U-Pb zircon in situ dating with LA-MC-ICP-MS using a mixed detector configuration

,


INTRODUCTION
Accurate U-Pb ages have been obtained from zircons by isotope dilution thermal ionization mass spectrometry (ID-TIMS) (as described by, e.g., Bowring et al.1998, Mundil et al. 2001, 2004).The advent of in situ analyses at the microscale by laser ablation (LA)-ICP-MS and ion microprobe (e.g., SHRIMP) has shown the complexity of zircons grains, which often exhibit more than one crystallization phase associated with different geological processes.Of these methods, ID-TIMS presents more precise and accurate data when a single phase zircon is dated (i.e., associated with one main geological event).However, ion microprobe analyses (such as those provided by SHRIMP) have better spatial resolution for dating the different growth phases on single zircon grains, often producing U-Pb data as precise as the ID-TIMS method.
LA-ICP-MS is well accepted as a reliable and more convenient method of dating detrital zircons (provenance information applied to sedimentary basin FARID CHEMALE Jr et al. evolution), especially after the introduction of the 213 nm laser wavelength.This method allows the analysis of spots at the size conventionally used during the SHRIMP analysis, which is the best method for in situ age determinations.During the last decade, numerous studies were published presenting successful U-Pb zircon data using ICP-MS (e.g., Koesler andSylvester 2003, Jackson et al. 2004).The development of MC-ICP-MS configured with Faradays cup and multiplier ion counting channels has introduced the possibility of generating U-Pb zircon data that may be comparable to those using SHRIMP (Cocherie and Robert 2008).For the present work, the amounts of 238 U, 232 Th and 206 Pb were obtained with Faraday cups, while 202 Hg, 204 Pb, 204 Hg and 207 Pb were obtained with MIC (multiplier ion counting) channels because these isotopes occur in very low amounts in most zircons.To test the suitability of our system and the U-Pb age reproducibility of the Concordia ages, zircon samples covering a wide age range from 2.2 to 0.2 Ga with a large variation of 206 Pb/ 238 U, 232 Th/ 238 U and 207 Pb/ 206 Pb ratios were analyzed by both SHRIMP and LA-MC-ICPMS.The analyzed samples comprised the following: i) zircons from the early Paleozoic gabbrodiorite (Temora II sample provided by the Australian Geological Survey); ii) zircons from Paleoproterozoic Tandilla gneisses (provided by Orestes Santos); iii) Neoproterozoic Piquiri Syenite; and iv) zircon samples of the volcanic rock of the Rincon Blanco basin.

ANALYTICAL PROCEDURES
Zircons were separated by conventional procedures using heavy liquids and magnetic separator after the concentration by hand panning.The most clear and inclusion-free zircons from the least magnetic fractions were hand picked for U-Pb SHRIMP and LA-MC-ICP-MS analyses.SHRIMP U-Pb SHRIMP zircon geochronology was carried out at the Research School of Earth Sciences, Australian National University, and at the Department of Geology and Geophysics, University of Western Australia by using SHRIMP II and RG equipments.Handpicked zircons were mounted in epoxy discs along with zircon standards, ground and polished, microphotographed in transmitted and reflected light, and and their internal zoning imaged by cathodoluminescence (CL) using scanning electron microscope.The mounts were then cleaned and gold-coated in preparation for the SHRIMP analysis.Analytical methods and data treatment can be found elsewhere (Compston et al. 1984, Williams 1998).Zircons grains were analyzed with a 2-3nA, 10kV primary O 2 -beam focused to a ~ 25 to ~20μm diameter spot.At mass resolution ~ 5500 the Pb, Th and U isotopes were resolved from all major interferences.The reduction of raw data and age calculation were carried out using Squid 2.02 and Isoplot-Ex (Ludwig 2003).U and Th concentrations were determined relative to those measured in the RSES standard SL13

LA-MC-ICP-MS
The same samples used for SHRIMP analyses were dated by LA-MC-ICP-MS with the New Wave UP213 Laser Ablation Microprobe coupled to a MC-ICP-MS Neptune at the Isotope Geology Laboratory of the Rio Grande do Sul University.We have tried to do the spot with the same size or somewhat larger and at the same site or close as regarding those did with SHRIMP.In some cases, due to the low content of Pb, either larger spots were used or increased the number of spots (in the same zircon phase) in order to get more precise data with the LAM method (see discussion of results).Isotope data were acquired in static mode with spot sizes of 25 and 40 µm.Laser-induced elemental fractional and instrumental mass discrimination were corrected using the reference zircon (GJ-1) U-PB ZIRCON DATING WITH LA-MC-ICPMS (Jackson et al. 2004).Two GJ-1 analyses were measured after every four or ten sample zircon spots.The external error was calculated after the propagation error of the GJ-1 mean and the individual sample zircon (or spot).

Collector configuration
The collector configuration used for simultaneous measurements of Th, U, Pb and Hg isotopes is as follows: The gain calibration of Faraday cups and the cross calibration between the L4 cup against the MIC3, 4 and 6 were carried out before the laser running was started.The MIC3 to 4 are attached to the L4 Faraday cup, and the MIC6 is attached to the L3.
Because the multicollector system involves 3 ion counters and 5 conventional Faraday collectors, the gain and cross calibration must be performed routinely.The gain factor used to calibrate the Faraday measurements is calculated by applying a constant signal of 33.0 volts.The cross calibration used to calculate the necessary conversion factors (voltage to cps) is achieved using a 220 ppt Neptune solution with the addition of 200 ppt Th and an efficient nebulizer system.A calculated conversion value of 62,500 cps/mV was used.
The various ratios are obtained simultaneously and appropriately corrected.However, because of inherent elemental and isotopic fractionations during laser ablation, these ratios vary during the analysis and require different approaches to estimate reliable data.As illustrated for the standard zircon, the 207 Pb/ 206 Pb ratios do not fractionate visibly like the 206 Pb/ 238 U ratios, which involve two different elements with their own chemical and physical properties (Fig. 1).Pb is more volatile than U, which condenses progressively on the walls of the pit formed during the laser ablation process.We routinely adopt the average of the 207 Pb/ 206 Pb determinations as the representative value for the sample, and for 206 Pb/ 238 U, we assume the extrapolated value for t (time) =0.Outliers that do not show a good alignment are also discarded.Other ratios, such as 206 Pb/ 207 Pb and 232 Th/ 238 U, are also taken into account in the extrapolated ratios when they are applicable or exhibit the same trend of fractionation.These ratios are usually quite close to the expected values.
The conversion factors are calculated based on the available data for the standard used and applied to unknown samples.Thus, a homogeneous standard is of paramount importance.The GJ-1 standard (GEMOC ARC National Key Center) meets the requirements for the methods used in our laboratory, and the ratios of 206 Pb*/ 238 U, 207 Pb*/ 206 Pb* and 232 Th/ 238 U are homogeneous during the entire "bracket" technique, a standardsamples-standard analysis.

Mass bias correction, external correction, and laser conditions
The isotope ratios and inter-element fractionation of data obtained by the MC-ICP-MS instrument were evaluated by interspersing the GJ-1 standard zircon on every set of 4,6,8  depending on the zircon homogeneity and the amount of Pb and U in the zircon.The GJ-1 standard zircon was used to estimate the necessary corrections for the external corrections and the internal instrumental fractionation.The GJ-1 zircon and sample were assembled in the same mounting.The spot size of the laser was usually 25 µm, but the spot sizes were 40 µm and 15 µm for the zircon phases with a low amount of 207 Pb (under 10,000 cps) and for small zircon grains (<30µm of diameter), respectively.
The repetition rate of the laser was 10 Hz.The energy varied from 0.3 to 1.1 mJ/pulse, and the corresponding spot sizes varied from 15 µm to 40 µm.The data acquisition occurred in 50 cycles of 1.048 s of integration time, and the masses 202, 204, 206, 207, 208, 232, and 238 were collected simultaneously.For every standard and sample set, blank values in the same conditions as the standard and sample were also measured.The average blank values were subtracted from all individual cycle measurements.The 204 Pb value was corrected for 204 Hg by assuming a 202 Hg/ 204 Hg ratio equal to 4.355.Laser operation conditions are shown in Table I.

Common Pb correction
The usual method for common-lead corrections on zircon grains (based on the non-radiogenic 204Pb isotope) is not appropriate when using the laser technique because the 204 Pb signal is strongly affected by 204 Hg.The majority of the 204 Hg comes from gases (Ar and He) that are required in the ICP and ablation procedures.After the Hg correction based on 202 Hg is measured, the common 204 Pb is insignificant in most situations.For instance, a typical signal intensity of the 204 Hg during laser ablation of the standard zircon is in the 600-1000 cps range, and the calculated count rate for 204Pb is less than the statistical error of ca.25-33 cps.We assume that the 204 Pb values obtained from zircon grains contain some common Pb, and we also assume a concordant age of 206   For the common lead isotope composition, we assume the isotope compositions evolve as proposed by Stacey and Kramers (1975).This assumption is required to determine an initial estimated age.
The 207 Pb/ 206 Pb and 206 Pb/ 238 U ratios were corrected after the f 206 and f 207 were determined for each cycle.The cycles with values of f 206 above 0.0025 are not usually included in the age calculation.

Calculation of the ratios and error estimation
After the blank and common Pb corrections, the ratios and their absolute errors (one sigma level) of 206 Pb*/ 238 U, 232 Th/ 238 U, and 206 Pb*/ 207 Pb* were calculated using an Excel sheet.Because the 206 Pb*/ 238 U usually produces a linear fractionation, we used the intercept method for laser-induced Pb/U fractionation to correct the ratio according to the formulation proposed by Youden (1951) and adopted by Koesler et al. (2002).The uncertainty of the fractionation-corrected ratio was calculated as one SD (standard deviation) of the intercept (σR(o)), which is the isotope ratio at the start of laser ablation.The internal derived errors were calculated in the conventional way by taking account of the uncertainties (1 SD) of the respective background signals (see Fig. 1).
For the 232 Th/ 238 U and 207 Pb*/ 206 Pb* ratios, the mean values were used after discarding the outliers.
In some cases, the 232 Th/ 238 U and 207 Pb*/ 206 Pb* ratios show a slight fractionation.Laser-induced fractionation was applied to obtain the R(o) of these ratios.

RESULTS
After reduction, the raw U-Pb data and the calculated 206 Pb*/ 238 U, 232 Th/ 238 U and 206 Pb*/ 207 Pb* ratios, along with the U-Pb Concordia diagrams, were obtained using Isoplot version EX (Ludwig 2003).We used the Tera-Wasserburg Concordia diagram (Tera and Wasserburg 1972) for the Phanerozoic to late Neoproterozoic rocks and the normal Concordia diagram for Precambrian samples.A summary of the results is presented in Table II.
PALEOPROTEROZOIC TANDILLHA BASEMENT Two samples were dated where the spots were of the same size and at the same site in both LA-MC-ICPMS and SHRIMP (Figures 2 and  3, Tables III and IV).The SHRIMP data have already been presented in Hartmann et al. (2002).The obtained ages for sample TA-01, a tonalitic gneiss outcropping in the Tandilla region, are 2,239±18 Ma (SHRIMP) and 2,243±20 Ma (LA-MC-ICP-MS).The younger syenogranite, which cuts the tonalitic gneiss (TA-01), presents ages of 2,114±12 Ma (SHRIMP) and 2,112+38/-28 Ma (LA-MC-ICPMS).In both samples, the   results have similar accuracy and precision, but the individual analyses obtained by SHRIMP have lower uncertainties (Fig. 4).The youngest rocks presented two zircons with greater errors, which were attributed to the unstable analytical conditions of the laser during the analyses.Figure 5 shows the results of the GJ-1 standard and zircon spot TA-1 (4-1) before the application of the conversion factor and error propagation.
NEOPROTEROZOIC PIQUIRI SYENITE The Piquiri Syenite, located in the Sul-Rio-Grandense Shield in southernmost Brazil, has two zircon phases.The youngest zircon phase, interpreted to be the igneous crystallization of the Piquiri Syenite, was dated by LA-MC-ICP-MS at 592±2.3 Ma and by SHRIMP at 594±2.7 (Fig. 6 and Table V).Although both ages are in the    error bar, the most reliable age determination for this rock is the 594 Ma obtained by SHRIMP, as suggested by Babinski et al. (1997).We recognize two zircon phases in the Piquiri Syenite zircons (Fig. 7), where the dated phase is the younger igneous age and interpreted to be the magmatism age of the studied syenitic body.
TEMORA GRABBRODIORITE Zircons grains of the Temora II, a gabbroic diorite (Australia), have been dated by the LA-MC-ICP-MS method (Table VI).Because the crystallization age of this unit is well constrained by Black et al. (2004) at 416.5±0.22Ma, the obtained U-Pb ratios yielded an      age of 418.6±4.3Ma (Fig. 8), which is close to the ID-TIMS data.However, because the zircons of Temora II have little Pb207, we had to use a spot size of 40 µm to obtain reliable isotopic data.The dated zircon grains are highly homogeneous and inclusion-free.

TRIASSIC VOLCANIC ROCKS
A rhyolite (sample RB-06) is situated at the basal portion of the Rincon Blanco Basin and located in the Precordillera fold and thrust belt at 31°24' -31° 33' south.A dating of the youngest zircons of this sample yielded ages of 244.1±3.5 Ma (LA-MC-ICP-MS) and 246.4 ±1.1 Ma (SHRIMP, Barredo et al. 2012) (Fig. 9 and tables VII and VIII), which are coeval to those basal rhyolites of the Cuyo Basin (Ávila et al. 2006), a contemporaneous volcano-sedimentary basin of the Triassic Intracontinental rifting exposed in the Precordillera region, Argentina.
The Triassic sample was mostly dated with a 40 µm spot size with the laser method because the 207 Pb* content was lower than 5,000 cps using a spot size of 25 µm.In this case, the analytical data presents high statistical errors, and the ratios cannot be estimated precisely.Figures 10 and 9 present Cl, BSE and reflected light (RE) images of dated zircons as well as the calculated individual ages for the spots obtained by SHRIMP and Laser.It is noteworthy that the age errors by the Laser method are larger than those obtained by the SHRIMP method for the same zircon phases (see Figure 10).

DISCUSSION
The analytical strategies using laser ablation coupled to an ICP-MS instrument vary according to certain parameters that must be taken into account in the available system.As a result,  we have no "best" parameters to date zircons.
The basic principles in such systems are a good control of the spot size (using CI, BSE and Rl images), the quality of the measurements and adequate fractionation (elemental and isotopic) corrections.Another important feature is to select with well-established criteria the ratios needed to obtain the mean or extrapolated parameters.
The mixed configuration with MC-ICP-MS (Neptune) and the coupled Laser Microprobe (New Wave 213 nm) have some advantages for dating zircons of different ages.The 207 Pb mass is therefore collected with a MIC channel, while the 206 Pb, 232 Th and 238 U are collected in the Faraday cups because the amount of 207 Pb* in the zircons, especially those of Neoproterozoic to Phanerozoic ages, is too low to be collected with Faraday cups with the laser conditions applied in this work.However, the obtained values for the 206 Pb, 232 Th and 238 U are from some mV up to one hundred mV, and uncertainties are less than 1 % for the obtained ratios.In addition, the calculated ratios for 238 U/ 207 Pb for zircons formed from 1.0 Ga to 20 Ma vary from 80 to 666.67, which can introduce large uncertainties if we analyze 238 and 207 with the same type of ion collectors.To obtain more precise ratios, we used the suggested mixed configuration in this present work.In the case in which the amount of 207 Pb was to high (over 15 mV or 1,000,000 cps), we decreased the size of the spot (to 15 µm) or the energy of the laser to maintain the same configuration.Alternatively, we can use different configurations of mixed collectors (masses 202, 204, 206 and 207 in MIC channels and mass 238 in Faraday, Buhn et al. 2009) or with the same type of collector (only MIC's channels or Faraday cups).
The present sample dating supports the idea that the LA-MC-ICP-MS method presented here can also be used for dating igneous rocks.We found that the results are somewhat comparable to the SHRIMP data (see Table I).In broad terms, SHRIMP is more precise and has a greater accuracy because it does not have the strong fractionation provoked by the laser (e.g., U and Pb isotopes).Also, there is no Hg interference, and it has a much higher resolution (MC-ICP-MS operated here at low resolution (800), while SHRIMP was routinely up to 5,000).Thus, the greater reliability of the SHRIMP data is also due to the size of the analyzed particle because SHRIMP produces a hole no more than 3 µm deep for 15 minutes of analysis, while the laser goes up to 15 µm or more in ca.50 seconds.Therefore, the particle size produced by the laser is too large, producing in most cases some variation of the obtained signal, which certainly increases the error estimation and the age of the analyzed zircon (spot).If the laser could produce smaller particles, the obtained data would be much better.This goal may be attainable in the near future using the Fenton Laser (Koesler andSylvester 2003, Cocherie andRobert 2008).
The presence of large amounts of common Pb was identified either with SHRIMP or with LA-MC-ICP-MS.In this case, the spot data cannot be used.The presence of common Pb is mostly related to metamictic zircon or the mounting resin.In some cases, the laser produces some instability that results in very few usable cycles for the spots, which makes it difficult to obtain a single spot age and the acquired data not useful at all.Therefore, for dating igneous rocks, it is necessary to have a larger number of zircon spots to obtain comparable results with SHRIMP.Based on our results and comparing the SHRIMP and LA-MC-ICPMS analytical data, we observe that the individual error obtained by LA-MC-ICPMS is usually several times larger (3 to 4 times) than those obtained by the SHRIMP method.It is also emphasized that the calculated age based on LA-MC-ICPMS presents an uncertainty over 1 % (for Phanerozoic rocks, it can be 2% or more), whereas the SHRIMP age uncertainties are less than 1%.This assumption is valid for homogeneous zircon, as we used in this study.FARID CHEMALE Jr et al.
The GJ-1 zircon was a very suitable standard with relatively high U content and homogeneous U-Th/Pb ratios, as already presented by Jackson et al. (2004).In our case, we used a large crystal (ca.36 mm2) mounted in a small cylinder (8 mm diameter) inserted in the sample mounting which provides consistent analytical data during a session day, even analyzing with 15 µm and 25 µm spot sizes in the same analytical conditions.In the case of the grain size of the zircon standard, like those used for SHRIMP (Temora II, FC1), the resulting data are much more variable during the Laser analyses due to the small size of the zircon crystal.

CONCLUSIONS
The analytical data obtained by the LA-MC-ICPMS method using a mixed collector configuration are here compared to results of the SHRIMP method.
The analyzed in situ zircons are from igneous rock samples formed at 2.2 Ga, 2.1 Ga, 0.59, 0.41 and 0.24 Ga, and the data obtained by both methods are in agreement when the analytical errors are taken into account.The same spot sites or zircon phases were analyzed for both methods, which is a strong argument for the age comparison between both methods.
The spot sizes for both methods are almost the same, but the calculated individual age errors for each spot are 2 to 4 times greater for the LA-MC-ICPMS method when compared to those of the SHRIMP method.The SHRIMP method is therefore a more accurate and precise method, especially due to the flatter spots (~ 3 µm = SHRIMP and ~ 15 =µm for LA-MC-ICPMS) and higher resolution.
The use of masses 202, 204, and 207 in the MIC channels, and 206, 208, and 238 in the Faraday cups are the best configuration for U-Pb zircon dating of samples formed in the Phanerozoic or with small amounts of radiogenic 207 Pb.One of the main problems of U-Pb zircon dating by LA-ICPMS methods is the common Pb correction, so it is recommended that those cycles with 206Pb/Pb204 ratios < 1,000 would be excluded to reduce data.
In this study, the GJ-1 also proved to be a suitable zircon standard for the laser ablation studies.
The advances of LA-MC-ICPMS will be mainly associated with the improvement of the Laser Microprobes (e.g., Fenton Laser), as well as the improvement of MIC channels or similar devices for acquiring small amounts of isotopes with a better precision.

Figure 1 .
Figure 1.Set of two standard measurements and 8 zircon spots of the AB-06 tuff a) Plot of 207 Pb*/ 206 Pb* and 206 Pb*/ 238 U ratios obtained on GJ-1 standard zircon; b) Summary of calculated results on standards; c) Plot of 207Pb*/206Pb* and 206Pb*/238U ratios obtained on 8 zircons (AB-06 tuff zircons) in the same analytical conditions as compared to those for GJ-1 standard zircon; d) Concordia age of the analyzed zircons on Tera-Wasserburg diagram.

Figure 8 .
Figure 8. Concordia diagrams with LA-MC-ICP-MS results for sample of Temora II (Australia).

TABLE I MC-ICP-MS and Laser operating conditions
. U-PB ZIRCON DATING WITH LA-MC-ICPMS

TABLE II Summary of U-Pb zircon age (Ma) of the analyzed samples. * Forced at origin (0±50 Ma), ** Age and error after Black et al
. (2004).FARID CHEMALE Jr et al.

TABLE V (
continuation) U-PB ZIRCON DATING WITH LA-MC-ICPMS