Geochemistry and δB evolution of tourmaline from tourmalinite as a record of oceanic crust in the Tonian Ibaré ophiolite, southern Brasiliano Orogen

The isotopic and geochemical evolution of tourmaline constrain the processes of paleo-oceanic lithosphere in ophiolites. The Brasiliano Orogen is a major structure of South America and requires characterization for the understanding of Gondwana supercontinent evolution. We made a pioneering investigation of tourmaline from a tourmalinite in the Ibaré ophiolite by integrating fi eld work with chemical analyses of tourmaline by electron microprobe (EPMA) and δB determinations via laser ablation inductively coupled plasma mass spectrometer (LA-ICP-MS). Remarkably massive tourmalinite (>90 vol.% tourmaline, some chlorite) enclosed in serpentinite has homogeneous dravite in chemical and isotopic composition (δB = +3.5 to +5.2‰). These results indicate a geotectonic environment in the altered oceanic crust for the origin of the tourmalinite. This fi rst δB characterization of tourmaline from tourmalinite sets limits to the evolution of the Neoproterozoic to Cambrian Brasiliano Orogen and


INTRODUCTION
Tourmaline is a most useful mineral because it is robust and can retain a record of geological processes (van Hinsberg et al. 2011). The mineral is helpful for understanding both continental (Chaussidon & Albarède 1992, Trumbull et al. 2008, Cabral et al. 2017) and oceanic (Smith et al. 1995, Farber et al. 2015 settings. Most studies focused on accessory tourmaline in granitic and mineralized volcanicsedimentary rocks, including sulphide ore (e.g., Namaqualand, South Africa -Plimer 1987, Sullivan, British Columbia -Palmer & Slack 1989, Broken Hill, Australia -Slack et al. 1993. Fewer studies concentrated on oceanic lithosphere, including crust and mantle or on volatile transfer processes from the subduction setting to the mantle wedge and arc magmatism (Palmer 1991, Rosner et al. 2003, Savov et al. 2005, Boschi et al. 2008, Yamaoka et al. 2012. In Brazil, stratiform tourmalinite occurs associated to mineralized quartz-tourmaline veins in the Calymmian Serra do Itaberaba (Ribeira belt, SE Brazil - Garda et al. 2009). Tourmaline was also described in pegmatites (Borborema Province, NE Brazil -Trumbull et al. 2013) and as platiniferous goldtourmaline aggregates (Gold-palladium belt of Minas Gerais, Brazil -Cabral et al. 2017).
Tourmaline (geochemistry, δ 11 B) was only studied in continental rocks in the continent, not in ophiolites. In the Brasiliano Orogen Neoproterozoic ophiolites were studied with zircon U-Pb isotopes in the Araçuaí Belt (oceanic plagiogranite - Queiroga et al. 2007) and Dom Feliciano Belt (albitite, chloritite, tourmalinite, rodingite blackwall -Arena et al. 2016, 2017. Tourmalines from ophiolites were not studied. We selected a massive tourmalinite (>90 vol.% tourmaline) from southern Brazil, because the rock is part of the Ibaré ophiolite. The sample was previously investigated with zircon U-Pb-Hf isotopes by Arena et al. (2017). Large (5-30 m diameter) tourmalinites remain undescribed in the oceanic crust or ophiolites, so this is a pioneering investigation of mantle interaction with oceanic water in the Tonian.
The tourmalinite was described in the field and large tourmaline crystals (up to 10 cm) were studied by EPMA for major elements and LA-ICP-MS for δ 11 B. The results indicate a remarkably homogeneous dravite, including boron isotopes (δ 11 B = +3.5 to +5.2‰). We interpret the tourmalinite as formed in the Tonian oceanic crust by alteration of mantle rocks in contact with oceanic water. This characterization of the geotectonic environment may have large impact on studies of the Early Brasiliano Orogen in the continent and reconstruction of Rodinia and Gondwana.

MATERIALS AND METHODS
Field study of the Ibaré ophiolite included collection of selected tourmalinite IB14 sample. Petrography of the tourmalinite preceded the determination of chemical and boron isotopic analyses of tourmaline. Petrography was done with a transmission petrographic microscope Olympus BX51, UC30. One polished thin section of the sample was studied for elemental mapping of tourmaline by electron microprobe at Laboratório de Microssonda Eletrônica, Universidade Federal do Rio Grande do Sul. A block was cut from the massive tourmalinite sample (Figs. 2d,e) and measures 10 cm in length by ~1.5 cm in width. The block was divided into parts and placed in 5 mounts in sequential order, each measuring 2 cm x 1.5 cm (Fig. 2f). Mounts from tablets 3 and 4 (Figs. 2f,3a,b) were selected for analyses by scanning electron microprobe and boron isotopes at Departamento de Geologia, Universidade Federal de Ouro Preto (UFOP), Minas Gerais. All these spot analyses were controlled by backscattered electron images (Fig. 3).

Electron microprobe
Electron microprobe analyses of mounted tourmalines were performed at UFOP using a JEOL JXA-8230 Superprobe equipped with 5 spectrometers. Operating conditions were 15 kV accelerating voltage, 20 nA beam current and 10 μm beam diameter and a selection of measurement spots ensured that the stimulated volume was not contaminated by phases other than tourmaline. Counting times on the peaks/background were 10/5 s for all elements. Background intensities were collected at higher and lower energies relative to the corresponding Kα line. Appropriate natural and synthetic reference materials were used for calibration (Supplementary Material -Table SI). Tourmaline structural formulae were calculated by normalizing to 15 cations in the tetrahedral and octahedral sites (T + Z + Y) and assuming 3 boron apfu (Henry et al. 2011) using the Excel spreadsheet of Tindle et al. (2002).

Boron isotopes
Boron isotope ratio measurements were carried out at UFOP on a Thermo-Scientific Neptune Plus multi-collector ICP-MS coupled to a Photon Machines 193 Excimer laser ablation system. Samples were ablated in He atmosphere using 20 um diameter spot and 15 Hz frequency at 7 J/cm2. In the mass spectrometer, 10 B and 11 B intensities were measured (in low resolution) on the L2 and H2 detectors, respectively. The measurements consisted of 98 cycles (or integration) and 0.5 s of integration time. Data were processed after the daily run using an inhouse spreadsheet by A. Gerdes (e.g., Devulder et al. 2015). The measured, background signal of the unknown sample was corrected for instrumental mass fractionation (IMF) using a standard-sample bracketing method and tourmaline B4 (schorlite, δ 11 B = -8.62 ‰) (Tonarini et al. 2003) as the primary reference material. The drift-corrected ratios were referenced to the published 11 B/ 10 B value of the reference material and the results are reported as δ 11 B values relative to NIST SRM 951 boric acid using the certified 11 B/ 10 B value of 4.04362 ± 0.00137 (Catanzaro et al. 1970). Matrix effects, known to occur in B isotope measurements (e.g., Mikova et al. 2014) and the reproducibility of   1995, Babinski et al. 1996, Leite et al. 1998, Saalmann et al. 2005a, b, Hartmann et al. 2011, Lena et al. 2014, Lopes et al. 2015, Arena et al. 2016. The studied tourmalinite is part of the Ibaré ophiolite (Arena et al. 2017), associated with volcano-sedimentary rocks which are remnants of a Tonian intra-oceanic arc.
The Ibaré ophiolite and associated rocks (Figs. 1b,2a) show evidence of regional greenschist facies metamorphism strongly overprinted by contact metamorphism caused by the intrusion (Naumann & Hartmann 1984, Naumann 1985

Sample description
Within the Ibaré ophiolite, massive tourmalinite (>90 vol.% tourmaline) is associated with chloritite, serpentinite, magnesian schist, rodingite, and albitite (Naumann 1985, Arena et al. 2016, 2017. Santa Rita Granite intruded the ophiolite in the northern portion and a tongue of the Jaguari Granite intruded in the southern portion. The tourmalinite (2 x 5 m large) occurs immersed in chloritite and serpentinite (Figs. 2b,c). Black color of tourmalinite is distinctive and contrasts with serpentinite and the surrounding chloritite blackwall. The tourmalinite is composed of tourmaline, chlorite, some ilmenite and zircon.

RESULTS
Representative electron microprobe analyses of tourmaline are listed in Supplementary Material -Table SIII. All analyzed points (n = 60) in tourmaline showed chemical homogeneity. . Some Al defficiency (<6 atoms per formula unit) in the Z site is due to Mg substitution and no Fe 3+ . In addition, the X site occupancy is predominantly Na. The tourmaline is dravite in composition (Supplementary Material - Table SIII) and shows no chemical zoning (e.g., Fig. 4). Tourmalines are of alkali group (Na + K dominant at X site) and Mgrich (Figs. 5a, b). The array is most consistent with operation of the MgFe 2+ −1 substitution (Figs. 6a, b). The studied IB14 tourmalines show homogenous isotopic compositions ranging from δ 11 B = +3.5 to +5.2‰ (Fig. 8b, Table I).

Dravite from Ibaré tourmalinite (IB14 sample)
is Mg-rich and shows homogeneous chemistry and boron isotopic composition. Because tourmalinites may also form during regional and contact metamorphism, careful evaluation is necessary to determine their origin relative to hydrothermal, metamorphic and granitic process. Particular care is required for the identification of processes in the oceanic realm.
A  (Fig. 2) were considered possible sources in case tourmalinization occurred in the continent, after the emplacement of the ophiolite. Our results discard this hypothesis. Tourmaline in most granites and pegmatites in the continents has δ 11 B values close to average continental crust (δ 11 B = −10 ±3‰; Marschall and Jiang 2011). Besides, tourmaline compositions in granitic rocks plot in fields 1 to 3 (Figs. 7a, b). In contrast, IB14 tourmalines plot in field 6 on Al-Fe-Mg(total) and Ca-Fe-Mg(total) diagrams (Henry & Guidotti 1985), unrelated to granitic rocks.
Tourmalines from Borborema pegmatites (Trumbull et al. 2013) overlap fields 5 and 6 on Al-Fe-Mg diagram (Fig. 7a); on Ca-Fe-Mg diagram (Fig. 7b), they overlap fields 1, 2 and 6 because of their preserved source chemical signature from surrounding rocks. The full range of δ 11 B is −20.2 to +1.6‰ with main range −17 to −9‰ (Fig. 8b). Trumbull et al. (2013) suggest that the strong isotopic contrast between the main range and the heavy B resulted from mixing with enclosing marble and calc-silicate gneisses. It is significant that the origin of IB14 tourmaline was unrelated to granitic fluids. Ibaré tourmaline is Mg-rich but no associated sedimentary or exhalative rocks were identified. Tourmalines from many environments are strongly zoned and varied in trace elements and boron isotopes, but IB14 tourmaline is chemically homogeneous (Figs. 4,7,8). The uniformity in composition of IB14 tourmalines is highlighted by comparison with tourmalines from tourmalinites associated with augen gneiss, leucogranite and garnet-micaschist In hydrothermal systems (Slack et al. 1993), Mg-rich tourmalines may form either by pre-metamorphic replacement from seawater-derived fluids under high fluid/rock conditions, or by sulfide-silicate reactions during metamorphism. Our results are more consistent with seawater-derived fluids. The IB14 dravite is interpreted as formed from entrained seawater (Slack & Trumbull 2011) (Fig. 7c) which caused serpentinization and subsequent tourmalinization.
The IB14 dravite has boron isotope composition (δ 11 B = + 3.5 to +5.2 ‰) typical of blackwall metasomatism between altered oceanic crust and serpentinite. The tourmalinite formed below the seawater-sediment (or volcanic) interface with no exhalative component (closely associated or in contact with chemical sediments such as metachert and iron-formation) as described by Slack et al. (1993). This is in agreement with the values of modern bulk oceanic crust (Fig. 8a) by Farber et al. (2015), which range between +3.7‰ and +7. 9‰ (Smith et al. 1995, Yamaoka et al. 2015. The B isotope composition of IB14 tourmalinite is more enriched in δ 11 B than the slab materials from which they likely originated (δ 11 B MORB = −3‰ to −14‰, average −7.1‰; Chaussidon & Jambon 1994, Marschall et al. 2017. This enrichment indicates that hydrothermal marine fluids altered bulk oceanic crust in the Ibaré tourmalinite. The boron concentration of fresh MORB has a δ 11 B value of −7‰, but when circulating seawater interacts with ocean crust at ~100 °C, then the boron is taken up into secondary minerals. Compilations of ocean cores and ophiolite sections give average boron contents of δ 11 B = +3‰ for the upper oceanic crust (Smith et al. 1995) and the upper mantle may be altered to serpentinite by circulating  seawater, particularly at slow-spreading midocean ridges (Palmer 2017). The boron isotopes composition of the oceanic crust is the result of crust interaction with seawater, through capture of boron during hydrothermal alteration (Spivack & Edmond 1987, Smith et al. 1995. The average δ 11 B calculated for altered oceanic crust is +3.7‰ The array of data is along the schorl-dravite join (more Mg-rich) and is most consistent with operation of the MgFe 2+ −1 substitution; (b) Al (total) versus Fe (total). The data classify the tourmaline as dravite end-member. some geological settings may carry B enriched in 11 B (Table II).
An evaluation is made of the mobility of boron in subduction-zone environments to test the hypotheses of origin of Ibaré tourmalinite. Boron isotopes can be used to unravel transfer processes from the subducting oceanic crust to the mantle wedge and to arc magmatism (Palmer 1991, Ishikawa & Tera 1999, Rosner et al. 2003, Savov et al. 2005. Studies by Ishikawa & Tera (1999) suggested a central role of altered oceanic crust and sediments as sources of fluids and boron in subduction zones. Additionally, Savov et al. (2005) and Tonarini et al. (2007) showed the potential role of subducted serpentinites as an important source of B-rich fluids to suprasubduction setting. Peacock & Hervig (1999) suggest that subduction-zone metamorphic dehydration reactions decrease the δ 11 B value of subducted altered oceanic crust as well as subducted sediments through continuous dehydration reactions (Table II). Boron uptake in oceanic rocks occurs by direct incorporation of boron during crystallization of structurally favorable, B-rich hydrothermal-diagenetic minerals (phyllosilicates -Williams et al. 2001) or by adsorption from seawater by secondary minerals such as clays and during lowtemperature interaction of fluids with crustal, mantle and sedimentary rocks (You et al. 1996). Chemical and boron isotopic characteristics of subduction-zone mobility of elements were not observed in Ibaré samples.
The isotopic and geochemical characteristics of IB14 dravite indicate parental fluid sourced from seawater. Lithospheric mantle was intensely metasomatised by large volumes of fluids during associated serpentinization (Fig. 8b). The characteristic depleted mantle composition of zircon contained in the tourmalinite and chloritite (Arena et al. 2017) supports the derivation of studied tourmaline from altered oceanic crust.
We envisage a geological history of the Ibaré ophiolite starting with oceanic crust formation in the Tonian during initial rupturing of Rodinia.

Author contributions
Karine da Rosa Arena did field work, prepared the samples, organized and interpreted the data, and wrote the text of the article. Léo Afraneo Hartmann participated in field work, samples study, supervision of data handling and revision of English of text. Cristiano Lana prepared samples, did δ 11 B analyses on the LA-ICP-MS and reduced data. Gláucia N. Queiroga supervised EPMA analyses, and data reduction and interpretation. Marco P. Castro operated the EPMA for the analyses of minerals and participated in interpretation of data.