Eudialyte-group minerals from the Monte de Trigo alkaline suite , Brazil : composition and petrological implications

Manuscript: 20160075. Received on: 06/13/2016. Approved on: 08/09/2016 ABSTRACT: The Monte de Trigo alkaline suite is a SiO2undersaturated syenite-gabbroid association from the Serra do Mar alkaline province. Eudialyte-group minerals (EGMs) occur in one nepheline microsyenite dyke, associated with aegirine-augite, wöhlerite, låvenite, magnetite, zircon, titanite, britholite, and pyrochlore. Major compositional variations include Si (25.09– 25.57 apfu), Nb (0.31– 0.76 apfu), Fe (1.40–2.13 apfu), and Mn (1.36– 2.08 apfu). The EGMs also contain relatively high contents of Ca (6.13– 7.10 apfu), moderate enrichment of rare earth elements (0.38–0.67 apfu), and a relatively low Na content (11.02–12.28 apfu), which can be correlated with their transitional agpaitic assemblage. EGM compositions indicate a complex solid solution that includes eudialyte, kentbrooksite, feklichevite, zirsilite-(Ce), georgbarsanovite, and manganoeudialyte components. EGM trace element analyses show low Sr and Ba contents and a negative Eu/Eu* anomaly, which are interpreted as characteristic of the parental magma due to the previous fractionation of plagioclase and/or alkali feldspar. The EGMs from the dyke border have higher contents of Fe, Sr (2,161–2,699 ppm), Mg (1,179–3,582 ppm), and Zn (732– 852 ppm) than those at the dyke center. These differences are related to the incorporation of xenoliths and xenocrysts of melatheralitic host rock into the nephelinesyenitic magma followed by crystal-melt diffusive exchange.

The present paper reports the occurrence of EGMs from the alkaline suite of Monte de Trigo, a small island (ca. 1 km 2 ) situated near the towns of Sao Paulo and Rio de Janeiro, at the western border of the Santos Basin (23º53'S, 45º47'W), in southeastern Brazil.They describes major and trace element analyses of these EGMs, and their composition is discussed in terms of site assignment and possible ideal components.The objectives of the study were to characterize the EGMs at Monte de Trigo and to understand how its petrological environment influences the major and trace element compositions of the EGMs.
The Monte de Trigo alkaline suite (86.6 Ma, Enrich et al. 2009) is a small, SiO 2 -undersaturated syenite-gabbroid intrusive suite (Fig. 1A) that belongs to the Serra do Mar alkaline province.According to Enrich et al. (2009), the alkaline suite is composed of different pulses of undersaturated alkaline magmas.The oldest unit comprises a cumulate nepheline-bearing olivine melagabbros, melatheralites (i.e., melanocratic nepheline gabbro, Le Maitre 2002) and clinopyroxenites with wide modal variations and layering structures.A leucocratic medium-to-coarse-grained nepheline syenite and nepheline-bearing alkali feldspar syenite intrude the cumulate mafic-ultramafic unit and compose most of the area of the suite.Alkali feldspar, hedenbergite, hastingsite, and biotite are early magmatic phases, whereas nepheline and analcime are late-crystallized phases.Accessory minerals include apatite, magnetite, and titanite and characterize a miaskitic association.Miaskitic and agpaitic nepheline microsyenite dykes cut the cumulate mafic-ultramafic body and the nepheline syenite stock and are interpreted as late-stage differentiates of the nepheline syenite (Enrich et al. 2009).
Near the border, the EGM-bearing dyke shows an increase in the mafic mineral content, locally up to 25 vol%, as well as some mafic aggregates (Figs.1C and 3A).The latter is composed of aegirine-augite and magnetite (Fig. 3B) and is interpreted as corroded and re-equilibrated xenoliths and/ or xenocrysts of the melatheralite host-rock.Accessory minerals at the dyke border are EGMs and låvenite (Figs.2C  and 3B).Late-magmatic phases include poikilitic titanite, zirconolite, pyrochlore, and apatite.
The melatheralite host rock was transformed into a metasomatic granofels on a centimeter scale near the contact with the dyke, by its reaction with the alkaline volatile-rich liquid.Olivine, augite, and plagioclase from melatheralite were replaced by aegirine-augite, magnesiokatophorite, biotite, magnetite, and potassic feldspar (Fig. 3A).Aegirine-augite is the most abundant mineral within the first millimeter of the granofels.The granofels also has relict cores of augite with andradite rims, biotite, and magnesiokatophorite pseudomorphs after augite and masses of potassic feldspar after plagioclase (Fig. 3A and 3C).

ANALYTICAL METHODS
Major element analyses of the EGMs were performed on standard carbon-coated thin sections in a Jeol JXA-8600 electron microprobe using a Noran Voyager 4.1 automation system at the Geoanalítica USP Core Facility of the University of São Paulo (USP).The analyses of standards and unknown samples were performed according to the electron microprobe settings of Williams (1996) (accelerating voltage of 25 kV and beam current of 100 nA) to properly quantify REE, HFSE, and other trace elements.The peak counting times were 10-20 s (F, Cl, Na, Si, K, Ca, Mn, Fe, Mg, Ti, and Al) and 40-60 s (La, Ce, Pr, Nd, Gd, Er, Yb, Y, Zr, Hf, Nb, Ta, U, Th, and Pb); the background counting times were half of the corresponding peak values.The Na and F contents were measured in the first minute of the analysis to prevent volatilization/migration due to the high probe current.A 30-μm beam diameter was also employed to avoid volatilization/ migration.The small EGM grain size in the studied samples together with the large beam diameter precluded a core-to-rim characterization.A combination of well-characterized natural and synthetic standards was used.The data were processed using the PROZA procedure (Bastin et al. 1984).Special care was taken in the selection of the peak and background position to eliminate peak overlap.The detection limits were in the range of 0.01-0.03wt%, except for F (up to 0.15 wt%).
After the removal of carbon coating from the thin sections, in situ trace-element analyses were performed on the same grain and as close as possible to the microprobe spot analysis using the laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) at the Geoanalítica-USP Facility of the USP.The analytical work utilized a Perkin Elmer Elan 6100 DRC ICP-MS instrument coupled with a neodymium-doped yttrium aluminum garnet (Nd:YAG) laser (New Wave UP 213 AF) operating at a wavelength of 213 nm in a He+Ar atmosphere.Laser ablation was performed at 1.5 J/cm 2 at a pulse rate of 10 Hz.Due to the small grain size, and to avoid mineral inclusions, a predetermined line raster of 120 µm with a 15-µm beam diameter  (Van Achterbergh et al. 2001) was used for data processing with National Institute of Standards and Technology (NIST) SRM 610 as the external standard and 42 Ca as the internal standard using values for CaO from the electron microprobe analysis for the minerals.A linear adjustment was selected as the interpolation method for drift control and quantification.The relative errors (1σ) were in the range of 5-10% for most of the trace elements.Detection limits are usually between 0.1 and 2 ppm, except for Mg and Zn, with values up to 20 ppm.Whenever an element was analyzed with microprobe and LA-ICP-MS (e.g., Hf, Ta, Pb, Mg, and some REE), the more precise data from LA-ICP-MS were preferred.EGM structural formulae were calculated according to Johnsen & Grice (1999) and Johnsen et al. (2001) based on 29 cations (sum of Si, Al, Zr, Ti, Hf, Nb, W and Ta) and constraining all Fe and Mn to be divalent.
The bulk-rock compositions of major and some trace elements were analyzed in a fused glass disc and pressed powder pellets using a Philips PW 2400 X-ray fluorescence spectrometer at the Geoanalítica-USP Facility of the USP using standard methods as described by Mori et al. (1999).Weight loss on ignition (LOI) was evaluated using standard gravimetric techniques.Trace elements and REE were analyzed in a Perkin Elmer Elan 6100 DRC ICP-MS at the Geoanalítica-USP Facility of the USP following the procedures delineated in Navarro et al. (2008).
The EGM-bearing nepheline microsyenite dyke (peralkaline index = 1.08) has a sodic character (Na 2 O/K 2 O = 2.5), with the lowest K 2 O content and highest Na 2 O content

EGM compositional variation and molecular components
According to the calculated structural formulae (Table 2), Si contents are sufficient to fill 25 tetrahedral positions, whereas the Zr (plus minor content of Hf) and Ca amounts fill the Z and M( 1) positions, respectively.In the X site, Cl prevails over F.
The most extensive variation in the EGM composition from Monte de Trigo involves Fe and Mn at the M(2) site and Si and Nb at the M(3,4) site (Fig. 4).Both variations correspond to the solid solution between eudialyte and kentbrooksite components, recognized as the major substitution of the EGMs worldwide by Johnsen et al. (1998) and Johnsen & Grice (1999), although excluding the F and Cl contents.In fact, some analyses present Mn as dominant cation in M(2) site with the X site occupied by Cl, which probably is a mineral still not approved by the International Mineralogical Association.A slight excess of Zr from the Z site could be located with Nb and Si in the M(3,4) site, together with minor to trace contents of Ti, Ta, and W.
Other significant major element variations in the EGMs from Monte de Trigo occur at the N(1-5) site.Most of the 15 positions of the N(1-5) sites are filled with Na, which varies from 11.02 to 12.28 apfu.The remaining positions may include other elements, such as REE, Y, Ca, and Mn, probably at the N(4) site (Rastsvetaeva 2007).The sum of the Fe and Mn contents (3.259 to 3.613 apfu) exceeds the three available M(2) positions.This excess suggests that small amounts of the georgbarsanovite component are present in the solid solution, in which Mn is located at the N(4) site (Khomyakov et al. 2005, Rastsvetaeva 2007).Similarly, the Ca content (6.129 to 7.098 apfu) exceeds the required amount to fulfill the six positions of the M( 1) site (Fig. 4), with the surplus Ca most likely located at the N(4) site, composing the feklichevite member (Pekov et al. 2001).
The N(4) site also includes REE (0.381 to 0.673 apfu) and Y (0.106 to 0.244 apfu), which indicates the presence of a zirsilite-(Ce) component (Khomyakov et al. 2003).Light REE enrichment is observed in the C1 chondrite-normalized REE diagram (Fig. 5A), with La N /Sm N ratios ranging from 9.1 to 13.7 and Gd N /Lu N ratios ranging from 0.71 to 1.95.A clearly marked negative Eu/Eu* anomaly is present, with values varying from 0.17 to 0.23.
Thus, the EGMs from Monte de Trigo are best represented by a complex solid solution involving eudialyte, feklichevite, zirsilite-(Ce), georgbarsanovite, and kentbrooksite components (Fig. 4).Part of the Mn content of the EGMs may be responsible for the formation of minor quantities of the manganoeudialyte member (Nomura et al. 2010) rather than kentbrooksite to compensate for the Nb deficiency, assuming all of the Nb-bearing components are present.The EGMs from the dyke border are more enriched in the eudialyte component, whereas the EGMs from the dyke center are more enriched in the feklichevite, zirsilite-(Ce) and kentbrooksite components (Fig. 4).
The trace element contents normalized according to the bulk composition of the host dyke are shown in Figure 5B.The EGMs from the dyke border have significantly higher contents of Sr, Th, Zn, and Mg and significantly lower contents of Hf, Ta, and heavy REE than those from the dyke center.The low K, Rb, and Ba EGM/bulk-rock ratios suggest an incompatible behavior, i.e., they are not incorporated in the EGM structure.Thus, the andrianovite component, an EGM similar to kentbrooksite and zirsilite-(Ce), but with K located at the N(4) site (Khomyakov et al. 2008), is irrelevant in the composition of the EGMs from Monte de Trigo, despite the high K and Rb contents of the bulk rock (Table 1).This finding suggests that K and Rb prefer other minerals, such as alkali feldspar.The contents of Mg, Zn, Ga, Sr, Th, and U in EGMs are slightly higher than the ones in the host rock, which suggests that they are neutral to compatible in the EGM structure, whereas Pb is highly compatible, with EGM/bulkrock ratios near 100, probably because the ionic radius of Pb (Shannon 1976) is similar to the size of the N( 4) site (Johnsen & Grice 1999).
The Na-poor and Ca-rich composition of the analyzed EGMs may be a consequence of transitional agpaitic magma composition.The accessory mineral assemblage from Monte de Trigo, as well as those from Langesundsfjord and North Qôroq, includes complex Na-Ca-Zr-F silicate minerals, such as wöhlerite and låvenite (Coulson 1997, Enrich et al. 2009, Andersen et al. 2010), which is typical of transitional agpaitic rocks (Sørensen 1997, Andersen et al. 2010, Marks et al. 2011).This assemblage suggests that the miaskitic to agpaitic transition in the Monte de Trigo alkaline suite follows the increasing fluorine trend of Andersen et al. (2010).The EGM crystallization along this fluorine trend, according to these authors, can occur if chlorine or HCl activity is sufficiently high.This seems to be the case for the EGMs from Monte de Trigo, as evidenced by their relatively high Cl content.
Most of the trace element contents of the EGMs from Monte de Trigo vary within the ranges of EGMs from other localities (Wu et al. 2010, Schilling et al. 2011b), especially the HFSEs and REE.Major differences of the analyzed EGMs include low-Sr and Ba contents and a negative Eu/ Eu* ratio (Fig. 6).These differences are also found in the bulk composition of the EGM-bearing nepheline microsyenite dyke (Table 1), which suggests that EGMs crystallized in an environment where Eu, Sr, and Ba were already depleted.This depletion could be assigned to previous extensive fractionation of plagioclase and/or alkali feldspar (e.g., Blundy & Wood 1991, Bédard 2006, Brotzu et al. 2007, Henderson & Pierozynski 2012, Carvalho & Janasi 2012) from the parental melt, similar to the interpretation by Schilling et al. (2011b) regarding the same features observed in the EGMs from Ilímaussaq and Mont Saint-Hilaire.
Extensive plagioclase fractionation is usually related to alkaline massifs with a basanite/alkali basalt parental melt, whereas it is absent for nephelinite parental magma (Marks et al. 2011, Schilling et al. 2011a, 2011b).Thus, EGMs that are crystallized from agpaitic residual liquids derived from basanite/alkali basalt and nephelinite parental magmas should present Sr-poor and Sr-rich trends, respectively.The compositional trends for EGMs from several occurrences are shown in the FeO-SrO-REE 2 O 3 diagram (Fig 7).The FeO-rich corner characterizes the composition of early magmatic EGMs, as indicated by Schilling et al. (2011b).Both REE-and Sr-rich end-members of EGMs are related to late-magmatic and/or post-magmatic hydrothermal fluids, as are, for example, the final compositions of the EGM trends from the Pilanesberg and Saint-Hilaire complexes (Olivio & Williams-Jones 1999, Mitchell & Liferovich 2006, Grice & Gault 2006, Schilling et al. 2011b).The Sr-poor EGM trends found at Saint-Hilaire and Ilímaussaq reflect their basanite/ alkali basalt parental magma (Larsen & Sørensen 1987, Bailey et al. 2001, Schilling et al. 2011a).Conversely, the Sr-rich EGM trends found at Poços de Caldas, Gardiner, Khibina and Lovozero could be linked to their nephelinite parental magma (Nielsen 1980, Kramm & Kogarko 1994, Sørensen 1997, Ulbrich et al. 2005, Kogarko et al. 2010).The EGMs from Monte de Trigo have relative Fe and REE enrichment that, together with the EGM texture, agrees with its magmatic origin.The Sr-poor EGM trends found in Monte de Trigo reflect their basanite/alkali basalt parental magma.A mafic dyke with basanite composition has already been described in Monte de Trigo (Thompson et al. 1998, Enrich et al. 2009), and it probably represents the parental magma for this alkaline suite.Again, the EGMs from Monte de Trigo are plotted close to the north Qôroq and Langesundsfjord EGM data.Some additional considerations about the compositional differences between the EGMs from the dyke border in relation to those from the center also could be made.The EGMs from the dyke border have the highest Mg (up to 3,582 ppm) and Zn (up to 852 ppm) contents of the worldwide EGMs (up to 1,000 ppm and 200 ppm for Mg and Zn, respectively; Deer et al. 1986, Gualda & Vlach 1996, Johnsen et al. 1998, Johnsen & Grice 1999, Ridolfi et al. 2003, Bulakh & Petrov 2004, Marks et al. 2008, Wu et al. 2010, Schilling et al. 2011b).Such high Mg and Zn contents, associated to relatively high Fe and Sr in these EGMs, could be related to the incorporation and reaction of the melatheralite xenoliths, as these elements are more abundant in the melatheralite (Table 1).The local-scale reaction due to the diffusive exchange between the melatheralite and the agpaitic magma at dyke border are suggested by the presence of corroded xenoliths (Fig. 1C), large mafic aggregates (Fig. 3B) and metasomatic granofels (Fig. 3A), and by the increase in modal mafic mineral content, up to 25% in volume.In addition, låvenite from the dyke border has higher Mg and Zn contents than wöhlerite from the dyke center (Table 2).
Other difference in EGM composition is related to the lower heavy REE contents of the EGMs at the dyke border than at the dyke center (Fig. 5).This seems to be a consequence of the crystallization of låvenite rather than of wöhlerite.Our data show that låvenite preferentially incorporates the heavy REE, whereas wöhlerite incorporates both the light and heavy REE (Fig. 5).The låvenite concomitant crystallization leads to a concurrence in the heavy REE partitioning between EGMs and låvenite.

CONCLUSIONS
The studied EGMs from Monte de Trigo reveal characteristics of their crystallization environment, i.e., magma composition  Gualda & Vlach (1996), Johnsen & Gault (1997), Johnsen & Grice (1999), Marks et al. (2008), Schilling et al. (2009Schilling et al. ( , 2011b)), and Wu et al. (2010).Blue squares, the EGMs from the dyke border; open circles, the EGMs from the dyke center.and concomitant crystallizing phases.The EGMs are characterized by moderate variations in Fe and Mn at the M(2) site and Si and Nb at the M(3,4) site, which are recognized as the major substitution of EGMs worldwide.The relatively Na-poor, Ca-rich EGM compositions seem to be related to the Monte de Trigo transitional agpaitic rocks association, which contain wöhlerite and låvenite, and thus follow an agpaitic fluorine-rich trend.The low-Ba, Sr, and Eu contents are related to previous plagioclase fractionation during magmatic evolution and support the interpretation of a basanite parental magma for the suite.The incorporation and reaction of melatheralite xenoliths into the agpaitic magma could lead to higher Fe, Sr, Mg, and Zn contents in the EGMs found near the dyke border.Finally, the relative depletion of heavy REE in the EGMs from the dyke border seems to be related to the presence of låvenite, which strongly partitions heavy REE.(1996), Gualda & Vlach (1996), Coulson (1997), Johnsen & Gault (1997), Olivio & Williams-Jones (1999), Johnsen & Grice (1999), Mitchell & Liferovich (2006), Marks et al. (2008), Schilling et al. (2009Schilling et al. ( , 2011b)), and Wu et al. (2010).Blue squares, the EGMs from the dyke border; open circles, the EGMs from the dyke center.

Figure 1 .
Figure 1.(A) Geological map of the Serra do Mar alkaline province and surrounding area in southeastern Brazil, modified from Almeida (1983).Inset: simplified geological map of Monte de Trigo Island (after Enrich et al. 2009); (B) Field relations of the EGM-bearing nepheline microsyenite dyke of the Monte de Trigo suite; (C) Detail of (B) showing corroded melatheralite xenoliths at the dyke border.

Figure 5 .
Figure 5. (A) C1 chondrite-normalized Ree diagram for EGMs.Chondrite values are from McDonough & Sun (1995); (B) Bulk-rock-normalized multi-element diagram of the EGMs from Monte de Trigo.The bulk-rock values are from the EGM-bearing nepheline microsyenite dyke center sample (Table1).Blue squares, the EGMs from the dyke border; open circles, the EGMs from the dyke center.The bulk-rock data from the dyke center (blue diamonds), låvenite (open triangle) and wöhlerite (black cross) are also plotted for comparison.

Table 2 .
Compositions of the EGMs from the Monte de Trigo alkaline suite.Representative analysis of låvenite (Låv) and wöhlerite (Wöh) were also included Representative analyses of wöhlerite and låvenite are presented in Table2.These phases have lower SiO 2 and higher ZrO 2 contents than EGMs.Wöhlerite has higher Nb 2 O 5 and CaO contents than låvenite does, whereas the latter has high MnO, FeO, and Na 2 O contents.Both minerals have almost no LILE, but they are enriched to a variable degree in REE.Wöhlerite contains greater amounts of light REE than låvenite does.The La N /Yb N ratios of wöhlerite and låvenite are 2.05 and 0.12, respectively, which are significantly lower than the values for EGMs from the dyke center (4.8-8.0),EGMs from the dyke border (12.5-25.0)and the bulk-rock composition (9.9).The Eu/Eu* anomaly is somewhat similar to that observed in the EGMs and bulk-rock analysis.Låvenite also has high contents of Mg (3,819 ppm) and Zn (329 ppm).
a The EGM analyses 1-11 and the wöhlerite are from the dyke center (sample mtr01b); the EGM analyses 12-15 and the låvenite are from the dyke border (sample mtr01a); b major element contents (wt.%) were determined by electron probe micro-analyzer (EPMA); c estimated considering Cl+F+OH=5 a.p.f.u; d includes H 2 O and trace elements; e structural formulae calculated on the basis of 29 cations (Si+Al+Zr+Ti+Hf+Nb+W+Ta), with all Fe and Mn constrained to be divalent; f trace element (ppm) as well as HfO 2 , Ta 2 O 5 and PbO (wt%) were determined by inductively coupled plasma mass spectrometry (ICP-MS), except from analyses 4, 8, 9, 10, 11, and 14, that were only EPMA data are available.

Table 1 )
. Blue squares, the EGMs from the dyke border; open circles, the EGMs from the dyke center.The bulk-rock data from the dyke center (blue diamonds), låvenite (open triangle) and wöhlerite (black cross) are also plotted for comparison.