Continuous Difusion Pb Loss in Igneous Zircon: A BSE, ID-TIMS and Raman Spectroscopy Study on a Mesoproterozoic Granite from the SW Amazonian Craton

This work presents a detailed investigation on zircon crystals of a Mesoproterozoic granite from the SW Amazonian Craton using a combination of back scattered electron (BSE) imaging, thermal ionisation mass spectrometry (TIMS) and Raman scattering spectroscopy. Six zircon grains were analysed. The results plotted in the Concordia diagram yielded an upper intercept age of 1423.0 ± 3.8 Ma, which is interpreted as the crystallization age of the rock. The scattering of the analyses in such diagram indicates that U-Pb isotopes are volume-dependent in the studied zircon grains. BSE reveals zonation in the zircon crystals, and Hf is the main element causing the variability in the BSE intensity and U has a secondary efect. Both elements have much higher atomic number than the principal constituents of zircon (Zr, Si and O), so the substitution of Zr by U and Hf results in increasing image brightness. BSE images from the zircon grains exhibit a euhedral external shape and ine-scale euhedral oscillatory zoning. The predominance of magmatic zoned crystals points out that the granite has not sufered metamorphic heating or any other processes that induce the formation of new phases. The results of the Raman scattering technique indicate that a SiO 4 tetrahedral internal vibrational structure is suiciently sensitive to determine the increasing degree of metamictization. Furthermore, evidence for metamictization is given by the heterogeneity of the half-width values of the Raman scattering peaks, which exhibit a decreasing trend from the core to the rims of the analysed grains. We concluded that the combination of the three techniques provides information on the Pb loss and that the degree of U-Pb isotopic discordance correlates closely with the volume of the zircon grain. The data, in addition to the lateral degree of metamictization detected by the Raman scattering technique, indicate that the loss of radiogenic Pb may be linked to the continuous difusion of this element. The correlation of the U-Pb discordance and metamictization emphasizes the importance of the Raman scattering spectroscopy analysis to performing zircon dating. Indeed, it is a helpful tool for geochronologists.


Introduction
The metamictization of zircon has been the subject of extensive research (Nasdala et al. 1996(Nasdala et al. , 1998a(Nasdala et al. , 1998b(Nasdala et al. , 1999Wopenka et al. 1996;Hartmann et al. 1997;Poller et al. 1997;Balan et al. 2001). Understanding how the zircon structure responds to radiation damage is important for both the storage of high-level radioactive waste (Ewing et al. 1995) and the use of U-Th-Pb isotopic systematics. In the latter case, fracturing due to the volume expansion that accompanies metamictization may lead to difuse loss or heterogeneous redistribution of radiogenic Pb (Pidgeon, 1992).
The metamictization of zircon can be characterised by Raman scattering spectroscopy (Knittle & Willians, 1993;Jollif et al. 1995). The Raman scattering method has advantages: little or no sample preparation is necessary, and the analysis can be performed in a reasonable amount of time. Particularly important is the ability of Raman spectroscopy to determine the crystallinity of micro-areas, allowing for the determination of such information by proiling across zircon grains (Jollif et al. 1995). In addition, BSE images provide complete information about the internal structures of zircon grains (Poller et al. 1997). As earlier investigations have demonstrated (Harchar & Miller, 1993), the parallel use of the BSE pre-investigating method allows for the inherited cores, desorbed areas and magmatic zonation to be distinguished. Thus, BSE is an important tool to solve the main problem involved in achieving successful zircon dating: the selection of the appropriate grains.
We present a detailed investigation on zircon grains from a Mesoproterozoic granite of the SW Amazonian Craton using a combination of back scattered electron (BSE) imaging, thermal ionisation mass spectrometry (TIMS) and Raman scattering spectroscopy. The objective of this work is to study the correlation between metamictization and the U-Pb isotopic discordance of zircon.

Procedures
For the U-Pb analyses, the separation of zircon grains was performed using the standard procedures. U-Pb zircon analyses were performed at the Isotope Geochemistry Laboratory (IGL), Department of Geology, University of Kansas (USA). The details of the procedures are found in Geraldes et al. (2001). A smaller magnetic fraction was abraded, and handpicked single grains were spiked with 205 Pb -235 U mixed tracer. The zircon grains were dissolved, and then, Pb and U were separated using procedures modiied after Krogh (1973, 1982) and Parrish (1987. The zircon weights varied from 0.002 to 0.018 mg. The isotopic ratios were measured using a VG Sector Multi-collector mass spectrometer in the single collector mode using a Daly detector. Table 1 exhibits the obtained analyses of the selected zircon crystals. Pb isotope compositions were analysed on single Re ilaments using silica gel and phosphoric acid. Uranium was loaded with Pb in the same ilament and analysed as UO 2 + . The amounts of radiogenic 208 Pb, 207 Pb, and 206 Pb were calculated using Pb laboratory blank correction (from 7 to 17 pg total Pb during the analyses) and for non-radiogenic common Pb corresponding to the model of Stacey & Krammers (1975) for the approximate age of the sample.
The theoretical difraction-limited spot diameter on the sample surface was approximately 1 μm for the Olympus 80 × long working distance objective. The laser power impinging on sample surface is 10 mW, which is typically non-destructive to samples of geologic interest. For BSE, 180 o backscattered light was detected with a 25.4 mm, 1024 element diode-array optical detector.

ID-TIMS Dating
U-Pb (single grain) zircon geochronology was performed on a granitic sample. The analysed zircon grains were clear, slightly caramel in colour, with rounded ends and short prismatic lateral faces. The studied grains present sizes varying from large, medium and small grains. Six zircon crystals were analysed, and the results, when plotted in the Concordia diagram, yielded an upper intercept age of 1423.0 ± 3.8 Ma ( Figure 1A), which may be interpreted as the crystallization age of the rock, as discussed below. Each one of the analysed zircon grains had a diferent weight, and when the analytical results were plotted in the Concordia diagram, the heaviest zircon grain (0.018 mg) was the most concordant (99%).
To understand the isotopic patterns of the grains, a second diagram was plotted ( Figure 1B) comparing Grain Weight (mg) versus 205 Pb/ 235 U Age Concordance (%) so that the results indicate that U-Pb isotopes are volume-dependent in the studied zircon grains.

BSE Imaging
BSE images reveal brightness contrasts related to the average atomic number of areas studied in a mineral: the higher the atomic number, the more electrons will be relected and the brighter that region will appear. BSE imaging is used for a variety of geologic studies and is recognised as a powerful tool for studying zonation in accessory minerals (Hanchar & Miller, 1993). In zircon, Hf is the main element responsible for the variability of the BSE intensity, with U having a secondary efect. Both elements have a much higher atomic number than the For BSE imaging and Raman scattering spectroscopy, zircon grains were hand-picked and prepared in a standard mount. To avoid the loss of crystals during the polishing process, only grains of the same size were mounted together using low luminescence resin. No special sample preparation is required for Raman scattering analysis, whereas the BSE technique requires a carbon coating on the polished surface. The zircon grains were analysed using a CCD T64000 Jobin-Yvon laser-Raman microprobe at the Fluid Inclusion Laboratory of the Geosciences Institute of the State University of Campinas (Brazil). The Ar-ion laser wavelength used in the experiments was 514.5 nm (the strongest green line).

Continuous Diffusion Pb Loss in Igneous Zircon: A BSE, ID-TIMS and Raman Spectroscopy Study on a Mesoproterozoic Granite from the SW Amazonian Craton Armando Dias Tavares Júnior; Mauro Cesar Geraldes; Anderson Costa dos Santos & Werlem Holanda dos Santos
principal constituents of zircon (Zr, Si and O); thus, the substitution of Zr by U and Hf results in increased brightness (Nicola et al. 1974).

Figure 2 shows zircon grains (BSE images)
from the granitic sample used in this study. The grains have a euhedral external shape and ine-scale euhedral oscillatory zoning. Magmatic zoned zircon crystals were found, indicating that the granitic samples have not sufered metamorphic heating or any other process resulting in new phases. Other than that, BSE images can provide further information, especially about inclusions showing they are frequently apparent in the sampled zircon crystals and might cause problems in performing U-Pb dating and Raman scattering analysis (Parkinson & Katayama, 1999).

Raman Scattering Monitoring
The Raman scattering spectrum of zircon is sensitive to its degree of metamictization. With decreasing crystallinity, the Raman bands become less intense (half-width peak variation) and increasingly broadened (area peak variation) and exhibit a shift to lower frequencies (shift of wavelength). In a synthetic single crystal of zircon, there are 36 irst-order Raman-active modes, which are subdivided into internal and external vibrations (Syme et al., 1977). External vibrations are related to massive silicate complexes (rotational and translational motions), and internal vibrations are related to silicon-oxygen and metal-silicate bonds and sites. We analysed the metamictization degree using the half-width of the Raman bands calculated in three diferent (modes) peaks: (Eg) 360 cm -1 , (A1g) 978 cm -1 and (B1g) 1012 cm -1 . The objective was to identify the best or the most sensitive Raman scattering parameter to metamictization.
Conventional Raman scattering spectroscopy analysis in zircon involves determining the peak shift, half-width and peak area. In this study we get, from the Raman scattering spectra, a plot of the intensity of the inelastically scattered radiation (measured in photons per second) as a function of the halfwidth in each analysed spot (Fialin et al., 1977). The mechanical coniguration of a high-resolution monochromator in combination with an argon ion laser as the excitation source enables the monitoring of the Raman scattering shift within the scale of the proile on the zircon surface.
In this way, the Raman scattering technique was used to obtain vibrational spectroscopic characteristics from the zircon lattice that is correlated to structural variations. The individual measurement spots can be seen in the backscattered electron images ( Figure 2). Three zircon grains were analysed with proiles through the zoning areas on the zircon surface. The results illustrated in Figure 3, indicate that the increasing width of the internal (B1g mode) vibration (anti-symmetric stretching of SiO 4 tetrahedral) is sensitive enough to determine the increasing degree of metamictization. Furthermore, evidence for metamictization is shown in Figure 4. These results indicate a marked heterogeneity of the halfwidth values of the represented three peaks, with a decreasing trending from the centre to the border in the three studied grains.

Discussion and Conclusions
We report the results of a combined ID-TIMS and Raman microprobe study in order to evaluate the correlation between metamictization and U-Pb isotopic discordance of zircon. The spatial resolution of the Raman scattering probe allows for a direct comparison of the U-Pb isotopic discordance and the metamictization of the zoned areas within a zircon crystal. The combination of the two measurements also provides information on the Pb loss. We found that the degree of U-Pb isotopic discordance is closely correlated with the volume of the zircon grain. The data, in addition to the lateral degree of metamictization detected by the Raman scattering technique, indicate that the loss of radiogenic Pb may be linked to continuous difusion of this element (Watson & Harrison, 1983;Levchenkov et al., 1998).
The correlation of U-Pb discordance and metamictization emphasizes the importance of the Raman scattering spectroscopy analysis to performing zircon dating. Indeed, it is a helpful tool for geochronologists. Because discordant isotopic ratios may provide ambiguous ages, it is advantageous to select zircon with well-ordered lattices for the age measurements. The results of the investigation here reported also suggest that a linear array of data points on the Concordia diagram is linked to the Pb loss. According to the literature (Krough, 1973), the concordance is directly proportional to the magnetic susceptibility of the zircon grain and also linked to metamictization.
Oscillatory zoning is a common feature in zircon from acidic igneous rock and is believed to form during zircon crystallisation from magma. The ID--TIMS and Raman scattering spectra herein presented above indicate that Pb difusivity in slightly damaged zircon is not negligible and that Pb loss is not restricted to zircon that have experienced episodic Pb loss (e.g., U-Pb mineral systems opened during a secondary event) or a protracted low-temperature (annealing) history (Gerbauer & Grunefelder, 1979). The analytical points deining a Discordia reported here ( Figure   1) corroborate the hypothesis in which the radiogenic lead was lost by continuous difusion (Cherniak et al., 1996(Cherniak et al., , 1997, and the upper intercept age closely approaches the age of the formation of the rock. Figure 3 Raman results for three zircon grain proiles (shown in Figure 2). Three peaks (360 cm -1 , 978 cm -1 and 1012 cm -1 ) were studied using half-width as a parameter. Metamictization evidence is given by a signiicant heterogeneity of the half-width values.