm) and a sharp symmetrical increase
in almandine, with corresponding decreases in pyrope and
XMg (from 0·30 to 0·21), within 800–1100 µm
of the rim. Some garnets also show a small increase in spessartine
at the rim. The relatively flat compositional profiles in the
cores of garnets from the restites can be attributed to high-temperature
homogenization, as has been widely observed in high-temperature
(above
750°C) garnets elsewhere (Tracy, 1982
; Barnes &
Carlson, 2001
). The sharp increase in almandine and decrease
in pyrope and
XMg near the rim are probably caused by volume
diffusion resetting during cooling (Tracy, 1982
; Kotkova &
Harley, 1998
), presumably because of Fe–Mg exchange with
hornblende and/or clinopyroxene.
View larger version (15K):
[in this window]
[in a new window]
|
Fig. 8. (a) Major-element profile through a large garnet from garnet-rich restite 02-132; (b) trace-element composition profile through the same garnet as illustrated in (a). The larger symbols refer to the garnet composition found at the outermost rim of the large garnet.
|
|
View this table:
[in this window]
[in a new window]
|
Table 3: Representative major- (electron microprobe, wt %) and trace-element (laser-ablation ICP–MS, ppm) concentrations of minerals from the garnet-rich segregations (01-292 and 02-132)
|
|
Amphibole in the restites is ferroan pargasitic and edenitic
hornblende (Leake, 1978
), with Si ranging from 6·29 to
6·56 cations p.f.u.
XMg values range from 0·51
to 0·58 and overlap with
XMg values from hornblende in
the unmigmatized rock. Individual hornblende grains show small
internal major-element zonation. This tends to be variable,
with hornblende from 02-132 showing an increase in Ti p.f.u.
(from 0·21 to 0·27) and a corresponding increase
in
XMg (0·54 to 0·56) from core to rim, whereas
01-292 shows a decrease in Ti p.f.u. (0·23 to 0·20)
and a corresponding decrease in
XMg (0·52 to 0·51)
from core to rim.
Clinopyroxene has an XMg between 0·64 and 0·72. The Al content in matrix clinopyroxene generally decreases from core (0·10 cations p.f.u.) to rim (0·08 cations p.f.u.). Clinopyroxene inclusions within garnet have Al contents (0·08–0·14 cations p.f.u.) that are slightly higher than, but overlapping with, clinopyroxene from the matrix. Plagioclase is andesine in composition (An44–49) and has a higher anorthite content than the unmigmatized host rock (An41–44). Most plagioclase grains show minor zoning, with a decrease in anorthite from core to rim, similar to the unmigmatized rock. Major-element compositions of titanite and ilmenite grains are unzoned. Titanite has Al contents ranging from 0·04 to 0·05 p.f.u., which is higher than in the unmigmatized rock. There is no apparent compositional distinction between titanite as inclusions in garnet and matrix titanite. Ilmenite has a Mn content of <0·01 cations p.f.u., which is considerably lower than Mn contents in ilmenite in the unmigmatized rock (0·05).
Trace elements
Representative trace-element compositions for minerals from the restites are given in Table 3. Garnet from the restite samples has typical chondrite-normalized HREE-enriched patterns (Fig. 9a). Garnets have very low concentrations of Rb, Sr and the LREE, and these show little zonation. They do not have a noticeable negative Eu anomaly and cores locally display even a small positive Eu anomaly. This indicates that garnet formed at a time when no significant amount of plagioclase was present. Plagioclase strongly fractionates Eu and hence minerals coexisting with plagioclase display a negative Eu anomaly. This suggests that plagioclase reacted away in order to produce garnet and melt. Trace elements that are more compatible in garnet, such as Ti, Y and the HREE, display complex zoning patterns from core to rim of large garnets. The cores in some garnets show an outward increase in Ti (from 500 to 1400 ppm; Fig. 8b), whereas others have relatively flat Ti profiles. Garnet rims always display a strong decrease in Ti to below 500 ppm and the outermost rim contains only 200 ppm Ti. HREE and Y display sympathetic patterns from core to rim, as illustrated by a Y profile over a large garnet (Fig. 8b). Y is generally constant from core to rim at a 100–150 ppm level, but locally shows a slight increase within these values from core to rim. The outer part of the rim then shows a complex evolution, with first a decrease in the Y content to 60 ppm, before a final increase to the highest levels observed (260 ppm). The highest values of Y and REE in garnet rims are always associated with the lowest Ti contents. The difference between core and rim analyses is also well documented in a Yb vs Ti diagram (Fig. 10). Similar to Y, the highest Yb contents are found in rims that show Ti contents of <500 ppm.
View larger version (42K):
[in this window]
[in a new window]
|
Fig. 9. Chondrite-normalized (Sun & McDonough, 1989) REE patterns for minerals from the peak metamorphic garnet + clinopyroxene-bearing restites (01-292 and 02-132): (a) garnet; (b) hornblende (01-292); (c) hornblende (02-132); (d) clinopyroxene; (e) titanite; (f) plagioclase. Symbols used for (b)–(f): triangles, inclusions within garnet; circles, matrix minerals; filled symbols, cores; open symbols, rims of minerals. Dashed lines in (f) indicate the inferred pattern when trace-element concentrations were below the detection limit.
|
|
View larger version (26K):
[in this window]
[in a new window]
|
Fig. 10. Ti vs Yb plot showing garnet compositions from the peak metamorphic garnet + clinopyroxene-bearing restites (01-292 and 02-132) and rehydrated amphibole–garnet segregation (02-174).
|
|
Hornblende from the garnet-rich segregations has extremely variable
trace-element compositions in the different textural sites (
Fig. 9b and c).
Hornblende inclusions within garnet show a strong
HREE depletion, indicating that these grains formed in equilibrium
with peak metamorphic garnet. Inclusions in sample 01-292 (
Fig. 9b)
also display a significant Eu anomaly and a LREE depletion
with respect to MREE. This pattern is quite different from patterns
in the matrix as well as hornblende from the unmigmatized amphibolite,
suggesting that the inclusions might preserve a prograde amphibole
composition that is not seen in the completely recrystallized,
strongly foliated amphibolites. In fact, the observed LREE depletion
suggests that these amphiboles were in equilibrium with an LREE-rich
mineral such as epidote or allanite (Hermann, 2002
), which is
stable in greenschist- to amphibolite-facies conditions in mafic
rocks. Matrix hornblende generally does not display any negative
Eu anomaly and is variably depleted in HREE. It is remarkable
that one analysed hornblende displays a clear zoning from core
to rim, with the core preserving a similar pattern to the inclusions
in garnet. Hornblende from sample 02-132 (
Fig. 9c) displays
additional complexity. Some of the cores display a positive
Eu anomaly, indicating that they might have formed at the expense
of plagioclase, whereas the rims lack any Eu anomaly, indicating
that they did not coexist with plagioclase. However, some of
the amphibole rims display a clear negative Eu anomaly, suggesting
that plagioclase was again a stable phase during the retrograde
evolution, when melts started to crystallize (
Fig. 5). The large
variation in amphibole composition as a function of textural
position suggests that the trace-element composition of amphibole
was not completely homogenized and that prograde trace-element
zoning patterns are able to survive. Additionally, it demonstrates
that amphibole was not completely reacted away during partial
melting of the amphibolites.
REE patterns of clinopyroxene in the restites (Fig. 9d) are characterized by either no, or small, positive Eu anomalies and HREE depletion, with the clinopyroxene inclusion in garnet from 01-292 having a stronger HREE depletion than matrix clinopyroxene. Titanite inclusions within garnet (Fig. 9e) are also strongly HREE depleted and have either no, or positive, Eu anomalies. Titanite in the matrix is less HREE depleted and also has similar Eu anomalies. The matrix titanite and clinopyroxene display higher HREE contents and probably formed together with the outermost garnet rims that are also enriched in HREE. Apart from high REE contents, titanite also contains significant amounts of V (400–500 ppm), Zr (300–450 ppm), Nb (80–200 ppm), Ta (6–20 ppm), Th (50–500 ppm) and U (10–100 ppm). Ilmenite in the two samples has low trace-element contents close to the detection limit, apart from V (200–600 ppm), Zr (2–3 ppm) Nb (20–60 ppm) and Ta (2–8 ppm). It is interesting to note that ilmenite contains lower Nb and Ta contents than titanite, although its Ti content is significantly higher. Also, there seems to be a bimodal composition of ilmenite. One group is characterized by overall low Nb contents and high Nb/Ta, similar to amphibole, whereas the other has high Nb contents and low Nb/Ta, similar to titanite. This suggests that ilmenite can form as a reaction product of either amphibole or titanite and inherit the Nb/Ta of the precursor mineral. Plagioclase has a strong positive Eu anomaly, displays a decrease of REE with increasing atomic number, and HREE are mainly below the detection limit.
Amphibole–garnet segregation (02-174)
Major elements
Representative major-element compositions for minerals from the hornblende-dominated segregation are presented in Table 4. Garnet is almandine–grossular-rich (Alm52–57Gr22–30Prp15–17Sps4–7) and has slightly higher almandine, and significantly higher spessartine and lower pyrope contents than garnet from the restites. Minor zoning of the major elements is evident in larger garnets, with an increase in almandine and a decrease in grossular and XMg from core to rim. This is similar to garnet from the restites and is therefore interpreted to represent retrograde diffusion during cooling. Some garnets show a reversal of this zoning pattern at the outer rims of the grain. XMg in garnet ranges from 0·21 to 0·24. The Mn in garnet generally increases from core to rim.
View this table:
[in this window]
[in a new window]
|
Table 4: Representative major- (electron microprobe, wt %) and trace-element (laser-ablation ICP–MS, ppm) compositions of minerals from the amphibole–garnet segregation (02-174)
|
|
The amphibole is mostly ferroan pargasitic hornblende (Leake,
1978
), with Si ranging from 6·39 to 6·57 cations
p.f.u. Amphibole has
XMg between 0·50 and 0·54,
which overlaps with hornblende from the restite (
XMg 0·51–0·58).
A core-to-rim transect of major- and trace-element compositions
was obtained from a large hornblende grain that partially replaced
a large garnet. Only relict garnet is now present, as small
fragments within the replacement hornblende.
XMg and Ca cations
p.f.u. show decreases from core to rim across the hornblende,
whereas
XMn shows an increase (from 0·01 to 0·04;
Fig. 11a). The increase in
XMn in the replacement hornblende
is presumably a result of the release of Mn during garnet breakdown.
View larger version (13K):
[in this window]
[in a new window]
|
Fig. 11. Core-to-rim profile across a large hornblende grain that partially replaces garnet from the rehydrated amphibole–garnet segregation (02-174): (a) major elements; (b) trace elements.
|
|
Clinopyroxene has a
XMg of
0·67 and Al ranges from 0·07
to 0·08 cations p.f.u., which overlaps that of matrix
clinopyroxene from the restites. Plagioclase is andesine in
composition (An
40–45) and has similar anorthite contents
to the unmigmatized rock. Plagioclase grains show minor zoning,
with a decrease in anorthite from core to rim, similar to the
unmigmatized rock and the restite. Titanite and ilmenite grains
are generally unzoned in major elements. Titanite has Al between
0·04 and 0·05 cations p.f.u., which is indistinguishable
from Al contents in the restites. Ilmenite has a Mn content
of 0·02 cations p.f.u., which is slightly higher than
ilmenite in the restites.
Trace elements
Representative trace-element compositions for minerals from the amphibole–garnet segregation are given in Table 4. Garnet trace-element compositions are HREE-enriched and more fractionated than garnet from the restites (Fig. 12a). Ti contents decrease from core to rim, whereas Y and HREE concentrations generally increase from core to rim. Most of the concentrations are similar to the outermost portion of the peak metamorphic garnets investigated in the previous samples. This is well documented in the Ti vs Yb plot, where the majority of the garnet analyses from the amphibole–garnet segregation plot on the low Ti, high Yb segment (Fig. 10). Garnets from the amphibole–garnet segregation have a small negative Eu anomaly (Eu/Eu* = 0·62–0·76), unlike garnets from the restites (Eu/Eu* = 0·75–1·68, but generally >1·0).
View larger version (32K):
[in this window]
[in a new window]
|
Fig. 12. Chondrite-normalized (Sun & McDonough, 1989) REE patterns for minerals from the amphibole–garnet segregation (02-174): (a) garnet; (b) hornblende; (c) clinopyroxene; (d) titanite; (e) plagioclase. Symbols for (b)–(e): triangles, inclusions within garnet; circles, matrix minerals; filled symbols, cores; open symbols, rims of minerals. Additional symbols used for (b): squares, large replacement hornblende grain; filled symbols, cores; open symbols, rims. Dashed lines in (e) indicate the inferred pattern when trace-element concentrations were below the detection limit.
|
|
Hornblende in different textural sites has strongly varying
HREE concentrations (
Figs 11b and
12b). Hornblende inclusions
within garnet and the core of the large replacement hornblende
both show a strong HREE depletion and a negative Eu anomaly,
suggesting that this hornblende formed early in the metamorphic
history, in equilibrium with garnet cores and in coexistence
with plagioclase. Therefore, these hornblende trace-element
concentrations are consistent with a prograde metamorphic origin.
Trace-element compositions across the large replacement hornblende
grain (
Fig. 11b) show a strong increase of Yb from core to rim.
This change is associated with the flattening of the HREE pattern
shown in
Fig. 12b. It is suggested that the source for this
HREE increase derives from the breakdown of garnet, which is
the main host of HREE at peak-metamorphic conditions. The outermost
rim of the amphibole displays lower HREE contents and might
represent a late-stage amphibole rim, related to the partial
breakdown of clinopyroxene. Although Yb increases by more than
100% from core to rim, the changes seen in Ca and
XMg are of
the order of 5% only. However, Mn shows a strong increase from
core to rim. Because the main host of Mn is also garnet, this
observation further supports the hypothesis that amphibole composition
is monitoring the breakdown of garnet.
Matrix clinopyroxene from the amphibole–garnet segregation (Fig. 12c) is also relatively enriched in HREE compared with clinopyroxene from the garnet-rich segregations. It has a negative Eu anomaly, indicating that clinopyroxene grew after peak metamorphism whereas garnet was being partially resorbed, and in equilibrium with plagioclase. Similarly, matrix titanite is relatively HREE enriched and also has a prominent negative Eu anomaly, suggesting that it also formed after peak metamorphism. In contrast, titanite inclusions within garnet (Fig. 12d) are strongly HREE depleted and have a negative Eu anomaly, indicating that they grew in equilibrium with garnet before plagioclase reacted away during prograde metamorphism. The HFSE are very similar to what has been analysed in titanite from the garnet-rich segregation. Plagioclase has a strong positive Eu anomaly and shows a peculiar increase in HREE with increasing atomic number (Fig. 12e). This pattern suggests that plagioclase crystallized at the time of garnet resorption, which is characterized by the liberation of significant amounts of HREE.
Tonalite (02-180)
Major elements
Representative major-element compositions for minerals from the tonalite are presented in Table 5. Amphibole is edenitic hornblende (6·51–6·69 Si p.f.u.), with XMg ranging from 0·53 to 0·56, which is similar to the unmigmatized rock and the restite. The Mn content of hornblende in the tonalite is higher than in the garnet-rich restite and is very similar to hornblende in the segregation, showing extensive back-reaction. Plagioclase is andesine in composition (An34–40) but has significantly lower anorthite contents than the unmigmatized rock, restitic garnet-rich segregation and the rehydrated amphibole–garnet segregation (An40–49). Plagioclase grains show minor zoning, with a decrease in anorthite from core to rim, which is similar to zoning in plagioclase from all other rock types. Ilmenite and titanite grains are unzoned in major elements. Titanite has an Al content of 0·04 cations p.f.u., which is indistinguishable from Al contents in the other rock types. Ilmenite has a Mn content of 0·05 cations p.f.u., which is higher than for ilmenite from rocks where garnet is present (<0·02).
View this table:
[in this window]
[in a new window]
|
Table 5: Representative major- (electron microprobe, wt %) and trace-element (laser-ablation ICP–MS, ppm) concentrations of minerals from the tonalite (02-180)
|
|
Trace elements
Representative trace-element compositions for minerals from
the tonalite are given in
Table 5. Hornblende from the tonalite
is homogeneous and has relatively flat REE patterns and a strong
negative Eu anomaly (
Fig. 13a). This is similar to REE patterns
of replacement hornblende outer rims and matrix hornblende from
the amphibole–garnet segregation, but hornblende from
the tonalite has overall lower trace-element concentrations.
Titanite from the tonalite has a strong negative Eu anomaly
and similar REE patterns and abundances to titanite that formed
during garnet resorption in the amphibole–garnet segregation
(
Fig. 12b). Titanite has even higher Th (350–1000 ppm)
and U (100–180 ppm) than titanite from the peak garnet-rich
segregation. Nb (300–400 ppm) is very similar in titanite
and in ilmenite, whereas Ta is always higher in titanite (30–100
ppm) than in ilmenite (
20 ppm). Ilmenite has a high V content
of about 2000 ppm. LREE in apatite are similar to those in titanite,
but HREE are significantly lower (
Fig. 13b). The Th and U contents
of apatite are also significantly lower than those of titanite.
View larger version (21K):
[in this window]
[in a new window]
|
Fig. 13. Chondrite-normalized (Sun & McDonough, 1989) REE patterns for mineral cores (filled circles) and rims (open circles) from the tonalite (02-180): (a) hornblende; (b) titanite and apatite.
|
|
|
DISCUSSION
|
|---|
Monitoring partial-melting processes
The Mt Ruby metabasites provide an excellent natural laboratory
to study partial-melting processes in mafic rocks. The rock
types include unmelted amphibolites, peak-metamorphic garnet-rich
restites, an amphibole–garnet segregation caused by back
reaction of melt with garnet and an associated tonalitic melt,
representing different stages of the partial-melting event.
Additionally, the minerals in these rocks occur in texturally
distinct domains that can be attributed to prograde, peak and
retrograde metamorphism (
Fig. 5). This allows evaluation of
the extent to which the trace-element compositions of such minerals
are able to document the partial-melting process.
Major vs trace elements
Detailed analysis of minerals from the Mt Ruby metabasites indicates that there is very little change in major-element compositions of minerals that we view to be texturally different. Even the minerals from different rock types with different textural relationships all have very similar major-element compositions. The main variations in major-element composition are related to Fe–Mg exchange between garnet, amphibole and clinopyroxene that result in profiles that are attributed to diffusion during cooling (Fig. 8a). The homogeneous compositions of all rock-forming minerals suggest that prograde and peak metamorphic variations have been completely erased by diffusion at the high peak metamorphic temperature of 800°C. This agrees with detailed studies, which document loss of major-element zoning in garnets above 700–750°C (Tracy, 1982; Barnes & Carlson, 2001). This makes it impossible to effectively monitor partial-melting processes in amphibolites using major-element composition, because partial melting of mafic rocks normally takes place at relatively high temperatures of 800–1000°C (Rapp et al., 1991; Wolf & Wyllie, 1993, 1994). However, our study shows that garnet, amphibole and titanite retain different trace-element compositions in texturally different positions. In particular, the preservation of zoning in single grains of garnet and amphibole provides evidence that peak metamorphic temperatures were not high enough to erase trace-element zoning. This is in agreement with experimental studies showing slow diffusion of REE in silicate minerals (Van Orman et al., 2001, 2002). In addition, several other trace-element studies using natural samples have shown that REE are not reset by diffusion, even at granulite-facies conditions, and hence provide a powerful tool to monitor partial-melting processes (Ottamendi et al., 2002; Hermann & Rubatto, 2003).
Mineral stability
Detailed investigations of textures coupled with the trace-element analysis of amphibole and titanite have revealed that these two minerals were stable during prograde amphibolite-, peak granulite- and retrograde amphibolite-facies metamorphic conditions (Fig. 5). Figure 14 summarizes the REE characteristics of the different generations of amphibole and titanite. The large observed variation in REE provides detailed information about partial-melting processes that affected the amphibolite. One of the most distinctive features of the investigated minerals is the change of Eu anomaly. Plagioclase displays a strong positive Eu anomaly (Fig. 6) and, hence, coexisting minerals tend to have a negative Eu anomaly. Prograde amphibole and titanite included in garnet all display such a negative Eu anomaly, indicating that they coexisted with plagioclase. This is also supported by the unmigmatized amphibolite, which shows the paragenesis quartz + amphibole + clinopyroxene + plagioclase + titanite that has not been affected by partial melting. However, peak metamorphic garnet, amphibole, titanite and rare clinopyroxene do not display any Eu anomaly. This suggests that plagioclase was completely consumed during garnet formation occurring because of reaction (1). It is only at the stage when garnet resorption is observed that a negative Eu anomaly occurs, again in clinopyroxene, amphibole and titanite. This indicates that the trapped interstitial melt, which reacted with the peak garnet, started to crystallize plagioclase. Plagioclase from the tonalite clearly formed during crystallization from a melt. Both plagioclase in the tonalite and in the leucocratic layers of the garnet-rich rocks shows a similar trend of decreasing anorthite content from core to rim. This supports the hypothesis from the REE systematics that plagioclase formed during retrograde crystallization of trapped melt and was not present when garnet formed. This is important for the correct application of thermobarometric methods on plagioclase–garnet assemblages. Our trace-element approach shows that these minerals did not really coexist and, thus, P–T estimations involving garnet and plagioclase will be in error in this case.
View larger version (38K):
[in this window]
[in a new window]
|
Fig. 14. Chondrite-normalized REE (Sun & McDonough, 1989) patterns for (a) hornblende; and (b) titanite, from the garnet-rich restites (01-292 and 02-132) and the rehydrated amphibole–garnet segregation (02-174), showing the characteristics of each stage of metamorphism.
|
|
Different generations of titanite display different trace-element
patterns and are able to retain different U and Th contents
up to very high grade. This offers the possibility that prograde
and retrograde stages of a high-grade metamorphic evolution
can be dated in mafic rocks. What needs to be tested is whether
or not titanite included in garnet is able to retain radiogenic
Pb to higher temperatures than matrix titanite. Our study shows
that titanite in equilibrium with garnet displays HREE depletion,
making it possible to relate titanite formation to metamorphic
conditions. As outlined by Frost
et al. (2000)
, titanite has
a great potential for linking age determinations to metamorphic
conditions. This study shows that by also analysing titanite
trace-element contents, additional information on the metamorphic
evolution can be obtained.
Garnet zoning
Garnet displays interesting zoning patterns that further constrain the partial-melting event. The Ti content of garnet is buffered by the coexistence of a Ti-rich mineral, such as titanite and/or ilmenite, and hence it is expected that Ti contents in garnet increase with increasing temperature. In fact, the Ti content of garnet in high-grade calcareous rocks is noted to show a positive correlation with temperature (Jamtveit et al., 1997). Therefore, the garnets that show an increase in Ti in the core may be recording prograde growth zoning. Ti then decreases towards the rim (to as low as 200 ppm) in all analysed garnets, which is likely to represent cooling in the presence of a Ti-rich phase, which buffers the garnet composition. Interestingly, we do not find evidence that HREE decrease during garnet growth, as is often the case during garnet growth associated with partial melting in pelitic rocks (Ottamendi et al., 2002; Hermann & Rubatto, 2003). In pelitic rocks, this phenomenon indicates that xenotime, which buffers the HREE content of garnet, is exhausted during partial melting. The absence of such a pattern suggests that the garnet HREE content might have been buffered by the peritectic melting reaction. During the initial part of retrogression, garnet that was stable at peak conditions starts to break down. This breakdown liberates trace elements, causing compatible elements that stabilize the garnet structure, such as Mn, Y and the HREE, to accumulate at the garnet rim (Pyle & Spear, 2003). Such garnets are visible as small grains in the matrix and rims of large garnet crystals from the garnet-rich restites, and relict garnet that underwent significant back-reaction. We therefore suggest that garnets that have considerably higher HREE in the rim than in the core in rocks that underwent partial melting might be used as indicators for back-reaction of garnet with melt. As cooling progresses, garnet starts to break down completely, and liberated HREE and Y are concentrated in clinopyroxene, amphibole and titanite (Fig. 12). The large majority of analysed clinopyroxene displays such enriched HREE contents, indicating that it was either in equilibrium with the resorbed garnet rims or not in equilibrium with garnet at all. Hence, the use of such garnet–clinopyroxene pairs for Sm–Nd geochronology may give misleading results, although the extent of this is yet to be fully investigated.
Composition of peak metamorphic partial melt
The detailed textural study, combined with trace-element analyses, allowed us to define garnet, titanite and hornblende that were present during melt extraction at peak metamorphic conditions. The fact that these compositions are markedly different from mineral compositions formed during the late retrograde evolution strongly suggests that the trace-element compositions were not affected by retrograde processes. Hence, using published mineral–melt partition coefficients, we were able to calculate the trace-element composition of the partial melt in equilibrium with the peak-metamorphic assemblage. It is not easy to choose appropriate mineral–melt partition coefficients, as the exact composition of the peak metamorphic melt is not well constrained and there are only limited published partition coefficients at relatively low temperatures. Detailed geochemistry and evidence presented below indicate that the peak metamorphic melt had an andesitic rather than tonalitic composition (Storkey, 2004; Storkey et al., in preparation). Therefore, we have chosen partition coefficients of titanite–melt (Tiepolo et al., 2002), garnet–melt (Green et al., 2000) and hornblende–melt (Brenan et al., 1995) that match, as closely as possible, our natural system. Average trace-element compositions of peak metamorphic titanite, garnet and hornblende from the garnet-rich restites and rehydrated (amphibole–garnet) segregations were combined with these appropriate partition coefficients. The calculated trace-element compositional field for the peak metamorphic melt and the trace-element patterns of the peak metamorphic minerals used to construct this melt composition are shown in Fig. 15a.
Figure 15b shows the bulk-rock primitive mantle-normalized patterns
of each sample compared with the calculated peak melt. The amphibole–garnet
segregation represents a composite rock consisting of a restitic
part, which dominates the HREE budget, and a melt component,
which dominates the more incompatible elements. The pattern
of the highly incompatible elements is similar to the calculated
melt pattern, suggesting that the rehydrated amphibole–garnet
segregation contains
30–50% trapped melt, in agreement
with mass balance based on major elements (Storkey, 2004
; Storkey
et al., in preparation). A similar pattern is present in the
garnet-rich segregations, which also represent a composite rock
with restitic garnet and minor amounts of trapped melt that
crystallized into the plagioclase–hornblende–quartz
matrix. The tonalite has much lower trace-element abundances
than the calculated melt and a positive Sr anomaly, whereas
the calculated melt has a strong negative Sr anomaly. These
characteristics indicate that the tonalite is not the peak partial
melt that has been extracted from the refractory restitic layers.
The mineral-scale trace-element geochemistry also provides evidence
to suggest that the tonalite is not a peak metamorphic partial
melt. Hornblende and titanite from the tonalite are characterized
by relatively HREE-enriched patterns and negative Eu anomalies
(
Fig. 12), and are very similar to late-stage hornblende and
titanite REE patterns from the amphibole–garnet segregation.
Therefore, the mineral trace-element geochemistry suggests that
the tonalite equilibrated during the retrograde history in equilibrium
with plagioclase after garnet had been partially to completely
resorbed, and was then extracted as a late-stage melt feature.
This is in agreement with a detailed study on the major- and
trace-element composition of these rocks, indicating that the
tonalites might have evolved from a more mafic melt present
at peak metamorphic conditions (Storkey, 2004
; Storkey
et al.,
in preparation). The overall low REE concentrations in the tonalite
may indicate that the tonalite contains some cumulate plagioclase.
This is supported by the observed positive Sr anomaly in bulk-rock
analyses. Interestingly, despite evidence of the accumulation
of these late-stage melts as numerous tonalitic bodies, there
is no significant body of the early trace-element-rich melts
that formed in equilibrium with the peak metamorphic assemblage
in the mostly anhydrous restites. Therefore, we suggest that
these melts were efficiently extracted from the Mt Ruby system.
Phase controls on partial melting and crystallization
In order to constrain the control of different phases on the trace-element distribution during partial melting, we have performed trace-element mass balance calculations for each sample (Fig. 16). Modal abundances of major phases were first estimated from thin section by point counting. Then, a least-squares method, based on major-element compositions of bulk rock and minerals, was used to
TOP
ABSTRACT
INTRODUCTION