- © 2016 The Mineralogical Society
Geochemical characteristics and rare-earth element (REE)-bearing minerals of calc-alkaline granites in southern Myanmar were investigated to identify the minerals controlling fractionation between light and heavy REE (LREE and HREE) during magmatic differentiation and weathering. The granites were classified on the basis of the mineral assemblages into two contrasting groups: allanite-(Ce)- and/or titanite-bearing granites; and more HREE-enriched granites characterized by hydrothermal minerals including synchysite-(Y), parisite-(Ce), bastnäsite-(Ce), xenotime-(Y), monazite-(Ce), Y-Ca silicate, waimirite-(Y) and fluorite. This suggests that allanite-(Ce) and titanite are not stable in differentiated magma and HREE are eventually preferentially incorporated into the hydrothermal minerals. The occurrence of the REE-bearing minerals is constrained by the degree of magmatic differentiation and the boundary of two contrasting granite groups is indicated by SiO2 contents of ∼74 wt.% or Rb/Sr ratios of ∼3–8. Fractionation between LREE and HREE during weathering of the granites is influenced by weathering resistance of the REE-bearing minerals, i.e. allanite-(Ce), titanite, the REE fluorocarbonates and waimirite-(Y) are probably more susceptible to weathering, whereas zircon, monazite-(Ce) and xenotime-(Y) are resistant to weathering. Ion-exchangeable REE in weathered granites tend to be depleted in HREE relative to the whole-rock compositions, suggesting that HREE are more strongly adsorbed on weathering products or that HREE remain in residual minerals.
Calc-alkaline granites in convergent plate boundaries generally contain fewer rare-earth elements (REE: yttrium and lanthanoids) compared to intra-plate alkaline rocks and carbonatites associated with REE mineralization (Taylor and Pollard, 1996; Wall, 2014; Linnen et al., 2014). As calc-alkaline granites undergo not only magmatic differentiation but also weathering processes resulting in REE enrichment, economic REE deposits associated with granites are almost exclusively confined to placer (Kanazawa and Kamitani, 2006) and ion-adsorption types (Wu et al., 1990; Bao and Zhao, 2008; Sanematsu and Watanabe, 2016). The economic significance of REE deposits is dependent not only on mineral resources and ore grades but also on the relative abundances of the more precious heavy REE (HREE: Y and Gd to Lu) in the ores, because HREE are less abundant in the crust except Y compared to LREE (LREE: La to Eu) (Rudnick and Gao, 2003). Ion-adsorption type deposits consist of weathered granites and include ion-exchangeable REE, which are adsorbed on the surface of clays or other weathering products (e.g. Yang et al., 1987; Wu et al., 1990; Bao and Zhao, 2008; Sanematsu and Watanabe, 2016). The ion-adsorption type deposits are moderately or strongly enriched in HREE, and have attracted attention recently (Bao and Zhao, 2008; Wall, 2014; Sanematsu and Watanabe, 2016).
In intermediate to felsic rocks, significant amounts of REE are contained in accessary minerals such as apatite, zircon, titanite, allanite-(Ce) and monazite-(Ce) (Bea, 1996). The partition coefficients of the mineral/melt are different between LREE and HREE because of the progressive decrease in ionic radii from La to Lu through Y (Henderson, 1984; Green, 1994; Tiepolo et al., 2002; Hoskin and Schaltegger, 2003; Gieré and Sorensen, 2004; Prowatke and Klemme, 2005; Rubatto and Hermann, 2007; Luo and Ayers, 2009). These studies suggest that allanite-(Ce), monazite-(Ce) and apatite incorporate more LREE than HREE and that titanite and zircon tend to incorporate more HREE. The occurrences and relative abundances of these REE-bearing minerals in granitic rocks are influenced strongly by fractionation between LREE and HREE (during magmatic differentiation (Miller and Mittlefehldt, 1982; Sawka et al., 1984; Wark and Miller, 1993). Additionally, a variety of REE minerals including REE fluorocarbonates occur in granitic rocks during a magmatic-hydrothermal alteration stage (Förster, 2000; Agangi et al., 2010; Sheard et al., 2012; Hoshino et al., 2013). Previous studies have noted that REE fluorocarbonates (bastnäsite-(Ce), parasite-(Ce) and synchysite-(Y)) in calc-alkaline granites are probably the REE source for ion-adsorption ores (Huang et al., 1989; Bao and Zhao, 2008; Ishihara et al., 2008; Sanematsu et al., 2013). These ores, typically weathered calc-alkaline granites, include significant amounts of ion-exchangeable REE because REE fluorocarbonates dissolved in acidic soil water during weathering (Huang et al., 1989; Sanematsu et al., 2013; Sanematsu and Kon, 2013). Other minerals could also be a source for ion-exchangeable REE because the underlying parent granites in these deposits do not necessarily include many REE fluorocarbonate minerals (Sanematsu and Watanabe, 2016 and references therein). The geochemistry of granites associated with ion-adsorption deposits has been discussed in some previous studies (Huang et al., 1989; Bao and Zhao, 2008; Sanematsu and Watanabe, 2016). However the geochemical and mineralogical characteristics of the LREE-rich and HREE-rich granites suitable for ion-adsorption ores are not well understood outside of China.
In this study, we discuss fractionation between LREE and HREE during magmatic differentiation and weathering processes of granites in order to identify minerals controlling the REE fractionation and minerals which could be a source for ion-adsorption REE in weathered granites. Calc-alkaline granites in southern Myanmar were selected for this study (Fig. 1), because the gradual fractionation of REE and enrichment of HREE by magmatic differentiation has been recognized in this area. The results of this study will help to understand LREE-rich and HREE-rich ion-adsorption type deposits developed on calc-alkaline granites.
The study area (Fig. 1) is the southern part of the north–south trending granite belt of Myanmar (Khin Zaw, 1990; Mitchell et al., 2012) in the western Sibumasu Terrane, which collided with the Sukhothai Terrane in the west of the Indochina in the Late Triassic (Sone and Metcalfe, 2008; Metcalfe, 2013; Khin Zaw et al., 2014). The N–S trending granite belt is attributed to be associated with plate subduction (Mitchell et al., 2012). Reported ages of granites range from 94 to 45 Ma by whole-rock Rb-Sr geochronology (Cobbing et al., 1992). In the study area, the granites are hosted in the Carboniferous Mergui Group (Myanmar Geosciences Society, 2014), and are associated locally with Sn-W mineralization.
Thirty one granite samples were collected from outcrops in the Tanintharyi Region of southern Myanmar (Fig. 1). The granite plutons are composed of magnetite-series and ilmenite-series granites on the basis of their magnetic susceptibility and the occurrence of magnetite (Ishihara, 1977). The biotite granites consist of quartz, plagioclase, K-feldspar, biotite and/or hornblende (Fig. 2a–c) with smaller amounts of accessary minerals such as titanite, allanite-(Ce), apatite, zircon, epidote, ilmenite, magnetite and/or pyrite. In contrast with the (hornblende-bearing) biotite granite, leucocratic biotite granite, muscovite-bearing biotite granite and garnet-bearing muscovite granite contain less allanite-(Ce) and rarely titanite. These granites consist of quartz, plagioclase, albite, K-feldspar, biotite and/or muscovite (Fig. 2d–f) with lesser amounts of apatite, zircon, ilmenite and pyrite. Sericitization of plagioclase and chloritization of biotite are found in most of the granites. Dissemination of fine-grained calcite and fluorite can occur in some of the muscovite-bearing granites.
Sixty three samples of weathered granite, were collected from different locations and different depths of weathering profiles on the granites. The sample locations are listed in Supplementary Table 1 which has been deposited with the Principal Editor of Mineralogical Magazine and is available from www.minersoc.org/pages/e_journals/dep_mat_mm.html. Six weathering profiles termed 26-2, 26-7, 28-5, 02-1, 02-3 and 27-8 were investigated (Fig. 3) these being underlain by granite samples 26-1G, 26-8G, 28-4G, 02-2G, 02-4G and 27-10G, respectively. The weathering profiles 26-2, 26-7 and 28-5 exhibit a brownish colour (Fig. 3a–c), whereas the profiles 02-1, 02-3 and 27-8 are pale khaki to whitish grey in colour (Fig. 3d–f), although the oxidized surface is brown. These whitish weathering profiles suggest the presence of leucogranite (Fig. 2d–f) and/or hydrothermal alteration prior to weathering. These six pairs of weathering profiles and granite protoliths were selected to determine fractionation between LREE and HREE during weathering and adsorption processes.
Scanning electron microscopy – energy dispersive X-ray spectrometry
In order to identify accessary minerals including REE-bearing minerals, we employed a field emission (FE) scanning electron microscopy (SEM) with energy dispersive X-ray spectrometry (EDS) after observation by plane-polarized light microscopy. Imaging and semi-quantitative analyses of REE-bearing minerals were performed using a Hitachi SU-70 FE-SEM-EDS system at the Central Science Laboratory, University of Tasmania. Imaging was carried out on carbon-coated polished thin sections at an accelerating voltage of 15 kV and beam current of 33 nA.
Electron microprobe analysis
The compositions of allanite-(Ce), titanite, apatite and a hydrothermal mineral were determined using a JEOL JXA-8530M electron microprobe analyser (EPMA) equipped with five wavelength-dispersive spectrometers (WDS) at AIST, because they could not be identified by optical microscopy and SEM-EDS. Elements detected were measured quantitatively using an accelerating voltage of 15 kV (for apatite) or 20 kV (for the other REE-bearing minerals), beam current of 20 nA, count times of 10–20 s and beam diameter of 1–10 µm. The following peak positions were selected: the Kα series of X-ray spectra was used for Si, Ti, Al, Ca, Mn and Fe; the L series for REE. Measurements of the Lα lines for La, Ce, Nd, Er and Yb and the Lβ line for Pr, Sm, Gd and Dy required no peak-overlap corrections. Synthetic Ca-Al silicate glasses containing each REE, which are available from P and H Development Ltd., were used for REE standards by following Hoshino et al. (2006). The standards for REE-bearing silicate minerals were Ca2Na(Mg,Fe)4TiSi6Al2O22(OH2) (kaersutite, Kα) for SiO2, Al2O3, CaO and MgO of allanite-(Ce), CaSiO3 (wollastonite, Kα) for SiO2 and CaO of titanite, (Mn,Ca)SiO3 (bustamite, Kα), TiO2 (rutile, Kα), Fe2O3 (Hematite, Kα), CaF2 (fluorite, Kα) and ThO2 (Th glass, Kα). The standards for apatite are Ca5(PO4)3F (apatite, Kα) for CaO and P2O5, CaF2 (fluorite) for F and CeP5O14 (Lα) for Ce2O3. All the data were corrected with a ZAF full matrix correction program.
For Raman spectrometry, excitation was accomplished using a wavelength of 532 nm from a solid-state ion laser with an output power of 50 mW. The Raman scattered light was detected by a laser Raman spectrometer (Micro Raman spectroscopy system Ram532) at AIST, Japan. The laser light was standardized with silicon using a band maximum at 520±5 cm–1.
Sample preparation and X-ray fluorescence (XRF) analysis were conducted in AIST, Japan. Granite and weathered granite samples of over 200 g were dried at room temperatures and crushed by an Fe-Mn alloy jaw crusher. Individual ∼30 g fractions of the crushed samples were pulverized using a rod mill (CMT TI-100) with alumina containers and rods for 3 min. Glass beads for XRF analysis were prepared by mixing 0.5 g of the sample powder with 5.0 g of lithium tetraborate flux (Li2B4O7, Spectromelt A10). The sample mixture was fused at 1250°C in a Pt crucible using a high-frequency fusion instrument (HERZOG HAG-M-HF), and cooled on a Pt disk. Major elements in the glasses were analysed by XRF Spectrometry (Rigaku ZSX Primus III+) at a voltage of 50 kV and a current of 50 mA. Calibration lines of the major elements were established using the GSJ geochemical reference samples of igneous rock series (Imai et al., 1995). Pressed powder pellets were prepared and analysed by XRF for some samples, for which results were under detection limits or over a calibration curve used for following trace-element analysis.
Trace elements and fluorine were analysed by Activation Laboratories Ltd., Canada. The powdered samples were mixed with a flux of lithium metaborate and tetraborate and fused in an induction furnace. The fused sample was poured into a 5% NHO3 solution with an internal standard and they mixed until completely dissolved. Element concentrations of the sample solution were analysed by inductively coupled plasma mass spectrometry (ICP-MS). Another fraction of the granite powders was also fused and dissolved with dilute HNO3 and used to determine the F– activity in the solution using a Mandel Scientific ion selective electrode.
Extraction experiments and analysis of ion-exchangeable REE
The REE extraction experiments and solution ICP-MS analysis were undertaken at the Geological Survey of Japan and basically follow the single-step extraction procedure of Sanematsu and Kon (2013). A 1 g powder of weathered granite was soaked in 40 ml volume of 0.5 mol/l ammonium sulfate [(NH4)2SO4] solution of pH = ∼5.7 in a 50 ml centrifuge tube. The tubes were mechanically shaken at room temperature for 24 h. The extract is separated from the solid samples by centrifugation. The supernatant solution was filtered by using a membrane filter (φ = 0.22 μm), and the filtered solution was acidified using HNO3 and kept in a polypropylene container. Ultra-pure water and an indium standard solution (Wako) as an internal standard were added to the acidified sample solution, and it was prepared to 1% HNO3 equivalent before analysis.
Element concentrations were determined by Agilent Technologies 7500cx ICP-MS. Flow rates of carrier gas and the ion-lens setting of the ICP-MS were optimized to maximize the signal intensity of Ce and to minimize the oxide production rate (140Ce16O/140Ce < 0.01). We monitored 7Li, 23Na, 24Mg, 27Al, 29Si, 31P, 39K, 43Ca, 45Sc, 47Ti, 51V, 53Cr, 55Mn, 57Fe, 59Co, 60Ni, 63Cu, 66Zn, 69Ga, 72Ge, 75As, 85Rb, 88Sr, 89Y, 90Zr, 93Nb, 95Mo, 107Ag, 111Cd, 115In (internal standard), 133Cs, 137Ba, 139La, 140Ce, 141Pr, 146Nd, 147Sm, 153Eu, 157Gd, 159Tb, 163Dy, 165Ho, 166Er, 169Tm, 172Yb, 175Lu, 181Ta, 182W, 205Tl, 208Pb, 209Bi, 232Th and 238U. Calibration lines were made by multi-element standard solutions of XSTC-1, -8 and -15 (SPEX).
Identification of REE-bearing minerals and classification of the granites
The results of the optical microscopy and SEM observation of typical granite samples are listed in Table 1. On the basis of the occurrences of REE-bearing minerals, the granites were classified into two groups.
One of the granite groups commonly contains both or either of coarse-grained (0.1–1 mm) ‘allanite-(Ce)’ and ‘titanite’ (Fig. 4a–c). This group is hereafter named ‘the AT Granite’ (Table 1). Euhedral-to-subhedral habits of allanite-(Ce) and titanite without hydrothermal alteration indicate that these minerals crystallized from magma in common with the rock-forming minerals. The AT Granite consists of hornblende-bearing biotite granodiorite and biotite granite. In the samples of 28-4G and 28-6G, allanite-(Ce) is common, although titanite was found rarely. In contrast, sample 26-8G contains abundant titanite, with allanite-(Ce) rarely occuring.
Results of the EPMA quantitative analysis of allanite-(Ce) and titanite are summarized in Table 2. The compositions indicate that allanite-(Ce) is distinctly rich in LREE and poor in HREE. In contrast, some titanite crystals are typically rich in HREE (Table 2). No REE were detected by EPMA in titanite in the hornblende-biotite granodiorites (Fig. 4a; 16-9G and 26-1G), whereas titanite in biotite granites (Fig. 4b) is rich in HREE and relatively poor in LREE (Table 2). Apatite commonly occurs in the AT Granite and the compositions indicate that it is fluoroapatite (Table 3). Ce2O3 is detected in some apatite crystals and it is significantly lower than the Ce2O3 content of allanite-(Ce).
The other granite group contains much less allanite-(Ce), which occurs as fine-grained anhedral crystals and rarely contains titanite (Fig. 4d,e). As this granite group commonly contains ῾hydrothermal minerals’ including REE fluorocarbonates such as synchysite-(Y) and parisite-(Ce), xenotime-(Y), monazite-(Ce), fluorite, Y-Ca silicate and Y-Ti oxide (Fig. 4f–j), it is hereafter named ‘the HM Granite’ (Table 1). The HM Granite is leucocratic and more differentiated than the AT Granite, and consist of biotite granite, muscovite-bearing biotite granite and garnet-bearing muscovite granite. Because the hydrothermal minerals exhibit anhedral-to-subhedral habits and were rarely found in the AT Granite, they are interpreted as secondary minerals crystallizing from hydrothermal fluids in a magmatic-hydrothermal alteration stage of granite. Synchysite-(Y), parisite-(Ce) and bastnäsite-(Ce) were distinguished on the basis of composition and Raman spectra. Compositions of the hydrothermal minerals determined by EPMA suggest that a Y-Ca silicate mineral is probably hingganite-(Y) in sample 02-4G (Table 2), although Be was not determined. Small amounts of Y are detected in Th silicate by EDS (ThSiO4: probably thorite). Some fluorite crystals are relatively rich in Y (up to 3.5 wt.%). Fine-grained waimirite-(Y) (YF3) crystals were commonly found in the most differentiated granite of sample 27-10G (Fig. 4l). These REE-bearing minerals commonly occur with or within fluorite and chlorite (Fig. 4), which are alteration products of biotite, in the HM Granite. Occurrences of apatite and zircon are less common than those of the AT Granite as estimated by the whole-rock geochemical data (Fig. 6e,f,k,l).
Compositions of granites
Whole-rock compositions of the granites are listed in Table 3. These data indicate that SiO2 and K contents range from 68.9 to 76.6 wt.% and range from 3.7 to 5.2 wt.%, respectively. These granites are classified into high-K calc-alkaline series on K2O vs. SiO2 diagrams (Fig. 5). The whole-rock SiO2 contents and Rb/Sr ratios (Fig. 6) indicate that the HM Granite is more differentiated than the AT Granite. The geochemical boundary of the two granite groups is the SiO2 content of 74 wt.% or Rb/Sr ratio, approximately between 3 and 8 (Table 4). The most differentiated garnet-bearing muscovite granite samples show remarkably high Rb/Sr ratios of 312 and 330, respectively, compared to the other granites (Table 4). Most of the granites typically show I-type features (Chappell and White, 1974) on the basis of the alumina saturation index [ASI = molar Al2O3/(CaO+Na2O+K2O) ratio] being <1.1 and absence of garnet (Table 4), except for the two garnet-bearing muscovite granites (27-10G and 27-28G) that show S-type features.
Whole-rock LREE and HREE contents differ between the AT and HM Granites (Fig. 6). The LREE contents of the AT and HM Granites range from 87 to 224 ppm and from 37 to 115 ppm, respectively (Table 4, Fig. 6a,g). The HREE contents of the AT and HM Granites range from 31 to 62 ppm and from 52 to 267 ppm, respectively (Table 4, Fig. 6b,h). The REE contents are not significantly different between the AT and HM Granites (Fig. 6c,i). The fractionation between LREE and HREE is represented by LaN/YbN ratios (Fig. 4c), wherein the subscript N represents normalization to CI chondrite (Sun and McDonough, 1989). The AT Granite has LaN/YbN ratios from 3.7 to 15.9. The HM Granite shows significantly lower LaN/YbN ratios from 0.15 to 4.6 (Table 4, Fig. 6d, j). The P2O5 contents of the AT Granite and HM Granite range from 0.04 to 0.10 wt.% and are <0.04 wt.%, respectively (Table 4, Fig. 6e,k). The Zr contents of the AT Granite and HM Granite range from 102 to 192 ppm and from 44 to 110 ppm, respectively (Table 4, Fig. 6f,l).
Fractionation of REE can be recognized by the chondrite-normalized REE distribution patterns of the granites (Fig. 7). The AT Granite is enriched in LREE, while in contrast, the HM Granite is more depleted in LREE and enriched in HREE relative to the AT Granite, showing a nearly-flat REE pattern with a distinct negative Eu anomaly. The two most differentiated granite samples 27-10G and 27-28G display strong HREE enrichment. In all the granite samples, Ce anomalies [Ce/Ce* = CeN/(LaN·PrN)1/2] are not significant (Fig. 7), showing a narrow range from 0.95 to 1.05. Strong negative Eu anomalies [Eu/Eu* = EuN/(SmN·GdN)1/2] are recognized in all the granites (Fig. 7; Table 4).
Compositions of weathered granites and ion-exchangeable REE
Whole-rock compositions of the weathered granites are given in Supplementary Table 1 (available from www.minersoc.org/pages/e_journals/dep_mat_mm.html). The REE contents of the weathered granites range widely from 30 to 364 ppm. The LaN/YbN ratios vary from 0.39 to 18. The results of REE extraction experiments indicate concentrations of ion-exchangeable REE from the weathered granites (Supplementary Table 2, available from www.minersoc.org/pages/e_journals/dep_mat_mm.html), which range widely from 2.3 to 175 ppm. The LaN/YbN ratios of the ion-exchangeable REE vary from 0.79 to 50. The percentages of ion-exchangeable REE concentrations relative to the whole-rock REE contents of weathered granites range widely from 1.5 to 59% (Supplementary Table 2).
REE fractionation during magmatic differentiation
Fractionation between LREE and HREE in granitic magma is generally constrained by crystallization of REE-bearing minerals in a magmatic differentiation process (Wark and Miller, 1993; Bea, 1996; Hoskin et al., 2000). In our study, the AT and HM Granites contain different minerals as a result of magmatic differentiation, reflecting LREE-rich and HREE-rich whole-rock compositions, respectively (Figs 5 and 6). The LREE-rich AT Granite, which has the higher LaN/YbN ratios (> 8), is characterized by lower SiO2 contents of 67–74 wt.% (Fig. 6d) and lower Rb/Sr ratios (< 3 (Fig. 6j). In contrast, the HREE-rich HM Granite has lower LaN/YbN ratios (< 4) and are characterized by higher SiO2 contents of 74–78 wt.% (Fig. 6d) and higher Rb/Sr ratios (>8) (Fig. 6j). A similar enrichment of HREE during magmatic differentiation has been commonly recognized in other granites, for example, in the Lachlan Fold Belt of southeastern Australia (Chappell, 1999), and in southwestern Japan (Ishihara and Murakami, 2006). These granites are significantly enriched in HREE in granites with over 74–75 wt.% SiO2 and this is consistent with the HREE enrichment at ∼74 wt.% SiO2 in our study results.
The AT Granite contains allanite-(Ce) and/or titanite in addition to apatite and zircon. The LREE and HREE contents are controlled by relative abundances of these REE-bearing minerals. Allanite-(Ce) is generally rich in LREE relative to HREE (Gieré and Sorensen, 2004). In our study, all the analysed allanite-(Ce) crystals are significantly rich in LREE and it is the dominant LREE-bearing mineral in the AT Granite. Titanite is moderately enriched in HREE, as its partition coefficients for HREE are relatively high (Green and Pearson, 1986; Prowatke and Klemme, 2005). The EPMA results indicate that titanite is one of the dominant HREE-bearing minerals in the AT Granite (Table 2) and its abundance varies between different granites. Titanite is significantly enriched in HREE in the differentiated AT Granite (samples 09-11G, 26-1G and 27-4G; Table 2). A back-scattered electron image of the HREE-rich titanite (Fig. 4b) is typically brighter than the HREE-poor titanite (Fig. 4a). The occurrence of HREE-rich titanite is reflected in the LaN/YbN ratios and slope of the REE distribution patterns of the granites. The titanite-rich and allanite-free granite 26-8G has a LaN/YbN ratio of 3.7, which is significantly lower than the ratios of the other AT Granite samples (Fig. 4c).
Titanite occurs in oxidized felsic magmas, whereas it becomes unstable in reduced magmas, where ilmenite becomes stable instead of titanite (Wones, 1989). An experimental study indicates that titanite is unstable in fluorine-rich melts because it reacts with fluorine to form fluorite (Price et al., 1999). These studies suggest that titanite, a HREE-bearing mineral, rarely crystallizes in the differentiated magma forming the HM Granite in our study (Fig. 6). Crystallization of zircon is not common due to a paucity of zirconium in the differentiated magma (Fig. 6f,i). These data suggest that a reduced and fluorine-bearing residual melt may become progressively enriched in HREE (Fig. 6) by fractional crystallization because crystallization of these magmatic HREE-bearing minerals are not common. In the HM Granite, eventually, most HREE are probably incorporated in other minerals excluding titanite and zircon (Fig. 4).
The HM Granite contains a variety of REE-bearing minerals including parisite-(Ce), synchysite-(Y), monazite-(Ce) and xenotime-(Y) (Fig. 4d–l) in contrast to the AT Granite (Fig. 4a–c). The occurrence of HREE minerals is reflected in the molar LREE/HREE ratios of the granite samples (Fig. 8). A gradual depletion of LREE during magmatic differentiation inevitably led to incorporation of HREE into newly crystallizing hydrothermal minerals in the HM Granite. The HREE minerals in the HM Granite crystallized dominantly in a magmatic-hydrothermal alteration stage, and this crystallization was constrained by whole-rock compositions of the granites as represented by SiO2 contents and Rb/Sr ratios (Fig. 6). The whole-rock granite geochemistry is suggestive of occurrences of hydrothermal REE minerals including HREE minerals.
The REE fluorocarbonates recognized in our study are mainly synchysite-(Y), parisite-(Ce) and bastnäsite-(Ce) coexisting with fine-grained xenotime-(Y), monazite-(Ce), Y-Ca silicate, Y-Ti oxide and fluorite (Fig. 4). Because Zr is incorporated almost exclusively in zircon, the low Zr contents in the HM Granite (Fig. 6f,l) reflect that HREE are dominantly contained in these HREE minerals (synchysite-(Y), xenotime-(Y), Y-Ca silicate and Y-Ti oxide) crystallizing during deuteric alteration of granite. The most strongly differentiated granite samples (27-10G and 27-28G) contain waimirite-(Y) (YF3), which is a recently found new mineral (Atencio et al., 2015).
Silicate melt tends generally to become depleted in P2O5 during magmatic differentiation and LREE are scavenged by phosphate minerals including apatite and/or monazite-(Ce) (Miller and Mittlefehldt, 1982; Montel, 1986; 1993; Rapp and Watson, 1986; Rapp et al., 1987; Chappell, 1999). In the granites investigated, phosphate contents progressively decrease by fractional crystallization and the HM Granite is significantly depleted in P2O5 (<0.04 wt.%; Fig. 6e,k). The phosphate decrease is predominantly due to crystallization of apatite because magmatic monazite-(Ce) is only rarely found. The microscopy observations and geochemical data indicate that the REE-bearing minerals of the AT Granite are dominated by allanite-(Ce) and apatite for LREE, and titanite and zircon for HREE. These minerals generally account for most of the REE content of monazite- and carbonate-free granitic rocks (Bea, 1996), which are typically characterized by I-type features.
REE fractionation during granite weathering
Mobility of REE during weathering of granitic rocks is influenced strongly by the degree of weathering and weathering resistances of REE-bearing minerals (e.g. Nesbitt, 1979; Braun et al., 1990, 1993; Aubert et al., 2001; Laveuf and Cornu, 2009). Particularly, REE fluorocarbonates in acidic soil water during weathering of granites, and which are rarely found in weathering profiles on the granites (Huang et al., 1989; Sanematsu et al., 2013; Sanematsu and Kon, 2013). Allanite-(Ce), titanite and apatite are also likely to break down during weathering of granitic rocks (Banfield and Eggleton, 1989; Braun and Pagel, 1994; Condie et al., 1995; Gieré and Sorensen, 2004; Price et al., 2005; Sanematsu and Watanabe, 2016). This is also indicated by placer minerals that contain lesser amounts of allanite-(Ce), titanite and apatite than weathering-resistant minerals including monazite-(Ce) and zircon (Orris and Grauch, 2002). Apatite is one of the ubiquitous REE-bearing minerals in igneous rocks and occurs more or less in all the granites studied. Abundances of allanite-(Ce) and titanite are more variable than apatite and zircon, and differ between the AT Granite and HM Granite (Table 1; Fig. 6).
Fractionation between LREE and HREE during weathering is strongly influenced by relative abundances of allanite-(Ce) and titanite in the parent granites (Fig. 9a–c). A columnar diagram of LREE and HREE contents of the weathering profile 26-2 and underlying allanite- and titanite-bearing granite (sample 26-1G) suggests that fractionation of REE is not significant and both allanite-(Ce), titanite and apatite break down during weathering (Fig. 9a). The weathering profile 26-7 and underlying titanite-rich granite (sample 26-8G) tends to contain less LREE and more HREE (Fig. 9b) than the profile 26-2 and parent granite 26-1G (Fig. 9a), respectively. In contrast, the weathering profile 28-5 on allanite-rich granite (sample 28-4G) shows strong depletion of LREE relative to HREE (Fig. 9c). The depletion of LREE results dominantly from allanite-(Ce) and partially from apatite. Less depletion of HREE is due to residual zircon because this granite rarely contains titanite. The fractionation of REE is also reflected in the LaN/YbN ratios in individual weathering profiles (Fig. 10). These data indicate that allanite-(Ce) and titanite strongly influenced the mobility and fractionation of REE during weathering of the AT Granite.
A variety of REE-bearing minerals occur in the HM Granite (Fig. 4, Table 1) and this leads to REE fractionation during weathering. REE fluorocarbonates dissolve in acidic soil water, whereas REE phosphates including monazite-(Ce) and xenotime-(Y) are resistant to weathering because of low dissolution rates of these phosphate minerals (Oelkers and Poitrasson, 2002; Oelkers et al., 2008). In the weathering profiles of 02-1 and 02-3, fractionation between LREE and HREE is not significant (Figs 9d,e, 10). The weathering profile overlying the most differentiated and HREE-enriched granite (sample 27-10G; Fig. 9f) is characterized by abundant HREE minerals such as xenotime-(Y) and waimirite-(Y) (Table 1). The strong depletion of HREE during weathering is attributed to decomposition of waimirite-(Y). Although its instability in weathering is not well understood, YF3 (waimirite-(Y)) is likely to dissolve in acidic solutions at ambient temperatures (Lokshin and Tareeva, 2007).
Weathered granites do not necessarily contain all the REE in mineral phases because REE are more or less adsorbed on the surface of clays or other weathering products as represented by ion-adsorption ores (Yang et al., 1987; Wu et al., 1990; Bao and Zhao, 2008; Sanematsu and Watanabe, 2016). The adsorbed REE concentrations of the weathered granite samples were evaluated by an ion-exchange reaction using an ammonium sulfate solution and results indicated that percentages of ion-exchangeable REE concentrations relative to whole-rock REE contents vary from 1.5 to 59% (Supplementary Table 2, available from www.minersoc.org/pages/e_journals/dep_mat_mm.html). The ion-exchangeable REE tend to be depleted in HREE relative to the whole-rock compositions (Fig. 10). The depletion of HREE in the ion-exchangeable fraction was recognized in another weathering profile in Thailand (Sanematsu et al., 2013). These results suggest that HREE are more strongly adsorbed on clays or other weathering products in weathered granites or that HREE are more contained in residual minerals including zircon and xenotime-(Y).
The calc-alkaline granites in southern Myanmar were classified into two contrasting groups: the AT Granite and HM Granite, on the basis of the mineral assemblages present. The AT Granite consists of (hornblende-bearing) biotite granites containing both or either of allanite-(Ce) and titanite in addition to apatite and zircon. The HM Granite, consists mainly of biotite granite, muscovite-biotite granite and garnet-bearing muscovite granite, rarely contains allanite-(Ce) and titanite, and is characterized by occurrences of hydrothermal minerals including synchysite-(Y), parisite-(Ce), bastnäsite-(Ce), xenotime-(Y), monazite-(Ce), Y-Ca silicate (probably hingganite-(Y)), Y-Ti oxide, fluorite and waimirite-(Y). The absence of allanite-(Ce) and titanite in the HM Granite may be attributed to their instability in differentiated magma.
The AT Granite is high in LREE (87–224 ppm) and low in HREE (37–115 ppm). In contrast, the more differentiated HM Granite is low in LREE (31–62 ppm) and high in HREE (52–267 ppm). The boundary between these two granite groups is indicated by whole-rock geochemistry of ∼74 wt.% SiO2 or Rb/Sr ratios of 3–8 because occurrences of allanite-(Ce) and titanite are constrained by the degree of magmatic differentiation.
Compositions of weathered granites and previous studies on weathering of REE-bearing minerals suggest that fractionation between LREE and HREE during granite weathering is influenced by the stabilities and relative abundances of the REE-bearing minerals. Allanite-(Ce), titanite, apatite, REE fluorocarbonates and waimirite-(Y) are likely to break down during weathering. The ion-exchangeable REE, which probably exist on clays or other weathering products in weathered granites, are depleted in HREE relative to the whole-rock compositions. This suggests that HREE are more strongly adsorbed on clays or other weathering products or that HREE are more contained in residual minerals.
The authors thank Dr. Karsten Goemann and Dr. Sandrin Feig at the Central Science Laboratory in the University of Tasmania for supporting image-capturing and analysis using a FE-SEM-EDS. They thank two anonymous reviewers for comments and Principal Editor Dr Roger Mitchell is thanked for copy editing the text. This study was financially supported by the JSPS Grant-in-Aid for Young Scientists (B) of No. 23760796, JSPS Postdoctoral Fellowships for Research Abroad and Japan Mining Promotive Foundation.
- Manuscript received 12 March 2015.
- Manuscript Accepted for publication 7 February 2016.