| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Journal of Petrology | Volume 44 | Number 4 | Pages 659-678 | 2003
© Oxford University Press 2003
Re–Os and Sm–Nd Isotope and Trace Element Constraints on the Origin of the Chromite Deposit of the Ipueira–Medrado Sill, Bahia, Brazil
1 DEPARTAMENTO DE GEOLOGIA, INSTITUTO DE GEOCIÊNCIAS, UNIVERSIDADE FEDERAL DO RIO GRANDE DO SUL, AV. BENTO GONÇALVES 9500, PORTO ALEGRE-RS, 91509-900, BRAZIL
2 INSTITUTO DE GEOCIÊNCIAS, UNIVERSIDADE DE BRASÍLIA-UNB, CAMPUS UNIVERSITÁRIO, ASA NORTE, BRASÍLIA-DF, 70910-900, BRAZIL
3 DEPARTMENT OF TERRESTRIAL MAGNETISM, CARNEGIE INSTITUTION OF WASHINGTON, 5241 BROAD BRANCH ROAD, WASHINGTON, DC 20015, USA
Present address: Rua Marcelo Gama 1038/404, Porto Alegre-RS, 90540-041, Brazil. Telephone: +55-51-3029-5030. Fax: +55-51-3221-3298. E-mail: juliana.marques{at}ufrgs.br
RECEIVED JANUARY 30, 2002; ACCEPTED OCTOBER 18, 2002
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONTECTONIC SETTINGGEOLOGY AND STRATIGRAPHY OF...ANALYTICAL METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The chromite deposit of the Paleoproterozoic Ipueira–Medrado sill is hosted in a single, thick (5–8 m), massive layer, which sets severe constraints for the origin of chromitites. It is divided from bottom to top into: (1) a Marginal Zone (5–20 m); (2) an Ultramafic Zone (<250 m) consisting of dunite and harzburgite that host the chromitite layer, in which intercumulus amphibole is important and more abundant toward the top; (3) a Mafic Zone (<40 m). The parental magma was large ion lithophile element and light rare earth element enriched and high field strength element depleted. Sm–Nd isotopic compositions are consistent with a 2 Ga age, but suggest a variable initial Nd isotopic composition that correlates with the abundance of amphibole. The more negative
Nd (mean –6·5) of the amphibole-rich intervals argues for crustal contamination, although the Nd (mean –4·4) of the amphibole-free samples suggests an old, enriched, subcontinental lithospheric mantle source. Chromite separates have initial 
Os values that range from –4·6 to +3. The negative Os values are typical of old, Re-depleted, lithospheric peridotitic mantle and give Re-depletion model ages of up to 2·75 Ga. An integrated assessment suggests that the very high-Mg parental magma probably originated from Archean, subcontinental, metasomatized, peridotitic lithospheric mantle and was subsequently contaminated with up to 30% of crust, which triggered the chromitite crystallization. KEY WORDS: mafic–ultramafic sill; isotopes and trace elements; chromite deposit
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONTECTONIC SETTINGGEOLOGY AND STRATIGRAPHY OF...ANALYTICAL METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The origin of chromitites in layered mafic–ultramafic intrusions has been addressed in many studies and distinct models have been proposed to explain their origin. These models include liquid immiscibility (McDonald, 1965
), increase in fO2 (Ulmer, 1969), changes in total pressure in the magma chamber (Cameron, 1977; Lipin, 1993), original magma composition (Cameron, 1978), crustal contamination (Irvine, 1975; Alapieti et al., 1989; Rollinson, 1997), mixing of fresh primitive magma with fractionated magma (Irvine, 1977; Campbell & Murck, 1993) and mixing of two different magmas, e.g. the ‘U’ type magma with ‘A’ type magma, at the Bushveld Complex (Sharpe & Irvine, 1983). The challenge for any model is to account appropriately for geochemical, field-based, and petrological constraints.
Chromitites occur at the base of cyclic units in several layered complexes, suggesting that they are formed by mixing between new influxes of primitive liquid and residual, more fractionated, liquid in the magma chamber (Irvine, 1977; Campbell & Murck, 1993). Evidence for new influxes of primitive magma is provided by changes in the composition of the main cumulus phases and in the order of crystallization of cumulus phases; such as those observed at the Bushveld, South Africa (Sharpe & Irvine, 1983), Great Dyke, Zimbabwe (Wilson, 1982), and Bacuri, Brazil (Spier & Ferreira Filho, 2001) intrusions. Recent studies of the Ipueira–Medrado chromite deposit, Brazil (Marques, 2001; Marques & Ferreira Filho, 2002) do not support such a model for the origin of the continuous (up to 7 km long) and thick (up to 8 m thick) chromitite layer currently mined in the complex. In fact, compositional variation of orthopyroxene and olivine in the stratigraphic interval below this thick chromitite is characterized by slow upward increase of Mg/(Mg + Fe) with no reversal before the onset of chromite crystallization. The resident magma was most primitive immediately below the chromitite and mixture with a new influx of primitive magma cannot account for its formation.
To evaluate the processes of chromitite formation in layered complexes we analyzed samples from one stratigraphic section through the Ipueira–Medrado sill for their Re–Os and Sm–Nd isotopic composition, as well as for rare earth and other trace element abundances. Our study contributes to the understanding of magmatic processes associated with the formation of chromitites, suggesting that the origin of the thick chromitite at the Ipueira–Medrado sill results from crustal contamination occurring in a dynamic magma chamber undergoing successive replenishment with primitive magma.
|
TECTONIC SETTING |
|---|
|
TOPABSTRACTINTRODUCTIONTECTONIC SETTINGGEOLOGY AND STRATIGRAPHY OF...ANALYTICAL METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The Ipueira–Medrado sill is part of a north–south-trending (70 km long and 20 km wide) swarm of mafic–ultramafic chromite mineralized bodies of the Paleoproterozoic Jacurici Complex, São Francisco Craton, Brazil. The complex has been considered to intrude the granulite–gneiss terranes of the Caraíba Granulite Complex (Barbosa et al., 1996
), located at the NE corner of the São Francisco Craton (Fig. 1a). However, a more recent review of the tectonic evolution of the São Francisco Craton suggests that the Jacurici intrusions are hosted by the Archean, medium-grade, gneiss–migmatitic and granite–greenstone belt sequences of the Serrinha Block (Teixeira et al., 2000).
|
The minimum age of the Jacurici intrusions is 2038 ± 19 Ma (2
), as indicated by zircon Pb–Pb evaporation dating of mafic rocks from the top of the sill (Oliveira & Lafon, 1995). The Jacurici Complex parallels the Cu-mineralized mafic–ultramafic intrusions of the Caraíba Complex (Oliveira & Tarney, 1995), which are separated by the 2·1 Ga (Rb/Sr age) Itiúba Syenite (Conceição, 1993). About 50 km west, another large chromite deposit occurs hosted in the Campo Formoso mafic–ultramafic intrusion (Duarte & Fontes, 1986). The Caraíba Complex has a 2001 ± 35 Ma (2) minimum age yielded by zircon Pb–Pb evaporation dating of mafic rocks (Oliveira & Lafon, 1995) and field relations suggest 2·0 Ga as a minimum age for the Campo Formoso Intrusion, implying that significant mafic magmatism took place in this region during the Paleoproterozoic. The Cana Brava, Niquelândia and Barro Alto layered intrusions are further examples of important Paleoproterozoic mafic magmatism in other regions of Brazil and similarities in isotopic composition (Ferreira Filho et al., 1994; Correia et al., 1996, 1997; Ferreira Filho & Pimentel, 2000) suggest that mafic magmatism was in fact widespread in Brazil during this period.
The estimated chromite reserves in the Jacurici Complex amount to >30 Mt (Marinho et al., 1986), making it the largest chromite deposit in Brazil. An updated estimate of the chromite reserve in the Ipueira–Medrado sill is >4·5 Mt grading 30–40 wt % Cr2O3 (Mineração Vale do Jacurici, unpublished data, 1998). The ore is mined from a continuous single layer averaging 5–8 m in thickness.
|
GEOLOGY AND STRATIGRAPHY OF THE SILL |
|---|
|
TOPABSTRACTINTRODUCTIONTECTONIC SETTINGGEOLOGY AND STRATIGRAPHY OF...ANALYTICAL METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The Ipueira–Medrado sill crops out as a synform structure 7 km long, 0·5 km wide and 300 m thick (Fig. 1b). The sill is oriented parallel to the regional basement foliation, is disrupted into two segments, and is conformable with the underlying quartzo-feldspathic gneiss and with the overlying rocks, including olivine-bearing marble, calc-silicate rocks and metacherts (Deus & Viana, 1982
). Detailed mapping and extensive drilling carried out by Mineração Vale do Jacurici (FERBASA Group) has provided access to fresh rocks and to the complete stratigraphy of the sill. Primary igneous textures and minerals are largely preserved, except for the olivine-rich rocks (dunites) that are highly serpentinized and the noritic rocks that show the mineralogical consequences of regional amphibolite-facies metamorphism.
The sill is subdivided into three geographical segments (Ipueira Sul, Ipueira II and Medrado) and into three main zones from the base to the top (Figs 1b and 2): (1) the Marginal Zone; (2) the Ultramafic Zone; (3) the Mafic Zone. This subdivision results from a detailed study carried out by Marques (2001) and Marques & Ferreira Filho (2002) that integrated geology, petrography and compositional variation of olivine, orthopyroxene and chromite. The zonal subdivision was mainly based on rock composition. The Marginal Zone consists mainly of gabbroic rocks that could represent the lower margin of the sill. The Ultramafic Zone consists of ultramafic rocks and the Mafic Zone of only noritic rocks. The Marginal Zone (MZ) is 5–20 m thick and consists of highly sheared gabbros in the Ipueira Sul and Ipueira II segments and pyroxene-rich harzburgite containing silicates with relative low Mg in the Medrado segment. The MZ occurs at the underlying basement contact.
|
The Ultramafic Zone (UZ) is up to 250 m thick and is subdivided into three members: the Lower Ultramafic Unit (LUU), the Main Chromitite Layer (MCL) and the Upper Ultramafic Unit (UUU). The LUU is 100–180 m thick and consists mainly of dunite interlayered with minor harzburgite and with rare chromitite seams (<0·5 m thick). The MCL is 5–8 m thick and is subdivided into three sublayers. The lowest is a 0·5–1 m thick massive chromitite that is not continuous throughout the sill, but is thickest in the Medrado segment. A 0·3–0·6 m thick chain-textured chromitite (Fig. 3) forms a distinct sublayer in the lower part of the MCL that is continuous throughout the Ipueira–Medrado sill and is located either above the LUU or above the massive chromitite from the lower sublayer in the Medrado segment. The upper sublayer is a continuous and homogeneous 4–6 m thick massive chromitite. The MCL has sharp contacts with the host rocks and between sublayers. The UUU is up to 50 m thick and is composed of harzburgites interlayered with minor dunites. The amount of pyroxene increases with stratigraphic height grading to an orthopyroxenite at the top of the UUU, which is more consistent in the Ipueira II segment (pegmatoidal layer
5 m thick). Chromite-rich seams in the UUU are 0·3–1·1 m thick and are absent in the Medrado segment. Magmatic intercumulus amphibole (hornblende–tschermakite to Mg-hornblende) occurs in both LUU and UUU harzburgites but becomes more frequent toward the top, where it reaches 40 vol. % in orthopyroxenite. Amphibole occurs as large oikocrysts enclosing olivine, orthopyroxene and embayed chromite, and is clearly an intercumulus phase. The harzburgites have no plagioclase and only minor clinopyroxene, which is normally absent.
|
The Mafic Zone, at the upper part of the sill, consists of up to 40 m of norite with variable proportions of orthopyroxene and plagioclase with meso- to adcumulate primary igneous texture. Mg-hornblende is also common but the amount decreases toward the top, where magmatic amphibole is absent. When metamorphic transformation occurs the assemblage consists of plagioclase, hornblende, biotite and rare garnet, indicating the regional amphibolite-facies metamorphism of the area.
Magmatic evolution
The compositional variation of olivine and orthopyroxene in harzburgite samples allows identification of two magmatic regimes (Marques, 2001; Marques & Ferreira Filho, 2002). Regime 1 occurs through the LUU and is characterized by a slow upward increase in MgO/FeO (olivine Fo 89–93·5, orthopyroxene En 88–94·5) intensified immediately below the MCL. Regime 2 is expressed in the UUU and is typified by a fast upward decrease in MgO/FeO (Fo 90–84, En 90–82) (Fig. 2). Regime 1 is interpreted as an open magmatic system with concomitant injections of fresh primitive magma and extrusion of fractionated melt, and regime 2 as a closed magmatic system with minor injections of fresh magma (Marques, 2001; Marques & Ferreira Filho, 2002). The gradual shift toward more primitive composition that occurs in the LUU immediately below the MCL reflects variation in magma supply rate, which appears to be important in instigating the chromite crystallization that resulted in the formation of the MCL.
Sampling
All samples analyzed in this study were taken from a single drill core (I-328-55) from the Ipueira Sul area. The drill core is representative of the Ipueira–Medrado sill and intersected chain-textured chromitites both in the LUU and the UUU. The petrography and mineral composition of the samples analyzed for trace element and isotopic composition have been described by Marques (2001) and Marques & Ferreira Filho (2002).
Twelve whole-rock samples taken from the drill core were analyzed for trace elements (Table 1) and for Sm/Nd isotopic composition (Table 2). Sample numbers represent their depth in the drill core measured in meters. It should be noted that the stratigraphy is inverted and therefore shallower depths represent the base of the sill and greater depths the top of the sill. The samples were selected from the less serpentinized and altered core sections representing the various zones and units of the intrusion; including samples above and below all the chromitite seams intersected. The strategy of the sampling had the following aims: (1) to allow a comparison between the Lower Ultramafic Unit and the Upper Ultramafic Unit avoiding lateral variations (only one drill core); (2) to observe possible changes below and above each chromitite seam; (3) to constrain the composition of the Marginal Zone; (4) to assess the significance of intercumulus amphibole. The location of each sample according to its relative stratigraphic position and a brief description are given in Fig. 4.
|
|
|
Twelve additional samples were selected for Re/Os isotopic analyses (Table 3). The freshness and the relative stratigraphic position were considered during the sampling. Five whole-rock samples of silicate rocks (two harzburgites from the LUU, two harzburgites from the UUU and one pyroxenite from the UUU) and six samples from chromitites (four from chain-textured seams and two from massive layers) were analyzed. All the chain-textured samples had their chromites separated for Re–Os analyses. The massive samples were analyzed using whole-rock aliquots and one sample (304.25) was analyzed as both whole rock and separated chromite. The location of each sample and brief description are shown in Fig. 4.
|
|
ANALYTICAL METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONTECTONIC SETTINGGEOLOGY AND STRATIGRAPHY OF...ANALYTICAL METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Sample preparation
All the whole-rock samples were reduced to representative rock powders (<200 mesh) using ceramic crushing equipment to avoid contamination. Chromite separates were obtained in a four-stage process. First, the chromitite samples were broken using a Teflon shield and rough crushed by hand using ceramic tools to a grain size where most of the grains were free of silicate. After being washed with deionized water and dried, the chromite was magnetically separated by hand and by Isodynamic Franz equipment. Further purification was obtained by hand-picking under a binocular microscope. The separate was finally powdered to less than 200 mesh in ceramic equipment by hand.
Trace element procedures
Trace element compositions were determined by inductively coupled plasma mass spectrometry (ICP-MS) in the Activation Laboratories Ltd., Canada. The sample digestion followed the lithium metaborate–tetraborate fusion technique. The resulting molten bead was rapidly digested in a weak nitric acid solution. The fusion technique ensures that the entire sample is dissolved, particularly elements such as rare earth elements (REE) in resistate phases. The sample solution was spiked with internal standards to correct for minor drift and contains a light, medium, and heavy element across the mass range and is incorporated into the final analyses. The solution was diluted and introduced into a Perkin Elmer SCIEX ELAN 6000 ICP-MS using a proprietary sample introduction technique. Twelve international certified reference materials were run with the samples including the following: W2 and MRG-1 as control materials, and MAG-1, IR1, DNC1, GXR-2, LKSD-3, Mica Fe, GXR1, SY3, STM-1 and IFG-1 as calibration standards. The detection limits are presented in Table 1.
Sm–Nd analytical procedures
Sm–Nd isotopic analyses were performed at the Laboratório de Geocronologia of the Instituto de Geociências of the Universidade de Brasília. The procedures used were similar to those described by Gioia & Pimentel (2000). Sample digestion was executed using 100–130 mg of rock sample mixed with the spike. The mixture was dissolved in Teflon bombs in a steel jacket, using 1 ml of distilled concentrated HNO3 and 4 ml of distilled concentrated HF. After evaporation, the sample was dissolved again in an HF–HNO3 (4:1) mixture and placed back in the oven at 190°C for 4 days. The separation of the REE was performed in a quartz column (i.d. 8 mm, height 15 cm) packed with 2·2 g or 12 cm of Bio-Rad AG 50W-X8 200–400 mesh cation resin in aqueous solution. The sample solution (750 µl) was eluted in the column using HCl. The REE were collected in the fraction between 1 and 15 ml of 6N HCl, after elution with 32 ml of 2·5N HCl. The separation of Sm and Nd was performed in a Teflon (Savilex) column (i.d. 5 mm, height 10 cm) packed with LN-Spec resin (liquid resin HDEHP 270–150 mesh powdered Teflon coated with di-ethylexil phosphoric acid). The REE fractions were totally evaporated and redissolved in 200 µl of 0·18N HCl. This solution was loaded into the LN-Spec column. The Nd fraction was collected in 4 ml of 0·3N HCl after the initial 10 ml of 0·18N HCl. After the extraction of Nd, 2 ml of 0·3N HCl were discarded and the Sm fraction was collected in 3 ml of 0·4N HCl. The Nd solution was loaded onto Re filaments and slowly evaporated. The mass spectrometer used was a Finnigan MAT 262, equipped with seven Faraday cup collectors. The analyses were performed in static mode using double-filament assemblies. The 143Nd/144Nd ratio was normalized to 146Nd/144Nd = 0·7219 and the decay constant used was 6·54 x 10-12/yr (Lugmair & Marti, 1978). The internal uncertainties for 143Nd/144Nd ratios varied from 0·010 to 0·003% and the external analytical uncertainty in 147Sm/144Nd determination is 
0·10%, based on repeated analysis of the international rock standard BHVO-1. Standard basalt BHVO is used as an internal standard in this laboratory. During the period of these analyses, our average 147Sm/144Nd and 143Nd/144Nd with 2 uncertainties for BHVO were 0·151 ± 0·002 and 0·512980 ± 0·000022, respectively, which we take to represent the external reproducibility of our Sm–Nd measurements. Samples in Table 2 with errors larger than this value reflect the low Nd concentration of these samples and hence the small amounts of Nd used for isotopic analysis.
Re–Os analytical procedures
The Re–Os analyses were performed at the Department of Terrestrial Magnetism of the Carnegie Institution of Washington. The procedures used were similar to those described by Carlson et al. (1999). A weighed amount of sample powder was added to a Pyrex Carius tube (Shirey & Walker, 1995) kept at low temperature in a dry-ice–methanol slurry. Also added to the Carius tube was a weighed quantity of a mixed 185Re–190Os tracer solution followed by the dissolution acid consisting of 2 ml concentrated HCl followed by 4 ml concentrated HNO3. All chromite samples were completely dissolved by this procedure to produce a clear deep green solution. Because of their large size (2 g), the silicate samples were taken from the oven, ultrasonicated for 4 h to break up the sample powder and then placed back in the oven for an additional night of heating. Oxidized OsO4 was extracted from the aqua regia three times into CCl4 (a total of 9 ml CCl4). With each step, the CCl4 solution was removed by pipet and added to a Teflon beaker containing 4 ml concentrated HBr. The Os was further purified by microdistillation (Roy-Barman & Allègre, 1994) and then loaded onto Pt filaments, which were then heated to 500°C in vacuum for 30 min. BaNO3 (20 µg) was added to the filament, which was then loaded into the mass spectrometer. Re remained in the aqua regia solution. Re was separated from the aqua regia using a two-column anion exchange procedure with HNO3 and HCl as elutants (Carlson et al., 1999).
Mass spectrometry of OsO3- and ReO4- was performed by peak hopping on the single pulse counting electron multiplier on the DTM 15-inch mass spectrometer. Signal sizes of 192Os > 100 Kilo-Count-per-second (KCps) were obtained for all chromite samples whereas the silicate samples were run at 192Os signal sizes from 10–70 KCps. Concentration uncertainties for whole-rock Re–Os analyses generally range from 1 to 5% because of inhomogeneous distribution of the trace phases containing these elements. Re/Os ratios in standard solutions show a reproducibility of 0·1%. Blanks for Re and Os were both 1 ± 0·5 pg. Many of the chromites had Re concentrations near blank levels, which explains the high uncertainties listed for these analyses, as the blank is assumed to have an uncertainty of ±50%, and the effect of this uncertainty is included during blank correction. Os blank corrections were insignificant for the chromite samples, but were significant for the low Os content silicate rocks. For these samples, blank correction was performed on both the Os concentration and isotopic composition using a blank isotopic composition of 187Os/188Os = 0·1805. The uncertainty of the blank correction was quadratically added to the uncertainty of the reported isotopic composition and dominates the error for samples 335.33 and 348.30. Errors reported in Table 3 are the larger of the 2 SE deriving from mass spectrometry or the error after blank correction.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONTECTONIC SETTINGGEOLOGY AND STRATIGRAPHY OF...ANALYTICAL METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Trace element geochemistry
All of the samples selected for geochemical analysis are cumulate rocks. Therefore their compositions are dependent on the type and amount of the cumulus phases and the amount of trapped intercumulus liquid. All the ultramafic samples exhibit some degree of serpentinization and some level of element mobility is expected, except for the highly immobile high field strength elements (HSFE).
Primitive mantle-normalized trace element diagrams for the mafic and ultramafic rocks of the Ipueira–Medrado sill are shown in Fig. 5. Trace element abundances for samples from the UUU and the LUU are similar, suggesting that ultramafic cumulates from both units crystallized from similar parental magmas. Disregarding the scattering of the data caused by the obvious mobility of some elements (i.e. Cs, Rb, Sr), abundances of the large ion lithophile elements (LILE, e.g. Ba, Th, U) are typically greater than those for the HFSE (e.g. Zr, Ta, Nb) and REE. Another notable feature is the very low Ta and relative high Zr. The concentration of incompatible elements in the harzburgite cumulates is related mainly to the amount of postcumulus material; thus the ultramafic cumulates show similar patterns, but broad concentration variation.
|
The cumulate gabbro from the Marginal Zone has a distinct composition when compared with the mafic and ultramafic rocks from the LUU and the UUU. The normalized incompatible element abundance pattern shows strong enrichment in the most incompatible elements with negative abundance anomalies for the HFSE Ta and also for Zr.
Rare earth elements (REE)
Mafic and ultramafic rocks of the Ipueira–Medrado sill, excluding the marginal gabbro, show similar chondrite-normalized REE patterns with enrichment in light REE (LREE) and a generally flat pattern for the heavy REE (HREE) (Fig. 6) with (Gd/Yb)N about unity. The LREE concentration in most of the samples is 5–10 times chondritic and the HREE concentration is <5 times chondritic. Eu anomalies are both positive and negative in the ultramafic rocks and slightly positive to strongly positive in the mafic rocks. The gabbro from the Marginal Zone shows a distinct chondrite-normalized pattern with the slope in LREE similar to that of the ultramafic rocks [(Ce/Sm)N = 3·05], but a strong depletion in HREE [(Gd/Yb)N = 6·84] (Fig. 6a).
|
The Marginal Zone gabbro has a distinct trace element content and chondrite-normalized REE pattern when compared with other rocks from the sill. The significant enrichment in elements such as Sr, Rb, Th, U, Pb and Ba, the relative depletion in Zr, Cr and Ni, and the highly depleted HREE pattern strongly suggest that this sample derives from a different parental magma than do most of the other rocks of the Ipueira–Medrado sill.
The ultramafic rocks from the Lower Ultramafic Unit have the lowest content of REE with 
REE ranging from 3·58 in olivine-rich harzburgites to 9·69 in harzburgite with intercumulus amphibole. The LREE enrichment among the harzburgites is similar, with (Ce/Sm)N ratio in the range of 2·7–2·9 (Fig. 6c). The olivine-rich harzburgites have a distinct pattern with high LREE slope, low middle REE (MREE) content and HREE enrichment resulting in a U-shaped pattern, most evident in sample 267.12, which is the most olivine rich. A similar pattern occurs in sample 183.83, except that this sample has a higher REE content and a positive Eu anomaly.
The harzburgites from the Upper Ultramafic Unit have a LREE slope similar to those from the Lower Ultramafic Unit, but show slightly more variability with (Ce/Sm)N ranging from 1·9 to 2·42 (Fig. 6b). The REE of the rocks from the UUU is also variable, ranging from 4·75 to 34·62. The lowest contents occur in harzburgite 310.88 immediately above the Main Chromitite Layer and below the first chain-textured chromitite of the UUU, which also has a distinctive REE pattern (Fig. 6b). The orthopyroxenite (sample 348.30) has a REE pattern similar to the harzburgites, except for its steeper LREE slope, with (Ce/Sm)N ratio of 3·6. This sample is chromite rich, has a high amount of intercumulus amphibole and marks the transition to the mafic rocks.
The two samples of norites from the Mafic Zone have distinct REE patterns (Fig. 6a). Sample 355.50 has lower total REE content (REE = 22·94 ppm) and shallower LREE slope [(Ce/Sm)N = 1·86], with a very small positive Eu anomaly. Sample 369.75 is very different, with REE 35 ppm, steep LREE slope [(Ce/Sm)N ratio = 6·5], flat MREE to HREE pattern and a strong positive Eu anomaly, probably related to the fact that it is a leuconorite with cumulus plagioclase.
Sm–Nd systematics
The Sm–Nd isotopic compositions of the analyzed samples are given in Table 2. The main characteristics are the low Sm (<1·5 ppm) and Nd (<6 ppm) contents of all samples except for the marginal gabbro (149.69), the small range in Sm/Nd ratios and the variable Archean Nd (CHUR) model ages and negative Nd for most of the samples.
Including all analyses, the data show considerable scatter in a Sm–Nd isochron diagram. Seven samples define a reasonable trend that yields a 2229 ± 68 Ma age and -5·1 initial Nd with mean square weighted deviation (MSWD) of 4·9 (Fig. 7a). This age could reflect the crystallization age of the sill; however, the scatter about the best-fit line, the strongly negative initial Nd and the high volume of amphibole in some samples could indicate that the magma was contaminated with crust and thus did not maintain a constant Nd isotopic composition during the crystallization of the sill. A combination of fractional crystallization and assimilation of LREE-rich crust could result in an artificially old isochron. Considering this possibility, another approach to interpreting the age systematics of the data is attempted.
|
The samples were separated into two groups; amphibole-free samples and samples containing
25% amphibole. Samples from the norites and those that were strongly serpentinized (267.12) were excluded. In addition, samples with >5% and <20% of amphibole (169.48) were excluded, to examine possible isotopic differences between the parental magma at the time it was crystallizing amphibole-free rocks as opposed to later crystallization of amphibole-rich rocks. The isochron obtained for the amphibole-free group yields a good linear correlation with an indicated age of 1985 ± 45 Ma and -4·7 initial Nd (MSWD = 0·25) (Fig. 7b). The amphibole-rich group, including the sample from the margin, yields an isochron of 2063 ± 84 Ma age and -6·5 Nd (MSWD = 0·67) (Fig. 7c). These almost parallel isochrons, both with very low MSWD and relatively low age uncertainty despite the small range in the Sm/Nd ratios and the small number of samples, probably reflect the true crystallization age of the sill. The lower initial Nd for the amphibole-rich and margin samples is consistent with the suggestion made previously that these samples have a larger crustal contribution than the amphibole-free samples. Both of these ages overlap the 2038 ± 19 Ma zircon Pb age reported by Oliveira & Lafon (1995), which we from now on consider as the crystallization age of the sill.
The amphibole-rich and amphibole-free samples are also clearly distinguished if Nd is calculated for each sample using the 2038 Ma zircon age as the crystallization age of the sill. The mean Nd for the amphibole-free harzburgites is -4·4, whereas the mean for the amphibole-rich samples is -6·5 (Fig. 8). The marginal gabbro (144.69) from the base of the sill has Nd = -6·7 and the norites from the top of the sill have strongly negative Nd up to -9·5 (see Table 2).
|
Re–Os systematics
Re and Os concentrations and Os isotopic compositions for the analyzed samples are given in Table 3. The harzburgite and pyroxenite samples have extremely low Re concentrations and low to moderate Os concentrations. All these samples have moderately radiogenic measured Os isotopic compositions, and their relatively low Re/Os ratios cause these samples to have calculated initial 187Os/188Os (
Os(2038 Ma) from +18 to +235) substantially higher than typical of the mantle (Os(2038 Ma) = 0 to +3). In light of the sub- to near-chondritic initial Os isotopic compositions of the chromites (Os(2038 Ma) = -4·6 to +3·3), we interpret the high calculated initial 187Os/188Os of the silicate samples to reflect Re loss during serpentinization of these samples leading to erroneous values for the initial Os isotopic composition, although it is possible that the elevated initial Os isotopic compositions of the silicate rocks reflect crustal contamination. The silicate samples do show a general correlation between 187Os/188Os and 1/Os content that might be expected for contamination. However, this explanation would require that the chromitites that lie immediately above or below these silicate rocks would have formed from a distinct, more primitive magma than that crystallizing the silicate rocks just before and after chromite crystallization.
The very low Re concentrations (near blank levels) in chromite mineral separates, which have high Os concentrations, agree with previous studies that suggest that Os can be structurally accommodated in chromite or occur as very small inclusions of Os-rich platinum group minerals within chromite (Talkington et al., 1983; Capobianco & Drake, 1990; Fleet et al., 1991; Marcantonio et al., 1993). The whole rock from sample 304.25 has slightly less Os than the chromite mineral separate from the same sample, which can be attributed to some dilution with Os-poor mineral phases as previously considered by Marcantonio et al. (1993) for Stillwater chromitites. The low 187Re/188Os values of the chromites preclude definition of a Re–Os isochron.
Os(2038 Ma) is the percent deviation of the sample's Os isotopic composition from that of a chondritic reservoir at the time of crystallization (Walker et al., 1989). Os(2038 Ma) values of the chromite samples show differences between the LUU and MCL chromitites and those from the UUU. The Os values of the LUU and MCL chromites are negative (Os = -4·6 to -3·2) whereas those of the UUU chromites are slightly positive (Os = 1·4 to 3·3). This up-stratigraphy increase in initial Os isotopic composition is present even in the MCL chromites, where the lowest sample (302.85) has the most negative Os(2038 Ma) of -4·6, the middle sample (304.25) has initial Os of -3·3 and the upper MCL sample (305.67) has initial Os of -0·3 (Fig. 9).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONTECTONIC SETTINGGEOLOGY AND STRATIGRAPHY OF...ANALYTICAL METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Retrieving the parental magmacomposition
The parental magma composition of the Jacurici Complex can only be assessed indirectly as no chilled margin has been identified. Many studies have attempted to determine the parental magma composition from mafic–ultramafic complexes using the composition of cumulate rocks or cumulate mineral separates and mineral–melt partition coefficients (Lambert & Simmons, 1987
; Cawthorn et al., 1991; Chai & Naldrett, 1992; Grant & Chalokwu, 1992; Bédard, 1994; Papike et al., 1995; Ferreira Filho et al., 1998). Such calculations are hampered by the difficulty in estimating the amount of intercumulus liquid present and the degree of interaction between the cumulate phases and the intercumulus liquid. Barnes (1986) and Cawthorn (1996) have studied the effect on the composition of cumulus phases produced by different amounts of trapped liquid.
The parental magma composition of the Ipueira–Medrado sill was calculated using a method similar to that applied by Cawthorn et al. (1991) and Ferreira Filho et al. (1998). The whole-rock REE composition, modal proportion of cumulus phases and estimated amount of intercumulus liquid are considered in the following equation:
 | (1) |
j is the wt % of the cumulus phase j in the rock,
 |
is the partition coefficient of element i for cumulus phase j, 
is the concentration of element i in the magma, and 
is the concentration of element i in the rock.
An example of the application of equation (1) is demonstrated in equation (2), where the Ce concentration is calculated for a harzburgite:
 | (2) |
The Ce, Sm and Yb concentrations of the parental magma were calculated from five harzburgite samples (169.48, 183.83, 232.70, 326.40 and 335.33) using the orthopyroxene partitioning coefficients from Schwandt & McKay (1998) and olivine partitioning coefficients from McKay (1986) and Beattie (1994). To evaluate the range of variation caused by intercumulus trapped liquid, four intercumulus liquid amounts were used. The chondrite-normalized REE pattern of the calculated parental magma (Fig. 10a and b) does not vary in shape when different degrees of trapped liquid are considered, as previously shown by Cawthorn et al. (1991). Furthermore, the calculated parental magma compositions exhibit virtually the same REE patterns as the cumulate rock (Fig. 10a and b), only Yb is relatively higher in the cumulate rocks than in the respective calculated parental magma because olivine and, mainly, orthopyroxene have higher partition coefficients for this element.
|
The above calculations indicate that the shapes of the cumulate REE patterns are controlled by the parental magma composition, with only minor influence of the cumulate phases present in the harzburgites. However, the total amount of REE varies significantly, being
5–10 times more in the calculated parental liquid than in the cumulate rock, depending on the amount of trapped intercumulus liquid. In the Ipueira–Medrado sill we propose that the amount of trapped intercumulus liquid correlates with the amount of intercumulus amphibole (also indicated by the higher abundances of incompatible elements such as Hf, Zr and Th in those samples). On this basis, we estimate the presence of 5–10% intercumulus liquid for samples 169.48 and 183.83, and 15–20% for samples 232.70, 326.40 and 335.33. The calculated parental liquid for these samples defines an envelope that is probably the best approximation to the parental magma of the sill (Fig. 10c). The REE pattern calculated for the parental magma does not show the HREE depletion of the marginal gabbro 144.69, supporting the conclusion that this gabbro derives from a different parental magma than the majority of the sill. The composition obtained for the parental magma of the ultramafic cumulates is very similar to that from the B1/U type magma from the Bushveld Complex (Harmer & Sharpe, 1985) and one of the parental magmas calculated for the Stillwater Complex (Lambert & Simmons, 1987) (Fig. 10c).
Nd and Os isotope constraints
The Sm–Nd systematics for all samples analyzed from the Ipueira–Medrado sill show considerable scatter about an isochron, even when the most serpentinized samples are removed. We interpret this as an indication that the rocks did not have the same Nd isotopic composition at the time they formed. The separation of the samples into amphibole-free and amphibole-rich samples yields nearly parallel Sm–Nd isochrons that overlap in age the 2038 Ma zircon age obtained by Oliveira & Lafon (1995). These two isochrons have distinctly different initial Nd isotopic compositions.
The more negative Nd of the amphibole-rich samples at 2038 Ma suggests that these samples may have formed from a magma that was more crustally contaminated than that which formed the amphibole-free samples. This enhanced degree of crustal contamination is reinforced by the fact that the marginal gabbro and the norites from the top of the sill, the rocks most likely to interact with the surrounding crustal rocks, have even more negative Nd than the amphibole-rich samples. Even the amphibole-free samples, however, have strongly negative Nd (mean -4·4). This indicates either that all the magmas involved in the formation of the cumulates had experienced substantial amounts of crustal contamination, or, alternatively, that the uncontaminated parental magma was derived not from the convecting mantle, but from old metasomatically enriched subcontinental lithospheric mantle.
At 2038 Ma, the Os isotopic compositions of the chromite separates give Os values that range from negative (-4·6) to slightly positive (+3·3). Os is compatible in the mantle during melting, whereas Re is incompatible (Walker et al., 1988, 1989). Continental lithospheric mantle generally is composed of the refractory residue of partial melting and is characterized by low Ca, Al and Re abundances, and consequently will evolve to negative Os (Walker et al., 1989; Carlson & Irving, 1994). Thus, the negative initial Os of some of the chromitites is compatible with a parental magma derived from a Re-depleted peridotitic lithospheric mantle source (Ellam et al., 1993; Lambert et al., 1994). Lithospheric mantle also commonly experiences metasomatic enrichment of the incompatible lithophile elements, and thus low Sm/Nd and negative Nd are a common characteristic of old metasomatized lithospheric mantle. In fact, continental lithospheric mantle seems to be the only terrestrial reservoir that combines negative Os and Nd. The oldest Re-depletion model ages for the chromites, and most of the Sm–Nd model ages (depleted mantle model ages will be about 100–200 Myr older than the CHUR ages listed in Table 2) for the silicate rocks from the Ipueira–Medrado sill are Archean. Whereas the Sm–Nd model ages need have no strict age significance because of both cumulate fractionation of Sm/Nd and crustal modification of the magma's Nd isotopic composition, the old Sm–Nd and Re depletion model ages suggest that the parental melt to the sill may have been derived by melting of previously melt depleted, but metasomatically enriched, subcontinental lithospheric mantle and hence was characterized by negative Os and Nd before undergoing any crustal contamination.
Both the chromites and the silicate rocks of the Ipueira–Medrado sill display a fairly wide range in initial Nd and Os isotopic compositions. Nd becomes more negative with the presence of increasing amounts of modal amphibole, and Os becomes less negative with increasing stratigraphic height, both of which are suggestive of a magma composition evolving through combined fractional crystallization and crustal assimilation. Figure 11 shows the modeled change in the Nd and Os isotopic compositions of possible parental magmas to the sill undergoing crustal contamination. The amount of crust needed to produce the variations shown in Fig. 11 depends strongly on the assumed Nd and Os concentrations of both the parental magmas and the crustal component. The Nd concentration in the parental magma is constrained to be high (20 times chondritic) according to the modeling discussed above (Fig. 10c). Such high Nd concentrations in the parental magma make the Nd isotopic composition of the magma relatively insensitive to changes caused by crustal contamination. In the example shown in Fig. 11, if the parental magma started with Nd = +4, a value typical of depleted MORB-like mantle at 2 Ga, almost 50% by mass assimilation of an Archean crustal component, even one characterized by the evolved Nd and Os characteristics as defined in Fig. 11, is needed to produce Nd = -4, the average value observed for the amphibole-free samples studied here. To go from Nd = -4 to Nd = -7 as observed in the amphibole-bearing samples would require an additional 10% assimilation, bringing the total crustal component to 60%. At this ratio of crust to primary magma, the major element composition of the magma is likely to be too siliceous and Mg poor to precipitate the ultramafic cumulates that make up the sill. One could appeal to a more mafic crustal component than used in the modeling shown in Fig. 11, but such a component probably would have lower Nd concentration and higher Nd isotopic composition requiring even more crustal contamination to create the observed range in Nd isotopic composition of the sill samples. Similarly, a more mafic crustal component might have higher Os concentration, which would cause the mixing curves shown in Fig. 11 to flatten, reaching negative Nd only at very positive Os. Another alternative is that the parental magma was derived from a source more like that of modern ocean island basalts (i.e. lower Nd, higher Nd concentration), but most modern ocean-island basalts have positive Os (e.g. Pegram & Allègre, 1992; Reisberg et al., 1993), and thus could not explain the negative Os of some of the chromites. The amount of crustal contamination needed to reach the negative Nd of the samples could be reduced by assuming a lower Nd concentration in the primary magma, but the Nd concentration would then not rise sufficiently quickly with fractionation or contamination to produce the magma REE abundances shown in Fig. 10c.
|
A convecting mantle source for the parental magma to the sill is difficult to reconcile both with the amount of crustal contamination needed to explain the Nd isotopic composition of the cumulates and with the negative
Os seen in some of the chromites. Crustal rocks are characterized by high Re/Os and hence radiogenic Os isotopic compositions (Walker et al., 1991; Ravizza & Turekian, 1992). Thus, crustal contamination will not lead to negative Os starting from the zero to positive Os typical of melts of the convecting mantle (Martin, 1991; Pegram & Allègre, 1992; Reisberg et al., 1993; Roy-Barman & Allègre, 1994). If instead of assuming a convecting mantle source for the primary magma, we assume this magma was generated from Archean subcontinental lithospheric mantle, the primary magma may have started with Nd = -4 and Os = -4·6. Because the peridotitic portions of subcontinental lithospheric mantle are a refractory residue enriched in such compatible elements as Mg, Cr and the platinum group elements, and metasomatism brings in elements such as the LREE, partial melts of the subcontinental lithospheric mantle will be Mg rich with high concentrations of both compatible elements such as Cr and Os, and incompatible elements such as Nd. If sufficiently water rich, subcontinental lithospheric mantle melts also may be silica rich (Tatsumi et al., 1986). This combination of compositional characteristics will make the primary magma likely to precipitate both olivine and orthopyroxene, as is observed in the cumulate sequence in the Ipueira–Medrado sill. Crustal contamination acting on this primary magma also could instigate orthopyroxene crystallization and take the Nd isotopic composition from Nd = -4 to -7 and Os from -4·6 to zero with as little as 30% assimilation (Fig. 11). This crustal assimilation could have initiated the chromite crystallization that formed the Main Chromitite Layer, as the chromites from this layer show stratigraphically correlated variation in initial Os that ranges from -4·6 at the base to -0·3 near the top of the layer.
Petrogenesis of Ipueira–Medrado sill
An evaluation of the petrogenesis of the Ipueira–Medrado sill is not easy because the parental magma composition can only be indirectly inferred from the cumulate samples. Nevertheless, the coherent pattern of the trace elements throughout the sill and the REE modeling suggest that the parental magma was enriched in LREE and LILE, depleted in Ta and enriched in Zr.
The geotectonic setting of the region in which the Jacurici Complex intruded is not well constrained and to understand the nature of the magmatism is even more difficult. Nevertheless, the general characteristics of the parental magma of the Ipueira–Medrado sill argue for two main possibilities. One possibility is that the parental magma of the Jacurici Complex was a high-Mg basalt or komatiitic magma generated from partial melting of the convecting mantle and subsequently enriched in LREE and LILE by interaction with an old Archean crustal component. Some evidence such as the enrichment in Si, Al and the high amount of fluids, as indicated by the crystallization of orthopyroxene instead of clinopyroxene, the high Al content in pyroxene and in chromite, and the presence of amphibole as an intercumulus phase (up to 30% in harzburgites) could be explained by a transfer of fluids from the surrounded supracrustal rocks to a very hot and dry mafic or ultramafic magma via assimilation of wall rocks. Such contamination could explain the enrichment in LILE and LREE, without adding substantial HFSE. Experimental studies have demonstrated that the HFSE are fairly immobile in fluids (Tatsumi et al., 1986; Keppler, 1996; You et al., 1996; Kogiso et al., 1997).
Another option would be a magma extracted from an old metasomatized subcontinental lithospheric mantle source forming the root of the Archean craton. In this case, the high LILE and LREE contents and the anomalously low Ta content and relative high Zr could be explained by a Ti-rich residual phase in the enriched mantle as the style of the mantle enrichment is strongly dependent on the minerals involved (O'Reilly & Griffin, 1988; O'Reilly et al., 1991). In this option, the parental magma need not have been contaminated by crustal rocks before intrusion into the sill.
The Sm–Nd systematics are consistent with a Paleoproterozoic crystallization age for the sill. The negative initial Nd in all samples suggests that the parental magma either originally suffered crustal contamination or was derived from an old metasomatically LREE-enriched subcontinental lithospheric mantle source. The more negative Nd of the amphibole-rich samples (mean -6·5) and marginal rocks of the sill along with the variable Archean Nd model ages argue for crustal contamination occurring during crystallization of the sill, as does the increasing Os of the chromites up-stratigraphy. These indications of crustal contamination do not necessarily preclude the possibility that the negative initial Nd of the amphibole-free samples (mean -4·4) implies a primary, uncontaminated magma from the subcontinental lithospheric mantle characterized by negative Nd and Os. The negative Os, in particular, cannot be generated by crustal contamination of a primary magma from the convecting mantle, but instead is suggestive of a magma source in an old, Re-depleted, peridotitic lithospheric mantle.
The Os and Nd isotopic results are interesting when the stratigraphy of the sill is considered. In the LUU, amphibole-rich harzburgite samples are scarce and more common close to the base of the sill. The LUU chromitite and the lower part of the MCL both have chromite separates with negative Os (-3 to -4). On the other hand, the UUU has abundant amphibole-rich, harzburgite samples with more negative Nd and chromitites with positive Os (up to +3·3).
Negative Os and Nd in a primary magma could be achieved in a situation where a very primitive high-Mg magma suffered little or no crustal contamination. In this case, the negative Os associated with negative Nd in the LUU would suggest that this interval has suffered little or no crustal contamination, whereas the positive Os and more negative Nd from the UUU indicates that this interval experienced more crustal contamination. The more negative Nd values from the marginal gabbro and norites from the top of the sill are consistent with this model.
Given a primary magma with low Nd, but high Os content, the negative Nd and Os of the LUU could be explained by crustal contamination with the Nd and Os concentrations of the crustal end-member adjusted to primarily affect the Nd isotopic compositions. This type of contamination, however, cannot produce the observed enrichment in LREE and LILE if the source of the magma was the convecting mantle. The stronger effect of contamination on Nd than on REE content can be demonstrated by the decrease of Nd with increase of modal proportion of amphibole, while the total REE content remains roughly constant (Fig. 12).
|
Trace element abundances, normalized REE patterns, and Nd and Os isotopic data from the Ipueira–Medrado