- © 2014 E. Schweizerbart'sche Verlagsbuchhandlung Science Publishers
Apatites from carbonatites, related alkaline silicate rocks, a carbonate-bearing melilititic dyke rock (bergalite), and a diatreme breccia (containing both carbonate and silicate fragments) of the Miocene Kaiserstuhl Volcanic Complex, SW Germany, are used to reconstruct the petrogenetic relationship among these rocks. Apatites from carbonatites reach higher Sr and Nb contents but are generally lower in Fe, Mn, Th, U, Si, S, Cl and Br compared to apatites from associated silicate rocks, whilst Na, REE and F contents are overlapping. Apatites from bergalite show a systematic and discontinuous core-rim zonation, with the core being compositionally similar to apatites from silicate rocks and the rim corresponding to carbonatitic apatites. These observations imply that the bergalite apatites nucleated in a silicate melt and continued to crystallize from an evolving CO2-enriched melt probably with carbonatitic affinity. Apatites from a diatreme breccia comprise three populations: (1) similar to the apatites from silicate rocks, (2) similar to the carbonatitic apatites, and (3) resembling apatite population (1) partially replaced by apatite (2). We infer that apatite (1) was derived from silicate-rock fragments and apatite (2) crystallized from a later intruding carbonatitic melt, which metasomatized the silicate-rock fragments and caused the replacement textures as observed in apatite population (3). We conclude that apatites from the Kaiserstuhl complex preserve important information on the petrogenetic relationship between carbonatitic and silicate melts. The carbonatitic melts at the Kaiserstuhl complex are probably the products of protracted fractionation of a CO2-rich nephelinitic melt.
From the approximately 500 carbonatite occurrences worldwide, more than 75% are associated with alkaline silicate rocks, ultramafic lamprophyres or kimberlites (e.g. Woolley & Kjarsgaard, 2008) and, in many cases, carbonatites postdate the silicate rocks (e.g., Keller et al., 1990; Bailey, 1993; Bell et al., 1999; Woolley, 2003; Woolley & Kjarsgaard, 2008; Halama et al., 2005). Despite nearly a century of research on carbonatites, their origin and relationships to associated silicate rocks is not completely understood yet (Bell et al., 1999; Gittins & Harmer, 2003). Experimental work showed that carbonatites can be generated by (1) liquid immiscibility of carbonatite and silicate melts (e.g., Koster van Groos & Wyllie, 1966; Freestone & Hamilton, 1980; Brooker & Kjarsgaard, 2011); (2) partial melting of carbonate-bearing mantle peridotite (e.g., Wallace & Green, 1988; Dalton & Presnall, 1998; Ghosh et al., 2009); and (3) fractional crystallization of a carbonate-rich alkaline silicate magma (Watkinson & Wyllie, 1971; Lee & Wyllie, 1994). In some localities, unusual transitional rocks (cf. carbonate-rich/bearing silicate rocks), such as bergalites, turjaites and okaites, are found temporally and spatially close to the carbonatite and associated silicate rocks (Eby, 1975; Keller et al., 1990; Keller, 1991, 1997; Bell & Simonetti, 1996). These rocks may provide important clues for the petrogenetic link between carbonatite and associated silicate rocks (Keller et al., 1990; Keller, 1991, 1997; Chakhmouradian & Zaitsev, 2004; Moore et al., 2009).
Apatite is a good indicator for magma evolution (Seifert et al., 2000; Piccoli & Candela, 2002; Zhang et al., 2012) and occurs in a wide range of magmatic systems such as mafic rocks (e.g., Brown & Peckett, 1977; Binder & Troll, 1989; Tribuzio et al., 1999), granitic rocks (e.g., Watson, 1980; Sha & Chappell, 1999; Marks et al., 2012) and syenitic rocks (e.g., Liferovich & Mitchell, 2006; Rønsbo, 2008). In particular, carbonatite (e.g., Bühn et al., 2001; Brassinnes et al., 2005; Chen & Simonetti, 2013) and related rocks (Eby, 1975; Keller et al., 1990; Bell & Simonetti, 1996; Moore et al., 2009) contain appreciable amounts of apatite. Experimental studies imply that many trace elements (e.g. Sr, rare earth elements [REE], Th, U) exhibit distinct partitioning behavior in carbonatitic and silicate melt systems (Klemme & Dalpé, 2003; Prowatke & Klemme, 2006; Hammouda et al., 2010). Thus, chemical differences between apatites from carbonatites and from associated silicate rocks are expected, and detailed investigations of apatite from the transitional dyke rocks might help to decipher the petrogenetic relationships between the two rock types.
The Kaiserstuhl Volcanic Complex (KVC) comprises a series of alkaline silicate rocks, associated with carbonatites as well as transitional dyke rocks (bergalites) and polygenic diatreme breccias (mixtures of carbonatite and silicate-rock fragments, e.g., Keller, 1984; Schleicher et al., 1990; Wimmenauer, 2003). In these rocks, apatite is a common accessory mineral. Hence, in the present work we studied the apatites from various rocks of KVC with the aim to (1) constrain the chemical variations of these apatites; (2) shed more light on the petrogenetic relationship between carbonatite and associated silicate rocks; and (3) investigate the abundance and variability of volatile elements (F, Cl, Br, S) in apatites from alkaline silicate rocks and carbonatite.
2. Geological setting and sample description
The KVC is located in the southern part of Upper Rhine Graben (Fig. 1a) and is part of extensive Cenozoic volcanism in Central Europe (Keller et al., 1990; Wilson & Downes, 1991, 2006; Riley et al., 1999; Wimmenauer, 2003). Ages of the Kaiserstuhl rocks range between 18 and 13 Ma (Lippolt et al., 1963; Baranyi et al., 1976; Kraml et al., 1995, 2006; Keller et al., 2002). The KVC is a sequence of early intrusion and extrusion of alkaline silicate rocks followed by late formation of carbonatites (Keller et al., 1990; Wimmenauer, 2003; Keller, 2008).
The silicate rocks are dominated by tephrites, essexites and phonolites (Fig. 1b). Sub-dominant rock types include olivine-melilitites, olivine-nephelinites, limburgites (olivine + augite + glassy groundmass), hauynophyres, syenites, shonkinite porphyries, and mondhaldeites (Wimmenauer, 2003; Keller, 2008). Shonkinite porphyry and mondhaldeite are fractionated dyke rocks from the tephritic magma. Carbonatites consist of sövite intrusions (Badberg and Orberg), alvikitic dykes and rare extrusive carbonatites (Keller, 1981, 2001; Hubberten et al., 1988). These are spatially and temporally associated with (1) magmatic diatreme breccias, some of which are polygenic mixtures consisting of mafic cumulates, carbonatite and silicate rock fragments, fill pipe structures in the sövite bodies (Baranyi, 1977; Katz & Keller, 1981; Hubberten et al., 1988; Keller et al., 1990; Sigmund, 1996), and (2) bergalite dyke rocks (Hubberten et al., 1988; Schleicher et al., 1990; Keller, 1991, 1997), which are carbonate-bearing, silica-undersaturated rocks with a mineral assemblage of melilite +sodalite +perovskite + biotite ± nepheline + apatite + magnetite + calcite (Keller, 2001, 2008).
In this study we investigate apatite separates from twelve KVC rocks, including four sövitic carbonatites from surface outcrops and drill cores, one diatreme breccia, one bergalite, one essexite, one mondhaldeite, one shonkinite porphyry, and three phonolitic rocks (Table 1a). Details on the petrography, mineralogy and geochemistry of these samples are given in Hubberten et al. (1988); Keller et al. (1990) and Schleicher et al. (1990). Apatite separates were produced by using standard heavy liquid methods and were subsequently mounted in epoxy, polished and carbon coated for further cathodoluminescence studies (CL), electron microprobe analysis (EMPA), secondary ion mass spectrometry (SIMS), and total reflection X-ray fluorescence analysis (TXRF).
3. Analytical methods
3.1. Electron microprobe analysis
Major and minor element compositions of apatites were determined using a JEOL 8900 electron microprobe operated in wavelength-dispersive mode at Tübingen University. A beam current of 10 nA and an acceleration voltage of 15 kV were used in connection with a defocused beam diameter of 10 μm. Durango apatite was used as the standard for Ca, P and F. Other standards included albite for Na, diopside for Si, hematite for Fe, rhodonite for Mn, barite for S, tugtupite for Cl, La-glass (REE16G) for La, Ce-glass (REE16G) for Ce, and synthetic GaAs and SrTiO3 for As and Sr, respectively. Counting times were 16 s for the Ca peak; 30 s for P, F, Na, Si, S and Sr peaks; and 60 s for Fe, Mn, As, La, Ce and Cl peaks, resulting in detection limits that are very similar to those reported by Marks et al. (2012). Data reduction was performed using the internal ZAF matrix correction software of JEOL (Armstrong, 1991).
Cathodoluminescence imaging, a powerful tool to reveal internal textures of minerals (e.g., Götze et al., 2013), was used prior to EMPA to document the internal structures in apatite crystals. According to Stormer et al. (1993) and Goldoff et al. (2012), long-time exposure under an electron beam might cause F and Cl diffusion depending on the orientation of the apatite crystal. To monitor these effects, we performed two tests. In the first test we analyzed apatite grains of sample M2 by EMPA along profiles parallel to the c- and a-axes of a crystal. Subsequently, these grains were exposed for around 3 minutes to the low-intensity (10 nA) CL-imaging tool integrated into the electron microprobe. The same grains were then reanalyzed with the analysis points set close to the previous ones. For the second test the sample holder was re-polished and carbon coated again. Additional apatite grains were analyzed and subsequently these grains were exposed for around 3 minutes to the high-intensity electron gun of a CL microscope (12–15 mA). These grains were then reanalyzed close to the previous points with the electron microprobe. The results from these two tests are shown in Fig. 2. Low-intensity CL did not cause statistically significant diffusion effects for F and Cl. In contrast, the high-intensity CL enhanced F diffusion. Thus, we conclude that the F and Cl diffusion effects are related to the intensity of the electron beam, and the use of a low (<10 nA) beam current for the CL study should not affect the quality of F and Cl analyses.
3.2. Secondary-ion mass spectrometry
Trace elements were analyzed using an upgraded CAMECA ims-3f ion microprobe at the Max-Planck-Institute for Chemistry in Mainz. The primary beam consisted of negative oxygen ions at a nominal accelerating potential of 12.5 kV and a beam current of 20 nA, resulting in a sputtering surface of 15–20 μm. Positive secondary ions of 16O, 23Na, 30Si, 44Ca, 88Sr, 89Y, 90Zr, 93Nb, 139La, 14°Ce, 141Pr, 146Nd, 147Sm, 153Eu, 157Gd, 163Dy, 167Er, 174Yb, 232Th, 238U were extracted in that order, using an acceleration potential of 4.5 kV with a 25 eV energy window and fully open entrance and exit slits. Ions were counted in a peak jumping mode and ratioed to 44Ca to quantify element abundances. Each measurement consisted of a six-cycle routine. At the beginning of each measurement, the energy distribution of 16O and subsequent peak centers for 44Ca, 88Sr, 14°Ce and 232Th were determined by scanning the peak in 20 steps across a 1.5 per mille wide B-field and the neighboring masses adjusted to these new peak centers. Measurement times were 30 s for Nb, Er, Yb; 20 s for Sm, Eu, Gd, Dy, U; 8 s for Y and Nd, 5 s for Ba, La, Ce, Pr, Th and 1 s for all other elements. Ion yields in phosphates are comparable to those from silicate glasses (Sano et al., 2002) and sensitivity factors were determined with MPI-DING reference glasses KL2-G, ML3B (Jochum et al., 2006) and NIST 610 (Jochum & Stoll, 2008). Energy filtering of molecular ion species was achieved with an offset of −80 eV. In apatite, the only molecular ions in the mass range of the REE that are not suppressed by energy filtering are monoxides and fluorides. However, the only substantial fluoride interference in the apatites was found at mass 159 where 140Ce19F interferes with Tb whereas hydride, hydroxide, and chloride peaks were found to be insignificant (Crozaz & Zinner, 1985; Zinner & Crozaz, 1986). Accordingly, Tb was not analyzed and oxide interferences on Eu, Gd, Dy, Er and Yb were corrected offline. Here, the general light-REE-enriched apatite pattern produces significant interferences where BaO+ interferes with 153Eu+, 141PrO+ with 157Gd+; 147SmO+ with 163Dy+; 151EuO+ with 167Er+; and both 158GdO+ and 158DyO+ with 174Yb+. The MO+ to M+ ratios used for the correction are 0.057, 0.165, 0.127, 0.058, 0.049, 0.145 and 0.127, respectively. The largest uncertainties are introduced by the correction of GdO and DyO interferences on Yb.
3.3. Total reflection X-ray fluorescence analysis
These analyses were performed using a S2 PICOFOX TXRF (Bruker AXS) at Universität Tübingen. Hand-picked apatite separates were milled to a fine-grained powder (< 15 μm). Around 1 mg apatite powder was dispensed in 1 ml Triton (1 vol.%) as solvent in an Eppendorf tube; 10 μl of a 10 mg/l Se solution was added as an internal standard. These suspensions were then homogenized for about 2 minutes, and aliquots of these suspensions were pipetted onto quartz glass sample discs and dried on a hot plate at 70°C. The dried samples where then analyzed for 1000 s (live time) using a Mo X-ray tube with an operating voltage of 50 kV and a beam current of 600 μA. The spectra were fitted with the Spectra 6.2 software (Bruker Nano GmbH). Further technical details can be found in Marks et al. (2012). TXRF is barely used to analyze solid materials (minerals and rocks) and available analytical approaches are limited (e.g., García-Heras et al., 1997; Misra et al., 2002; Marks et al., 2012). Here, we compare the results and uncertainties of apatite compositions analyzed by EPMA, SIMS and TXRF. Our data show good consistency between the three methods (see supplementary material, linked to this article and freely available online on the GSW website of the journal: http://eurjmin.geoscienceworld.org/), demonstrating that TXRF is reliable in quantifying REE.
4.1. CL imaging
The observed variations in CL intensity and color (Fig. 3) are generally related to the contents of luminescence activator elements, e.g. REE, Mn and Eu (Mariano, 1988; Koberski & Keller, 1995; Mitchell et al., 1997; Waychunas, 2002; Dempster et al., 2003). To better illustrate the different apatite types from various KVC rocks, we use the abbreviations listed in Table 1b.
4.1.1. Apatite from silicate rocks and carbonatites
Apatites from silicate rocks (S.Ap) generally show complex zonations, either oscillatory or patchy (Fig. 3a–c), which, however, do not reflect any detectable systematic compositional variation. Apatites from carbonatites (C.Ap) appear brighter than those from silicate rocks and occasionally contain micro-holes and calcite inclusions (Fig. 3d–e). Sample KB3, which has an unusual orbicular texture in hand-specimen, shows varying CL intensities in apatite separates, and at least two types of apatite (brightly luminescent and dull) can be distinguished (Fig. 3f).
4.1.2. Apatite from bergalite
Most apatite grains from bergalite show regular zoning with a gray inner core, darker outer core, and a bright rim (Fig. 3g–h). The boundaries between these zones are generally abrupt and often display euhedral outlines. In rare cases, the core appears to be partially resorbed (rounded shape; Fig. 3h). Apparently unzoned grains show a similar CL brightness as the cores (Fig. 3g) and have similar chemical compositions (see below). We assume that this is an effect of different orientations of the crystals in the sample mount.
4.1.3. Apatite from diatreme breccia
Apatites from a diatreme breccia comprise three different populations (Fig. 3i): (1) grains with lower CL brightness (D.Ap); (2) higher-brightness grains with occasional micro-holes and calcite inclusions (B.Ap, Fig. 3j); and (3) grains with replacement textures (R.Ap), resembling apatite population (1) partially replaced by apatite (2) preferentially around the margins of crystals and along cracks (Fig. 3k–l).
4.2. Compositional variation
4.2.1. Comparison of C.Ap and S.Ap
Strontium is generally higher in C.Ap than in S.Ap (Fig. 4a; Table 2). Iron and Mn are low in C.Ap (mostly <0.005 apfu) with more than 50 % of values below the respective detection limits (360 ppm for Fe, 220 ppm for Mn), whereas they are higher in S.Ap (mostly >0.005 apfu, Fig. 4a). The C.Ap and S.Ap datasets generally overlap in their REE and Na contents (Fig. 4b and 4e), with apatites from sövite KB3 showing a REE variation (from 0.06 to 0.15 apfu) that is distinct from other C.Ap (Table 2). Thorium in C.Ap is commonly lower than 10 ppm with some cases below the detection limit of 0.4 ppm (Table 3, Fig. 4f). Uranium in C.Ap is always below the detection limit (0.04 ppm). In contrast, the S.Ap has much higher Th (40 to 200 ppm) and U (6 to 72 ppm) contents (Fig. 4f). Silicon and S contents are generally lower in C.Ap than in S.Ap (Fig. 4c), reaching 0.17 apfu Si and 0.09 apfu S in S.Ap. An exception, again, is sövite KB3, with Si contents as high as 0.27 apfu. Arsenic is higher in C.Ap (23–47 ppm) than in S.Ap (<10 ppm; Table 3). Niobium in C.Ap is always higher than 2 ppm, compared to < 1ppm in S.Ap (Fig. 4f).
Variation in F is very similar in C.Ap and S.Ap (Fig. 4d), whereas Cl is significantly lower in C.Ap (<0.01 apfu) than in S.Ap (0.02-0.2 apfu). In both groups, F and Cl are negatively correlated to each other (Fig. 4d). Bromine is always below the detection limit of TXRF in C.Ap (around 0.5 ppm), but generally between 1 and 3 ppm in S.Ap with an exceptionally high Br content of 12 ppm in apatites from a phonolitic dyke (Table 4).
4.2.2. Apatite from bergalite
The CL zonation in bergalite apatites (Fig. 3i–k) correlates with compositional differences (Fig. 5; Table 2). The inner cores are characterized by low Sr and REE and intermediate Na contents (Fig. 5; Table 2). The rims are highest in Sr and lowest in Na. The outer cores generally overlap with the inner cores in terms of Sr, REE and Na. Silicon contents increase from inner to outer core and decrease towards the rims (Fig. 5). Sulfur concentrations in the inner cores show a large variation and overlap considerably with those in the outer cores. The rims, however, are notably poorer in S (Table 2). The average F content in the inner cores (0.35 apfu) is slightly lower than in the outer cores (0.47 apfu; Fig. 5). The highest values were found in the rims (0.75 apfu). The contents of Cl and calculated OH are lower in the rims than in the cores (Table 2).
4.2.3. Apatite from diatreme breccia
The B.Ap grains show large variation in many elements and commonly overlap with the data for D.Ap (Fig. 6). Fluorine and Cl contents in both groups exhibit significant differences. R.Ap.rel always shows the same compositional range as defined by D.Ap, whereas R.Ap.sec is compositionally similar to B.Ap.
5.1. Interpretation of the compositional differences between C.Ap and S.Ap
Chemical variation of magmatic apatites depends on apatite–melt partitioning coefficients, melt composition, charge-balance constraints for multi-element substitutions, and possibly other factors (e.g., Peng et al., 1997; Sha & Chappell, 1999; Harlov et al., 2002). In the following, we evaluate the influences of these factors on the chemical differences between C.Ap and S.Ap.
5.1.1. Apatite–melt partition coefficients and host melt compositions
According to experimental work (Klemme & Dalpé, 2003; Prowatke & Klemme, 2006; Hammouda et al., 2010), apatite in the carbonatitic system has lower Dapatite-melt values for Sr, La and Ce and higher values for Nb in comparison with those from the silicate-melt system (Supplementary Table; Klemme & Dalpé, 2003; Prowatke & Klemme, 2006; Hammouda et al., 2010). Indeed, C.Ap from the Kaiserstuhl has higher Nb concentrations than S.Ap (Table 3), implying that partition coefficients play an important role. However, the Sr, La and Ce concentrations in C.Ap are not lower than those in S.Ap, although they have lower Dapatite-melt values in C.Ap (Table 2). We attribute this to the effect of melt composition, since carbonatitic melts are usually enriched in Sr and REE compared to silicate melts (Hubberten et al., 1988; Keller et al., 1990; Martin et al., 2013). For instance, according to Keller et al., 1990 and unpublished), carbonatites from the KVC contain up to 16 000 ppm Sr, whereas the silicate rocks contain < 1000 ppm.
5.1.2. Substitution mechanisms
A positive correlation between S and Si for C.Ap and S.Ap (Fig. 7a) indicates that sulfur is mainly incorporated according to the substitution
(Pan & Fleet, 2002). However, the high Si contents in apatite from sövite KB3 cannot be explained this way. Apatites from this sample also show the highest La + Ce contents, which clearly correlate with Si (Fig. 7b), pointing to the importance of the substitution
as proposed for apatites from other localities (e.g., Comodi et al., 1999; Harlov et al., 2002, 2005; MacDonald et al., 2013). The combination of these two substitutions (Fig. 7c) demonstrates that both mechanisms can explain the variation observed in C.Ap and S.Ap. However, for sample KB3, a significant deviation from the 1:1 line exists. This is probably because only La + Ce were analyzed by EPMA, thereby underestimating the total REE. Alternatively, other substitution mechanisms may play a role for apatites from this sample. The substitution mechanism REE3+ + Na+ = 2 Ca2+ (Harlov et al., 2002, 2005; MacDonald et al., 2013), however, does not play a major role since Na does not correlate with La + Ce (Fig. 4b) and apatites from KB3 are not exceptionally Na-rich (Table 2).
5.2. Interpretation of the zoned apatites from bergalite
Zoning patterns in apatite may be related to a variety of processes such as fractional crystallization, magma mixing, magmatic degassing, diffusion, disequilibrium crystallization, partial dissolution, and recrystallisation (e.g., Jolliff et al., 1989; Brenan, 1994; Sha & Chappell, 1999; Tepper & Kuehner, 1999; Boyce & Hervig, 2008, 2009; Chakhmouradian et al. 2008; Rønsbo, 2008). The zoning textures found in the bergalite apatites (Fig. 5) display discontinuous concentration changes in both volatile (F, Cl and S) and non-volatile elements (e.g., Sr and Si). Therefore, it seems unlikely that magma degassing can fully explain these zonations.
Clear compositional differences between inner and outer cores are only observed for Si (Fig. 5). The difference may be caused by fractionation of the parental melt combined with changing trace element substitution mechanisms in apatite. However, the observed CL textures and chemical profiles show abrupt changes (Fig. 3g–h) and rounded shapes and embayments in some cores (Fig. 3g–h). This can be explained by the stirring of magmas during ascent in high-level magma bodies. Such stirring could result in partial thermal and chemical changes of the magma and lead to disequilibrium crystallization of apatite.
The chemical differences between core and rim probably rather reflect the evolving magma composition during bergalite crystallization. The rims of the bergalite apatites are very similar to those of C.Ap (e.g., relatively high CL brightness, low Si, Cl and S, but high Sr contents; Fig. 8), whereas the cores are chemically similar to S.Ap. To further evaluate the compositional similarity, we calculated Sr concentrations of the related parental melts from the apatite compositions. We assume that the cores crystallized from a melilite nephelinite melt, whereas the rims crystallized from a late-stage melt with carbonatitic affinity. The DSr partition coefficients used here are 5.1 for apatite/phonolite, 1.56 for apatite/tephrite and 0.53 for apatite/carbonatite (Prowatke & Klemme, 2006; Hammouda et al., 2010). A comparison of the modeling result with the whole-rock Sr contents (Keller, unpublished data) shows that the calculated Sr concentrations are in agreement with, or close to, those from whole-rock data (Fig. 9). Moreover, it also supports the hypothesis that the cores of the bergalite apatites correspond to S.Ap., whereas the rims correspond to C.Ap.
The overgrowth of C.Ap-like rims on S.Ap-like cores in the apatite crystals from bergalite can be explained by the following processes: (1) fractional crystallization of a melilite nephelinite magma; and (2) mixing of two independent magmas, a silicate and a carbonatitic one. These two magmas will either mix to form a homogeneous intermediate product, or remain immiscible. Watkinson & Wyllie (1971) produced calcite + cancrinite + melilite at 1 kbar using a carbonated nepheline-rich liquid at ca. 600 ° C, indicating that carbonatite can be generated by fractional crystallization of a carbonated nephelinite magma. Accordingly, it was assumed that the bergalite melt fractionated from a parental CO2-rich olivine melilite nephelinite mantle-derived magma (Keller et al., 1990; Keller, 2008). This fractionation caused the enrichment of CO2, Sr, and REE in the evolved melt (Keller, 1991, 1997). The residual melt from which the C.Ap-like rims crystallized would, in this case, have carbonatitic affinity. Dempster et al. (2003) demonstrated that discontinuously zoned apatite can monitor changing permeability in granites, in which apatite cores grew early in the magma chamber, whereas rims record late-stage crystallization within more isolated interstitial melt pockets in a highly solidified and compacted crystal mush. Similarly, C.Ap-like rims may represent the late-stage crystallization from carbonate-rich melt pockets.
Mixing of carbonatitic and melilititic magmas could theoretically yield an intermediate product such as bergalite. However, the Sr contents in the rims do not support this assumption, because they generally plot outside the interval defined by the two assumed end-members, C.Ap and S.Ap (Fig. 8). Moore et al. (2009) interpreted the genesis of silicocarbonatite dykes by mingling between carbonatitic and silicate magmas, involving liquid immiscibility. If we assume a carbonatitic melt was injected into a melilititic melt with immiscibility during the late stage, it is still not clear whether core-rim zoned apatite or replaced apatite (as observed in diatreme breccias; see below) will form during this mingling process.
In summary, the zoning textures in bergalitic apatites imply that they initially nucleated in a silicate melt and later developed a rim while in equilibrium with a carbonatitic melt. We favour hypothesis (1) to explain the observed apatite textures.
5.3. Interpretation of the replacement textures in apatite from diatreme breccias
Mineral replacement reactions are generally related to dissolution-reprecipitation processes resulting from chemical weathering, leaching, alteration, metamorphism or metasomatism (Putnis, 2002; Engvik et al., 2009). Replacement of chlorapatite by hydroxy-fluorapatite during metasomatism in metagabbro has been described by Harlov et al. (2002) and Engvik et al. (2009) and has been experimentally reproduced under alkaline hydrothermal conditions (Yanagisawa et al., 1999; Harlov et al., 2002). However, the replacement of hydroxyapatite by fluorapatite, as documented experimentally by Rendón-Angeles et al. (2000), has so far not been reported for natural apatites. In the present study, the replacement textures found in apatites from the diatreme breccia reveal hydroxy-fluorapatite replaced by fluorine-hydroxyapatite (Figs. 3i–l and 7). The R.Ap.sec represents a late-stage fluid/melt re-equilibration product. Compositional and CL similarities between R.Ap.rel and S.Ap indicate that the former represent relict grains of early-formed apatites crystallized from a silicate melt. In contrast, R.Ap.sec is similar to C.Ap, implying formation from a carbonatitic melt. These observations suggest that a carbonatitic magma captured silicate-rock fragments and rapidly rose to the surface to form the diatreme breccias. During this process, metasomatism of the silicate xenoliths by the carbonatitic melt caused the replacement textures observed in some of the apatites.
5.4. Abundances and variability of volatile elements (F, Cl, Br, S) in apatite
Apatites from the KVC are mostly fluorine-hydroxyapatite or hydroxy-fluorapatite (Fig. 10). Their Cl contents are quite low, especially in the C.Ap (Fig. 11a). This agrees well with previous studies where apatites from carbonatite complexes were reported as Cl-poor with mostly less than 0.1 wt.% Cl (e.g., Eby, 1975; Hogarth, 1989; Seifert et al., 2000; Patiño Douce et al., 2011). The low Cl content was explained by Cl partitioning into an aqueous fluid phase, which generally coexisted with the carbonatitic melt (Gittins, 1989; Seifert et al., 2000). Bromine contents generally correlate with Cl: in Cl-poor C.Ap Br is below the detection limit but reaches around 2.5 ppm in relatively Cl-rich S.Ap. This is consistent with previously found positive correlations between Cl and Br (O’Reilly & Griffin, 2000; Marks et al., 2012), implying Cl and Br behave similarly during incorporation by apatite.
Sommerauer & Katz-Lehnert (1985) determined carbon concentrations in some apatites from Kaiserstuhl using coulometric titration and Fourier-transformed infrared (FTIR) spectroscopy, revealing up to 3.9 wt.% CO2 in carbonatitic apatites and up to 0.9 wt.% in apatites from the silicate rock. However, the calcite inclusions in C.Ap observed in the present work and the strong zoning in apatite from the bergalite samples make the interpretation of Sommerauer & Katz-Lehnert (1985) problematic, because their work was performed on bulk powder samples.
Sulfur is incorporated in apatite as sulfate (Pan & Fleet, 2002), and most natural apatites contain <0.5 wt.% SO3 (Broderick et al., 2007), although concentrations of up to 2 wt.% have been reported (Imai et al., 1993; Streck & Dilles, 1998; Broderick et al., 2007; Parat et al., 2011). The S.Ap from KVC are rich in SO3 (0.74–1.11 wt.%), indicating that the KVC silicate magma was enriched in S and crystallized under relatively oxidizing conditions (Imai et al., 1993; Peng et al., 1997; Parat et al., 2002). In contrast, C.Ap is relatively poor in SO3 (0.02–0.5 wt.%). These low contents of both SO3 and Cl (Fig. 11b) probably reflect the relative depletion of carbonatitic magma in both elements. It is unlikely to represent a Cl and S loss during degassing, because when nephelinitic magma fractionated to carbonatitic magma, non-volatile elements (e.g., Si) vary with Cl and S as well (see the compositional variation in core-rim zoned apatite from bergalite in Fig. 5).
Textural and chemical variations of apatites from the KVC provide important information for the genesis of their host rocks: Apatites from bergalite allow for the reconstruction of the genetic relations between carbonatites and associated alkaline silicate rocks. Their textural and compositional core-rim zonation implies that these apatites initially crystallized from a silicate melt and further developed in a melt with carbonatitic affinity. This carbonatitic melt is likely a product of prolonged fractional crystallization of an initial CO2-rich melilite nephelinite melt. On the other hand, apatites from silicate rock fragments found in a diatreme breccia were subsequently metasomatized by late-injected carbonatitic melt, as shown by their replacement textures. Overall, our study shows that apatite is an equally sensitive monitor for primary magmatic processes (e.g., fractional crystallization) and secondary processes, such as metasomatic overprint.
Dr. M. Rahn and Prof. H. Schleicher are gratefully thanked for kindly providing some of the investigated apatite separates and for insightful discussions. Prof. Ch. Ma is thanked for the helpful suggestions and comments. We also thank B. Walter for assistance with apatite separation and M. Mangler for discussions about the TXRF method. The China Scholarship Council (CSC) is thanked for granting scholarship (2010641006) to the first author. H. Teiber is supported by the Deutsche Forschungsgemeinschaft (grant MA 2563/3-1), which is gratefully acknowledged. We appreciate the detailed reviews of A. Chakhmouradian and T. Hammouda as well as the editorial handling of R. Gieré.
- Received 18 June 2013.
- Modified version received 14 January 2014.
- Accepted 15 January 2014.