Phase Equilibrium Modeling of MT–UHP Eclogite: a Case ...Yangkou Bay in the Sulu Belt of eastern China. The Sulu Belt is a classic example of deep continen-tal subduction and the

Phase Equilibrium Modeling of MT–UHP

Eclogite: a Case Study of Coesite Eclogite at

Yangkou Bay, Sulu Belt, Eastern China

Bin Xia1,2*, Michael Brown2,3, Lu Wang1,3, Song-Jie Wang1,2,3,4 and

Philip Piccoli2

1School of Earth Sciences, State Key Laboratory of Geological Processes and Mineral Resources, China

University of Geosciences, Wuhan 430074, China; 2Department of Geology, Laboratory for Crustal Petrology,

University of Maryland, College Park, MD 20742, USA; 3Center for Global Tectonics, China University of

Geosciences, Wuhan 430074, China; 4College of Earth Science and Engineering, Shandong University of Science

and Technology, Qingdao 266590, China

*Corresponding author. Present address: College of Earth Sciences, State Key Laboratory of Geological

Processes and Mineral Resources, China University of Geosciences, Wuhan 430074, China. E-mail:

[email protected]

Received December 3, 2016; Accepted May 28, 2018

ABSTRACT

In this study, we present an example of phase equilibrium modeling of medium-temperature–ultra-

high-pressure (MT–UHP) eclogites that includes consideration of the influence of ferric iron (O) and

H2O on the phase equilibria. As a case study, we focus on the intergranular coesite-bearing eclogites atYangkou in the Sulu Belt. Based on phase equilibrium modeling of four eclogites, we monitor changes

in phase relations during deep subduction and exhumation, and investigate fluid behavior during de-

compression. To determine the appropriate O and H2O contents to use in calculating P–T pseudosec-

tions for these eclogites, we use an iterative process in which calculated temperature/pressure (T/P)–O/

H2O phase diagrams are combined with constraints from petrological observations. P–T pseudosec-

tions were calculated for each of the four eclogites to constrain the P–T conditions. The highest P–T con-

ditions retrieved were P> 5�5 GPa at T>850�C, although variation in mineral compositions suggeststhat the maximum P–T conditions could have been higher. A P–T path was reconstructed based on

microstructural evidence, mineral compositions that constrain P–T conditions within phase assemblage

fields, average P calculations and mineral thermobarometry. During exhumation, the retrograde P–T

path passed through metamorphic conditions of P¼ 4�0–3�4 GPa at T¼ 850–800�C and P¼2�4–1�7 GPa

at T¼ 800–750�C, before reaching crustal levels at P¼ 1�3–0�9 GPa at T¼ 730–710�C. The prograde evo-

lution is suggested to have followed a high dT/dP path during the early stage of subduction, followedby a low dT/dP segment to the metamorphic peak. During exhumation, the eclogites at Yangkou be-

came domainal, made up of host-rock with low a(H2O) in which garnet and omphacite have partially re-

equilibrated and intergranular coesite has been preserved, cut by veins and veinlets where a(H2O) was

higher and new mineral assemblages have developed. In the veins, the new assemblage comprises

coarse phengite and quartz with symplectites of K-feldspar þ plagioclase þ biotite þ quartz around the

phengite. By contrast, the veinlets comprise symplectites of hornblendeþ plagioclase 6 quartz 6 clino-

pyroxene after omphacite; similar symplectites occur at the edges of the phengite–quartz veins againsthost eclogite. We interpret the coarse phengite and quartz, which previously could have been coesite,

to have formed by precipitation of solutes from fluid migrating under UHP conditions, whereas we in-

terpret the symplectites around the phengite to have formed by local melting and crystallization during

exhumation from HP eclogite- to HP amphibolite-facies conditions. The symplectites in the veinlets and

along the edges of the phengite–quartz veins are interpreted to have formed by reaction of local grain-

boundary fluid with the host under HP amphibolite-facies conditions.

VC The Author(s) 2018. Published by Oxford University Press. All rights reserved. For permissions, please e-mail: [email protected] 1253

J O U R N A L O F

P E T R O L O G Y

Journal of Petrology, 2018, Vol. 59, No. 7, 1253–1280

doi: 10.1093/petrology/egy060

Advance Access Publication Date: 11 June 2018

Original Article

Dow

nloaded from https://academ

ic.oup.com/petrology/article-abstract/59/7/1253/5035807 by U

niversity of Maryland C

ollege Park user on 04 October 2018

https://academic.oup.com/

Key words: intergranular coesite; MT–UHP eclogite; phase equilibrium modeling; P–T path;Yangkou, Sulu Belt

INTRODUCTION

Ultrahigh-pressure (UHP) metamorphic rocks, particu-

larly eclogite and associated country rock gneisses in

orogenic belts, demonstrate that continental crust can

be subducted to and returned from mantle depths

(Chopin, 2003; Liou et al., 2004; Brown & Johnson,

2018). Pressure–temperature–time (P–T–t) paths tracing

the deep subduction and exhumation of these UHP

rocks provide insight into geological processes during

continental collision at convergent plate boundaries

and form the basis for geodynamic modeling of these

processes (Gerya & Stockhert, 2006; Warren et al.,

2008; Roda et al., 2012; Sizova et al., 2012). Thus, robust

quantification of P–T–t paths from natural samples of

UHP metamorphic rocks is important if our geodynamic

modeling is to provide deeper understanding of proc-

esses during continental collision at convergent plate

boundaries.

Although the P–T conditions of UHP metamorphism

can be qualitatively constrained by the presence of indi-

cative UHP minerals such as coesite and diamond, the

quantitative estimation of these conditions is under-

taken using conventional thermobarometry (e.g. Krogh

Ravna & Terry, 2004) and/or phase equilibrium model-

ing (e.g. Wei & Clarke, 2011; Wei et al., 2013). Forward

modeling involves the calculation of phase equilibria

for a given rock composition using an internally consist-

ent thermodynamic dataset and appropriate activity–

composition models for the phases of interest (Holland

& Powell, 1998; Powell et al., 1998), which may then be

related to the observed mineral assemblages, mineral

proportions and mineral compositions for that particu-

lar sample. In addition, we may calculate phase equili-

bria for a representative composition (e.g. mid-ocean

ridge basalt; MORB) to investigate how variables such

as H2O content and oxidation state affect these equili-

bria (Rebay et al., 2010).

During the last decade, phase equilibrium modeling

has become the preferred thermobarometric method in

many studies because it utilizes the maximum informa-

tion available from the sample being studied, and, in

many cases, allows the evolution of mineral assemb-

lages to be quantified to determine a robust P–T path

(Powell & Holland, 2008). This method has proven use-

ful in the study of UHP eclogites, in part because some

minerals, such as garnet and phengite, may retain pro-

grade or peak stage compositional information that has

allowed quantification of these P–T conditions as well

as those recorded during exhumation (e.g. Wei et al.,

2009, 2013; Massonne, 2011, 2012; Li et al., 2016b).

One particular challenge in modeling UHP eclogites

is the large P–T stability field of high-variance mineral

assemblages at peak conditions [e.g. Grt þ Omp þ Coe

6 Ph 6 Ky; mineral abbreviations follow Whitney &

Evans (2010)] with little change in the proportions or

compositions of the rock-forming minerals, particularly

garnet and omphacite (e.g. �Stıpska & Powell, 2005; Wei

et al., 2013; Groppo et al., 2015). Another challenge for

medium-temperature eclogites (MT eclogites) with peak

T of 550–900�C (Carswell, 1990) is the common re-

equilibration of rock-forming minerals during exhum-

ation after residence at elevated temperatures

(>700�C), which may limit our ability to determine pro-

grade, peak and retrograde P–T conditions using con-

ventional thermobarometry (Caddick et al., 2010).During the past 20 years, phase equilibrium model-

ing has extended to complex compositional systems

that are geologically realistic. To optimize the results,

the amount of ferric iron, which is commonly not ana-

lyzed or is uncertain, and an appropriate H2O content,

which is commonly less than the loss on ignition (LOI),

to be used in the bulk-rock composition must be deter-

mined; both of these may significantly influence the

phase equilibria of eclogites (e.g. Proyer et al., 2004;

Clarke et al., 2006; Rebay et al., 2010). Because these

issues have not been considered fully in recent exam-

ples of phase equilibrium modeling of MT–UHP eclo-

gites (e.g. Massonne, 2011, 2012; Wei et al., 2013;

Groppo et al., 2015), we have made a particular effort to

address them in this study of MT–UHP eclogites from

Yangkou Bay in the Sulu Belt of eastern China.

The Sulu Belt is a classic example of deep continen-

tal subduction and the UHP eclogite at Yangkou Bay is

the sole example where intergranular coesite is pre-

served in crustal eclogites (e.g. Liou & Zhang, 1996; Ye

et al., 1996; Wang et al., 2018). Despite much effort, the

peak P–T conditions and the P–T path for these eclogites

remain uncertain. For instance, using conventional ther-

mobarometry [e.g. Grt–Cpx thermometer after Holland

(1983); Grt–Cpx thermometer after Krogh Ravna (2000);

Grt–Cpx–Ph thermometer after Krogh Ravna & Terry

(2004)], P–T conditions for the peak stage were esti-

mated to be 2�9–3�4 GPa at 560–760�C (Liu et al., 2009a)

or 2�8–4�5 GPa at 800–890�C (Liou & Zhang, 1996; Zhang

& Liou, 1997; Wang et al., 2014). By contrast, reintegra-

tion of exsolved minerals in host garnet in lenses of

eclogite that occur within peridotite has suggested a

much higher pressure of >7 GPa at the peak stage (Ye

et al., 2000). Therefore, one aim of our study is to assess

whether the peak pressure that is retrieved from phase

equilibrium modeling is higher than that retrieved by

conventional thermobarometry and potentially closer to

the 7 GPa derived from reintegration of majoritic garnet.

In addition, microstructures attributed to low-volume

partial melting have been observed in the eclogite and

associated country rock gneisses from the Sulu Belt.

Partial melting in eclogites has commonly been inter-

preted as the result of breakdown of hydrous minerals

1254 Journal of Petrology, 2018, Vol. 59, No. 7

Dow





such as white micas and/or zoisite (e.g. Zeng et al.,

2009; Zheng et al., 2011; Li et al., 2014; Wang et al.,

2014, 2016). However, primary phengite in eclogites in

these localities commonly exhibits equilibrium micro-

structures and has high Si contents that indicate forma-

tion at UHP conditions—features that suggest stability

during exhumation (e.g. Chen et al., 2014; Wang et al.,

2014). Alternatively, partial melting might be due to the

exsolution of molecular H2O and structural hydroxyl

from nominally anhydrous minerals (e.g. garnet and

omphacite) during decompression (e.g. Chen et al.,

2013, 2014; Wang et al., 2017). Therefore, another aim

of our study is to assess the extent to which melt might

have been present during the later stages of exhum-

ation of the eclogites at the Yangkou locality.

In this contribution, we use phase equilibrium model-

ing combined with average P calculations and mineral

thermobarometry to quantify the P–T conditions for four

eclogites, reconstruct a common P–T path for this locality

from these data, and discuss melting and melt crystal-

lization. The sample set comprises one bimineralic, one

kyanite-bearing and two phengite-bearing eclogites,

which are the types of eclogite that occur most common-

ly in subduction settings elsewhere, potentially making

our study of wider applicability. The P–T conditions of

such rocks have important geodynamic implications, as

discussed by Hacker (2006) and Brown & Johnson

(2018). The results of our study also provide further

understanding of melting during exhumation.

GEOLOGICAL BACKGROUND

Located in eastern China, the Sulu Belt marks the colli-

sion zone between the Yangtze and North China

Cratons. The belt is generally subdivided into a UHP

sector in the center and north, and an HP sector in the

south (Fig. 1a). Lithologies in the UHP sector comprise

dominantly orthogneiss and paragneiss, with minor

eclogite, garnet peridotite, quartzite and marble (e.g.

Zhang et al., 2009; Zheng, 2009), whereas rocks in the

HP sector comprise mainly kyanite- and topaz-bearing

quartzite, marble, paragneiss and granite gneiss, with

rare blueschist (Zhang et al., 1995).

Eclogite occurs mainly as blocks and lenses within

the gneisses, garnet peridotite and marble, and has

variable mineral assemblages including Grt þ Omp 6

Ph 6 Ky 6 Ep/Zo 6 Coe/Qz. Coesite and/or pseudo-

morphs after coesite are extensively recognized as

inclusions in rock-forming minerals in eclogite and in

zircons from various lithologies, including gneisses,

kyanite-bearing quartzite and marble (e.g. Wang et al.,

1993; Liou & Zhang, 1996; Nakamura & Hirajima, 2000;

Ye et al., 2000; Zhang et al., 2005; Liu et al., 2006b; Liu &

Liou, 2011; Wang et al., 2014). Diamonds have been

suggested to be present as inclusions in garnet (Xu

et al., 2003, 2005), as single grains in heavy mineral con-

centrates derived from eclogite (e.g. Zhang et al., 2007).

Calculated pressures and temperatures vary widely,

with estimates of pressure for peak metamorphism

from 2�8 to 4�5 GPa at temperatures ranging from

�600�C to >850�C (Liou & Zhang, 1996; Zhang & Liou,

1997; Zhang et al., 2006; Zhu et al., 2007; Liu et al.,

2009a; Liu & Liou, 2011; Wang et al., 2014). Higher peak

pressures of >7 GPa and >5�5 GPa have also been sug-

gested for unusual eclogite lenses in peridotite at

Yangkou Bay (Ye et al., 2000) and for marble at

Shanqingge (Liu et al., 2006b), respectively. Eclogites

record variable retrogression during exhumation, such

as overprinting by granulite-facies assemblages (e.g.

Wang et al., 1993; Banno et al., 2000; Nakamura &

Hirajima, 2000) or (high-pressure) amphibolite-facies

assemblages (e.g. Zhang et al., 2005; Zhu et al., 2007;

Liu et al., 2009a; Nakamura & Hirajima, 2010).

Supercritical fluid or hydrous melt may have been pre-

sent in eclogites and the associated country rocks, ei-

ther at the peak stage (e.g. Ferrando et al., 2005; Zhang

et al., 2008) or generated during decompression owing

to exhumation (e.g. Zheng et al., 2011; Chen et al., 2012,

2014; Li et al., 2014; Wang et al., 2014, 2016, 2017).Zircon geochronology, summarized by Liu et al.

(2006a) and Liu & Liou (2011), reveals the timing of the

prograde, peak and retrograde stages of the meta-

morphic evolution as follows. The prograde segment of

the P–T path into the coesite stability field occurred be-

tween c. 245 and 235 Ma, the peak segment of the P–T

path in the coesite stability field lasted from c. 235 to c.

225 Ma, and the retrograde stage through the quartz

eclogite-facies to the amphibolite-facies lasted from c.

225 to c. 210 Ma.

The eclogites in this study were sampled from

Yangkou Bay in the central Sulu Belt (Fig. 1b). The litholo-

gies include metagabbro, coesite-bearing eclogite, ser-

pentinized peridotite and quartzo-feldspathic gneiss (Liou

& Zhang, 1996; Ye et al., 2000). Eclogite is enclosed in the

gneiss and garnet peridotite. All of the UHP units are cut

by lamprophyre and quartz porphyry dikes.

PETROLOGY

Four eclogites were chosen for detailed petrographic ana-

lysis and phase equilibrium modeling. The samples com-

prise two phengite-bearing eclogites (YK128-12 and YKS-

5), one kyanite-bearing eclogite (WYKK-3) and one bimi-

neralic eclogite (YK5-2). Except for YKS-5, the samples are

characterized by intergranular coesite and inclusions of

coesite in both rock-forming and accessory minerals.

PetrographyPhengite-bearing eclogitesPhengite-bearing eclogites have the mineral assem-

blage Grt þ Omp þ Ph þ Qz 6 Coe 6 Ky, with <5 vol. %

phengite in YK128-12, but >5 vol. % phengite in YKS-5.

Both rocks have a granular microstructure, but

YK128-12 is finer grained (grain size 0�3–0�5 mm) than

YKS-5 (grain size 0�4–0�7 mm).

In YK128-12, narrow veinlets (width <0�05 mm) filled

with a symplectite of Hbl þ Pl 6 Qz 6 Cpx replace

Journal of Petrology, 2018, Vol. 59, No. 7 1255

Dow





omphacite along grain boundaries in the primary eclog-

ite (Fig. 2a and b). The primary eclogite is also cut by

wider veins (width >0�5 mm) filled with dominantly

coarse-grained (0�3–0�7 mm) Ph þ Qz (Fig. 2a and c) and

subordinate fine-grained symplectite (normally

<0�01 mm, locally 0�03–0�1 mm) of Kfs þ Pl þ Bt þ Qz

around the coarse phengites (Fig. 2c and d). In places,

cuspate K-feldspar has small dihedral angles against

coarse quartz and phengite (Fig. 2d). Symplectites of Hbl

þ Pl 6 Qz 6 Cpx also develop along the margins of the

veins (Fig. 2a and e) and clinopyroxene separates the

symplectite from quartz along the outer edges of the

veins (Fig. 2c and e). Sometimes near veinlets, primary

omphacite or omphacite inclusions in garnet show a re-

action relationship suggesting replacement by horn-

blende, plagioclase and/or quartz (Fig. 2f), indicating that

the reaction Grt þ Omp! Hbl þ Plþ Qz may occur.

In this rock type, coesite is either included in garnet

and omphacite or occurs as intergranular grains with a

number density of >14 grains in a circle of diameter

4 mm (Fig. 2g and h). Radiating quartz fibers and/or pal-

isade quartz around polycrystalline aggregates indicate

the former presence of intergranular coesite. Granular

garnet has inclusions of phengite, omphacite, rutile and

coesite/quartz, sometimes with zircon and rarely kyanite

(Fig. 2f and i). Omphacite occurs in several microstruc-

tural settings. Primary omphacite occurs as inclusions

in garnet and as one of the main rock-forming minerals

(Cpx1, Fig. 2b, e and f), whereas secondary omphacite

occurs in veinlets around garnet or primary omphacite

(Cpx2, Fig. 2b and e), in symplectites (Cpx3) with horn-

blende, plagioclase and quartz in veinlets (Fig. 2b) or

veins (Fig. 2e), and separating symplectites from coarse

quartz at the outer edge of veins (Cpx4, Fig. 2e). Coarse

rock-forming phengite is mainly euhedral and un-

altered, whereas finer-grained phengite is subhedral to

anhedral (Fig. 2h and i).

In YKS-5 garnet and omphacite (Cpx1) grains are elon-

gated parallel to the foliation defined by phengite

(Fig. 3a), and garnet contains inclusions of phengite,

omphacite (Cpx1), quartz and rutile (Fig. 3a and b).

Coesite or quartz pseudomorphs after coesite are

absent. Although not as well developed, similar to

YK128-12, veinlets with a symplectite of Hbl þ Pl 6 Qz

occur locally along omphacite grain boundaries, but the

larger veins present in YK128-12 are absent from YKS-5.

Kyanite-bearing eclogiteSample WYKK-3 is medium-grained (0�3–1�0 mm) with

a mineral assemblage of Grt (35–40 vol. %) þ Omp (40–

45%) þ Ph (10–15%) þ Ky (3–5%) þ Coe/Qz (3–5%), and

accessary rutile and ilmenite. Anhedral to subhedral

garnet and omphacite (Cpx1) are commonly elongated

and fractured. In plane-polarized light, some larger gar-

nets show an inclusion-rich area surrounded by a clear

rim (Fig. 3c). Inclusions of omphacite (Cpx1), kyanite,

coesite/quartz and phengite, together with rutile/ilmen-

ite and zircon, are common in garnet (Fig. 3c and d).

Kyanite also replaces garnet, containing fine-grained

garnet relics (<0�01 mm; Fig. 3e). Intergranular coesite

and coesite inclusions in garnet and omphacite are par-

tially or completely replaced by quartz aggregates.

Hornblende and plagioclase occur in the veins that cut

the primary eclogite (Fig. 3f).

Fig. 1. (a) Simplified geological map of the Sulu Belt in eastern China. (b) Geological map of the Yangkou locality (based on Wanget al., 2016). TB, Tarim Block; NCB, North China Block; SCB, South China Block.


Dow





Bimineralic eclogiteThe bimineralic eclogite (YK5-2) has an equigranular

granoblastic microstructure (grain size 0�2–0�8 mm, but

mostly 0�3–0�5 mm) with a mineral assemblage of Grt

(45–50 vol. %) þ Omp (40–45%) þ Coe/Qz (5–10%), and

accessory rutile, ilmenite, apatite and zircon. Some larger

garnets show an inclusion rich-area inferred to represent

a core surrounded by a clear mantle (Fig. 3g). The inclu-

sions are mostly omphacite (Cpx1), rutile/ilmenite and

quartz/coesite, with subordinate apatite and zircon.

Fig. 2. Photomicrographs showing the mineralogy and microstructures of phengite-bearing eclogite YK128-12 under plane-polar-ized light (a, g), cross-polarized light (h, i) and in BSE images (b–f). (a) Veins and veinlets in eclogite. Light-colored minerals in veinsare mainly coarse phengite and quartz; dark-colored minerals along veinlets and at the edge of veins are mainly Hbl þ Pl 6 Qz 6Cpx3. (b) In veinlets, a symplectite of Hbl þ Pl 6 Qz 6 Cpx3 replaces primary omphacite. Cpx1 represents primary omphacite oromphacite inclusions in garnet. Cpx2 represents retrograde omphacite with higher j(o) content along garnet or primary omphacitegrains, whereas Cpx3 represents retrograde omphacite with lower j(o) content in the symplectites. (c, d) Symplectite of Kfs þ Pl þBt 6 Qz occurs around coarse phengite in veins. (e) At the edges of veins, symplectite of Hbl þ Pl 6 Qz 6 Cpx3 develops after Cpx1.Cpx4 represents clinopyroxene with low Jd content but high Ae content rimming coarse quartz in veins. (f) Adjacent to veinlets, asymplectite of Hbl þ Pl 6 Qz replaces primary omphacite and omphacite inclusions in garnet. Ky occurs as inclusions in garnet orreplacing garnet and omphacite. (g, h) Intergranular coesite between garnet and omphacite. Quartz with radiating fibers surroundscoesite. Coesite shows higher relief than quartz. (i) Garnet showing inclusions of omphacite, phengite and quartz. Coesite occurs asinclusions in primary garnet and omphacite.


Dow





Intergranular coesite (Fig. 3h), with a number density of

up to 13 grains in a circle of diameter 4 mm, and inclu-

sions of coesite in garnet and omphacite (Cpx1) are ubi-

quitous. Most coesite grains, regardless of location, have

been retrograded to palisade quartz or mosaic quartz at

the rims. Veinlets of symplectite comprising Cpx3þ Hbl þPl 6 Qz occur along Cpx1 or Cpx2 grain boundaries

(Fig. 3i). Adjacent to these veinlets, rare kyanite together

with quartz are present replacing garnet (Fig. 3i), whereas

intergranular coesite is absent.

Metamorphic evolutionBased on petrographic observations, several stages in

the metamorphic evolution of the eclogites may be

Fig. 3. Photomicrographs showing the mineralogy and representative microstructures of phengite eclogite YKS-5 (a, b), kyanite-bearing eclogite WYKK-3 (c–f) and bimineralic eclogite YK5-2 (g–i) under cross-polarized light (a, b, g), plane-polarized light (c, f, h),and in BSE images (d, e, i). (a, b) Inclusions of omphacite, phengite, rutile and quartz in garnet, and pristine rock-forming phengiteshowing no evidence of retrograde reaction. (c) Garnet showing color differences. The inner part is inclusion-rich, interpreted torepresent the garnet core, whereas the mantle and rim are devoid of inclusions. The inclusions are mostly omphacite, phengite andquartz, with subordinate rutile, ilmenite and apatite. Rock-forming phengite is subhedral in shape and pristine. (d) Kyanite as inclu-sions in garnet. (e) Kyanite with garnet inclusions. (f) Retrogression along infilled fractures with hornblende and plagioclase in theveins. (g) Garnet showing an inclusion-rich core surrounded by a clear rim. The inclusions are mostly omphacite, rutile/ilmeniteand quartz/coesite, with subordinate apatite and zircon. (h) Intergranular coesite (showing higher relief than quartz) between garnetand omphacite. (i) Veinlets with symplectite comprising Cpx þ Hbl þ Pl 6 Qz; there are rare kyanite relics.


Dow





inferred (Table 1). For both phengite-bearing eclogites,

YK128-12 and YKS-5, a prograde stage is evidenced by

inclusions of Cpx1, Ph, Ky, Coe and Rt in Grt. The peak

stage without Ky in the mineral assemblage (Grt, Cpx1,

Ph, Coe and Rt, Table 1) could be due to the consump-

tion of Ky during the prograde evolution. An early de-

compression stage is represented by the retrograde

mineral assemblage Cpx2, Qz, Rt and secondary Ky

replacing Cpx1, and is followed by a late decompres-

sion stage recorded in the veins and veinlets by sym-

plectites of Kfs þ Bt þ Pl þ Qz around coarse phengite,

symplectites of Hbl þ Pl 6 Qz 6 Cpx3 replacing Cpx1

and Cpx2, and the development of Cpx4 against Qz. The

Ky-bearing eclogite, WYKK-3, has a prograde mineral

assemblage similar to that in the Ph-bearing eclogite, as

indicated by inclusions of Cpx1, Ph, Ky, Coe and Rt in

Grt. The peak stage has a mineral assemblage of Grt þCpx1 þ Ph þ Coe þ Ky þ Rt. In the veins, the mineral as-

semblage of Hbl þ Pl þ Qz probably indicates a late hy-

dration stage. For the bimineralic eclogite, YK5-2, the

prograde evolution to the peak stage did not change the

mineral assemblage (Grt, Cpx1, Coe and Rt). The early

decompression stage is identified by Cpx2, Qz, Rt and

secondary Ky, and replacement of Cpx1, followed by

the late decompression stage recorded by symplectites

of Hbl þ Pl 6 Qz replacing Cpx1 and Cpx2.

Mineral compositionsAnalytical methodsMineral analyses were performed using a JEOL 8900

electron probe microanalyzer at the Advanced Imaging

and Microscopy Laboratory at the University of

Maryland. The following operating conditions were uti-

lized: 15 kV accelerating voltage, 10 nA (phengite) or

25 nA (remaining phases) cup current, and a 1–10 lm

beam diameter. A series of natural and synthetic stand-

ards was utilized: plagioclase [Lake County plagioclase

(Si, Al, Na, K), Kakanui hornblende (Fe, Mg, Mn), micro-

cline (K), Broken Hill rhodonite (Mn)]; phengite

[Muthuen Township muscovite (K, Al, Si), Tiburon

Peninsula albite (Na), Kakanui hornblende (Fe, Mg, Ti),

Geophysical Lab synthetic Ba-glass (Ba), Lake County

plagioclase (Ca), and Broken Hill Rhodonite (Mn)]; kyan-

ite [Minas Gerais kyanite (Si, Al), Kakanui hornblende

(Fe, Mg, Ti, Ca), Broken Hill rhodonite (Mn), and Tiburon

Peninsula albite (Na)]; garnet [USGS GTAL garnet (Si,

Al, Fe, Ca, Mg), Kakanui hornblende (Na, K, Ti),

Johnstown hypersthene (Cr), and Minas Gerais spes-

sartine (Mn)]; and hornblende, biotite and omphacite

[Kakanui hornblende (Si, Til, Al, Fe, Mg, Ca, K and Na),

Johnstown hypersthene (Cr), and Minas Gerais spes-

sartine (Mn)]. Raw X-ray intensities were corrected

using a ZAF algorithm.

GarnetIn the phengite-bearing eclogite YK128-12, element

mapping of one of the larger garnets, supported by a

traverse of point analyses, shows relict zoning in cal-

cium, but only weak or no zoning in other cations

(Fig. 4a and b). In the apparent core the grossular con-

tent is relatively low (29 mol %), increasing to 33 mol %

in the mantle before decreasing outwards to 30 mol %

at the edge; there is an increase to 32 mol % in some

parts of the outermost rim or adjacent to fractures

(Table 2); XFe [¼ Fe/(Fe þ Mg)] changes little from cen-

ter to edge (0�67–0�70). For the phengite-bearing eclog-

ite YKS-5, calcium is seen to be zoned (Fig. 4c and d);

the grossular content in the apparent core is 29 mol %,

which decreases to 25 mol % in the mantle. The highest

grossular content (33 mol %) develops in the outmost

rim or adjacent to fractures (Fig. 4c; Table 2). XFe

changes little from center to edge (0�59–0�61).

For the kyanite-bearing eclogite (WYKK-3), element

mapping of a large garnet with an inclusion-rich appar-

ent core and a clear mantle/rim, supported by a traverse

of point analyses, shows relict zoning in calcium but

less distinct zoning of magnesium and iron (Fig. 4e and

f). In one half of the garnet (from core to A in Fig. 4f), the

grossular content in the core is relatively high (31–

32 mol %), decreasing to 26 mol % in the inner rim and

then increasing to 34 mol % in the outer rim (Table 2),

resembling the zoning pattern in the garnet from one of

the phengite-bearing eclogites (YKS-5); XFe varies little

Table 1: Metamorphic evolution inferred for the eclogites at Yangkou Bay, Sulu Belt

Prograde stage Peak stage Early decompression stage Late decompression stage

Samples YK128-12 and YKS-5Grt and inclusions

(Cpx1, Ph, Ky, Coe, Rt)Grt, Cpx1, Ph, Coe, Rt Ky (replacing Cpx1),

Cpx2, (Ph), Qz, Rt(4a) Hbl, Pl, Qz, (Cpx3) (symplectite

around Cpx1 and Cpx2)(4b) Kfs, Bt, Pl, Qz (between Ph and Qz)(4c) Cpx4 around Qz

Prograde stage Peak stage Early decompression stage Late hydration stage

Sample WYKK-3Grt and inclusions

(Cpx1, Ph, Ky, Coe, Rt)Grt, Cpx1, Ph, Ky, Coe, Rt Ky (replacing Grt þ Cpx1),

Cpx2, (Ph), Qz, RtHbl, Pl, Qz (in the fractures)

Prograde stage Peak stage Early decompression stage Late decompression stage

Sample YK5-2Grt and inclusions

(Cpx1, Coe, Rt)Grt, Cpx1, Coe, Rt Ky (replacing Cpx1),

Cpx2, Qz, RtHbl, Pl, Qz, (Cpx3) (symplectite

around Cpx1 and Cpx2)


Dow





Fig. 4. Ca element maps and traverses of point analyses for selected garnets from the four studied eclogites. Ca content in garnetincreases from green to yellow to red in the images (see online version for color images).


Dow





(0�64–0�66) from core to rim. In the other half of the gar-

net (from core to B in Fig. 4f), the grossular content

increases from core (30–31 mol %) to rim (34 mol %),

with XFe of 0�65–0�68.

In the bimineralic eclogite (YK5-2), a larger garnet

shows complex compositional zoning from an inclusion-

rich apparent core across the associated mantle (Fig. 4g

and h). The traverse shows that in one half of the garnet,

the grossular content decreases from an inner core (31–

32 mol %) to an outer core (28–29 mol %), before increas-

ing to 33 mol % in the mantle and then decreasing to

29 mol % in the inner rim (Table 2). In the outer rim, the

grossular content increases to 32 mol % again. The other

half of the garnet shows similar grossular variations.

Detailed examination of the core shows that inclusions

are rare in the inner part with the higher grossular con-

tent, whereas the outer part with the lower grossular con-

tent has numerous omphacite inclusions. Although care

was taken during analysis to avoid inclusions, the low

grossular content could reflect contamination of the anal-

yses in this area by cryptic fine-grained inclusions of

omphacite. Overall, the grossular zoning is similar to that

in garnet from one of the phengite-bearing eclogites

(YK128-12). The almandine and pyrope contents are simi-

lar to those of the other samples, but there is no systemat-

ic zoning, and XFe varies only slightly (0�67–0�71).

In summary, garnet in all four eclogites retains relict

zoning in calcium content whereas other bivalent cati-

ons preserve only weak zoning or are unzoned. Two

contrasting calcium zoning profiles are present. The

first shows an increase, followed by a decrease and

then an increase in calcium from the core to the rim, as

recorded by the larger garnets in YK128-12 and YK5-2.

The second shows a decrease followed by an increase

in calcium from core to rim or against fractures, as

recorded by the larger garnets in YKS-5 and WYKK-3.

OmphaciteWithin each sample, the primary omphacite, whether as

inclusions in garnet or as a main rock-forming mineral,

has indistinguishable compositions (Table 3, Fig. 5a).

The j(o) [where j(o) ¼ Jd þ Ae] content of the rock-

forming omphacite generally shows no systematic var-

iations from core to rim.For the phengite-bearing eclogite YK128-12, ompha-

cite inclusions (cpx1) in garnet have j(o) of 0�62 and

rock-forming omphacite (Cpx1) has j(o) of 0�60–0�64.

Secondary omphacite (Cpx2) in veinlets around garnet

or primary omphacite has slightly lower j(o) of 0�52–

0�57, whereas sodic clinopyroxene in symplectites in

veinlets (Cpx3) or against coarse quartz in veins (Cpx4)

has a significantly lower j(o) of 0�26–0�28 and 0�24–0�27,

Table 2: Representative compositions of garnet from the eclogites at Yangkou, Sulu Belt

YK128-12 YKS-5 WYKK-3 YK5-2

Position: core core mantle rim core mantle mantle rim core core mantle rim core core mantle rim

SiO2 39�10 38�88 38�61 38�87 39�26 38�99 39�35 39�13 38�95 38�84 38�99 39�64 38�51 38�60 38�83 38�45TiO2 b.d. b.d. 0�04 0�03 b.d. b.d. 0�03 b.d. 0�07 0�09 b.d. b.d. 0�03 0�06 b.d. 0�05Al2O3 21�89 21�54 21�72 21�81 22�03 22�18 22�29 22�19 21�28 21�34 21�74 22�32 21�52 21�30 21�63 21�58Cr2O3 b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. 0�05 b.d. b.d. b.d. b.d. b.d. b.d. b.d.FeO 22�24 21�90 21�29 21�07 20�00 20�89 20�47 19�27 21�51 21�49 21�77 19�67 21�93 22�11 22�15 21�73MnO 0�46 0�41 0�45 0�41 0�40 0�50 0�46 0�44 0�51 0�50 0�59 0�46 0�40 0�43 0�44 0�43MgO 6�05 5�79 5�53 5�59 7�31 7�95 7�73 7�02 5�94 6�12 6�85 6�07 5�65 5�68 5�82 5�55CaO 10�52 10�96 11�90 11�67 10�63 9�15 9�96 11�35 11�55 11�34 9�38 12�34 11�45 11�32 10�69 11�59Na2O b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d.K2O b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d.Total 100�26 99�48 9�54 99�45 99�63 99�69 100�28 99�38 99�81 99�72 99�31 100�49 99�48 99�49 99�58 99�37Oxygens 12�00 12�00 12�00 12�00 12�00 12�00 12�00 12�00 12�00 12�00 12�00 12�00 12�00 12�00 12�00 12�00Si 3�00 3�01 2�99 3�01 3�00 2�98 2�99 2�99 3�00 3�00 3�01 3�01 2�98 2�99 3�01 2�98Ti b.d. b.d. 0�00 0�00 b.d. b.d. 0�00 0�00 0�00 0�01 0�00 0�00 0�00 0�00 0�00 0�00Al 1�98 1�97 1�98 1�99 1�99 2�00 2�00 2�00 1�93 1�94 1�98 2�00 1�97 1�95 1�97 1�97Cr b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. 0�00 b.d. b.d. b.d. b.d. b.d. b.d. b.d.Fe3þ 0�01 0�01 0�04 0�00 0�01 0�04 0�03 0�02 0�05 0�06 0�00 0�00 0�07 0�06 0�01 0�06Fe2þ 1�42 1�41 1�34 1�36 1�27 1�29 1�28 1�23 1�34 1�33 1�41 1�25 1�35 1�37 1�42 1�35Mn 0�03 0�03 0�03 0�03 0�03 0�03 0�03 0�03 0�03 0�03 0�04 0�03 0�03 0�03 0�03 0�03Mg 0�69 0�67 0�64 0�64 0�83 0�91 0�88 0�79 0�68 0�70 0�79 0�69 0�65 0�66 0�67 0�64Ca 0�87 0�91 0�99 0�97 0�87 0�75 0�81 0�95 0�95 0�94 0�78 1�01 0�95 0�94 0�89 0�96Na b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d.K b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d.Sum 8�00 8�00 8�00 8�00 8�00 8�00 8�00 8�00 8�00 8�00 8�00 7�99 8�00 8�00 8�00 8�00Alm 0�47 0�47 0�45 0�45 0�42 0�43 0�43 0�41 0�44 0�44 0�47 0�42 0�45 0�46 0�47 0�45Grs 0�29 0�30 0�33 0�32 0�29 0�25 0�27 0�32 0�32 0�31 0�26 0�34 0�32 0�31 0�29 0�32Pyr 0�23 0�22 0�21 0�21 0�28 0�30 0�29 0�26 0�23 0�23 0�26 0�23 0�22 0�22 0�22 0�21Sps 0�01 0�01 0�01 0�01 0�01 0�01 0�01 0�01 0�01 0�01 0�01 0�01 0�01 0�01 0�01 0�01Fe# 0�67 0�68 0�68 0�68 0�60 0�59 0�59 0�61 0�66 0�65 0�64 0�64 0�67 0�68 0�68 0�68

Fe# ¼ Fe2þ/(Fe2þ þ Mg); Alm, almandine; Grs, grossular; Spss, spessartine; Pyr, pyrope. The mineral formulae were calculatedusing the program AX (T. Holland’s AX software page: http://www.esc.cam.ac.uk/research/research-groups/research-projects/tim-hollands-software-pages/ax). b.d., below detection.


Dow





http://www.esc.cam.ac.uk/research/research-groups/research-projects/tim-hollands-software-pages/ax

http://www.esc.cam.ac.uk/research/research-groups/research-projects/tim-hollands-software-pages/ax

Fig. 5. Compositions of omphacite and hornblende in the eclogites. (a) Ternary classification diagram for sodic clinopyroxene.Black filled symbols represent omphacite inclusions in garnet (Cpx1) and open symbols represent primary rock-forming omphacite(Cpx1). (b, c) Compositions of hornblende (after Leake et al., 1997).

Table 3: Representative compositions of omphacite and hornblende from the eclogites at Yangkou, Sulu Belt

YK128-12 YKS-5 WYKK-3 YK5-2

Cpx1 Cpx1 Cpx2 Cpx2 Cpx3 Cpx4 Hbl Cpx1 Cpx1 Cpx1 Cpx1 Hbl Cpx1 Cpx1 Hbl Hbl

SiO2 56�93 56�91 56�05 54�83 52�99 52�97 44�11 56�45 56�24 56�54 56�63 49�95 56�55 56�14 46�67 43�47TiO2 0�05 0�03 0�04 0�06 0�13 0�04 0�18 0�06 b.d. 0�04 b.d. 0�29 0�04 0�04 0�13 0�55Al2O3 15�48 15�60 13�55 12�50 6�87 2�37 14�85 13�35 12�19 14�53 13�64 7�20 13�81 15�39 10�87 13�60Cr2O3 b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d.FeO 4�10 3�62 4�89 5�84 6�14 12�59 11�06 3�52 3�06 4�06 3�52 11�45 3�89 3�82 9�99 13�49MnO b.d. b.d. b.d. 0�03 b.d. 0�09 0�05 0�04 0�05 b.d. 0�03 0�20 b.d. b.d. 0�05 0�07MgO 5�94 5�58 6�91 7�29 11�82 10�06 12�46 7�69 8�24 6�44 7�24 15�40 7�04 5�83 15�45 12�49CaO 9�46 8�88 11�05 12�63 18�72 18�34 11�98 11�91 12�54 10�19 11�06 11�33 10�77 9�57 11�64 11�20Na2O 9�01 9�18 8�21 7�22 3�41 3�62 3�68 7�66 7�22 8�61 8�00 1�79 8�26 8�81 2�32 2�98K2O b.d. b.d. b.d. b.d. b.d. b.d. 0�06 b.d. b.d. b.d. b.d. 0�22 b.d. b.d. 0�08 0�04Total 100�97 99�79 100�70 100�41 100�07 100�09 98�37 100�67 99�55 100�42 100�11 97�61 100�37 99�61 97�12 97�86Oxygens 6�00 6�00 6�00 6�00 6�00 6�00 23�00 6�00 6�00 6�00 6�00 23�00 6�00 6�00 23�00 23�00Si 1�98 1�99 1�96 1�94 1�92 1�96 6�38 1�98 1�99 1�98 1�99 7�15 1�98 1�98 6�71 6�33Ti 0�00 0�00 0�00 0�00 0�00 0�00 0�02 0�00 b.d. 0�00 b.d. 0�03 0�00 0�00 0�02 0�06Al 0�63 0�65 0�56 0�52 0�29 0�10 2�53 0�55 0�51 0�60 0�57 1�22 0�57 0�64 1�84 2�34Cr b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d.Fe3þ 0�01 0�00 0�07 0�09 0�11 0�23 0�00 0�02 0�00 0�03 0�00 0�26 0�03 0�01 0�29 0�34Fe2þ 0�11 0�11 0�07 0�09 0�08 0�16 1�34 0�09 0�09 0�09 0�10 1�11 0�09 0�11 0�91 1�31Mn b.d. b.d. b.d. 0�00 b.d. 0�00 0�01 0�00 0�00 b.d. 0�00 0�02 b.d. b.d. 0�01 0�01Mg 0�31 0�29 0�36 0�39 0�64 0�56 2�69 0�40 0�44 0�34 0�38 3�29 0�37 0�31 3�31 2�71Ca 0�35 0�33 0�42 0�48 0�73 0�73 1�86 0�45 0�48 0�38 0�42 1�74 0�40 0�36 1�79 1�75Na 0�61 0�62 0�56 0�50 0�24 0�26 1�03 0�52 0�50 0�58 0�55 0�50 0�56 0�60 0�65 0�84K b.d. b.d. b.d. b.d. b.d. b.d. 0�01 b.d. b.d. b.d. b.d. 0�04 b.d. b.d. 0�02 0�01Sum 4�00 3�99 4�00 4�00 4�00 4�00 15�86 4�00 4�00 4�00 4�00 15�44 4�00 4�00 15�64 15�81Jd 0�62 0�64 0�54 0�49 0�23 0�07 0�54 0�50 0�59 0�56 0�56 0�62Ae 0�00 0�00 0�03 0�03 0�03 0�20 0�00 0�00 0�00 0�00 0�01 0�00WEF 0�38 0�36 0�42 0�48 0�75 0�73 0�46 0�50 0�40 0�44 0�43 0�38Jo¼JdþAe

0�62 0�64 0�57 0�52 0�26 0�27 0�54 0�50 0�59 0�56 0�57 0�62

Jd, jadeite; Ae, aegirine; WEF, wollastonite þ enstatite þ ferrosilite. Other details are as in the footnote to Table 2.


Dow





respectively. However, Cpx4 is rich in the Ae compo-

nent (0�16–0�20), whereas its Jd content is only 0�07–

0�08, whereas Cpx1–Cpx3 are rich in the Jd component

with a low Ae of <0�04 (Fig. 5a). For the other three eclo-

gites, omphacite as inclusions in garnet and rock-

forming omphacite (Cpx1) has j(o) of 0�51–0�56 and

0�50–0�54 (phengite-bearing eclogite YKS-5), 0�51–0�59

and 0�53–0�56 (kyanite-bearing eclogite WYKK-3) and

0�57–0�60 and 0�56–0�63 (bimineralic eclogite YK5-2),

respectively.

PhengiteThe compositional variation of phengite from different

microstructural settings is shown in Supplementary

Data Electronic Appendix Fig. S1 (supplementary data

are available for downloading at http://www.petrology.

oxfordjournals.org). For YK128-12, phengite included in

garnet has TiO2 contents from 0�48 to 0�58 wt % and Si

contents from 3�48 to 3�54 p.f.u. (per formula unit; 11 O

basis). Coarse phengite flakes have TiO2 contents of

0�34–0�62 wt % and Si contents of 3�49–3�54 p.f.u. Rarer

fine-grained phengite has TiO2 contents of 0�54–0�71 wt

% and Si contents of 3�43–3�44 p.f.u., slightly lower than

the other types. The large phengite flakes (>0�7 mm) in

the vein have TiO2 contents of 0�34–0�68 wt % and Si

contents of 3�44–3�53 p.f.u., similar to the coarse phen-

gite and the inclusions in garnet. For YKS-5 and WYKK-

3, phengite included in garnet has TiO2 contents of

0�61 and 0�42–0�49 wt %, and Si contents of �3�51 and

3�47–3�53 p.f.u., respectively, whereas phengite outside

garnet has TiO2 contents of 0�37–0�61 and 0�45–0�62 wt

%, and Si contents of 3�47–3�58 and 3�42–3�53 p.f.u,

respectively. BaO contents in all phengite are in the

range from 0�02 to 0�71 wt % (Table 4).

HornblendeHornblende in the symplectites has Al2O3 contents of

10�87–17�98 wt %. It is pargasite (Fig. 5c; Leake et al.,

1997) in YK128-12, with Mg# [Mg/(Mg þ Fe2þ)] of 0�55–

0�72 and (Na þ K)A from 0�85 to 0�94 p.f.u. (23 O basis),

whereas it ranges from edenite to pargasite (Fig. 5c) in

YK5-2, with Mg# of 0�66–0�78 and (Na þ K)A of 0�54–

0�80 p.f.u. Hornblende bordering the veins in the

kyanite-bearing eclogite (WYKK-3) has lower Al2O3 con-

tents (5�09–8�50 wt %) than the hornblende in YK128-12

and YK5-2. It is magnesiohornblende (Fig. 5b; Leake

et al., 1997), with Mg# of 0�72–0�79 and (Na þ K)A of

0�28–0�42 p.f.u.

Other mineralsThe sodic feldspar in the symplectites in the two

phengite-bearing eclogites (YK128-12 and YK5-2) is alb-

ite with Ab0�91–0�98, whereas the plagioclase in the veins

in the kyanite-bearing eclogite (WYKK-3) is oligoclase

with Ab0�81–0�90 (Table 4). Kyanite in all samples is nearly

pure Al2SiO5 (Table 4).

PHASE EQUILIBRIUM MODELING

The purpose of the phase equilibrium modeling is to

quantify the P–T evolution of the eclogites at Yangkou

Bay in the Sulu Belt.

Table 4: Representative composition of phengite, biotite, plagioclase and kyanite from the eclogites at Yangkou, Sulu Belt

YK128-12 YKS-5 WYKK-3

Mineral:. Phflake

Phtiny

Ph inGrt

Ph inveins

Ph inveins

Bt inveins

Bt inveins

Pl inveins

Phflake

Ph inGrt

Phflake

Ph inGrt

Pl Ky inGrt

Kyout

SiO2 53�02 51�98 53�72 51�83 53�11 39�99 39�01 66�36 54�06 52�02 53�32 53�46 64�58 36�87 37�31TiO2 0�34 0�54 0�50 0�68 0�34 2�35 2�86 b.d. 0�42 0�61 0�48 0�47 0�04 b.d. b.d.Al2O3 24�31 26�56 24�25 26�43 24�40 17�22 16�69 20�62 23�47 24�50 24�63 24�85 20�32 62�47 63�08FeO 1�73 2�06 1�84 2�07 1�72 11�92 12�65 0�42 1�52 1�43 1�62 1�57 0�81 0�48 0�54MnO b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d.MgO 4�47 3�64 4�64 3�55 4�50 15�24 13�91 b.d. 5�02 4�28 4�55 4�46 0�96 b.d. b.d.CaO b.d. 0�04 b.d. b.d. b.d. 0�06 b.d. 1�72 b.d. b.d. b.d. b.d. 2�86 b.d. b.d.Na2O 0�15 0�57 0�29 0�19 0�12 0�43 0�11 10�81 0�11 0�27 0�16 0�21 9�63 b.d. b.d.K2O 10�60 10�02 10�68 10�92 11�03 9�86 9�67 0�04 10�70 10�51 10�54 10�53 0�07 b.d. b.d.BaO 0�49 0�67 0�32 0�23 0�31 0�19 0�46 b.d. 0�14 0�67 0�34 0�56 b.d. b.d. b.d.Total 95�12 96�08 96�23 95�88 95�53 97�25 95�36 99�96 95�44 94�30 95�64 96�11 99�28 99�82 100�93O 11�00 11�00 11�00 11�00 11�00 11�00 11�00 8�00 11�00 11�00 11�00 11�00 8�00 5�00 5�00Si 3�54 3�44 3�54 3�44 3�53 2�87 2�87 2�92 3�58 3�51 3�53 3�53 2�87 1�00 1�00Ti 0�02 0�03 0�03 0�03 0�02 0�13 0�16 b.d. 0�02 0�03 0�02 0�02 0�00 b.d. b.d.Al 1�91 2�07 1�89 2�07 1�91 1�45 1�45 1�07 1�83 1�95 1�92 1�93 1�07 1�99 1�99Fe2þ 0�10 0�11 0�10 0�11 0�10 0�71 0�78 0�02 0�08 0�08 0�09 0�09 0�03 0�01 0�01Mn b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d. b.d.Mg 0�44 0�36 0�46 0�35 0�45 1�63 1�52 0�00 0�50 0�43 0�45 0�44 0�06 b.d. b.d.Ca b.d. 0�00 b.d. b.d. b.d. 0�00 b.d. 0�08 b.d. b.d. b.d. b.d. 0�14 b.d. b.d.Na 0�02 0�07 0�04 0�02 0�02 0�06 0�02 0�92 0�02 0�04 0�02 0�03 0�83 b.d. b.d.K 0�90 0�85 0�90 0�92 0�94 0�90 0�91 0�00 0�91 0�90 0�89 0�89 0�00 b.d. b.d.Ba 0�01 0�02 0�01 0�01 0�01 0�01 0�01 b.d. 0�00 0�02 0�01 0�01 b.d. b.d. b.d.Total 6�95 6�96 6�96 6�97 6�97 7�76 7�71 5�01 6�94 6�96 6�94 6�94 5�01 3�00 3�00

The details are the same as in the footnote to Table 2.


Dow





https://academic.oup.com/petrology/article-lookup/doi/10.1093/petrology/egy060#supplementary-data




http://www.petrology.oxfordjournals.org

http://www.petrology.oxfordjournals.org

MethodsPhase diagrams were calculated using the

THERMOCALC software (Powell & Holland, 1988; ver-

sion 3.40, released on 20 March 2014) and the associ-

ated internally consistent thermodynamic dataset ds62

(Holland & Powell, 2011; updated at 6 February 2012)

for compositions in the NC(K)FMASHTO [Na2O–CaO(–

K2O)–FeO–MgO–Al2O3–SiO2–H2O–TiO2–O] system.

There are no potassic minerals in the bimineralic eclog-

ite (YK5-2), so K2O was not included in the modeling; in

the natural sample, the trivial amount of K2O is

assumed to be incorporated into pyroxene and horn-

blende. A–x relationships used in the modeling are as

follows: clinopyroxene (Green et al., 2016); garnet,

orthopyroxene and phengitic muscovite (White et al.,

2014); paragonite (Coggon & Holland, 2002); K-feldspar

and plagioclase (Holland & Powell, 2003). Pure phases

include lawsonite, kyanite, quartz/coesite, rutile, talc

and aqueous fluid (H2O). The modeling to constrain the

late prograde, peak and early retrograde P–T evolution

was done without melt as a phase, because there is no

appropriate activity–composition (a–x) model for melt

in basic rocks at these ultrahigh pressures (see Green

et al., 2016).

The bulk-rock compositions of YK5-2 and YKS-5

were obtained by inductively coupled plasma optical

emission spectrometry. For the modeling, the CaO and

FeO contents were corrected for the P2O5 and TiO2 con-

tained in apatite and ilmenite, respectively. For samples

YK128-12 and WYKK-3, the bulk-rock compositions

were considered to be inappropriate for modeling

owing to the scale of heterogeneity in these two sam-

ples. Instead, an effective domainal composition for the

primary eclogite away from the veinlets and veins was

calculated by integrating mineral compositions and vol-

umes. Primary sample compositions for YKS-5 and

YK5-2 (in wt%) together with all modified compositions

used to calculate phase diagrams (in mol %) are given

in Table 5.

H2O-present versus H2O-absent metamorphismIn their phase equilibrium modeling of MT–UHP eclo-

gites from the south Dabie orogen, a continuation to the

west of the same continental collisional event that

formed the Sulu Belt, Wei et al. (2013) assumed H2O

was present in excess. Thus, we begin the phase equi-

librium modeling by calculating a P–T pseudosection

with H2O in excess for one of the phengite-bearing eclo-

gites (YK128-12). This exercise also provides a founda-

tion for making decisions with respect to the T/P–O and

T/P–H2O modeling that follows.

The P–T pseudosection in Fig. 6 was calculated for a

range of P¼ 1�5–5�5 GPa and T¼ 600–900�C in the

NCKFMASHTO system, using the modified composition

given in Table 5 (for Fig. 6) with excess H2O and an

assumed value of 1�14 mol % for O, which is the aver-

age for eclogites retrieved from the Chinese Continental

Scientific Drilling (CCSD) main hole in the southern

Table 5: Primary sample compositions (in wt %) together with modified bulk-rock compositions used for phase diagram calculation(in mol %) for eclogites at Yangkou, Sulu Belt

Sample Figures H2O SiO2 Al2O3 CaO MgO FeOT K2O Na2O TiO2 O

YKS-5 (wt %) * 50�09 17�01 10�48 6�58 10�33 0�74 3�39 1�38 *YK5-2 (wt %) * 49�97 17�33 10�13 5�46 11�38 0�12 4�05 1�55 *YK128-12 (mol %) 6, P–T excess 52�94 10�74 10�93 7�93 9�87 0�13 4�44 1�87 1�14

7, T/P–O excess 53�55 10�87 11�06 8�02 9�99 0�13 4�49 1�89 0�0051�95 10�54 10�73 7�78 9�69 0�13 4�35 1�83 3�00

8, T–H2O 0�00 53�42 10�84 11�03 8�00 9�96 0�13 4�48 1�89 0�251�00 52�88 10�73 10�92 7�92 9�86 0�13 4�43 1�87 0�25

9, P–T 0�26 53�28 10�81 11�00 7�98 9�94 0�13 4�46 1�88 0�2511, P–T 7�15 76�62 7�96 0 3�17 0�81 3�70 0 0�19 0�40S3a, T–O 0�26 53�41 10�84 11�03 8�00 9�96 0�13 4�48 1�89 0�00S3b, P–O 0�26 52�34 10�62 10�81 7�84 9�76 0�13 4�39 1�85 2�00S4, T–O 7�17 76�93 7�99 0�00 3�18 0�82 3�71 0�00 0�20 0�00

7�10 76�16 7�91 0�00 3�15 0�81 3�67 0�00 0�19 1�00YKS-5 (mol %) S1a, T–O excess 53�18 10�64 11�51 10�42 9�17 0�50 3�48 1�11 0�00

51�58 10�32 11�16 10�11 8�89 0�48 3�38 1�07 3�00S2a, T–H2O 0 52�96 10�60 11�46 10�38 9�13 0�50 3�47 1�10 0�40

6�00 48�28 9�66 10�45 9�46 8�32 0�45 3�16 1�00 0�4010a, P–T 0�96 52�45 10�50 11�35 10�28 9�04 0�49 3�44 1�09 0�40

WYKK-3 (mol %) S1b, T–O excess 53�36 12�48 10�71 9�08 8�75 0�75 3�79 1�07 0�0051�76 12�10 10�39 8�81 8�49 0�73 3�67 1�04 3�00

S2b, T–H2O 0�00 53�20 12�44 10�68 9�05 8�73 0�75 3�77 1�07 0�3010�00 47�87 11�19 9�61 8�14 7�85 0�68 3�40 0�96 0�30

10b, P–T 1�45 52�43 12�26 10�53 8�92 8�60 0�74 3�72 1�05 0�30YK5-2 (mol %) S1c, T–O excess 53�56 10�94 11�11 8�72 10�20 0�00 4�21 1�25 0�00

51�95 10�62 10�78 8�46 9�89 0�00 4�09 1�22 3�0010c, P–T 0�00 53�34 10�90 11�07 8�69 10�16 0�00 4�20 1�25 0�40

FeOT represents total iron. O (in mol %) is equal to Fe2O3 (in mol %). Figures S1a–c, S2a, b, S3a, b and S4 are Supplementary DataElectronic Appendix at Journal of Petrology online.* Represents undetected elements in the bulk rock composition.


Dow





Sulu Belt, �280 km south of Yangkou Bay (Zhang et al.,

2006). This pseudosection is dominated by a small

number of high-variance phase assemblage fields; inparticular, over a range of pressure, Grt þ Omp þ Law

þ Coe þ Ph þ H2O is stable at lower temperature and

Grt þ Omp þ Ky þ Coe þ Ph þ H2O is stable at higher

temperature. Wei et al. (2013) suggested using the high-

est Si content of phengite and the highest grossular

content in garnet to constrain close to peak P–T condi-

tions, although the amount of inevitable re-equilibrationof prograde zoning in garnet while reaching peak T is

unknown (Caddick et al., 2010).

For YK128-12, isopleths for the highest Si content of

phengite (3�54 p.f.u.) and the highest grossular content

in garnet (0�33) intersect in the phase assemblage field

Grt þ Omp þ Law þ Ph þ Coe þ H2O, yielding apparentP–T conditions of �4�5 GPa at �800�C. Calculated min-

eral mole proportions (on a one-oxide basis, approxi-

mating vol. %) at this P–T condition are Grt ¼ 38�4%,

Omp ¼ 42�3%, Law ¼ 9�1%, Ph ¼ 1�8%, Coe ¼ 6�5% and

Rt ¼ 1�9 %. This mineral assemblage includes �9 mol %

lawsonite, but lawsonite (or pseudomorphs after law-

sonite) is not present in the samples of this study andhas not been reported previously in eclogites from the

Yangkou locality or from the Sulu Belt. Thus, we con-

clude that the assumption of H2O present in excess is

inappropriate for these eclogites.

In support of this conclusion, we note also that at

Yangkou Bay, the presence of intergranular coesite inthe eclogites and the preservation of igneous mineral-

ogy and texture in the protolith gabbro and the

surrounding gneisses (Liou & Zhang, 1996; Wallis et al.,

1997; Zhang & Liou, 1997; Wang et al., 2018) indicate

that at the metamorphic peak these rocks were very

probably fluid-absent (Mosenfelder et al., 2005; Young

& Kylander-Clark, 2015). In addition, during eclogite-fa-

cies metamorphism both the stability of lawsonite and

the proportion that may be present are determined by

the amount of H2O in the system and the oxidation con-

ditions during metamorphism (Rebay et al., 2010).

Thus, before calculating the remainder of the P–T pseu-

dosections used in this study, we followed an iterative

process to estimate appropriate values for the ferric

iron (O) and H2O contents in the bulk compositions to

be modeled.

Estimation of Fe31 (O)To evaluate the influence of O on phase relations and

mineral stability, T–O and P–O pseudosections were cal-

culated for YK128-12 using the bulk-rock composition

listed in Table 5 (for Fig. 7). For these calculations, H2O

was assumed to be present in excess. We considered a

range of O contents from 0 to 3 mol %, where zero rep-

resents all Fe as Fe2þ and 3 represents 58�9 mol % Fe as

Fe3þ. The T–O diagram was calculated at P¼ 4�0 GPa

and the P–O diagram was calculated at T¼800�C. Using

slightly different values for P or T in modeling these T/

P–O diagrams has little influence on the results.The calculated T–O diagram is dominated by four-

and five-variant fields (Fig. 7a). Lawsonite-bearing

phase assemblages develop at T<790�C and Ky-

bearing phase assemblages develop at T> 720�C, at

O> 0�6 mol %. The mineral assemblage observed in the

sample (Grt þ Omp þ Coe þ Ph þ Rt) is present in a six-

variant phase assemblage field where O< 0�60 mol %

and T is above �725�C. The P–O pseudosection calcu-

lated at T¼ 800�C yields a similar result (Fig. 7b). Thus,

any arbitrary value for O that is <0�60 mol % in the bulk-

rock composition can be used for further modeling. As

a further constraint, we used the mineral modes and

compositions (with Fe3þ �0�03 p.f.u. for both garnet and

omphacite) to calculate an O content, which gives a

value of about 0�2 mol %. Given the low Fe3þ content of

the rock-forming minerals in this sample (Cpx with

Acm0–3 and Grt with Adr0–2), a low O content of 0�25 mol

% in the bulk-rock composition was used for further

modeling.

Following the same procedure, T–O pseudosections

were calculated for the other three samples

(Supplementary Data Electronic Appendix Fig. S2a–c).

For the phengite-bearing eclogite, YKS-5, the modeled

mineral assemblage of Grt þ Omp þ Ph þ Coe þ Rt

occurs in a field with an O content of <0�8 mol %

(Supplementary Data Electronic Appendix Fig. S2a). For

the bimineralic eclogite, YK5-2, the T–O pseudosection

was calculated in the K-absent system. Nonetheless,

the phase relations are similar to those for the phengite-

bearing eclogites (Supplementary Data Electronic

Appendix Fig. S2c). The mineral assemblage of Grt þ

Fig. 6. P–T pseudosection for phengite-bearing eclogite YK128-12. The bulk-rock composition is given in Table 5. ‘cg’ repre-sents a calculated isopleth for the highest value of grossularcontent in garnet and ‘Si’ represents a calculated isopleth forthe highest value of Si in phengite.


Dow











Omp þ Coe þ Rt is present in the field with an O content

of <0�50 mol %. A value of O¼ 0�40 mol % was chosen

for further modeling of both samples. For the kyanite-

bearing eclogite (WYKK-3), the modeled mineral as-

semblage of Grt þ Omp þ Ky þ Ph þ Coe þ Rt occurs

over the full range of O contents investigated at T above

�790�C (Supplementary Data Electronic Appendix Fig.

S2b). A low bulk-rock O content is suggested by the low

Fe3þ content of the rock-forming minerals in this sam-

ple (Cpx with Acm0–1 and Grt with Adr0–5). Thus, we use

a value of 0�30 mol % in further modeling of this sample,

which is similar to the other samples and consistent

with the mineral compositions.

Estimation of H2OTo estimate the appropriate H2O content for modeling,

we calculated T–H2O pseudosections at P¼ 4�0 GPa for

YK128-12, YKS-5 and WYKK-3. A T–H2O pseudosection

was not calculated for the bimineralic eclogite (YK5-2)

because all the rock-forming minerals (Grt þ Omp þCoe/Qz) are considered anhydrous in THERMOCALC.

For YK128-12, the range of H2O contents considered

is from 0�00 to 1�00 mol %. The calculated T–H2O dia-

gram is dominated by five-variant fields (Fig. 8). Free

H2O is present at T above �740�C and H2O above

0�26 mol %. Lawsonite-bearing phase assemblages de-

velop at T below �750�C and H2O above 0�26. The

observed mineral assemblage of Grt þ Omp þ Coe þPh þ Rt develops only in a narrow field with an H2O

content of �0�26 mol %. At H2O < 0�26 mol %, K-feldspar

develops because insufficient phengite is present to ac-

commodate all available K (phengite is the only hy-

drous mineral in the phase assemblage). Given the

likelihood that the peak mineral assemblage was fluid-

absent, the H2O content in this sample can be con-

strained only to be less than saturated; that is, less than

0�26 mol %. Although Kfs will be present in phase as-

semblage fields where H2O <0�26 mol %, the amount is

nominal (Kfs ¼ 0�27–0�00 mol % at H2O ¼ 0�20–0�26 mol

%) and its presence does not upset the topology of the

phase diagram; thus, Kfs may be neglected in the phaseassemblages (see Schorn & Diener, 2017). T–H2O pseu-

dosections calculated for YKS-5 and WYKK-3 give

Fig. 7. (a) T–O diagram calculated at P¼4�0 GPa for phengite-bearing eclogite YK128-12. (b) P–O diagram calculated at T¼800�Cfor YK128-12. The bulk-rock composition used is given in Table 5, with H2O in excess and O varying from 0 to 3 mol %. The fieldlabeled in bold type represents the observed peak assemblage of interest.

Fig. 8. T–H2O diagram calculated at P¼4�0 GPa for phengite-bearing eclogite YK128-12, using the bulk-rock compositiongiven in Table 5, with O¼0�25 mol % and H2O varying from 0to 1 mol %. The field labeled in bold type represents theobserved peak assemblage of interest.


Dow








similar results, but different minimum H2O values

owing to different amounts of modal phengite

(Supplementary Data Electronic Appendix Fig. S3). For

YKS-5, the phase assemblage Grt þ Omp þ Coe þ Ph þRt þ Kfs develops in the phase assemblage field with

H2O of <0�96 mol %. For WYKK-3, the phase assem-

blage of Grt þ Omp þ Ky þ Coe þ Ph þ Rt þ Kfs devel-

ops in the phase assemblage field with H2O of

<1�45 mol %.

In summary, we find that the H2O content in the bulk-

rock compositions can be constrained only to a max-

imum value by phase equilibrium modeling to avoid

free water at the peak stage. This value corresponds to

all the water incorporated in the hydrous mineral phen-

gite. Any arbitrary value of H2O higher than this max-

imum will allow free H2O or additional hydrous

minerals in the modeled phase assemblage at the peak

stage, such as lawsonite, that are not present in the

samples of this study. However, any lower value will

promote the development of K-feldspar. Therefore, the

H2O contents chosen for further modeling are 0�26 mol

% for YK128-12, 0�96 mol % for YKS-5 and 1�45 mol %

for WYKK-3. Given the absence of hydrous minerals in

the peak assemblage of YK5-2, for further modeling the

H2O content was set to zero.

P–T diagrams with constrained O and H2Ocontents for eclogite-facies metamorphismPhengite-bearing eclogite YK128-12The P–T pseudosection for phengite-bearing eclogite

YK128-12 (Fig. 9a) was calculated in the NCKFMASHTO

system using the composition given in Table 5 (for

Fig. 9). Compared with Fig. 6, phase relations in the new

diagram are much simpler and are dominated by higher

variance assemblage fields (five- or six-variant). In

Fig. 9a, Law is absent because of the low H2O content

and kyanite is absent at P above �3�1 GPa owing to the

reduced O content in the bulk-rock composition.

Although Kfs is present in all fields, the modeled con-

tent is less than 0�01 mol % and may be neglected.

Plagioclase and Opx are present at low P (<2�0 GPa)

and high T (>660�C). The observed mineral assemblage

of Grt þ Omp þ Coe þ Ph þ Rt occurs at P above

�3�1 GPa, but over the full range of T from 600 to 900�C.

Grossular and XFe [¼ Fe/(Fe þMg)] isopleths for garnet

(0�29–0�33 and 0�66–0�72, respectively), Si isopleths for

phengite (3�43–3�58 p.f.u.) and j(o) isopleths for ompha-

cite (0�58–0�66) were calculated for the modeled P–T

range (Fig. 9b). In general, the Si in phengite, the gros-

sular in garnet and the j(o) of pyroxene all increase with

increasing P and record the highest values at the high-

est P shown. By contrast, XFe decreases with increasing

T, with isopleths that have negative slopes in kyanite-

bearing mineral assemblages at lower P but positive

slopes in kyanite-absent mineral assemblages at higher

P. In the phase assemblage field of Grt þ Omp þ Ph þCoe þ Rt þ Kfs, isopleths of grossular and j(o) are wide-

ly spaced and mineral proportions show little or no

change (less than 1 mol %) across the field, although

Fe–Mg exchange continues between garnet and

omphacite with increasing T.

The grossular content in a large garnet increases

from core to mantle from 0�29 to 0�33, and then

decreases from 0�33 to 0�30 at the inner rim. How much

these variations in composition have been flattened by

volume diffusion is unknown (Caddick et al., 2010).

Taking the changes in grossular content at face value,

they suggest an increase in P followed by a decrease.

Thus, the grossular content in the apparent core of the

garnet might record part of the prograde evolution.

Because primary kyanite occurs as inclusions in garnet,

it is likely that the prograde path passed through a

kyanite-bearing field.

Based on the petrography, the peak mineral assem-

blage corresponds to the field Grt þ Omp þ Ph þ Coe

(þ Kfs) in Fig. 9a. Although changes in the composition

of garnet and omphacite are small and variations in

mineral proportions are minor across this high-variance

field, restricting our ability to quantify the P–T condi-

tions more tightly (Fig. 9b), variation in the Si content in

phengite is dominantly due to changes in pressure.

Thus, we may use isopleths of Si content in phengite to

estimate a minimum P. In this sample, the highest Si

content of 3�54 p.f.u. was retrieved from phengite

included in garnet and from large phengite flakes. This

value corresponds to P of 3�7–4�7 GPa at T from 600 to

900�C (Fig. 9b), indicating that the peak P for this sam-

ple was >3�7 GPa at T of 600–900�C. The lowest Si con-

tent of 3�43–3�44 p.f.u. was retrieved from fine-grained

phengite, which corresponds to P of 3�8–4�2 GPa at

T> 660�C. We interpret the small decrease in P implied

by the change in phengite composition to record de-

compression during exhumation, consistent with the

decrease in grossular content of garnet from the mantle

(0�33) to inner rim (0�30) discussed above. XFe in garnet

varies from 0�67 to 0�70 from mantle to rim and the j(o)

in primary omphacite varies from 0�60 to 0�64. Isopleths

for these compositions define a field with P of 2�1–

3�4 GPa and T of 620–815�C (Fig. 9b). The mineral as-

semblage in this field is Grt þ Omp þ Ph þ Ky þ Coe/Qz

(þ Kfs), consistent with the observed assemblage in the

sample. Kyanite in this field was formed via the reaction

Grt þ Coe! Ky þ Omp during decompression.

Phengite eclogite YKS-5The P–T pseudosection for phengite eclogite YKS-5

(Fig. 10a) was calculated in the NCKFMASHTO system

using the composition given in Table 5 (for Fig. 10a).

Phase relations for this sample are similar to those for

YK128-12. The peak mineral assemblage field of Grt þOmp þ Coe þ Ph þ Rt (þ Kfs) is located at P above

�3�1 GPa at T of 600–900�C. In this field, the Si in phengite

occurring as inclusions in garnet and as coarse flakes out-

side garnet (3�47–3�58 p.f.u.) records P of 3�6–4�8 GPa at

T>640�C (Fig. 10a). Isopleths for XFe (0�59–0�61) and

grossular content (0�25–0�29) in garnet from core to


Dow







mantle, as well as j(o) of omphacite (0�50–0�56), define a

field with P of 1�7–2�7 GPa and T of 670–760�C (Fig. 10a).

Kyanite-bearing eclogite WYKK-3The P–T pseudosection for the kyanite-bearing eclogite

(WYKK-3; Fig. 10b) was calculated in the NCKFMASHTO

system using the composition given in Table 5 (for

Fig. 10b). Figure 10b has been calculated up to P¼ 7�5 GPa

because the high XAl [Al2O3/(CaO þ NaO þ K2O) in mole

per cent] of the bulk-rock composition stabilizes kyanite

over a much larger range of pressures (up to 7�5 GPa over

the full T range) than in the phengite eclogites. The occur-

rence of kyanite as inclusions in garnet, as a rock-forming

mineral in the matrix and replacing garnet suggests that it

was stable during the prograde, peak and retrograde

stages. In Fig. 10b, isopleths for Si in phengite have rela-

tively steep positive slopes. In the field with the phase as-

semblage Grt þ Omp þ Ky þ Coe þ Rt (þ Kfs), the

isopleth for the highest Si in phengite (3�53 p.f.u.) occurs

at P of 3�9–5�9 GPa and T of 650–900�C (Fig. 10b). Isopleths

for XFe (0�64–0�66) and grossular content (0�26–0�32) in

garnet from core to mantle, and the j(o) of primary

omphacite inclusions in garnet and rock-forming ompha-

cite (0�51–0�59) define a field with P of 1�6–3�4 GPa and T

of 620–715�C in the mineral assemblage field Grt þ Omp

þ Kyþ Coe/Qz þ Rt (þ Kfs) (Fig. 10b).

Bimineralic eclogite YK5-2The P–T pseudosection for the bimineralic eclogite (YK5-

2; Fig. 10c) was calculated in the NCFMASTO system

using the composition given in Table 5 (for Fig. 10c). The

phase relations in this diagram are similar to those for

the phengite-bearing eclogites (Figs 9 and 10a), suggest-

ing that omission of K2O in the model system makes lit-

tle difference at the low concentration in this sample.

At P below 3�7–4�2 GPa, kyanite forms via the reaction

Grt þ Coe! Ky þ Omp during decompression, consistent

with the observed assemblage in this sample. The grossu-

lar content in garnet from an apparent core to the mantle

(0�31–0�33) corresponds to isopleths in the mineral assem-

blage field of Grtþ Ompþ Kyþ Coeþ Rt (Fig. 10c). Based

on petrological observations, the peak mineral assemblage

was Grt þ Omp þ Coe þ Rt, which is located in a field at P

above �3�7 GPa (Fig. 10c). In this field, neither mineral pro-

portions nor mineral compositions show much change.

The grossular content of garnet from the mantle to the

inner rim decreases from 0�33 to 0�29, corresponding to

decompression across isopleths in the Ky-bearing mineral

assemblage field (Fig. 10c). Given the garnet core–rim

structure shown by petrological observations, the grossu-

lar content from garnet core to mantle then to rim may re-

cord P–T information from the prograde, peak and

retrograde stages. Isopleths for XFe (0�67–0�71) and gros-

sular (0�29–0�33) in garnet from mantle to inner rim, and

the j(o) of primary omphacite (0�56–0�63) define a field with

P<3�7 GPa and T< 720�C in the mineral assemblage field

of Grtþ Ompþ Kyþ Coe/Qzþ Rt (Fig. 10c).

Assessing the presence of melt duringexhumationIn the literature, it has been suggested that low-volume

partial melting occurred during exhumation in the eclo-

gites at Yangkou Bay and at other localities in the Sulu

Belt (Zheng et al., 2011; Chen et al., 2014; Li et al., 2014;

Fig. 9. (a) P–T pseudosection calculated for phengite-bearing eclogite YK128-12, using the bulk-rock composition given in Table 5.(b) Shows calculated isopleths for grossular content (cg) and XFe (xg) in garnet, Si in phengite (Si) and j(o) of omphacite (jo). Thefields constrained by bold dashed lines indicate P–T constraints for near peak and retrograde stages, based on mineral assemb-lages and isopleths for the compositions of phengite and garnet and omphacite, respectively. The field labeled in bold type in (a)represents the observed peak assemblage of interest.


Dow





Wang et al., 2014, 2016, 2017). In the veins in the

phengite-bearing eclogite YK128-12, we interpret sym-plectites of Kfs þ Bt þ Qz þ Pl around coarse phengite,

separating it from quartz, as evidence of partial melting

during decompression.

We use phase equilibrium modeling of this low-vol-

ume partial melting to provide an important constraint

on the low-pressure part of the P–T evolution by defin-

ing an appropriate solidus for crystallization of themelt generated during the late decompression stage.

To determine the solidus and assess the presence of

melt during exhumation, we calculated phase dia-

grams involving melt for YK128-12. The effective com-

position for modeling (Table 5) was calculated by

integrating an equal volume of coarse phengite and

quartz, using the molar volumes of phengite and

quartz from Holland & Powell (1998) and the averagecomposition of phengite in the veins from Table 4. The

pseudosections were calculated in the KFMASHTO

system. There is negligible sodium and no calcium in

the coarse phengite in the veins, so Na2O and CaOwere not included in the modeling. A–x relationships

(melt model for metapelite, biotite, orthopyroxene,

muscovite and cordierite) used in the modeling are

from White et al. (2014). The a–x model for melt is

based on the a–x model for haplogranite melt of

Holland & Powell (2001), which was calibrated up to

1�0 GPa. Accordingly, supra-solidus phase equilibriamodeled at P> 1�0 GPa should be interpreted with due

caution. Pure phases include K-feldspar, quartz, kyan-

ite, sillimanite and aqueous fluid (H2O).

The H2O content in the effective composition was

calculated from the average analysis of phengite in the

veins from Table 4. The presence of hematite, goethite

and barite in thin grain boundary veinlets in the eclo-

gites at Yangkou Bay demonstrates the presence of anoxidizing fluid during retrograde metamorphism (Wang

et al., 2016). In this study, the high Ae content in sodic

Fig. 10. P–T pseudosection calculated for phengite eclogite YKS-5 (a), kyanite-bearing eclogite WYKK-3 (b) and bimineralic eclogite YK5-2(c). The bulk-rock compositions are given in Table 5. Calculated isopleths for grossular content (cg) and XFe (xg) in garnet, Si in phengite(Si) and j(o) of omphacite (jo) are shown in the figures. The fields constrained by bold dashed lines indicate P–T constraints for near peakand retrograde stages, based on mineral assemblages and isopleths for the compositions of phengite and garnet and omphacite, respect-ively. The field labeled in bold type represents the observed peak assemblage of interest. (See online version for color images.)


Dow





clinopyroxene against coarse quartz in the veinlets also

suggests that conditions during the retrograde meta-

morphism were relatively oxidizing. Therefore, the O

content in the effective composition was probably

higher than that used for the peak stage P–T pseudosec-

tion of Fig. 9. Following the method introduced above,

we calculated a T–O pseudosection at P¼ 1�2 GPa to de-

termine an appropriate O content (Supplementary Data

Electronic Appendix Fig. S4). The O content is con-

strained to be in the range of 0�0–0�4 mol %. Because

the O content was higher than that used for modeling

the peak eclogite-facies conditions (0�25 mol %), we

chose the maximum permissible value of O¼ 0�4 mol %

for modeling the P–T pseudosection discussed below.

The P–T pseudosection was calculated for P of 0�2–

1�6 GPa and T of 600–900�C (Fig. 11). Extrapolating the

a–x model for hydrous melt to very high pressures may

introduce errors (Holland & Powell, 2001), so we limit

the calculation to P of 1�6 GPa. In Fig. 11, the phase dia-

gram is dominated by five-variant fields. The solidus

occurs at T of 706–737�C, with the lowest T at

P¼ 1�47 GPa, where the slope changes from positive at

higher P to negative at lower P. At high P (>1�15 GPa)

and low T (<737�C), H2O is absent at sub-solidus condi-

tions. With increasing T at low to high P or decreasing P

at low to high T, Sill and then Ky replaces Ms via the re-

action Qz þ Ms ! Sill/Ky þ Bt þ Kfs þ H2O or Liq.

Cordierite is stable at T>640�C and P<0�7 GPa, and

orthopyroxene is stable at T> 814�C and P< 0�6 GPa.

In Fig. 11, calculated isopleths of mol % melt are

shown in the field of interest with the assemblage Bt þKfs þ Ms þ Qz þ Liq. The modeled melt content

increases with decreasing P (or increasing T) to a max-

imum of �8 mol % around 0�6 GPa and the isopleths be-

come slightly more closely spaced owing to the

progressive consumption of muscovite. We interpret

that the observed mineral assemblage of Kfs þ Bt þ Qz

[without Pl in the (Na2O þ CaO)-free system] that forms

the symplectites around the coarse phengite flakes in

the veins may have crystallized from melt under P–T

conditions corresponding to the solidus between the as-

semblage fields Bt þ Kfs þ Ms þ Qz þ Liq and Bt þ Kfs

þMs þ Qz þ H2O (Fig. 11). We estimate the symplectite

in the veins to be no more than 5% by volume. For a

modeled melt content of 1–5 mol %, P–T conditions at

the solidus are P of 0�9–1�3 GPa and T of 730–710�C

(Fig. 11).

P estimates using the average P–T methodIn addition to phase equilibrium calculations, the aver-

age P (avP) method of Powell & Holland (2008) was

used to estimate the P of equilibration for the main

rock-forming mineral assemblages in the eclogites at

Yangkou Bay. AvP was calculated using THERMOCALC

3.40 and the ds55 dataset (Holland & Powell, 1998;

updated November 2003). The equilibrated mineral

assemblages for calculation were Grt, Cpx, Ph, Rt and

Coe for YK128-12 and YKS-5, with Ky in addition for

WYKK-3. A–x models used were Grt (White et al., 2007),

Omp (Diener & Powell, 2012) and Ph (Coggon &

Holland, 2002). Rt, Coe and Ky are considered pure

phases, and fluid was assumed to be absent.

Calculations were performed at P of 1�5–5�5 GPa and T

of 600–900�C. Starting guesses for datafile coding of the

a–x relationships were calculated using the observed

mineral compositions, following the definitions for the

compositional variables for the minerals used in the a–x

relationships.As notified by E. Green (personal communication,

2017), when running THERMOCALC in AvP–T mode

there is a problem with the T-dependence of K.

Therefore, we calculated avP at a range of specified T, ra-

ther than AvP–T. A set of avP results using different start-

ing guesses shows that the maximum P can be obtained

using compositions of Grt with maximum mGrt zGrt2 (cor-

responding to maximum aPyr aGrs2), Cpx with maximum

jOmp/xOmp [jOmp corresponds to Na/(Na þ Ca) and xOmp

corresponds to Fe/(Fe þ Mg)] and Ph with minimum yPh

(maximum aCel), consistent with the logic outlined by

Carswell et al. (2000) for calculating the maximum

recorded pressure conditions for MT–UHP eclogite.

Results of the AvP calculations and the corresponding

standard deviations (SD) and sigfits are given in Table 6.

The pressures are similar for YK128-12 (4�106 0�44 GPa

at T¼800�C) and YKS-5 (4�15 6 0�47 GPa at T¼ 800�C),

but the pressure for WYKK-3 (4�406 0�5 GPa at

T¼ 800�C) may be slightly higher. AvP calculations for

Fig. 11. P–T pseudosection calculated for veins in the phengite-bearing eclogite YK128-12. The bulk-rock compositions usedare given in Table 5. Calculated isopleths for the amount ofmelt (green dotted line, in mol %) at supra-solidus conditionare modeled. The field constrained by a bold dashed line at thesolidus indicates P–T constraints for symplectite of Kfs þ Bt þQz þ Pl around coarse phengite in veins. The field labeled inbold type represents the observed peak assemblage of inter-est. (See online version for color images.)


Dow







the mineral assemblage Di þ Hbl þ Pl þ Qz in the sym-

plectite in YK128-12 gave results with SD and sigfits too

high to be useful, which may imply disequilibrium.

DISCUSSION

The application of phase equilibrium modeling toUHP eclogitesO and H2OAs we have demonstrated in this study, it is important

to determine the appropriate O and H2O contents in the

bulk-rock composition as the first step in phase equilib-

rium modeling of UHP eclogites, as these variables in-

fluence the phase relations, and the proportions and

compositions of phases (Clarke et al., 2006; Rebay et al.,

2010). Although the Fe2O3 in the bulk-rock composition

can be determined analytically, studies have shown

that the oxidation state of eclogites, and, therefore, the

Fe2O3 content, may be affected by retrograde meta-

morphism (e.g. Li et al., 2016a). In addition, fluid cannot

simply be assumed to have been present in excess dur-

ing the metamorphic evolution, particularly for those

eclogites with evidence implying fluid-deficient condi-

tions, such as the survival of intergranular coesite and

the preservation of igneous mineralogy and texture in

the outcrops at Yangkou Bay.

For these reasons, it is necessary to use an iterative

process that combines T/P–O/H2O diagrams with the

observed mineral assemblages, and the mineral pro-

portions and compositions, accepting the caveats dis-

cussed above, to constrain O and H2O contents (Rebay

et al., 2010). Our phase equilibrium modeling shows

that the P–T stability of the common mineral assem-

blage Grt þ Omp þ Coe/Qz þ Rt 6 Ph 6 Ky, as well as

lawsonite-bearing assemblages, in MT–UHP eclogites is

dependent on the oxidation state and whether H2O is

present in excess. In this study, using reasonable values

for O and H2O in the phase equilibrium modeling, we

do not predict lawsonite to be stable at any stage of the

P–T evolution of the eclogites at Yangkou Bay, in agree-

ment with petrological observations. By contrast, based

on phase equilibrium modeling with H2O in excess, Wei

et al. (2013) argued that lawsonite was present at the

metamorphic peak in MT–UHP eclogites from the South

Dabie belt, despite its absence from the samples they

studied.

Isopleth thermobarometryThe advantage of pseudosection thermobarometry is

that it uses all the available information from the

samples studied—the bulk-rock composition, the phase

assemblage and, potentially, the phase proportions and

compositions (Powell & Holland, 2008). Thus, in prin-

ciple, the phase relations combined with compositional

isopleths for critical minerals may yield reliable P–T

constraints for different stages of the metamorphic

evolution.

For MT–UHP eclogites, the residence time at an ele-

vated temperature (>700�C) is likely to promote cation

diffusion and mineral re-equilibration (e.g. Chakraborty

& Ganguly, 1991; Caddick et al., 2010), potentially

resulting in elimination of information relating to the

prograde and/or peak stages of the metamorphism

from those rock-forming minerals that exhibit solid so-

lution, such as garnet and pyroxene. This problem has

been well characterized in granulite-facies metamorph-

ism and methods have been proposed to mitigate the

effects of diffusive re-equilibration (e.g. Frost & Chacko,

1989; Fitzsimons & Harley, 1994; Pattison & Begin,

1994; Pattison, 2003). The problem has not been as

widely appreciated in UHP metamorphism, but we note

that Baldwin et al. (2008) were not able to retrieve P–T

conditions in the coesite stability field using conven-

tional thermobarometry on eclogite with preserved

coesite. This failure to retrieve P–T conditions compat-

ible with the presence of coesite using conventional

thermobarometry was most probably due to post-peak

diffusive re-equilibration. Similarly, in pseudosection

thermobarometry, if the stability field for the peak min-

eral assemblage occurs over a wide range of P and T,

such as the Grt þ Omp þ Rt þ Coe 6 Ky 6 Ph (þ Kfs)

assemblages in this study, diffusive re-equilibration

may limit our ability to narrow down the range of P–T

conditions using isopleths of mineral composition, as

discussed by Wei et al. (2013).

Compositional isopleths commonly used in pseudo-

section thermobarometry of HP–UHP eclogites are the

divalent cations in garnet and Si in phengite, together

with the j(o) of omphacite (e.g. Wei et al., 2009, 2013;

Bruand et al., 2010; Groppo et al., 2015; Li et al., 2016b).

Although cation diffusion should be expected at

T> 700�C (Caddick et al., 2010), modified garnet growth

zoning can still be identified in MT–UHP eclogites from

both the Sulu and Dabie UHP belts (e.g. Enami &

Nagasaki, 1999; Schmid et al., 2000; Zhang et al., 2005),

including those with evidence of a granulite-facies over-

print (e.g. Chen et al., 1998), in eclogites from the

Variscan basement of the Bohemian Massif (O’Brien,

1997), and in migmatites from the Himalaya (Harris

et al., 2004) and the Variscan belt (Jones & Brown,

1990; Kalt et al., 2000). In these examples, the

Table 6: AvP results for the phengite-bearing eclogite YK128-12 and YKS-5 and the kyanite-bearing eclogite WYKK-3 at Yangkou

YK128-12 YKS-5 WYKK-3

T (�C) 700 750 800 850 900 700 750 800 850 900 700 750 800 850 900avP (GPa) 3�78 3�94 4�10 4�25 4�40 3�83 3�99 4�15 4�31 4�46 3�93 4�17 4�40 4�64 4�87SD 0�28 0�35 0�44 0�54 0�65 0�30 0�37 0�47 0�58 0�70 0�34 0�41 0�50 0�60 0�73Sigfit (rfit) 1�2 1�4 1�6 1�9 2�2 1�3 1�5 1�9 2�2 2�5 1�8 2�0 2�2 2�5 2�8


Dow





preservation of modified growth zoning is interpreted

to be due to either the large grain size of the garnet or

decompression and fast cooling.

Other factors that may influence the retention of

compositional properties developed before or at the

metamorphic peak include diffusion kinetics of the ions

of interest, deformation, the presence of a fluid and the

oxygen fugacity (e.g. Chakraborty & Ganguly, 1991;

Caddick et al., 2010). Numerical modeling based on

experiments indicates that volume diffusion of Ca in

garnet is significantly more sluggish than that of the

other bivalent cations owing to its larger ionic size

(Chakraborty & Ganguly, 1991; Schwandt et al., 1996).

Therefore, Ca in garnet tends to better preserve pro-

grade chemical variation, whereas Mg, Fe and Mn are

commonly re-equilibrated and retain little or no chem-

ical zoning.

In this study, element mapping of garnet from two

of the samples shows internal zoning of Ca and, less

strongly, Mg and Fe, but not Mn (Fig. 4b and h). The

rate of exhumation of the eclogites at Yangkou Bay

and the fluid-deficient conditions may have

contributed to the preservation of Ca zoning in some

garnets. Although we are uncertain about the extent

to which the Ca zoning profiles in garnet have been

modified, the modeled Ca isopleths for garnet com-

positions from inner core to outer core from samples

YK128-12 and YK5-2, in combination with information

derived from mineral inclusions in garnet, suggest

preservation of a segment of the prograde path for

these eclogites (Figs 9b and 10c). By contrast, the

modeled Grs isopleths for garnet compositions from

mantle to inner rim in combination with the j(o) iso-

pleths for omphacite compositions indicate that re-

equilibration may have occurred along the retrograde

path (Figs 9b and 10a–c). Whether the zoning profiles

of these garnets have been modified during the sub-

sequent evolution of the eclogites cannot be

assessed, so the P–T results we present above based

on the zoning are taken as minima.

The Si content of phengite is more dependent on

pressure than temperature (e.g. Massonne & Schreyer,

1987; Schmidt et al., 1997, 2001). In experimental and

modeling studies of a variety of bulk compositions,

Fig. 12. P–T path for the eclogites at Yangkou Bay. The P–T path includes five stages based on phase diagrams for the four eclogitesmodeled, average P calculations and TiO2-in-phengite thermometry. Patches and the ellipse in different colors represent P–Tranges derived from different samples, corresponding to the P–T fields in the individual P–T pseudosections. The reaction Ph þ Cpxþ Coe/Qz þ (Ky)!Melt is from Liu et al. (2009b). (See online version for color images.)


Dow





phengite exhibits systematic changes of Si content with

P (Hermann, 2003; Caddick & Thompson, 2008;

Hermann & Spandler, 2008; Auzanneau et al., 2010; Wei

et al., 2013). The increase in Si in phengite with increas-

ing P is due to a Tschermaks-type exchange where Mg

þ Si replaces Al between muscovite and celadonite in

mineral assemblages buffered by a SiO2 phase

(Hermann, 2003). However, during decompression,

change in the Si content of phengite is expected to be

limited, because change requires a coupled substitution

involving the tetrahedral site, which is rate limiting for

diffusion. Therefore, the Si in phengite is expected to be

a much more robust constraint on the highest pressure

preserved by the peak mineral assemblage than the Ca–

Fe–Mg proportions in garnet, where compositional

change during decompression involves only cation dif-

fusion on the dodecahedral site. Phase equilibrium

modeling of the samples in this study shows that at a

given T, Si in phengite depends mainly on P, whereas

changes in H2O and bulk-rock composition have little ef-

fect. Thus, the Si content of phengite is believed to be a

robust barometer that records reliable peak or near

peak pressures. To constrain the full P–T path for the

eclogites at the Yangkou locality, we rely on Si-in-

phengite isopleth barometry within the appropriate

phase assemblage fields for each of the three phengite-

bearing samples.

P–T path for UHP eclogites at Yangkou BayIf we accept the possibility that prograde compositions

may be preserved in the cores of some coarse-grained

garnets from the phengite-bearing (YK128-12) and bimi-

neralic (YK5-2) eclogites, we may define a segment of

the prograde P–T evolution using these data. Then by

combining the results from all four samples at UHP and

HP conditions for the retrograde P–T evolution, and

using the result from the symplectites around coarse

phengite flakes in the veins from sample YK128-12, we

may propose a full P–T path for the eclogites at

Yangkou Bay, from the prograde stage to the peak

metamorphic conditions and through several retro-

grade stages (Fig. 12).

The prograde stageBased on primary inclusions of Ky þ Omp þ Ph þ Coe

in garnet from YK128-12 and WYKK-3, the prograde P–T

path (Fig. 12, stage 1) passed through the assemblage

field Grt þ Omp þ Ky þ Ph þ Coe þ Rt (þ Kfs) (Figs 9b

and 10b). Although we cannot constrain the prograde

evolution within this assemblage field, increasing gros-

sular from inner core to outer core of garnet in YK128-

12 is consistent with increasing P and/or T through this

field. Accordingly, we propose a close to isobaric heat-

ing path at �2�8 GPa with a shallow dT/dP slope, fol-

lowed by a rapid compression to the peak metamorphic

conditions as shown in Fig. 12. Such a pattern is similar

to modeled P–T paths for the ocean crust in modern

subduction zones (Syracuse et al., 2010). In these

models, the low dT/dP paths follow a high dT/dP seg-

ment at intermediate depths, with of the order of a

300�C increase in temperature from 2�5 to 3�0 GPa. This

is attributed to heating at the plate interface at the tran-

sition from partial to full coupling between the slab and

the overlying mantle wedge in the subduction zone

(Syracuse et al., 2010). The similar change in the P–T

path during stage 1 is broadly consistent with prograde

Zr-in-rutile temperatures at 3�0 GPa of �680�C obtained

from rutile inclusions in garnet from eclogite at

Yangkou Bay (Wang et al., 2016).

The maximum P–T conditions retrievedThe peak metamorphic paragenesis in the phengite-

bearing (YK128-12 and YKS-5) and bimineralic (YK5-2)

eclogites comprises Grt þ Omp þ Coe þ Rt 6 Ph. In the

calculated P–T pseudosections these mineral assemb-

lages are stable over a wide range of P–T conditions at

P> 3�1 GPa and T of 600–900�C (Figs 9 and 10a, c), but

the proportions and compositions of garnet and

omphacite change little. By contrast, the Si content in

phengite changes systematically with P and, therefore,

may be used to estimate the P at or near the meta-

morphic peak (Fig. 12, stage 2).

In samples YK128-12, YKS-5 and WYKK-3, phengite

shows a limited range of covariation of Al and TiO2 con-

tents, and Mg/(Mg þ Fe) ratios with Si, implying some

resetting of phengite composition during decompres-

sion (Supplementary Data Electronic Appendix Fig. S1).

Nevertheless, a minimum value of the near peak P can

be estimated from the highest Si content in phengite in

each of the phengite-bearing eclogites. Phengite from

YK128-12 has a maximum Si of 3�54 p.f.u. and that from

YKS-5 has a maximum Si of 3�58 p.f.u., which give simi-

lar P of 4�0–4�6 GPa at T of 600–900�C, indicating that

the P should be no less than �4�5 GPa at T �850�C. By

contrast, the highest Si content (3�53 p.f.u.) in phengite

from the kyanite-bearing eclogite (WYKK-3) suggests P

of 3�9–5�9 GPa at T of 650–900�C (Fig. 10b). The Si in

phengite isopleths for each sample have different P–T

slopes. Thus, for an appropriate range of peak tempera-

tures of 800–900�C, we may estimate pressure based on

the crossing isopleths for the highest values of Si in

phengite, which yields a range from 4�2 to 5�9 GPa.

Given the possibility that the Si content in phengite may

have been reset during decompression, these pressures

probably represent minimum values, which allows the

possibility that the maximum P–T conditions may have

exceeded �5�5 GPa at >850�C.

Using avP, a similar range of P of 3�8–4�5 GPa at T of

700–900�C was retrieved from the assemblage Grt þ Ph

þ Omp þ Coe þ Rt in the phengite-bearing eclogites

YK128-12 and YKS-5, and a slightly higher P of 3�9–

4�9 GPa at T of 700–900�C was calculated for the mineral

assemblage Grt þ Ph þ Omp þ Ky þ Coe þ Rt in the

kyanite-bearing eclogite WYKK-3. The avP calculations

are broadly consistent with the pressures constrained

by the Si content in phengite. However, using isopleths


Dow







of Grs and XFe in Grt with j(o) in Omp from the four

eclogites yields P–T results that are inconsistent with

those derived using Si in Ph (Figs 9b and 10a–c), which

may imply failure to maintain equilibrium between

these minerals during decompression. Therefore, the

avP results may not record the maximum P.

Disequilibrium plagues conventional thermobarome-

try owing to diffusional re-equilibration during retro-

gression, and the recovery of peak T is considered

unlikely (see Hacker, 2006). Although �Stıpska & Powell

(2005) used Grt–Cpx Fe–Mg exchange thermometry to

retrieve the peak T for an eclogite with the assemblage

Grt þ Omp þ Qz, the calculated uncertainty was of the

order of 100�C. Alternatively, experimental studies have

shown a strong positive correlation of TiO2 (or Ti) in

phengite with temperature when buffered with rutile

and quartz/coesite (Hermann & Spandler, 2008;

Auzanneau et al., 2010). Therefore, the TiO2 content in

phengite may be used to constrain the T at the peak or

near peak conditions for MT–UHP eclogite. The range of

TiO2 contents in phengite from all four samples in this

study suggests that the phengite compositions may

have been partially re-equilibrated during exhumation,

as also suggested by the range of Si contents in phen-

gite. The highest TiO2 contents in phengite are 0�71,

0�68 and 0�61 wt % for samples YK128-12, YKS-5 and

WYKK-3, respectively, constraining a maximum T of

800–900�C, at 4�5–5�9 GPa (Fig. 12).

In this study of the eclogites at Yankgou Bay, the max-

imum P–T conditions were potentially >5�5 GPa and

>850�C (Fig. 12, stage 2), which is higher than those esti-

mated by previous studies based on conventional thermo-

barometry, but closer to the P of >7�0 GPa at T> 1000�C

suggested by Ye et al. (2000), based on reintegration of

exsolved minerals with their host garnet. Furthermore,

our result is consistent within uncertainty with the min-

imum P of 5�5 GPa at T of 740–870�C retrieved by Liu et al.

(2006b) at Shanqingge, also in the Sulu Belt, but a few

hundred kilometers to the SW of Yangkou Bay.

The retrograde stagesEarly decompression (Fig. 12, stage 3) is recorded by

coarse phengite flakes and quartz in the veins from

YK128-12. The Si content of the coarse phengite ranges

from 3�53 to 3�44 p.f.u., corresponding to P of 4�5–

3�4 GPa (Hermann, 2003) or 4�0–3�2 GPa [equation (8) of

Caddick & Thompson, 2008] at T¼ 850–800�C (Fig. 12).

The TiO2 content in the phengite was not used to con-

strain temperature because rutile was absent and the

activity of TiO2 was unbuffered. Given the calculated

pressure, we speculate that quartz in these veins may

have been initially precipitated as coesite.

Subsequently, the UHP eclogites decompressed across

the coesite to quartz phase change (Fig. 12, stage 3 to

stage 4), facilitating re-equilibration of the main rock-

forming minerals.

During decompression and cooling (Fig. 12, stage 4),

isopleths for the Grs content of garnet, the XFe of the

garnet core or mantle and the j(o) of omphacite indicate

re-equilibration in the phase assemblage field Grt þOmp þ Ky þ Coe/Qz þ Kfs 6 Ph, consistent with the de-

velopment of secondary kyanite in YK128-12 (Fig. 2f),

WYKK-3 (Fig. 3e) and YK5-2 (Fig. 3i). These compos-

itional variations define a wide P–T range for each sam-ple (Figs 9b and 10), but whether the mineral

compositions represent equilibrium is uncertain and

our preferred P–T path passes through the high-T edge

of the range of P–T conditions determined.

Rock-forming phengite in the eclogites maintains a

euhedral shape and shows no evidence of instability,

such as breakdown by partial melting, implying that it

was stable throughout the post-peak P–T evolution. InFig. 12 we plot the experimentally determined solidus

for a phengite-bearing eclogite from the Dabie UHP ter-

rane (Liu et al., 2009b). This eclogite had a low H2O con-

tent (�0�29 wt %, essentially only the hydroxyl in

phengite and zoisite) such that initial melting of the

phengite-bearing assemblage occurred at relatively

high temperatures of 800–920�C, at P of 1�5–3�0 GPa.

The H2O content of the eclogites at Yangkou Bay is also

low (<0�5 wt %, Table 5), which could be the reason forthe stability of phengite during the decompression. This

expectation is consistent with the proposed P–T path,

which is located to the low-T side of the phengite-

breakdown melting reaction from Liu et al. (2009b).

Further decompression and local fluid migration

along veins and veinlets led to the development of a

symplectite of Kfs þ Bt þ Qz þ Pl around coarse phen-

gite in veins in YK128-12 and Hbl þ Pl 6 Qz 6 Cpx3 afteromphacite in veinlets in YK128-12 and YK5-2 (Fig. 12,

stage 5). Phase equilibrium modeling of the simplified

vein composition in YK128-12 yielded P–T conditions of

1�3–0�9 GPa at 710–730�C for the symplectite assem-

blage of Kfs þ Bt þ Qz (þ Pl) around phengite. These P–

T conditions provide the final constraint on the exhum-

ation P–T path for the eclogites.

During exhumation from HP eclogite- to amphibo-lite-facies conditions, the retrograde P–T path passed

through metamorphic conditions of P¼2�4–1�7 GPa at

T¼ 800–750�C and P¼ 1�3–0�9 at T¼ 730–710�C from

stage 4 to stage 5 (Fig. 12). The symplectites of Hbl þ Pl

6 Qz 6 Cpx3 after omphacite in veinlets and along the

edges of veins are interpreted to have formed by reac-

tion of local grain boundary fluid with the host under

amphibolite-facies metamorphic conditions, althoughthe P–T conditions cannot be constrained.

The final stage of the retrograde evolution is repre-

sented by coarse hornblende and plagioclase bordering

the veins in sample WYKK-3. Because the hornblende

and plagioclase compositions are different from those

in the symplectites along the veinlets in YK128-12 and

YK5-2, we infer that fluid migration along fractures

through the eclogite led to this retrogression at am-phibolite-facies conditions in lower crust.

In summary, this P–T evolution is similar to that in

many UHP terranes in the world (Ernst & Liou, 2008;

Liou et al., 2009).


Dow





Uncertainties on the maximum P–T resultsIt is essential to assess the uncertainties when interpret-

ing the P–T results of phase equilibrium modeling

(including the avP calculations). Systematic and ran-

dom uncertainties should be propagated through all

calculations. Sources of systematic uncertainty include

the internally consistent dataset and mineral a–x rela-

tionships, whereas sources of random uncertainty are

mainly related to the analysis of minerals and determin-

ation of the effective composition of the bulk-rock or the

equilibration volume (Powell & Holland, 2008; Palin

et al., 2016). Normally both systematic and random

uncertainties need to be considered. However, when

comparing P–T results determined using the same

methods for several samples, systematic uncertainties

are common to all results, so that only relative errors

owing to random uncertainties need to be considered;

relative errors are generally small.

P–T conditions based on thermodynamic modeling

rely on rock, domain and mineral compositions. To min-

imize the uncertainties stemming from the electron

microprobe analysis of minerals, it is advisable to use

isopleths and independent sets of reactions that do not

involve Fe-bearing end-members (Hacker, 2006).

Uncertainties related to the effective composition may

be assessed using T/P–X pseudosections to investigate

those elements that influence the position of assem-

blage fields, and the mineral proportion and compos-

ition in phase diagrams (Powell & Holland, 2008).

For basic rocks, O and H2O are the main constituents

in the effective composition that affect the phase rela-

tions of high-variance assemblages at UHP conditions.

Using the appropriate O and H2O contents minimizes

the uncertainty associated with this composition. Thus,

in this study we calculated T/P–O/H2O pseudosections

to determine appropriate values for these two variables

and, based on the results of the modeling, we adjusted

the bulk-rock compositions accordingly.

For eclogite mineral assemblages such as Grt þOmp þ Coe/Qz þ Rt 6 Ky 6 Ph, the error derived from

mineral compositions is mainly due to uncertainty asso-

ciated with the ferric iron content in omphacite, which

might be 10% of the calculated result at UHP conditions,

whereas uncertainties associated with garnet and phen-

gite increase the compound error by only an additional

10% (Holland & Powell, 2008). In pseudosection ther-

mobarometry, the advantage of using isopleths of Si in

phengite, grossular in garnet and j(o) in omphacite is

that they involve no Fe-bearing end-members and the

uncertainty related to the Fe3þ in omphacite (and in gar-

net) becomes moot. In this study, the highest P was

retrieved from the Ky-bearing eclogite WYKK-3, where

Si in phengite in the assemblage Grt þ Omp þ Ky þCoe þ Ph (þ Kfs) indicates P> 5�5 GPa at T> 850�C. In

the analyses of the secondary standards, elements

were all within about 1% relative, with the exception of

Fe (4% relative). Therefore, the propagated uncertainty

on P constrained by Si-in-phengite barometry is

�0�20 GPa at T of 850–900�C.

For WYKK-3, the systematic errors on the P derived by

pseudosection thermobarometry are �0�07 GPa (at 1r) for

Si ¼ 3�53 p.f.u. in Ph, �0�07 GPa for j(o) ¼ 0�66 in Omp and

�0�12 GPa for z(g) ¼ 0�32 in Grt. However, we note that

these errors are based on propagation of uncertainties on

the thermodynamic data of the end-members of the

phases and do not include those associated with formula-

tion of the a–x relationships of the phases (see tutorials on

THERMOCALC website). The absolute errors on P derived

by combining the systematic and random uncertainties

discussed above for sample WYKK-3 are �0�3 GPa for the

maximum P of 5�5–5�9 GPa at T of 850–900�C; this esti-

mate should be taken as a minimum. Errors for the other

two phengite-bearing eclogites (samples YKS-5 and

YK128-12) are similar.

For avP calculations, the error on the results is

propagated from the uncertainties in the internally con-

sistent dataset, the interaction energies and the input

mineral compositions (Powell & Holland, 1994, 2008).

For WYKK-3, the maximum avP result of 4�4–4�6 GPa at

T of 800–850�C gives absolute errors of 0�5–0�6 GPa (at

1r; Table 6). The errors for the maximum avP calcula-

tions for the other two samples (YK128-12 and YKS-5)

are similar (Table 6).

In addition, differential re-equilibration of the main

rock-forming minerals during decompression may intro-

duce uncertainties into the results. Therefore, the errors

discussed above are almost certainly underestimated and

the results should be taken as minimum estimates.

Behavior of fluid during decompressionThe survival of intergranular coesite in eclogite and

preservation of relict igneous mineralogy and texture in

the protolith gabbro and surrounding gneisses at

Yangkou Bay (Liou & Zhang, 1996; Wallis et al., 1997;

Zhang & Liou, 1997; Wang et al., 2018), combined with

the conservation of pre-metamorphic O-isotope signa-

tures at General’s Hill (<2 km south of Yangkou; Wang

et al., 2017) indicate that the UHP metamorphism

occurred under fluid-deficient or fluid-absent conditions

(see Schorn & Diener, 2017). Indeed, the survival of

intergranular coesite requires essentially ‘dry’ condi-

tions (Mosenfelder et al., 2005; Wang et al., 2018).

However, the common development of veins and vein-

lets in the eclogites during the early and intermediate

stages of the retrograde evolution implies the presence

and local migration of some fluid. We interpret this ap-

parent contradiction to mean that during exhumation

the eclogite at Yangkou Bay became domainal, with low

a(H2O) host-rock domains where garnet and omphacite

have partially re-equilibrated during decompression

and intergranular coesite has survived, and high a(H2O)

domains represented by the veins and veinlets where

new mineral assemblages were formed, and if coesite

was present in these domains it was inverted to quartz.


Dow





As discussed above, it is likely that the coarse phen-

gite flakes in the veins were precipitated under UHP

conditions (stage 3 in Fig. 12) close to the second critical

endpoints (P of 3�0–3�5 GPa at T of 750�C) in basalt/gran-

ite–H2O systems (Hermann et al., 2006; Hack et al.,

2007; Mibe et al., 2011). At UHP conditions and

T> 750�C, the nature of the fluid depends on the solute

content and it might be called supercritical fluid or hy-

drous melt according to preference at high solute con-

tent (Hermann et al., 2006, 2013; Wang et al., 2017).

Hereafter, we will simply use the term fluid. The source

of the fluid that precipitated the coarse phengite is un-

certain. As proposed by Wang et al. (2018), it may have

been exsolved from the nominally anhydrous minerals

that incorporate water as both structural hydroxyl and

molecular water in point defects within the crystal lat-

tice (Xia et al., 2005; Johnson, 2006; Chen et al., 2007,

2011; Zheng, 2009; Zhang et al., 2011). However, the

amount of fluid is likely to have been limited. The sur-

vival of intergranular coesite only millimeters away

from the veins and veinlets suggests the fluid was gen-

erated locally and that migration of this fluid was lim-

ited. Nevertheless, the higher a(H2O) in the veins may

have promoted partial melting at lower pressure during

the retrograde evolution, as recorded by the symplec-

tites of Kfs þ Pl þ Bt þ Qz around coarse phengite in the

veins (Fig. 2c and d).

In the simplified system used to model the symplec-

tites (Fig. 11), crystallization of melt at the solidus would

develop the phase assemblage Kfs þ Bt þ Qz þMs. The

presence of Pl in the symplectites in the veins indicates

that a small amount of fluid probably migrated into the

veins, introducing sodium (6 Ca) into the melt. The so-

dium (and calcium) could have been a product of the

transition from Na-rich Cpx in the host to Na-poor Cpx

in the veinlets.

Tectonic implicationsBased on conventional thermobarometry and/or Ti-in-

zircon and Zr-in-rutile thermometry, three patterns of

exhumation P–T path have been proposed for the Sulu

Belt: isothermal decompression followed by decreasing

P and T (e.g. Zhang & Liou, 1997; Banno et al., 2000;

Nakamura & Hirajima, 2000; Zhang et al., 2009), decom-

pression with significant cooling (e.g. Nakamura &

Hirajima, 2010; Li et al., 2014), and decompression with

heating (e.g. Wang et al., 1993; Zheng et al., 2011; Li

et al., 2014; Wang et al., 2016). At Yangkou Bay, previ-

ous studies of the eclogites have resulted in proposals

for all three types of exhumation P–T path (Zhang &

Liou, 1997; Liu et al., 2009a; Wang et al., 2014, 2016).

This variety of P–T paths may be partly due to the

large differences in calculated temperatures for the

peak or retrograde stages. One explanation for these

differences may be the application of different thermo-

barometric methods, combined with uncertainty over

the amount of Fe3þ in the ferromagnesian minerals.

Alternatively, different units in the Sulu Belt could have

experienced different exhumation processes. Eclogites

from Weihai and Rongcheng in the northeastern Sulu

Belt record evidence of a granulite-facies overprint with

the development of Opx þ Pl þ Qz (e.g. Wang et al.,

1993; Banno et al., 2000; Nakamura & Hirajima, 2000).

By contrast, at Yangkou Bay and Taohang, and for sam-

ples from the CCSD, symplectites of Cpx þ Pl þ Hbl þQz indicate an amphibolite-facies overprint during the

last stage of the exhumation (e.g. Zhang et al., 2006;

Zhu et al., 2007; Nakamura & Hirajima, 2010; this study).

The rare occurrence of blueschists in the southern sec-

tor of the Sulu Belt indicates lower P–T conditions than

further north. Therefore, a common P–T path during ex-

humation should not be expected for the HP/UHP rocks

in the Sulu Belt.

The new P–T path defined in this study is unlike

those proposed previously for Yangkou Bay and other

localities in the central and northern Sulu Belt. Here we

suggest that the prograde P–T path comprised a seg-

ment with a high dT/dP gradient, followed by a segment

with a low dT/dP gradient, similar to the P–T paths mod-

eled for the ocean crust during active subduction at the

present day (Syracuse et al., 2010). Based on peak min-

eral assemblages and mineral compositions, meta-

morphism at Yangkou Bay is confirmed as MT–UHP

type, with maximum P–T of >5�5 GPa at >850�C. These

P–T conditions cannot be retrieved using conventional

thermobarometry, although slightly higher P–T condi-

tions have been proposed based on interpretations of

textural and mineralogical features in eclogites (Ye

et al., 2000). In addition, similar P–T conditions to those

at Yangkou Bay have been retrieved from marble in

other units in the Sulu Belt (Liu et al., 2006b). The ex-

humation P–T path shows a continuous decrease in P

and T to high-pressure amphibolite-facies conditions.

The maximum P of >5�5 GPa indicates that different

units in the Sulu Belt may have been exhumed from dif-

ferent portions of the subduction system, as has been

discussed in other orogens such as the Alps, where dif-

ferent P–T paths are registered in adjacent segments of

the same tectonic unit (e.g. Bucher et al., 2005; Groppo

et al., 2009). To address this issue, additional quantita-

tive thermobarometry should be conducted on other

units in Sulu in the future. Based on the success of this

study, phase equilibrium modeling of eclogites at other

localities in the Sulu Belt may be useful in addressing

the subduction and exhumation processes that oper-

ated along the length of the belt.

CONCLUSIONS

1. We have shown that it is necessary to constrain the

O and H2O contents in the bulk-rock composition

prior to calculating a P–T pseudosection for MT–UHP

eclogite, otherwise the P–T pseudosection may be

unsuitable for the intended study. The results of this

study show that the P–T stability of the commonly

developed mineral assemblages Grt þ Omp þ Coe/


Dow





Qz 6 Ph 6 Ky in MT–UHP eclogites are dependent

on both the O and H2O contents.

2. A new P–T path is proposed for the eclogites at

Yangkou Bay, extending from the prograde evolution

through the peak UHP conditions to the HP and am-

phibolite-facies retrograde conditions. The modeled

peak phase assemblage of Grt þ Omp þ Coe 6 Ky 6

Ph (þ Kfs) is in agreement with the petrological obser-

vations. Modeled isopleths for Si in phengite com-

bined with TiO2 contents in phengite yield minimum

P–T conditions of P> 5�5 GPa at T> 850�C. Modeled

isopleths for the compositions of garnet and ompha-

cite indicate that re-equilibration occurred in the phase

assemblage field of Grt þ Omp þ Ky þ Coe/Qz (þ Kfs).

Further decompression led to the development of Kfs

þ Bt þ Qz þ Pl around coarse phengite in veins and

symplectites of Hbl þ Pl 6 Qz 6 Cpx after omphacite

in veinlets. The retrograde P–T path passed through

metamorphic conditions of P¼ 4�0–3�4 GPa at T¼ 850–

800�C to 2�4–1�7 GPa at T¼ 800–750�C, before exhum-

ation to shallower crustal levels at P¼ 1�3–0�9 GPa and

T¼ 730–710�C.3. During exhumation, the eclogites at Yangkou Bay

became domainal, with fluid-deficient host-rock

domains where intergranular coesite was preserved

and fluid-present domains (veins and veinlets)

where new mineral assemblages developed. In

veins where a(H2O) was higher, decompression led to

partial melting of coarse phengite þ quartz in veins

and the crystallization of Kfs þ Bt þ Qz (þ Pl) around

coarse phengite at the solidus. In veinlets, migration

of fluid into the host along grain boundaries pro-

moted the limited development of symplectites com-

prising Hbl þ Pl 6 Qz 6 Cpx replacing omphacite.

4. This study demonstrates that phase equilibrium

modeling combined with hybrid thermobarometric

methods allows reconstruction of the P–T path for

MT–UHP eclogites at Yangkou Bay and the monitor-

ing of fluid behavior during decompression. The

new P–T path contrasts with those proposed previ-

ously for this locality and other localities in the Sulu

Belt. Additional studies using phase equilibrium

modeling are required to recognize spatial differen-

ces in P–T evolution in the Sulu Belt and to relate

these differences to subduction and exhumation

processes for different HP/UHP units within the belt.

ACKNOWLEDGEMENTS

Reviews by Eleanor Green, Johann Deiner and Gisella

Rebay, plus comments by Editor Joerg Hermann, sub-

stantially improved the original paper.

FUNDING

This study was supported by the National Natural Science

Foundation of China (Grants 41502043, 41272225 and

41572182), the Fundamental Research Funds for the

Central Universities, China University of Geosciences,

Wuhan (Grants CUG-G1323511572 and CUGL150815) and

China Postdoctoral Science Foundation (Grant

2016T90742), the MOST Special Fund (MSFGPMR02), and

the Opening fund (GPMR210703) from the State Key

Laboratory of Geological Processes and Mineral

Resources, China University of Geosciences Wuhan. B.X.

has been supported by the China Scholarship Council for

the postdoctoral program in the University of Maryland,

College Park.

SUPPLEMENTARY DATA

Supplementary data for this paper are available at

Journal of Petrology online.

REFERENCES

Auzanneau, E., Schmidt, M. W., Vielzeuf, D. & Connolly, J. A. D.

(2010). Titanium in phengite: a geobarometer for high tem-

perature eclogites. Contributions to Mineralogy and

Petrology 159, 1–24.Baldwin, S. L., Webb, L. E. & Monteleone, B. D. (2008). Late

Miocene coesite-eclogite exhumed in the Woodlark Rift.

Geology 36, 735.Banno, S., Enami, M., Hirajima, T., Ishiwatari, A. & Wang, Q. C.

(2000). Decompression P–T path of coesite eclogite to

granulite from Weihai, eastern China. Lithos 52, 97–108.Brown, M. & Johnson, T. (2018). Secular change in metamorph-

ism and the onset of global plate tectonics. American

Mineralogist https://doi.org/10.2138/am-2018-6166.Bruand, E., Stuwe, K. & Proyer, A. (2010). Pseudosection model-

ling for a selected eclogite body from the Koralpe (Hohl),

Eastern Alps. Mineralogy and Petrology 99, 75–87.Bucher, K., Fazis, Y., De Capitani, C. & Grapes, R. (2005). Blueschists,

eclogites, and decompression assemblages of the

Zermatt–Saas ophiolite: high-pressure metamorphism of sub-

ducted Tethys lithosphere. American Mineralogist 90, 821–835.Caddick, M. J. & Thompson, A. B. (2008). Quantifying the

tectono-metamorphic evolution of pelitic rocks from a wide

range of tectonic settings: mineral compositions in equilibrium.

Contributions to Mineralogy and Petrology 156, 177–195.Caddick, M. J., Konopasek, J. & Thompson, A. B. (2010).

Preservation of garnet growth zoning and the duration of pro-

grade metamorphism. Journal of Petrology 51, 2327–2347.Carswell, D. A. (1990). Eclogite Facies Rocks. Glasgow: Blackie,

396 pp.Carswell, D. A., Wilson, R. N. & Zhai, M. (2000). Metamorphic

evolution, mineral chemistry and thermobarometry of

schists and orthogneisses hosting ultra-high pressure eclo-

gites in the Dabieshan of central China. Lithos 52, 121–155.Chakraborty, S. & Ganguly, J. (1991). Compositional zoning

and cation diffusion in garnets. In: Ganguly, J. (ed.)

Diffusion, Atomic Ordering, and Mass Transport. New York:

Springer, pp. 120–175.Chen, N. S., Sun, M., You, Z. D. & Malpas, J. (1998).

Well-preserved garnet growth zoning in granulite from the

Dabie Mountains, central China. Journal of Metamorphic

Geology 16, 213–222.Chen, R. X., Zheng, Y. F., Gong, B., Zhao, Z. F., Gao, T. S., Chen,

B. & Wu, Y. B. (2007). Origin of retrograde fluid in

ultrahigh-pressure metamorphic rocks: constraints from

mineral hydrogen isotope and water content changes in

eclogite–gneiss transitions in the Sulu orogen. Geochimica

et Cosmochimica Acta 71, 2299–2325.


Dow






https://doi.org/10.2138/am-2018-6166

Chen, R. X., Zheng, Y. F. & Gong, B. (2011). Mineral hydrogen

isotopes and water contents in ultrahigh-pressure metaba-

site and metagranite: constraints on fluid flow during contin-

ental subduction-zone metamorphism. Chemical Geology

281, 103–124.Chen, R. X., Zheng, Y. F. & Hu, Z. C. (2012). Episodic fluid action

during exhumation of deeply subducted continental crust:

geochemical constraints from zoisite–quartz vein and host

metabasite in the Dabie orogen. Lithos 155, 146–166.Chen, Y. X., Zheng, Y. F. & Hu, Z. C. (2013). Synexhumation ana-

texis of ultrahigh-pressure metamorphic rocks: petrological

evidence from granitic gneiss in the Sulu orogen. Lithos

156–159, 69–96.Chen, Y. X., Zheng, Y. F., Gao, X. Y. & Hu, Z. C. (2014).

Multiphase solid inclusions in zoisite-bearing eclogite: evi-

dence for partial melting of ultrahigh-pressure metamorphic

rocks during continental collision. Lithos 200–201, 1–21.Chopin, C. (2003). Ultrahigh-pressure metamorphism: tracing

continental crust into the mantle. Earth and Planetary

Science Letters 212, 1–14.Clarke, G. L., Powell, R. & Fitzherbert, J. A. (2006). The lawsonite

paradox: a comparison of field evidence and mineral equilibria

modelling. Journal of Metamorphic Geology 24, 715–725.Coggon, R. & Holland, T. J. B. (2002). Mixing properties of phen-

gitic micas and revised garnet–phengite thermobarometers.

Journal of Metamorphic Geology 20, 683–696.Diener, J. F. A. & Powell, R. (2012). Revised activity–composi-

tion models for clinopyroxene and amphibole. Journal of

Metamorphic Geology 30, 131–142.Enami, M. & Nagasaki, A. (1999). Prograde P–T path of kyanite

eclogites from Junan in the Sulu ultrahigh-pressure prov-

ince, eastern China. Island Arc 8, 459–474.Ernst, W. G. & Liou, J. G. (2008). High- and ultrahigh-pressure

metamorphism: past results and future prospects. American

Mineralogist 93, 1771–1786.Ferrando, S., Frezzotti, M. L., Dallai, L. & Compagnoni, R. (2005).

Multiphase solid inclusions in UHP rocks (Su–Lu, China): rem-

nants of supercritical silicate-rich aqueous fluids released dur-

ing continental subduction. Chemical Geology 223, 68–81.Fitzsimons, I. C. W. & Harley, S. L. (1994). Garnet coronas in sca-

polite–wollastonite calc-silicates from East Antarctica: the

application and limitations of activity-corrected grids.

Journal of Metamorphic Geology 12, 761–777.Frost, B. R. & Chacko, T. (1989). The granulite uncertainty prin-

ciple: limitations on thermobarometry in granulites. Journal

of Geology 97, 435–450.Gerya, T. & Stockhert, B. S. (2006). Two-dimensional numerical

modeling of tectonic and metamorphic histories at active

continental margins. International Journal of Earth Sciences

95, 250–274.Green, E. C. R., White, R. W., Diener, J. F. A., Powell, R., Holland,

T. J. B. & Palin, R. M. (2016). Activity–composition relations

for the calculation of partial melting equilibria in metabasic

rocks. Journal of Metamorphic Geology 34, 845–869.Groppo, C., Beltrando, M. & Compagnoni, R. (2009). The P–T

path of the ultra-high pressure Lago Di Cignana and adjoin-

ing high-pressure meta-ophiolitic uniinsights into the evolu-

tion of the subducting Tethyan slab. Journal of

Metamorphic Geology 27, 207–231.Groppo, C., Rolfo, F., Liu, Y., Deng, L. & Wang, A. (2015). P–T

evolution of elusive UHP eclogites from the Luotian dome

(North Dabie Zone, China): how far can the thermodynamic

modeling lead us? Lithos 226, 183–200.Hack, A. C., Thompson, A. B. & Aerts, M. (2007). Phase relations

involving hydrous silicate melts, aqueous fluids, and minerals.

In: Liebscher, A. & Heinrich, C. A. (eds) Fluid–Fluid Interactions.

Mineralogical Society of America and Geochemical Society,

Reviews in Mineralogy and Geochemistry 65, 129–185.Hacker, B. R. (2006). Pressures and temperatures of

ultrahigh-pressure metamorphism: implications for UHP

tectonics and H2O in subducting slabs. InternationalGeology Review 48, 1053–1066.

Harris, N. B. W., Caddick, M., Kosler, J., Goswami, S., Vance, D.

& Tindle, A. G. (2004). The pressure–temperature–time pathof migmatites from the Sikkim Himalaya. Journal ofMetamorphic Geology 22, 249–264.

Hermann, J. (2003). Experimental evidence for diamond-faciesmetamorphism in the Dora–Maira massif. Lithos 70, 163–182.

Hermann, J. & Spandler, C. J. (2008). Sediment melts at sub-arc

depths: an experimental study. Journal of Petrology 49,717–740.

Hermann, J., Spandler, C., Hack, A. & Korsakov, A. V. (2006).

Aqueous fluids and hydrous melts in high-pressure andultra-high pressure rocks: implications for element transferin subduction zones. Lithos 92, 399–417.

Hermann, J., Zheng, Y. F. & Rubatto, D. (2013). Deep fluids insubducted continental crust. Elements 9, 281–287.

Holland, T. J. B. (1983). The experimental determination of activ-

ities in disordered and short-range ordered jadeitic pyrox-enes. Contributions to Mineralogy and Petrology 82, 214–220.

Holland, T. J. B. & Powell, R. (1998). An internally consistent

thermodynamic data set for phases of petrological interest.Journal of Metamorphic Geology 16, 309–343.

Holland, T. J. B. & Powell, R. (2001). Calculation of phase rela-tions involving haplogranitic melts using an internally con-

sistent thermodynamic dataset. Journal of Petrology 42,673–683.

Holland, T. J. B. & Powell, R. (2003). Activity–composition rela-

tions for phases in petrological calculations: an asymmetricmulticomponent formulation. Contributions to Mineralogyand Petrology 145, 492–501.

Holland, T. J. B. & Powell, R. (2011). An improved and extendedinternally consistent thermodynamic dataset for phases ofpetrological interest, involving a new equation of state for

solids. Journal of Metamorphic Geology 29, 333–383.Johnson, E. A. (2006). Water in nominally anhydrous crustal

minerals: speciation, concentration, and geologic signifi-

cance. In: Keppler, H. & Smyth, J. R. (eds) Water inNominally Anhydrous Minerals. Mineralogical Society ofAmerica and Geochemical Society, Reviews in Mineralogyand Geochemistry 62, 117–154.

Jones, K. A. & Brown, M. (1990). High temperature ‘clockwise’P–T paths and melting in the development of regional mig-matites: an example from southern Brittany, France. Journal

of Metamorphic Geology 8, 551–578.Kalt, A., Corfu, F. & Wijbrans, J. R. (2000). Time calibration of a

P–T path from a Variscan high-temperature low-pressure

metamorphic complex (Bayerische Wald, Germany), andthe detection of inherited monazite. Contributions toMineralogy and Petrology 138, 143–163.

Krogh Ravna, E. J. K. (2000). The garnet–clinopyroxeneFe2þ–Mg geothermometer: an updated calibration. Journalof Metamorphic Geology 18, 211–219.

Krogh Ravna, E. J. K. & Terry, M. P. (2004).Geothermobarometry of UHP and HP eclogites and schist-s—an evaluation of equilibria among garnet–clinopyroxe-ne–kyanite–phengite–coesite/quartz. Journal of

Metamorphic Geology 22, 579–592.Leake, B. E., Woolley, A. R., Arps, C. E. S., et al. (1997).

Nomenclature of amphiboles; report of the Subcommittee

on Amphiboles of the International MineralogicalAssociation, Commission on New Minerals and MineralNames. European Journal of Mineralogy 9, 623–1037.


Dow





Li, D. Y., Xiao, Y. L., Li, W. Y., Zhu, X., Williams, H. M. & Li, Y. L.

(2016a). Iron isotopic systematics of UHP eclogites respond

to oxidizing fluid during exhumation. Journal of

Metamorphic Geology 34, 987–997.Li, J. L., Klemd, R., Gao, J. & John, T. (2016b). Poly-cyclic meta-

morphic evolution of eclogite: evidence for multistage buria-

l–exhumation cycling in a subduction channel. Journal of

Petrology 57, 119–146.Li, W. C., Chen, R. X., Zheng, Y. F. & Hu, Z. C. (2014).

Dehydration and anatexis of UHP metagranite during con-

tinental collision in the Sulu orogen. Journal of

Metamorphic Geology 32, 915–936.Liou, J. G. & Zhang, R. Y. (1996). Occurrences of intergranular

coesite in ultrahigh-P rocks from the Sulu region, eastern

China: implications for lack of fluid during exhumation.

American Mineralogist 81, 1217–1227.Liou, J. G., Tsujimori, T., Zhang, R. Y., Katayama, I. &

Maruyama, S. (2004). Global UHP metamorphism and con-

tinental subduction/collision: the Himalayan model.

International Geology Review 46, 1–27.Liou, J. G., Ernst, W. G., Zhang, R. Y., Tsujimori, T. & Jahn, B. M.

(2009). Ultrahigh-pressure minerals and metamorphic ter-

ranes: the view from China. Journal of Asian Earth Sciences

35, 199–231.Liu, D. Y., Jian, P., Kroner, A. & Xu, S. T. (2006a). Dating of pro-

grade metamorphic events deciphered from episodic zircon

growth in rocks of the Dabie–Sulu UHP complex, China.

Earth and Planetary Science Letters 250, 650–666.Liu, F. L. & Liou, J. G. (2011). Zircon as the best mineral for P–T–time

history of UHP metamorphism: a review on mineral inclusions

and U–Pb SHRIMP ages of zircons from the Dabie–Sulu UHP

rocks. Journal of Asian Earth Sciences 40, 1–39.Liu, F. L., Gerdes, A., Liou, J. G., Xue, H. M. & Liang, F. H.

(2006b). SHRIMP U–Pb zircon dating from Sulu–Dabie dolo-

mitic marble, eastern China: constraints on prograde,

ultrahigh-pressure and retrograde metamorphic ages.

Journal of Metamorphic Geology 24, 569–589.Liu, J. B., Liu, W. Y., Ye, K. & Mao, Q. (2009a). Chlorine-rich

amphibole in Yangkou eclogite, Sulu ultrahigh-pressure

metamorphic terrane, China. European Journal of

Mineralogy 21, 1265–1285.Liu, Q., Jin, Z. & Zhang, J. (2009b). An experimental study of de-

hydration melting of phengite-bearing eclogite at

1�5–3�0 GPa. Chinese Science Bulletin 54, 2090–2100.Massonne, H. J. (2011). High-pressure/ultrahigh pressure rocks

in the Bohemian Massif; Occurrences and P–T conditions of

high and ultrahigh pressure rocks in the Bohemian Massif.

Geolines 23, 18–26.Massonne, H. J. (2012). Formation of amphibole and clinozoisi-

te–epidote in eclogite owing to fluid infiltration during ex-

humation in a subduction channel. Journal of Petrology 53,

1969–1998.Massonne, H. J. & Schreyer, W. (1987). Phengite geobarometry

based on the limiting assemblage with K-feldspar, phlogo-

pite, and quartz. Contributions to Mineralogy and Petrology

96, 212–224.Mibe, K., Kawamoto, T., Matsukage, K. N., Fei, Y. W. & Ono, S.

(2011). Slab melting versus slab dehydration in

subduction-zone magmatism. Proceedings of the National

Academy of Sciences of the USA 108, 8177–8182.Mosenfelder, J. L., Schertl, H., Smyth, J. R. & Liou, J. G. (2005).

Factors in the preservation of coesite: the importance of

fluid infiltration. American Mineralogist 90, 779–789.Nakamura, D. & Hirajima, T. (2000). Granulite-facies overprinting

of ultrahigh-pressure metamorphic rocks, northeastern Su–Lu

region, Eastern China. Journal of Petrology 41, 563–582.

Nakamura, D. & Hirajima, T. (2010). Significant cooling during

exhumation of UHP eclogite from the Taohang area in theSulu region of eastern China and its tectonic significance.

Island Arc 19, 707–717.O’Brien, P. J. (1997). Garnet zoning and reaction textures in

overprinted eclogites, Bohemian Massif, European

Variscides: a record of their thermal history during exhum-ation. Lithos 41, 119–133.

Palin, R. M., White, R. W., Green, E. C. R., Diener, J. F. A.,

Powell, R. & Holland, T. J. B. (2016). High-grade metamorph-ism and partial melting of mafic and intermediate rocks.Journal of Metamorphic Geology 34, 871–892.

Pattison, D. R. M. (2003). Petrogenetic significance oforthopyroxene-free garnet þ clinopyroxene þ plagioclase 6

quartz-bearing metabasites with respect to the amphibolite and

granulite facies. Journal of Metamorphic Geology 21, 21–34.Pattison, D. R. M. & Begin, N. J. (1994). Zoning patterns in ortho-

pyroxene and garnet in granulites: implications for geother-mometry. Journal of Metamorphic Geology 12, 387–410.

Powell, R. & Holland, T. J. B. (1988). An internally consistent

dataset with uncertainties and correlations: 3. Applicationsto geobarometry, worked examples and a computer pro-gram. Journal of Metamorphic Geology 6, 173–204.

Powell, R. & Holland, T. J. B. (1994). Optimal geothermometryand geobarometry. American Mineralogist 79, 120–133.

Powell, R. & Holland, T. J. B. (2008). On thermobarometry.

Journal of Metamorphic Geology 26, 155–179.Powell, R., Holland, T. J. B. & Worley, B. (1998). Calculating

phase diagrams involving solid solutions via non-linearequations, with examples using THERMOCALC. Journal ofMetamorphic Geology 16, 577–588.

Proyer, A., McCammon, C. & Dachs, E. (2004). Pitfalls in geo-thermobarometry of eclogites: Fe3þ and changes in the min-eral chemistry of omphacite at ultrahigh pressures.

Contributions to Mineralogy and Petrology 147, 305–318.Rebay, G., Powell, R. & Diener, J. F. A. (2010). Calculated phase

equilibria for a MORB composition in a P–T range,

450–650�C and 18–28 kbar: the stability of eclogite. Journalof Metamorphic Geology 28, 635–645.

Roda, M., Spalla, M. I. & Marotta, A. M. (2012). Integration ofnatural data within a numerical model of ablative subduc-tion: a possible interpretation for the Alpine dynamics of the

Austroalpine crust. Journal of Metamorphic Geology 30,973–996.

Schmid, R., Franz, L., Oberhansli, R. & Dong, S. (2000). High-Si

phengite, mineral chemistry and P–T evolution ofultra-high-pressure eclogites and calc-silicates from theDabie Shan, eastern China. Geological Journal 35, 185–207.

Schmidt, M. W., Poli, S., Comodi, P. & Zanazzi, P. F. (1997).High-pressure behavior of kyanite; decomposition of kyanite intostishovite and corundum. American Mineralogist 82, 460–466.

Schmidt, M. W., Dugnani, M. & Artioli, G. (2001). Synthesis andcharacterization of white micas in the join muscovite–alumi-

noceladonite. American Mineralogist 86, 555–565.Schorn, S. & Diener, J. F. A. (2017). Details of the

gabbro-to-eclogite transition determined from microtex-

tures and calculated chemical potential relationships.Journal of Metamorphic Geology 35, 55–75.

Schwandt, C. S., Cygan, R. T. & Westrich, H. R. (1996). Ca

self-diffusion in grossular garnet. American Mineralogist 81,448–451.

Sizova, E., Gerya, T. & Brown, M. (2012). Exhumation mechan-

ism of ultrahigh pressure crustal rocks during collision ofspontaneously moving plates. Journal of Metamorphic

Geology 30, 927–955.�Stıpska, P. & Powell, R. (2005). Constraining the P–T path of a

MORB-type eclogite using pseudosections, garnet zoning and


Dow





garnet–clinopyroxene thermometry: an example from theBohemian Massif. Journal of Metamorphic Geology 23, 725–743.

Syracuse, E. M., van Keken, P. E. & Abers, G. A. (2010). The glo-bal range of subduction zone thermal models. Physics of theEarth and Planetary Interiors 183, 73–90.

Wallis, S. R., Ishiwatari, A., Hirajima, T., Ye, K., Guo, J. H.,Nakamura, D., Kato, T., Zhai, M. G., Enami, M., Cong, B. L. &Banno, S. (1997). Occurrence and field relationships ofultrahigh-pressure metagranitoid and coesite eclogite in theSu–Lu terrane, eastern China. Journal of the GeologicalSociety, London 154, 45–54.

Wang, L., Kusky, T. M., Polat, A., Wang, S. J., Jiang, X. F., Zong,K. Q., Wang, J. P., Deng, H. & Fu, J. M. (2014). Partial meltingof deeply subducted eclogite from the Sulu orogen in China.Nature Communications 5, 5604.

Wang, L., Wang, S. J., Brown, M., Zhang, J. F., Feng, P. & Jin, Z.M. (2018). On the survival of intergranular coesite in UHPeclogite. Journal of Metamorphic Geology 36, 173–194.

Wang, Q. C., Ishiwatari, A., Zhao, Z. Y., Hirajima, T., Hiramatsu, N.,Enami, M., Zhai, M. G., Li, J. J. & Cong, B. L. (1993).Coesite-bearing granulite retrograded from eclogite in Weihai,eastern China. European Journal of Mineralogy 5, 141–152.

Wang, S. J., Wang, L., Brown, M. & Feng, P. (2016).Multi-stage barite crystallization in partially melted UHPeclogite from the Sulu belt, China. American Mineralogist101, 564–579.

Wang, S. J., Wang, L., Brown, M., Piccoli, P., Johnson, T. E.,Feng, P., Deng, H., Kitajima, K. & Huang, Y. (2017).Composite granite–quartz veins in the Sulu belt, China: evi-dence of supercritical fluid during exhumation of deeplysubducted UHP continental crust. Journal of MetamorphicGeology 35, 601–629, doi:10.1111/jmg.12248.

Warren, C. J., Beaumont, C. & Jamieson, R. A. (2008). Deep sub-duction and rapid exhumation: role of crustal strength andstrain weakening in continental subduction andultrahigh-pressure rock exhumation. Tectonics 27,CrossRef][10.1029/2008TC002292]

Wei, C. J. & Clarke, G. L. (2011). Calculated phase equilibria forMORB compositions: a reappraisal of the metamorphic evo-lution of lawsonite eclogite. Journal of MetamorphicGeology 29, 939–952.

Wei, C. J., Yang, Y., Su, X. L., Song, S. G. & Zhang, L. F. (2009).Metamorphic evolution of low-T eclogite from the NorthQilian orogen, NW China: evidence from petrology and cal-culated phase equilibria in the system NCKFMASHO.Journal of Metamorphic Geology 27, 55–70.

Wei, C. J., Qian, J. & Tian, Z. L. (2013). Metamorphic evolutionof medium-temperature ultra-high pressure (MT–UHP) eclo-gites from the South Dabie orogen, Central China: an insightfrom phase equilibria modelling. Journal of MetamorphicGeology 31, 755–774.

White, R. W., Powell, R. & Holland, T. J. B. (2007). Progressrelating to calculation of partial melting equilibria for meta-pelites. Journal of Metamorphic Geology 25, 511–527.

White, R. W., Powell, R., Holland, T. J. B., Johnson, T. E. &Green, E. C. R. (2014). New mineral activity–compositionrelations for thermodynamic calculations in metapelitic sys-tems. Journal of Metamorphic Geology 32, 261–286.

Whitney, D. L. & Evans, B. W. (2010). Abbreviations for names ofrock-forming minerals. American Mineralogist 95, 185–187.

Xia, Q. K., Sheng, Y. M., Yang, X. Z. & Yu, H. M. (2005).Heterogeneity of water in garnets from UHP eclogites, east-ern Dabieshan, China. Chemical Geology 224, 237–246.

Xu, S. T., Liu, Y. C., Chen, G. B., Compagnoni, R., Rolfo, F., He,M. C. & Liu, H. F. (2003). New finding of micro-diamonds in

eclogites from Dabie–Sulu region in central–eastern China.Chinese Science Bulletin 48, 988–994.

Xu, S. T., Liu, Y. C., Chen, G. B., Ji, S. Y., Ni, P. & Xiao, W. S. (2005).Microdiamonds, their classification and tectonic implicationsfor the host eclogites from the Dabie and Su–Lu regions in cen-tral eastern China. Mineralogical Magazine 69, 509.

Ye, K., Hirajima, T., Ishiwatari, A., Guo, J. H. & Zhai, M. G.(1996). Significance of interstitial coesite in eclogite fromYangkou, Qingdao city, eastern Shandong province.Chinese Science Bulletin 41, 1047–1048 [in Chinese].

Ye, K., Cong, B. & Ye, D. (2000). The possible subduction of con-tinental material to depths greater than 200 km. Nature 107,734–736.

Young, D. J. & Kylander-Clark, A. R. C. (2015). Does continentalcrust transform during eclogite facies metamorphism?Journal of Metamorphic Geology 33, 331–357.

Zeng, L. S., Liang, F. H., Asimow, P., Chen, F. Y. & Chen, J.(2009). Partial melting of deeply subducted continental crustand the formation of quartzofeldspathic polyphase inclu-sions in the Sulu UHP eclogites. Chinese Science Bulletin54, 2580–2594.

Zhang, R. Y. & Liou, J. G. (1997). Partial transformation of gabbroto coesite-bearing eclogite from Yangkou, the Sulu terrane,eastern China. Journal of Metamorphic Geology 15, 183–202.

Zhang, R. Y., Liou, J. G. & Cong, B. L. (1995). Talc-, magnesite-and Ti-clinohumite-bearing ultrahigh-pressure meta-maficand ultramafic complex in the Dabie Mountains, China.Journal of Petrology 36, 1011–1037.

Zhang, R. Y., Liou, J. G. & Ernst, W. G. (2009). The Dabie–Sulucontinental collision zone: a comprehensive review.Gondwana Research 16, 1–26.

Zhang, Z. M., Xiao, Y. L., Shen, K. & Gao, Y. J. (2005). Garnetgrowth compositional zonation and metamorphic P–T pathof the ultrahigh-pressure eclogites from the Sulu orogenicbelt, eastern central China. Acta Petrologica Sinica 21,809–818 [in Chinese with English abstract].

Zhang, Z. M., Xiao, Y. L., Hoefs, J., Liou, J. G. & Simon, K.(2006). Ultrahigh pressure metamorphic rocks from theChinese Continental Scientific Drilling Project: I. Petrologyand geochemistry of the main hole (0–2,050 m).Contributions to Mineralogy and Petrology 152, 421–441.

Zhang, Z. M., Yang, J. S., Rong, H., Hu, J. Z., Su, J. F. & Mao, H.K. (2007). Discovery of diamond in eclogite from the ChineseContinental Scientific Drilling Project Main Hole (CCSD–MH)in the Sulu UHPM belt. Acta Petrologica Sinica 23,3201–3206 [in Chinese with English abstract].

Zhang, Z. M., Shen, K., Sun, W. D., Liu, Y. S., Liou, J. G., Shi, C.& Wang, J. L. (2008). Fluids in deeply subducted continentalcrust: petrology, mineral chemistry and fluid inclusion ofUHP metamorphic veins from the Sulu orogen, easternChina. Geochimica et Cosmochimica Acta 72, 3200–3228.

Zhang, Z. M., Shen, K., Liou, J. G., Dong, X., Wang, W., Yu, F. &Liu, F. (2011). Fluid–rock interactions during UHP metamorph-ism: a review of the Dabie–Sulu orogen, east–central China.Journal of Asian Earth Sciences 42, 316–329.

Zheng, Y. F. (2009). Fluid regime in continental subduction zones:petrological insights from ultrahigh-pressure metamorphicrocks. Journal of the Geological Society, London 166, 763–782.

Zheng, Y. F., Xia, Q. X., Chen, R. X. & Gao, X. Y. (2011). Partialmelting, fluid supercriticality and element mobility inultrahigh-pressure metamorphic rocks during continentalcollision. Earth-Science Reviews 107, 342–374.

Zhu, Y. F., Massonne, H. J. & Theye, T. (2007). Eclogites from theChinese Continental Scientific Drilling borehole, their pet-rology and different P–T evolutions. Island Arc 16, 508–535.


Dow





10.1029/2008TC002292

Documents

Phase Equilibrium Modeling of MT–UHP Eclogite: a Case ...Yangkou Bay in the Sulu Belt of eastern China. The Sulu Belt is a classic example of deep continen-tal subduction and the