Boron isotopic composition of tourmaline, prismatine, and grandidierite from granulite facies paragneisses in the Larsemann Hills, Prydz Bay, East Antarctica: Evidence for a non-marine evaporite source

Boron isotopic composition of tourmaline, prismatine, and grandidierite from granulite facies paragneisses in the Larsemann Hills, Prydz Bay, East Antarctica: Evidence for a non-marine evaporite source

Available online at www.sciencedirect.com Geochimica et Cosmochimica Acta 123 (2013) 261–283 www.elsevier.com/locate/gca Boron isotopic composition ...

4MB Sizes 143 Downloads 193 Views

Available online at www.sciencedirect.com

Geochimica et Cosmochimica Acta 123 (2013) 261–283 www.elsevier.com/locate/gca

Boron isotopic composition of tourmaline, prismatine, and grandidierite from granulite facies paragneisses in the Larsemann Hills, Prydz Bay, East Antarctica: Evidence for a non-marine evaporite source JohnRyan MacGregor a, Edward S. Grew a,⇑, Jan C.M. De Hoog b, Simon L. Harley b, Piotr M. Kowalski c, Martin G. Yates a, Chris J. Carson d a

School of Earth and Climate Sciences, University of Maine, Orono, ME 04469, USA School of Geosciences, University of Edinburgh, Edinburgh EH9 3JW, United Kingdom c Forschungszentrum Ju¨lich, Institute of Energy and Climate Research (IEK-6), Wilhelm-Johnen-Strasse, 52425 Ju¨lich, Germany d Geoscience Australia, P.O. Box 378, Canberra, ACT 2601, Australia b

Received 9 October 2012; accepted in revised form 12 May 2013; available online 1 June 2013

Abstract A unique assemblage of B-rich rocks (<2 wt.% B) containing the borosilicate minerals tourmaline (Tur), prismatine (Prs) and grandidierite (Gdd) is exposed in the ca. 1000 Ma Brattstrand Paragneiss in the Larsemann Hills, East Antarctica. The B isotope composition of these three minerals in 21 paragneisses and 6 anatectic pegmatites were measured in situ using secondary ion mass spectrometry. d11B ranges from 2.8 to 14.4& in tourmaline, from 9.6 to 17.8& in prismatine and from–2.8 to 8.7& in grandidierite (weighted uncertainties mostly ±1–2& per sample). In most cases, average d11B increases in a given sample Prs < Tur < Gdd. Whether the observed B-isotope distribution has attained equilibrium were assessed using two criteria: microstructural equilibrium and regular distribution of Mg and Fe. In samples for which the two criteria are met, we assume the measured distribution of isotopes D11BA–B (= d11BA  d11BB) represents equilibrium: D11BTur-Prs = + 5.0 ± 1.4& (5 of 12 pairs), D11BTur -Gdd = 3.3 ± 0.8& (3 of 5 pairs), and D11BGdd-Prs = + 7.2 ± 1.3& (2 of 2 pairs), consistent with the preference of 10B for tetrahedral sites (prismatine) and 11B for trigonal sites (tourmaline, grandidierite). Ab initio computations of B isotope fractionation factors, D11BTur-Prs = + 6.4 ± 1.3&, D11BTur -Gdd = 1.8 ± 1.1&, and D11BGdd-Prs = + 8.2 ± 1.1&, provide confirmation ˚ ) B–O bond length of the measured fractionations. The computed, relaxed ionic structure of grandidierite gave a shorter (1.368 A 11 ˚ than in case of dravite (1.385 A), which explains the measured enrichment of grandidierite in B relative to tourmaline. The precursor of the B-rich rock least changed by metamorphism, tourmaline metaquartzite (d11B(Tur) = 8.6& to 5.9&), is interpreted to be a product of pre-metamorphic, hydrothermal B-metasomatism. Boron sources consistent with this d11B range include oceanic crust, clastic sediments, volcanic rocks and non-marine evaporite borate, but not marine evaporite. However, dominance of metasediments in the Brattstrand Paragneiss, major and trace element compositions of the B-rich rocks and presence of abundant B are consistent with non-marine evaporite being the most important source of B for the B rich lithologies of the Larsemann Hills. A possible scenario for precursors of these rocks is a succession of clastic sediments, in part tuffaceous, intercalated with B-rich evaporite deposits in a continental rift basin, in which circulating hydrothermal fluids leached B from the evaporite and precipitated it as tourmaline in the associated clastic rocks to form the precursors to the tourmaline metaquartzites. Ó 2013 Elsevier Ltd. All rights reserved.

⇑ Corresponding author.

E-mail addresses: [email protected] (J.R. MacGregor), [email protected] (E.S. Grew), [email protected] (J.C.M. De Hoog), [email protected] (S.L. Harley), [email protected] (P.M. Kowalski), [email protected] (M.G. Yates), [email protected] (C.J. Carson). 0016-7037/$ - see front matter Ó 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.gca.2013.05.030

262

J.R. MacGregor et al. / Geochimica et Cosmochimica Acta 123 (2013) 261–283

submarine hydrothermal processes analogous to those Slack et al. (1989, 1993) proposed for the tourmaline-rich rocks associated with Pb–Zn–Ag deposits at Broken Hill, Australia. Wang et al. (2004) obtained B isotopic data on tourmaline, prismatine and grandidierite from the Larsemann Hills using thermal ionization mass spectrometry (Xiao et al., 1988) and cited the low 11B/10B ratios as evidence for a non-marine evaporite source of the boron. Grew et al. (2013a) presented a larger set of geochemical data, together with preliminary B isotopic data measured in situ, to explain in greater detail how enrichment in boron could have resulted from premetamorphic hydrothermal alteration, either in a rifted basin or in a mud volcanic system. The purpose of the present paper is to report the full set of B isotope data that Grew et al. (2013a) considered preliminarily to (1) determine the equilibrium distribution of B isotopes between tourmaline, prismatine, and grandidierite and (2) provide constraints on the possible source of B in the precursors of the B-rich rocks.

1. INTRODUCTION Boron has two naturally occurring stable isotopes, 10B (ca. 20%) and 11B (ca. 80%) differing significantly in atomic weight, and as a result boron isotopes are subject to readily measurable fractionation. Because geologic reservoirs differ markedly in isotopic composition, boron isotopes are a powerful means to infer possible sources of boron for the precursors of boron-bearing metamorphic rocks and to understand the processes that affected the precursor during burial, metamorphism, anatexis and retrograde rehydration (e.g., Palmer and Slack, 1989; Slack et al., 1993; Leeman and Sisson, 2002; Palmer and Swihart, 2002). The challenge is to establish how each process contributed to the boron isotopic signatures measured today. In contrast to most granulite-facies rocks, paragneisses containing borosilicate minerals in the Larsemann Hills, a coastal exposure on Prydz Bay, East Antarctica (Fig. 1), contain up to 2 wt.% B, and thus provide an excellent opportunity to apply in situ measurements of boron isotopes using secondary ion mass spectrometry (SIMS) to sort out the effects of metamorphic processes on isotopic fractionation and infer the source of boron. Tourmaline and two high-temperature borosilicate minerals, prismatine, (h,Mg,Fe,Na)(Al,Mg,Fe)9(Si,B,Al)5O21(OH,F), where h indicates vacancy, and grandidierite, (Mg,Fe2+)Al3BSiO9, were first reported from the Larsemann Hills by Ren and Zhao (1992), Ren et al. (1992) and Ren and Liu (1993, 1994). Carson et al. (1995a) and Carson and Grew (2007) mapped the regional extent of these borosilicate minerals in the Larsemann Hills. On the basis of whole-rock geochemical data, Grew et al. (2006) suggested that the precursors to the B-rich rocks in the Larsemann Hills could be clastic and volcanogenic rocks altered by 74˚E

2. GEOLOGIC AND TECTONIC SETTING The Larsemann Hills (Figs. 1 and 2), one of the larger exposures on the southeast coast of Prydz Bay, are composed primarily of granulite-facies rocks, including the Brattstrand Paragneiss, which extends from the Brattstrand Bluffs to the Bølingen Islands (e.g., Fitzsimons and Harley, 1991; Fitzsimons, 1997; Carson and Grew, 2007). The Brattstrand Paragneiss includes a unique suite of boronand phosphorous-rich rocks in the Larsemann Hills (Grew et al., 2013a). The paragneisses were affected by several episodes of deformation and intrusion of granite and pegmatite (Stu¨we et al., 1989; Fitzsimons and Harley, 1991; Dirks et al., 1993; Carson et al., 1995b, 1997). Peak

76˚E

78˚E

68˚S

Antarctica

Prydz Vestfold Hills

Bay

Sørsdal Glacie

r

helf eS y Ic

s

nd

la

Is

ills

n

H

ge

n

lin

an

fs

uf

em

Bø rr Ke s ro ain un nt M ou M

Am er

69˚S

Bl

rs

La

nd

tra

ts

at

Br

Rauer Islands

0

40

Statler Hills

Kilometres 70˚S

Fig. 1. Map showing exposures along the Prydz Bay coast, East Antarctica.

J.R. MacGregor et al. / Geochimica et Cosmochimica Acta 123 (2013) 261–283

263

Antarctica

PRYD

s

Z BAY

“Eliza Kate Island”

N

7 8 Seal Cove 9-16

s kne a Bro insul n Pe

6 Fisher Island

17

s rne a Sto insul n Pe

McCarthy Point

ID GR

1-3

5

s

E NS

TE

4

Wilcock Bay

T

AS

O NC

18-20

IN

IS HR

C

Legend - Paragneiss locality - Pegmatite locality

Fig. 2. Sketch map of the Larsemann Hills, showing locations of the samples studied (location numbers given in Table 1). Horizontal Datum: WGS84 Projection: UTM Zone 43.

metamorphic conditions are estimated to have reached 0.6– 0.7 GPa and 800–860 °C (Thost et al., 1994; Fitzsimons, 1997; Carson et al., 1997). Most investigators conclude there were two major high-temperature metamorphic events, one at ca.1000–900 Ma and the second at ca. 530 Ma, but the relative intensities and regional extent of the two events remains controversial, as does the timing of sedimentation of the precursors to the Brattstrand Paragneiss. The scenario proposed by Kelsey et al. (2008) envisaged a Neoproterozoic basin developing along the eastern shore of Prydz Bay at the time of closure along a suture zone to the southeast during amalgamation of eastern Gondwana from ca. 600 to ca. 575 Ma. In contrast, building on the scenario proposed by Wang et al. (2008), Grew et al. (2012) cited SHRIMP U-Pb data as evidence that precursors of the Brattstrand Paragneiss were deposited in a back-arc basin located inboard of a ca. 1000 Ma continental arc that was active along the leading edge of the Indo-Antarctic craton prior to a collisional orogeny at 930–900 Ma (Boger, 2011; Grew et al., 2012), which may correlate with that in the nearby Rayner Complex of East Antarctica and termination of the 1000–900 Ma tectonothermal episode in the Eastern Ghats Belt. Syntectonic pegmatites in the Larsemann Hills are one product of extensive partial melting during decompression from peak conditions near 0.7 GPa, 800 °C to about 0.4– 0.5 GPa, 750 °C in the Early Palaeozoic (Carson et al., 1997), most likely, at 530–510 Ma. Our field studies of pegmatites containing borosilicate minerals revealed three generations (Grew et al., 2008, 2013b; Wadoski et al., 2011). The earliest overlapped the D2 and D3 deformations of Carson et al. (1997) and forms irregular pods and veins up to a meter thick, which are either roughly concordant or crosscut S2 and S3 fabrics and are locally

folded. The second generation is associated with D4 and forms planar, discordant veins up to 20–30 cm thick, whereas the youngest generation forms discordant veins and pods. The D2–D3 and D4 pegmatites are abyssal class ˇ erny´ and Ercit, 2005) characterized by (BBe subclass, C tourmaline + quartz intergrowths and boralsilite (Al16B6Si2O37); the borosilicates prismatine, grandidierite, werdingite and dumortierite are locally present. In contrast, tourmaline (no intergrowths), beryl and primary muscovite are present in the youngest pegmatites, i.e., beryl subclass of the rare-element class. Spatial correlation between tourmaline-bearing pegmatites and borosilicate-bearing rocks in the host Brattstrand Paragneiss is greatest for the D2–D3 pegmatites, less for the D4 pegmatites, and non-existent for the youngest pegmatites, which leads us to conclude the D2–D3 and D4 pegmatites are anatectic melts close to their source. 3. SAMPLES AND PETROGRAPHIC DESCRIPTIONS The 27 samples analyzed for B isotopes (Fig. 2; Table 1) belong to seven rock types containing tourmaline, prismatine or grandidierite from five of the stratigraphic units mapped by Carson and Grew (2007) in the Brattstrand Paragneiss sequence: (1) tourmaline metaquartzites, (2) borosilicate gneisses, (3) leucogneisses, (4) iron-rich rocks, (5) miscellaneous silica-undersaturated rocks, (6) biotite gneiss with prismatine segregations and (7) anatectic pegmatites (MacGregor, 2012; Grew et al., 2013a). Rock names follow Carson and Grew (2007) and Grew et al. (2013a). The six studied pegmatites belong to the deformed D2–D3 generation, which is the earlier of the two generations containing a diverse borosilicate assemblage (Grew et al., 2008; Wadoski et al., 2011).

264

J.R. MacGregor et al. / Geochimica et Cosmochimica Acta 123 (2013) 261–283

Table 1 List of analyzed samples. Location number

Latitude and longitude

011704

69° 76° 69° 76°

120902

123001 010205

121801 121602 112701 112704 121103 121105 012101 121202 121102 112801 120603 120405 121301 011401 120102 112906 SH/06/10 SH/06/22 SH/06/01R

Location (Fig. 2)

26.7120 S 06.4000 E 26.7010 S 06.4150 E

Thin section designations

Rock type or mineral

Map unit (Carson and Grew, 2007)

1

C

Tourmaline metaquartzite

Stornes Gneiss

2

A

Tourmaline metaquartzite Borosilicate gneiss Pegmatite Tourmaline metaquartzite

Thala Tourmaline Metaquartzite Wilcock Bay Metapelite – Thala Tourmaline Metaquartzite Thala Tourmaline Metaquartzite – Stu¨we Metapelite

69° 76° 69° 76°

26.7010 06.4150 26.6840 07.7500

S E S E

3

B1, I, J G2 L1

4

A, I

Tourmaline metaquartzite

69° 76° 69° 76° 69° 76° 69° 76° 69° 76° 69° 76° 69° 76° 69° 76° 69° 76° 69° 76° 69° 76° 69° 76° 69° 76° 69° 76° 69° 76° 69° 76° 69° 76° 69° 76° 69° 76°

27.7830 S 04.1120 E 23.5060 S 15.2500 E 22.6620 S 10.8080 E 22.835 S 11.1280 E 24.2700 S 06.7050 E 24.2310 S 06.8330 E 24.2700 S 06.7050 E 24.4120 S 06.7020 E 24.2090 S 06.7080 E 24.4580 S 06.5360 E 24.4740 S 07.9680 E 24.5460 S 07.5310 E 24.6230 S 05.8350 E 25.0370 S 03.7910 E 25.0160 S 03.8440 E 24.8950 S 04.4290 E 24.2150 S 06.7270 E 24.5550 S 067250 E 23.180 S 23.900 E

5

J5 A

6

C

Pegmatite Garnet-cordierite-biotite gneiss Iron-rich rock

7

C

8

B4

9

Stu¨we Metapelite

B

Prismatine porphyroblast in biotite gneiss Prismatine–orthopyroxenerock Borosilicate gneiss

Stornes Gneiss

Stornes Gneiss

10



Biotite gneiss

Stornes Gneiss

11

M

Borosilicate gneiss

Stornes Gneiss

12

C, D

Tourmaline metaquartzite

Stornes Gneiss

13

C1

Pegmatite



14

D

Pegmatite



15

F2

Leucogneiss

16

D2

Pegmatite

Donovan Prismatine Leucogneiss –

17

B

Borosilicate gneiss

Stornes Gneiss

18

D3

Leucogneiss

19

G

Leucogneiss

20

B

Pegmatite

Donovan Prismatine Leucogneiss Donovan Prismatine Leucogneiss –





Prismatine separate







Grandidierite separate







Prismatine separate



Stornes Gneiss

Note: Samples with the prefix SH were collected by Simon Harley: SH/06/10 and SH/06/22 from the northeast end and east side, respectively, of the lake to the east of Lake Ferris, northern Stornes Peninsula (close to points 9–16, Fig. 2); SH/06/01R from Seal Cove, Broknes Peninsula.

3.1. Tourmaline metaquartzites (7 samples) Tourmaline metaquartzites are composed almost entirely of tourmaline and quartz, with accessory plagioclase, apatite and ilmenite. However, sample 123001L contains a porphyroblast of prismatine, and sample 010205A has sub-

ordinate grandidierite, which is mostly altered to phyllosilicate. The abundance of tourmaline is approximately equal to or greater than quartz. Tourmaline grain sizes range from 0.5 to 1 mm, and are roughly equant and subrounded, although in some cases grains exhibit highly irregular grain boundaries (Fig. 3a). Aggregates of tourmaline

J.R. MacGregor et al. / Geochimica et Cosmochimica Acta 123 (2013) 261–283

grains show triple junctions with grain boundaries intersecting at about 120°. Tourmaline is olive green to light brown in color and shows no color zoning, although a few grains have light-blue rims. 3.2. Borosilicate gneisses (6 samples) All three borosilicates prismatine, grandidierite and tourmaline are commonly present in a given thin section, but are generally found segregated in separate layers, which are typically a few millimeters thick. The matrix is composed of quartz, plagioclase and K-feldspar; sillimanite is also commonly present, and rutile is a widespread accessory. Several samples are blue-green in hand specimen due to the abundance of well-aligned grandidierite prisms. In three sections (012101M; 121301B; 120902I), all three borosilicates are found in association with one another. In section 012101M prismatine and grandidierite occur in patches containing sillimanite and abundant apatite, which are irregularly dispersed in a matrix of tourmaline and quartz (Fig. 3b). Within the patches, grandidierite is

265

commonly surrounded by prismatine and apatite, and is isolated from quartz. Tourmaline inclusions are uncommon in prismatine. Quartz grains are commonly large (2 mm), and grain boundaries intersect at triple junctions forming 120° angles. In section 121301B, centimeter-sized prisms of prismatine are well-aligned in a matrix of feldspar, and sillimanite (Fig. 3c). These prisms border a plagioclase domain with irregularly distributed prismatine, tourmaline and grandidierite. Grandidierite in the plagioclase domain is commonly overgrown by prismatine and occurs as inclusions in prismatine (Fig. 4a). The abundance of tourmaline is greatest within the feldspar domain, in which it forms isolated grains. Section 120902I has several layers, one of which is dominated by well-aligned sillimanite; a second contains prismatine and sillimanite in a quartz matrix, whereas a third contains grandidierite and sillimanite in a matrix of abundant quartz and minor plagioclase. Grandidierite and prismatine are commonly segregated, but locally occur in contact. Rare, fine-grained tourmaline and tourmaline overgrowths of grandidierite and prismatine are present.

Fig. 3. Photomicrographs illustrating microstructures interpreted to be consistent with equilibrium crystallization of the borosilicate minerals. Plane-polarized light. (a) Sample 011704C, tourmaline metaquartzite. (b) Sample 012101M, borosilicate gneiss. (c) Sample 121301B, borosilicate gneiss. Despite the disparity in size, the three borosilicates appear to be in microstructural equilibrium, but grandidierite and tourmaline did not equilibrate in terms of Fe-Mg distribution. (d) Sample 120102G, leucogneiss. Tourmaline aggregate is polygonal with grain boundaries intersecting at 120°. There is no evidence of reaction with prismatine. Ap – apatite, Crd- cordierite, Gdd – grandidierite; Mgt – magnetite, Pl – plagioclase, Prs – prismatine, Qtz – quartz.

266

J.R. MacGregor et al. / Geochimica et Cosmochimica Acta 123 (2013) 261–283

3.3. Leucogneisses (3 samples) The leucogneisses consists primarily of plagioclase and quartz, with subordinate prismatine, tourmaline, biotite and cordierite, as well as accessory ilmenite-hematite, hercynite and magnetite. Prismatine commonly forms irregularly distributed porphyroblastic patches, locally surrounded by polygonal aggregates of tourmaline (Fig. 3d). Less common are large centimeter-scale crystals of prismatine, surrounded by cordierite segregations. Fine tourmaline grains are locally isolated in a quartz–plagioclase matrix. 3.4. Iron-rich rocks (1 sample) The iron-rich rocks, which are restricted to Fisher Island, contain layers rich in magnetite with biotite, orthopyroxene, prismatine, monazite, apatite and, less commonly, grandidierite, and tourmaline (Grew et al., 2013a). Magnetite comprises roughly 20 modal percent, and is commonly associated with hercynite. In the studied

sample of this rock type, No. 121602C, centimeter sized prisms of prismatine enclose coarse orthopyroxene grains up to nearly 1 cm across. Tourmaline is found as rounded inclusions in prismatine and orthopyroxene. Finely fibrous grandidierite occurs in fractures in orthopyroxene (Fig. 4c). 3.5. Miscellaneous silica-undersaturated rocks (2 samples) Relatively SiO2-poor rocks lacking quartz include samples 112704B4 and 121801A. Sample 112704B4 consists of prismatine prisms and orthopyroxene porphyroblasts, and abundant millimeter-scale biotite grains between porphyroblast grain boundaries. Tourmaline fills fractures in orthopyroxene. Section 121801A (39.30 wt% SiO2, Grew et al., 2013a) is a prismatine bearing garnet–cordierite–biotite gneiss. Cordierite and biotite are the most abundant phases, with lesser garnet, magnetite and hercynite. Prismatine forms prisms 5 mm in length, and trace tourmaline is found either as inclusions in garnet (Fig. 4b) or adjacent to garnet or to biotite.

Fig. 4. Photomicrographs illustrating microstructures indicating non-equilibrium crystallization between the borosilicate minerals. Planepolarized light. (a) Sample 121301B, borosilicate gneiss. Grandidierite is overgrown by prismatine. (b) Sample 121801A, Si-poor garnet– cordierite–biotite gneiss with prismatine. Tourmaline is found only in garnet. (c) Sample 121602C, Fe-rich rock. Fibrous grandidierite filling fractures cutting orthopyroxene appears to be secondary. (d) Sample 121105, prismatine porphyroblast in biotite gneiss. Tourmaline infilling fractures cutting prismatine appears to be secondary. Crd- cordierite, Grt – garnet, Gdd – grandidierite; Mgt – magnetite, Pl – plagioclase, Prs – prismatine, Qtz – quartz.

J.R. MacGregor et al. / Geochimica et Cosmochimica Acta 123 (2013) 261–283

3.6. Prismatine segregations in biotite gneiss (2 samples) Section 121105 consists of centimeter-sized prismatine prisms surrounded by cordierite and segregated from a domain of plagioclase, K-feldspar, quartz and well-aligned biotite. Prismatine is fractured and contains inclusions of tourmaline that has grown in fractures (Fig. 4d), and uncommon relict prismatine is present in cordierite. Rare, fine fibrous grandidierite occurs sparingly in cordierite. The cordierite matrix is cut by microveinlets of tourmaline-quartz intergrowth that resemble the graphic tourmaline-quartz intergrowths characteristic of the anatectic pegmatites (Wadoski, 2009; Wadoski et al., 2011). Sample 112701C contains a prism of prismatine nearly 2 cm across in a matrix of plagioclase, biotite and sparse cordierite. The prism encloses monazite, apatite, magnetite, and traces of tourmaline and grandidierite, the last in association with magnetite and biotite. 3.7. Anatectic pegmatites (6 samples) The most widespread minerals in the D2–D3 anatectic pegmatites are quartz, plagioclase, K-feldspar, apatite, tourmaline and boralsilite, Al16B6Si2O37; prismatine and grandidierite are less common, whereas two borosilicate minerals, werdingite and dumortierite, have been found, but not in the analyzed sections (Grew et al., 1998, 2008; Wadoski et al., 2011). Tourmaline in three sections forms primary graphic intergrowths with quartz; the primary tourmaline is overgrown in places by a later generation of tourmaline. In the two sections with associated tourmaline and prismatine (112801D and 121102C1), tourmaline forms overgrowths on prismatine. Section 112801D is characterized by a graphic intergrowth of prismatine with quartz (Wadoski et al., 2011, Fig. 4B), whereas section 121102C1 has prismatine prisms in plagioclase. Grandidierite is present in section 010205J5 as 0.5 mm grains in fractures in plagioclase, and in section 112906B as grains 1–2 mm across in a matrix of quartz and K-feldspar (Grew et al., 2008, Fig. 11e). 3.8. Crystallization sequence in the paragneisses The microstructural relationships exhibited in the B-enriched paragneisses suggest three phases of tourmaline crystallization during metamorphism: (1) on the prograde path, (2) at peak metamorphic conditions and (3) on the retrograde path during the Early Cambrian event in the Larsemann Hills. Evidence for early prograde formation is the presence of tourmaline inclusions in peak metamorphic minerals, such as garnet (Fig. 4b), prismatine and orthopyroxene; these inclusions imply that peak-metamorphic minerals crystallized and overgrew previously formed tourmaline. Evidence for formation at peak conditions include the equigranular aggregates of tourmaline and quartz that make up the tourmaline metaquartzites (Fig. 3a) and isolated grains (Fig. 3c) and equigranular tourmaline aggregates adjacent to prismatine in other rock types (Fig. 3b and d). Post-peak retrograde metamorphic growth of tourmaline is indicated by (1) tourmaline overgrowths of prism-

267

atine and grandidierite; (2) tourmaline filling fractures in prismatine (Fig. 4d) and orthopyroxene; (3) microveinlets of intergrown tourmaline and quartz cutting cordierite. Cordierite in the Larsemann Hills is commonly associated with post-peak decompression (e.g., Carson et al., 1997). Prismatine is inferred to have crystallized primarily at peak metamorphic conditions. It locally crystallized contemporaneously with orthopyroxene, and following the crystallization of grandidierite and sillimanite, which it has overgrown in some sections (e.g., Fig. 4a). In addition, prismatine is locally embayed by and forms relicts in cordierite. Grandidierite microstructures suggest three stages of growth during early, peak, and retrograde metamorphism. Prismatine overgrowths as displayed in Fig. 4a are suggestive of early crystallization of grandidierite. Growth at the metamorphic peak is represented by grandidierite occurrences in the same section as prismatine. However, both borosilicate minerals are commonly segregated into separate layers and are rarely associated in microstructural equilibrium (Fig. 3b). Late stage formation is suggested by the occurrence of fine aggregates of randomly oriented grandidierite fibers in fractures in orthopyroxene (Fig. 4c), cordierite, K-feldspar, and quartz. Tourmaline metaquartzites are the only metamorphic rock that is interpreted to have reached peak metamorphic conditions with negligible change in mineral assemblage. The peak metamorphic assemblages containing two or three borosilicate minerals are most relevant for constraining equilibrium B isotopic fractionation among coexisting borosilicate minerals. Tourmaline and prismatine in microstructural equilibrium at peak metamorphic conditions are found in many of the rock types (e.g., Fig. 3c and d), and there are a few sections that contain coexisting grandidierite and tourmaline. Least common are sections, i.e., 012101M (Fig. 3b) and 120902I, in which all three borosilicates appear to have crystallized together in microstructural equilibrium, and thus are interpreted to represent the coexistence of all three borosilicate minerals at peak metamorphic conditions. 4. ANALYTICAL METHODS 4.1. Electron microprobe analysis (EMPA) Compositions of tourmaline, prismatine and grandidierite in the 21 paragneisses, (Table 1) as well as of the prismatine and grandidierite used as standards for SIMS (see Section 4.2.), were determined using the Cameca SX100 electron microprobe using wavelength dispersive spectroscopy (WDS) at the University of Maine. The operating conditions were a 15 kV acceleration voltage, 10 nA beam current and a 5 lm beam diameter, with the standards polylithionite (FKa), tugtupite (NaKa, ClKa), diopside (MgKa, CaKa), kyanite (AlKa, SiKa in grandidierite and prismatine); anorthite (AlKa, SiKa in tourmaline), sanidine (KKa), rutile (TiKa), Cr2O3 (CrKa), rhodonite (MnKa),and almandine (FeKa). Data were processed using the X-Phi correction of Merlet (1994). All iron was assumed to be Fe2+, and B contents were assumed. Tabulated analyses

268

J.R. MacGregor et al. / Geochimica et Cosmochimica Acta 123 (2013) 261–283

for a given section are averages of several grains, each of which was analyzed at 10 spots. The reader is referred to Wadoski (2009) and Wadoski et al. (2011) for compositions of tourmaline, prismatine and grandidierite in the anatectic pegmatites. 4.2. In-situ secondary ionization mass spectroscopy (SIMS) Boron isotopes were measured in situ in tourmaline, prismatine and grandidierite by secondary ionization mass spectroscopy (SIMS) using a Cameca ims-4f ion microprobe at the Edinburgh Ion Microprobe Facility. Secondary ions were generated by sputtering with a 3 nA 16O primary ion beam of 20 lm in diameter. Analytical spots were pre-sputtered for 2 min and then were analyzed for twenty acquisition cycles with integration times of 8 s for 10 B and 2 s for 11B during each cycle. No energy filtering was applied to the secondary ion signal. We used a mass resolution of M/DM = 1200 (defined as 11B peak width at 5% peak height) to resolve the peaks for 10B and 9Be1H and the peaks for 10B1H and 11B. The 10B1H/10B signal ratio was monitored during each analysis and varied from 0.01% to 0.16% depending on the sample. Performing the analysis without resolving molecular interferences would have added a 0.4 per mil uncertainty to the results, which underlines the need for using high mass resolution. Secondary ion detection was conducted by a single electron multiplier with a dead time of 14.5 ns. Dead-time corrected 10B and 11B count rates were converted to a 11B/10B ratio for each cycle by dividing the 11B signal by the 10B signal from the current and previous cycles to offset signal drift during analysis. Internal precision is calculated as the standard erp ror of the mean (1s/ n) of all measured cycles. Boron isotope composition is expressed as per mil deviation from the standard SRM 951 boric acid in 11 delta notation: d11B (= {[sample B/10B  11 10 SRM951 B/ B]  1}  1000). The unknowns were run with four boron isotope standards analyzed using thermal ionization mass spectrometry (TIMS) for B isotopes by Leeman and Tonarini (2001) and by electron microprobe and other methods by Dyar et al. (2001) for major constituents: elbaite No. 98144, dravite No. 108796, schorl No. 112566, and prismatine No. 112233; and three new secondary standards analyzed using the same TIMS method by Samuele Agostini (personal communication, 2011): two of prismatine and one of grandidierite from the Larsemann Hills (Table 1). During a typical daily session, all standards were run at least twice, whereas elbaite No. 98114 was repeated at least eight times throughout the day to monitor mass drift (see below). Other investigators who used the same elbaite, dravite and schorl standards reported matrix-dependent mass fractionation of 2& or less, i.e., comparable to other errors, and thus averaged the measured ainst for use in calibration (e.g., Marschall, 2005; Marschall et al., 2006; Ludwig et al., 2011; Cabral et al., 2012). In contrast, we found non-negligible matrix-dependent mass fractionations for tourmaline, as compositions were offset by 3.6& for schorl 112566 and +1.6& for elbaite No. 98114 relative to dravite No. 108796 (Table 2). Chaussidon and Albare`de (1992) reported

matrix-dependent mass fractionation of on average 15& between elbaite and schorl, although this was a different set of standards than the ones used here. There is also a significant matrix-dependent mass fractionation for prismatine as the Fe-bearing samples SH/06/01R and SH/06/10 show offsets of 4.1& and 4.9&, respectively, compared to No. 112233, which contains little Fe (XMg = Mg/(Fe2+ + Mg = 0.96). The tourmaline unknowns vary little in major element chemistry except for Na/Ca ratio (see below) and are closest to dravite 108796 in terms of Mg/Fe2+ ratio assuming Fe is predominantly Fe2+ in the unknowns, i.e., XMg = 0.70 vs. 0.64–0.78, and consequently we used the dravite as the primary calibration standard for the tourmaline unknowns. Similarly, we used SH/06/01R and SH/06/10 prismatine standards, which have XMg = 0.66–0.67, close to our prismatine unknowns with XMg = 0.63–0.77. Finally, the grandidierite standard from the Stornes Peninsula (SH/06/22, Table 2), which has a major-element chemical composition similar to our samples, was used to correct instrumental mass fractionation for grandidierite. Instrumental mass drift was monitored by repeatedly measuring one tourmaline standard (elbaite No. 98114; 11 B/10B = 4.001; Leeman and Tonarini, 2001) during each analytical session, typically 2 or 3 repeats each time we re-analyzed the standards, generally 3 or 4 times each day. This elbaite standard was chosen for this purpose because it gave the smallest standard deviation on initial homogeneity tests. As the mass fractionation between elbaite and dravite was constant throughout the analytical session, reliance on elbaite and dravite is expected to have introduced a smaller error than using a standard that is less homogeneous. Typical drift during each daily session (generally 10–12 h) of instrumental mass fractionation was 2 permil due to aging of the detector, which was therefore corrected assuming it was linear with total cumulative counts on the detector. Compilation of all drift-corrected measurements of the elbaite standard gives 1r uncertainty of 1.1& (n = 94). In a given section, each mineral was analyzed at 4–10 spots depending on microstructures and abundance, commonly at more than one spot per grain. Instrumental standard deviation ranged from 0.2& to 1.6&. Using the program Isoplot, measurements were weighted according to the standard deviations (2r) on each grain and averaged over the 5–12 grains in a given sample to give an average for the sample. The uncertainties at the 95% confidence level for sample averages ranged from 0.4& to 2.9&. In two cases, different tourmaline microstructures yield significantly different B isotope compositions and these are reported separately, whereas in cases of matrix tourmaline vs. included tourmaline vs. overgrowth tourmaline, differences are within the uncertainties, i.e., cannot be resolved by the ion microprobe SIMS method, so the average for all grains is given. Two clear cut outlier datasets obtained on one tourmaline grain each in sections 121202C and 011704C, as well as one outlier prismatine grain in section 120102G, for a total of 5 spots, were rejected for having been located too close to the edge of the section, where the beam is affected by the sample holder; nonetheless,

J.R. MacGregor et al. / Geochimica et Cosmochimica Acta 123 (2013) 261–283

269

Table 2 Boron isotope compositions and ratios for the ion microprobe standards. Mineral number

NBS 951

Elbaite 98144

Schorl 112566

Dravite 108796

Prismatine 112233

Prismatine SH/ 06/01R

Prismatine SH/06/10

Grandidierite SH/06/22

TIMS 11B/10B TIMS d11B (&) Data source Measured 11 B/10B Measured d11B (&) 1s (&) n aIMF amtx

4.0436

4.0014 10.4

3.9931 12.5

4.0169 6.6

4.0002 10.7

3.9799 15.8

3.9881 13.7

4.0126 7.7

1

2 3.9947

2 4.0077

2 4.0169

2 3.9815

3 3.9784

3 3.9896

3 4.0126

12.1

8.9

6.6

15.4

16.1

13.4

7.7

1.1 93 0.9647 0.9983

1.1 24 0.9647 1.0037

1.6 42 0.9647 1.0000

1.4 67 0.9630 0.9953

1.8 41 0.9630 0.9996

1.1 24 0.9630 1.0004

1.5 27 0.9621 1.0000

Note: measured ratio = raw ratio/aIMF, where aIMF is determined for each mineral, using dravite for tourmaline, SH/06/01R and SH/06/10 for prismatine, and SH/06/22 for grandidierite. amtx = measured ratio/TIMS ratio. Sources: (1) Catanzaro et al. (1970); (2) Leeman and Tonarini (2001); (3) Samuele Agostini (personal communication).

additional entries giving the averages with these outliers are also included in the tables. In addition, we rejected the B isotope data from section 121301B. This section had to be trimmed to afford access to critical grains, as a result of which the section started to peel from the glass plate, which is likely to have distorted the impact field of the primary ion beam. Total counts on prismatine varied by a factor of three. 5. MICROANALYSIS RESULTS 5.1. Major element composition 5.1.1. Tourmaline Tourmaline formulae were calculated on the basis of 24.5 oxygen excluding B and (OH + F + Cl) using the idealized formula X(Na,Ca, h)Y(Fe,Mg)3Z(Al)6(BO3)3TSi6O18V (OH)3W(OH,F,Cl), where T, V, W, X, Y and Z are crystallographic sites and h denotes cation vacancy (Henry et al., 2011). H2O content was calculated assuming (OH + F + Cl) = 4 anions per formula unit (apfu). All B is assumed to be trigonally coordinated, and its content is assumed to be stoichiometric. EPMA analyses of tourmaline (Table 3, Fig. 5) reveals that compositions are highly variable from sample to sample, and in some cases, from grain to grain in a given sample, in terms of Na/(Na + Ca), which ranges from 0.32 to 0.80. In contrast, XMg = Mg/(Mg + Fe) only ranges from 0.65 to 0.79. Tourmaline is thus dominantly dravite, i.e., XMg = Mg/(Mg + Fe) > 0.5, (Na + K) > Ca at the X-site and the X site is 89.3–98.6% occupied (Fig. 5). Uvite, in which (Na + K) < Ca, is found only in sections 121801A and 121602C. Titanium content ranges from 0.041 to 0.268 apfu, and is highest in the tourmaline metaquartzites and borosilicate gneisses. Analytical totals range from 99.75 to 103.4 wt% and average 101.7 wt% (e.g., Table 3), despite careful selection and checking of standards. Other than undetected analyti-

cal error, the most plausible explanation is that water content was overestimated. Assuming that OH + F + Cl = 4 apfu, calculations of the formulae give total cations (excluding the X-site occupants Na, K and Ca) ranging from 17.82 to 17.97 apfu and averaging 17.88 apfu, which is below the ideal cation content of 18 assuming only the X site is partially vacant. If OH + F + Cl = 3.6 apfu and the idealized formula written X(Na,Ca, h)Y(Fe,Mg,Ti)3Z (Al)6(BO3)3Si6O18V(OH)3W(OH,F,O), then calculated analytical totals would be lower by nearly 0.5 wt%. Thomson (2006) cited compositional data as evidence that substitution of Mg and Fe by Ti results in deprotonation of dravite in granulite-facies rocks in south-central Massachusetts, USA. Substitution of Mg and Fe2+ by Al or Fe3+ (an oxy-tourmaline component, Henry et al., 2011) and substitution of Al by Ti could also contribute to deprotonation. 5.1.2. Prismatine Prismatine structural formulae were calculated on the basis of 20.228 oxygens excluding B and (OH + F) using the general formula (h,Mg,Fe,Na)(Al,Mg,Fe)9(Si,B,Al)5O21(OH,F), where h indicates vacancy (Table 4). Water content was calculated assuming that OH + F + Cl = 1 apfu. Since the tetrahedrally coordinated B is non-stoichiometric, we assumed its content to be 0.848 apfu on the basis of the crystal refinement of a prismatine from the Larsemann Hills reported by Cooper et al. (2009), whence the fractional number of O used to calculate the formulae. Prismatine grains show relatively little variation in XMg = Mg/(Mg + Fe), which ranges from 0.63 to 0.77, either within a section or from sample to sample. A traverse across a prismatine porphyroblast in section 112701C shows some zonation in terms of Fe and Mg (Fig. 6). 5.1.3. Grandidierite Grandidierite structural formulae were calculated on the basis of 7.5 oxygens excluding B using the formula (Mg, Fe2+)Al3BSiO9 (Table 5). B is trigonally coordinated and

270

J.R. MacGregor et al. / Geochimica et Cosmochimica Acta 123 (2013) 261–283

Table 3 Selected electron microprobe analyses of tourmaline. Sample Rock type Grain No. anal

121202D Tur Qtzite 1 10

120902A Tur Qtzite 5 10

012101M B-sil Gn 3 10

121301B B-sil Gn 3 10

120102G Leuco 2 10

120603F2 Leuco 3 10

121801A Q-under 5 10

121602C Fe-rich 17 10

Weight% SiO2 TiO2 B2O3calc Al2O3 Cr2O3 MgO CaO MnO FeO Na2O K2O F Cl H2Ocalc O = F, Cl Total

35.39 1.69 10.63 31.00 0.08 8.10 1.80 0.00 6.77 1.94 0.10 0.47 0.01 3.44 0.20 101.22

35.28 2.03 10.69 31.41 0.02 8.30 2.29 0.01 6.01 1.59 0.10 0.66 0.01 3.37 0.28 101.49

35.18 2.11 10.66 30.51 0.38 9.09 2.62 0.01 5.40 1.45 0.10 0.97 0.01 3.22 0.41 101.31

35.33 1.80 10.71 31.03 0.77 8.74 1.89 0.02 5.86 1.77 0.11 0.75 0.00 3.34 0.32 101.80

35.43 1.48 10.58 29.66 0.01 8.86 2.44 0.00 7.33 1.61 0.07 0.83 0.02 3.25 0.35 101.21

35.01 1.45 10.53 30.02 0.07 8.55 2.34 0.02 7.42 1.57 0.10 0.69 0.02 3.30 0.30 100.78

34.81 0.53 10.52 30.34 0.01 8.50 3.53 0.02 8.14 0.93 0.10 0.60 0.00 3.35 0.25 101.12

35.00 1.02 10.54 29.43 0.03 9.28 3.40 0.03 7.56 1.10 0.11 0.70 0.03 3.30 0.30 101.22

Formulae per 24.5 O (excluding B and H) Bcalc 3.000 3.000 Si [iv]

Al Sum T Ti [vi] Al Cr Mg Mn Fe Sum Y + Z Na K Ca hcalc Sum X Cation sum F Cl OHcalc Anion sum XMg XCa

5.822 0.178 6.000 0.183 5.566 0.001 2.171 0.000 1.007 8.929 0.513 0.014 0.430 0.043 1.000 18.886 0.432 0.004 3.564 4.000 0.683 0.456

5.738 0.262 6.000 0.248 5.758 0.003 2.013 0.002 0.817 8.842 0.501 0.022 0.399 0.078 1.000 18.764 0.342 0.001 3.657 4.000 0.711 0.444

3.000

3.000

3.000

3.000

3.000

3.000

5.734 0.266 6.000 0.259 5.597 0.049 2.209 0.001 0.737 8.852 0.459 0.022 0.458 0.061 1.000 18.791 0.499 0.002 3.498 4.000 0.750 0.499

5.736 0.264 6.000 0.220 5.673 0.099 2.116 0.002 0.796 8.907 0.558 0.022 0.328 0.091 1.000 18.815 0.387 0.001 3.612 4.000 0.727 0.370

5.822 0.178 6.000 0.183 5.566 0.001 2.171 0.000 1.007 8.929 0.513 0.014 0.430 0.043 1.000 18.886 0.432 0.004 3.564 4.000 0.683 0.456

5.780 0.220 6.000 0.180 5.621 0.009 2.103 0.002 1.024 8.939 0.503 0.021 0.414 0.062 1.000 18.878 0.360 0.007 3.633 4.000 0.673 0.451

5.751 0.249 6.000 0.065 5.658 0.002 2.094 0.002 1.124 8.945 0.297 0.020 0.625 0.057 1.000 18.888 0.313 0.000 3.687 4.000 0.651 0.678

5.773 0.227 6.000 0.126 5.493 0.004 2.280 0.004 1.043 8.951 0.352 0.023 0.601 0.024 1.000 18.927 0.363 0.010 3.627 4.000 0.686 0.630

Note: Tur Qtzite – tourmaline metaquartzite, B-sil Gn – borosilicate gneiss, Leuco – leucogneiss. Q –under – quartz-undersaturated All Fe as FeO. h – cation vacancy. XMg = Mg/(Mg + Fe), XCa = Ca/(Ca + Na).

its content is assumed to be stoichiometric. Grandidierite exhibits a narrow range of XMg from 0.76 to 0.83, and is relatively homogenous from grain to grain in a sample. 5.1.4. Distribution of major constituents Eleven metamorphic rocks were analyzed with associated prismatine and tourmaline, eight of which show a relatively consistent distribution of Mg and total Fe between tourmaline (Tur) and prismatine (Prs) and give a KD (Mg/Fe Prs-Tur) = (Mg/Fe)Prs/(Mg/Fe)Tur = 1.066 or nearly equal distribution for the inferred trend, whereas

the three outliers show greater fractionation of Mg relative to Fe into tourmaline (Fig. 7a). Compositions of prismatine and tourmaline in two pegmatite samples are close to the inferred trend. Tourmaline and grandidierite (Gdd) are present together in six samples, four of which give KD (Mg/Fe Gdd-Tur) = 1.538 for the inferred trend (Fig. 7b). Sample 121103B was included in calculating the inferred trend in KD after the data for a tourmaline rim around grandidierite were taken out. There are three sections containing associated grandidierite and prismatine, too few for calculating a meaningful KD. A theoretical KD (Mg/Fe

J.R. MacGregor et al. / Geochimica et Cosmochimica Acta 123 (2013) 261–283

271

Fig. 5. X-site occupancy of tourmaline analyzed for B isotopes.

Gdd-Prs) = 1.443 can be calculated from the inferred trends cited above: magnesium-Fe distribution in samples 120902I and 012101M gives a nearly coincident preferred trend, whereas sample 121301B plots away from both trends (Fig. 7c). Fluorine-OH distribution between prismatine and tourmaline is less regular with KD (F/OH Prs-Tur) = 0.806 (r2 = 0.704) for the samples giving the inferred trend in Mg-Fe distribution. Adding the three off-trend pairs in Fig. 7a has a negligible effect: KD (F/OH Prs-Tur) = 0.839 (r2 = 0.718). 5.2. Boron isotopic composition

on tourmaline in a nodule of graphic tourmaline-quartz intergrowth on northern Stornes Peninsula. In our experience, such intergrowths are characteristic of the pegmatites. Three of our samples show differences exceeding uncertainties in isotopic composition between microstructural varieties: (1) 7.0 ± 0.9& in matrix tourmaline vs. 9.1 ± 0.7& in tourmaline filling fractures in prismatine in sample 121105 (Fig. 4d), (2) 6.5 ± 0.6& in tourmaline filling fractures in one grain of orthopyroxene in sample 112704B4, but 10.5 ± 0.6& in tourmaline in a second orthopyroxene grain, and (3) 10.1 ± 0.7& for tourmaline in graphic intergrowth with quartz vs. 7.2 ± 0.4& for tourmaline overgrowing prismatine in pegmatite sample 12110C1.

5.2.1. Tourmaline Average tourmaline d11B ranges from 2.8& to 14.4& in 20 metamorphic rocks (Table 6, Fig. 8), but the ranges for individual rock types differ from one another. For example, tourmaline metaquartzite d11B varies minimally from sample to sample (5.9& to 8.6&), whereas average borosilicate gneiss tourmaline ranges more widely, 2.8& to 11.2&, and leucogneiss tourmaline is lighter, 9.6& to 14.4& (Fig. 8). d11B of tourmaline in graphic intergrowths with quartz in the anatectic pegmatites ranges from 4.6& to 11.9&, mostly heavier than the 12.0 ± 0.3& that Wang et al. (2004) obtained by TIMS

5.2.2. Prismatine Average prismatine d11B ranges from 9.7& to 17.8& in 12 metamorphic rocks and two pegmatites (Table 7), compositions which overlap the compositions of two reference prismatines measured by TIMS (Table 2), 15.75 ± 0.12& (Seal Cove) and 13.73 ± 0.07& (Stornes), but not those Wang et al. (2004) obtained by TIMS on metamorphic prismatine: 26.2 ± 0.7& and 34.6 ± 0.7& from prismatine–biotite–plagioclase gneiss on northern Stornes Peninsula and from an unspecified rock in Seal Cove on Broknes Peninsula (Fig. 2), respectively. As is the case of most tourmaline, leucogneiss

272

J.R. MacGregor et al. / Geochimica et Cosmochimica Acta 123 (2013) 261–283

prismatine (d11B = 13.6& to 17.8&) contains lighter B than borosilicate gneiss prismatine (d11B = –9.6 to –16.9& Fig. 8). In a traverse across a prismatine porphyroblast d11B is greater in the core relative to the rims. Although the difference is less than 2r of the measurements at individual spots (Fig. 6), it could be greater than the combined uncertainty and, thus, the porphyroblast could be considered slightly zoned in both d11B and Fe/Mg ratio. 5.2.3. Grandidierite Grandidierite in the six borosilicate gneisses has a d11B composition that ranges from 1.4& to 8.7& (Table 8, Fig. 8), which overlaps the reference grandidierite measured by TIMS, 7.7 ± 0.6& (Table 2), but not that Wang et al. (2004) obtained by TIMS on metamorphic grandidierite: 30.5 ± 0.5& from grandidierite-prismatine-quartz gneiss in Wilcock Bay, presumably the same area where our samples 1–4 were collected (Fig. 2). Grandidierite in two anatectic pegmatites displays a narrow range of d11B, from 6.4 ± 0.9& to 7.3 ± 1.6&. 6. DISCUSSION 6.1. Comparison with TIMS data from the Larsemann Hills In summary, our ion microprobe boron isotope compositions for metamorphic prismatine and grandidierite are consistent with TIMS compositions on reference standards of these minerals (Table 2) from similar rocks in the Larsemann Hills. However, our SIMS and TIMS data diverge significantly from the TIMS compositions for metamorphic prismatine and grandidierite measured by Wang et al. (2004), who reported d11B about 20–30& more negative than ours. In contrast, our d11B values on tourmaline from intergrowths with quartz almost overlap with the d11B reported by Wang et al. (2004) for such tourmaline. Marschall and Ludwig (2006) encountered an analogous situation in their unsuccessful attempt to replicate using SIMS the highly negative TIMS d11B values ranging from 37.2& to 21.3& reported by Jiang et al. (2003) for tourmaline from the Lavicky leucogranite, Czech Republic. Marschall and Ludwig (2006) obtained less negative and less variable d11B of 10.77 ± 1.24& and suggested that the discrepancy might be due to difficulties experienced by Jiang et al. (2003) during sample preparation instead of during mass spectrometry because tourmaline is such refractory mineral. Jiang (2006) attributed the problems to the use of an HF dissolution digestion method, during which volatile BF3 may escape, leading to B fractionation. Wang et al. (2004) used an alkali fusion method, which would avoid B evaporation. Nonetheless, it is still possible that prismatine and grandidierite could have behaved differently than tourmaline using this method, which might explain why our results are much more discrepant for these minerals than for tourmaline when compared with the isotopic compositions reported by Wang et al. (2004). Because of these discrepancies we have not included the d11B reported by Wang et al. (2004) in our discussion and Fig. 8.

6.2. Distribution of boron isotopes The ion microprobe analyses reported in this study are the first measurements of B isotopes among associated tourmaline, grandidierite, and prismatine. This provides a unique opportunity to both determine the equilibrium distribution under conditions of the granulite facies where all three minerals appear to have been stable, and to assess the factors affecting B isotopic fractionation. Fig. 9 and Table 9 give the distributions expressed as D11BA–B (= d11BA  d11BB) for 18 pairs in 14 samples in which two or three borosilicate minerals could be measured; the three pairs in sample 121301B were discarded due to damage to the section (see Section 4.2.). D11BA–B is reported as an average of the d11BA  d11BB data with errors given at the 95% confidence level and with individual d11BA  d11BB weighted according to the errors in the individual d11BA  d11BB values. Whether the observed isotopic distributions represent equilibrium depends on whether there is independent evidence for equilibrium, specifically microstructural and chemical. Microstructural equilibrium was assessed by petrographic observation. The best indication of microstructural equilibrium would be that grains form polygonal aggregates (Vernon, 2004). Such clear cut evidence is not often observed, nonetheless, we interpret the microstructures illustrated in Fig. 3 to represent equilibrium. Relationships such as overgrowths of one mineral around another or inclusions of one mineral in another are commonly assumed to indicate disequilibrium. However, this assumption does not appear to be valid in all cases in the borosilicate rocks, and thus we have looked for additional evidence for disequilibrium. Examples considered to indicate disequilibrium include (1) complete isolation of grandidierite by prismatine in sample 121301B (Fig. 4a), (2) differences in grain size combined with tendency to be isolated as is tourmaline in 121801A (Fig. 4b), and (3) fracture fillings such as finely fibrous grandidierite in orthopyroxene (Fig. 4c) or tourmaline in prismatine (Fig. 4d). Six of the 18 pairs exhibit microstructural disequilibrium, including both tourmaline-prismatine pairs from the pegmatites. One measure of chemical equilibrium is distribution of Mg and Fe (total) using the partition coefficient between one borosilicate and another, e.g., KD (Mg/Fe PrsTur) = (Mg/Fe)Prs/(Mg/Fe)Tur, which is reasonably regular (Fig. 7). Three of the six pairs inferred to represent microstructural disequilibrium appear as outliers in Fig. 7, as do two pairs showing microstructural equilibrium. Another measure of chemical equilibrium is Na–Ca distribution between tourmaline and plagioclase, which is not regular (MacGregor, 2012); the wide range in tourmaline Na/Ca ratio in a given sample (Fig. 5) also indicates that distribution of Na and Ca did not attain equilibrium in most samples. In summary, five of 12 tourmaline–prismatine pairs, three of four tourmaline pairs and both prismatine–grandidierite pairs show evidence for microstructural equilibrium and chemical equilibrium assessed in terms of Mg and Fe distribution. These 10 pairs are thus likely to represent

J.R. MacGregor et al. / Geochimica et Cosmochimica Acta 123 (2013) 261–283

273

Table 4 Selected electron microprobe analyses of prismatine. Sample Rock type Grain No. anal

SH/06/01R Std – 10

SH/06/22 Std – 10

123001L1 Tur Qtzite 17 10

012101M B-sil Gn 24 10

121301B B-sil Gn 2 10

120102G Leuco 27 10

120603F2 Leuco 20 10

121801A Q-under 5 10

121602C Fe-rich 4 10

112701C Porph 4 10

Weight% SiO2 TiO2 B2O3calc Al2O3 Cr2O3 MgO CaO MnO FeO Na2O K2O F Cl H2Ocalc O = F, Cl Sum

29.35 0.28 3.88 40.43 0.01 12.77 0.05 0.16 11.91 0.08 0.00 0.67 0.00 0.87 0.28 100.16

29.62 0.23 3.84 38.57 0.07 13.50 0.03 0.06 11.81 0.06 0.00 0.67 0.01 0.85 0.28 99.02

29.73 0.25 3.94 41.82 0.05 14.42 0.02 0.01 7.95 0.13 0.00 1.23 0.01 0.62 0.52 99.64

30.21 0.31 3.95 40.50 0.18 15.39 0.02 0.07 8.37 0.07 0.00 1.15 0.01 0.66 0.48 100.39

29.35 0.31 3.91 41.00 0.15 14.57 0.02 0.04 8.89 0.08 0.00 1.00 0.01 0.72 0.42 99.62

29.91 0.21 3.85 37.68 0.02 14.85 0.03 0.08 11.16 0.06 0.00 0.75 0.00 0.82 0.32 99.09

29.29 0.21 3.81 38.17 0.00 13.16 0.03 0.04 12.67 0.05 0.00 0.62 0.01 0.87 0.26 98.64

28.74 0.06 3.82 38.96 0.01 12.90 0.05 0.09 13.53 0.02 0.01 0.30 0.00 1.02 0.13 99.36

29.68 0.16 3.84 37.26 0.03 15.17 0.08 0.06 11.57 0.04 0.00 0.55 0.02 0.90 0.24 99.13

29.53 0.23 3.86 38.91 0.08 13.58 0.04 0.09 12.00 0.04 0.00 0.49 0.02 0.94 0.21 99.59

3.754 0.029 0.848 5.932 0.017 2.851 0.002 0.007 0.869 0.017 0.000 14.327

3.685 0.029 0.848 6.067 0.015 2.726 0.002 0.004 0.933 0.019 0.000 14.330

3.820 0.020 0.848 5.672 0.002 2.827 0.003 0.009 1.192 0.015 0.000 14.407

3.781 0.020 0.848 5.807 0.000 2.533 0.003 0.004 1.368 0.011 0.000 14.377

3.700 0.006 0.848 5.910 0.001 2.475 0.006 0.009 1.456 0.004 0.001 14.417

3.800 0.015 0.848 5.623 0.003 2.896 0.011 0.007 1.239 0.010 0.000 14.452

3.764 0.022 0.848 5.844 0.008 2.581 0.006 0.009 1.279 0.010 0.000 14.370

0.450 0.002 0.548 1.000 0.766

0.398 0.001 0.601 1.000 0.745

0.304 0.000 0.696 1.000 0.703

0.252 0.002 0.746 1.000 0.649

0.121 0.000 0.879 1.000 0.630

0.224 0.004 0.772 1.000 0.700

0.197 0.004 0.799 1.000 0.669

Formulae per 20.228 O (excluding B and H) Si 3.714 3.792 3.709 Ti 0.027 0.022 0.024 Bcalc 0.848 0.848 0.848 Al 6.029 5.820 6.150 Cr 0.001 0.007 0.005 Mg 2.409 2.576 2.682 Ca 0.006 0.004 0.003 Mn 0.017 0.006 0.001 Fe 1.261 1.265 0.829 Na 0.019 0.014 0.030 K 0.000 0.000 0.000 Cation sum 14.330 14.355 14.280 F Cl OH Anion sum XMg

0.268 0.001 0.731 1.000 0.656

0.270 0.002 0.728 1.000 0.671

0.485 0.003 0.512 1.000 0.764

Note: Std – used as standard for ion microprobe analyses, Tur Qtzite – tourmaline metaquartzite, B-sil Gn – borosilicate gneiss, Leuco – leucogneiss. Q-under – quartz-undersaturated. Porph – porphyroblast in biotite gneiss. All Fe as FeO. XMg = Mg/(Mg + Fe).

isotopic equilibrium (Fig. 9). d11B decreases grandidierite > tourmaline > prismatine, with D11BGdd-Prs close to +7.2& and D11BGdd-Tur close to +3.3&, whether all pairs are considered or only those pairs more likely to be equilibrium, as indicated by other evidence for equilibrium. Fractionation of 11B into tourmaline and grandidierite relative to prismatine is consistent with the preference of 11 B for trigonally coordinated sites, and 10B for tetrahedrally coordinated sites (e.g., Kakihana et al., 1977). However, the fractionation between tourmaline and grandidierite cannot be explained by differences in coordination. Because isotopic fractionation is correlated with the B–O bond length of which boron coordination is one factor, we performed ab initio calculations of B isotope fractionation factors following the method developed by Kowalski and Jahn (2011) and Kowalski et al. (2013). We have applied the same pseudofrequency analysis technique

that was successfully applied in recent investigation of B isotope fractionation between tourmaline, boromuscovite and aqueous fluids (Kowalski et al., 2013). We used plane-wave density functional theory (DFT) code CPMD (Marx and Hutter, 2000) with BLYP generalized gradient approximation (Becke, 1988; Lee et al., 1988), energy cutoff of 140Ryd and norm conserving Goedecker pseudopotentials for the description of the core electrons (Goedecker et al., 1996). The beta factor (a factor describing isotope fractionation between species and an ideal monoatomic gas) for tourmaline (dravite) is the one computed in Kowalski et al. (2013). The beta factors for grandidierite and prismatine were computed using the crystal lattice parameters of Dzikowski et al. (2007) and Cooper et al. (2009), respectively. The computed super-cells contained 120 atoms for grandidierite (Mg8Al24Si8B8O72) and 146 atoms for prismatine (Mg14Al24Si16B4O88) and the ions were relaxed

274

J.R. MacGregor et al. / Geochimica et Cosmochimica Acta 123 (2013) 261–283

Fig. 6. Photomicrograph (plane polarized light) of a traverse across prismatine porphyroblast in sample 112704C, a biotite gneiss (bottom panel), together with electron microprobe data on Fe and Mg (top panel) and ion microprobe data on B isotopes (middle panel) collected at p six points across the porphyroblast. The uncertainty in d11B at each point is 2r, where r is the standard error of the mean (1s/ n) of all measured cycles for this point. Bt – biotite, Mgt – magnetite, Pl – plagioclase.

to the equilibrium positions. The error of the computed fractionation factors was estimated following the procedure outlined in Kowalski and Jahn (2011) and Kowalski et al. (2013). The values of the B isotope fractionation factors computed assuming T = 1000 K between tourmaline, grandidierite and prismatine agree within the uncertainties with the measured values (Table 9; Fig. 9), providing theoretical confirmation of the measured fractionations. The unexpected small enrichment of grandidierite in 11B relative to tourmaline is related to the computed, relaxed ionic struc˚ ) B– ture of grandidierite, which results in shorter (1.368 A ˚ , Kowalski O bond length than in case of dravite (1.385 A et al., 2013). 6.3. The effect of metamorphism on boron content and boron isotope compositions 6.3.1. Retention of boron In contrast to most granulite-facies rocks, which are characteristically depleted in B (e.g., Leeman and Sisson, 2002), the Larsemann Hills granulite facies paragneisses include units unusually rich in B, up to 2 wt.% (Grew et al., 2013a). Most likely, high initial B concentrations hosted in tourmaline facilitated B retention with increasing grade of metamorphism. Tourmaline is more refractory than other common hosts of B such as clays and muscovite, which break down at relatively low temperatures, releasing

B that would be transported out of the system by aqueous fluids. Studies by Kawakami (2001, 2004) and Kawakami and Ikeda (2003) show that in pelitic rocks with typical concentrations of B (ca. 120 lg/g B) in the Ryoke metamorphic belt, SW Japan, tourmaline breaks down in the upper amphibolites-facies, releasing B into anatectic melt. However, there are examples where tourmaline persists into the granulite facies, the most notable being the extensive tourmalinites (tourmaline metaquartzites) at Broken Hill, Australia (Slack et al., 1993). These occurrences are consistent with experimental studies showing that dravite tourmaline is stable up to nearly 900 °C at 0.5 GPa (e.g., van Hinsberg et al., 2011). At temperatures of 865–895 °C, dravite tourmaline breaks down to the high-temperature borosilicates, prismatine and grandidierite, as well as to cordierite, sapphirine and melt (Robbins and Yoder, 1962; Werding and Schreyer, 2002). In the Larsemann Hills rocks, the retention of B under granulite-facies conditions can be explained by the persistence of tourmaline in B-rich rocks such as the tourmaline metaquartzites together with the formation of grandidierite and prismatine as breakdown products of tourmaline. 6.3.2. Tourmaline breakdown reactions Tourmaline breakdown to prismatine and grandidierite is a potential driving force of isotopic fractionation only if tourmaline entirely recrystallizes in the process. We

J.R. MacGregor et al. / Geochimica et Cosmochimica Acta 123 (2013) 261–283

assume that tourmaline did recrystallize during the reaction if there is microstructural evidence for equilibrium together with regular distribution of Mg and total Fe (see Section 6.2.). Zoning is rare in individual tourmaline grains, consistent with equilibrium recrystallization of tourmaline, although there is variation in Na/Ca ratio from grain to grain (Fig. 5). Here we assume that the primary source of B for prismatine and grandidierite was tourmaline. Other possible Bbearing reactants, including B-bearing muscovite and other B-bearing layered silicates, as well as water-soluble borates in evaporates, would not have survived the upper-amphibolite-facies and granulite-facies temperatures needed for prismatine and grandidierite to crystallize in the aluminous, low-Ca paragneisses (Grew, 2002; Werding and Schreyer, 2002) that are the subject of this study. Possible reactions for tourmaline breakdown in granulite- facies rocks (Grew, 2002) are:

Table 5 Selected electron microprobe analyses of grandidierite. Sample Rock type Grain No. anal

SH/06/10 Std – 10

012101M B-sil Gn 30 10

121301B B-sil Gn 1 10

120902I B-sil Gn 4 10

Weight% SiO2 TiO2 B2O3calc Al2O3 Cr2O3 MgO CaO MnO FeO Na2O K2O Total

20.46 0.00 11.82 51.34 0.12 11.40 0.00 0.02 5.03 0.00 0.00 100.20

20.15 0.03 11.73 51.03 0.11 11.70 0.00 0.04 4.46 0.00 0.01 99.26

20.13 0.01 11.84 51.36 0.13 12.25 0.00 0.05 4.42 0.00 0.00 100.18

20.14 0.01 11.71 50.98 0.13 10.74 0.00 0.00 6.03 0.00 0.00 99.74

B) 0.995 0.001 1.000 2.969 0.004 0.861 0.000 0.002 0.184 0.000 0.001 6.018 0.824

0.985 0.000 1.000 2.963 0.005 0.894 0.000 0.002 0.181 0.000 0.000 6.030 0.832

Tourmaline + sillimanite + garnet or cordierite ! grandidierite + quartz + sodic plagioclase + H2O. Tourmaline + sillimanite + garnet ! prismatine + quartz + plagioclase ± H2O or melt.

0.996 0.000 1.000 2.972 0.005 0.792 0.000 0.000 0.249 0.000 0.000 6.015 0.761

Because the lighter B isotope fractionates into prismatine, tourmaline recrystallizing during breakdown to prismatine is expected to be enriched in the heavier isotope, whereas tourmaline recrystallizing during breakdown to grandidierite is expected to be enriched in the lighter isotope, as the heavier isotope preferentially partitions into the grandidierite. These predictions are consistent with the greater range in d11B in tourmaline from the borosilicate gneisses and leucogneisses compared to the narrow range in d11B in tourmaline from the tourmaline metaquartzite (Fig. 8). The gneisses contain relatively abundant grandidierite and prismatine as well as tourmaline, whereas the tourmaline metaquartzites typically have a uniform assemblage of tourmaline and quartz ± minor apatite and ilmenite. However, the range of d11B of tourmaline associated with prismatine in the leucogneisses, quartz-undersaturated rocks and the Fe-rich rocks is lower

Note: Std – used as standard for ion microprobe analyses, B-sil Gn – borosilicate gneiss. All Fe as FeO. XMg = Mg/(Mg + Fe).

would not expect the initial isotopic composition of B in tourmaline to be affected if the original tourmaline survived the reaction because of the refractory nature of tourmaline with respect to B isotopes (van Hinsberg et al., 2011; Marschall and Jiang, 2011). In this discussion we will

Inferred trend Off trend Pegmatite Linear (Inferred trend)

2.0

4.0

3.5

121801A

1.5

Off trend

3.0 2.5 Mg/Fe tourmaline

s) Pr

4.0

3.5

Inferred trend

120603F2

1.5

120902B1

)= 1 R 2 .45 =0 9M .97 g/F e( 8

M

3.5

2.0

2.5

3.0

Mg/Fe tourmaline

C

(G dd

2.5

121103B (Tur rim)

Calculated

4.5

/Fe

rs

(P

g/

4.5

Mg

Fe

0 M 96 66 0 1. )=

0.

(Tu r)

123001L1

/Fe

e

(G R 2 = dd ) = 0.98 1.5 2 38 Mg

R

=

121301B

Off trend

/Fe

2

Inferred trend

121301B

)

ur (T

F g/

3.0

5.0

B

Mg

Mg/Fe prismatine

3.5

5.0

A Mg/Fe grandidierite

4.0

Mg/Fe grandidierite

Formulae per 7.5 O (excluding Si 1.003 Ti 0.000 Bcalc 1.000 Al 2.965 Cr 0.005 Mg 0.832 Ca 0.000 Mn 0.001 Fe 0.206 Na 0.000 K 0.000 Cation sum 6.013 XMg 0.801

275

3.0 3.5

2.0

2.5

3.0

3.5

Mg/Fe prismatine

Fig. 7. Atomic Mg/Fe ratio in associated borosilicate minerals. Inferred trends (dashed lines) are based on least squares fits. (A) The inferred trend does not include the two pegmatite pairs (based on data from Wadoski, 2009). (B) The tourmaline rim in sample 121103B is not included in the average, but is plotted separately. (C) Calculated refers to a trend calculated from the preferred trends in A and B.

276

J.R. MacGregor et al. / Geochimica et Cosmochimica Acta 123 (2013) 261–283

Table 6 Boron isotopic composition of tourmaline. Sample 011704C (011704C 010205I 123001L1 010205A 121202C (121202C 121202D 120902A 120902I 012101M 120902B1 120902J 121103B 120102G 120603F2 011401D3 121602C 121801A2 112704B4 112704B4 121105 121105 112801D 120405D2 120902G2 121102C1 121102C1

d11B& 8.2 6.0 5.9 7.9 6.4 6.3 5.0 8.2 8.6 6.6 11.2 2.8 8.9 5.6 9.6 14.4 13.2 11.0 12.7 6.5 –11.1 9.1 7.0 10.0 11.9 4.6 7.2 –9.3

95% 1.4 3.5 0.8 1.8 1.0 0.5 1.5 1.1 1.6 1.5 1.2 1.7 1.1 1.0 0.6 1.4 0.9 1.3 1.7 0.6 1.1a 0.7 0.9 1.4 2.3 1.9 0.4 1.4

No. b

9 11c 9 11 6 8b 10c 6 6 6 8 5 7 7 12 8 6 8 4 5 1 4 4 5 6 6 4 6

Rock TQ TQ) TQ TQ TQ TQ TQ) TQ TQ BG BG BG BG BG LG LG LG FR QU QU QU BT BT P P P P P

Comments

Matrix and inclusions in Prs

Matrix and overgrowths on Prs Matrix and inclusions in Prs Matrix and overgrowths on Gdd Matrix and inclusions in Prs

Inclusions in Prs and Opx Inclusions in garnet. With biotite Fills fractures in Opx grain #1 Fills fractures in Opx grain #2 Fills fractures in Prs Matrix Overgrowths on Prs Intergrowth with Qtz Intergrowth with Qtz Overgrowths on Prs Intergrowth with Qtz

Notes: 95% = weighted uncertainty at 95% confidence level. No – number of analyses in the average, TQ – tourmaline metaquartzite, BG – borosilicate gneiss, LG – leucogneiss, QU – quartz-undersaturated rock, FR – Fe-rich rock, BT – biotite gneiss, P – pegmatite, Gdd – grandidierite, Opx – orthopyroxene, Prs – prismatine, Qtz – quartz. p a Uncertainty of 2r, where r is the standard error of the mean (1s/ n) of all measured cycles for this point. b Average excluding outliers attributed to points being too close to sample holder edge. c Average including outliers.

than the range of d11B of tourmaline in the metaquartzite; the range of d11B of tourmaline associated with grandidierite ± prismatine in the borosilicate gneisses overlaps it (Fig. 8). The simplest explanation for this result is that whole-rock d11B varied from one rock type to another, i.e., d11B in the leucogneisses, quartz-undersaturated rocks and Fe-rich rocks was lower than in the tourmaline metaquartzites and borosilicate gneisses. The impact of anatexis on d11B is difficult to assess, as the fractionation of isotopes between melt and tourmaline has not yet been determined experimentally (van Hinsberg et al., 2011). Citing similarity in d11B for paragneisses, granitoids and pegmatites in the central Andes, northwest Argentina, Kasemann et al. (2000) concluded that little fractionation is expected between tourmaline and granitic melt in this area. The overlapping d11B values for pegmatitic and metamorphic minerals (Fig. 8) suggest that a similar conclusion could be drawn for the Larsemann Hills. However, we calculate that fractionation between tourmaline and melt could be almost the same as rhyolite melt and fluid at the temperatures inferred for anatexis, 800– 860 °C, i.e., D11B = 4.5& (Fig. 10), by combining experi-

mental data on fractionation between rhyolite melt and fluid (Hervig et al., 2002) with fractionation between tourmaline and fluid (Meyer et al., 2008). In order to apply this D11B to the Larsemann Hills pegmatites, we have assumed that the ratio of tetrahedrally to trigonally coordinated B (IVB/IIIB) in the Larsemann Hills pegmatitic magma would have been roughly the same as in the experimental melts, because D11B between minerals and melt is sensitive to IVB/IIIB in the melt (Tonarini et al., 2003). A D11B of 4.5& implies that tourmaline crystallizing in the presence of melt would be enriched in the heavier isotope, thereby contributing further to the spread of d11B in tourmaline from the leucogneisses and borosilicate gneisses, rocks that would have more likely melted than the tourmaline metaquartzite. However, d11B in the resulting tourmaline is not higher; again implying that whole-rock d11B of the leucogneisses, quartz-undersaturated rocks and Fe-rich rocks was lower prior to melting than d11B of the tourmaline metaquartzites, while the whole-rock d11B of the borosilicate gneisses was the same or slightly higher. The overlapping d11B values for pegmatitic and metamorphic minerals can be interpreted to indicate that

J.R. MacGregor et al. / Geochimica et Cosmochimica Acta 123 (2013) 261–283

277

Fig. 8. Ranges of boron isotope composition of the borosilicate minerals by rock type (excluding the biotite gneisses and late tourmaline filling fractures) with filled circles representing individual samples (Tables 6–8). Tourmaline associations with prismatine (Prs) and grandidierite (Gdd) are distinguished by color of the filling.

Table 7 Boron isotopic composition of prismatine. Sample

d11B&

95%

No.

Rock

123001L1 120902I 012101M 120102G (120102G 120603F2 011401D3 121602C 121801A2 112704B4 121105 112701C 112801D 121102C1

14.2 9.6 –16.9 15.3 14.9 13.6 17.8 15.8 16.5 13.2 10.7 13.3 13.2 13.8

1.4 1.5 2.2 1.1 1.7 1.3 2.1 2.7 1.4 1.7 0.8 1.1 2.9 1.3

7 7 6 9a 10b 5 6 6 6 6 6 6 7 7

TQ BG BG LG LG) LG LG FR QU QU BT BT P P

Table 8 Boron isotopic composition of grandidierite. Comments

Sample

d11B&

95%

No.

Rock

120902I 012101M 120902B1 120902J 121103B 112906B 01205J5

2.8 8.7 3.4 5.5 3.9 6.4 7.3

0.4 0.9 2.2 0.8 0.7 0.9 1.6

7 5 5 7 8 6 6

BG BG BG BG BG P P

Notes: 95% = weighted uncertainty at 95% confidence level. No – number of analyses in the average, BG – borosilicate gneiss, P – pegmatite. Zoned (Fig. 6)

Notes: 95% = weighted uncertainty at 95% confidence level. No. – number of analyses in the average, TQ – tourmaline metaquartzite, BG – borosilicate gneiss, LG – leucogneiss, QU – quartz-undersaturated rock, FR – Fe-rich rock, BT – biotite gneiss, P – pegmatite. a Average excluding outliers attributed to points being too close to sample holder edge. b Average including outliers.

melting of tourmaline-bearing paragneiss and crystallization of tourmaline from melt together did not significantly fractionate B isotopes; in effect, fractionation during melting and fractionation during crystallization canceled out. 6.4. Source of boron The extensive studies of tourmaline (e.g., van Hinsberg et al., 2011; Marschall and Jiang, 2011) make it the most

suitable of the three borosilicate minerals on which to focus for constraining a source of B. We have selected the scenario proposed by Slack et al. (1993) for tourmaline metaquartzites (tourmalinites) associated with the Broken Hill stratiform lead–zinc-silver deposits in the Willyama Supergroup, Australia, as a point of departure for developing our scenario for the Larsemann Hills. Grew et al. (2013a) found that several distinctive rocks in Brattstrand Paragneiss in the Larsemann Hills have analogues in the Willyama Supergroup in the Broken Hill district, and thus possible precursors can be considered by comparing the Brattstrand Paragneiss with the more extensively studied Willyama Supergroup, despite some important differences, notably the Mn-rich garnet quartzite and gahnite-quartz rocks so typical of metamorphic rocks associated with the ore deposits at Broken Hill have not been reported in the Brattstrand Paragneiss. In brief, Slack et al. (1993) interpreted precursors of these tourmaline metaquartzites to be tourmaline-quartz

278

J.R. MacGregor et al. / Geochimica et Cosmochimica Acta 123 (2013) 261–283

Fig. 9. Plot of D11B = D11BA–B (= d11BA  d11BB), where phases A and B are tourmaline (Tur), prismatine (Prs) and grandidierite (Gdd). Filled symbols: distribution of B isotopes is inferred to be equilibrium according to the criteria in Table 9; unfilled symbols: distribution is not considered equilibrium from these criteria. Computed: fractionations are labeled with C.

rocks (tourmalinite) that resulted from pre-metamorphic, low-temperature hydrothermal B-metasomatism of clastic sediments with non-marine evaporite being the source of

B for the fluid. Fig. 11 (lines 1–3) illustrates schematically the evolution of B isotope composition in such a scenario: B extracted from evaporite by hydrothermal fluids would tend to be isotopically heavier than B in the evaporite, but changes in B isotope composition are difficult to quantify because factors other than isotopic fractionation, notably pH, also play important roles in the fractionation of B isotopes between borate minerals and fluids (Kasemann et al., 2004). Boron in tourmaline precipitated from the fluid in associated rocks is expected to be lighter than B in the fluid because of the preference of the lighter 10B isotope for tourmaline, which increases with decreasing temperature (Palmer et al., 1992; Meyer et al., 2008). Following Slack et al. (1993), we have selected tourmaline metaquartzites as the best rock type for constraining possible sources of B in the B-rich metamorphic rocks using B isotopes. The rationale for this choice is that the mineral assemblage in this rock type appears to be the least altered by metamorphism, and the tourmaline has the narrowest range of isotopic composition relative to the other rock types in this study (Fig. 8). Prismatine and grandidierite are either absent or much subordinate to tourmaline in this rock type, and thus the B isotope composition of tourmaline was minimally affected by recrystallization during

Table 9 Boron isotope fractionation and two criteria for equilibrium. Sample Tourmaline–Prismatine 123001L1 120902I 012101M 120102G 120603F2 011401D3 112704B4 121801A 121602C 121105 112801D (P) 121102C1 (P) Total (12 pairs) Equilibrium (5 pairs) Computed fractionation factor Tourmaline–Grandidierite 120902I 012101M 120902J 120902B1 121103B Total (5 pairs) Equilibrium (3 pairs) Computed fractionation factor

Microstr Y Y Y Y Y Y N N Y N N N

Chem

KD

D11B&

1s

N Y Y Y N Y Y N Y Y Y N

Tur/Prs 1.08 0.98 0.93 0.91 1.10 0.94 0.99 1.14 0.93 0.93 0.95 1.09

Tur-Prs +6.3 +3.0 +5.7 +5.7 –0.8 +4.6 +6.7 +3.8 +4.8 +1.6a +3.2 +6.6a +4.1 +5.0 +6.4

2.3 2.1 2.5 1.2 1.9 2.3 1.8 2.2 3.0 1.1 3.2 1.4 1.5 1.4 1.3

Y Y Y N Y

Tur/Gdd 0.65 0.65 0.64 0.74 0.66

Tur-Gdd –3.8 –2.5 –3.4 +0.6 –1.7 –2.5 –3.3 –1.8

1.5 1.5 1.4 2.8 1.2 1.6 0.8 1.1

Y Y

Gdd/Prs 1.50 1.44

Gdd-Prs +6.8 +8.2 +7.2 +8.2

1.5 2.4 1.3 1.5

at 1000 K Y Y Y N N

at 1000 K

Grandidierite–Prismatine 120902I Y 012101M Y Total and equilibrium (2 pairs) Computed fractionation factor at 1000 K

Note: Microstr – microstructural, Chem – chemical based on KD (Mg-Fe), (P) – pegmatite, Gdd – grandidierite, Prs – prismatine, Tur – tourmaline. Y - yes, N - no. a Used d11B of tourmaline in contact with prismatine (Table 6).

J.R. MacGregor et al. / Geochimica et Cosmochimica Acta 123 (2013) 261–283

breakdown to prismatine and grandidierite. In addition, the tourmaline metaquartzites probably underwent much less melting than feldspar-bearing borosilicate gneisses and leucogneisses. A third consideration is that isotopic fractionation resulted from interaction of tourmaline with an aqueous fluid in an open system where B and other constituents of tourmaline were lost from the rock. Assuming some open-system behavior in the Broken Hill rocks, Slack et al. (1993) attributed the 1–6& decrease in d11B between amphibolite-facies tourmaline (18& to 17&) and granulite-facies tourmaline (23& to 19&) to interaction with fluid. This decrease is greater than expected from experimental fluid-tourmaline fractionation at metamorphic temperatures of 700–800 °C when the fluid is sufficiently acid for B(OH)3 to be the dominant B species and for tourmaline to be stable, 2& (Palmer et al., 1992), which Slack et al. (1993) cited, or even lower according to newer data, 0.5& (Meyer et al., 2008). Slack et al. (1993) suggested that the discrepancy could be explained either by two-stage fractionation associated with two distinct metamorphic events or by d11B differences in the precursors to the rocks being compared. Given the difficulties in assessing the effect of devolatilization reported by Slack et al. (1993), we did not attempt to calculate the effect of metamorphism on tourmaline d11B, i.e., we assume closed-system behavior. Palmer and Slack (1989) concluded that although the effects of regional metamorphism on tourmaline d11B values could not be rigorously evaluated, it seemed reasonable to believe that tourmalines would be fairly resistant to isotopic fractionation during metamorphism due to slow diffusion rates (see also van Hinsberg et al., 2011; Marschall and Jiang, 2011); in any case, the metamorphic history would not be the dominant control determining d11B of tourmalines from metasedimentary rocks.

Fig. 10. Temperature-dependent fractionation of B isotopes between basaltic and rhyolitic (masucanite) and fluid from Hervig et al. (2002) and tourmaline-fluid from Meyer et al. (2008) in terms of D11B, the difference between the d11B of two minerals. Following London (2011), separate trend lines were calculated for basalt and rhyolite. The fractionation between tourmaline and rhyolite was calculated by subtracting the rhyolite-fluid trend line from the tourmaline fluid trend line.

279

The tourmaline B isotope compositions for the Larsemann Hills are shown in Fig. 11 to lie within the B isotope ranges of tourmaline inferred to be sourced from marine evaporites, non-marine evaporites, and clastic sediment (Palmer and Slack, 1989; Palmer, 1991). The Larsemann Hills tourmaline d11B also lies within the ranges given by Palmer and Slack (1989) for oceanic crust, which we have not considered because its low concentrations of B would yield low abundances of tourmaline (Palmer and Slack, 1989), and for volcanic rocks, which seem a less plausible source than sedimentary rocks due to the dominance of metasedimentary rocks in the Brattstrand Paragneiss (Fitzsimons and Harley, 1991; Carson and Grew, 2007; Grew et al., 2013a). A potential source of B not considered by Palmer and Slack (1989) and Palmer (1991) is B-rich mud volcanoes (Slack et al., 1998; Grew et al., 2013a), such as the onshore mud volcanoes described from the Kerch-Taman region, Ukraine and Russia, north of the Black Sea (Kopf and Deyhle, 2002; Kopf et al., 2003) or submarine mud volcanoes in the Black Sea basin (Slack et al., 1998). Boron isotopic composition in fluids of the onshore mud volcanoes in Kerch-Taman area are d11B = +16.5& to +42.4& (Kopf et al., 2003). Even allowing for marked fractionation between tourmaline and fluid under hydrothermal conditions, e.g., 5.4& to 3.8& at 200–300 °C (Meyer et al., 2008), d11B of tourmaline crystallizing from such fluids would be much heavier than B in the Larsemann Hills tourmaline, and thus we rule out mud volcanoes as a source of B for the Larsemann Hills tourmaline metaquartzites (cf. Grew et al., 2013a). Left with a choice among three alternatives of marine evaporites, non-marine evaporites, and clastic sediments, we must appeal to arguments other than boron isotopes to constrain the source of B. Bulk-rock compositions of some gneisses provide independent support for the presence of an evaporitic component in the Larsemann Hills metamorphic rocks (Grew et al., 2013a). For example, Larsemann Hills sodic leucogneisses could be metamorphosed volcanic detritus altered in an alkaline environment with B present, whereas the borosilicate gneisses could be metasediments with a large chemical component. A second consideration is the sheer abundance of B that is not associated with enrichment of other elements that would suggest a volcanic-exhalative source of B in precursors to the Larsemann Hills gneisses. If the hydrothermal fluids had been of volcanic exhalative origin, one would expect there to be enrichment of other elements normally found in exhalative deposits such as sulfides and rocks rich in Mn (spessartine), Zn (gahnite) or Fe (Slack, 2002), as reported from Broken Hill (Slack et al., 1993). Only Fe-rich rocks have been found in the Larsemann Hills, where there is just one locality (Fisher Island, Fig. 2). The Fisher Island Fe-rich rocks might have had an exhalative origin, but Grew et al. (2013a) conjectured these could be chemical sediments with little or no connection to volcanism. A source dominated by evaporite borate would supply little other than B in great abundance, which is what is observed in the Larsemann Hills with the possible exception of the Fisher Island Fe-rich rocks.

280

J.R. MacGregor et al. / Geochimica et Cosmochimica Acta 123 (2013) 261–283

Fig. 11. Boron isotope compositions of potential sources of boron and of tourmaline sourcing boron from these sources. Bar colors: black – borate mineral; blue – aqueous fluid; olive-green – tourmaline. Sources of data by line. (1) Miscellaneous borate minerals in evaporite deposits (Swihart et al., 1986), (2) Data for non-marine hydrothermal fluids are from measurements on geothermal fluids associated with salar deposits, northwestern Argentina (Kasemann et al., 2004) and data for marine hydrothermal fluids are from seafloor brine pools in the Red Sea (Palmer, 1991). (3) Tourmaline composition calculated from brine composition assuming fractionation (Meyer et al., 2008) at 200 °C. (4– 7) Compositions of metamorphic borate (suanite, szaibe´lyite), fluid calculated assuming no fractionation with borate, tourmaline measured, and tourmaline calculated assuming fractionation (Meyer et al., 2008) for 350–450 °C from the Zhuanmiao deposit, Liaoning-Jilin area, China (Peng and Palmer, 2002). (8–10) Compositions of tourmaline sourcing borate from non-marine borate, marine borate and clastic sediments (modified from Palmer (1991), Fig. 3, with data for tourmaline from marine borate from Frimmel and Jiang, 2001). (11) Compositions of tourmaline in tourmaline metaquartzite, Larsemann Hills (this paper, Fig. 10).

In order to distinguish between non-marine and marine evaporite, Palmer and Slack (1989) argued that tourmaline sourced from marine evaporites should have d11B > 0&, and Palmer (1991) suggested that in marine evaporite where d11B < 0&, then the B had a non-marine component that has overwhelmed the contribution from marine evaporite. In evaluating these criteria we must assume that seawater d11B has not changed significantly since the Archean, our only choice because there is no accurate knowledge of seawater d11B over long-time scales (Pagani et al., 2005); even variation over the last 50 Ma is contentious (e.g., Raitzsch and Ho¨nisch, 2013). A cutoff of d11B = 0& in tourmaline is consistent with range of d11B (+4 to +27.5&) in tourmaline inferred to have been sourced from marine evaporite reported by Frimmel and Jiang (2001), but not with the relatively heavy B in tourmaline from deposits in the Liaoning and Jilin Provinces of northeastern China (Fig. 11, lines 4– 7, Peng and Palmer, 2002). Tourmalinite (d11B = +1.8& to +9.3&) in the Zhuanmiao deposit formed by reaction during metamorphism at 350–450 °C of volcanogenic rocks with fluids rich in B derived from suanite, Mg2B2O5, and szaibe´lyite, MgBO2(OH), in which d11B ranges from +8.8& to +12.6&. These results suggest that the range of d11B values for tourmaline sourced from marine and nonmarine evaporate overlap, i.e., the d11B range for tourmaline formed from non-marine evaporite B extends to +10&. In summary, the Larsemann Hills tourmaline d11B

values (2.8& to 14.4&) are consistent with derivation from a non-marine evaporite source. 6.5. Comparison with other areas A possible analogue for the precursor environment of deposition is the Miocene lacustrine Jarandol basin in Serbia, which contains analcite-rich tuffaceous rocks, clays, and marls with bodies of magnesite and borates (Obradovic´ et al., 1992, 1999). According to these authors, volcanic emanations enriched pore waters in B, Si, Ca and Na, which deposited Ca and Na–Ca borates and borosilicates such as colemanite, Ca6B2O8(OH)62H2O, ulexite, NaCaB5O6(OH)65H2O, and howlite, Ca2B5SiO9(OH)5. Later fluids introduced P, which resulted in local deposition of lu¨neburgite, Mg3B2(PO4)2(OH)66H2O in fractures in magnesite. The presence of this borate-phosphate in the precursor could explain simultaneous enrichment of B and P in a few Larsemann Hills rocks, which was reported by Grew et al. (2013a). Although B is inferred to have survived metamorphism, compounds of B with Ca and Na have not. Instead we have borosilicates of Al, Mg and Fe, with little Ca and Na (tourmaline) or none at all (prismatine and grandidierite) in the borosilicate gneisses and leucogneisses that Grew et al. (2013a) suggested might have an evaporite precursor. That is, Ca and Ca–Na borates reacted with clays and possibly magnesite to form aluminoborosili-

J.R. MacGregor et al. / Geochimica et Cosmochimica Acta 123 (2013) 261–283

cates during metamorphism. The analogy with Jarandol and similar basins implies that precursors Grew et al. (2013a) considered to be clastic sediments could have had a substantial tuffaceous component. 6.6. Summary statement In summary, the tourmaline B isotopic compositions, together with the whole-rock compositions and abundance of B, are consistent with non-marine evaporite being the most important source of B. A possible scenario is a succession of clastic sediments, in part tuffaceous, intercalated with B-rich evaporite deposits in a rift basin, in which circulating hydrothermal fluids leached B from the evaporite and transported it to associated clastic rocks. This model differs from the scenario that Slack et al. (1993) proposed for Broken Hill in that volcanic exhalative rocks are absent or very much subordinate in the Larsemann Hills. As a result of this B metasomatism, tourmaline precipitated out in the clastic and tuffaceous sediments, which became the precursors to the tourmaline metaquartzites, while borosilicate gneisses and leucogneisses could be derived from sedimentary and tuffaceous rocks containing a chemical component, i.e., evaporitic borate (Grew et al., 2013a). ACKNOWLEDGEMENTS Fieldwork during the 2003/2004 field season and laboratory expenses were supported by NSF grants OPP-0228842 and EAR0837980 to University of Maine and ASAC 2350 to C.J.C. and E.S.G. We thank two anonymous reviewers and Horst Marschall for their constructive and thoughtful comments on an earlier version of the manuscript; Chris Gerbi for comments on the Master’s thesis (MacGregor, 2012) and valuable insight on uncertainties in the SIMS measurements, which improved the present manuscript; Richard Hinton for his considerable assistance in carrying out the boron isotope analyses; Thomas Ludwig for sharing results obtained in Heidelberg on the three tourmaline standards; the Davis station leader Bob Jones and members of the 2003–2004 ANARE for logistic support; Liudong Ren and Yanbin Wang for translation of passages in Wang et al. (2004) and Samuele Agostini, Istituto di Geoscienze e Georisorse – Consiglio Nazionale delle Ricerche, Pisa, Italy for providing TIMS B-isotope data on prismatine and grandidierite from the Larsemann Hills.

REFERENCES Becke A. D. (1988) Density-functional exchange-energy approximation with correct asymptotic behavior. Phys. Rev. A 38, 3098–3100. Boger S. D. (2011) Antarctica—Before and after Gondwana. Gondwana Res. 19, 335–371. Cabral A. R., Wiedenbeck M., Rios F. J., Seabra Gomes, Jr., A. A., Rocha Filho O. G. and Jones R. D. (2012) Talc mineralisation associated with soft hematite ore, Gongo Soco deposit, Minas Gerais, Brazil: petrography, mineral chemistry and boron-isotope composition of tourmaline. Mineral. Deposita 47, 411–424. Carson C. J. and Grew E. S. (2007) Geology of the Larsemann Hills region, Antarctica, first ed. (1:25 000 scale map). Geoscience Australia, Canberra.

281

Carson C. J., Dirks P. H. G. M. and Hand M. (1995a) Stable coexistence of grandidierite and kornerupine during medium pressure granulite facies metamorphism. Mineral. Mag. 59, 327–339. Carson C. J., Dirks P. H. G. M., Hand M. P., Sims J. P. and Wilson C. J. L. (1995b) Compressional and extensional tectonics in low-medium pressure granulites from the Larsemann Hills, East Antarctica. Geol. Mag. 132, 151–170. Carson C. J., Wilson C. J. L. and Dirks P. H. G. M. (1997) Partial melting during tectonic exhumation of a granulite terrane: an example from the Larsemann Hills, East Antarctica. J. Metamorph. Geol. 15, 105–126. Catanzaro E. J., Champion C. E., Garner E. L., Malinenko G., Sappenfield K. M. and Shields W. R. (1970) Boric acid, isotopic, and assay standard reference materials. U.S. National Bureau of Standards, Special Publication No. 260–17, U.S. Government Printing Office (Washington, DC), 70 pp. ˇ erny´ P. and Ercit T. S. (2005) The classification of granitic C pegmatites revisited. Can. Mineral. 43, 2005–2026. Chaussidon M. and Albare`de F. (1992) Secular boron isotope variations in the continental crust: an ion microprobe study. Earth Planet. Sci. Lett. 108, 229–241. Cooper M. A., Hawthorne F. C. and Grew E. S. (2009) The crystal chemistry of the kornerupine–prismatine series. 1. Crystal structure and site populations. Can. Mineral. 47, 233–262. Dirks P. H. G. M., Carson C. J. and Wilson C. J. L. (1993) The deformational history of the Larsemann Hills, Prydz Bay: the importance of the Pan-African (500 Ma) in East Antarctica. Antarct. Sci. 5, 179–192. Dyar M. D., Wiedenbeck M., Robertson D., Cross L. R., Delaney J. S., Ferguson K., Francis C. A., Grew E. S., Guidotti C. V., Hervig R. L., Hughes J. M., Husler J., Leeman W. P., McGuire A. V., Rhede D., Rothe H., Paul R. L., Richards I. and Yates M. (2001) Reference minerals for the microanalysis of light elements. Geostand. Newsl. 25, 441–463. Dzikowski T. J., Groat L. A. and Grew E. S. (2007) The geometric effects of VFe2+ for VMg substitution on the crystal structures of the grandidierite–ominelite series. Am. Mineral. 92, 863–872. Fitzsimons I. C. W. (1997) The Brattstrand Paragneiss and the Sostrene Orthogneiss: a review of Pan-African metamorphism and Grenvillan relics in Southern Prydz Bay. In The Antarctic Region: Geological Evolution and Processes (ed. C. A. Ricci). Terra Antarctica, Siena, Italy. pp. 121–130. Fitzsimons I. C. W. and Harley S. L. (1991) Geological relationships in high-grade gneiss of the Brattstrand Bluffs coastline, Prydz Bay, east Antarctica. Aust. J. Earth Sci. 38, 497–519. Frimmel H. E. and Jiang S.-Y. (2001) Marine evaporites from an oceanic island in the Neoproterozoic Adamastor ocean. Precambr. Res. 105, 57–71. Goedecker S., Teter M. and Hutter J. (1996) Separable dual-space Gaussian pseudopotentials. Phys. Rev. B 54, 1703–1710. Grew E. S. (2002) Borosilicates (exclusive of tourmaline) and boron in rock-forming minerals in metamorphic environments. In Boron Mineralogy, Petrology and Geochemistry, vol. 33 (eds. E. S. Grew and L. M. Anovitz). Mineral. Soc. Amer. Rev. Mineral., pp. 387–502. Grew E. S., McGee J. J., Yates M. G., Peacor D. R., Rouse R. C., Huijsmans J. P. P., Shearer C. K., Wiedenbeck M., Thost D. E. and Su S.-C. (1998) Boralsilite (Al16B6Si2O37): a new mineral related to sillimanite from pegmatites in granulite-facies rocks. Am. Mineral. 83, 638–651. Grew E. S., Christy A. G. and Carson C. J. (2006) A boronenriched province in granulite-facies rocks, Larsemann Hills, Prydz Bay, Antarctica. Geochim. Cosmochim. Acta 70(18 Suppl.), A217 (abstr.).

282

J.R. MacGregor et al. / Geochimica et Cosmochimica Acta 123 (2013) 261–283

Grew E. S., Graetsch H., Po¨ter B., Yates M. G., Buick I., Bernhardt H.-J., Schreyer W., Werding G., Carson C. J. and Clarke G. L. (2008) Boralsilite, Al16B6Si2O37, and “boronmullite”: compositional variations and associated phases in experiment and nature. Am. Mineral. 93, 283–299. Grew E. S., Carson C. J., Christy A. G., Maas R., Yaxley G. M., Boger S. D. and Fanning C. M. (2012) New constraints from UPb, Lu-Hf and Sm-Nd isotopic data on the timing of sedimentation and felsic magmatism in the Larsemann Hills, Prydz Bay, East Antarctica. Precambr. Res. 206–207, 87–108. Grew E. S., Carson C. J., Christy A. G. and Boger S. D. (2013a) Boron- and phosphate-rich rocks in the Larsemann Hills, Prydz Bay, East Antarctica: tectonic implications. In Antarctica and Supercontinent Evolution (eds. S. L. Harley, I. C. W. Fitzsimons and Y. Zhao). Geological Society of London Special Publication. http://dx.doi.org/10.1144/SP383.8. Grew E. S., Maas R. and Carson, C. J. (2013b) Isotopic constraints on the source of pegmatites with boron and beryllium minerals in the Larsemann Hills, Prydz Bay, Antarctica. Geological Association of Canada, Mineralogical Association of Canada, Joint Annual Meeting, Winnipeg, Abstracts http://gac.esd.mun.ca/gac_2013/search_abs. Henry D. J., Nova´k M., Hawthorne F. C., Ertl A., Dutrow B. L., Uher P. and Pezzotta F. (2011) Nomenclature of the tourmaline-supergroup minerals. Am. Mineral. 96, 895–913. Hervig R. L., Moore G. M., Williams L. B., Peacock S. M., Holloway J. R. and Roggensack K. (2002) Isotopic and elemental partitioning of boron between hydrous fluid and silicate melt. Am. Mineral. 87, 769–774. Jiang S.-Y. (2006) Reply to “Re-examination of the boron isotopic composition of tourmaline from the Lavicky granite, Czech Republic, by secondary ion mass spectrometry: back to normal” by H. R. Marschall and T. Ludwig: Critical comment on “Chemical and boron isotopic compositions of tourmaline from the Lavicky leucogranite, Czech Republic”. Geochem. J. 40, 639–641. Jiang S.-Y., Yang J.-H., Nova´k M. and Selway J. (2003) Chemical and boron isotopic compositions of tourmaline from the Lavicky leucogranite, Czech Republic. Geochem. J. 37, 545– 556. Kakihana H., Kotaka M., Satoh S., Nomura M. and Okamoto M. (1977) Fundamental studies on the ion-exchange separation of boron isotopes. Bull. Chem. Soc. Jpn. 50, 158–163. Kasemann S., Erzinger J. and Franz G. (2000) Boron recycling in the continental crust of the central Andes from the Palaeozoic to Mesozoic, NW Argentina. Contrib. Mineral. Petrol. 140, 328–343. Kasemann S., Meixner A., Erzinger J., Viramonte J., Alonso R. N. and Franz G. (2004) Boron isotope composition of geothermal fluids and borate minerals from salar deposits (central Andes/ NW Argentina). J. S. Am. Earth Sci. 16, 685–697. Kawakami T. (2001) Boron depletion controlled by the breakdown of tourmaline in the migmatite zone of the Aoyama area, Ryoke metamorphic belt, southwestern Japan. Can. Mineral. 39, 1529–1546. Kawakami T. (2004) Tourmaline and boron as indicators of the presence, segregation and extraction of melt in pelitic migmatites: examples from the Ryoke metamorphic belt, SW Japan. Trans. R. Soc. Edinburgh: Earth Sci. 95, 111–123. Kawakami T. and Ikeda T. (2003) Boron in Metapelites controlled by the breakdown of tourmaline and retrograde formation of borosilicates in the Yanai area, Ryoke metamorphic belt, SW Japan. Contrib. Mineral. Petrol. 145, 131–150. Kelsey D. E., Wade B. P., Collins A. S., Hand M., Sealing C. R. and Netting A. (2008) Discovery of a Neoproterozoic basin in the Prydz Belt in East Antarctica and its implications for

Gondwana assembly and ultrahigh temperature metamorphism. Precambr. Res. 161, 355–388. Kopf A. J. and Deyhle A. (2002) Back to the roots: source depths of mud volcanoes and diapirs using boron and B isotopes. Chem. Geol. 192, 195–210. Kopf A., Deyhle A., Lavrushin V. Y., Polyak B. G., Gieskes J. M., Buachidze G. I., Wallmann K. and Eisenhauer A. (2003) Isotopic evidence (He, B, C) for deep fluid and mud mobilization from mud volcanoes in the Caucasus continental collision zone. Int. J. Earth Sci. 92, 407–426. Kowalski P. M. and Jahn S. (2011) Prediction of equilibrium Li isotope fractionation between minerals and aqueous solutions at high P and T: an efficient ab initio approach. Geochim. Cosmochim. Acta 75, 6112–6123. Kowalski P. M., Wunder B. and Jahn S. (2013) Ab initio prediction of equilibrium boron isotope fractionation between minerals and aqueous fluids at high P and T. Geochim. Cosmochim. Acta 101, 285–301. Lee C., Yang W. and Parr R. C. (1988) Development of the Colle– Salvetti correlation-energy formula into a functional of the electron density. Phys. Rev. B 37, 785–789. Leeman W. P. and Sisson V. B. (2002) Geochemistry of boron and its implications for crustal and mantle processes. In Boron: Mineralogy, Petrology and Geochemistry, vol. 33 (eds. E. S. Grew and L. M. Anovitz). Mineral. Soc. Am. Rev. Mineral., pp. 645–707. Leeman W. P. and Tonarini S. (2001) Boron isotopic analysis of proposed borosilicate mineral reference samples. J. Geostand. Geoanal. 25, 399–403. London D. (2011) Experimental synthesis and stability of tourmaline: a historical overview. Can. Mineral. 49, 117–136. Ludwig T., Marschall H. R., Pogge von Strandmann P. A. E., Shabaga B. M., Fayek M. and Hawthorne F. C. (2011) A secondary ion mass spectrometry (SIMS) re-evaluation of B and Li isotopic compositions of Cu-bearing elbaite from three global localities. Mineral. Mag. 75, 2485–2494. MacGregor J. (2012) Boron isotopic study of the borosilicates tourmaline, prismatine, and grandidierite in granulite facies paragneisses, from the Larsemann Hills, Prydz Bay, East Antarctica. M. S. thesis, Univ. Maine. Marschall H. R. (2005). Lithium, beryllium, and boron in highpressure metamorphic rocks from Syros (Greece). InauguralDissertation, Universita¨t Heidelberg. Available at: . Marschall H. R. and Jiang S.-Y. (2011) Tourmaline isotopes: no element left behind. Elements 7, 313–319. Marschall H. R. and Ludwig T. (2006) Re-examination of the boron isotopic composition of tourmaline from the Lavicky granite, Czech Republic, by secondary ion mass spectrometry: back to normal. Critical comment on “Chemical and boron isotopic compositions of tourmaline from the Lavicky leucogranite, Czech Republic” by S.-Y. Jiang et al. Geochem. J. 37, 545–556, 631–638. Marschall H. R., Ludwig T., Altherr R., Kalt A. and Tonarini S. (2006) Syros metasomatic tourmaline: evidence for very highd11B Fluids in subduction zones. J. Petrol. 47, 1915–1942. Marx D. and Hutter J. (2000) Ab initio molecular dynamics: theory and implementation. In Modern Methods and Algorithms of Quantum Chemistry (ed. J. Grotendorst). NIC, FZ Juelich, pp. 301–449 (CPMD code: J. Hutter et al.) . Merlet C. (1994) An accurate computer correction program for quantitative electron probe micro-analysis. Mikrochim. Acta 114(115), 363–376. Meyer C., Wunder B., Meixner A., Romer R. L. and Heinrich W. (2008) Boron-isotope fractionation between tourmaline and

J.R. MacGregor et al. / Geochimica et Cosmochimica Acta 123 (2013) 261–283 fluid: an experimental re-investigation. Contrib. Mineral. Petrol. 156, 259–267. Obradovic´ J., Stamatakis M. G., Anicic S. and Economou G. S. (1992) Borate and borosilicate deposits in the Miocene Jarandol basin, Serbia, Yugoslavia. Econ. Geol. 87, 2169–2174. Obradovic´ J., Vasic´ N., Kasˇanin-Grubin M. and Grubin N. (1999) Neogene lacustrine sediments and authigenic minerals geochemical characteristics. Geolosˇki anali Balkanskoga poluostrva 63, 135–154. Pagani M., Lemarchand D., Spivack A. and Gaillaridet J. (2005) A critical evaluation of the boron isotope-pH proxy: the accuracy of ancient ocean pH estimates. Geochim. Cosmochim. Acta 69, 953–961. Palmer M. R. (1991) Boron isotope systematics of hydrothermal fluids and tourmaline: a synthesis. Chem. Geol. 94, 111–121. Palmer M. R. and Slack J. F. (1989) Boron isotopic composition of tourmaline from massive sulfide deposits and tourmalinites. Contrib. Mineral. Petrol. 103, 434–451. Palmer M. R. and Swihart G. H. (2002) Boron isotope geochemistry: an overview. In Boron: Mineralogy, Petrology and Geochemistry, vol. 33 (eds. E. S. Grew and L. M. Anovitz). Mineral. Soc. Am. Rev. Mineral., pp. 709–744. Palmer M. R., London D., Morgan G. and Babb H. (1992) Experimental determination of fractionation of 11B/10B between tourmaline and aqueous vapour: a temperature- and pressure-dependent isotopic system. Chem. Geol. 101, 123–129. Peng Q.-M. and Palmer M. R. (2002) The Paleoproterozoic Mg and Mg-Fe borate deposits of Liaoning and Jilin Provinces, Northeast China. Econ. Geol. 97, 93–108. Raitzsch M. and Ho¨nisch B. (2013) Cenozoic boron isotope variations in benthic foraminifers. Geology 41, 591–594. Ren L. and Liu X. (1993) An occurrence of the assemblage grandidierite, kornerupine, and tourmaline in Antarctica. Antarct. Res. 4(2), 21–28. Ren L. and Liu X. (1994) Tourmaline and its relationship with metamorphism, Zhongshan Station, Antarctica. Yanshi Kuangwuxue Zashi. Acta Petrol. Mineral. 13(2), 169–174 (in Chinese with English abstract). Ren L. and Zhao Y. (1992) The first occurrence of grandidierite in Antarctica. Explor. Geosci. 7, 1–6. Ren L., Zhao Y., Liu X. and Chen T. (1992) Re-examination of the metamorphic evolution of the Larsemann Hills, East Antarctica. In Recent Progress in Antarctic Earth Science (eds. Y. Yoshida, K. Kaminuma and K. Shiraishi). Terrapub, Tokyo, pp. 145–153. Robbins C. R. and Yoder H. S. (1962) Stability relations of dravite, a tourmaline. Carnegie Inst. Washington Yearbook, vol. 61, pp. 106–107. Slack J. F. (2002) Tourmaline associations with hydrothermal ore deposits. In Boron: Mineralogy, Petrology and Geochemistry, vol. 33 (eds. E. S. Grew and L. M. Anovitz). Mineral. Soc. Am. Rev. Mineral., pp. 559–643. Slack J. F., Palmer M. R. and Stevens B. P. J. (1989) Boron isotope evidence for the involvement of non-marine evaporites in the origin of the Broken Hill ore deposits. Nature 342, 913–916.

283

Slack J. F., Palmer M. R., Stevens B. P. J. and Barnes R. G. (1993) Origin and significance of tourmaline-rich rocks in the Broken Hill District, Australia. Econ. Geol. 88, 505–541. Slack J. F., Turner R. J. W. and Ware P. L. G. (1998) Boron-rich mud volcanoes of the Black Sea region: modern analogues to ancient sea-floor tourmalinites associated with Sullivan-type Pb–Zn deposits? Geology 26, 439–442. Stu¨we K., Braun H. M. and Peer H. (1989) Geology and structure of the Larsemann Hills area, Prydz Bay, East Antarctica. Aust. J. Earth Sci. 36, 219–241. Swihart G. H., Moore P. B. and Callis E. L. (1986) Boron isotopic composition of marine and nonmarine evaporate borates. Geochim. Cosmochim. Acta 50, 1297–1301. Thost D. E., Hensen B. J. and Motoyoshi Y. (1994) The geology of a rapidly uplifted medium and low pressure granulite facies terrane of Pan-African age: the Bolingen Islands, Prydz Bay, Eastern Antarctica. Petrology 2, 293–316. Thomson J. A. (2006) A rare garnet–tourmaline–sillimanite– biotite–ilmenite–quartz assemblage from the granulite-facies region of south-central Massachusetts. Am. Mineral. 91, 1730– 1738. Tonarini S., Forte C., Petrini R. and Ferrara G. (2003) Melt/biotite 11 B/10B isotopic fractionation and the boron local environment in the structure of volcanic glasses. Geochim. Cosmochim. Acta 67, 1863–1873. van Hinsberg V. J., Henry D. J. and Marschall H. R. (2011) Tourmaline: an ideal indicator of its host environment. Can. Mineral. 49, 1–16. Vernon R. H. (2004) A Practical Guide to Rock Microstructure. Cambridge University Press, New York. Wadoski E. (2009) Microstructural and chemical study of borosilicate minerals in pegmatites from the Larsemann Hills, Prydz Bay, East Antarctica. M. S. thesis, Univ. Maine. Wadoski E., Grew E. S. and Yates M. G. (2011) Compositional evolution of tourmaline-supergroup minerals from granitic pegmatites in the Larsemann Hills, East Antarctica. Can. Mineral. 49, 381–405. Wang Y. B., Ren L. D., Liu D. and Xiao Y. K. (2004) Boron isotopic compositions of borosilicates from the Larsemann Hills, East Antarctica. Geochimica 33, 215–218 (in Chinese with English abstract). Wang Y., Liu D., Chung S.-L., Tong L. and Ren L. (2008) SHRIMP zircon age constraints from the Larsemann Hills region, Prydz Bay, for a late Mesoproterozoic to early Neoproterozoic tectono-thermal event in East Antarctica. Am. J. Sci. 308, 579–617. Werding G. and Schreyer W. (2002) Experimental studies on borosilicates and selected borates. In Boron: Mineralogy, Petrology and Geochemistry, vol. 33 (eds. E. S. Grew and L. M. Anovitz). Mineral. Soc. Am. Rev. Mineral., pp. 117–163. Xiao Y.-K., Beary E. S. and Fassett J. D. (1988) An improved method for the high-precision isotopic measurement of boron by thermal ionization mass spectrometry. Int. J. Mass Spectrom. Ion Processes 85, 203–213. Associate editor: Edward M. Ripley