Diffusion-controlled metamorphic reaction textures ... - Ingenta Connect

2 downloads 0 Views 998KB Size Report
garnets developed a thick corona of epidote plus some quartz and calcite during ... Epidote in contact with quartz develops albite coronas as it decomposes into ...
Eur. J. Mineral. 2014, 26, 25–40 Published online January 2014

Diffusion-controlled metamorphic reaction textures in an ultrahigh-pressure impure calcite marble from Dabie Shan, China ALEXANDER PROYER1,*, FRANCO ROLFO2, DANIELE CASTELLI2 and ROBERTO COMPAGNONI2 1

Department of Earth Sciences, University of Graz, Universitaetsplatz 2/II, 8010 Graz, Austria *Corresponding author, e-mail: [email protected] Current address: Department of Geology, University of Botswana, Private Bag UB00704, Gaborone, Botswana 2 Department of Earth Sciences, University of Torino, Via Valperga Caluso 35, 10135 Torino, Italy

Abstract: The overall metamorphic reaction in an impure calcite marble from the ultra-high pressure (UHP) zone at Changpu, eastern Dabie Shan, China, is partitioned into local reaction domains defined by isolated grains of omphacite, garnet, epidote, quartz and ilmenite within the calcite matrix. These reaction domains witness different stages of the metamorphic evolution. Chemically homogeneous omphacite in some instances has partial or complete coronas of quartz which are interpreted as a prograde phenomenon, i.e. caused by replacement of a plagioclase precursor by omphacite with increasing pressure. Rare relics of talc in epidote indicate that the first breakdown reaction during exhumation was talc þ omphacite þ garnet ¼ amphibole þ epidote  quartz (still within the eclogite facies). During this early retrogression stage, omphacite was partially or completely rimmed by chemically heterogeneous amphibole. Most garnets developed a thick corona of epidote plus some quartz and calcite during this early retrogression or are completely pseudomorphed. Where garnet occurs near or next to omphacite, this corona consists partly or entirely of amphibole. The second retrograde stage is commonly characterized by the breakdown of omphacite þ quartz to albite þ diopside þ amphibole, often in the form of symplectite. Localized reactions become evident as omphacite at its site can decompose to three distinct types of symplectite (albite þ diopside, albite þ amphibole, albite þ diopside þ amphibole). Matrix quartz grains develop rims of individual diopside crystals at distances up to 2–3 millimetres from symplectites; additional outer rims of albite develop adjacent to garnet or corona epidote. Matrix quartz remote from omphacite sites is locally transformed into albite, where an Al-bearing mineral (mostly epidote) is found nearby. Epidote in contact with quartz develops albite coronas as it decomposes into albite þ minute grains of Cerich epidote (allanite). Primary epidote and corona epidote remote from any quartz develop a complex compositional zoning with enrichment of mainly Ce3þ at the outer rim, indicating exposure to a fluid undersaturated in alumina. Finally, large ilmenite grains in the matrix develop thick overgrowths of titanite, but in several instances an intermediate step of rutile formation is preserved. All these features indicate a direct correlation of textures with transport distances of dissolved chemical species in a non-pervasively infiltrating pore fluid. Simplified open-system reactions were derived for each site at each metamorphic stage and show that Al must have been quite mobile during prograde quartz corona formation, whereas both Si and Al were relatively immobile during the two retrograde stages – the relative sequence of mobility being Naþ . Mg2þ, Ca2þ, Fe2þ .. Si4þ . Al3þ, Ti4þ. Mass-balance between sites is consistent and no significant gain or loss of cations is indicated. Key-words: local reaction, diffusion-controlled, UHP-metamorphism, impure marble, symplectite, reaction corona.

1. Introduction

equilibration. The importance of these sub-reactions or partial reactions may vary locally as a function of kinetics factors, mainly the availability of or protection from fluids or partial melt. Reduced availability or even absence of fluid or melt will limit the supply of ions/species with low diffusion rates or long diffusion pathways at more remote reaction sites. In that case mineral assemblages and compositions will develop corresponding to thermodynamic equilibrium on a very small scale (mosaic equilibrium) which then may or may not develop towards a more mature stage (equilibration on a larger scale) depending on the continued or increased presence of fluid or melt. The

For a closed system, the overall reaction of a rock during metamorphism can be calculated by matrix operations from the compositions of the major minerals involved (Spear et al., 1982; Thompson, 1982a and 1982b; Fisher, 1989). The reaction textures of the rock, however, may not reflect that overall reaction directly. Reaction textures observed in rocks will rather correspond to localized reactions which are the kinetically most efficient partial reactions that may or may not add up to the theoretical total whole-rock reaction, depending on the degree of eschweizerbart_xxx

0935-1221/13/0025-2349 $ 7.20 DOI: 10.1127/0935-1221/2013/0025-2349

# 2013 E. Schweizerbart’sche Verlagsbuchhandlung, D-70176 Stuttgart

26

A. Proyer, F. Rolfo, D. Castelli, R. Compagnoni

combination of solubilities, diffusion coefficients (of species in a grain boundary fluid) and chemical potential gradients will determine the relative mobility and ‘‘reach’’ of each component. Limited supply of some components will result in mineral assemblages and mineral compositions that vary systematically with the distance between suppliers and recipients of components – a situation that may be described by the term ‘‘diffusion-controlled’’. In such a case the mineral assemblage forming at a reaction site is determined by the slowest diffusing components – thus giving a first idea about what components to hold constant when formulating an open system equation for a localized reaction. In well-equilibrated rocks the role of localized reactions is not easily recognized and in fact not relevant for most purposes. However, for anybody interested in deriving metamorphic-path information from partly equilibrated rocks containing relics of an earlier history in various stages of transformation, a good understanding of such sub-reactions and their characteristics may become quite important. Preserved diffusion-controlled reaction textures provide first-hand evidence about the relative mobility of components at metamorphic pressure (P) and temperature (T) conditions that can be used by experimentalists and theoretical modellers to devise their experiments and evaluate their results. More than fourty years ago, Dugald M. Carmichael (1969) presented some excellent examples where localized or partial reactions became obvious, like the transformation from kyanite- to sillimanite-bearing micaschists from the Whetstone Lake area in Canada, which did not proceed as a direct transformation but in a series of sub-reactions, with a mass-balancing exchange of ions in solution between the different reaction sites. Impure marbles present a particularly interesting subject because reactions amongst silicates and oxide minerals may become very localized and compartmented into subreactions to the degree that the individual silicate grains are separated from each other by the carbonate matrix. The study presented here demonstrates that the distance between individual grains of reaction partners and local variations in fluid permeation control the spectrum of textures and compositions actually formed. This variation of these parameters results in a wide spectrum of systematically and gradually differing reaction textures that has not been documented before. Even though impure marbles subjected to an ultra-high pressure (UHP) metamorphic evolution have received some attention, mainly in the Dabie-Sulu region of eastern China (Wang & Liou, 1993; Ye & Hirajima, 1996; Kato et al., 1997; Ogasawara et al., 1998; Tang et al., 2006) and the Kokchetav massif of Kazakhstan (e.g. Ogasawara et al., 2000; Katayama et al., 2002; Ogasawara et al., 2002; Sobolev et al., 2007), their potential to record and preserve relics of (ultra)high-pressure metamorphism, mainly because reaction partners are separated by a carbonate matrix that may impede the kinetics of retrograde reequilibration, has become more obvious only recently (cf. Proyer et al., 2013, and references therein). The present study also

shows how eclogite-facies minerals may remain well preserved, even though coesite relics or thermobarometric evidence for UHP conditions have not been found.

2. Sample petrography Sample RPC 598 was collected in the surroundings of Changpu (N30 420 43.100 , E116 150 19.400 , Fig. 1) in the UHP zone of Dabie Shan, east central China (Schmid et al., 2003). The roadcut outcrop exposes decimetre- to metre-thick eclogite boudins, sulphide-bearing silicate fels and jadeite-bearing fels embedded in pure marbles that only occasionally display millimetre-thick bands containing micas or garnet plus pyroxene. RPC 598 is a sample of impure calcite marble from a band of the latter type (Fig. 1d). Altogether, these lithologies belong to a large coherent unit, which, as a whole, underwent a UHP metamorphic event. The metamorphic history of this region is well established, with peak metamorphic conditions of 3.3 GPa and 720  C inferred by Rolfo et al. (2004). On hand specimen, the impure marble is whitish and consists of up to mm-sized calcite grains in which reddish and greenish grains of garnet and omphacite appear both isolated from each other and aggregated in small blebs. The petrographic microscope reveals additional larger grains of colourless quartz as well as rims or coronas around these three primary minerals that will be described in detail below, and symplectite after some omphacite. Apatite, rutile, titanite, ilmenite, chalcopyrite and pyrrhotite are the most prominent accessory minerals. On a microstructural basis, the oldest mineral assemblage preserved consists of garnet, omphacite, quartz (after coesite), rutile, REE-enriched epidote, calcite (prior aragonite) and perhaps talc, which was found as a rare tiny inclusion in epidote. All these minerals except talc and epidote are chemically homogeneous, coarse-grained and most often separated from each other by the carbonate matrix. Epidote is commonly internally zoned and, where not entirely included in another mineral, tends to develop rims that are enriched in Fe and REE, mainly Ce (Fig. 2a) and so brighter in backscattered-electron (BSE) imaging. Quartz forms individual grains and also partial or complete rims around omphacite. Three examples of such rims are shown in Fig. 2b–d. Rutile is preserved as intergrowths with matrix ilmenite, mantled by titanite, or as inclusions within garnet where it is often marginally replaced by titanite which may show a mesh texture between different rutile sites and towards garnet grain boundaries, most likely following pre-existing cracks in garnet (Fig. 2e). Ilmenite commonly forms large opaque grains, both in the matrix and as inclusions in omphacite, and is invariably replaced at its rims by titanite (Fig. 2f). The age relationship between rutile and ilmenite is not clear as the textures are ambivalent (Fig. 3a). Ilmenite may be an even older relic of a pre-eclogite stage that was partly replaced by rutile in the eclogite facies, with most of this rutile transformed later into titanite. eschweizerbart_xxx

Diffusion-controlled metamorphic reaction textures

27

Fig. 1. A) Simplified tectono-metamorphic sketch map of the Dabie-Tongbai-Qinling belt. Insert: location of the belt in Eastern China. B) Geologic sketch map of east-central Dabie Shan (after Rolfo et al., 2000). 1, Yangtze craton basement; 2, granitic gneiss; 3, Dabie basement gneiss; 4, WPU Unit (Rolfo et al., 2004); 5a, jadeite-bearing granofels; 5b, marble; 6, Mesozoic granitoids. C) Detailed geological setting of the areas around Changpu (after Qian & Huang, 2000, modified). Symbols: jd, jadeite-bearing granofels (purple); m, marble (blue); ec, fresh and retrogressed eclogite (green); aa ¼ alluvium; og ¼ oligoclase-bearing granitic gneiss; gg ¼ monzo-granitic gneiss; dg ¼ granodioritic gneiss; bg ¼ biotite-plagioclase gneiss; mp ¼ muscovite-plagioclase gneiss; mg ¼ two micas – plagioclase gneiss; gb ¼ garnet-biotiteplagioclase gneiss; gs ¼ garnet-muscovite schist; sp ¼ serpentinite; ts ¼ talc schist; vb ¼ volcanic breccia. D) Typical appearance of marbles at Changpu. Decimetre to metre-thick boudins of eclogite (E) are enveloped by the planar fabric of the marble (M) locally underlined by millimetre-thick bands (B) containing micas and/or garnet þ pyroxene. eschweizerbart_xxx

28

A. Proyer, F. Rolfo, D. Castelli, R. Compagnoni

Fig. 2. Representative SEM images of the studied sample microstructures. a) Zoned epidote in the matrix has a thin BSE-bright rim; b) omphacite with a partial quartz corona; c) full quartz corona around omphacite with small crystals of amphibole and diopside at the interface; d) fine-grained retrogression products of omphacite within a quartz corona; e) rutile inclusions in garnet, partly replaced by titanite; f) ilmenite inclusion in former omphacite, rimmed by titanite.

Both garnet and omphacite developed coronas during subsequent metamorphic stages and display various degrees of replacement. In the simplest case, garnet acquired a rim of epidote plus calcite and quartz, and omphacite a rim of amphibole (Fig. 3b, c), even though omphacite sufficiently protected by the carbonate matrix or by a quartz-corona may show only very limited replacement (Fig. 2b, c). Such simple coronas develop where garnet and omphacite are separated from each other by at least several hundred micrometres of calcite matrix. Where distances are smaller, more complex types of coronas developed around garnet: amphibole occurs in the upper part of the corona in Fig. 3d and more evenly distributed through a mixed corona of clinozoisite þ amphibole þ calcite þ quartz around a former garnet-omphacite

intergrowth (Fig. 3e). In some instances, also close to omphacite, garnet is rimmed by amphibole only (Fig. 3f). Omphacite inside such relatively coarse-grained coronas or omphacite not affected by the first type of replacement commonly shows partial to complete decay to symplectite with a surprisingly great variety of products. Some symplectites consist entirely of albite and diopside (Fig. 4a), and larger single actinolite crystals are found only along the outer margin. Other symplectites consist entirely of albite and amphibole (Fig. 4b), the latter being considerably aluminous. Finally a mixed type of symplectite occurs, consisting of albite-rich plagioclase, amphibole and diopside, the latter two minerals being almost indistinguishable in grain size and BSE contrast (Fig. 4c–e). However, amphibole sometimes coarsens along eschweizerbart_xxx

Diffusion-controlled metamorphic reaction textures

29

Fig. 3. Scanning electron microscope images of the studied microstructures. a) Ilmenite-rutile-titanite intergrowth in calcite matrix, b) garnet rimmed by a corona of epidote þ calcite þ quartz; c) omphacite and quartz rimmed by compositionally heterogeneous amphibole; d) corona around garnet is amphibole-bearing at the upper end; e) former garnet-omphacite intergrowth replaced by a mixed corona; f) garnet with an omphacite corona near symplectite after omphacite.

symplectite margins and along some transects which may correspond to former fluid pathways such as cracks, subgrain boundaries or intracrystalline strain zones. Amphibole-bearing symplectites commonly contain additional calcite. Symplectitization was neither pervasive nor very intense, as demonstrated by the preservation of omphacite relics near or even within symplectites (Fig. 4c, d). Napoor clinopyroxenes in symplectites have an XFe similar to that of the original omphacite (Table 1). Clinopyroxene of similar composition grows as almost continuous rims along the grain boundaries of nearby matrix quartz up to a distance of 1–2 millimetres from symplectite sites. Single

but larger actinolite crystals are found sporadically as an outer part of this clinopyroxene rim, which becomes thinner and finer grained with distance from the decomposing omphacite (Fig. 4f). At grain boundaries between quartz and an Al-rich mineral like garnet or epidote, a second rim of albite develops between the clinopyroxene rim around quartz and the Al-mineral (Fig. 4g). Amphibole is absent in such settings, and the clinopyroxenes are significantlty higher in XFe (around 0.45). Symplectites of amphibole þ diopside þ quartz (Fig. 4h) are exceptional and perhaps developed during the earlier retrograde stage (corona stage), before albite became stable. No obvious changes occur at former garnet sites

eschweizerbart_xxx

30

A. Proyer, F. Rolfo, D. Castelli, R. Compagnoni

Fig. 4. Scanning electron microscope images of the studied microstructures. a) Albite-diopside symplectite after omphacite with larger actinolite grains at the outer margin and a thin rim of diopside around adjacent quartz; b) albite-amphibole symplectite adjacent to matrix quartz that is mainly rimmed by abundant actinolite and subordinate diopside; c) close proximity of symplectite and omphacite preserved during the symplectitization stage; d) partly symplectized omphacite within an amphibole corona; symplectite consists of amphibole þ diopside þ albite; e) symplectite after omphacite near garnet; f) decreasing amount of amphibole rim around quartz with increasing distance from the site of former omphacite; g) a double rim of amphibole þ albite between quartz and garnet with epidote corona; the amphibole rim ends but the albite rim continues further; h) amphibole-diopside-quartz symplectite replacing omphacite inside an amphibole corona.

eschweizerbart_xxx

54.22 b.d. 4.87 8.78 b.d. 10.69 19.05 2.82 b.d. n.a.

56.81 0.14 10.51 5.58 b.d. 8.22 12.69 6.99 b.d. n.a.

100.94 2.00 0.00 0.44 0.03 0.14 0.00 0.43 0.48 0.48 0.00 0.00 4.00 0.24 1.09 4.59

SiO2 TiO2 Al2O3 FeO MnO MgO CaO Na2O K2O F

Total Si Ti Al Fe3þ Fe2þ Mn Mg Ca Na K F SumCat X(Fe) Na/Al Si/Al

eschweizerbart_xxx

100.79 1.99 0.00 0.27 0.00 0.26 0.00 0.53 0.70 0.25 0.00 0.00 4.00 0.33 0.91 7.41

54.79 b.d. 6.28 8.68 b.d. 9.67 17.88 3.49 b.d. n.a.

cpx grt cor

99.80 2.01 0.00 0.07 0.01 0.23 0.00 0.74 0.86 0.09 0.00 0.00 4.00 0.23 1.36 30.41

54.25 b.d. 1.52 7.64 b.d. 13.36 21.77 1.26 b.d. n.a.

cpx sympl

98.80 1.99 0.00 0.07 0.10 0.30 0.00 0.54 0.87 0.11 0.00 0.00 3.98 0.35 1.64 28.87

52.29 b.d. 1.54 12.39 0.06 9.57 21.27 1.53 b.d. n.a.

cpx qtz rim

99.22 2.01 0.00 0.03 0.02 0.42 0.00 0.55 0.89 0.08 0.00 0.00 4.00 0.44 2.45 64.97

52.53 b.d. 0.69 13.69 b.d. 9.61 21.67 1.02 b.d. n.a.

cpx qtz-grt

97.63 6.94 0.00 1.57 0.35 1.37 0.00 2.88 1.70 0.60 0.00 0.16 15.40 0.32 0.38 4.42

47.89 b.d. 9.19 14.13 b.d. 13.32 10.97 2.13 b.d. 0.34

amp 1 omp cor

97.33 6.57 0.06 2.11 0.16 1.71 0.00 2.44 1.77 0.83 0.00 0.00 15.65 0.41 0.40 3.12

44.44 0.52 12.08 15.15 b.d. 11.05 11.19 2.90 b.d. b.d.

amp 2 omp cor

98.40 7.73 0.00 0.41 0.17 1.20 0.00 3.54 1.85 0.17 0.00 0.00 15.07 0.25 0.41 18.95

54.73 b.d. 2.45 11.56 b.d. 16.83 12.22 0.61 b.d. b.d.

amp 3 omp cor

97.68 6.19 0.00 2.81 0.40 1.94 0.00 1.78 1.65 0.88 0.00 0.00 15.64 0.52 0.31 2.20

41.55 b.d. 16.00 18.78 b.d. 7.99 10.33 3.03 b.d. b.d.

amp grt cor

cor ¼ corona; sympl ¼ from symplectite; mx ¼ matrix; b.d. ¼ below detection limit; n.a. ¼ not analysed; SumCat ¼ sum of cations.

100.43 1.99 0.00 0.21 0.02 0.25 0.00 0.58 0.75 0.20 0.00 0.00 4.00 0.30 0.95 9.47

cpx omp rim

omp fresh

Mineral Site

Table 1. Representative analyses of clinopyroxenes, amphiboles and plagioclase from sample RPC598.

96.69 7.93 0.00 0.11 0.05 1.78 0.00 3.10 1.97 0.07 0.01 0.12 15.02 0.36 0.67 74.85

53.96 0.04 0.61 14.87 b.d. 14.14 12.49 0.25 0.04 0.25

amp qtz rim

98.78 7.64 0.00 0.66 0.21 1.10 0.00 3.46 1.72 0.30 0.00 0.16 15.08 0.24 0.46 11.53

54.50 b.d. 4.01 11.15 b.d. 16.54 11.47 1.11 0.00 0.36

act sympl

96.63 6.71 0.00 1.87 0.28 1.61 0.00 2.61 1.89 0.69 0.00 0.24 15.66 0.38 0.37 3.58

44.89 b.d. 10.63 15.09 b.d. 11.73 11.90 2.39 b.d. 0.50

edenite sympl

99.80 2.98 0.00 1.03 0.00 0.00 0.00 0.00 0.02 0.95 0.00 0.00 4.98

68.08 n.a. 19.92 b.d. b.d. b.d. 0.47 11.25 0.08 n.a.

plag 1 sympl

100.27 2.75 0.00 1.25 0.00 0.00 0.00 0.00 0.25 0.75 0.00 0.00 4.99

62.26 n.a. 24.01 b.d. b.d. b.d. 5.18 8.82 b.d. n.a.

plag 2 sympl

Diffusion-controlled metamorphic reaction textures 31

grt core

38.74 b.d. 22.01 0.00 23.28 0.52 3.90 12.12 0.00 100.57

3.00 0.00 2.01 0.00 1.51 0.03 0.45 1.01 0.00 8.00

Mineral Site

SiO2 TiO2 Al2O3 Fe2O3 FeO MnO MgO CaO F Total

Si Ti Al Fe3þ Fe2þ Mn Mg Ca F SumCat

eschweizerbart_xxx

0.00 1.00 0.00 0.00 0.98 0.02 0.00 0.00 0.00 2.00

0.00 53.44 0.00 0.00 46.91 0.76 b.d. 0.12 0.00 101.23

ilm core

1.02 0.88 0.13 0.01 0.00 0.00 0.00 0.99 0.13 3.03

31.38 35.93 3.27 0.22 0.00 b.d. b.d. 28.33 1.28 100.41

titanite core

Table 1. cont.: Representative analyses of garnet, ilmenite, titanite, epidote

1.01 0.98 0.02 0.02 0.00 0.00 0.00 0.96 0.05 2.99

31.10 40.26 0.59 0.76 0.00 b.d. b.d. 27.46 0.46 100.63

titanite core

3.04 0.00 2.61 0.43 0.00 0.00 0.00 1.87 0.00 7.95

38.95 b.d. 28.44 7.28 0.00 b.d. b.d. 22.38 0.00 97.05

epidote grt cor

3.09 0.00 2.42 0.56 0.00 0.00 0.00 1.83 0.00 7.91

34.77 b.d. 23.12 8.40 0.00 b.d. b.d. 19.22 0.00 85.51

epidote mx core

3.07 0.00 2.50 0.48 0.00 0.00 0.00 1.91 0.08 7.95

37.43 b.d. 25.88 7.74 0.00 b.d. b.d. 21.70 0.29 93.04

epidote mx rim

32 A. Proyer, F. Rolfo, D. Castelli, R. Compagnoni

Diffusion-controlled metamorphic reaction textures

33

Fig. 5. Scanning electron microscope images of the studied microstructures. a) Albitized quartz: porous parts are albite; b–d) albitized quartz and epidote: replaced parts of epidote are decorated by bright spots of Ce-rich epidote; e) X-ray distribution image for Si (same region as in 5b): quartz (white), albite (medium grey), epidote (dark grey) and calcite (black); f) BSE-image detail of 5b/e; g) pseudomorphs after garnet: quartz is not albitized but epidote is marginally enriched in REE; h) partial pseudomorphs after garnet: quartz is partly replaced by plagioclase in its upper and left portion.

during symplectitization, with the possible exception of thin BSE-bright rims at epidote grain margins, which is a general, site-independent phenomenon.

Finally, quartz grains remote from omphacite or garnet can be partially transformed to albite (‘‘albitization’’). This phenomenon is difficult to detect during scanning electron eschweizerbart_xxx

34

A. Proyer, F. Rolfo, D. Castelli, R. Compagnoni

100 s and beam diameter 1–2 mm. The following natural and synthetic standards were used: titanite (Ca, Ti), rhodonite (Mn), garnet (Mg, Fe, Al, Si), F-phlogopite (F), adularia (K) and jadeite (Na). Representative chemical compositions of minerals are given in Table 1. Fresh omphacite is relatively homogeneous in composition. The individual grains are unzoned and have the same composition throughout the sample (ca. Jd44Di42Hed14), even as inclusions in garnet. Relict omphacite in coronas around garnet and retrograde clinopyroxene along omphacite grain margins and in symplectites are increasingly diopside-rich respectively. Almost continuous inner rims around quartz grains in the vicinity of symplectized omphacite sites also consist of Na-poor clinopyroxenes (Fig. 6). The Fe2þ/ (Fe2þþMg2þ) ratio of retrograde clinopyroxene is similar to that of original omphacite (0.30–0.35); only diopside in symplectites can have XFe ratios as low as 0.25 and diopside near garnet lies around 0.45. Garnet grains are compositionally homogeneous (ca. Alm50Gro38Pyr12) throughout the sample, similar to ilmenite, which has almost end-member composition (Table 1). Titanite has XF between 0.05 and 0.17 and XAl between 0.02 and 0.13 and apparently contains small amounts of Fe3þ. Plagioclase is commonly albite (XAn , 0.10) but may occasionally range in composition up to XAn ¼ 0.26 in both symplectite and ‘‘albitized’’ epidote. According to Leake et al. (1997) all amphiboles are calcic (Na in the B-position is less than 0.5 a.p.f.u.). Those from coronas around garnet are Al- and Fe-rich (pargasites, one edenite), those in coronas around omphacite and in symplectites cover a similar composition range from pargasite or magnesiohornblende via edenite to actinolite at the Al-poor end of the spectrum (Fig. 7). Larger amphibole crystals in contact with quartz and symplectized omphacite are generally actinolites. As epidote has not been analysed for REE, the analytical totals are often below ideal values in Table 1, even for those parts of grains with the lowest concentrations of REE and Fe3þ (darkest BSE contrast). Only the epidote from garnet coronas, which is very similar in composition to the outer zone of matrix epidote and was

microscope (SEM) imaging because the BSE-contrast of quartz and albite is almost identical. However, albitized domains are often characterized by a porous texture and in about 30 % of the cases the boundary between quartz and albite is outlined by micro- to submicrometre-sized particles of very REE-rich epidote (Fig. 5a, b). In some of these cases, relics of actual clinozoisite/epidote are preserved inside these allanite-delimited zones, indicating that albite also replaced former epidote, preferably next to quartz (Fig. 5b–d). Detailed investigation of several such reaction sites reveals that albitization is restricted to areas where matrix epidote occurs in direct contact with or in close vicinity to quartz, as illustrated by the characteristic X-ray map for Si and BSE-image in Fig. 5e, f. These epidotes grow a rim of albite and the fine grained, BSE-bright Ceepidote spots described above are considered to outline pre-albitization grain boundaries between epidote and quartz. Epidote-calcite-quartz pseudomorphs after garnet can be unaltered (Fig. 5g) or albitized, too (Fig. 5h).

3. Mineral compositions Chemical analyses of the minerals were carried out with a JEOL JSM-6310 scanning electron microprobe equipped with a LINK ISIS energy dispersive system and a MICROSPEC wave length dispersive system (for Na and F) at Institute of Earth Sciences, Karl-Franzens University of Graz, Austria. The operating conditions were: acceleration voltage 15 kV, beam current 6 nA, live counting time

Fig. 6. Clinopyroxene compositions plot in the upper left section of the triangle jadeite–Q(diopside þ hedenbergite)–aegirine. Jadeite content decreases from fresh omphacite (dots) via cpx in garnet coronas (crosses) and omphacite coronas (open squares) to diopside in symplectite (filled triangles).

Fig. 7. Compositions of amphiboles from omphacite coronas (diamonds), garnet coronas (crosses), quartz rims (open squares) and symplectites (circles), according to Leake et al. (1997). eschweizerbart_xxx

Diffusion-controlled metamorphic reaction textures

most likely coeval in growth, contains no significant amounts of REE and 0.43–0.48 a.p.f.u. Fe3þ (based on eight cations per formula unit).

35

have decided not to use dissolved halide salts or associated neutral species like NaCl0(aq) for mass and charge balancing just as a matter of convenience. Changes in speciation and pH will occur if halide anions are used instead of (OH)–, as illustrated in one example below, but the fundamental transport necessities of cations to and from the reaction sites remain unchanged. Based on the above, we derived our open-system reaction equations as follows: From textural and compositional information we conclude on relative mobilities of components. Reaction equations are then formulated as conservative for the least mobile components. All solid phases present at a reaction site are expressed in mineral formulae. Mass and charge balance is then attained using those fluid species that are considered both dominant and most mobile. We derive (sometimes step by step) the equation that explains best the observed textures and mineral compositions and then check if there is a reasonable degree of mass balance between sites (where more than one site is known for a stage).

4. Petrological interpretations The sequence of replacement events is considered to reflect at least three different stages: 1) a prograde stage when albite-rich plagioclase was replaced by omphacite that partly developed a quartz corona; 2) a first retrograde stage during which the peak eclogite-facies assemblage partly transformed into a relatively coarse grained lowerpressure eclogite-facies assemblage characterized by growth of amphibole and epidote, but absence of plagioclase; and 3) the stage of symplectite formation from omphacite, perhaps coeval with the albitization of epidote þ quartz in the matrix. A detailed interpretation of these stages is given below. Open-system reactions are used to describe the replacement processes at each site (‘‘site’’ meaning an isolated grain of albite, quartz, garnet or omphacite). As we have insufficient information to derive the P-T conditions of the stages, as thermodynamic data about speciation in the fluid at these elevated P-T conditions (within or just below the eclogite facies) are scarce to absent and as thermodynamic treatment of complex system requires a lot of assumptions even at low P-T conditions (e.g. Bach & Klein, 2009) we do not attempt to calculate speciations nor to formulate thermodynamic equations. A physico-chemical description of the formation of reaction coronas as given for instance by Joesten (1977) is not possible due to the same reasons and because in this study the reaction geometry is much more complex (no adjacent minerals between which coronas develop but variable distances between sites in three dimensions and unknown fluid pathways and transport properties). Therefore we try to approximate the chemical changes at each site by simple, generalized open-system reaction equations in the style of Carmichael (1969): for simplicity, we use the mono- and divalent cations (Naþ, Mg2þ, Fe2þ, Ca2þ) and neutral associated complexes for Al and Si (Al(OH)3, Si(OH)4) as representatives for all the species of that cation potentially present at the conditions of metamorphism. It is clear, however, that the reaction stoichiometries would change as a function of the dominant species (and speciation in general) at the P-T-X conditions of interest. There is considerable latitude on how open-system equations can be formulated if there are no constraints from thermodynamics and/or textural and compositional observations, and the apparent element mobilities indicated by such equations can differ significantly. Not only speciation, but also the local pH or redox conditions, as well as the importance of halide anions (mainly Cl– and F–) are largely unknown, but as will be shown below, we take the indications we have from textural and mineral chemical observations to narrow down the possible choices. We

4.1. Quartz coronas around omphacite Most omphacite grains occur as single crystals isolated in a calcite matrix. A small number of these, which are commonly at a distance of at least a few millimetres from other silicates or oxides, have developed a partial or complete corona of quartz. The occurrence of such quartz coronas around omphacite has an apparently simple explanation as the prograde decay of albite produces jadeite þ quartz, and thus might be the result of isolated albite grains in the marble becoming unstable during prograde pressure increase. However, the situation changes if the clinopyroxene contains a significant amount of diopside component, as in the studied sample where the compositionally homogeneous omphacites have an Xjd of 0.44. An expanded albite breakdown reaction can be written for this case, which involves the transport of ionic species to and from the albite reaction site: ð1  nÞ NaAlSi3 O8 þ n Mg2þ þ n Ca2þ þ 4nðOHÞ ¼ albite ðNa1n Can ÞðAl1n Mgn ÞSi2 O6 þ ð1  3nÞ SiO2 þ 2n H2 O: omphacite

quartz (1)

where n is always a mole fraction, i.e. 0 , n , 1. This equation has a positive reaction coefficient for SiO2 if n , 1/3, which corresponds to Xjd . 0.67 in omphacite. The measured Xjd ¼ 0.44 in sample RPC598 cannot be explained this way, so an additional supply of SiO2 and/ or loss of Na and Al from the albite site are required. Local supply of Ca from decomposing CaCO3 (at the time of reaction most likely aragonite) does not change equation (1) significantly:

eschweizerbart_xxx

36

A. Proyer, F. Rolfo, D. Castelli, R. Compagnoni

corona around garnet consists of epidote þ quartz þ calcite and indicates that Fe and Mg must have been mobile whereas Al, Si and Ca perhaps remained at the reaction site. Corona formation can be explained schematically by the reaction

ð1  nÞ NaAlSi3 O8 þ n Mg2þ þ 2n CaCO3 ¼ ðNa1n Can ÞðAl1n Mgn ÞSi2 O6 þ ð1 3nÞSiO2 þ 2n CO2 þ n Ca2þ :

ð1aÞ

but the coronas could be explained in this case by the fact that reaction (1a) leads to a volume decrease because CO2 leaves the reaction site. In such a case quartz precipitation might be an effective means to fill the nascent void. A reaction equation for actual loss of Na and Al from the reaction site can be balanced for Si and Ca preservation (both represented by local solids). Reaction coefficients in terms of ‘‘n’’ become a bit cumbersome, but clearly show that quartz is a precipitate for all Xjd . 1/3:

3 Ca3 Al2 Si3 O12 þ H2 O þ 5 CO2 ¼ grossular 2 Ca2 Al3 Si3 O12 ðOHÞ þ 5 CaCO3 þ 3 SiO2 : clinozoisite

¼ ðNa1n Can ÞðAl1n Mgn ÞSi2 O6 þ ð1  3n=2ÞSiO2

1:5 grt þ 2:79 Ca2þ þ 0:5 H2 O þ 2:5 CO2

ð1bÞ

¼ 1 czo þ 1:5 qtz þ 2:5 cal þ 0:54 Mg2þ þ 2:25 Fe2þ :

In other words, Al must be quite mobile for quartz coronas to form. It may have been consumed (together with Na) at other sites where omphacite perhaps formed from (Ca-)Mg-silicates, like amphibole. Finally, if both Mg2þ and Ca2þ are supplied from afar, i.e. there is no local participation of calcite in the reaction, the very simple reaction equation for this case implies that omphacites of any composition could form together with quartz, but Al mobility is still a requirement:

(2a) This reaction reproduces the observed texture and shows that additional Ca2þ is transported to the garnet corona site in exchange for Fe2þ and Mg2þ. Amphibole coronas around omphacite seem to be contemporaneous and indicate loss of Ca (the Ca/Mg ratio in omphacite is higher than in the amphiboles). Starting again with a simplified model reaction, this would be:

NaAlSi3 O8 þ n Mg2þ þ n Ca2þ ¼ albite

4 CaMgSi2 O6 þ Mg2þ þ H2 O þ CO2 ¼ diopside

ðNa1n Can ÞðAl1n Mgn ÞSi2 O6 þ SiO2 þ n Al3þ þ n Naþ : omphacite

quartz

As this reaction only considers the grossular endmember of garnet, another reaction including Fe and Mg can be written for an average garnet composition of ca. Alm50Gro38Pyr12 and the condition of Al-conservation (assuming Al to be the least mobile component):

ð1  n=2Þ NaAlSi3 O8 þ n Mg2þ þ n CaCO3 þ n CO2 þ n=2 Al3þ þ n=2 Naþ :

calcite

(2)

Ca2 Mg5 Si8 O22 ðOHÞ2 þCaCO3 þ Ca2þ :

quartz

(3)

tremolite

(1c)

Hence, the marginal replacement of omphacite by amphiboles at this stage apparently just compensates the ionic budget at the garnet site. As the composition of amphiboles growing at omphacite rims during this stage can be quite variable in terms of Na-, Al- and Fe-content (Fig. 7), the stoichiometry of a chemically more complex reaction depends on the amphibole composition chosen. As for general trends, the Na/Al ratio of these amphiboles ranges from 0.32 to 0.68 (most scatter between 0.35 and 0.40), indicating that Na from decomposing jadeite component was preferentially partitioned into the fluid and partly transported elsewhere. XFe of amphiboles is generally higher (0.25–0.44) than that of omphacite (0.26  0.03), which is consistent with the fact that Fe2þ dominates over Mg2þ in the divalent cations released from decomposing garnet during its corona-formation. Hence, a generalized type of equation can be written for the omphacite site as:

The amount of quartz forming (partial) rims around omphacite is quite variable and is consistent with a process characterized by reactions (1b) and (1c). Omphacites without a quartz corona would develop when any of the limiting n-values mentioned above was overstepped or where omphacites grew at the site of another precursor mineral like amphibole. As there are no preserved hints as to what the other reaction sites in the rock could have been, we do not attempt to mass balance the ionic transport between sites for this case.

4.2. Coronas around garnet and omphacite Apparently, the assemblage omphacite þ garnet þ quartz/ coesite þ epidote (rich in Fe3þ and REE) þ rutile  talc was passive during the metamorphic evolution through the eclogite facies, perhaps to UHP conditions and back, as indicated by P-T conditions derived from the same field area (Schmid et al., 2003; Rolfo et al., 2004). Exhumation was then accompanied by at least two separate stages of chemical reaction, the first one represented by corona formation around garnet and omphacite. The

omp þ H2 O þ Fe2þ þ Mg2þ ¼ amp þ Naþ þ Ca2þ : The Si/Al-ratio of amphiboles is either similar to the value in omphacite (~4.7) or lower and only rarely higher (actinolites). This is a considerable spread, but at least the lower values can be explained by the occasional presence of eschweizerbart_xxx

Diffusion-controlled metamorphic reaction textures

additional diopside or quartz (Fig. 4d). Apparently the variety of amphibole compositions caused by the low solubility and/or diffusivity of Al3þ and Si4þ can become quite complex. The coronae around garnet and omphacite can merge when the two minerals occur close to each other, and perhaps where they are connected by preferred fluid pathways. In such cases amphibole becomes part of the garnet coronas. Such ‘‘mixed’’ epidote-amphibole coronas correspond more closely to what one normally sees in rocks with less distinct single reaction site types, as in eclogites. Coarse-grained blasts of amphibole and epidote and somewhat finer-grained epidote and (slightly Al- and Fe-richer) amphibole at garnet grain boundaries are typical for eclogites in the low-pressure eclogite facies, near the transition to the amphibolite facies.

37

time-integrated availability of dissolved species from the fluid. The first step of omphacite decay in an isolated site can be formulated as: 2 ðNaCaÞðAl; MgÞSi4 O12 þ 2 H2 O ¼ 2 NaAlSi3 O8 þ CaMgSi2 O6 þ Mg2þ 2þ

þ Ca

(4)



þ 4 ðOHÞ :

The following alternative equation uses Cl– instead of (OH): 2 ðNaCaÞðAl; MgÞSi4 O12 þ 4 H3 Oþ þ 4 Cl (40 )

¼ 2 NaAlSi3 O8 þ CaMgSi2 O6 þ Mg2þ þ Ca2þ þ 4Cl :

and shows that apart from possible changes in pH and speciation, which cannot be expressed in the equation, the predicted transport directions of cations are not changed. The reaction is pH-sensitive in both cases. The second stage seems to be, from a microstructural viewpoint, decomposition to albite þ amphibole or rather recrystallization of diopside þ albite to amphibole by hydration. Using measured mineral compositions and setting Mg2þþ Fe2þ ¼ M, Al3þþ Fe3þ ¼ Al one finds that both omphacite and diopside would react with plagioclase to form the more Al-rich amphiboles:

4.3. The second retrograde stage Symplectitization of omphacite, coronas around quartz and albitization of matrix epidote near quartz are all considered to be contemporaneous because of the similarity in grain size of the reaction products and because of chemical consistency of the localized reactions forming them (see below). The diopside/tremolite coronas around quartz grains embedded in a calcite matrix indicate that Ca and Mg were mobile. Albitization of epidote far from the Nasource omphacite corroborates significant mobility of Na. Fine dots of allanite outlining the pre-albitization grain boundary between quartz and epidote often run close to the present quartz-albite grain boundary. This means that the replacement front moved mainly towards epidote, hence Si was a bit more mobile than Al. However, as volume was more or less preserved in the slightly porous albitized parts, Al must have also been removed from that site, indicating a certain mobility of this element which is considered to have precipitated in the form of non-porous, allanite-free albite just nearby (see Fig. 5d, f, h). Hence the mono- and divalent cations were considered as mobile and those of higher valence as relatively immobile in the following reaction equations. As a consequence, (OH)– was introduced to charge balance the mobile cations. This seems to be arbitrary but we consider it justified by the textural observations just mentioned and by the consistency and simplicity of the resulting equations (see below)

1:6 di þ 1:3 plag þ 0:04 Ca2þ þ 2:7 M2þ þ 4:7 ðOHÞ ¼ 1:0 amph þ 1:4 H2 O þ 0:71 Naþ : ð4aÞ 2þ 2þ 1:4 omp þ 1:4 plag þ 0:99 Ca þ 3:4 M þ 7:7 ðOHÞ ¼ 1:0 ed þ 2:8 H2 O þ 1:1 Naþ :

ð4bÞ

a process that requires some of the Ca2þ and M2þ released by the initial omphacite decay and thus might occur to some extent coeval with it, reducing the necessity for ion exchange with the quartz site (see below). The growth of Al-poor amphiboles (actinolites) then indicates a more mature stage, i.e. a better approximation of equilibrium with the other reaction sites in the rock for the given P-T conditions: 2:66 di þ 0:82 plag þ 1:74 M2þ þ 0:95 ðOHÞ ¼ 1:0 act þ 1:0 Ca2þ þ 0:44 Naþ :

(4c)

Hence the type of symplectite developing depends on the relative supply of cations to the reaction site and the maturity of the process (approximation of equilibrium compositions on a larger scale), which is partly reflected in the textural position (along former omphacite margins and cracks) and the larger grain size of amphibole, compared to diopside. A third type is fine-grained amphibole-albite symplectite with no trace of diopside, but usually additional calcite. Such a symplectite could form by addition of CO2 that combines with the excess Ca released from omphacite decay. The amphiboles measured in such symplectites

4.3.1. Symplectitization of omphacite Symplectitization of omphacite indicates the last major stage of retrograde reactions. According to textural observations it is a two-stage process: a first decay to almost pure albite þ diopside, and a second stage of transformation of diopside to amphibole and of albite to a more Ca-rich plagioclase. These two subsequent stages occur side by side and are only controlled by the access of fluid and – as one aspect of this – by the

eschweizerbart_xxx

38

A. Proyer, F. Rolfo, D. Castelli, R. Compagnoni

absorbed there for corona diopside formation. Some of the Ca2þ released gets resorbed by the decomposing garnet in its corona formation of epidote þ calcite þ quartz, and Mg2þ and Fe2þ are exchanged in turn towards the symplectite sites where they trigger the growth of amphiboles. 3:86 omp þ 2:14 M2þ þ 0:35 CO2 þ 3:3 ðOHÞ Hence, even though the continued decomposition of garnet during the symplectite stage is not apparent from the reac¼ 1:0 amp þ 0:18 plag þ 0:35 CaCO3 þ 0:62 H2 O þ ð4cÞ tion texture itself, it is strongly indicated by mass balance þ 1:04 Na : requirements during this stage. The growth of amphibole in the symplectite also causes the Na/Al ratio in the symplectite solids to drop below 1.0, which signals release of a 4.3.2. Reaction rims around quartz near omphacite significant amount of Naþ to the pore fluid. Epidote is not The main corona mineral around matrix quartz is diopside, stable in the presence of such fluids and decomposes to which forms from the divalent cations and hydroxyl-ions in albite wherever additional Si-species are available as reacexactly the same ratio as expelled during omphacite tants. Obviously, this is just the case in close proximity to decomposition by reaction (4): matrix quartz, sometimes with a thin diopside corona in between. 2 SiO2 þ Ca2þ þ Mg2þ þ 4ðOHÞ ¼ CaMgSi2 O6 þ 2 H2 O: have a composition intermediate between those of the actinolite and edenite analyses given in Table 1, which results in the localized reaction equation 4c (again for Alconservation):

(5) Additional, less common amphiboles around quartz are compositionally very close to pure tremolite, as would be expected: 8 SiO2 þ 5 Mg2þ þ 2 Ca2þ þ 14 ðOHÞ ¼ 2 Ca2 Mg5 Si8 O22 ðOHÞ2 þ6 H2 O:

5. Element mobility during prograde and retrograde metamorphism The fact that the observed reaction textures can be described best by open-system reactions for each type of reaction site (‘‘site’’ meaning an isolated grain of albite, quartz, garnet or omphacite) means that diffusional transport of species in the fluid is the main controlling factor. The composition of minerals stable at various metamorphic stages is largely determined by thermodynamics, though sometimes only on a small scale, as indicated by variable amphibole compositions during the first retrograde stage and by the ‘‘coexistence’’ of metastable amphibole and diopside in the symplectites after omphacite. Apart from that, the differences in diffusion rate of the various chemical species of Na, Ca, Mg, Fe, Al and Si and Ti and their solubilities and chemical potential gradients determine, which of the numerous possible simultaneous reactions in open systems are fastest and hence most efficient. The discussion about quartz coronas around omphacite shows that Al – even though assumed as immobile in most of the above equations – was mobile to a certain extent, as otherwise SiO2 should not have accumulated. Relative diffusion rates during the first stage of retrogression are very apparent: coronae of epidote þ quartz þ calcite around garnet indicate relative immobility of Al and even Si. Mg2þ and Fe2þ ions released from the garnet sites trigger amphibole nucleation and growth at the grain boundaries of omphacite, the breakdown of which feeds back Ca2þ ions to the garnet site. These amphibole grains are chemically heterogeneous, with small-scale zoning in Na-, Al-, and Fe-content, indicating Na- release into the pore fluid (the Na/Al ratio in amphibole drops below 1), irregular supply of ions and perhaps a chromatographic effect causing variations in Mg2þ versus Fe2þ supply from the remote garnet sites. Nevertheless, Fe2þ, Mg2þ and Ca2þ were transported on a scale of several millimetres between garnet and ompacite sites during this stage. Oxidation must have played a role, too, because garnet is virtually free of ferric

(5a)

Textures indicate that the reacting volume for this kind of ion exchange is limited to grain boundaries within a few millimetres from omphacite sites. In the vicinity of an Alsource, like garnet or epidote, an outer rim of albite (Fig. 4f) can develop, indicating short-distance mobility of Si and significant mobility of Na: Ca2 Al3 Si3 O12 ðOHÞ þ 3 Naþ þ 6 SiðOHÞ4 ¼ 3 NaAlSi3 O8 þ 2 Ca2þ þ ðOHÞ þ13 H2 O:

(6)

4.3.3. Albitization of quartz and epidote remote from omphacite Two textural types of plagioclase are produced during the last stage of retrogression: one is part of symplectites after omphacite and the second can develop where epidote and quartz occur next to each other. Only a minor amount of quartz gets albitized and epidote also remains as a relic, suggesting that transformation only occurred where fluid was passing through, carrying a significant amount of Na generated from omphacite breakdown and allowing a limited mobility of Si and Al. The reaction at the quartz-epidote site can be formulated as: 6 SiO2 þ 3 Naþ þ Ca2 Al3 Si3 ðOHÞ ¼ 3 NaAlSi3 O8 þ 2 Ca2þ þ OH :

(7)

and is equivalent to reactions (6). There seems to be a reasonable mass balance of ions between reaction sites during the symplectite stage: Mg2þ, Ca2þ and (OH)– are transferred from the most primitive diopside-albite symplectites to the quartz sites and eschweizerbart_xxx

Diffusion-controlled metamorphic reaction textures

iron, while epidote contains about 7–8 wt% Fe2O3 (around 0.45 a.p.f.u. Fe3þ). This might explain why amphiboles are less ferrous than one would expect from the ionic balance with the garnet site as shown above. This does not necessarily mean that the fluid entering the rock was oxidizing. Oxidation may have been enforced as the fastest process leading to decomposition of garnet in a situation of significant overstepping of an equilibrium reaction curve. As the structure of epidote-group minerals accommodates ferric much rather than ferrous iron, contact with the fluid may have resulted to some extent in direct oxidation (2 FeO þ H2O ¼ Fe2O3 þ H2) while the remaining FeO from garnet was dissolved (FeO þ H2O ¼ Fe2þþ 2 OH–). Such changes in the redox potential and pH of the fluid may have modified the other reactions mentioned above to some extent. The greatest number of different reaction sites develops during the symplectite stage: There are four different symplectites, two reactions at the quartz site, two (very similar) reactions at combined quartz þ epidote sites and most likely an ongoing transformation of garnet to epidote because transformation of diopside to amphiboles in the symplectites results in a constant excess of Ca2þ and a demand for Mg2þ and Fe2þ, which the garnet in turn would supply for diopside and amphibole formation. There is a clear difference on transport distances for the various cations at this stage: The Naþ released in the symplectites is transported on a millimetre to centimetre scale before it precipitates at least partly as albite where an Al- and Si-source are in contact. Ca, Mg and Fe2þ are transported over distances of a few millimetres along grain boundaries from omphacite to quartz sites to form diopside and actinolite coronas. The exchange of Ca2þ versus Mg2þþFe2þ between symplectites and garnet sites covers distances on the order of several millimetres. Si mobility is low and that of Al even lower, as indicated by the limited extent and asymmetry of albite formation at quartz-epidote contacts.

39

fluid access had to trigger symplectite formation and the pertinent reactions at the other sites. The type and maturity of symplectites after omphacite depends on cation availability and may range from albite þ diopside via a combined albite þ diopside þ Al-rich amphibole symplectite to albite þ actinolitic amphibole symplectite at a mature stage with sufficient supply of M2þ ions, or a symplectite of albite þ actinolite þ calcite, where CO2 was supplied. We have no direct indication that acid-base reactions were operative, but the fact that H3Oþ and OH-are fastmoving species means that they are kinetically relevant, particularly in cases where significant overstepping of equilibrium conditions supplies a driving force for fast reactions, once the kinetic barriers are overcome by the infiltration of fluid. Redox reactions seem to play a certain role, as indicated by the partial oxidation of ferrous iron from garnet to form epidote, but there is not enough analytical evidence to derive more detailed information about the importance of redox processes. Oxidation enforced by Gibbs free energy minimization is nevertheless an interesting conclusion from our observations. The relative diffusivities of ionic species in the pore fluid cannot be estimated with precision because there is no information about the spatial and temporal availability of fluids in the various settings during the different metamorphic stages. A tentative qualitative sequence, however, would be Naþ . Mg2þ, Ca2þ, Fe2þ .. Si4þ . Al3þ, Ti4þ, which is essentially in agreement with general theory and earlier findings from field and experiment.

Acknowledgements: Two anonymous reviewers provided valuable suggestions to improve the manuscript. The responsibility for any remaining flaws is fully ours. The first author wants to gratefully acknowledge the financial support for this work by Austrian Science Fund project P22479-N21. The studied samples were collected during fieldwork funded by the Italian M.U.R.S.T. and National Research Council.

6. Conclusions References

Marbles with isolated specks of non-carbonate minerals provide fine examples to study the role of localized reactions in metamorphic systems subject to a variety of thermobaric conditions, which may in turn yield substantial information about metamorphic reaction mechanisms in general. Provided these lithologies escaped widespread deformation, both prograde and retrograde coronas or reaction rims can be preserved and themselves commonly preserve older minerals in their interior, which makes such rocks also valuable recorders of high-pressure to UHP conditions. The first retrograde reactions may have been triggered by influx of fluid or, as indicated by scarce relics of talc, by the reaction talc þ omphacite þ garnet ¼ amphibole þ epidote  quartz. Corona formation around omphacite and garnet stopped with the end of fluid infiltration or with complete talc consumption, and a separate stage of minor

Bach, W. & Klein, F. (2009): The petrology of seafloor-rodingites: insights from geochemical reaction path modeling. Lithos, 112, 103–117. Carmichael D.M. (1969): On the mechanism of prograde metamorphic reactions in quartz-bearing politic rocks. Contrib. Mineral. Petrol., 20, 244–267. Fisher G.W. (1989): Matrix analysis of metamorphic mineral assemblages and reactions. Contrib. Mineral. Petrol., 102, 69–77. Joesten R. (1977): Evolution of mineral assemblage zoning in diffusion metasomatism. Geochim. Cosmochim. Acta, 41, 649–670. Katayama, I., Ohta, M., Ogasawara, Y. (2002): Mineral inclusions in zircon from diamond-bearing marble in the Kokchetav massif, northern Kazakhstan. Eur. J. Mineral., 14, 1103–1108. Kato, T., Enami, M., Zhai, M. (1997): Ultra-high-pressure (UHP) marble and eclogite in the Su-Lu UHP terrane, eastern China. J. metamorphic Geol., 15, 169–182. eschweizerbart_xxx

40

A. Proyer, F. Rolfo, D. Castelli, R. Compagnoni

Sobolev, N.V., Schertl, H.-P., Neuser, R.D., Shatsky, V. (2007): Relict unusually low iron pyrop-grossular garnets in UHPM calc-silcate rocks of the Kokchetav Massif, Kazakhstan. Int. Geol. Rev., 49, 717–731. Spear, F.S., Rumble III, D., Ferry, J.M. (1982): Linear algebraic manipulation of N-dimensional composition space. in ‘‘Characterization of metamorphism through mineral equilibria’’, J.M. Ferry, ed., Rev. Mineral., 10, 53–104, Mineralogical Society of America. Tang, J., Zheng, Y.-F., Wu, Y.-B., Gong, B. (2006): Zircon SHRIMP U-Pb dating, C and O isotopes for impure marbles from the Jiaobei terrane in the Sulu orogen: implication for tectonic affinity. Precambrian Res., 144, 1–9. Thompson J.B., Jr. (1982a): Composition space: an algebraic and geometric approach. in ‘‘Characterization of metamorphism through mineral equilibria’’, J.M. Ferry, ed., Rev. Mineral., 10, 1–31, Mineralogical Society of America. Thompson J.B., Jr. (1982b): Reaction space: an algebraic and geometric approach. in ‘‘Characterization of metamorphism through mineral equilibria’’ J.M. Ferry, ed., Rev. Mineral., 10, 33–52, Mineralogical Society of America. Wang, X. & Liou, J.G. (1993): Ultra-high pressure metamorphism of carbonate rocks in the Dabie Mountains, central China. J. metamorphic Geol., 11, 575–588. Ye, K. & Hirajima, T. (1996): High-pressure marble at Yangguantun, Rongcheng County, Shandong Province, eastern China. Mineral. Petrol., 57, 151–165.

Leake, B.E., Woolley, A.R., Arps, C.E.S., Birch, W.D., Gilbert, M.C., Grice, J.D. et al. (1997): Nomenclature of amphiboles; report of the Subcommittee on Amphiboles of the International Mineralogical Association Commission on New Minerals and Mineral Names. Eur. J. Mineral., 9, 623–651. Ogasawara, Y., Zhang, R.-Y., Liou, J.G. (1998): Petrogenesis of dolomitic marbles from Rongcheng in the Su-Lu ultrahigh-pressure metamorphic terrane, eastern China. Island Arc, 7, 82–97. Ogasawara, Y., Ohta, M., Fukasawa, K., Katayama, I., Maruyama, S. (2000): Diamond-bearing and diamond-free metacarbonate rocks from Kumdy-Kol in the Kokchetav Massif, northern Kazakhstan. Island Arc, 9, 400–416. Ogasawara, Y., Fukasawa, K., Maruyama, S. (2002): Coesite exsolution from supersilicic titanite in UHP marble from the Kokchetav massif, northern Kazakhstan. Am. Mineral., 87, 454–461. Proyer, A., Rolfo, F., Zhu, Y.F., Castelli, D., Compagnoni, R. (2013): Ultrahigh-pressure metamorphism in the magnesite þ aragonite stability field: evidence from two impure marbles from the DabieSulu UHPM belt. J. metamorphic Geol., 31, 35–48. Qian, C. & Huang, J. (2000): Geological map of Changpu area, Yuexi County, Anhui Province, scale 1:10.000. Anhui People’s Publishing House. Rolfo, F., Compagnoni, R., Xu, S., Jiang, L. (2000): First report of felsic whiteschist in the ultrahigh-pressure metamorphic belt of Dabie Shan, China. Eur. J. Mineral., 12, 883–898. Rolfo, F., Compagnoni, R., Wu, W., Xu, S. (2004): A coherent lithostratigraphic unit in the coesite eclogite complex of Dabie Shan, China: geologic and petrologic evidence. Lithos, 73, 71–94. Schmid, R., Romer, R.L., Franz, L., Oberha¨nsli, R., Martinotti, G. (2003): Basement-cover sequences within the UHP unit of the Dabie Shan. J. metamorphic Geol., 21, 531–538.

Received 3 May 2013 Modified version received 16 October 2013 Accepted 23 October 2013

eschweizerbart_xxx