U-Pb Ages of Zircon Inclusions in Sapphires from Ratnapura and Balangoda (Sri Lanka) and Implications for Geographic Origin
ABSTRACT
Five sapphires from the secondary placer deposits of Ratnapura and Balangoda in Sri Lanka were classified as being of metamorphic/metasomatic/non-basalt-related origin based on trace-element analysis (LA-ICP-MS) and inclusion characterization. Two sapphires—one from each deposit—contained suitable zircon inclusions that were dated using the LA-ICP-MS method. They yielded U-Pb ages of approximately 549 Ma. The results suggest that the zircon enclosed in the sapphires probably formed in a high-temperature event at the end of the Precambrian granulite facies metamorphism of the ancient Gondwana continent and that this event coincides with the crystallization of the sapphires. The metamorphic character of these sapphires is confirmed from the trace-element composition as well as the solid- and liquid-phase inclusions.
INTRODUCTION
Blue sapphire is one of the most appreciated gemstones, and its geographic-geological origin significantly affects its value. Most gem-quality sapphires are found in secondary placer deposits worldwide, and little is known about their crystallization in the geological environment. Although the geological context of gem-quality sapphires from secondary deposits is unknown, studying their geochemistry and the type of inclusions in combination with dating zircon inclusions could increase our understanding of their geological formation and geographic origin.
Zircon is probably the most powerful chronometer, since it can record peak temperature(s) but will generally survive post-cooling history (for a summary of zircon and its use as a geochronometer, see Hanchar and Hoskin, 2003). Two independent clocks or radiogenic U-decay lines record the time elapsed starting with the crystallization of zircon. Complications arise if the zircon inclusion is zoned, displaying multiple growth events at different times. For example, a zircon in a granitic magma chamber may have grown continuously during magmatic crystallization pulses, as indicated by oscillatory zoning (fine growth bands best viewed with cathodoluminescence), and then be eroded and washed into a sediment that is metamorphosed to high-grade metamorphic conditions. New zircon growth may occur around the former magmatic zircon under favorable conditions. The new zircon may recrystallize homogenously, reflecting the metamorphic event of its formation. In the latter case, the U-Pb clock will be reset and the U-Pb dating of this inclusion in sapphire will provide age information about the crystallization event of the sapphire. In the case of dating an inherited detrital zircon—normally in the center of the crystal—the maximum age of sapphire crystallization obtained may be close to the actual crystallization age or a considerable interval away. Zircon inclusions are often reported from gem-quality sapphires (Coenraads et al., 1990; Sutherland et al., 1998a, 1998b, 2008, 2015; Link, 2015; Zeug et al., 2017). Zircon in sapphire has been dated in alkaline basalts of Australia’s Central Province by Coenraads et al. (1990); from the Mercaderes–Rio Mayo area in southwest Colombia, related to northern Miocene Andean uplift and volcanism (Sutherland et al., 2008); from the New England sapphire fields in Australia (Abduriyim et al., 2012); from the Lava Plains area of northeast Australia, related to a complex interplay between initial metasomatized mantle involvement with infiltrations of felsic and mafic melts into the crustal levels (Sutherland et al., 2015); from the alkali basalt-related placers at Primorye in Russia (Akinin et al., 2017); and from Mogok and Madagascar (Link, 2015, 2016).
This study reports on the gemological and geochemical characteristics of Sri Lankan sapphires from the secondary deposits of Balangoda and Ratnapura, as well as the results of U-Pb dating of zircon inclusions in establishing some characteristic parameters for sapphire from these deposits. The sapphires were characterized by a combination of classical gemological methods such as solid and fluid inclusion study, polarized ultraviolet/visible/near-infrared absorption (UV-Vis-NIR) spectroscopy, Fourier-transform infrared (FTIR) spectroscopy, Raman spectroscopy, and geochemical methods such as laser ablation–inductively coupled plasma–mass spectrometry (LA-ICP-MS) for trace-element analysis and U-Pb dating.
MATERIALS AND METHODS
Five rough sapphires (table 1) from the secondary placer deposits of Ratnapura and Balangoda were selected from the Gübelin Reference Stone Collection at the Gübelin Gem Lab (GGL). Samples SASLR016, SASLR017, SASLR019, and SASLA01 were acquired by GGL at the Ratnapura gem market in May 2005, while SASLR050_01 was purchased directly at the Marapona mine in the mining area of Ratnapura in February 2009. One sapphire from Balangoda (SASLA01, violet to dark blue) and one from Ratnapura (SASLR050_01, milky white to light purple-blue) presented zircon inclusions (>30 µm) suitable for U-Pb dating using LA-ICP-MS (table 1).
The sapphires were studied using a standard gemological microscope at GGL, and their specific fluorescence characteristics were monitored in a dark room (using 3 W lamps emitting 365 nm long-wave and 254 nm short-wave UV). Specific gravity was measured using the hydrostatic method. Mineral inclusions were identified using a Renishaw Raman 1000 spectrometer with an Ar+ laser and a wavelength of 514 nm (without the use of a polarizer), connected to a Leica DMLM optical microscope. The laser excitation used was from 10 to 15 mW, in confocal mode, with 20 to 50× magnification, from 200 to 2000 cm–1 (three cycles with an acquisition time from 10 to 30 seconds). Rayleigh scattering was blocked by a holographic notch filter, and the backscattered light was dispersed on an 1800 groove/mm holographic grating with a 50 micron slit width (approximately 1.5 cm–1 spectral resolution). The system was calibrated using a diamond and its single characteristic first-order band at 1331.8 cm–1, which designates its normal mode of vibration, usually referred to as LO=TO mode (longitudinal optical = transversal optical). LO and TO have the same frequency due to the high symmetry of the diamond lattice (Nasdala et al., 2005). UV-Vis-NIR absorption spectra were collected in the range between 250 and 1000 nm using a Varian Cary 5000 UV-Vis-NIR spectrophotometer with diffraction grating polarizers. We used a data sampling interval and spectral bandwidth of 0.5 nm and a scan rate of 150 nm/min. FTIR absorption spectra were acquired from 6000 to 400 cm–1 using a Varian 640 spectrometer with a resolution of 4 cm–1 and 200 scans and a diffuse reflectance (DRIFT) accessory as the beam condenser. Trace-element analysis (Mg, Ti, V, Cr, Fe, and Ga) was performed on a Perkin Elmer ELAN DRC-e single collector quadrupole mass spectrometer combined with a 193 nm ESI Excimer gas laser ablation system. A set of three single-spot analyses (120 μm diameter) was collected on each sample using a laser frequency of 10 Hz, an ablation time of 50 seconds, and a laser energy of 6.2 J/cm2. The mass spectrometer performance was optimized to maximum intensities at U/Th ratios of ~1 and ThO/Th less than 0.3 using 16.25 liters per minute Ar plasma gas, 0.88 L/min argon as nebulizer gas, and 1 L/min helium as sample gas. Multi-element NIST610 was used as the glass standard for external calibration; internal calibration was done by normalizing to 100% cations of stoichiometric corundum. The data reduction was carried out using an in-house spreadsheet following Longerich et al. (1996).
Zircon inclusions were carefully localized and cut with a special saw using a fine cord as a cutting blade. This cutting method allowed us to obtain the largest possible surface of the zircon crystals. Sapphires were embedded into a special epoxy to avoid degassing under vacuum. The zircons reached up to 120 µm in size. Cathodoluminescence images of zircon inclusions were obtained at the University of Geneva, with a JEOL JSM7001F scanning electron microscope. Acceleration voltage was 15 kV, probe current was 3.2 nA, and emission current was approximately 90 µA. Cathodoluminescence imaging is a powerful tool in identifying multiple growth zones in zircon related to different crystallization pulses during magmatic or metamorphic events.
The Raman spectrum of the dated zircon SASLR050_01 was obtained at the University of Geneva using a confocal LABRAM spectrometer equipped with a green 532.12 nm Nd-YAG laser coupled to a charge-coupled device (CCD) detector and an optical microscope (Olympus BX51, 100× objective lens). The system was calibrated using a diamond and its single characteristic first-order band at 1331.8 cm–1. The parameters were: a spot diameter of 2.5 µm, a depth resolution of 5 µm, and a spatial resolution of <10 µm3. A laser power of 10 mW was used. The spectrum was acquired in the range from 200 to 1700 cm–1 in three cycles with an acquisition time of 30 seconds.
U-Pb dating of zircon inclusions was carried out on an Element XR sector field ICP-MS (Thermo Fisher Scientific) interfaced to an UP 193-FX ArF 193 nm excimer ablation system (New Wave Research) at the University of Lausanne. Two spot analyses (diameter <25 µm) were carried out on one zircon inclusion in sample SASLA01 from Balangoda, and three spots were measured on two zircon inclusions in SASLR050_1 from Ratnapura. The zircon samples GJ1 (206Pb/238U age 609.7 ± 1.8 Ma; Jackson et al., 2004) and Plešovice (206Pb/238U age 337.13 ± 1.8 Ma; Sláma et al., 2008) were used as external and secondary reference, respectively. The dwell time was 10 ms for 207Pb, 8 ms for 208Pb, 8 ms for 235U, and 5 ms for all other masses. Laser pulse duration was approximately 5 ns, with a repetition rate of five pulses per second. The data reduction was done with LAMTRACE, a Lotus 1-2-3 spreadsheet written by Simon Jackson of Macquarie University in Sydney.
GEMOLOGICAL PROPERTIES AND SPECTROSCOPY
Five rough to partially cut sapphires were selected from the Ratnapura and Balangoda deposits in Sri Lanka (figure 1A and table 1). They ranged in color from bluish white to blue and dark blue. Under short- and long-wave UV, they displayed a weak to strong reaction with yellow to orange colors. Their UV-Vis-NIR spectra showed a similar band distribution. In a representative UV-Vis-NIR spectrum of the sapphire from Ratnapura (SASLR050_01), the absorption bands at 377, 380, and 450 nm (Ferguson and Fielding, 1971; Schwarz et al., 2000, 2008) are attributed to Fe3+ (figure 2). In addition, the Cr3+ absorption band was present at 560 nm (Schwarz et al., 2008). This sample also showed the broad Fe2+→Ti4+ intervalence charge transfer band between 570 and 700 nm that is responsible for the blue color. We did not observe the pronounced band at around 800 nm, often attributed to the Fe2+→Fe3+ intervalence charge transfer, ruling out a possible basalt-related origin (Schwarz et al., 2000; Kan-Nyunt et al., 2013; Emmett et al., 2017). Bands due to artifacts are also observed around 550 nm as well as at 600 nm due to Wood’s anomaly, and at around 800 nm due to detector change.
Solid- and Liquid-Phase Inclusions. The Ratnapura samples contained abundant inclusions (figure 1), and a summary is given in table 2 for both deposits. The importance of mineral inclusions and their link to the genesis of sapphire deposits has been emphasized by many studies, such as De Maesschalck and Oen (1989), Beran and Rossman (2006), Palanza et al. (2008), and Khamloet et al. (2014). FTIR analysis of samples from Ratnapura revealed hematite, carbonate, kaolinite, boehmite, and CO2 inclusions (figure 1B), which confirm the observations of De Maesschalck and Oen (1989) for Ratnapura. Zircon, apatite, spinel, ilmenite (figure 1C), and graphite were identified by Raman spectroscopy. Spinel, ilmenite, and CO2 inclusions point to a medium- to high-grade metamorphic host rock. According to Beran and Rossman (2006), kaolinite-group minerals are the most common phases in turbid parts of corundum. Rakotondrazafy et al. (2008) observed the occurrence of quartz-free, kaolinite-bearing rocks hosting veins of sapphire in Andilamena, Madagascar. The occurrence of kaolinite might be related to the mineral paragenesis of the host rock, or it could reflect a late, low-temperature event after the formation of the sapphire.
Raman Spectrum of Dated Zircon Inclusion. The Raman spectrum of the dated zircon inclusion SASLA01 (Balangoda) in figure 3 indicates the main band positions for the molecular vibrational modes. The spectrum displays the typical Ʋ3 band near 1006 cm–1, which is considered to reflect the anti-symmetric stretching of the SiO4 group (Dawson et al., 1971). Its position near 1006 cm–1 demonstrates the well-crystallized nature of this zircon crystal (Davies et al., 2015). The Ʋ2 band at 440 and 356 cm–1 is related to bending vibration around the SiO4 groups (Dawson et al., 1971). The Ʋ1 band near 973 cm–1 is caused by the Si-O symmetric stretching band (Dawson et al., 1971). Generally, all of these bands are quite sharp—the Ʋ2 band at 356 cm–1 has a FWHM of 13.6, and the Ʋ3 band near 1006 cm–1 has a FWHM of 8.8—which is characteristic for crystalline material (Davies et al., 2015).
Trace-Element Chemistry. Trace-element geochemistry, in particular the elements Fe, Ti, Ga, Cr, Mg, and V, is used for discriminating the geographic origin and/or type of sapphire deposit—i.e., metamorphic, non-basalt-related versus basalt-related (e.g., Sutherland et al., 1998a,b, 2002; Schwarz et al., 2000, 2008; Saminpanya et al., 2003; Abduriyim and Kitawaki, 2006; Peucat et al., 2007; Giuliani et al., 2014). The results of the discriminating elements and ratios used in figure 4 are given in table 3.
The Fe content of the sapphires from Ratnapura ranges from 286 to 1213 ppmw, Ga content from 20 to 99 ppmw, and Cr content from 2 to 102 ppmw. Ti shows a large variation from 92 to 253 ppmw, while V content measured between 14 and 70 ppmw. For the Balangoda sample, the following concentrations were obtained: 514 ppmw Fe, 110 ppmw Ga, 17 ppmw Cr, 42 ppmw Ti, and 6 ppmw V. These values lie within the range reported by Abduriyim and Kitawaki (2006).
The Ga/Mg ratio has been applied to classify metamorphic and magmatic deposits (Peucat et al., 2007). The value is generally high for magmatic and basalt-hosted sapphires (10 or much higher) and low for metamorphic and metasomatic ones (10 or much lower). Sapphires from Ratnapura display Ga/Mg ratios of 0.18 to 0.84 (table 3), coinciding with the value of 0.6 reported for a blue sapphire from Ratnapura by Peucat et al. (2007). The Ga/Mg value of the sapphire from Balangoda was determined to be 5.03 (table 3). Both values lie within the range indicated for metamorphic sapphire. In figure 4A, Fe (ppmw) is plotted versus the Ga/Mg ratio and the field for metamorphic sapphire occurrences is outlined. Values for both analyzed deposits are located well within the field defined by Peucat et al. (2007) for metamorphic sapphires from Sri Lanka. In the Fe-(Ti × 10)-(Mg × 100) ternary diagram (figure 4B), the sapphires plot in the field for metamorphic sapphires, as well as in the area characteristic for Sri Lankan sapphires defined by Schwarz et al. (2008). In the Cr2O3/Ga2O3 versus Fe2O3/TiO2 diagram (Abduriyim and Kitawaki, 2006), three samples from Ratnapura plot within the compositional field for a non-basalt-related origin and another plots outside the field. The Balangoda sample plots within the Ratnapura field (figure 4C). When the samples are displayed in the FeO + TiO2 + Ga2O3 versus FeO-Cr2O3-MgO-V2O3 diagram (Giuliani et al., 2014), they lie at the lower end of the field, characteristic for corundum in metasomatites (figure 4D).
U-Pb Dating of Zircon Inclusions. In characterizing the Sri Lankan sapphires (Elmaleh, 2015), one sample from Ratnapura (SASLR050_01, figure 5) and one from Balangoda (SASLA01, figure 6) provided zircon inclusions ranging up to 120 µm in size that were suitable for dating with conventional LA-ICP-MS. Under the scanning electron microscope, the backscattered images of the polished surface of sapphire SASLR050_01 (figure 5, A and B) are shown with its zircon inclusions observed in backscattered and cathodoluminescence mode (figure 5, C–F). The dated zircon inclusion in sapphire SASLA01 is shown in figure 6 in the scanning electron microscope (left), under cathodoluminescence (center), and with the craters of the LA-ICP-MS measurement (right). It showed sector zoning under cathodoluminescence and a small overgrowth along the rim that could not be dated due to its size of 10 µm (figure 6, center). Radial cracks around zircon inclusions were observed in the host sapphire of both deposits (figure 5, D and F). During cooling and decompression of the host rock, the zircon inclusion expands more rapidly than the enclosing sapphire crystal, resulting in stress along the wall and the formation of cracks in the host sapphire (Noguchi et al., 2013). Crack formation due to heating is considered very unlikely since the samples were purchased at the mine as rough sapphires and from reputable sources (at the mine or near the mine). In addition, the presence of inclusions such as kaolinite and boehmite, which degrade at relatively low temperatures, is a strong indication that the samples are unheated. In the concordia diagram for the zircon inclusion in the Balangoda sapphire (SASLA01; see figure 7, left, and table 4 for results of zircon dating), the axes are defined by the ratios of the radiogenic daughter Pb isotopes (206Pb and 207Pb) to their corresponding parent U isotopes (236U and 235U). At time zero, when the two U-Pb clocks were set or when the zircon formed, there was no radiogenic Pb in the zircon. Once the zircon was below a critical temperature, radiogenic Pb (206Pb and 207Pb) would accumulate due to the decay of 236U and 235U. The ratios of 207Pb/235U and 206Pb/238U would increase with time, and by analyzing the appropriate isotopic ratios one can determine the age of the zircon inclusion using the established equation with the decay constant for the corresponding decay series (e.g., Harley and Kelly, 2007). Two analyses were obtained on the same zircon inclusion (figure 5, right, with the ablation craters of sample SASLA01), and the Pb/U ratios (figure 7, left) correspond to a concordia age of 547.7 ± 5.7 Ma (table 4).
A weighted mean diagram (figure 7, right) is used to determine the mean age from U-Pb analysis of the zircons in sapphire SASLR050_01 from Ratnapura, since the analyses plot in the conventional concordia diagram above the concordia line. This may be related to an excess of radiogenic 206Pb or a loss of 238U. The younger range of uncertainty is then applied to approximate the crystallization age of zircon. In other words, the obtained age of 557 Ma should be reduced by the uncertainty of approximately 8.4 Ma, resulting in an age of approximately 549 Ma (table 4).
In summary, the U-Pb zircon ages obtained for the sapphires from Ratnapura and Balangoda show a similar crystallization age of around 549 Ma and are therefore probably related to the same metamorphic event (Elmaleh, 2015; Elmaleh et al., 2015a,b; Schmidt et al., 2017).
Geological Considerations: Time of Sapphire Crystallization. The Ratnapura and Balangoda deposits are located in the Highland Complex of Sri Lanka. The Highland Complex is part of an extensive Precambrian granulite belt of the ancient Gondwana continent related to the Pan-African evolution and comprising the Kerala Khondalite Belt in southern India, Madagascar, Mozambique, and Tanzania as well as the Lütz-Holm Bay area in Antarctica (Baur et al., 1991; Dissanayake and Chandrajith, 1999). Baur et al. (1991) systematically analyzed zircons of the Central Highland in the granulite facies formation and concluded that localized charnockitization or the local influx of CO2-enriched fluids are responsible for the granulite facies metamorphism. The main phase of granulite facies metamorphism is bracketed between approximately 660 Ma and 550 Ma and related to the Pan-African orogeny (Baur et al., 1991; Dissanayake and Chandrajith, 1999). In addition, Baur et al. (1991) record in their zircon population a younger event at 547 ± 35 Ma, which they assign to a high-grade regional metamorphic event at the end of the granulite facies metamorphism. In a zircon inclusion in a pink sapphire of Madagascar, Link (2015) reports an age of 652 ± 41 Ma and relates it to the Pan-African tectonic-metamorphic event, which occurred within a time window of 730–550 Ma (Black and Liegeois, 1993). This would indicate an earlier crystallization event of zircon in sapphire from Madagascar than from Sri Lanka, in line with the time windows for the regional high-grade metamorphic events in the respective areas.
Unfortunately, placer deposits do not provide sufficient information about the geological background. Nevertheless, the obtained ages for the zircons probably correspond to this granulite facies event. Evidence for a granulite facies metamorphism and a possible charnockitization comes also from the homogenization temperatures from primary CO2-fluid inclusions in sapphire suggesting temperatures of >630°C and pressures of 5.5 kbar, as reported from Balangoda by De Maesschalck and Oen (1989). The zircon in our sample from Balangoda displayed a small overgrowth that could not be dated, but could reflect a metamorphic event at the end of the granulite facies metamorphism or a post-metamorphic or post-tectonic event, which would make it younger than the dated core. The sapphires probably crystallized during or at the end of the granulite facies metamorphism.
CONCLUSIONS
The trace-element composition of the Sri Lankan sapphire samples, the presence of CO2 inclusions, the radial cracks around the zircon inclusions, and the obtained zircon ages of approximately 549 Ma all support a metamorphic origin related to granulite facies metamorphism or a later metamorphic or magmatic event. The zircons probably crystallized during or at the end of the Precambrian granulite facies metamorphism of Gondwana related to the Pan-African evolution. They are younger than the zircons in sapphire from the Pan-African rocks of Madagascar. The zircon probably formed when the sapphire crystallized or shortly before. The determined ages of the zircon inclusions coincide or nearly coincide with the age of the sapphire crystallization within the host rock.
Dating zircon inclusions in individual sapphires using high-resolution SIMS, TIMS, or SHRIMP methods, detailed cathodoluminescence imaging, and characterization by Raman spectroscopy does increase our understanding of sapphire formation. However, it would be desirable to study sapphire and its geological context at the extraction site as well as investigate the pressure-temperature-time (P-T-t) conditions of sapphire formation directly on sapphire occurrences in high-grade metamorphic rocks.
