斜长角闪岩

本文为一篇很有价值的英文文献,文中对斜长角闪岩特征做了详细介绍。

Contrib Mineral Petrol (2014) 168:1060

DOI 10.1007/s00410-014-1060-0

Rare earth element–SiO2 systematics of island arc crustal amphibolite migmatites from the Asago body of the Yakuno Ophiolite, Japan: a field evaluation of some model predictions

Xiaofei Pu · James G. Brophy · Tatsuki Tsujimori

Received: 25 April 2014 / Accepted: 21 August 2014 / Published online: 9 September 2014 © Springer-Verlag Berlin Heidelberg 2014

Abstract The two most commonly invoked processes for generating silicic magmas in intra-oceanic arc environments are extended fractional crystallization of hydrous island arc basalt magma or dehydration melting of lower crustal amphibolite. Brophy (Contrib Mineral Petrol 156:337–357, 2008) has proposed on theoretical grounds that, for liquids >~65 wt% SiO2, dehydration melting should yield, among other features, a negative correlation between rare earth ele-ment (REE) abundances and increasing SiO2, while frac-tional crystallization should yield a positive correlation. If correct, the REE–SiO2 systematics of natural systems might be used to distinguish between the two processes. The Per-mian-age Asago body within the Yakuno Ophiolite, Japan, has amphibolite migmatites that contain felsic veins that are believed to have formed from dehydration melting, thus forming an appropriate location for field verification of the proposed REE–SiO2 systematics for such a process. In addi-tion to a negative correlation between liquid SiO2 and REE abundance for liquids in excess of ~65 % SiO2, another important model feature is that, at very high SiO2 contents (75–76 %), all of the REE should have abundances less than

Communicated by T. L. Grove.

X. Pu · J. G. Brophy (*)

Department of Geological Sciences, Indiana University, Bloomington, IN 47405, USAe-mail: brophy@indiana.eduX. Pu

e-mail: pux@indiana.edu

T. Tsujimori

The Pheasant Memorial Laboratory for Geochemistry

and Cosmochemistry (PML), Institute for Study of the Earth’s Interior, Okayama University, Misasa, Tottori-Ken 682-0193, Japan

e-mail: tatsukix@misasa.okayama-u.ac.jp

that of the host rock. Assuming an initial source amphibolite that is slightly LREE-enriched relative to the host amphibo-lites, the observed REE abundances in the felsic veins fully support all theoretical predictions.

Keywords Amphibolite · Migmatite · Rare earth elements · Partial melting · Silicic magma

Introduction

The generation of silicic magmas in intra-oceanic arc envi-ronments is a subject that has received a lot of attention. Though many possible origins have been suggested, the two most commonly invoked processes for generating such magmas are extended fractional crystallization of hydrous, mantle-derived island arc basalt (IAB) magma or dehydra-tion melting of lower crustal amphibolite. As is often the case, when two different processes are suggested for the same phe-nomenon, both processes are probably occurring. In a former study, Brophy (2008) proposed that the REE–SiO2 system-atics of mafic to felsic magmas in oceanic arc environments could be used to distinguish between these two processes in natural arc lavas and/or intrusives. Because the 2008 study was entirely model based, it requires some form of field veri-fication before REE–SiO2 systematics can be accepted as a useful geochemical tool. The present study consists of a pet-rologic and geochemical study of a natural example of partial melting of island arc crust from the Asago Body within the Yakuno ophiolite located on Honshu Island, Japan (Fig. 1). The goal of the study is to evaluate the proposed REE–SiO2 systematics of Brophy (2008) for dehydration melting of lower crustal amphibolite. The Yakuno migmatites were first studied by Suda (2004) who concluded that they were gener-ated by dehydration melting of island arc crustal amphibolite.

1 3

本文为一篇很有价值的英文文献,文中对斜长角闪岩特征做了详细介绍。

1060 Page 2 of 12

Fig. 1 Generalized map show-ing the location of the Asago Body in the Yakuno Ophiolite (dark colored mafic units) and the amphibolite migmatites that are the focus of this study (after Suda 2004

)

Contrib Mineral Petrol (2014) 168:1060

REE–SiO2 systematics for amphibolite melting and basalt fractionation in oceanic island arcs

It is commonly assumed that amphibolite melting is simply the reverse process of basalt fractionation, and therefore, the two processes cannot be distinguished from one another on chemical grounds. This is not true. During the dehydra-tion melting of amphibolite, silicic melts are generated by one or more reactions of the type shown below (Rushmer 1991; Beard and Lofgren 1991; Rapp and Watson 1995)

Hbd+Na-richplag+Fe-ox=cpx+opx+Ca-richplag+Fe-ox+silicicmelt.This reaction will continue to produce silicic melt until all of the original hornblende is consumed. Thus, the REE–SiO2 systematics of the silicic melt will always

be controlled by the combined presence of hornblende, plagioclase, orthopyroxene and clinopyroxene. During fractional crystallization of hydrous, IAB magma, horn-blende is most likely present as a crystallizing phase in the lower crust (Davidson et al. 2007), but the extent to which it is a significant crystallizing phase is still uncer-tain (e.g., Davidson et al. 2013). Thus, in many arc sys-tems, extended basalt fractionation may be dominated by olivine, clinopyroxene, orthopyroxene and Fe-oxide, but not hornblende. From the standpoint of REE–SiO2 system-atics, the potentially different role played by hornblende in dehydration melting of amphibolite and extended IAB fractionation may be very significant. The reason for this is emphasized in Fig. 2, which shows the variation in horn-blende-liquid D values [plotted as log(D)] versus liquid SiO2 for La and Yb. For both elements, there is a steady

1 3

本文为一篇很有价值的英文文献,文中对斜长角闪岩特征做了详细介绍。

Contrib Mineral Petrol (2014) 168:1060 Page 3 of 12 1060

Fig. 2 Hornblende–liquid La and Yb D values versus SiO2 content of co-existing liquid (glass). Data sources include Sisson (1994), Klein et al. (1997), Dalpe and Baker (2000) and Brophy et al. (2011)

Fig. 3 Predicted REE–SiO2 variations (from Brophy 2008) for lower crustal amphibolite melting, hornblende-free mid-crustal fractionation of basalt and hornblende-bearing mid-

crustal fractionation of basalt

increase in log(D) and therefore D with increasing liquid SiO2. Furthermore, the partitioning behavior eventually changes from incompatible (D < 1) to compatible (D > 1). For La, this switch occurs at around 75 wt% SiO2, but reaches as low as ~60 % SiO2 for Yb. Of the four minerals mentioned above (olivine, plagioclase, clinopyroxene and hornblende), hornblende is the only one that displays this behavior to such a large extent. What this means is that, for SiO2-rich liquids, the presence or absence of horn-blende can translate into dramatically different partitioning behavior for the REE.

Brophy (2008) modeled the variation in La and Yb with increasing liquid SiO2 for both fractional crystalliza-tion of IAB basalt and dehydration melting of lower arc crust amphibolite. The modeling combined major element mass-balanced fractional crystallization models and exper-imentally based amphibolite melting models with quanti-tative expressions describing the log (D)–SiO2 variations for La and Yb shown in Fig. 2. The results are summarized in Fig. 3, which shows the predicted variation in La and Yb abundances for batch melting of amphibolite and frac-tional crystallization of IAB basalt. Fractional melting and accumulated fractional melting show similar overall results and are not shown here. In both cases, the initial gabbro or basalt was assumed to contain 50 wt% SiO2. The results are expressed in terms of an element enrichment factor in the liquid, Cl/Co, where Cl is the concentration of the element in the liquid and Co is the original concen-tration of the element in the source rock (for melting) or the original basalt magma (for crystallization). The most important feature of Fig. 3 is that, for liquids in excess of ~65 % SiO2, amphibolite melting reveals a negative cor-relation between La and Yb abundances and SiO2 content, while IAB basalt fractionation shows a positive correla-tion up to SiO2 contents of around 70 wt% after which a negative correlation is observed. These differences reflect (1) the increasing compatibility of La and Yb in horn-blende with increasing liquid SiO2 and (2) the dominant role played by hornblende during melting and the minor role during crystallization. If these model predictions are correct, then they could provide an important geochemical tool for distinguishing between a melting and fractional crystallization origin for natural silicic lavas in intra-oce-

anic arc environments.

1 3

本文为一篇很有价值的英文文献,文中对斜长角闪岩特征做了详细介绍。

1060 Page 4 of 12Fig. 4 Predicted variations in liquid enrichment factors (Cl/Co) for La (HREE), Gd (MREE) and Yb (HREE) from the model of Brophy (2008

)

Predicted REE–SiO2 systematics for amphibolite melting

The most controversial part of the Brophy (2008) modeling is the predicted REE–SiO2 systematics for amphibolite melting, the field verification of which constitutes the essence of the current study. The gist of the predicted systematics is sum-marized in Fig. 4, which shows predicted liquid enrichment factors (Cheavy (La) rare earth elements (REEs) as a function of liquid l/Co) for representative light (La), middle (Gd) and SiO2 content. Two major predictions can be made. First, for liquids greater than around 65 % SiOshow a negative correlation with increasing liquid SiO2, all of the REE should ond, at very high liquid SiO2. Sec-2 contents (around 75–76 %), the LREE should all show enrichment values <1. It is these two theoretical predictions that are to be evaluated in this study.

Analytical methods

Bulk rock major and trace element analyses were conducted by XRF and LA-ICPMS at Michigan State University fol-lowing analytical procedures described in Deering et al. (2008). Major element compositions of minerals were determined by electron microprobe analysis (EMPA) on a Cameca SX50 located at Indiana University. Operating con-ditions included a 15 kV acceleration voltage, 20 nA sample current and a beam diameter of 3 microns. Na2O was always analyzed first to reduce any effects of volatilization/diffu-sion. Accessory mineral percentages in the mafic amphibo-lite were determined by initial EMPA of individual miner-als followed by image analysis of multiple (15) back scatter electron images for four different amphibolite samples.

The Yakuno Ophiolite

The Permian-aged Yakuno Ophiolite is located on Honshu Island, Southwest Japan (Fig. 1). Three different segments have

1 3

Contrib Mineral Petrol (2014) 168:1060

Fig. 5 Field photograph showing one of the amphibolite migmatite outcrops that are the focus of this study

been identified within the ophiolite including unusually thick oceanic crust, island arc crust and oceanic back-arc basin crust (Ishiwatiri 1985; Hayasaka 1990; Ichiyama and Ishiwatari 2003). The Asago body is a fault-bounded tectonic slice within the ophiolite suite. It is comprised of three structural units: a Lower Unit (L-Unit), Middle-Unit (M-Unit) and Upper-Unit (U-Unit) (not shown in Fig. 1). The M-Unit is thought to repre-sent a lower to middle crustal section of an intra-oceanic island arc (Hayasaka 1990). The amphibolite migmatites sampled for this study were located in the lower part of the M-Unit, near the boundary of the M- and L-Unit (Suda 2004). They were col-lected along a short southwesterly transect following the bot-tom of a small stream valley (Fig. 1). A total of ten samples were collected. Two samples represent mafic amphibolites without felsic veins, while the rest of the samples consist of intermixed mafic amphibolite and felsic veins. According to Suda (2004), the P–T conditions during the formation of these migmatites are estimated to be around 850 °C and 3.5–5.5 kbar with a peak metamorphic grade of granulite facies.

Sample description

The amphibolites show a clear migmatite signature with felsic veins and pockets dispersed throughout the amphibo-lite host (Fig. 5). The contacts between the mafic amphibo-lite host and the felsic material are always sharp, suggest-ing that the felsic melts were actually generated elsewhere and intruded into their current position through joints and fractures in the host rock. The scale and shape of the fel-sic intrusions vary from sample to sample. The amphibolite host consists of primary amphibole, plagioclase and minor amounts of augite. Detailed analysis of multiple samples

indicates very small volumes of sphene (0.5 wt%), apatite

本文为一篇很有价值的英文文献,文中对斜长角闪岩特征做了详细介绍。

Contrib Mineral Petrol (2014) 168:1060 (0.15 wt%) and zircon (0.05 wt%). The amphibole is calcic and ranges from magnesio-hornblende to ferro-hornblende (Leake et al. 1997). The plagioclase has been extensively albitized with some crystals reaching Ab100 in composition. Most crystals show some degree of saussuritization. A few plagioclase grains have compositions around Abing that the original (pre-albitized) plagioclase was of this 35, suggest-general composition. Augite is largely unaltered and com-positionally uniform at around Wominerals include actinolite, prehnite, epidote, chlorite and 46En36Fs18. Secondary saussurite minerals. Actinolite, though present in only small amounts, can partially or completely replace hornblende. Hornblende is occasionally replaced by chlorite. Small amounts of epidote and prehnite occur as discrete crystals.The felsic veins consist of primary plagioclase and quartz. Quartz shows up as aggregates of fine-grained crys-tals and or intergrowths with plagioclase. Individual pla-gioclase crystals are still recognizable under cross-polar in microscope, but they have been mostly saussuritized. Most plagioclase crystals have been completely albitized (Ab99–Ab100). Subhedral hornblende xenocrysts can be found floating in the felsic material.

The hydrous secondary minerals in both the amphibo-lite and felsic intrusions point toward a significant prehnite to greenschist facies retrograde metamorphism that post-dated the peak metamorphic event (when partial melting occurred) by some undetermined amount of time.

Whole rock geochemistry

Whole rock major and trace element data are listed in Table 1. Total iron is reported as FeO*. SiO2 contents (on an anhydrous basis) range from 47 to 55 % in the mafic amphibolites and from 76 to 81 % in the felsic intrusion samples. Figure 6 shows Harker variation diagrams for selected oxides. Within both the mafic and felsic groups, Al2O3, TiO2, FeO*, MgO and CaO steadily decrease with increasing SiO2, while Na2O and K.

2O show a rough increase with increasing SiOFigure 7 shows a chondrite-normalized trace element 2spider diagram for both mafic and felsic samples using the normalization factors of McDonough and Sun (1995). The mafic and felsic samples have similar overall patterns, but the felsic samples are enriched in the incompatible ele-ments and depleted in the compatible elements compared with the mafic samples. The overall patterns are flat with several important anomalies including positive Ba, slightly negative Nb and Ti, and strong P depletion. Also seen are positive Ba anomalies more negative Ti (mostly in felsic samples) and Nb anomalies. Felsic samples have higher incompatible element abundances and lower compatible element abundances compared with the mafic samples.

Page 5 of 12 1060

The overall flat patterns in the chondrite-normalized pro-files and the presence of positive Ba and negative Nb and Ti anomalies are consistent with an island arc setting as origi-nally suggested by Hayasaka (1990) and Suda (2004).

Figure 8 shows chondrite-normalized REE patterns for both the mafic and felsic samples. The patterns for the mafic samples are flat with general abundances ranging from about 6 to 20 times those of chondrite. The patterns for the felsic samples are somewhat fractionated and con-cave upwards. Relative to the mafic samples, the inclined patterns for the felsic samples have resulted primarily from HREE depletion with only slight LREE enrichment.

Relationship between amphibolite host and felsic veinsTwo important questions for this study are (1) do the felsic veins represent partial melts of amphibolite and; (2) do the felsic veins represent in situ partial melting of the amphi-bolites in which they are hosted? Several lines of evidence support an amphibolite melting origin for the felsic veins. First, the field relations are those of a classic migmatite, which are almost universally viewed as representing partial melting (e.g., Sawyer 2008). Second, relative to the host amphibolites, the felsic veins show slight LREE enrichment and significant HREE depletion. Furthermore, the HREE display a concave upwards pattern. This pattern is entirely consistent with melting of a source rock that contained sig-nificant residual hornblende at the time of melt extraction (i.e., an amphibolite). Finally, Fig. 9 compares the major element chemistry of the host amphibolites and felsic veins with the experimental melts of Beard and Lofgren (1991) for the dehydration melting of amphibolite over a pressure range of 3–6.9 kbar. All of the felsic veins have SiO-rich experimental 2 con-tents that are greater than the most SiOliquid. However, for all of the oxides, the felsic veins lie on 2extensions of the experimental liquid trends. When taken together, these three lines of reasoning strongly support an amphibolite melting origin for the felsic veins.

It is noteworthy that all of the felsic samples display a significant positive Eu anomaly (Fig. 8), the origin of which is unclear. Plagioclase preferentially retains Eu2+, while hornblende and augite preferentially exclude Eu2+ (e.g., Brophy et al. 2011). Thus, the positive anomaly could be a result of plagioclase accumulation, suggesting a fractional crystallization origin for the silicic veins, or partial melt-ing of amphibolite wherein the silicic melt is in equilibrium with hornblende and/or augite during the melting process. Because there is so much field and textural evidence sug-gesting a partial melting origin, the positive Eu anomalies are most likely a reflection of the latter.

Several lines of evidence also suggest that the host amphibolites do not represent the true source rock for the

1 3