1
2
3
Trace elements transport during subduction-zone
4
metamorphism: Evidence from ultrahigh pressure
5
metamorphic rocks, Western Tianshan, China
6
7
Yuanyuan Xiao a,*, Shaun Lavis b, Yaoling Niu a, c, *, Julian A. Pearce b, Huaikun
8
Li d, Huichu Wang d, Jon Davidson a
9
a
Department of Earth Sciences, Durham University, Durham, DH1 3LE, UK
10
b
School of Earth and Ocean Sciences, Cardiff University, Cardiff, CF10 3YE, UK
11
c
School of Earth Sciences, Lanzhou University, Lanzhou, 730000, China
12
d
Tianjin Institute of Geology and Mineral Resources, Tianjin 300170, China
13
14
Manuscript prepared for <<The Geological Society of America Bulletin>>
15
First submission: June 13, 2010
16
Revised version submission: March 15, 2011
17
18
Abstract (213 words), text (8741 words), 85 references, 14 Figures, 3 Tables, 3
19
Appendices
20
21
22
23
24
*Corresponding
authors:
Miss Yuanyuan Xiao (yuanyuan.xiao@durham.ac.uk)
Professor Yaoling Niu (yaoling.niu@durham.ac.uk)
-1-
25
ABSTRACT
26
We report the petrography and results of a geochemical study of blueschists and
27
eclogites from the Western Tianshan ultrahigh pressure metamorphic belt, Northwest
28
China. The protoliths are mostly oceanic basalts, including those resembling oceanic
29
island basalts and mid-ocean ridge basalts, some of which show geochemical
30
signatures of continental crust contamination (or volcanic arc basalts-like). Together
31
with metamorphic rocks of sedimentary protolith, this lithological assemblage most
32
likely represents a subducted and exhumed tectonic mélange.
33
By examining correlations between incompatible elements, we show that high
34
field strength elements, rare earth elements, Th and U are relatively immobile,
35
whereas Pb and Sr are mobile in both basaltic and sedimentary protoliths during
36
subduction-zone metamorphism (SZM). The immobility of U may suggest its +4 form
37
and a more reduced condition in subduction zones than previous thought. K, Rb, Cs
38
and Ba are mobile in rocks of basaltic protolith, but immobile in rocks of sedimentary
39
protolith due to the presence and the persistent stability of white mica throughout their
40
petrologic history after the diagenesis.
41
The lack of Rb/Sr - Sm/Nd correlation cannot explain the observed first-order Sr-
42
Nd isotope correlation in oceanic basalts, suggesting that the residual ocean crust that
43
has undergone SZM cannot be the major source material for oceanic basalts although
44
they can contribute to mantle compositional heterogeneity.
45
INTRODUCTION
46
One of the major advances in the field of solid earth geochemistry over the past
47
~ 40 years is the recognition of mantle chemical and isotopic heterogeneity through
48
studies of oceanic basalts, including mid-ocean ridge basalts (MORB) and basalts
49
erupted on intraplate volcanic islands, or oceanic island basalts (OIB). OIB are
50
particularly variable in composition such that several isotopically distinct end-2-
51
members, e.g., enriched mantle I (EMI), enriched mantle II (EMII), high μ (HIMU; μ
52
= 238U/204Pb) are required to explain (e.g., White, 1985; Zindler and Hart, 1986). The
53
depleted MORB mantle (DMM) is thought to have resulted from the extraction of
54
continental crust from the primitive mantle early in Earth’s history (e.g., O’Nions et
55
al., 1979), but the origin of OIB end-members remains enigmatic. Isotopic ratio
56
differences among these end-members reflect the differences in the radioactive
57
parent/radiogenic daughter (P/D) ratios (e.g., Rb/Sr, Sm/Nd, U/Pb and Th/Pb)
58
(Zindler and Hart, 1986) in their ultimate mantle sources, which with time evolve to
59
distinctive fields in isotopic ratio spaces (see Niu and O’Hara, 2003). Significant P-D
60
element fractionations in the solid state are considered unlikely in the deep mantle due
61
to extremely slow diffusion rates (Hofmann and Hart, 1978), hence processes known
62
to occur in the upper mantle and crust (e.g., partial melting, magma evolution,
63
metamorphism, weathering, transportation and sedimentation) are possible causes of
64
P-D fractionations (Niu and O’Hara, 2003). Among other processes, mantle
65
metasomatism can effectively cause mantle compositional heterogeneity (e.g., Sun
66
and McDonough, 1989), and may take place at the base of the growing oceanic
67
lithosphere (e.g., Niu and O’Hara, 2003; Pilet et al., 2008) or in the mantle wedge
68
above subduction zones (Donnelly et al., 2004).
69
As an inevitable consequence of plate tectonics, subduction of shallow and
70
surface processed materials (SSPM) must volumetrically be the major agent that
71
causes mantle compositional heterogeneity. This has led to the contention that mantle
72
heterogeneity can be related to specific subducted components. For example, (1)
73
altered ocean crust (AOC) with high U/Pb ratio and appropriate values for other P/D
74
ratios (e.g., Weaver, 1991; Hofmann, 1997) could produce mantle sources with HIMU
75
characteristics, (2) land-derived sediments or upper continental crust materials (e.g.,
76
Weaver, 1991; Hofmann, 1997; Willbold and Stracke, 2006) may contribute to EMII
-3-
77
type lavas; and (3) pelagic sediments (e.g., Weaver, 1991) or lower continental crust
78
(Willbold and Stracke, 2006) may contribute to EMI type lavas.
79
While these interpretations are apparently reasonable, they cannot be validated
80
without understanding the geochemical effects of subduction-zone metamorphisms
81
(SZM). This is because metamorphic dehydration of subducting ocean crust, which
82
has been widely accepted to cause mantle wedge melting for arc magmatism (e.g.,
83
Gill, 1981; McCulloch and Gamble, 1991), may be accompanied by chemical
84
modification resulting from elemental transport in the released fluids. As illustrated in
85
Fig.1, the subducted component (i.e., [P/D][in]) is modified as it passes through the
86
“subduction factory” (e.g., Tatsumi and Kogiso, 2003) (i.e., [P/D][out]), so this process
87
needs to be accounted for in any realistic mass balance models. We cannot
88
indiscriminately treat the compositions of subducted/subducting sediments (of various
89
types) or AOC as source materials for oceanic basalts without understanding the
90
effects of SZM because it is the geochemical properties of [P/D][out] that contribute to
91
mantle compositional heterogeneity with P/D and isotopic imprints.
92
Experimental and modelling approaches (e.g., Pawley and Holloway, 1993;
93
Schmidt and Poli, 1998) to SZM are useful, but their applications are not
94
straightforward because the natural system is open and complex as there are many
95
potential protolith types and metamorphic conditions (Zack and John, 2007; Schmidt
96
and Poli, 2003). Thus, analysis of natural rocks, which have actually subducted and
97
been exhumed, remains essential in understanding the effects of SZM. However,
98
many rocks studied previously did not reach the depth of ultrahigh pressure
99
metamorphism (UHPM) (e.g., Sorensen et al., 1997; Becker et al., 2000; Spandler et
100
al., 2003), and thus are not suitable for studying the geochemical consequences of
101
SZM on recycling and mantle heterogeneity. Some ultrahigh pressure (UHP; > 2.5 –
102
3.0 GPa) metamorphic rocks of ocean crust protolith have been reported in recent
103
studies (Bebout, 2007), e.g., 2.7 – 2.9 GPa in Lago di Cignana (e.g., Reinecke, 1998),
-4-
104
2.6 – 2.8 GPa in central Zambia (John et al., 2004), but the results are inconclusive
105
and additional work is still required. This study of UHP metamorphic rocks, from the
106
Western Tianshan, China, as evidenced by the presence of coesite (e.g., Lü et al.,
107
2008, 2009), contributes to our understandings of the geochemical effects of UHPM
108
and offers an insight into processes occurring beneath the volcanic arc.
109
Based on mineral modal and detailed correlation analyses of most analyzed
110
elements in these rocks, we conclude that the sequences of appearance and
111
disappearance of hydrous phases (e.g., white mica, epidote group minerals) are more
112
important than the simple blueschist-eclogite dehydration reactions in controlling the
113
transportation of fluid-mobile elements in subducting/subducted oceanic crust (and
114
sediments) during SZM. Furthermore, the residual ocean crust that has passed through
115
SZM cannot be the major source material for OIB in terms of P/D ratios and isotopic
116
systematics.
117
GEOLOGICAL BACKGROUND
118
Overview of Western Tianshan Mountains
119
The Chinese Western Tianshan orogenic belt separates the Junggar Plate to the
120
north from the Tarim Plate to the south. Between the Tarim and Junggar plates is the
121
Yili-Central Tianshan Plate defined by the North Central and the South Central
122
Tianshan Suture for its northern and southern boundaries respectively (Fig.2) (e.g.,
123
Gao and Klemd, 2000; Zhang et al., 2001). The UHP metamorphic belt is located
124
along the South Central Tianshan Suture, which is thought to continue to the west in
125
Kazakhstan (e.g., Volkova and Budenov, 1999). It defines the palaeo-convergent
126
margin associated with northward subduction of the south Tianshan palaeo-oceanic
127
floor beneath the Yili-Central Tianshan Plate, followed by the northward subduction
-5-
128
of the Tarim Plate upon closure of the south Tianshan ocean basin (e.g., Gao et al.,
129
1999; Gao and Klemd, 2003) (Fig.2).
130
Much effort has been expended to date the peak metamorphic event using a
131
variety of techniques including the Ar-Ar plateau method (e.g., Gao and Klemd, 2003)
132
and the Sm-Nd and Rb-Sr isochron methods (Gao and Klemd, 2003), the latter of
133
which gave an age of ~ 350 – 345 Ma (Carboniferous) for peak metamorphism and ~
134
311 – 310 Ma for exhumation (i.e., cooling age) (detailed discussion can be found in
135
Gao et al. (2006)). Much younger ages of 233 ± 4 to 226 ± 4.6 Ma (Triassic) have
136
been obtained using the sensitive high-resolution ion microprobe (SHRIMP) U-Pb
137
method on metamorphic zircons extracted from both blueschist and UHP eclogite
138
(Zhang et al., 2007). Based, however on the most recent SHRIMP U-Pb ages reported
139
by Gao et al., (2009) and Su et al. (2010), as well as the constraints of the cooling age
140
of ~311 Ma (also personal communication with Lifei Zhang, 2010), it is apparent that
141
the peak metamorphism happened in the Carboniferous (vs. Triassic) Period in
142
Western Tianshan.
143
Existing work has argued that the Western Tianshan high pressure (HP) and
144
UHP metamorphic rocks have experienced subduction and ultimate exhumation of
145
oceanic crust in response to the continental collision, with a clockwise P-T-t path on a
146
regional scale and nearly isothermal decompression from peak eclogite facies to
147
epidote-amphibolite facies (e.g., Gao et al., 1999). The protoliths are argued to be
148
dominated by seafloor basalts, including OIB, normal (N)-type and enriched (E)–type
149
MORB, as well as volcanic arc basalts (VAB) (e.g., Gao and Klemd, 2003) with some
150
terrigenous sediments (e.g., Ai et al., 2006). The highly variable compositions of
151
meta-basaltic rocks are consistent with the geochemistry of seamounts like those near
152
the East Pacific Rise (see Niu and Batiza, 1997). Intermingled with terrigenous
153
sediments and volcaniclastic rocks, those dismembered basalts formed a tectonic
-6-
154
mélange within an accretionary wedge, and then the subsequent subduction of this
155
complex is responsible for the UHPM (e.g., Gao and Klemd, 2003; Ai et al., 2006).
156
UHPM in Western Tianshan Mountains
157
Despite the debate on the peak P-T conditions of Western Tianshan
158
metamorphism over the past ~ 10 years (e.g., Zhang et al., 2002a-b; Klemd, 2003), the
159
new discovery of coesite from Western Tianshan (Lü et al., 2008, 2009) gives
160
definitive evidence in support of a peak pressure in excess of 2.5 GPa, indicating that
161
UHPM did take place.
162
Coesite is stable at pressures > 2.5 GPa and is a ‘smoking gun’ for UHPM. The
163
recognition of coesite relics in garnet of Tianshan eclogites has been inferred to be
164
widespread (Lü et al., 2008, 2009; also personal communication with Lifei Zhang,
165
2010), and thus it is indisputable for UHPM in Western Tianshan. However, previous
166
geothermobarometric calculations along with the lack of convincing evidence to show
167
the common presence of UHPM index minerals (e.g., coesite, kyanite) in samples
168
from Tianshan eclogites (including our samples) led some authors (e.g., Klemd et al.,
169
2003, Lavis, 2005) to argue against the UHPM. Hence, this requires some
170
straightforward conceptual clarifications here. Coesite is volumetrically rare, and it
171
can be widespread throughout an UHP metamorphic belt, but is not necessarily
172
present in each random sample. For example, if we make 10 thin sections from a
173
single hand specimen, we may find coesite only in one of the 10 thin sections. This is
174
also the reason why it may take a long time to confirm one UHP metamorphic belt. It
175
is thus incorrect to argue that the thin section that contains coesite had subducted to
176
depth > 80 km whereas other 9 thin sections of the same hand specimen without
177
coesite had not subducted to that depth. Furthermore, because of the rapid subduction
178
and exhumation, the rock may not stay at peak P-T conditions long enough relative to the
179
quartz-coesite transformation, which result in even rarer occurrence of coesite. In
-7-
180
addition, because of the widespread retrograde metamorphism and overprints, it is
181
also likely that a portion of previous coesite crystals may have completely reverted to
182
quartz during the re-equilibrium at low pressures.
183
Although not observed in this work, previous studies identified other lines of
184
evidence indicating the UHPM in the Western Tianshan, i.e., the presence of quartz
185
and coesite exsolution lamellae due to the decompression of Ca-Eskola in omphacite
186
of Tianshan eclogites (e.g., Zhang et al., 2002a), and the coexistence of dolomite,
187
magnesite and calcite (aragonite pseudomorph) in rocks of sedimentary protolith
188
within Tianshan eclogite bodies (e.g., present between eclogitized basaltic flows and
189
pillows; Zhang et al., 2002b, 2003). The latter two arguments, however, are not
190
definitive evidence for UHPM. For example, Ca-Eskola may be stable in the quartz
191
(vs. coesite) stability field (Page et al., 2005) and the decomposition of dolomite may
192
take place at lower pressures (e.g., 2.3 - 2.8 GPa; Smit et al., 2008). Nevertheless, the
193
presence of coesite, together with the arguments considering Ca-Eskola composition
194
and dolomite decomposition, ~ 2.5 GPa is likely the minimum for peak pressure.
195
Hence, there should be no dispute on the occurrence of UHPM in Western Tianshan.
196
PETROGRAPHY
197
The UHP metamorphic belt in Tianshan is dominated by blueschist and eclogite
198
pods and lenses, surrounded by greenschist facies rocks (Fig.2). A total of 50 samples
199
were taken from locations broadly similar to those reported by Gao et al. (1999) and
200
Zhang et al. (2003), including one rodingite TS02-34, which is dominantly made up
201
of garnet (69.4 vol%). An additional 9 samples were provided by Lifei Zhang from
202
localities described in Zhang (2003). Detailed petrography is given in Table 1. Rocks
203
of sedimentary protolith (Fig.3g) can be readily distinguished from rocks of basaltic
204
protolith (Figs.3a-f, h) by their very high modal quartz and carbonate, as well as more
205
obvious foliation defined by the relatively high modal white mica, but less modal
-8-
206
(clino)zoisite which is probably due to the dilution caused by the significantly higher
207
quartz and white mica modes. Many samples of sedimentary protolith have ~ 50 vol%
208
quartz, and these are termed as quartzite (Fig.3g and Table 1). Compared with rocks
209
of basaltic protolith, rocks of sedimentary protolith tend to show stronger
210
retrogressive overprints, particularly the extent of chloritization of minerals such as
211
garnet and white mica.
212
Samples TS02-3B, TS02-15A, TS02-17b and TS02-32A belong strictly to
213
eclogite classification (where garnet + omphacite > 70 vol% of the bulk rock). They
214
are composed of garnet porphyroblasts and fine-grained omphacite matrix, along with
215
varying amounts of accessory minerals, including amphibole (Na-amphibole with a
216
little Mg-hornblende), (clino)zoisite, white mica (both phengitic muscovite and
217
paragonite), rutile, titanite, quartz and carbonate. Glaucophane and omphacite
218
inclusions in garnet indicate that eclogites have undergone a prograde blueschist
219
facies metamorphism prior to reaching eclogite facies, i.e., Stage 1 in Fig.4.
220
Glaucophane also occurs as discrete relicts of prograde metamorphism in the matrix
221
and only some of them show the overgrowth of amphibole (e.g., TS02-15A),
222
indicating a highly progressive metamorphism and limited retrograde metamorphic
223
overprints. The relatively low modal (clino)zoisite (no more than 1.07 vol%) may be
224
attributed to less retrograde (clino)zoisite formation and more pronounced breakdown
225
of prograde (clino)zoisite evidenced by resorption features (Gao et al., 1999; John et
226
al., 2008), representing the Stage 2 in Fig.4.
227
Blueschist is dominated by glaucophane and (clino)zoisite with or without
228
omphacite. (Clino)zoisite could be locally abundant (up to 20 vol%), forming
229
omphacite-garnet epidosite. Glaucophane, omphacite, white mica, (clino)zoisite,
230
carbonate and quartz are also typical inclusions in garnet poikiloblasts (e.g., Figs.3a-
231
b). Of particular note is the occurrence of (clino)zoisite and paragonite preserved as
232
inclusions in garnet, forming box-shaped lawsonite pseudomorphs (e.g., Fig.3a; this is
-9-
233
also observed in eclogites) and suggesting a previous lawsonite-bearing blueschist
234
facies metamorphism, i.e., Stage 1 in Fig.4.
235
Except being evidenced by individual inclusions in garnet, prograde
236
(clino)zoisite also shows amoeboid grain boundaries caused by replacement of
237
omphacite (also see Fig.5 in John et al., 2008). The presence of prograde (clino)zoisite
238
in lawsonite stability field may result from the relatively low H2O content in the
239
system (Poli and Schmidt, 1997). Prograde white micas are also present (e.g., as
240
inclusions in glaucophane in Fig.3c), which together with prograde (clino)zoisite, may
241
reflect the Stage 1 as well (Fig.4). Sample TS02-01, a chloritoid-bearing omphacitic-
242
epidosite (not shown in Table 1) has chloritoid poikiloblasts, which contain
243
glaucophane, idioblastic (clino)zoisite and white mica as its variable inclusions, and
244
indicate the transition from Stage 1 to Stage 2 in Fig.4.
245
The idioblastic retrograde (clino)zoisite as well as retrograde white micas (e.g.,
246
one paragonite porphyroblast including one garnet relict in Fig.3d) are also present,
247
reflecting Stage 3 in Fig.4. (Sodic-)calcic amphiboles are common, most of which
248
have clearly replaced other minerals like garnet. Clear zonings, i.e., (sodic-)calcic
249
amphibole rimmed Na-amphibole (reflecting Stage 5 in Fig.4), sometimes even with
250
well-preserved omphacite (e.g., Figs.3c & e reflecting Stage 4 in Fig.4), indicate
251
retrograde metamorphisms from eclogite to (epidote-)amphibolite facies. This is also
252
supported by the replacement of rutile at the rim by titanite (e.g., Fig.3b). Some
253
carbonate minerals are well crystallized, showing an equilibrated texture with the
254
coexisting silicate phases, reflecting their protolith inheritance and metamorphic
255
recrystallization. However, retrograde carbonate is also occurred, as irregular patches
256
or fine veinlets replacing other minerals (including garnet).
257
Based on above discussed mineral assemblages, Fig.4 shows a schematic P-T
258
path to illustrate the concept, although individual samples are likely to have
259
undergone different extents of retrograde metamorphism. The peak metamorphic P-T
- 10 -
260
condition we chose here are the corrected estimates of Lü et al. (2009), i.e., 470 ~
261
510°C at 2.4 ~ 2.7 GPa.
262
In some samples dominated by blueschist facies assemblages (garnet +
263
glaucophane + (clino)zoisite + rutile + titanite + minor white mica and omphacite),
264
omphacite concentrates at the core of veinlets (locally formed omphacitite), which is
265
mainly composed of white mica, quartz and, occasionally, epidote (e.g., TS02-02 &
266
TS02-42; Fig.3h). The veinlet represents fluid channel that was formed at initial stage
267
of dehydration from blueschist to eclogite facies during prograde metamorphism,
268
while omphacite-dominated core represents the residue of this dehydration.
269
ANALYTICAL METHODS
270
Abundances of both major and trace elements in bulk-rock samples were
271
measured at Cardiff University. An Ultima 2 Inductively Coupled Plasma-Optical
272
Emission Spectrometer (ICP-OES) was used for major element analysis and Thermo
273
Element-x7 series Quadrupole Inductively Coupled Plasma-Mass Spectrometer (ICP-
274
MS) was used for trace element analysis. Hand specimens were first processed using a
275
diamond wheel to remove saw marks and weathered surfaces before being rinsed with
276
deionised water. The dried samples were crushed by a clean jaw crusher and
277
powdered with either an agate Tema mill or agate ball mills.
278
To ensure high quality data, complete sample dissolution, especially for
279
refractory minerals such as zircon and rutile, is essential (e.g., Yu et al., 2001). We
280
chose two methods: lithium metaborate fusion for most elements (by which zircon,
281
rutile, titanite etc are readily dissolved) (e.g., Yu et al., 2001), and HF-HNO3
282
dissolution of sample powders for Pb, low abundance elements and elements that are
283
not analysable in LiBO2-fused solutions (such as Cs because of its volatility). Through
284
comparisons with Ultrapure SPEX CertiPrep 100% lithium metaborate, Spectroflux
285
ultraclean 100% lithium metaborate was found to be superior and was chosen as the
- 11 -
286
flux in LiBO2-fused solution method because of its lower Nb content. In the second
287
method, the routine HNO3 washing was used to wash laboratory ware and flush the
288
instrument between analysis runs. Additional flushing of the instrument with HCl was
289
also used to remove Pb from the sample lines. We used JB-1a (a Geological Survey of
290
Japan basalt reference rock standard) as an unknown monitor (APPENDIX 1; Ando et
291
al., 1987). The precision and the accuracy for all the analyzed trace elements are
292
within 5% and ±10% respectively (see APPENDIX 1).
293
Mineral compositions were analyzed using a Cambridge (Leo) S360 scanning
294
electron microscope (SEM) at Cardiff University equipped with an INCA Energy
295
Dispersive X-ray Spectrometry (EDS) of Oxford Instruments monitored with a Co
296
standard. The analyses were done with 20 kV accelerating voltage, 50 seconds
297
counting time and 45% dead time. Data reduction was done using ZAF correction
298
with INCA Energy + software. Natural and synthetic standards from Micro-Analysis
299
Consultancy LTD were used for calibrations. Garnet, ompacite, amphibole, white
300
mica, and epidote groups were analyzed quantitatively. The representative mineral
301
analyses and formula calculations are given in APPENDIX 2. All these data were
302
reported by Lavis (Cardiff University PhD thesis, 2005).
303
MINERALOGY AND BULK COMPOSITION
304
Mineral Compositions
305
Garnet shows a sharp compositional variation with increasing pyrope
306
composition at the rim in most samples (Table A2.1). Clinopyroxene is omphacitic
307
(Fig. 5a and Table A2.1). Amphiboles vary in composition, from typical glaucophane
308
to calcic and sodic-calcic amphiboles including actinolite, barroisite/katophorite and
309
magnesio-hornblende (Fig. 5b). The specific types of amphibole representatives are
310
given in Table A2.2. Both phengitic muscovite and paragonite mica exist (Table
- 12 -
311
A2.1), and are classified in terms of K/Na ratio. Compositions of epidote group
312
minerals are also variable, especially in total FeO (Table A2.2).
313
Major Element Analyses and Reconstructed Bulk-Rock Composition
314
Rocks of sedimentary protolith have high and variable SiO2 (52.09 to 76.66
315
wt%), and relatively low MgO (1.81 to 4.82 wt%). Samples of basaltic protolith have
316
low SiO2 (40.96 to 53.19 wt%) and relatively high MgO (up to 8.56 wt%). One
317
sample of ultramafic protolith (TS02-34) has only 34.01 wt% SiO2.
318
To understand the redistribution of elements (including flux in and out of the
319
system) between mineral phases in our subsequent papers using in situ major and
320
trace element analyses, it is first necessary to perform a quantitative analysis of
321
mineral modes. Detailed modal analyses of these rocks have been done by point-
322
counting using a James Swift automatic point counter, model F 415C. The modal data
323
are given in Table 1. Comparisons of reconstructed bulk-rock major element
324
compositions (the calculation method and the calculated results are given in
325
APPENDIX 3) with actual analyses indicate that our estimations of mineral modes are
326
reliable, although some uncertainties still exist (see discussion in APPENDIX 3).
327
TRACE ELEMENT MOBILITY AND GEOCHEMISTRY OF SAMPLE
328
GROUPS
329
Sample Grouping
330
Experimental studies (e.g., Manning, 2004; Kessel et al., 2005; Hermann et al.,
331
2006) have shown that high field strength elements (HFSE, i.e., Ti, Zr, Hf, Nb, Ta)
332
together with Y and the middle to heavy rare earth elements (M-HREE), are usually
333
not mobile during fluid-rock interactions. These studies also show that Th and LREE
334
are usually immobile until temperatures approach the solidus. Alkali and alkali earth
335
elements such as Rb, Cs and Ba are usually mobile, as are most of the primary
- 13 -
336
elements used for the classification of recent lavas (i.e., Na and K). For this reason,
337
immobile elements should be used for the classification and the tectonic
338
discrimination of metamorphic protoliths (e.g., Pearce and Cann, 1973; Winchester
339
and Floyd, 1977).
340
The Zr/Ti vs. Nb/Y diagram (Winchester and Floyd, 1977) is often used as an
341
immobile trace element equivalent of the TAS (total alkali vs. silica) diagram to
342
classify volcanic rocks with varying degrees of alteration/metamorphism in the
343
geological records. In Figure 6a, which uses the updated boundaries of Pearce (1996),
344
samples of basaltic protolith plot as three clusters mainly in the sub-alkalic basalt field.
345
They display a small range in Zr/Ti ratio (equivalent to small SiO2 variations) but a
346
large range in Nb/Y ratio (equivalent to large total alkali variations). We term these
347
Group 1, Group 2 and Group 3 (see Table 1 and 2 for mineral assemblages and
348
geochemical data, respectively).
349
Provided Th, Nb, Ti and Yb are immobile throughout the metamorphic process,
350
the Th/Yb vs. Nb/Yb and Ti/Yb vs. Nb/Yb diagrams (Pearce, 2008) are effective in
351
finger-printing the tectonic setting of magma eruption. The first of these diagrams
352
features a diagonal MORB-OIB array containing N-MORB, E-MORB and OIB
353
compositions (Fig.6b). Basalts with a subducted or assimilated continental crustal
354
component plot above this array; these include VAB, some back-arc basin basalts and
355
basalts contaminated with continental crust-like materials. For our samples, the three
356
groups defined in Fig.6a are also well separated on this projection. Samples from
357
Group 1 and Group 3 plot within or at the field edge of the MORB-OIB array. Their
358
protoliths are most likely OIB- or E-MORB-like and depleted MORB-like rocks
359
respectively. The high Ti/Yb ratios of the former on the Ti/Yb-Nb/Yb diagram (not
360
shown) demonstrate that these are OIB-like rather than E-MORB-like.
361
In contrast, Group 2 samples define a distinct trend away from the MORB-OIB
362
array into the volcanic arc/contamination field. Because this trend includes samples
- 14 -
363
within the MORB-OIB array, it is likely to reflect the effect of contamination of
364
MORB-like magma by continental crust or a mixing relationship between MORB and
365
arc magmas (Pearce, 2008). Group 2 samples will be excluded from the study of
366
element mobility (below) because alteration effects and contamination (or mixing)
367
effects cannot be easily resolved in the bulk rock studies.
368
Determination of Element Mobility/Immobility
369
Although HFSE and HREE have been widely considered to be immobile, there is
370
some evidence that they are not always conservative during SZM (e.g., Gao et al.,
371
2007; John et al., 2008), a possibly predicated consequence of the extreme
372
compositions of some subduction-zone fluids (e.g., supercritical H2O-rich fluids, CO2-
373
rich fluids, and perhaps, even SiO2-rich hydrous melt; Manning, 2004; Kessel et al.,
374
2005). Therefore, to properly understand the geochemical consequence of SZM, it is
375
necessary to examine the actual mobility/immobility of each element in our samples.
376
Because element behaviour during formation of the protoliths is governed by
377
partition coefficient during partial melting/crystallisation, elements with similar
378
incompatibility in igneous protolith are expected to be significantly correlated. Based
379
on this principle, one simple way to examine the element mobility/immobility is to
380
see if these statistically significant correlations are preserved or not in our
381
metamorphic samples, i.e., whether these elements have been fractionated from each
382
other during metamorphism (e.g., Niu, 2004). If the significant relationship between
383
similarly incompatible elements is preserved in the metamorphic rocks, it indicates
384
that these two elements have remained immobile, otherwise, one of them or both may
385
have be mobilised during metamorphism. Note that two mobile elements with similar
386
incompatibility may also exhibit significant correlations. Therefore, we select two
387
HFSE with different incompatibilities (i.e., Nb and Zr), and investigate their
- 15 -
388
correlations with other trace elements with varying incompatibility, i.e., Nb with LILE,
389
LREE, Th, U and Ta, while Zr with MREE, some LREE and Hf (Fig.7).
390
Investigations of correlation coefficients are carried out following the methods of
391
Niu (2004) and Lavis (2005) for co-genetic rocks of basaltic protolith from Group 1
392
(Figs.7a-b) and Group 3 (Figs.7c-d), also for rocks of sedimentary protolith in Table 3
393
(although the method is less effective in meta-sedimentary rocks than that for meta-
394
basaltic rocks). Fig.7 shows correlation coefficients of the immobile incompatible
395
elements Nb and Zr with other trace elements, which are sorted in the order of
396
decreasing incompatibility from left to right following the order reported by Niu and
397
Batiza (1997). For N = 23 (Group 1) and 6 (Group 3), R > 0.482 and R > 0.882
398
respectively are statistically significant at > 99 % confidence levels, and we treat
399
values above these as statistically significant correlations. In Fig. 7, we find that Ba,
400
Cs, Rb, K, Pb and Sr are not significantly correlated with Nb and Zr, and have
401
therefore, been mobilized in our samples of basaltic protolith, during either prograde
402
or retrograde metamorphism, or prior to SZM (i.e., during seafloor alterations). All
403
other elements (i.e., REE, HFSE, Th and U) show statistically significant correlations
404
with Nb and Zr have thus remained largely immobile. All these element geochemical
405
behaviours are consistent with the experimental results as a function of interactions of
406
rocks with aqueous fluids (vs. supercritical liquids, e.g., Kessel et al., 2005; Hermann
407
et al., 2006).
408
Previous studies (e.g., John et al., 2004) have demonstrated LREE, U and Th can
409
be fractionated from HFSE and HREE during SZM. However, Fig.8e shows that the
410
U/Nb ratio of our samples is constrained within a small range similar to that of N-
411
MORB, except for samples from Group 2 that have been attributed to the
412
contamination or mixing with materials with high, possibly variable U/Nb ratios, such
413
as continental crust (see above). In addition, Nb/La ratios of our samples are also
414
relatively constant, i.e., 0.94±0.15 (1σ; Group 1) and 0.45±0.06 (1σ; Group 3).
- 16 -
415
Together with the significant linear correlations between U and Th (Fig.8f), all of
416
these suggest that U, Th and LREE in our samples, have not been mobilized since the
417
formation of their protolith, as predicted for HFSE and HREE. The compositional
418
range of AOC is relatively large with respect to the U content, which may be enriched
419
by up to 5 times in AOC, as shown in the super-composite of tholeiitic section of
420
ocean crust (Kelley et al., 2003). Thus, AOC may show a poor correlation of U with
421
immobile elements and have lower Th/U ratios (or Nb/U) in some rocks due to U
422
enrichment during seafloor alterations. Therefore, the significant linear correlation of
423
U with Th (or Nb, not shown) and the inferred immobility of U, as well as the
424
distinctive Th/U ratios of all our samples from AOC (Fig.8f) may suggest that the
425
component of AOC is not involved greatly in our samples of protolith. Thus the
426
mobility of LILE is most likely caused by SZM, i.e., during prograde or/and
427
retrograde metamorphism, rather than seafloor alteration.
428
The significant inter-element correlations of HFSE and REE (Table 3), and the
429
good correlations of K–Rb–Cs–Ba with HFSE at > 99% confidence levels in rocks of
430
sedimentary protolith (Figs.8a-d) suggest their rather similar behaviour during SZM.
431
As HFSE are immobile, K–Rb–Cs–Ba and REE in samples of sedimentary protolith
432
are also immobile. The poor correlations of Pb and Sr with HFSE may indicate their
433
mobility during SZM. Therefore, based on above discussion, Pb and Sr are mobile
434
while REE, HFSE, Th and U are immobile in both lithologies. The
435
mobility/immobility of K–Rb–Cs–Ba are dependent on different lithologies, i.e.,
436
immobile in rocks of sedimentary protolith vs. mobile in rocks of basaltic protolith.
437
Recently, Gao and co-workers (Gao et al., 2007; John et al., 2008) concluded
438
that Nb, Ta, Ti, REE and LILE are all mobile during SZM based on a localized
439
fluid/vein assemblage from Tianshan. Although their observations may be valid, we
440
emphasize that caution is necessary to generalize local phenomena to the prevailing
- 17 -
441
nature of SZM. Our data demonstrate clearly that REE and HFSE are largely, if not
442
entirely, immobile during SZM (Figs.7-8).
443
Geochemistry of Different Groups
444
Fig.9 shows N-MORB normalised multi-element diagrams for rocks of basaltic
445
protolith (a-c) and for rocks of sedimentary protolith (d). All the mobile elements
446
discussed above (i.e., LILE, Pb and Sr in the grey areas in Figs.9a-c) display greater
447
variability than immobile elements due to the “fluid effect” during SZM in the case of
448
samples of basaltic protolith. Mobile elements will not be used in the following
449
discussion on the geochemistry of protoliths and their tectonic settings. Immobile
450
incompatible elements are generally enriched in Group 1, with a pattern similar to the
451
average composition of OIB. Group 2 displays a relatively flat pattern with stronger
452
enrichments of U and Th relative to Nb and Ta (i.e, higher [Th/Nb]N-MORB and
453
[U/Ta]N-MORB), which are comparable (or even much higher in most samples) to the
454
model
455
contamination/mixing by continental or arc materials (see Fig.6b) as well as the above
456
discussion that the component of AOC is absent in the protoliths (in Group 1 and
457
Group 3), Group 2 protolith may thus be likely derived from an N-MORB source with
458
subsequent contaminations by continental crust or arc materials rather than simple
459
seafloor alterations. Two samples (i.e., TS02-15A & TS02-15B) from Group 2
460
(Fig.9b) show overall elevated trace element abundances and lower MgO contents of
461
4.15 and 4.39 wt% respectively (Table 2), which are consistent with the protolith
462
being more evolved basalt. Samples from Group 3 are depleted in the progressively
463
more incompatible elements and enriched in the less incompatible elements (such as
464
HREE), which shows a similar pattern observed in highly depleted MORB-like
465
basalts (Fig.9c).
composition
of
AOC
(Kelley
et
al.,
2003).
Considering
the
- 18 -
466
As discussed above, LILE are considered as immobile elements in rocks of
467
sedimentary protolith, therefore the large concentration range of LILE shown in
468
Fig.9d is most likely caused by the compositional heterogeneity of sedimentary
469
protolith, rather than the mobility of LILE. In contrast, variations of Pb and Sr
470
concentrations are most likely attributed to their fractionations owing to their mobility.
471
Most samples of sedimentary protolith have similar compositions to global oceanic
472
subducted sediments (GLOSS; Plank and Langmuir, 1998) with enriched LILE and
473
depleted HFSE, while those depleted in LILE are likely rocks of greywacke protolith
474
with arc-basaltic lithic fragments.
475
Considering both the field study in Western Tianshan (e.g., Volkova and
476
Budenov, 1999; Ai et al., 2006) and the geochemical variability of modern seamounts
477
from enriched OIB-like to depleted MORB-like basalts (such as those observed at
478
near East Pacific Rise (EPR)-seamounts, Niu and Batiza, 1997), we suggest that the
479
UHP metamorphic rocks from the Western Tianshan may resulted from the
480
subduction and exhumation of a tectonic mélange, which is composed of normal
481
seafloor with dismembered seamounts, sediments as well as volcanic arc derived
482
materials.
483
DISCUSSION
484
Constraints on Trace Element Budgets
485
General Thoughts
486
In the previous standard model of SZM, the transition from blueschist to eclogite
487
facies was thought to mark the major dehydration reaction within the subducted
488
oceanic crust (e.g., Peacock, 1993; Liu et al., 1996). However, in recent studies (e.g.,
489
Spandler et al., 2003, 2004), it has been proposed that this transition does not
490
necessarily result in the liberation of mobile elements with released fluids, which is
- 19 -
491
consistent with the poor correlations of fluid-soluble element contents with modal
492
glaucophane and LOI in our samples (e.g., Figs.10d & h).
493
Recognition and experimental studies of HP-UHP metamorphic hydrous
494
minerals (e.g., phengite, lawsonite, (clino)zoisite and allanite) under subduction zone
495
conditions (e.g., Pawley and Holloway, 1993; Schmidt and Poli, 1998) and studies on
496
the distribution of fluid-soluble elements in different minerals (e.g., Spandler et al.,
497
2003; Bebout et al., 2007; El Korh et al., 2009), all indicate that these minerals are
498
capable of accommodating fluids in the form of hydroxyl or H2O as well as their
499
preferential fluid-soluble elements. Specifically, phengite is the dominant host of K–
500
Rb–Cs–Ba in the whole rock (e.g., Zack et al., 2001); epidote group minerals (i.e.,
501
epidote, (clino)zoisite, and allanite) and lawsonite, can host most Sr, Pb, U, Th and
502
LREE (e.g.,Hemann, 2002; Feineman et al., 2007). Therefore, the fluids and the
503
otherwise mobile elements can be carried in the subducting slab for as long as and as
504
deep as the host hydrous minerals remain stable, e.g., 3.2 GPa for zoisite (~ 100 km)
505
(even up to 5.0 GPa at 700°C and 6.6 GPa at 950°C according to Poli and Schmidt,
506
1998), up to 12 GPa for lawsonite (~ 360 km; Schmidt, 1995), 10 GPa for phengite (~
507
300 km) (e.g., Sorensen et al., 1997), K-hollandite to the deep mantle (i.e., > 16 - 23
508
GPa for LILE, Rapp et al., 2008). These depths are far greater than the depth of
509
amphibole stability (commonly 2 ~ 2.6 GPa at 700°C reported by Pawley and
510
Holloway, 1993, although it could be stable at higher pressures) and thus it is those
511
hydrous metamorphic minerals that are stable at the UHPM conditions that control
512
fluid-mobile element behaviour during SZM.
513
Constraints on the Budget of LILE
514
It is practically impossible to distinguish muscovite and paragonite
515
petrographically unless carrying out grain-by-grain analysis using an electron
516
microprobe. Thus, for convenience in our modal analyses, these minerals were
517
considered together as “white mica”. Importantly, our yet-to-be submitted data
- 20 -
518
suggest that both muscovite and paragonite, which occur as solid solutions, have
519
important controls on LILE, although LILE concentrations in muscovite are about an
520
order of magnitude greater than that of paragonite. Therefore, incorrect identification
521
of muscovite-rich or paragonite-rich grains could potentially affect estimated bulk-
522
rock major element and trace element concentrations based on mineral modal
523
abundances. We have found, however, that the expected positive trend between the
524
total modal white mica and the bulk-rock LILE does, to a first order, exist in our
525
samples of both sedimentary (Figs.10a-c) and basaltic protoliths (Figs.10e-g),
526
although correlations of the latter are insignificant above 99 % confidence levels.
527
The significant positive correlations of white mica modal abundances with K–
528
Rb–Cs–Ba in samples of sedimentary protolith (Figs.10a-c, not shown for K) indicate
529
that these elements are hosted by white mica as expected. The apparent immobility of
530
these elements in samples of sedimentary protolith discussed above suggests that
531
white mica, as the most important host of LILE, may have existed in the protolith and
532
re-crystallized subsequently during metamorphism. Alternatively, the protolith may
533
have had abundant clay minerals, with K–Rb–Cs–Ba adsorbed to the clay mineral
534
surfaces, and conserving these elements when these clay minerals converted to white
535
mica. Whatever the actual case, abundances of LILE in sedimentary protolith
536
remained unaffected by metamorphism because minerals into which they can partition
537
have been stable throughout the petrogenetic history of the host rocks.
538
For rocks of basaltic protolith, the positive trend but insignificant correlations
539
between Rb–Cs–Ba and white mica (Figs.10e-g), suggest that these elements may be
540
not simply controlled by white mica. Unlike sedimentary rocks, white mica is absent
541
in basaltic protolith and its presence in rocks of basaltic protolith is of metamorphic
542
origin. Therefore, both compositions of the protolith and later metamorphic processes
543
play important roles in determining the mobility/immobility of LILE. Given the low
544
concentrations of K–Rb–Cs–Ba in bulk-rock compositions of average ocean crust (see
- 21 -
545
Table 1 in Niu and O'Hara, 2003), little muscovite may be produced from this kind of
546
protolith during metamorphism. Nevertheless, E-MORB- or OIB-like protoliths, or
547
AOC (see Table 6 in Kelley et al., 2003), relatively enriched in K2O compared with
548
N-MORB, could supply enough K to produce significant volumes of white mica,
549
which is evidenced by more commonly present white mica in samples from Group 1
550
(OIB-like source) than that in samples from Group 3 (depleted N-MORB-like source).
551
On the other hand, the relative timing of white mica formation during metamorphism
552
could also dictate the behaviour of those mica-hosted elements. The origin of the poor
553
correlation between LILE and white mica could have resulted from adding elements
554
during retrograde metamorphism (see discussion below), or losing from the rocks
555
during the early stages of dehydration prior to the crystallization of white mica.
556
However, once white mica has been formed, especially phengitic muscovite, most
557
LILE will be strongly conserved, and only some LILE may be lost through the
558
decomposition of phengite above 5 GPa due to the increased potassium solubility in
559
clinopyroxene (Schmidt and Poli, 1998).
560
Constraints on the Budgets of Pb and Sr
561
The correlations of (clino)zoisite with its favoured trace elements, i.e., REE, U,
562
Th, Sr and Pb (not shown), are insignificant in both meta-sedimentary and meta-
563
basaltic rocks. Unlike white mica, which is the exclusive accommodation of K-Rb-Cs-
564
Ba, (clino)zoisite is not the sole host for REE, U, Th, Sr and Pb (note that the bulk-
565
rock contents of Pb and Sr are not significantly correlated at > 99% confidence levels).
566
Based on the results reported by recent literatures (e.g., van der Straaten et al., 2008;
567
Beinlich et al., 2010) and our LA-ICPMS analyses of these rocks, Sr (and maybe Pb)
568
can also be accommodated by carbonate and paragonite; REE can also be
569
incorporated into titanite, as can Sr, Th, U and Pb to a lesser extent. The influence of
570
these additional phases may therefore explain the observed poor correlations of
571
(clino)zoisite with its favoured trace elements. Alternatively, considering the
- 22 -
572
immobility of REE, U and Th in the bulk rock, the poor correlation is also likely
573
caused by heterogeneous distributions of trace elements in (clino)zoisite. In contrast,
574
the poor correlations with Sr and Pb may also reflect the loss/gain of these two
575
elements during the transition from lawsonite to (clino)zoisite or the transition
576
between different (clino)zoisite generations.
577
Implications for Subducion-Zone Magmatism
578
Elemental Contributions to Arc Lavas: LILE, Sr and Pb
579
From above discussion, it can be seen that although the transition from blueschist
580
to eclogite facies can theoretically release significant amounts of fluids (Pawley and
581
Holloway, 1993) and fluid-soluble elements for the formation of arc lavas, this may
582
not necessarily be always the case because some eclogites remain enriched in LILE,
583
e.g., TS02-41. This highly eclogitized sample without obviously retrograde overprints,
584
which contains 12.9 wt% white mica and only 0.67 wt% (clino)zoisite, has an
585
enriched K-Rb-Cs-Ba and depleted Sr signature (Fig.9a). It is white mica that survives
586
from the blueschist-to-eclogite dehydration retains/transports these elements into the
587
deep mantle. Therefore, the appearance and stability of the HP-UHP hydrous minerals
588
(e.g., white mica, (clino)zoisite etc.), which control the immobility/mobility of fluid-
589
soluble elements during SZM (the detailed process is illustrated by Figs.11d-e),
590
consequently affect the geochemical signatures of arc lavas. Furthermore, it is the
591
major element composition (including H2O) and the oxidation state of protolith that
592
control modal abundances of hydrous minerals (e.g., Rebay et al., 2010).
593
During the prograde P-T path of our samples, a significant portion of Sr and Pb
594
may have first been accommodated in lawsonite (as indicated by inclusions composed
595
of paragonite + clinozoisite in garnet, e.g., Fig.3a). Accompanied by prograde-
596
retrograde P-T path transition (see Fig.4) and continued heating for a period thereafter
597
with decreased P/T ratios, (clino)zoisite replaced lawsonite and scavenged the
- 23 -
598
liberated Sr and Pb. The presence of prograde (clino)zoisite indicates that some of the
599
Sr and Pb may also have been greatly hosed by (clino)zoisite because of low H2O
600
available in the original system (the H2O content decides the modal lawsonite under
601
blueschist facies, i.e., the higher H2O content, the more lawsonite there may be, e.g.,
602
Rebay et al., 2010; Poli and Schmidt, 1997). Up to about ~100 km or even deeper
603
depths in relatively cold subduction zones (~12 GPa; Schmidt, 1995), (clino)zoisite or
604
lawsonite will transform to garnet and omphacite, which cannot further accommodate
605
Sr and Pb. Consequently, these two elements will be liberated along with released
606
fluids into the subduction channel and perhaps, contribute to the partial melting of the
607
mantle wedge and the further formation of arc lavas/back-arc magmas (Fig.12).
608
If paragonite or
muscovite was present prior to SZM, LILE would be
609
transported within the subducted slab to a depth of at least ~ 70 km. This situation is
610
most likely expected in subducting sedimentary rocks or subducting E-MORB-like or
611
OIB-like rocks, with enough K to produce muscovite. Rb-Ba-Cs will be stabilised
612
then (refer to Figs.11d2-e2 for the trace element redistributed model) until the
613
muscovite breaks down at a depth up to ~ 300 km (e.g., Sorensen et al., 1997; Poli
614
and Schmidt, 2002). If there are no further hydrous minerals (e.g., K-hollandite, Rapp
615
et al., 2008) for hosting these elements, they will contribute to the enrichment of LILE
616
in back-arc lavas (Fig.12). Otherwise, if no paragonite or phengitic muscovite was
617
present before dehydration reactions during SZM, these elements will be released
618
with fluids beneath the forearc (refer to Figs.11d1-e1 for the trace element
619
redistributed results), which is most likely the case for subducting depleted N-MORB-
620
like rocks without significant seafloor alterations (i.e. low K and H2O).
621
Elements such as LREE, U and Th, which are typically enriched in VAB, have
622
previously been considered as being mobile elements in SZM (e.g., McCulloch and
623
Gamble, 1991). Our studies, however, have revealed the immobility of these elements
624
in both meta-sedimentary and meta-basaltic rocks, i.e., since the formation of the
- 24 -
625
protoliths. This may indicate that LREE, U and Th have not been released during
626
SZM for the formation of the subduction-zone magmatism, up to this stage at least.
627
Instead, these elements would be transported to the deeper mantle.
628
On the other hand, channelized fluids have been proposed as an important
629
medium for migrating those fluid-soluble elements in the system (John et al., 2004;
630
Zack and John, 2007; Bebout, 2007; John et al., 2008). The observed dehydration
631
veins in several samples on thin section scales (e.g., Fig.3h) are mainly composed of
632
white mica and quartz without or with only minor epidote, in the core of which is
633
exclusively made up of omphacite representing the dehydrated residue of blueschist.
634
Given that white mica is enriched in LILE while (clino)zoisite is enriched in REE, Th,
635
U, Pb and Sr, the dominant occurrence of white mica but the lack of (clino)zoisite in
636
the vein reflects the highly enriched LILE and the depleted LREE in the fluid. It
637
probably implies that there is only a great loss of LILE with the released fluids but
638
LREE are less mobile or even immobile, at least on thin section scales.
639
Redox Conditions in Subduction Zones
640
Another important feature in the standard subduction model is its presumed
641
highly oxidized state, as indicated by high Fe3+/Fe2+ observed in arc volcanic rocks
642
and in spinels from sub-arc mantle xenoliths. These oxidised conditions are thought to
643
be caused by the addition of slab-derived fluids (e.g., Wood et al., 1990; Brandon and
644
Draper, 1996). However, based on V/Sc ratios in sub-arc peridotites and arc lavas
645
rather than Fe3+/Fe2+, Lee et al. (2003, 2005) proposed that the oxygen fugacity in arc
646
magma sources is surprisingly similar to that of asthenospheric mantle beneath ocean
647
ridges, as low as -1.25 to 0.5 log units relative to the FMQ (Fayalite-Magnetite-Quartz)
648
buffer. They also pointed out that it is inappropriate to use Fe3+/Fe2+, which only
649
reflects the redox state in the last equilibrium process without any records of previous
650
equilibrium history. The observation of CH4 fluid inclusions in olivine from an
651
interpreted mantle wedge harzburgite sampled from the Qilian Suture Zone,
- 25 -
652
Northwest China (Song et al., 2009) also indicates that the actual redox state in
653
subduction zones may be more reduced than widely assumed.
654
As U becomes fluid soluble or mobile only at its +6 valence under oxidized
655
conditions (e.g., Langmuir, 1998; Niu, 2004), the existence of a significant U-Th
656
correlation (Fig.8f; not only for our studied subduction-zone metamorphic rocks) and
657
the absence of fractionations of U from Nb (Fig.8e) indicates that U may be in the
658
form of U4+, not U6+ during SZM. The above observations, together with the
659
identification of graphite inclusions in porphyroblastic garnet from eclogite in
660
Western Tianshan (Lü et al., 2008, 2009), all indicate that SZM takes place in a more
661
reducing environment than previous thought. Our claim that U is immobile due to
662
reducing SZM conditions, plus the argument by Lee et al. (2003, 2005) and Song et al.
663
(2009) suggest that more effort is needed to understand the oxidation state of
664
subduction zones at UHPM conditions.
665
Influence of Retrograde Metamorphism
666
Our samples have generally experienced different extents of retrograde
667
metamorphism during their exhumation. For example, TS02-03A and TS02-03B are
668
from the same hand specimen. The latter sample is a classic eclogite, while the former
669
has retrogressed to a blueschist assemblage with clear retrograde overprints (e.g.,
670
Figs.3c & e). Owing to the absence of white mica, TS02-03B has been strongly
671
depleted in Rb-Cs-Ba. However, as the result of localised rehydration, the relatively
672
dry eclogite has been transformed to wet blueschist facies only in TS02-03A. The
673
released K in the fluids (which may be from an external source) led to the formation
674
of greater white mica modal abundance and higher Rb-Cs-Ba concentrations, which in
675
TS02-03A can be up to ~ 100 times more enriched than that of TS02-03B (as
676
illustrated by the vertical dashed arrows in Fig.9a). Therefore, in order to understand
677
the actual composition of rocks that have undergone SZM (i.e., [P/D]out) and the
- 26 -
678
implications for mantle compositional heterogeneity, we need to evaluate the
679
geochemical effects of retrograde metamorphism and exclude these from the
680
geochemical features of samples what we observe now.
681
Whether the metasomatized rocks will be enriched (i.e., in LILE, Sr and Pb) or
682
not by the rehydration process during retrograde metamorphism, depends on the
683
composition of the rocks from which the fluids originated and through which the
684
rising fluids have passed, e.g., enriched OIB-/depleted N-MORB-like basaltic rocks,
685
AOC, sediments or within-slab serpentine like ‘‘abyssal peridotites’’.
686
The serpentinized lithospheric mantle of the subducted slab is one of the most
687
significant fluid reservoirs (e.g., Schmidt and Poli, 1998) for the retrograde
688
metamorphism and plays an important role in the exhumation of subducted oceanic
689
lithosphere due to its relatively lower density (e.g., Pilchin, 2005). Because the fluid
690
released from serpentines is likely rich in Mg and poor in LILE, the dehydration of
691
serpentinite itself cannot directly make the subducted crustal rocks more enriched (see
692
Niu, 2004; van der Straaten et al., 2008). Rather, lithologies modified by adding fluids
693
from serpentine will be even more depleted (Fig.11i). However, the depleted fluids
694
derived from serpentine could scavenge a significant portion of elements from the
695
reacted rocks they passed through, especially those enriched rocks (Figs.11g-h; van
696
der Straaten et al., 2008).
697
Alternatively, fluids derived from rocks that subducted beyond our samples and
698
migrated up-dip along the subduction channel (Fig.11g) to metasomatize shallower
699
rocks rather than just migrating into the overlying mantle wedge. Some of these fluids
700
can even induce transformation of the shallower subducted rocks from dry eclogites to
701
blueschist-like assemblages. As we have discussed above, if no white mica was
702
present prior to dehydrations, Rb-Ba-Cs are most likely to be released from mafic
703
eclogites. Then, these fluids with LILE could further rehydrate the dry eclogitic rocks
- 27 -
704
at shallower levels and lead them to be enriched in the relevant elements as well (see
705
Fig.11h).
706
Implications for Mantle Compositional Heterogeneities
707
We have argued that fluids and a portion of fluid-soluble elements (i.e., LILE, Sr
708
and Pb) could be liberated into the mantle wedge from rocks of basaltic protolith,
709
while rocks of sedimentary protolith may host K, Rb, Cs and Ba in phengite, stable up
710
to depths of ~ 300 km. At greater depths, K-hollandite could become the stable phase
711
hosting these elements (Rapp et al., 2008).
712
Because our samples have not actually subducted beyond the depth for the
713
formation of arc lavas (i.e., ~108 km according to Tatsumi and Kogiso, 2003), they
714
cannot represent the actual [P/D]out brought to the deep mantle (see Fig. 1 and 12).
715
However, systematic studies of the controls on the behaviour of different elements
716
and the direct geochemical consequence of SZM for the depth > 75 km in our study
717
still allow us to reasonably infer the [P/D]out into the deeper mantle. Therefore, we
718
selected several eclogitic samples with minimal retrograde metamorphic overprints
719
from Group 1 and Group 3 (the sample’s name shown in bold in Table 1), and plotted
720
their P/D ratios against ratios of immobile elements (i.e., Nb/Zr and Th/U) (Fig.13).
721
With relatively constant Nb/Zr and Th/U ratios in each group, Rb/Sr and U/Pb
722
ratios vary significantly (Figs.13a-d), which is expected as Rb, Sr and Pb are mobile.
723
In contrast, Sm/Nd and Lu/Hf ratios are almost constant (Figs.13e-h), consistent with
724
their inherited magmatic signature and immobile behaviour. Furthermore, the lack of
725
correlations of Sm/Nd and Lu/Hf with Rb/Sr (Fig.14a) implies that these rocks, when
726
retuning into mantle source regions, cannot contribute to the observed first-order
727
negative Sr-Nd isotope correlation seen in oceanic basalts, even though the significant
728
positive Sm/Nd - Lu/Hf correlation is apparently consistent with the
729
176
143
Nd/144Nd vs.
Hf/177Hf correlation in oceanic basalts (Fig.14b). Therefore, the subducted ocean
- 28 -
730
crust cannot contribute to the major source materials of oceanic basalts in terms of Sr-
731
Nd-Pb radiogenic isotopes (also see Niu and O’Hara, 2003; Niu, 2004).
732
The
immobility
of
U
during
SZM
and
Pb
fractionation
in
the
733
subducting/subducted slab, on the other hand, indicate that U/Pb in the subducted
734
crust will have increased [P/D]out and will lead to a higher ratio of U/Pb in the deep
735
mantle (there may be an elevated ratio of Th/Pb as well due to similar behaviour of Th
736
as U during SZM, e.g., Fig.8f). Therefore, such fractionations of U-Th (especially U)
737
from Pb may contribute to the source of HIMU.
738
CONCLUSIONS
739
We have attempted to characterize element behaviours in response to SZM using
740
detailed petrography and bulk-rock geochemistry of subduction-zone UHP
741
metamorphic rocks from Chinese Western Tianshan. Several tentative conclusions are:
742
1. Based on the studies of immobile elements, our samples of basaltic protolith
743
can be divided into three groups: 1) Group 1 points to a protolith resembling the
744
present-day average OIB (or enriched seamount lavas); 2) the protoliths of Group 2
745
are similar to N-MORB parental magmas contaminated by continental margin
746
materials; and 3) the protoliths of Group 3 are akin to highly depleted MORB. The
747
association of these diverse meta-basaltic compositions together with rocks of
748
sedimentary protolith indicate that the input into subduction zones can be very diverse.
749
We agree with previous authors (e.g., Gao and Klemd, 2003; Ai et al., 2006) that the
750
occurrence of these diverse lithologies in the UHP metamorphic belt, indicates that
751
subducted/exhumed complexes represent a tectonic mélange composed of normal
752
ocean crust with dismembered seamounts, sediments and arc derived materials. We
753
have also argued that, the composition of the protolith plays an important role in
754
determining the mineral assemblages and elemental fractionations during SZM.
- 29 -
755
2. We have demonstrated and confirmed that HFSE and REE are in general
756
immobile during SZM. The significant correlation of U with Th (and Nb) indicates
757
that U is immobile and points to a reduced environment, contrary to the common
758
perception that subduction zones are relatively oxidized. The immobility of LREE, U
759
and Th further implies that they cannot contribute to subduction-zone magmatism up
760
to 75 km at least (i.e., > 2.5 GPa). In addition, the immobility of U in this study may
761
suggest that the component of AOC is absent in our samples of protoliths.
762
3. LILE including Sr and Pb in rocks of basaltic protolith have been mobilised,
763
although it is not clear whether this happened at the same time for all samples; the
764
possible timing for the mobilisation is most likely during prograde or retrograde
765
metamorphism. LILE in meta-sedimentary rocks appear to have remained immobile,
766
most likely because white mica (or clay minerals) into which these elements partition,
767
has been present since the formation of their protolith. For meta-basaltic rocks, the
768
timing of the appearance and stability of white mica during seafloor subduction is
769
important in determining the mobility/immobility of their preferentially hosted
770
elements. However, further studies on trace element budgets in natural samples from
771
other HP and UHP metamorphic suites are needed to verify this result.
772
4. Compared with the abundances and ratios of immobile elements, it is
773
confirmed that both abundances and ratios of mobile elements have been altered
774
during SZM, hence the [P/D]in (i.e., Rb/Sr and U/Pb input into the trench) differs from
775
the [P/D]out of subduction-zone metamorphic system. Therefore, it is inappropriate
776
and misleading to use [P/D]out in the subducting slab, inferred from oceanic basalts, to
777
constrain the [P/D]in of surface rocks that have entered trenches.
778
- 30 -
779
APPENDIX
APPENDIX 1. DATA QUALITY ESTIMATIONS
780
781
TABLE A1. ANALYTICAL RESULTS OF UNKNOWN SAMPLE JB-1a
Method
LOD (ppm)
LOQ (ppm)
RSD*(%)
Average JB-1a
JB-1a (lab-cert)
Difference (%)
Sc
HCl washed
0.02
0.08
1.21
29.4
28.3
3.80
Nb
HCl washed
0.00
0.02
0.97
30.3
27.8
8.68
Ta
HCl washed
0.00
0.00
1.75
1.84
1.76
4.96
Pb
HCl washed
0.02
0.06
2.22
7.18
6.69
7.26
Rb
HCl washed
0.02
0.05
0.90
41.8
39.9
4.67
Cs
HCl washed
0.00
0.01
0.82
1.28
V
LiBO2-fused
0.66
2.20
4.64
208
1.19
193
7.59
7.46
MnO
LiBO2-fused
0.51
1.69
4.91
0.15†
0.15†
2.98
Zn
LiBO2-fused
0.82
2.73
3.75
80.4
Ga
LiBO2-fused
0.02
0.08
4.39
18.8
74.3
17.8
8.29
5.49
Sr
LiBO2-fused
0.30
1.00
1.40
461
451
2.09
Y
LiBO2-fused
0.42
1.42
2.23
24.2
23.7
1.95
Zr
LiBO2-fused
0.06
0.19
2.03
145
143
1.33
Ba
LiBO2-fused
0.41
1.36
2.27
512
513
-0.03
La
LiBO2-fused
0.01
0.04
4.36
39.2
38.9
0.75
Ce
LiBO2-fused
0.01
0.02
4.11
66.7
65.6
1.63
Pr
LiBO2-fused
0.00
0.01
3.94
7.38
7.20
2.56
Nd
LiBO2-fused
0.01
0.02
3.71
26.4
26.6
-0.70
Sm
LiBO2-fused
0.01
0.02
3.83
5.16
5.05
2.13
Eu
LiBO2-fused
0.00
0.01
3.73
1.55
1.51
2.60
Gd
LiBO2-fused
0.03
0.10
4.75
4.84
4.80
0.87
Tb
LiBO2-fused
0.01
0.03
3.54
0.73
0.72
1.24
Dy
LiBO2-fused
0.00
0.01
3.65
4.15
4.04
2.90
Ho
LiBO2-fused
0.00
0.00
3.90
0.81
0.78
3.56
Er
LiBO2-fused
0.00
0.01
4.45
2.17
2.17
-0.01
Tm
LiBO2-fused
0.00
0.00
4.78
0.33
0.33
-1.65
Yb
LiBO2-fused
0.00
0.01
3.45
2.12
2.08
1.87
Lu
LiBO2-fused
0.00
0.01
3.66
0.32
0.32
-1.52
Hf
LiBO2-fused
0.00
0.01
3.66
3.56
3.50
1.60
Th
LiBO2-fused
0.00
0.01
4.62
8.60
8.58
0.17
U
LiBO2-fused
0.00
0.02
3.44
1.74
1.62
6.96
Cr
HNO3 washed
0.28
0.94
2.23
428
468
-8.51
Co
HNO3 washed
0.01
0.04
1.90
37.9
38.2
-0.69
Cu
HNO3 washed
2.08
6.93
0.74
55.5
58.3
-4.79
Note: LOD and LOQ are calculated from 10 repeated analyses of the procedural blank used in the calibration curves, based on 3 and 10
respectively. HCl and HNO3-washed refer to HF-HNO3-dissolved samples that have been prepared and analysed in apparatus washed by
HCl and HNO3 respectively. RSD values are calculated from repeated analyses of multiple preparations of JB-1a.
* RSD: Relative Standard Deviation, defined as: 1/mean * 100
†Element in wt%. All the other elements are in ppm.
- 31 -
APPENDIX 2. REPRESENTATIVES OF MINERAL COMPOSITION ANALYSES AND FORMULAE
782
783
784
TABLE A2.1. SEM-EDS ANALYTICAL RESULTS OF REPRESENTATIVE GARNET, OMPHACITE AND WHITE MICAS
Garnet
TS02-03A
TS02-15A
Omphacite
TS02-17b
TS02-32
TS0217b
TS02-32
White mica
TS02-41
Core
Rim
Core
Rim
Core
Rim
Core
Rim
SiO2
36.80
37.41
36.34
36.87
36.42
37.27
36.59
36.94
55.68
54.41
56.07
54.26
54.11
TiO2
0.00
0.00
0.21
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.18
Al2O3
20.67
21.12
20.58
21.11
20.41
21.19
20.51
20.71
9.04
7.99
10.20
8.40
FeO*
34.19
30.31
35.01
32.63
32.33
30.35
34.13
33.55
6.63
7.63
3.44
MnO
0.56
0.27
0.37
0.43
3.38
0.27
0.74
0.00
0.00
0.00
0.00
MgO
1.19
3.01
1.22
2.52
0.97
3.27
1.56
2.39
8.07
7.95
CaO
7.08
8.67
7.30
7.34
6.84
8.10
6.76
6.99
13.98
Na2O
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
K2O
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
TS0242
TS023A
TS02-10
TS0217b
TS02-30
TS02-35
TS0241
mus
mus
pg
pg
pg
mus
pg
mus
mus
55.22
51.26
50.13
45.72
45.87
46.17
53.89
46.65
50.72
52.38
0.00
0.42
0.33
0.25
0.00
0.00
0.34
0.00
0.51
0.32
9.04
6.19
24.59
27.72
38.73
38.15
39.30
22.85
39.05
28.42
22.38
12.14
12.28
8.37
2.18
3.48
0.29
0.43
0.28
1.56
0.76
2.73
3.62
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
9.00
5.26
4.63
9.29
4.24
2.38
0.00
0.20
0.00
4.96
0.15
3.06
4.10
14.26
14.79
11.76
10.46
16.17
0.00
0.00
0.00
0.15
0.34
0.00
0.13
0.00
0.00
7.27
6.81
6.67
7.81
8.81
5.65
0.35
0.42
7.37
7.07
6.48
0.36
6.53
0.66
0.00
0.00
0.00
0.00
0.00
0.00
0.00
10.78
10.11
0.60
0.49
0.62
9.05
0.88
9.54
10.48
wt%
Cr2O3
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Total
100.5
100.8
101.0
100.9
100.3
100.5
100.3
100.6
100.7
99.05
100.2
99.63
99.51
100.9
93.82
94.94
92.96
92.35
93.19
93.01
94.16
95.64
93.29
Si
5.94
5.93
5.87
5.89
5.92
5.92
5.93
5.93
4.00
4.00
3.99
4.03
4.02
4.02
6.96
6.76
5.98
6.04
6.01
7.25
6.03
6.72
7.17
Ti
0.00
0.00
0.03
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.01
0.00
0.04
0.03
0.02
0.00
0.00
0.03
0.00
0.05
0.03
Al
2.95
2.96
2.94
2.98
2.93
2.97
2.94
2.94
0.57
0.52
0.64
0.55
0.59
0.40
2.95
3.30
4.48
4.44
4.52
2.72
4.46
3.33
2.71
Fe2+
2.31
2.01
2.36
2.18
2.20
2.01
2.31
2.25
0.20
0.23
0.10
0.38
0.38
0.25
0.12
0.20
0.02
0.02
0.02
0.09
0.04
0.15
0.21
Mn
0.04
0.02
0.03
0.03
0.23
0.02
0.05
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Mg
0.14
0.36
0.15
0.30
0.12
0.39
0.19
0.29
0.43
0.44
0.48
0.29
0.26
0.50
0.43
0.24
0.00
0.02
0.00
0.50
0.01
0.30
0.42
Ca
0.61
0.74
0.63
0.63
0.60
0.69
0.59
0.60
0.54
0.56
0.56
0.47
0.42
0.63
0.00
0.00
0.00
0.01
0.02
0.00
0.01
0.00
0.00
Na
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.25
0.24
0.23
0.28
0.32
0.20
0.02
0.03
0.47
0.45
0.41
0.02
0.41
0.04
0.00
K
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.47
0.43
0.03
0.02
0.03
0.39
0.04
0.40
0.46
Cr
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Note: Oxygen by stoichiometry. The formulae of garnet, omphacite and white micas are calculated based on 12, 6 and 11 oxygens respetively. FeO*- total Fe as FeO.
785
- 32 -
786
TABLE A2.2. SEM-EDS RESULTS OF REPRESENTATIVE AMPHIBOLES AND EPIDOTE GROUP MINERALS
Amphibole
TS02-3A
Epidote group minerals
TS02-15A
TS02-17b
TS02-35
TS02-01
Gln
Gln
Gln
TS02-35
TS0241
TS0247
TS0248
TS02-50B
Gln
Brs
Gln
Brs
Brs
Brs
Act
Gln
MgHorn
SiO2
57.19
51.64
57.00
47.03
47.22
46.47
51.34
56.09
48.22
57.19
57.79
51.41
57.74
58.16
36.94
37.29
38.35
38.25
38.67
37.07
38.28
38.46
38.16
TiO2
0.00
0.20
0.00
0.23
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.26
0.31
0.00
0.00
Al2O3
11.03
8.21
10.18
10.30
8.97
10.54
4.17
10.72
6.26
10.71
10.98
6.26
10.80
11.25
23.39
22.83
26.25
25.64
27.33
25.65
27.75
29.08
27.26
27.56
FeO*
10.17
12.39
12.89
17.69
17.35
17.56
15.23
11.97
20.77
8.04
8.30
11.39
11.71
10.09
13.74
13.30
8.55
10.14
8.85
9.15
6.92
5.28
7.88
7.26
MnO
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.27
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
MgO
10.46
13.34
9.19
9.72
10.09
9.48
13.45
9.40
9.66
11.81
11.43
14.98
9.96
11.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
CaO
0.45
8.03
0.47
8.75
8.90
8.49
10.45
0.76
9.65
1.33
0.76
10.21
0.57
0.59
22.64
22.63
24.18
23.60
23.95
23.08
24.26
24.29
23.89
24.09
Na2O
7.62
3.96
8.00
3.89
3.50
3.79
1.90
7.34
2.96
7.09
7.56
2.55
7.73
7.77
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
K2O
0.00
0.20
0.00
0.28
0.34
0.33
0.16
0.00
0.29
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Cr2O3
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Total
96.93
97.97
97.72
97.89
96.37
96.65
96.69
96.28
98.07
96.18
96.82
96.80
98.52
98.86
96.71
96.05
97.33
97.63
98.80
94.95
97.47
97.42
97.19
97.13
Si
7.93
7.38
7.96
6.97
7.10
6.97
7.57
7.91
7.25
7.93
7.95
7.43
7.94
7.91
5.96
6.05
6.04
6.03
5.99
6.00
5.99
5.98
6.00
6.00
Ti
0.00
0.02
0.00
0.03
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.03
0.04
0.00
0.00
Al
1.80
1.38
1.68
1.80
1.59
1.86
0.72
1.78
1.11
1.75
1.78
1.07
1.75
1.80
3.33
3.27
3.66
3.57
3.74
3.67
3.84
4.00
3.79
3.83
Fe2+
1.18
1.48
1.51
2.19
2.18
2.20
1.88
1.41
2.61
0.93
0.96
1.38
1.35
1.15
1.25
1.22
0.76
0.90
0.77
0.84
0.61
0.46
0.70
0.64
Mn
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.03
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Mg
2.16
2.84
1.91
2.15
2.26
2.12
2.96
1.98
2.17
2.44
2.34
3.23
2.04
2.23
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Ca
0.07
1.23
0.07
1.39
1.43
1.36
1.65
0.11
1.56
0.20
0.11
1.58
0.08
0.09
1.96
1.97
2.04
1.99
1.99
2.00
2.03
2.02
2.01
2.03
Na
2.05
1.10
2.17
1.12
1.02
1.10
0.54
2.01
0.86
1.91
2.02
0.71
2.06
2.05
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
K
0.00
0.04
0.00
0.05
0.07
0.06
0.03
0.00
0.06
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Cr
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
Gln
MgHorn
TS0217b
wt%
Note: Oxygen by stoichiometry. The formulae of amphibole and epidote group minerals are calculated based on 23 and 12.5 oxygens respectively. FeO*- total Fe as FeO.
787
- 33 -
38.22
APPEDIX 3. COMPARISONS OF RECONSTRUCTED BULK
COMPOSITIONS WITH ANALYTICAL COMPOSITIONS
788
789
790
Reconstructions of bulk-rock major element compositions use the formula Zj
i n, j m
791
=
X Y
i 1, j 1
i
j
, where i refers to different mineral phases (i.e., garnet, glaucophane,
792
omphacite, rutile, titanite, white mica, carbonate, chlorite, quartz, (clino)zoisite etc.), j
793
refers to different oxides (i.e., SiO2, TiO2, Al2O3, FeO, MnO, CaO, MgO, K2O, Na2O),
794
X refers to mineral weight proportions, Y refers to element abundances in mineral i, Z
795
refers to bulk concentrations of element j. Before the calculation of Z values, mineral
796
modes, which are in volume percent, must be converted to weight proportions by
797
multiplying their relevant mineral densities. Calculation details are given in note to
798
Table 1. Because all the Fe in mineral composition analyses are expressed as FeO, the
799
total Fe as Fe2O3 in the bulk rock are then converted into FeO by multiplying 0.89981
800
before all the other calculations. Comparisons of reconstructed bulk-rock composition
801
with actual (analytical) data are given in Fig.A1..
- 34 -
802
For N = 56 samples, the correlation coefficients R > 0.45 statistically significant
803
at > 99.9 % confidence levels. The significant correlations suggest that both mineral
804
compositional data (APPENDIX 2) and modal data (Table 1) are of good quality.
805
Deviations from unity of the slope for TiO2 (1.01), Al2O3 (0.93), MgO (1.03) and
806
TFeO (1.04) ensure that the reconstructed compositions are within 10% uncertainties,
807
although SiO2 (1.15), MnO (1.12) and CaO (1.20) may be slightly overestimated.
808
Several reasons would result in the uncertainties: (1). Heterogeneous mineral
809
compositions; (2). Heterogeneous mineral distributions; (3). Dark minerals which
810
have been omitted during estimations of mineral modes; (4). Tiny minerals which
811
have not been discerned under optical microscope. However, the absence of
812
systematic uncertainties for major elements together with the significant correlations
- 35 -
813
between reconstructed bulk-rock compositions and analytical compositions above the
814
confidence levels indicate the estimated mineral modes are reliable at least for major
815
elements, and this result is very encouraging for metamorphic rocks of multiple
816
phases with varying mineralogies and grain sizes. It proves the reliability of the
817
estimated mineral modes, which could be useful for understanding trace element
818
distribution/budget between phases in our subsequent studies.
819
ACKNOWLEDGEMENTS
820
The sampling was conducted in 2002 supported by a NERC Senior Research
821
Fellowship to Yaoling Niu, a NERC studentship to Shaun Lavis, and a Royal Society
822
China joint project, Chinese NSF and Chinese Geological Survey. The analytical
823
work was done by Shaun Lavis as a part of his PhD research work supported by
824
HEFCE/NERC to Julian A. Pearce and Yaoling Niu and other NERC grants to Julian
825
A. Pearce. Yaoling Niu also thanks Leverhulme Trust for a Research Fellowship.
826
Yuanyuan Xiao is supported by a Durham University Doctoral Fellowship.
827
Comments from the associate editor (William Peck) and reviewers (Timm John, Lifei
828
Zhang and an anonymous reviewer) are significant to improve this paper.
829
REFERENCES CITED
830
[1] Ai, Y. L., Zhang, L.F., Li, X.P., and Qu, J.F., 2006, Geochemical characteristics
831
and tectonic implications of HP-UHP eclogites and blueschists in Southwestern
832
Tianshan, China: Progress in Natural Science, v. 16, no. 6, p. 624 - 632.
833
[2] Ando, A., Mita, N., and Terashima, S., 1987, 1986 values for fifteen GSJ rock
834
reference samples, "igneous rock series": Geostandards Newsletter, v. 11, p. 159-
835
166, doi: 10.1111/j.1751-908X.1987.tb00023.x.
836
837
[3] Bebout, G. E., Bebout, A.E., and Graham, C.M., 2007, Cycling of B, Li, and
LILE (K, Cs, Rb, Ba, Sr) into subduction zones: SIMS evidence from micas in
- 36 -
838
high-P/T metasedimentary rocks: Chemical Geology, v. 239, p. 284-304, doi:
839
10.1016/j.chemgeo.2006.10.016.
840
[4] Becker, H., Jochum, K.P., and Carlson, R.W., 2000, Trace element fractionation
841
during dehydration of eclogites from high-pressure terranes and the implications
842
for element fluxes in subduction zones: Chemical Geology, v. 163, p. 65-99, doi:
843
10.1016/S0009-2541(99)00104-7.
844
[5] Beinlich, A., Klemd, R., John, T., Gao, J., 2010, Trace-element mobilization
845
during Ca-metasomatism along a major fluid conduit: Eclogitization of blueschist
846
as a consequence of fluid–rock interaction: Gechimica et Cosmochimica Acta, v.
847
74, p. 1892-1922, doi:10.1016/j.gca.2009.12.011.
848
[6] Brandon, A.D., and Draper, D.S., 1996, Constraints on the origin of the oxidation
849
state of mantle overlying subduction zones - An example from Simcoe,
850
Washington, USA: Geochimica et Cosmochimica Acta, v. 60, p. 1739-1749, doi:
851
10.1016/0016-7037(96)00056-7.
852
[7] Cann, J. R., 1970, Rb, Sr, Y, Zr and Nb in some ocean floor basaltic rocks: Earth
853
and Planetary Science Letters, v. 10, p. 7-11, doi: 10.1016/0012-821X(70)90058-
854
0.
855
[8] Donnelly, K.E., Goldstein, S.L., Langmuir, C.H., and Spiegelman, M., 2004,
856
Origin of enriched ocean ridge basalts and implications for mantle dynamics:
857
Earth and Planetary Science Letters, v. 226, p. 347-366, doi:
858
10.1016/j.epsl.2004.07.019.
859
[9] El Korh, A., Schmidt, S.T., Ulianov, A., and Potel, S, 2009, Trace element
860
partitioning in HP-LT metamorphic assemblages during subduction-related
861
metamorphism, Ile de Groix, France: a detailed LA-ICPMS Study: Journal of
862
Petrology, v. 50, p. 1107-1148, doi: doi:10.1093/petrology/egp034.
- 37 -
863
[10] Feineman, M.D., Ryerson, F.J., DePaolo, D.J., and Plank, T., 2007, Zoisite-
864
aqueous fluid trace element partitioning with implications for subduction zone
865
fluid composition: Chemical Geology, v. 239, p. 250-265, doi:
866
10.1016/j.chemgeo.2007.01.008.
867
[11] Gao, J., Klemd, R., Zhang, L., Wang, Z., and Xiao, X., 1999, P – T path of high –
868
pressure / low – temperature rocks and tectonic implications in the western
869
Tianshan Mountains, NW China: Jounal of Metamorphic Geology, v. 17, p. 621-
870
636, doi: 10.1046/j.1525-1314.1999.00219.x.
871
[12] Gao, J., and Klemd, R., 2000, Eclogite occurrences in the southern Tianshan high
872
pressure belt, Xinjiang, Western China: Gondwana Research, v. 3, p. 33-38, doi:
873
10.1016/S1342-937X(05)70055-1.
874
[13] Gao, J., and Klemd, R., 2003, Formation of HP-LT rocks and their tectonic
875
implications in the western Tianshan Orogen, NW China: geochemical and age
876
constraints: Lithos, v. 66, p. 1-22, doi: 10.1016/S0024-4937(02)00153-6.
877
[14] Gao, J., Long, L.L., Qian, Q., Huang, D.Z., Su, W., and Klemd, R., 2006, South
878
Tianshan: a Late Paleozoic or a Triassic orogen? : Acta Petrologica Sinica (in
879
Chinese), v. 22, no. 5, p. 1049-1061.
880
[15] Gao, J., John, T., Klemd, R., and Xiong, X.M., 2007, Mobilization of Ti-Nb-Ta
881
during subduction: Evidence from rutile-bearing dehydration segregations and
882
veins hosted in eclogite, Tianshan, NW China: Geochimica et Cosmochimica
883
Acta, v. 71, p. 4974-4996, doi: 10.1016/j.gca.2007.07.027.
884
[16] Gao, J., Long, L.L., Klemd, R., Qian, Q., Liu, D.Y., Xiong, X.M., Su, W., Liu,
885
W., Wang, Y.T., Yang, F.Q., 2009, Tectonic evolution of the South Tianshan
886
orogen and adjacent regions, NW China: geochemical and age constraints of
- 38 -
887
granitoid rocks: Int. J. Earth Sci., v. 98, p. 1221-1238, doi:10.1007/s00531-008-
888
0370-8.
889
890
[17] Gill, J., 1981, Orogenic Andesites and Plate Tectonics: Springer-Verlag, Berlin,
390 p.
891
[18] Hermmann, J., Spandler, C., Hack, A., and Korsakov, A.V., 2006, Aqueous fluids
892
and hydrous melts in high-pressure and ultra-high pressure rocks: Implications for
893
element transfer in subduction zones: Lithos, v. 92, p. 399-417, doi:
894
10.1016/j.lithos.2006.03.055 .
895
[19] Hofmann, A.W., and Hart, S.R., 1978, An assessment of local and regional
896
isotopic equilibrium in the mantle: Earth and Planetary Science Letters, v. 38, p.
897
44-62, doi: 10.1016/0012-821X(78)90125-5.
898
[20] Hofmann, A. W., Jochum, K.P., Seufert, M., White, W.M., 1986, Nb and Pb in
899
oceanic basalts: new constraints on mantle evolution: Earth and Planetary Science
900
Letters, v. 79, p. 33-45.
901
902
903
[21] Hofmann, A.W., 1997, Mantle geochemistry: the message from oceanic
volcanism: Nature, v. 385, p. 219-229, doi: 10.1038/385219a0.
[22] John, T., Scherer, E.E., Haase, K., Schenk, V., 2004, Trace element fractionation
904
during fluid-induced eclogitization in a subducting slab: trace element and Lu-Hf-
905
Sm-Nd isotope systematic: Earth and Planetary Science Letters, v. 227, p. 441-
906
456, doi: 10.1016/j.epsl.2004.09.009.
907
[23] John, T., Klemd, R., Gao, J., and Garbe-Schönberg, C.D., 2008, Trace-element
908
mobilization in slabs due to non steady-state fluid-rock interaction: Constraints
909
from an eclogite-facies transport vein in blueschist (Tianshan, China): Lithos, v.
910
103, p. 1-24, doi: 10.1016/j.lithos.2007.09.005.
- 39 -
911
[24] Kelley, K. A., Plank, T., Ludden, J., and Staudigel, H., 2003, Composition of
912
altered oceanic crust at ODP sites 801 and 1149: Geochemistry Geophysics
913
Geosystems, v. 4, doi: 10.1029/2002GC000435.
914
[25] Kessel, R., Schmidt, M.W., Ulmer, P., and Pettke, T., 2005, Trace element
915
signature of subduction - zone fluids, melts and supercritical liquids at 120-180
916
km depth: Nature, v. 437, p. 724-727, doi: 10.1038/nature03971.
917
[26] Klemd, R., 2003, Ultrahigh-pressure metamorphism in eclogites from the western
918
Tianshan high-pressure belt (Xinjiang, western China)-Comment: American
919
Mineralogist, v. 88, p. 1153-1156,
920
[27] Langmuir, D., 1978, Uranium solution-mineral equilibria at low temperatures
921
with applications to sedimentary ore deposits: Gechimica et Cosmochimica Acta,
922
v. 42, p. 547-569.
923
924
[28] Lavis, S., 2005, Recycling in subduction zones: Evidence from eclogites and
blueschist of NW China [Ph.D. thesis]: Cardiff University, 245 p.
925
[29] Leake, B.E., Woolley, A.R., Arps, C.E.S., Birch, W.D., Gilbert, M.C., Grice, J.D.,
926
Hawthorne, F.C., Kato, A., Kisch, H.J., Krivovichev, V.G., Linthout, K., Laird, J.,
927
Mandarino, J.A., Maresch, W.V., Nickel, E.H., Rock, N.M.S., Schumacher, J.C.,
928
Smith, D.C., Stephenson, N.C.N., Ungaretti, L., Whittaker, E.J.W., and Youshi,
929
G., 1997, Nomenclature of amphiboles: Report of the subcommittee on
930
amphiboles of the International Mineralogical Association, Commission on new
931
minerals and mineral names: The Canadian Mineralogist, v. 35, p. 219-246.
932
[30] Lee, C.-T. A., Brandon, A.D., and Norman, M., 2003, Vanadium in peridotites as
933
a proxy for paleo-ƒO2 during partial melting: Prospects, limitations, and
934
implications: Geochimica et Cosmochimica Acta, v. 67, p. 3045-3064, doi:
935
10.1016/S0016-7037(03)00268-0.
- 40 -
936
[31] Lee, C.-T. A., Leeman, W.P., Canil, D., and Li, Z.-X.A., 2005, Similar V/Sc
937
Systematics in MORB and Arc Basalts: Implications for the Oxygen Fugacities of
938
their Mantle Source Regions: Journal of Petrology, v. 46, p. 2313-2336, doi:
939
10.1093/petrology/egi056.
940
[32] Liu, J., Bohlen, S.R., and Ernst, W.G., 1996, Stability of hydrous phases in
941
subducting oceanic crust: Earth and Planetary Science Letters, v.143, p.161-171,
942
doi: 10.1016/0012-821X(96)00130-6.
943
[33] Lü, Z., Zhang, L.F., Du, J.X., and Bucher, K., 2008, Coesite inlusions in garnet
944
from eclogitic rocks in western Tianshan, northwest China: Convincing proof of
945
UHP metamorphism: American Mineralogist, v. 93, p. 1845-1850, doi:
946
10.2138/am.2008.2800.
947
[34] Lü, Z., Zhang, L.F., Du, J.X., Bucher, K., 2009, Petrology of coesite-bearing
948
eclogite from Habutengsu Valley, western Tianshan, NW China and its
949
tectonometamorphic implication: Journal of Metamorphic Geology, v. 27, p. 773-
950
787.
951
952
953
[35] Manning, C.E., 2004, The chemistry of subduction-zone fluids: Earth and
Planetary Science Letters, v. 223, p. 1-16, doi: 10.1016/j.epsl.2004.04.030.
[36] McCulloch, M.T., and Gamble, J.A., 1991, Geochemical and geodynamical
954
constraints on subduction zone magmatism: Earth and Planetary Science Letters,
955
v. 102, p. 358-384, doi: 10.1016/0012-821X(91)90029-H.
956
957
[37] Morimoto, N., 1988, Nomenclature of pyroxenes: Mineralogy and Petrology, v.
39, p. 55-76, doi: 10.1007/BF01226262
958
[38] Niu, Y.L., and Batiza, R., 1997, Trace element evidence from seamounts for
959
recycled oceanic crust in the eastern equatorial Pacific mantle: Earth and
- 41 -
960
Planetary Science Letters, v. 148, p. 471-484, doi: 10.1016/S0012-
961
821X(97)00048-4.
962
[39] Niu, Y.L., and O'Hara, M.J., 2003, Origin of ocean island basalts: A new
963
perspective from petrology, geochemistry, and mineral physics considerations:
964
Journal of Geophysical Research, v. 108, p. ECV5.1-ECV5.19, doi:
965
10.1029/2002JB002048.
966
[40] Niu, Y.L., 2004, Bulk-rock major and trace element compositions of abyssal
967
peridotites: Implications for mantle melting, melt extraction and post-melting
968
processes beneath mid-ocean ridges: Journal of Petrology, v. 45, p. 2423-2458,
969
doi: 10.1093/petrology/egh068.
970
[41] Niu, Y.L., 2009, Some basic concepts and problems on the petrogenesis of intra-
971
plate ocean island basalts: Chinese Science Bulletin, v. 54, p. 4148-4160, doi:
972
10.1007/s11434-009-0668-3.
973
[42] Niu, Y. L. and O'Hara, M.J., 2009, MORB mantle hosts the missing Eu (Sr, Nb,
974
Ta and Ti) in the continental crust: New perspectives on crustal growth, crust–
975
mantle differentiation and chemical structure of oceanic upper mantle: Lithos, v.
976
112, p. 1-17,
977
[43] O' Nions, R.K., Evensen, N.M., and Hamilton, P.J., 1979, Geochemical
978
modelling of mantle differentiation and crustal growth: Journal of Geophysical
979
Research, v. 84, p. 6091-6101, doi: 10.1029/JB084iB11p06091.
980
[44] Page, F. Z., Essene, E.J., Mukasa, S.B., 2005, Quartz exsolution in clinopyroxene
981
is not proof of ultrahigh pressures: Evidence from eclogites from the Eastern Blue
982
Ridge, Southern Appalachians, U.S.A.: American Mineralogist, v. 90, p. 1092-
983
1099, doi:10.2138/am.2005.1761.
- 42 -
984
[45] Pawley, A.R., and Holloway, J.R., 1993, Water Sources for Subduction Zone
985
Volcanism: New Experimental Constraints: Science, v. 260, p. 664-667, doi:
986
10.1126/science.260.5108.664.
987
[46] Peacock, S.M., 1993, The importance of blueschist-eclogite dehydration reactions
988
in subducting oceanic crust: Geological Society of America Bulletin, v. 105, p.
989
684-694, doi: 10.1130/0016-7606(1993)105<0684:TIOBED>2.3.CO;2.
990
[47] Pearce, J. A., and Cann, J.R., 1973, Tectonic setting of basic volcanic rocks
991
determined using trace element analyzes: Earth and Planetary Science Letters, v.
992
19, p. 290-300, doi: 10.1016/0012-821X(73)90129-5.
993
[48] Pearce, J.A., 1996, A user’s guide to basalt discrimination diagrams, in Trace
994
element geochemistry of volcanic rocks: implications for massive sulphide
995
exploration: Geol. Assoc. Canada Spec. Publ. v. 12, p. 79-113.
996
[49] Pearce, J. A., 2008, Geochemical fingerprinting of oceanic basalts with
997
applications to ophiolite classification and the search for Archean oceanic crust:
998
Lithos, v. 100, p. 14-48, doi: 10.1016/j.lithos.2007.06.016.
999
[50] Pilchin, A., 2005, The role of serpentinization in exhumation of high- to ultra-
1000
high-pressure metamorphic rocks: Earth and Planetary Science Letters, v. 237, p.
1001
815-828, doi: 10.1016/j.epsl.2005.06.053.
1002
[51] Pilet, S., Baker, M.B., Stolper, E.M., 2008, Metasomatized lithosphere and the
1003
origin of alkaline lavas: Science, v. 320, p. 916-919, doi:
1004
10.1126/science.1156563.
1005
[52] Plank, T., and Langmuir, C.H., 1998, The chemical composition of subducting
1006
sediment and its consequences for the crust and mantle: Chemical Geology, v.
1007
145, p. 325-394, doi: 10.1016/S0009-2541(97)00150-2.
- 43 -
1008
[53] Poli, S., Schmidt, M.W., 1997, The high-pressure stability of hydrous phases in
1009
orogenic belts: an experimental approach on eclogite-forming processes:
1010
Tectonophysics, v. 273, p. 169-184.
1011
[54] Poli, S., and Schmidt, M.W., 1998, The high-pressure stability of zoisite and
1012
phase relationships of zoisite-bearing assemblages: Contributions to Mineralogy
1013
and Petrology, v. 130, p. 162-175, doi: 10.1007/s004100050357.
1014
[55] Poli, S., and Schmidt, M.W., 2002, Petrology of subducted slabs: Annual Review
1015
Earth Planet Science, v. 30, p. 207-235, doi:
1016
10.1146/annurev.earth.30.091201.140550.
1017
[56] Rapp, R. P., Irifune, T., Shimizu, N., Nishiyama, N., Norman, M.D., and Inoue,
1018
T., 2008, Subduction recycling of continental sediments and the origin of
1019
geochemically enriched reservoirs in the deep mantle: Earth and Planetary
1020
Science Letters, v. 271, p. 14-23, doi: doi:10.1016/j.epsl.2008.02.028.
1021
[57] Rebay, G., Powell, R., Diener, J.F.A., 2010, Calculated phase equilibria for a
1022
MORB composition in a P-T range, 450-650°C and 18–28 kbar: the stability of
1023
eclogite: Journal of Metamorphic Geology, v. 28, p. 635-645, doi:10.1111/j.1525-
1024
1314.2010.00882.x.
1025
[58] Reinecke, T., 1998, Prograde high- to ultrahigh-pressure metamorphism and
1026
exhumation of oceanic sediments at Lago di Cignana, Zermatt-Saas Zone,
1027
western Alps: Lithos, v. 42, p. 147-189, doi:10.1016/S0024-4937(97)00041-8.
1028
[59] Schmidt, M.W., 1995, Lawsonite: Upper pressure stability and formation of
1029
higher density hydrous phases: American Mineralogist, v. 80, p. 1286-1292.
1030
1031
[60] Schmidt, M.W., and Poli, S., 1998, Experimentally based water budgets for
dehydrating slabs and consequences for arc magma generation: Earth and
- 44 -
1032
Planetary Science Letters, v. 163, p. 361-379, doi: 10.1016/S0012-
1033
821X(98)00142-3.
1034
[61] Schmidt, M.W., and Poli, S., 2003, Generation of mobile components during
1035
subduction of oceanic crust, in Rudnick, R.L., ed., The crust: Treatise on
1036
Geochemistry, v. 3, p. 567-591.
1037
[62] Smit, M. A., Brӧcker, M., Scherer, E.E., 2008, Aragonite and magnesite in
1038
eclogites from the Jæren nappe, SW Norway: disequilibrium in the system
1039
CaCO3–MgCO3 and petrological implications: Journal of Metamorphic Geology,
1040
v. 26, p. 959-979, doi:10.1111/j.1525-1314.2008.00795.x.
1041
[63] Song, S.G., Su, L., Niu, Y.L., Lai, Y., and Zhang, L.F., 2009, CH4 inclusions in
1042
orogenic harzburgite: Evidence for reduced slab fluids and implication for redox
1043
melting in mantle wedge: Geochimica et Cosmochimica Acta, v. 73, p. 1737-
1044
1754, doi: 10.1016/j.gca.2008.12.008.
1045
[64] Sorensen, S. S., Grossman, J.N., and Perfit, M.R., 1997, Phengite-hosted LILE
1046
enrichment in eclogite and related rocks: Implications for fluid-mediated mass
1047
transfer in subduction zones and arc magma genesis: Journal of Petrology, v. 38,
1048
p. 3-34.
1049
[65] Spandler, C., Hermann, J., Arculus, R., and Mavrogenes, J., 2003, Redistribution
1050
of trace elements during prograde metamorphism from lawsonite blueschist to
1051
eclogite facies: implications for deep subduction-zone processes: Contributions to
1052
Mineralogy and Petrology, v. 146, p. 205-222, doi: 10.1007/s00410-003-0495-5.
1053
[66] Spandler, C., Hermann, J., Arculus, R., and Mavrogenes, J., 2004, Geochemical
1054
heterogeneity and element mobility in deeply subducted oceanic crust; insights
1055
from high-pressure mafic rocks from New Caledonia: Chemical Geology, v. 206,
1056
p. 21-42, doi: 10.1016/j.chemgeo.2004.01.006.
- 45 -
1057
[67] Su, W., Gao, J., Klemd, R., Li, J.L., Zhang, X., Li, X.H., Chen, N.S., and Zhang,
1058
L., 2010, U–Pb zircon geochronology of Tianshan eclogites in NW China:
1059
implication for the collision between the Yili and Tarim blocks of the
1060
southwestern Altaids: European Journal of Mineralogy, v. 22, p. 473-478.
1061
[68] Sun, S.S., and McDonough, W.F., 1989, Chemical and isotopic systematic in
1062
ocean basalt: Implication for mantle composition and processes, in Saunders,
1063
A.D., and Norry, M.J., eds., Magmatism in the Ocean Basins: Geological Society,
1064
London, Special Publication, v. 42, p. 313-345, doi:
1065
10.1144/GSL.SP.1989.042.01.19.
1066
[69] Tatsumi, Y., and Kogiso, T., 2003, The subduction factory: its role in the
1067
evolution of the Earth's crust and mantle: Geological Society, London, Special
1068
Publications, v. 219, p. 55-80.
1069
[70] Van der Straaten, F., Schenk, V., John, T., and Gao, J., 2008, Blueschist-facies
1070
rehydration of eclogites (Tian Shan, NW-China): Implications for fluid-rock
1071
interaction in the subduction channel: Chemical Geology, v. 255, p. 195-219, doi:
1072
10.1016/j.chemgeo.2008.06.037.
1073
[71] Volkova, N. I., and Budanov, V.I., 1999, Geochemical discrimination of
1074
metabasalt rocks of the Fan-Karategin transitional blueschist/greenschist belt,
1075
South Tianshan, Tajikistan: seamount volcanism and accretionary tectonics:
1076
Lithos, v. 47, p. 201-216, doi:10.1016/S0024-4937(99)00019-5.
1077
[72] Weaver, B.L., 1991, The origin of ocean island end-member compositions: trace
1078
element and isotopic constraints: Earth and Planetary Science Letters, v. 104, p.
1079
381-397, doi: 10.1016/0012-821X(91)90217-6.
1080
1081
[73] White, W.M., and Patchett, J., 1984, Hf-Nd-Sr isotopes and incompatible element
abundances in island arcs: implications for magma origins and crust-mantle
- 46 -
1082
evolution: Earth and Planetary Science Letters, v. 67, p. 167-185,
1083
doi:10.1016/0012-821X(84)90112-2.
1084
[74] Winchester, J.A., and Floyd, P.A., 1977, Geochemical discrimination of different
1085
magma series and their differentiation products using immobile elements:
1086
Chemical Geology, v. 20, p. 325-343, doi: 10.1016/0009-2541(77)90057-2.
1087
[75] Willbold, M., and Stracke, A., 2006, Trace element composition of mantle end-
1088
members: Implications for recycling of oceanic and upper and lower continental
1089
crust: Geochemistry Geophysics Geosystems, v. 7, doi: 10.1029/2005GC001005.
1090
[76] Wood, B.J., Bryndzia, L.T., and Johnson, K.E., 1990, Mantle oxidation state and
1091
its relationship to tectonic environment and fluid speciation: Science, v. 248, p.
1092
337-345, doi: 10.1126/science.248.4953.337.
1093
[77] Yu, Z. S., Robinson, P., and McGoldrick, P., 2001, An evaluation of methods for
1094
the chemical decomposition of geological materials for trace element
1095
determination using ICP-MS: Geostandards Newsletter, v. 25, p. 199-217, doi:
1096
10.1111/j.1751-908X.2001.tb00596.x.
1097
[78] Zack, T., Rivers, T., and Foley, S.F., 2001, Cs-Rb-Ba systematics in phengite and
1098
amphibole: an assessment of fluid mobility at 2.0 GPa in eclogites from
1099
Trescolmen, Central Alps: Contributions to Mineralogy and Petrology, v.140,
1100
p.651-669, doi: 10.1007/s004100000206.
1101
[79] Zack, T., and John, T., 2007, An evaluation of reactive fluid flow and trace
1102
element mobility in subducting slabs: Chemical Geology, v. 239, p. 199-216, doi:
1103
10.1016/j.chemgeo.2006.10.020 .
- 47 -
1104
[80] Zhang, L. F., Gao, J., Ekebair, S., and Wang, Z.X., 2001, Low temperature
1105
eclogite facies metamorphism in Western Tianshan, Xinjiang: Science in China, v.
1106
44, p. 85-96.
1107
[81] Zhang, L.F., Ellis, D.J., and Jiang, W.B., 2002a, Ultrahigh-pressure
1108
metamorphism in western Tianshan, China: Part I. Evidence from inclustions of
1109
coesite pseudomorphs in garnet and from quartz exsolution lamellae in omphacite
1110
in elogites: American Mineralogist, v. 87, p. 853-860.
1111
[82] Zhang, L.F., Ellis, D.J., Williams, S., and Jiang, W.B., 2002b, Ultra-high pressure
1112
metamorphism in western Tianshan, China: PartⅡ. Evidence from magnesite in
1113
eclogite: American Mineralogist, v. 87, p. 861-866.
1114
[83] Zhang, L.F., Ellis, D.J., Arculus, R.J., Jiang, W., and Wei, C., 2003, ‘Forbidden
1115
zone’ subduction of sediments to 150 km depth – the reaction of dolomite to
1116
magnesite + aragonite in the UHPM metapelites from western Tianshan, China:
1117
Journal of Metamorphic Geology, v. 21, p. 523-529, doi: 10.1046/j.1525-
1118
1314.2003.00460.x.
1119
[84] Zhang, L.F., Ai, Y.L., Li, X.P., Rubatto, D., Song, B., Williams, S., Song, S.G.,
1120
Ellis, D., and Liou, J.G., 2007, Triassic collision of western Tianshan orogenic
1121
belt, China: Evidence from SHRIMP U-Pb dating of zircon from HP/UHP
1122
eclogitic rocks: Lithos, v. 96, p. 266-280, doi: 10.1016/j.lithos.2006.09.012.
1123
[85] Zindler, A., and Hart, S., 1986, Chemical Geodynamics: Ann. Rev. Earth Planet
1124
Sci., v. 14, p. 493-571, doi: 10.1146/annurev.ea.14.050186.002425.
1125
FIGURE CAPTIONS
1126
Fig.1. Schematic diagram to show the likely changes of the subducted materials and
1127
the possible effects on elemental ratios of radioactive over radiogenic isotopes (P/D)
1128
by dehydration dominated subduction-zone metamorphism (SZM) (ⅰ), which is also
- 48 -
1129
widely accepted to cause the mantle wedge melting for arc magmatism ( ⅱ ).
1130
Conceptually, it is P/D ratios of the residual materials passing through SZM that
1131
determine mantle chemical and, with time, consequent isotopic heterogeneities
1132
recognized in oceanic basalts (after Niu, 2009).
1133
Fig.2. Simplified map of Chinese Western Tianshan mountains (after Gao et al., 1999
1134
and Zhang et al., 2003). a. A geological map for the location of Western Tianshan; b.
1135
Distributions of our sampling positions.
1136
Fig.3. Photomicrographs of representative samples from Western Tianshan. Panels of
1137
(b-e, h) were taken under plane polarized light (PPL), others taken under crossed
1138
polarized light (XPL). (a-f, h), samples of basaltic protolith from Group 1 (see Table 1
1139
for classifications). (g), one sample of sedimentary protolith. The mineral
1140
abbreviations used in this paper are summed up as following: Ab – albite; Act –
1141
actinolite; Ae – Aegirine; Alla – allanite; Amp – amphibole; Ang – antigorite; Arg –
1142
aragonite; Brs – barroisite; Ca – carbonate; Chl – chlorite; Cld – chloritoid; Czs –
1143
clinozoisite; Cpx – clinopyroxene; Di – diopside; Dol – dolomite; Epi – epidote; Gln
1144
– glaucophane; Grt – garnet; Horn – hornblende; Jd – jadeite; Law – lawsonite; Mag –
1145
magnesite; Mica – white mica (including both phengitic muscovite and paragonite);
1146
Omp – omphacite; Qtz – quartz; Pg – paragonite; Phen – phengitic muscovite; Prp –
1147
pyrope; Rt – rutile; Tn – titanite; Tr – tremolite and Zs –(clino)zoisite.
1148
(a). Eclogitic blueschist. Coexisting paragonite and clinozoisite preserved as
1149
tabular inclusions in garnet poikiloblast represent lawsonite pseudomorphs. (b).
1150
Eclogitic blueschist. Idioblastic garnet with omphacite and glaucophane inclusions
1151
indicates a prograde blueschist facies metamorphism. In the lower right corner of this
1152
photo, rutile has been replaced by surrounded titanite during retrograde
1153
metamorphism. (c). In samples dominated by glaucophane, the presence of white
1154
mica and omphacite inclusions indicates the glaucophane in this sample, TS02-03A,
- 49 -
1155
is produced by retrograde metamorphisms rather than the prograde ones. This
1156
retrograde sample will also be used for late discussions of rehydration effects on re-
1157
enrichment in late text, compared with eclogite TS02-03B. (d). Epidote blueschist.
1158
Garnet inclusion in the paragonite, probably reflects a re-enriched process caused by
1159
rehydration during retrograde metamorphisms. (e). Glaucophane has barroisite
1160
overgrowth at rim, one omphacite inclusion as well, caused by retrograde overprints.
1161
(f). Epidote-bearing blueschist with chlorite overprinting on garnet porphyroblast. (g).
1162
Chloritized mica-bearing quartzite, characterized by abundant quartz and foliation
1163
fabrics. (h). One veinlet, generated as a result of the initial dehydration during
1164
prograde metamorphism. The inset photo shows the rough contact relationship
1165
between the omphacitite core, the mica dominated veinlet and the blueschist facies
1166
assemblage in host rocks.
1167
Fig.4. Schematic phase diagram (after El Korn et al. (2009), Schmidt and Poli (1998),
1168
Poli and Schmidt (2002) and Zhang et al. (2003)). Abbreviations: GS – greenschist;
1169
BS – blueschist; EA – epidote amphibolite; AMP – amphibolite; EC – eclogite; GR –
1170
granulite; other abbreviations can be referred to the caption of Fig.3.
1171
The light grey dotted curve with solid circles is the P-T path for Western
1172
Tianshan reported by Zhang et al. (2003). The dark grey curve with stars is a
1173
schematic P-T path for our samples based on detailed petrography, and the peak
1174
metamorphic conditions are chosen following Lü et al. (2009). Each numerical star
1175
represents one crucial step (i.e., “Stage x” mentioned in the relevant text) evidenced
1176
by different mineral assemblages and characterized by index minerals.
1177
Fig.5. Compositional classifications of clinopyroxene and amphibole. (a). WEF
1178
(wollastonite + enstatite + ferrosilite) - Jd - Ae diagram following Morimoto (1988)
1179
for clinopyroxene compositions, all of which have omphacitic compositions. (b).
- 50 -
1180
(Na)M4 - Si (p.f.u) diagram for amphiboles (Leake et al., 1997), showing
1181
compositional distinctions between glaucophane and other amphiboles.
1182
Fig.6. (a). Zr/Ti-Nb/Y diagram for rocks of basaltic protolith (after Winchester and
1183
Floyd, 1977; Pearce, 1996). (b). Th/Yb vs. Nb/Yb for rocks of basaltic protolith (after
1184
Pearce, 2008). OIB, E-MORB and N-MORB are plotted as well for comparisons
1185
(black solid circles; the values of oceanic basalts are from Sun and McDonough,
1186
1989).
1187
Fig.7. Correlation coefficient diagrams of Nb-Zr with other trace elements for
1188
samples of Group 1 (a-b) and Group 3 (c-d) from Western Tianshan mountains (Niu,
1189
2004; Lavis, 2005). Elements along X-axis are ordered in terms of decreasing
1190
incompatibility from left to right. Two immobile elements (Nb and Zr) are chosen
1191
considering their different incompatibilities, therefore significant correlation
1192
coefficients with Nb or Zr can reflect immobility of elements with varying
1193
incompabilities. The dashed lines represent the minimum significant correlation
1194
coefficients (refer to critical values of Pearson’s correlation coefficient using one tail
1195
test) at 99% confidence levels for 23 (Group 1) and 6 (Group 3) samples respectively
1196
(i.e., degrees of freedom are 21 and 4 respectively). Although correlation coefficients
1197
of U, Hf and Ti in dotted areas are also a little lower than the dashed lines in Fig.7c
1198
(also in Fig.7d for U), combined with the illustrations of Group 1 in Figs.7a-b, it is
1199
more likely that these elements have not been mobilised.
1200
Fig.8. Co-variation diagrams. (a-d) reveal significant correlations of LILE with Nb in
1201
rocks of sedimentary protolith, which may indicate relative immobility of K, Cs, Ba
1202
and Rb. Although they are not explicitly linear, correlations are still considered
1203
significant at > 99 % confidence levels (the critical value of Pearson’s correlation
1204
coefficient using one tail test for 11 samples is 0.685), which may indicate relative
1205
immobility of K-Cs-Ba-Rb. (e) U/Nb vs. Nb/Zr for rocks of both sedimentary and
- 51 -
1206
basaltic protoliths (after John et al., 2004). The U/Nb ratio of N-MORB (= 0.02) is
1207
referred to Sun and McDonough (1989). See the detailed explanation in the relevant
1208
text. (f) Correlations between U and Th for rocks of basaltic (Group 1 and Group 3)
1209
and sedimentary protolith in this study, as well as samples from previous studies
1210
(Spandler et al., 2003, 2004; Gao et al., 2007; John et al., 2008; El Korh et al., 2009),
1211
without considering the effect of their different protoliths. The Th/U ratio range for
1212
the average value of each group and the Th/U ratio of AOC (Super Composite of Site
1213
801, representing tholeiitic section of ocean crust in Kelley et al., 2003) are also given
1214
for comparisons. All the correlation coefficients shown in the box beneath the figures
1215
are those for correlations of U and Th at the confidence levels > 99 %. The significant
1216
correlations between U and Th (or Nb, which is not shown) indicate that U is
1217
immobile during SZM at its +4 valence, which suggests a reduced condition in
1218
subduction zones.
1219
Fig.9. N-MORB (Sun and McDonough, 1989) normalized multi-element diagrams of
1220
different groups for samples of basaltic protolith (a-c) and sedimentary protolith (d).
1221
The order of elemental incompatibility determined by X/Y-X plots (Niu and Batiza,
1222
1997) is followed here. Average OIB (a; Sun and McDonough, 1989), AOC (b-c;
1223
Super Composite of Site 801, representing tholeiitic section of ocean crust in Kelley
1224
et al., 2003) and GLOSS (d; Plank and Langmuir, 1998) are plotted for comparison.
1225
Three sample representatives are also highlighted in Fig.9a and the vertical dashed
1226
arrows represent the effect of re-enrichment caused by rehydration during retrograde
1227
metamorphisms. Elements in light grey areas are generally mobile as discussed in
1228
previous text.
1229
Fig.10. Correlations of mineral modal abundances and LOI with Cs-Rb-Ba. (a-c).
1230
Significant correlations of white mica modes (wt%) with Cs–Rb–Ba for rocks of
1231
sedimentary protolith (the critical value of Pearson’s correlation coefficient using one
- 52 -
1232
tail test for 10 samples is 0.715). (e-g). Positive trends between white mica modes
1233
(wt%) and Cs–Rb–Ba for rocks of basaltic protolith, although the correlations are
1234
insignificant > 99% confidence levels. Correlations of Cs with LOI (wt%) (d) and
1235
glaucophane modes (wt%) (h).
1236
Fig.11. Schematic diagrams to show trace element distributed patterns of rocks with
1237
theoretical N-MORB composition during hydrothermal alterations (a), prograde
1238
metamorphisms (b-f) and retrograde metamorphisms (g-i) in subduction zones.
1239
(d1-e1) show the geochemical consequences without considerations of capability
1240
of hydrous minerals (e.g., phengite and epidote group minerals) to accommodate
1241
mobile elements. It also explains the situation under which the hydrous minerals have
1242
not been readily formed prior to the liberation of mobile elements. Those elements
1243
(e.g., Cs, Sr etc.) are then released in the first pattern. (d2-e2) display the prograde
1244
process with the presence of hydrous minerals before mobilisations of these elements,
1245
and thus could reside their preferential elements.
1246
(g-i) illustrate the geochemical processes during retrograde metamorphisms. Two
1247
distinctive trace element distributed patterns (i vs. h) are controlled by the
1248
composition of fluids, as well as compositions of the reacted rocks that fluids passed
1249
through. The detail is illustrated by (g). The path labelled ‘No’ or ‘Yes’ reflects
1250
retrograde metamorphisms without or with re-enrichments of fluid-soluble elements
1251
respectively during rehydration. The widely accepted dehydration of serpentines
1252
within mantle sections of the subducted slab always represent a ‘No’ path. However, a
1253
‘Yes’ path may be followed if the fluids derived from serpentine could scavenge
1254
related mobile elements from the passed reacted rocks. Alternatively, it is also likely
1255
for those fluids to be originated from dehydrations at deeper depths within the
1256
subducted rocks (with or without those mobile elements) to rehydrate the subducted
1257
rocks at shallow depths, and lead to a ‘Yes’ path or a ‘No’ path respectively.
- 53 -
1258
Fig.12. Schematic diagram to show multistage SZM (modified from Poli and Schmidt,
1259
2002).
1260
Fig.13. Correlations of P/D ratios (i.e., Rb/Sr, U/Pb, Sm/Nd and Lu/Hf) with
1261
immobile element ratios (Nb/Zr and Th/U), for samples which seem to show only
1262
prograde effects (or with only limited retrograde overprints), i.e., those labelled in
1263
bold in Table 1.
1264
Fig.14. Comparisons of P/D ratios with radiogenic isotope ratios (the two insets are
1265
adapted from Hofmann (1997) and White and Patchett (1984)). Selected samples are
1266
the same as those in Fig.13. The positive Sm/Nd-Lu/Hf correlation is consistent with
1267
these rocks being potential source materials for the Hf-Nd isotopic relationship in
1268
oceanic basalts, but the lack of Rb/Sr-Sm/Nd (and Lu/Hf) correlation is inconsistent
1269
with the rocks being the major source materials for the Sr-Nd and Sr-Hf isotopic
1270
systematics in oceanic basalts.
- 54 -