Geochemistry, Geophysics, Geosystems

Routine low-damage apatite U-Pb dating using laser ablation–multicollector–ICPMS

Authors


Abstract

[1] Apatite is a common U-bearing accessory mineral with a U-Pb closure temperature of ∼500°C, making U-Pb dating of apatite a potentially valuable thermochronometer. However, its low U concentration and tendency to incorporate common lead has limited widespread application to destructive isotope dilution methods. We overcome previous limitations by using a Nu Plasma multicollector ICPMS with an attached short-pulse excimer laser, and by identifying two new matrix-matched reference apatites to correct for elemental fractionation: gem-quality 485 Myr old apatite from Madagascar which we independently characterized by ID-TIMS analysis, and 523.5 Ma apatite from the McClure Mountain syenite (source of the40Ar/39Ar reference MMhb). Common Pb is corrected using measured 204Pb isobarically corrected for Hg interference and a five-step iterative process using Stacey and Kramers' common Pb model. We accurately reproduce ages of numerous independently characterized apatites, regularly achieving precision of <2% (2σ) by pooling as few as five 30 μm spot analyses. Data quality in apatite with low U concentrations, low 206Pb/204Pb values (<∼30) and young ages (<∼75 Ma) is compromised by the goal of avoiding significant grain damage. Such limitations can be overcome by using spot sizes 65 μm or greater, but at the expense of substantial grain damage. For single detrital apatite grains with ages of ∼500 Ma, precision of <4% (2σ) was achieved by pooling 2 to 3 spots per grain. The accuracy of our detrital results is supported by a good age match with similar closure temperature 40Ar/39Ar detrital hornblende ages from the same sediment.

1. Introduction

[2] Apatite is a common U-bearing accessory mineral in magmatic, sedimentary, and metamorphic rocks, with a U-Pb closure temperature of 450–550°C [Cherniak et al., 1991] making U-Pb dating of apatite a potential widely applicable medium-temperature thermochronometer and a complement to the similar closure temperature40Ar/39Ar hornblende system [Harrison, 1982], especially in rocks where hornblende is absent. However, the low U concentration in apatite, its tendency to incorporate high amounts of common lead during crystallization, and the lack of good matrix-matched apatite reference material to monitor fractionation has previously limited the accuracy and hence widespread application of this method. The relatively low durability of detrital apatite during sedimentary transport means that U-Pb detrital apatite ages tend to represent first cycle provenance ages from the immediate source catchment, in contrast to the greater potential for polycyclic behavior of highly resistant detrital zircon. The ability to routinely apply detrital apatite U-Pb dating thus constitutes a highly complementary provenance tool to supplement routine U-Pb dating of detrital zircon from the same sediment. Our main motivations for exploring low-damage laser ablation U-Pb dating of apatite is to determine U-Pb ages on detrital apatite previously dated using the fission track dating method, and after U-Pb dating, to further analyze the same grains by fully destructive (U-Th)/He low temperature thermochronometry – hence producing three different ages using three different methods on the same detrital grain – a method we refer to as single grain triple dating. To avoid significant grain damage to allow further analysis involves some compromise regarding achievable maximum precision. The need to use small laser spot size thus has some limitations when applied to apatite of very young age and low uranium concentration. However, these limitations are outweighed by the possibility to obtain single-grain triple dates from detrital apatite to open up new frontiers, including high resolution provenance of immature modern and ancient sediments, and the remote study of the long-term geologic record of catchment erosion and landscape evolution.

2. Challenges to Reliable U-Pb Dating of Apatite

[3] U-Pb dating of apatite is made difficult by its generally low U concentrations, and incorporation of common lead during formation. This has previously limited successful and accurate apatite U-Pb dating to fully destructive isotope dilution methods [e.g.,Oosthuyzen and Burger, 1973; Krogstad and Walker, 1994; Bosch et al., 1996; Berger and Braun, 1997; Corfu and Stone, 1998; Amelin et al., 1999; Chamberlain and Bowring, 2001; Amelin and Zaitsev, 2002; Schoene and Bowring, 2006]. More recently attempts have been made to apply in situ SHRIMP, SIMS, and laser ablation (LA) ICPMS U-Pb techniques on apatite [e.g.,Sano et al., 1999; Willigers et al., 2002; Terada and Sano, 2003; Nishizawa et al., 2004; Sano et al., 2006; Storey et al., 2007; Carrapa et al., 2009; Nemchin et al., 2009; Chew et al., 2011]. However, in situ apatite U-Pb techniques have been hindered by the lack of well-characterized matrix-matched reference material to correct for elemental fractionation, as well as by the difficulty in accurately and precisely measuring204Pb to provide a robust common lead correction that does not rely on assumption of concordance.

3. Analytical Methods

[4] To overcome these limitations we have identified two new well-characterized matrix-matched reference apatites to correct for elemental fractionation and directly measure204Pb isobarically corrected for 202Hg to allow for common Pb correction without the assumption of concordance inherent in some other recent approaches to apatite LA-ICPMS dating [e.g.,Chew et al., 2011].

[5] Analyses were conducted at the Arizona LaserChron Center using a Nu Plasma HR multicollector ICPMS with either an attached New Wave UP193HE platform with a New Wave SuperCell and retrofitted ATL-Lasertechnik 4 ns short pulse 193 nm excimer laser, or for later analyses using larger spot sizes using a newly acquired Photon Machines Analyte G2 laser ablation platform equipped with HelEx ablation chamber that uses the same ATL-Lasertechnik 4 ns short pulse 193 nm excimer laser source. We employ the same mass spectrometer and laser set-up used for routine zircon analysis [Gehrels et al., 2008] (www.laserchron.org).

[6] Samples are prepared by mounting both unknowns and reference apatite in an 1″ epoxy plug. We include several chips of a primary reference apatite for elemental mass fractionation correction along with some secondary reference apatite. A small amount of Sri Lanka reference zircon is also included to check machine performance and tuning. The surface of the epoxy plug is polished down with 2500 grit sandpaper, followed by a final polish with 9 micron lapping film. Distilled water and alcohol are used final sample cleaning as dilute acid washes used for zircon etch and damage the apatite. Material is ablated from the surface of a polished apatite using a standard laser spot size of 30 μm for 15 s with a fluence of ∼4 J/cm2, a pulse frequency of 7 Hz, and pulse width of ∼4 ns (same for both laser platforms). Initial analyses using the newer HelEx ablation chamber show an approximately 20% improvement in Faraday collector signal, all other conditions being equal. Measurement of pit depth in grains plucked from the epoxy mount show excavation depths are ∼20 microns deep, compared to ∼15 micron deep pits in zircon under the same lasing conditions. The ablated material is carried in helium into the plasma source of the Nu HR MC-ICPMS, which is equipped with a flight tube of sufficient width that U, Th, and Pb isotopes are measured simultaneously. All measurements are made in static mode, using Faraday detectors with 3 × 1011 ohm resistors for 238U (Ex-Hi),232Th (H2), 208Pb (L6), 207Pb (L7), 206Pb (L8) and discrete dynode ion counters for 204Pb (IC0) and 202Hg (IC2). Ion yields are ∼0.8 mV per ppm.

4. Data Acquisition and Reduction

[7] Initial tuning the ICPMS uses a solution containing 7 ppb U, 5 ppb Th, and 0.1 ppb (for Faradays) or 0.01 ppb (for ion counters) radiogenic Pb (NBS SRM 983) introduced via a Nu DSN-100 desolvating nebulizer. This solution generates a sensitivity of ∼300 to 500 V/ppm for238U. When the instrument settings are optimized, the DSN line is replaced by a line that contains a mixture of He carrier gas (0.38 L/min) and Ar make-up gas (0.90 L/min). Optimum instrument settings are then checked by ablating an in-house Sri Lanka reference zircon. Values that are monitored include appropriate sensitivity (for apatite this is ∼50–60 kcps/ppm of U for a 30μm beam and ∼1.3 μm/sec ablation rate), 206Pb/204Pb value, 204Pb+204Hg background intensity (∼150 cps), 206Pb/238U age, and 206Pb/207Pb age. Data are acquired in two steps during each analysis: an initial measurement of backgrounds with the laser off, followed by measurement of peak intensities with the laser firing. Both acquisitions are the same duration of 15 s. Data reduction takes place off-line using raw count data imported into an Excel spreadsheet. Data are first corrected for background and any excess204Hg. 204Hg interference with 204Pb is accounted for by measurement of 202Hg during laser ablation and subtraction of 204Hg according to the natural 202Hg/204Hg of 4.35. This Hg correction is not significant for most analyses because our Hg backgrounds are low (generally ∼150 cps at mass 204). Data are then corrected for downhole laser fractionation, elemental fractionation, and common Pb correction (each of these is dealt with separately below).

[8] Uncertainties are accounted for and propagated separately depending on whether they are relevant for an individual analysis (internal errors) or a set of analyses (external errors). Internal error components for 206Pb*/238U, 208Pb*/232Th, and 206Pb*/207Pb* measurement include either the uncertainty returned for the downhole corrected ratio, or measurement uncertainty where appropriate and the uncertainty in the ratios that results from the measurement of 206Pb/204Pb. The latter adds uncertainty through the common Pb correction, and is different for each analysis because the composition of initial Pb is unknown. Four main external errors are incorporated into a final age uncertainty. In order of decreasing significance these are the elemental fractionation correction, common Pb composition, age of the primary reference material and the small uncertainty in the decay constants of 238U, 235U, and 232Th [Steiger and Jäger, 1977]. Final data reduction and calculation of concordia ages, isochron ages, weighted mean ages, and age probability density plots is done using Isoplot [Ludwig, 2003].

4.1. Downhole Laser Fractionation

[9] Typical apatite downhole laser fractionation trends using a polished epoxy mount are shown in Figure 1 for both 30 micron and 65 micron spot size using the New Wave and Photon Machine lasers, respectively. In contrast to zircon, most apatite shows either a slightly decreasing (30 micron spot) or flat (65 micron spot) 206Pb/238U ratio during downhole lasing. To account for any change we regress only the last 10 values to determine the 206Pb/238U ratio at integration number 6. For 206Pb/207Pb and 206Pb/204Pb the average of the one second integrations is used, with a 2σ filter removing any outliers. Note that the higher 207Pb signal acquired when using the larger 65 micron spot results in significantly reduced run-time scatter in206Pb/207Pb ratio (see section 6.7).

Figure 1.

Plots showing 15 s apatite laser downhole U and Pb fractionation for (a–c) 30 μm diameter beam using the New Wave 193 nm excimer laser with New Wave SuperCell and (d–f) for 65 μm beam using Photon Machines Analyte G2 193 nm excimer laser equipped with HelEx ablation chamber. The 206Pb/238U ratio the last 10 values are regressed through the sixth integration. For 206Pb/207Pb and 206Pb/204Pb the average of the one second integrations is used, with a 2σ filter removing any outliers.

4.2. Elemental Fractionation Correction Using Matrix-Matched Reference Material

[10] One of the most significant limitations to using in situ LA-ICMPS for U-Pb dating of apatite has been the lack of a well characterized matrix-matched reference to correct for the differential fractionation of U, Th, and Pb during laser ablation. Finding a suitable reference material is made difficult by the tendency of apatite to have low and variable U (and hence radiogenic Pb) concentration, variable common Pb, and to lose Pb by thermally induced diffusion at temperatures > ca. 500°C. Despite this, we have indentified two reliable natural reference apatite samples: gem rough apatite from Madagascar (details below) and apatite from the McClure Mountain syenite, Colorado [Schoene and Bowring, 2006]. During analysis we employ a bracketing approach (five reference apatite measurements at the start, one between every four or five unknowns, and three at the end) using the exact same lasing conditions and spot size for both reference apatite and unknowns. During each run we also obtain data from a secondary reference apatite (usually McClure Mountain) to test the run accuracy, reproducibility, and concordance when using the primary reference (usually Madagascar) for elemental mass fractionation correction.

4.2.1. Madagascar Fractionation Reference Apatite (MAD Apatite)

[11] In our exploration of potential apatite to correct for elemental fractionation, blue/green gem roughs of Madagascan apatite from the “1st Mine Discovery” showed favorable very low intragrain variability in U-Pb ratios. The apatite we employ was supplied as gem quality ca. 1 cm sized crystals taken from larger grains. We report ID-TIMS U-Pb data from randomly chosen mm-sized shards taken from two larger fragments of the MAD apatite as part of the NSF EARTHTIME initiative (www.earth-time.org) using the facilities at MIT (Table 1 and Figure 2) following experimental procedures outlined by Schoene and Bowring [2006]. The smaller shards from each larger MAD crystal show excellent internal consistency in ID-TIMS U-Pb age and206Pb/204Pb ratio. The data show some slight (ca. 1%) discordance most likely reflecting the common Pb correction being dependent of Stacey and Kramers' [1975] common lead evolution model rather than the true value. However, the quoted 206Pb/238U weighted mean ages are insensitive to the correction used. The weighted mean 206Pb/238U age for each crystal (MAD1 and MAD2) differ by about 11 Ma. One explanation for this age difference is that variable Pb diffusion occurred during cooling through ca. 500°C [Cherniak et al., 1991] owing to the dated crystal fragments being derived from different sized whole crystals. The sample batch as a whole is not ideal for distribution unless independent ID-TIMS U-Pb characterization is performed on each crystal fragment individually. Nevertheless, a limited supply of uncalibrated crystal fragments from the same batch of Madagascar apatite are available by contacting the first author.

Figure 2.

ID-TIMS U-Pb ages from two randomly selected chips from same batch of gem rough Madagascar apatite showing weighted average206Pb/238U ages and concordia plots. Data represented by the gray ellipses in the concordia plots were not used in the weighted mean age calculations.

Table 1. ID-TIMS U-Pb Isotopic Data for Randomly Chosen Chips From Two Larger Fragments of 807 Madagascar Apatite (MAD1 and MAD2)a
FractionCompositionIsotopic RatiosDates (Ma)
Th/UbPb*c (pg)Pbcd (pg)Pb*/Pbce206Pb/204Pbf206Pb/238Ug±2σ (%)207Pb/235Ug±2σ (%)207Pb/206Pbg±2σ (%)206Pb/238Uh±2σ (abs.)207Pb/235Uh±2σ (abs.)207Pb/206Pbh±2σ (abs.)Corr. Coef.
  • a

    Blank composition: 206Pb/204Pb = 18.24 ± 0.21; 207Pb/204Pb = 15.34 ± 0.16; 208Pb/204Pb = 37.35 ± 0.20. Corr. Coef., correlation coefficient.

  • b

    Th contents calculated from radiogenic 208Pb and the 207Pb/206Pb date of the sample, assuming concordance between U-Th and Pb systems.

  • c

    Total mass of radiogenic Pb.

  • d

    Total mass of common Pb.

  • e

    Ratio of radiogenic Pb (including 208Pb) to common Pb.

  • f

    Measured ratio corrected for fractionation and spike contribution only.

  • g

    Measured ratios corrected for fractionation, tracer, blank and initial common Pb.

  • h

    Isotopic dates calculated using the decay constants λ238= 1.55125E-10 andλ235= 9.8485E-10 [Jaffey et al., 1971].

  • i

    These analyses not included in weighted mean calculation.

MAD1 (Apatite)
A126.00106.015.637710.078470.6500.61010.000.0564009.7487.03.0484.038.0468.045.40.50
A2i26.11194.011.94161440.077870.2700.6234.000.0580003.9483.41.2492.016.0531.020.70.51
A426.00187.012.02161390.078390.2800.6314.100.0584004.0486.51.3497.016.0544.021.80.50
A526.03197.011.34171530.078370.4900.6273.800.0580003.6486.42.3494.015.0531.019.10.35
A626.05183.910.82171350.077950.8000.6293.900.0582403.7483.93.9495.319.3538.819.90.50
A726.07182.611.41161480.078460.4100.6204.100.0578943.5486.92.0489.620.1525.618.40.49
A8i26.10194.610.24191550.077660.3560.6143.700.0581203.9482.11.7486.218.0534.320.80.78
A926.04196.911.58171380.078560.5500.6304.400.0579453.8487.52.7496.021.8527.920.10.63
 
MAD2 (Apatite)
A126.34200.23.87524170.076220.1520.6131.400.0583061.307473.50.7485.35.4541.328.60.67
A226.56422.89.55443560.076470.2510.6101.520.0578741.447475.01.1483.75.8525.031.70.35
A326.99125.65.71221840.076120.5660.5853.420.0557213.266472.92.6467.512.8441.372.70.36
A526.8753.42.09262120.076460.1960.5942.340.0563852.248475.00.9473.78.9467.649.80.50
A626.55135.74.10332540.076330.1620.6011.460.0561652.114474.20.8477.97.0459.146.20.56
A726.49249.06.43393250.076410.1590.5921.350.0571561.986474.60.8472.46.4497.743.20.46
A826.35287.28.17354010.076250.2560.6011.990.0553152.895473.71.2477.69.5488.462.60.52
A926.78147.86.12243570.076350.1550.6062.090.5831551.521474.30.7481.010.1541.632.70.42
A1026.85321.05.81554020.076320.1850.6121.560.0561153.013474.10.9484.77.6457.165.50.40

[12] During a typical apatite U-Pb laser ablation run using MAD1 apatite as the primary fractionation reference, very consistent self-normalized data is obtained (∼1% uncertainty at the 2σ level from 132 spot analyses) (Figure 3a and Table S1 in the auxiliary material). The MAD apatite does have a relatively high Th/U ratio (∼26 in the ID-TIMS analyses, and between 15 and 30 in the LA-ICPMS runs). There is thus potential for excess206Pb introduced by incorporation of 230Th from the 238U-206Pb decay path during apatite crystallization. Corrections can be applied if the U/Th ration of the parent magma is known [e.g., Schärer, 1984; Amelin and Zaitsev, 2002]. However, this correction is usually only significant for very young samples, and the small corrections required are typically much less than the typical uncertainty of 5–10% for individual MAD1 spot analyses. This is supported by the concordance of the MAD1 ID-TIMS data. The self-normalized concordia age MSWD value for the combined MAD1 LA-ICPMS data is low (0.033,Figure 3a) indicative of uncertainty overestimation. The source of this excess uncertainty is most likely attributable to noise on the Faraday collector affecting the low 207Pb signal of ∼15 kcps for this apatite using a 30 μm spot size, as well as potential response time differences between the 204Pb ion counter signal and 206Pb Faraday signal.

Figure 3.

(a) Self-normalized MAD1 reference apatite data from 132 spots over 9 separate machine runs over course of 8 months showing excellent concordia age precision of 1.0% (2σ). The low MSWD value is indicative of ∼4% overestimation of uncertainties; (b) self-normalized MMap reference apatite data from 160 spots over 14 separate machine runs showing excellent concordia age precision of 1.2% (2σ). (c) Isochron plot from 169 individual spots from 14 machine runs (not corrected for common Pb).

4.2.2. McClure Mountain Fractionation Reference Apatite (MMap)

[13] The Cambrian McClure Mountain syenite of Colorado is the source of the widely used 40Ar/39Ar hornblende reference material MMhb [Alexander et al., 1978]. This syenite also contains numerous accessory minerals, the U-Pb systematics of which were examined in detail using ID-TIMS bySchoene and Bowring [2006]. Using a 3D total U-Pb isochron to correct for common lead these authors reported an ID-TIMS apatite U-Pb age of 523.51 ± 1.47/1.53/2.07 Ma (MSWD = 2.1; internal errors/with tracer calibration errors/with tracer calibration and decay constant errors).

[14] Apatite from the McClure Mountain syenite occurs as small euhedral grains of apatite varying from about 500 μm to less than 50 μm in axial length. Both inter- and intragrain U/Pb ratios of MMap are very consistent despite variable U concentrations of individual grains. Individual laser spots all yield concordant ages once corrected for common Pb usingStacey and Kramers' [1975] model (see section 4.3). Typical self-normalized concordia age uncertainty for 10–12 spots during a typical sample run is between 1.2 and 2.0% at the 2σ level. Over the course of 14 machine runs and 160 spot analyses the 2σ concordia age uncertainty is 0.6% (Figure 3b and Table S2). The variable U concentrations and hence μ (238U/204Pb) between grains allows production of a well-defined isochron plot that does not rely on a common Pb correction (Figure 3c). The initial 206Pb/204Pb composition obtained from this isochron (19.9 ± 3.3) although relatively imprecise, compares well with the value predicted by the Stacey and Kramers [1975] common Pb model of 17.90 at 523.5 Ma. This isochron shows a low MSWD value of 0.061 implying overestimation of uncertainties. Given the variable 206Pb/204Pb values in these grains (47 to 320), one potential cause of this excess uncertainty is response time differences between the 204Pb ion counter signal and 206Pb Faraday signal (hence out of phase 206Pb and 204Pb signals) combined with downhole intragrain 206Pb/204Pb variability during lasing.

4.2.3. Reference-Sample Fractionation Bracketing

[15] Bracketed fractionation factors are generated for 206Pb/238U, 206Pb/207Pb, and 208Pb/232Th based on a 30 point running average to incorporate at least five or six reference analyses (Figure 4). The vertical axis on these plots shows the value needed to correct a measured ratio (e.g., 206Pb/238U or 206Pb/207Pb) and the x axis represents each analyses in order of acquisition. The position of each unknown analysis is marked with a vertical gray line, and the value for each reference analysis is shown as a blue diamond. The sliding window average and its 1σ standard error are shown with red lines. The 206Pb/238U of each unknown is then corrected for fractionation based on the sliding window average. The 206Pb/207Pb ratio does show appreciably more scatter than the 206Pb/238U during each run. The 207Pb signal for the MAD1 reference apatite is typically ∼15 kcps, a value at which Faraday collector electronic noise can lead to higher than expected signal variation (hence overestimation of uncertainty) and low MSWD values. For example, during the 15 s run of each spot analysis the standard deviation of the 207 signal when ablating MAD1 apatite is ∼15% (1σ) compared to typical values of ca. 5% (1σ) for the 238, 232, 208, and 206 peak signals. While this can decrease the precision of our results, it has no bearing on their accuracy. Despite this scatter, the similar chemical behavior of 206Pb and 207Pb means they show little fractionation with a value close to 1.0. The U concentration values quoted in this study are approximations determined by comparison to these two reference apatites. Mean values of 20 ppm for MAD1 and 10 ppm for MMap are based on concentrations determined during independent fission track analysis calibrated against a dosimeter glass of known U concentration.

Figure 4.

Examples of reference-unknown206Pb/238U and 206Pb/207Pb fractionation correction bracketing using both the MAD1 and MMap reference apatite. The blue points represent the standard fractionation value (measured ICPMS ratio compared to ratio expected given known ID-TIMS age). The thick red line represents the 30 point running average of these values with 1σ standard error shown by the thin red lines. The gray bars represent a ±2% standard error about the average.

4.3. Danger of Using Non-matrix Matched Reference Materials

[16] To correct for elemental fractionation, the few previous attempts at apatite LA-ICPMS dating have used NIST reference material glass [Willigers et al., 2002], zircon [Storey et al., 2007], and titanite [Carrapa et al., 2009] or combinations of the above [Chew et al., 2011]. To test the robustness of using non-matrix reference materials, we undertook several runs incorporating in-house Sri Lanka (SL) zircon and Bear Lake titanite [Gehrels et al., 2008]. However, the results for MMap apatite yielded consistently discordant and inaccurate ages (Figure 5), demonstrating the danger of using non-matrix matched reference material for elemental fractionation correction of LA-ICPMS apatite U-Pb data. Although accurate LA-ICPMS apatite U-Pb ages have been produced using other non-matrix-matched reference materials and approaches [e.g.,Chew et al., 2011], to better guarantee accurate ages for unknown apatite, we advise using a matrix-matched apatite elemental fractionation reference material such as that from the McClure Mountain syenite or the TIMS characterized Madagascar gem rough apatite introduced in this study.

Figure 5.

Example concordia plots showing data for MMap apatite when using (a) the SL2 zircon fractionation standard and (b) the Bear Lake sphene standard. The ID-TIMS MMap apatite U-Pb age is 523.5 Ma, thus using these non-matrix matched reference minerals overestimates the true age by ca. 10–12%.

4.4. Common Pb Correction

[17] To accurately measure 204Pb, we assign its measurement to an independent ion-counter. Our low204Pb blank set-up for the Nu-Plasma HR multicollector ICPMS, coupled with the measurement of202Hg to correct for 204Hg isobaric interference, allows the 206Pb, 207Pb, and 208Pb peaks to be corrected for common Pb according to Stacey and Kramers' [1975] common lead model values for 206Pb/204Pb, 207Pb/204Pb, and 208Pb/204Pb given a suitable estimate of the grain age and requiring no assumption of concordance. However, owing to the high common to radiogenic Pb ratio (Pbc/Pb*) in most apatite, the initial 206Pb/238U age estimate from the raw measured data considerably overestimates the true age. To address this we apply a five-step iterative process to determine the206Pb/238U age used to estimate initial 206Pb/204Pb according to Stacey and Kramers' [1975] model [e.g., Chew et al., 2011]. We use the initial measured 206Pb/238U age to apply a Stacey and Kramers [1975] model value to calculate a revised corrected 206Pb/238U ratio and age. This corrected age is then used to calculate a new Stacey and Kramers [1975] 206Pb/204Pb model value, and calculate a new 206Pb/238U ratio and age. We repeat this process five times. We find that for almost all apatite, ages converge after 2 or 3 iterations. We additionally apply uncertainties of 1.5 for 206Pb/204Pb and 0.3 for 207Pb/204Pb based on the variation in Pb isotopic composition in modern crustal rocks. The improved effectiveness of using the iterative approach is shown in Figure 6for detrital apatite from a common Pan-African source yielding apatite U-Pb cooling ages of ca. 500 Ma. These detrital apatite data are discussed in more detail insection 7.

Figure 6.

(a) Concordia plot of some example detrital apatite data of common Pan-African (∼500 Ma) age. In this example common Pb was corrected according toStacey and Kramers' [1975] model 206Pb/204Pb value at the time represented by the raw measured 206Pb/238U age; (b) concordia plot with same data as in Figure 6a but using a five step iterative process to determine the age used to derive Stacey and Kramers' [1975] model common Pb composition.

5. Example Data

5.1. Bancroft Terrane Apatite, Canada

[18] Large centimeter-sized apatite crystals from various Grenville aged pegmatites in Ontario, Canada are widely available. We analyzed two crystals with varying common Pb and uranium concentrations, one from the Bear Lake region (Bear Lake 1) and one from the Wilberforce region. In all analyses, random small chips were selected from crushed fragments of larger cm-sized euhedral crystals. U-Pb sphene ages from the Bancroft terrane range from 1024 to 1074 Ma [Mezger et al., 1991] and hornblende 40Ar/39Ar total fusion and plateau ages with closure temperature of ca. 500°C similar to that of U-Pb in apatite [Harrison, 1982; Cherniak et al., 1991] range between 956 and 996 Ma [Cosca et al., 1995]. The U-Pb data from these apatite crystals and all other example apatite data are provided inTable S3.

[19] Bear Lake 1 has 206Pb/204Pb ratios ranging from 54 to 250 in 5 spots from 5 different fragments analyzed. Two spots with the highest U concentrations (59 and 60 ppm, respectively) yield higher precision concordant ages of 935 ± 24 and 969 ± 16 Ma (2σ). All 5 spots give a concordia age of 962 ± 13 Ma (2σ) with MSWD = 7.0 (Figure 7a). The weighted mean 206Pb/238U age of the 5 analyses is statistically equivalent at 958 ± 13 Ma (2σ), but with lower MSWD of 1.5 (Figure 7b). The dated fragments have variable μ (238U/204Pb) allowing calculation of an imprecise 238U-206Pb isochron age of 911 ± 160 Ma (95% confidence, MSWD of 0.20) uncorrected for common Pb (Figure 7c). The low precision and low number of spot analyses do not allow calculation of a meaningful initial 206Pb/204Pb value (26 ± 33 compared to Stacey and Kramers' [1975] model value of 17.066 for 1 Ga). An isochron anchored through an initial 206Pb/204Pb value that encompasses Stacey and Kramers' [1975] model Pb values between 0 Ga and 2 Ga (18.700 to 15.159) with an additional uncertainty of 1.5 to account for modern day variability in 206Pb/204Pb values, produces an age of 951 ± 53 Ma (95% confidence) with MSWD of 0.21.

Figure 7.

Apatite U-Pb data plots from (a–f) 2 crystals of the Grenville Bancroft terrane, Canada; (g–i) anorthosite from the Duluth Complex, Minnesota; (j–l) Tin Mountain pegmatite, South Dakota; (m–o) the Neoarchean Louis Lake Batholith of the Wind River Range, Wyoming; and (p–r) a Cretaceous granite of the Salinian block, California (see text for details).

[20] The Wilberforce apatite has 206Pb/204Pb values around 100 and relatively high uranium concentrations for apatite of 39 to 58 ppm. A concordant age of 937 ± 20 Ma (95% confidence) is obtained from 9 laser spots on randomly selected chips from a large ca. 5cm crystal (Figure 7d). The MSWD value of 0.093 is low indicative of uncertainty overestimation related to noise-related variability in the low 207 signal. However, the excess uncertainty does not affect the accuracy of our results. The ages do vary slightly along concordia. One potential cause is differential Pb diffusion (and hence retentivity) in different domains within the crystal during slow post-Grenville cooling. This spread in ages is seen more clearly on the weighted mean206Pb/238U age plot (938 ± 31 Ma, 95% confidence, MSWD of 4.0) with individual ages between 895 ± 32 Ma and 1039 ± 51 Ma (Figure 7e). One spot has a very different μ value to most of the other data, allowing the calculation of a 238U-206Pb isochron age of 1011 ± 48 Ma (Figure 7f), albeit with a initial 206Pb/204Pb value of 10.8 ± 3.1 that is too low relative to the expected Stacey and Kramers' [1975] model value of 17.066. Ignoring the anomalous low μ spot, but anchoring the isochron through the same initial 206Pb/204Pb value used above for Bear Lake 1 gives an age of 934 ± 71 Ma (95% confidence) with MSWD of 1.09.

[21] The weighted mean U-Pb ages obtained from both of these Bancroft terrane apatite crystals (958 ± 13 Ma and 938 ± 31 Ma, respectively) match well with similar closure temperature hornblende40Ar/39Ar cooling ages between 956 and 996 Ma from the same region [Cosca et al., 1995] supporting the accuracy of our U-Pb apatite ages.

5.2. Forest Center Anorthosite, Duluth Complex, Minnesota

[22] Zircon from anorthosite of the Duluth Complex, near Forest Center, Minnesota is a widely used fractionation reference material for both LA-ICPMS and SIMS U-Pb dating. High precision ID-TIMS analysis of these zircons yield a weighted207Pb/206Pb age of 1099.0 ± 0.6 Ma [Paces and Miller, 1993]. Apatite is also contained in these rocks typically as small euhedral grains some 100–200 μm in diameter. We analyzed 10 apatite grains, each with a single 30 micron spot. U concentrations varied from 4 to 14 ppm, and 206Pb/204Pb values between 49 and 230. The 10 spots gave a concordia age of 1079 ± 47 Ma (95% confidence) with MSWD of 2.1 (Figure 7g), and a weighted mean 206Pb/238U age of 1094 ± 34 Ma (2σ) with MSWD of 0.27 (Figure 7h). These ages are within the uncertainty of the zircon 207Pb/206Pb age and imply fast cooling of these rocks to below 500°C following their intrusion. Common Pb uncorrected data plotted as an 238U-206Pb isochron give an age of 1070 ± 140 Ma (95% confidence) with low MSWD of 0.056 (Figure 7i) implying some uncertainty overestimation.

5.3. Tin Mountain Pegmatite, South Dakota

[23] Apatite from large (1–2 cm) apatite crystals from the early Proterozoic Tin Mountain granitic pegmatite, Black Hills, South Dakota has been previously dated by ID-TIMS [Krogstad and Walker, 1994]. This study revealed these large apatite crystals to be complex, with older concordant core ages of ca. 1703 Ma related to slow cooling following intrusion of the parent rock at 1715 ± 3 Ma, and younger discordant rim ages as young as 1673 Ma. In addition, Tin Mountain apatite shows occasional extreme reverse discordance requiring either loss of U or gain of Pb that Krogstad and Walker [1994]attribute to the possible presence of a uranium-rich phase, such as uraninite.

[24] Our data (from randomly chosen fragments over two separate machine runs) show similar dispersed data around concordia, with both normal discordance and minor reverse discordance. When anchored through the origin (207Pb/206Pb ages) the mean upper intercept age is slightly older than the common Pb corrected core ages of Krogstad and Walker [1994] (1762 ± 69 Ma at 95% confidence, compared to 1703 ± 3 Ma – Figure 7j). These dispersed data result in a large MSWD of 14, and most likely represents core to rim age zonation seen by Krogstad and Walker [1994]that these authors attribute to open-system U-Pb behavior along cracks and close to the rims during slow cooling. The Tin Mountain apatite has very high amounts of common Pb (204Pb signal of >60 kcps). However, high U concentrations (193–410 ppm) and old age still provide enough radiogenic lead to provide reasonable 206Pb/204Pb ratios. The spread in ages is seen clearly on the 206Pb/238U weighted mean age plot (Figure 7k) where ages between 1431 Ma and 1765 Ma yield a weighted mean age of 1586 ± 25 Ma (95% confidence, MSWD of 2.1). The 238U-206Pb isochron (requiring no initial common Pb assumption) while relatively precise, gives an age of 1525 ± 65 Ma (Figure 7l) younger than any obtained by Krogstad and Walker [1994]including the 1715 ± 3 Ma intrusion age and ca. 1650 Ma Rb-Sr muscovite cooling ages [Krogstad and Walker, 1994]. The initial 206Pb/204Pb value of 20.7 ± 4.7 is higher that the muscovite value of 16.695 obtained by Krogstad and Walker [1994]- again perhaps reflecting some open-system U-Pb behavior and/or the presence of U-rich phases within the randomly chosen apatite fragments we analyzed.

5.4. Wind River Range, Wyoming

[25] The oldest apatite we have analyzed are small grains (ca. 100–200 μm) from a Neoarchean Louis Lake batholith granodiorite of the Wind River Range in Wyoming that has a published discordant (ca. 3%) upper intercept ID-TIMS zircon207Pb/206Pb crystallization age of 2629.5 ± 1.5 Ma [Frost et al., 1998]. Owing to its old age and relatively high U concentrations (17 to 124 ppm) this apatite has high 206Pb/204Pb values for apatite of between 191 and 1797. Of the 10 grains analyzed, most are concordant (Figure 7m), although show variation in age, with an upper intercept 207Pb/206Pb age of 2599 ± 26 Ma (95% confidence fit through the origin). This variation in age may reflect differential Pb diffusion (and hence retentivity) in different sized crystals during slow post-crystallization cooling. The weighted mean206Pb/238U age (Figure 7n) of 2599 ± 28 Ma (2σ, MSWD of 0.49) is indistinguishable from the upper intercept 207Pb/206Pb age. These ages are corroborated by the common Pb uncorrected 238U-206Pb isochron age of 2592 ± 150 Ma (95% confidence) (Figure 7o). These near identical apatite U-Pb ages indicate that this rock cooled to below ca. 500°C within ca. 30 Ma following crystallization at ca. 2630 Ma. Shallow intrusion is supported by contact metamorphism of the Louis lake batholith in older country rocks [Frost et al., 1998] and implies that this rock has not seen temperatures > ∼500°C since the Neoarchean.

5.5. Cretaceous Franitoids of the Salinian Block, California

[26] We analyzed some relatively high U apatite from a slowly cooled Cretaceous granite from the Salinian block of California from which an apatite isotope dilution 207Pb/206Pb isochron age of 70 ± 20 Ma and a decay-constant-corrected hornblende K-Ar age of 94 Ma were previously obtained [Mattinson, 1978]. Thirteen apatite grains from sample JM71–2 (sample BH-1 in the work ofMattinson [1978]) have U concentrations of between 32 ppm up to 153 ppm, and 206Pb/204Pb values of 32 to 99. The concordia and weighted mean 206Pb/238U ages are indistinguishable (91.4 ± 4.0 Ma and 91.5 ± 4.0, respectively) with good MSWD values of 0.90 and 0.47 (Figures 7p and 7q). The common Pb uncorrected 238U-206Pb isochron age of 87 ± 14 Ma (Figure 7r) shows good agreement, albeit with low MSWD indicative of some uncertainty overestimation. All these ages are within error of the similar closure temperature hornblende K-Ar age of ∼94 Ma implying accuracy of our results.

6. Apatite With High Common Pb, Low Uranium, and/or Young Age

[27] One limitation when trying to avoid significant damage to small apatite grains (e.g., by use of 30 micron spot size and 15 s acquisition time) is that data quality in terms of precision and concordance tends to break down in apatite with low U concentrations, low radiogenic to common Pb ratios (206Pb/204Pb values) and/or young age. Examples of analyses and data quality using 30 micron laser spot size obtained from such apatite are shown below including samples with mixed high and low U concentration grains (Figure 8) as well as samples of young age and/or very low U concentration (Figure 9). The causes of reduced data quality in such low Pb*/Pbc grains are discussed below, followed by additional analytical data from apatite with both young age and or high proportions of common Pb that demonstrate substantial improvement in data quality when using larger spot size, albeit at the expense of significant damage or destruction of smaller grains.

Figure 8.

(a–f) Apatite U-Pb data from 2 Proterozoic granites of the Gold Butte block, Nevada and (g–i) a Cretaceous granodiorite of the North Patagonian batholith, southern Chile.

Figure 9.

Apatite U-Pb data plots from (a, b) the Neoproterozoic Mud Tank apatite, central Australia, (c, d) gem apatite of the Cerro de Mercado iron ore deposit, Durango, Mexico, and (e, f) small apatite crystals from the Fish Canyon Tuff, Colorado.

6.1. Proterozoic Granite of the Gold Butte Block, Nevada

[28] We analyzed two samples of apatite from slowly cooled Proterozoic rapakivi granite of the Gold Butte block of southern Nevada (PRGB4 and PRGB18) that have published isotope dilution ICPMS 207Pb/206Pb ages of 1130 ± 45 Ma and 1389 ± 56 Ma, respectively [Reiners et al., 2000]. Small apatite grains (ca. 100–200 μm) from both samples have average apatite uranium concentrations varying from 1 to 27 ppm, and variable 206Pb/204Pb values from low values of 20 to a few grains having values over 100. Variable quality data is well demonstrated in these samples. In PRGB4, the grains with lower U (<6 ppm) and 206Pb/204Pb (<40) have high uncertainties giving a concordia age of 1263 ± 32 Ma (2σ) with MSWD of 2.4 (Figure 8a). The weighted mean 206Pb/238U age shows a better MSWD value of 1.02 and age of 1264 ± 32 Ma (Figure 8b). Two grains with higher U and 206Pb/204Pb values (>43) are concordant, giving an age of 1230 ± 39 Ma (2σ) with low MSWD of 0.056 (Figure 8a, inset). Only five grains were analyzed from sample PRGB18. The individual spots have large uncertainty (18–24% 2σ on the 206Pb/238U age) but when pooled yield a concordant age of 1114 ± 88 Ma (7.9% 2σ) with MSWD of 0.94 (Figure 8d). The one grain with high 206Pb/204Pb (∼68) gives a concordant 206Pb/238U age of 1188 ± 163 Ma (2σ). The weighted mean using just the better quality 206Pb/238U ages is 1116 ± 88 Ma (2σ, MSWD of 0.82) (Figure 8e). The data from both samples produce imprecise 238U-206Pb isochron ages with low MSWD that are consistent with the weighted mean ages (Figures 8c and 8f). The initial 206Pb/204Pb values are geologically reasonable, although also with high uncertainties (18.7 ± 2.1 and 17.7 ± 2.3, respectively).

6.2. Patagonian Batholith

[29] In addition to analysis of a Cretaceous granitoid of the Salinian Block, California (section 5.5), we analyzed a well-characterized Cretaceous granodiorite sample from the Patagonian batholith of southern Chile [Thomson et al., 2001]. In contrast to the Salinian granitoid, most of the 8 apatite grain analyses from this sample had low U concentrations (<11 ppm) leading, with one exception, to low precision spot ages. The data yield a concordant age of 129 ± 11 Ma (2σ) (Figure 8g). However, one grain with higher U (56 ppm) and higher 206Pb/204Pb (68) yields an 206Pb/238U age of 119.2 ± 17.0 Ma (2σ) (Figure 8h) more consistent with the previously determined zircon ID-TIMS crystallization age from this sample of 118.7 ± 0.9 Ma [Hervé et al., 2007] albeit still just within 2 σuncertainty of the pooled concordia age. This sample illustrates the potential danger of low precision analyses in some young samples, where minor discordance is hidden within large uncertainties yielding an older concordia age that may mask the true apatite U-Pb cooling age. An uncorrected238U-206Pb isochron produces an age in line with the zircon U-Pb age of this rock, albeit with large uncertainty (117 ± 30 Ma with 95% confidence) with an initial206Pb/204Pb value (Figure 8i) of 19.9 ± 3.6 that compares with Stacey and Kramers' [1975] model value of ca. 18.6 at 100 Ma.

6.3. Mud Tank Carbonatite

[30] Large zircon crystals from the Neoproterozoic Mud Tank carbonatite, Strangways Range, Northern Territory, Australia, are used widely as reference zircon [e.g., Horstwood, 2008]. Large centimeter to decimeter-sized apatite is also found in the same deposit, and has been used as a calibration reference in (U-Th)/He and fission track dating [e.g.,Green et al., 2006; Spiegel et al., 2009]. The apatite we analyzed (random fragments from a large 10 cm sized crystal) has very low uranium concentration (<3 ppm) and very low radiogenic to common Pb ratios (between 0.7 and 0.9, equivalent to more than 54% common Pb) leading to imprecise results using a 30 μm spot size. A Wetherill concordia plot (Figure 9a) shows concordant data with an age of 455 ± 51 Ma (2σ, MSWD = 1.01). The weighted mean 206Pb/238U age from 19 analyses is 463 ± 51 Ma (2σ, MSWD = 0.69) (Figure 9b). All these ages are younger than the zircon U-Pb age of 732 Ma [Black and Gulson, 1978], but older than late Ordovician to Carboniferous (450–300 Ma) metamorphism of the Alice Springs Orogeny in this region [Haines et al., 2001]. These ages are similar to Sm-Nd and other ages of 460–485 Ma associated with metamorphism up to granulite facies in the nearby Harts Range [Mawby et al., 1999].

6.4. Durango Apatite

[31] Gem apatite from the early Oligocene Cerro de Mercado iron ore deposit, Durango, Mexico (31.4 Ma [McDowell et al., 2005]) is a widely used reference material for both fission track and (U-Th)/He dating [Barbarand et al., 2003]. However, owing to its relatively low and variable uranium concentration (<12 ppm), and young age, using 30 μm spots only yields very imprecise, poor quality results (Figure 9c). To achieve meaningful and precise U-Pb ages from this apatite require larger laser spots (seesection 6.7.3). The very poor precision results in a mean 206Pb/238U age from 12 individual spot analyses of 25 ± 19 Ma (2σ) (Figure 9d). Using smaller spots is thus able to identify young grains, which, if grains or fragments are large enough, can then be further analyzed to obtain higher precision results.

6.5. Fish Canyon Tuff

[32] Apatite from the Fish Canyon Tuff is a common fission track dating reference material [Barbarand et al., 2003]. However, its low uranium concentration (∼10 ppm), young age (28 Ma), and very high proportion of common to radiogenic Pb (206Pb/204Pb < 24; Pb*/Pbc ratios < 0.6, equivalent to >70% common Pb) leads to mostly discordant data (Figure 9e) and a wide spread in 206Pb/238U ages (Figure 9f). Very similar 238U/204Pb (μ) values and very low 206Pb/204Pb values do not allow construction of a meaningful 238U-206Pb isochron. In such young samples with very low amounts of radiogenic Pb and very high amounts of incorporated common Pb, our techniques do not currently provide enough accuracy and precision on the measured Pb/U ratios and 206Pb/204Pb values to provide an accurate age.

6.6. Data Quality Limits Using a 30 μm Spot Size

[33] After applying the iterative approach to Stacey and Kramers' [1975] model common Pb correction using measured 204Pb (section 4.3), data quality in terms of percent departure from concordance still tends to break down at very low measured 206Pb/204Pb values < ∼30 and ages < ∼75 Ma (Figure 10a). Most discordance almost certainly relates to low 207 signal (itself dependent on amount of common Pb, age and U concentration). The typical Faraday 207 signal of the 485 Ma MAD1 reference apatite (with ∼ 20 ppm U) is ∼15 kcps using a 30 μm laser spot. Thus for an apatite with similar age and common Pb, but with only 1 ppm U, the 207 signal would be <1000 cps (approaching Faraday collector noise levels when using 3 × 1011 ohm resistors). Clearly in such cases only 206Pb/238U ages should be considered, assuming common Pb has been adequately corrected. When using small laser spot size of 30 μm, our results show that extreme caution needs to be applied when trying to interpret U-Pb data from apatite with low age and uranium concentration, and high proportions of common Pb (low206Pb/204Pb values).

Figure 10.

(a) Plot of 206Pb/238U age versus 206Pb/204Pb values for all example data presented in this study showing how quality of data in terms of % departure from concordance depreciates at ages below ∼75 Ma and at 206Pb/204Pb values below ∼30 (below and to left of dashed line); (b) plot showing little correlation between 204 beam intensity versus % departure from concordance.

[34] Another potential limitation is a reduction in the effectiveness of Stacey and Kramers' [1975] common Pb correction with amount of radiogenic lead (Pb*/Pbc). Such a correlation has been previously observed in ID-TIMS U-Pb apatite data [Schoene and Bowring, 2006]. These authors found that use of a 3D total Pb-U isochron [Ludwig, 1998] was most effective for correcting common Pb in apatite. However, the lower precision of our LA-ICPMS method has so far excluded the possibility of using the 3D total Pb-U isochron approach. The higher the proportion of common Pb, the more sensitive the common Pb correction is to the value used for the initial isotopic composition of the common Pb component. In such cases a model-based value may not be appropriate [e.g.,Schoene and Bowring, 2006]. Note that no correlation was found between background-corrected 204 signal intensity and the efficiency of the204Pb common Pb correction in terms of concordance for the analyses published in this study (Figure 10b) implying inaccurate measurement of 204Pb is not an issue.

6.7. Improved Data Quality Using Larger Laser Spot Size

[35] To improve data quality we carried out analyses using larger 65 μm and 110 μm laser spot sizes on several problematic apatites: a Grenville aged Bancroft Terrane apatite (Bear Lake 2) with high common Pb (206Pb/204Pb < 35), the low uranium Mud Tank apatite (U concentrations <3 ppm) and the young (31 Ma) Durango apatite (Figure 11 and Table S4). Each of these apatites are available as large cm-sized crystals, thus enabling dating using large spot sizes without worry of grain destruction. Typical detrital and magmatic apatite grains analyzed in this study are usually no larger than ca. 200 × 80 microns, and more typically around 100 × 60 microns, thus using a larger spot size in such samples is limited by apatite size. MAD1 apatite was used as the mass fractionation reference apatite. Note that in contrast to the other analyses in this study, these samples were ablated with a newly acquired Photon Machines Analyte G2 193 nm excimer laser equipped with a HelEx ablation chamber.

Figure 11.

(a, b) Large 65 μm and 110 μm laser spot size apatite U-Pb data on an older Grenville aged apatite with high proportion of common Pb, Bear Lake 2; (c, d) a low uranium concentration apatite, Mud Tank; (e, f) the young 31 Ma Durango apatite.

6.7.1. Bear Lake Apatite

[36] A second apatite crystal from the Bear Lake region (Bear Lake 2) has a very high proportion of common Pb given its Grenville age, with measured 206Pb/204Pb values varying from 29 to 35 and Pb*/Pbc ratios of 1.2 to 1.5 (equivalent to ca. 40–45% common Pb) and typical uranium concentration for apatite of around 10 ppm. The 65 μm spot data, while still showing some variation in individual spot precision, give a precise 10 spot concordant U-Pb age of 921.3 ± 8.7 Ma (2σ, MSWD = 0.20) (Figure 11a). The 110 μm spot data, albeit from only 5 analyses, yield a similar concordant age of 896 ± 21 Ma (2σ, MSWD = 0.69) (Figure 11b), although still with some variability in precision between each spot analysis. In addition, the uncertainty overestimation leading to low MSWD values is effectively eliminated when using larger laser spot size. We attribute this to reduced Faraday noise on the 207 signal. For example for the MAD1 reference apatite, the 207 signal increases from ∼15 kcps using a 30 μm spot to ∼50 kcps with the 65 μm spot and ∼140 kcps with the 110 μm spot. This increased signal manifests itself in the form of much less scatter on the 207Pb/206Pb ratio with both downhole fractionation (Figure 1e) and the bracketed reference-sample apatite elemental fractionation correction.

6.7.2. Mud Tank Apatite

[37] The Mud Tank apatite, with its low uranium concentrations (<3 ppm) and high proportion of common Pb (206Pb/204Pb values < 30; Pb*/Pbc ratios < 1) yielded poor data when using the 30 μm spot size with a 2σ age uncertainty of ∼11% (section 6.3). With larger spot size, a similar age is obtained. However, the data quality is much improved. A concordia age from ten 65 μm spots of 464.4 ± 9.6 Ma (2σ, MSWD = 1.4) is obtained – an uncertainly of 2.1% (Figure 11c), while using the large 110 μm spot, a concordia age of 457.2 ± 7.8 Ma (2σ, MSWD = 0.31) was acquired – a 2σ uncertainty of 1.7% from only 5 spots (Figure 11d).

6.7.3. Durango Apatite

[38] The improvement in data quality using larger spot size with the young Durango apatite is even more marked. While the uranium concentration and 206Pb/204Pb values of this apatite are unremarkable (ca. 10–15 ppm, and between 90 and 160, respectively), the amount of total Pb (radiogenic + common Pb) in this apatite is very low with typical 207 signal often close to 1000 cps using a 30 μm spot making the acquisition of good quality data challenging. However, when using a 65 μm spot a concordia age of 32.2 ± 5.3 Ma (2σ, MSWD = 0.21) was obtained from only 10 spots (uncertainty of 16%), albeit with two spots showing large uncertainties (Figure 11e). With the 110 μm spot size, 5 spots yielded a concordia age of 32.0 ± 3.1 (2σ, MSWD = 0.32) - a 2σ uncertainty of 9.7% (Figure 11f). These ages compare well with the precise sanidine-anorthoclase40Ar–39Ar reference age of 31.44 ± 0.18 Ma [McDowell et al., 2005].

7. Detrital Apatite U-Pb Dating

7.1. Methodology and Use of Apatite Fission Track Mounts

[39] One of the challenges with single grain triple dating is developing a robust procedure for mounting apatite fractionation reference materials along with a fission track epoxy mount in the laser ablation cell to perform a reliable and accurate fractionation correction. We found the best results were achieved by mounting reference apatite in a small ∼2 mm thick polished epoxy mount mounted next to the similar thickness epoxy fission track mount, such that He flow is similar across both mounts (Figure 12).

Figure 12.

Schematic illustration of laser ablation cell set-up of epoxy fission track mount and neighboring polished epoxy reference apatite mount. The fission track mount is approximately 15 × 15 mm in size, the reference apatite mount 15 × 5 mm, and each mount has a thickness of about 2 mm.

7.2. Single Grain, Multiple Spot Example Data

[40] We have found that reliable U-Pb ages can be obtained without significantly damaging detrital apatite grains by ablating several 30μm laser spots in a single grain. In the several detrital apatite samples we have analyzed, the polished internal surface of typically sized apatite (ca. 100 × 60 μm) allows two to three spots to be ablated without destroying the grain. Larger single spots may also be used particularly in younger and/or lower uranium grains, but we have found larger spots lead more often to grain disintegration. U-Pb data from multiple spots from a single grain also better accounts for any grains that have substantial intragrain spatial variation in uranium concentration and amounts of common Pb (and hence206Pb/204Pb ratios).

[41] We present here unpublished apatite U-Pb data from a Holocene diamict from Prydz Bay, offshore East Antarctica (JPC34) derived from a large cratonic catchment dominated by Pan-African (ca. 500 Ma) and older metamorphic rocks. InFigure 6b (in section 4.4, where we previously used these data to demonstrate the efficiency of the common Pb correction) all the data from every spot (122 spots total, from 52 single grains – Table S5) are plotted on a Wetherill concordia plot. Uranium concentrations for most grains are typical for apatite (varying from as low as 1 ppm to as high as 112 ppm (average of 17 ppm), with 206Pb/204Pb ratios varying from 21 (up to 60% common to radiogenic Pb) to 182 (average value of 47). Most of the spots fall on concordia, with only a few spots from low uranium grains showing some discordance, albeit forming an apparent discordia toward a lower concordia intercept of about 500 Ma.

[42] In Figure 13, we present multiple spot data from 6 different grains with variable U concentrations and 206Pb/204Pb values. Figures 13a, 13b and 13c show concordant data from three grains with near average U concentrations between 11 and 24 ppm, each with three spot analysis. Each individual spot has relatively low precision (between ∼7 and 20% with 2σ uncertainty). However, by combining three spots from the same grain, a higher precision can be achieved varying from 9.2% to 5.6% (2σ) in the illustrated samples, to as low as 3% in grain 33. In low uranium grains of <∼5 ppm (<10% of the total grains in this sediment sample) the data tend to be discordant (e.g., Figure 13d) and form discordia with a concordia lower intercept age with very large uncertainties (80% 2σuncertainty). In most cases such ages should be discarded. However, if, for example, the sole goal of a detrital apatite U-Pb dating study is to differentiate grains with an Cordilleran arc source (ca. 100 Ma) from grains from an Proterozoic to Archean source (>1000 Ma), data from such low U concentration grains may still be of some use. In a couple of grains, the ages show some small reverse discordance (Figure 13e). This likely reflects overcorrection for common Pb related to assumptions inherent in using Stacey and Kramers' [1975] model and age used in the correction. In the example shown, the weighted mean 206Pb/238U age (Figure 13f) gives a better MSWD value (1.9 compared to 3.2).

Figure 13.

Example multiple spot U-Pb apatite data from 5 single grains taken from JPC34, a Holocene sediment from offshore East Antarctica with a predominantly Pan-African (∼500 Ma) catchment. See text for more detailed discussion of data from each individual grain.

[43] Analyzing the multiple spot data from each grain provides us with an age spectrum that can be plotted as an age probability density plot (PDF) with binned age histogram in the standard manner used for detrital zircon U-Pb analysis (Figure 14a). Published detrital hornblende 40Ar/39Ar data has been obtained from this same sample [Roy et al., 2007]. The closure temperature for Ar in hornblende (ca. 500°C [Harrison, 1982]) is similar to that for Pb in apatite [Cherniak et al., 1991] thus comparison of the data provides a test of the accuracy of our detrital apatite U-Pb results. The data are compared inFigure 14b. The main age peaks of each data set match well, and thus support the accuracy of our approach to U-Pb dating detrital apatite epoxy fission track mounts, especially in older samples. Note that the individual grain age precision of the apatite U-Pb ages is lower than the hornblende40Ar/39Ar ages. However, the sample preparation and data collection (ca. 1 min per laser spot) using our routine laser ablation ICPMS approach is less involved and time-consuming than hornblende40Ar/39Ar dating.

Figure 14.

(a) Age probability density plot and binned age histogram for apatite U-Pb data obtained using the single grain multiple laser spot approach; (b) comparison of detrital apatite U-Pb ages with previously published detrital hornblende40Ar/39Ar ages obtained from the same sediment sample [Roy et al., 2007].

8. Conclusions

[44] We have been able to overcome previous limitations to routine in situ U-Pb dating of apatite by laser ablation ICPMS by identifying two well-characterized matrix-matched reference apatites to correct for elemental fractionation coupled with the use of a Nu Plasma multicollector ICPMS with an attached short-pulse 193 nm excimer laser that allows direct measurement of204Pb corrected for background 204Hg. Two new reference apatites are large fragments of gem-quality apatite from Madagascar that was independently characterized by TIMS U-Pb dating and apatite from the McClure Mountain syenite (source of the40Ar/39Ar reference material MMhb). Using these reference apatites and measured 204Pb with an iterative Stacey and Kramers common Pb correction has allowed us to accurately reproduce ages from several well-characterized apatite samples. In practice, we can achieve precision <2% (2σ) on concordant ages with as few as five 30 μm spot analyses. Thus in many different apatites we achieve our stated goal of acquiring reliable apatite U-Pb ages without significant damage to the individual grains using a 30μm spot size, and 15 s acquisition time. However, in apatite with low U concentrations, low radiogenic to common Pb ratios (206Pb/204Pb values < ∼35) and young ages (<∼75 Ma) data quality is compromised when using a small 30 μm spot size. Such limitations can be overcome by using spot sizes of 65 μm or greater, but at the expense of significant grain damage and/or destruction. If damage or destruction is not an issue, then use of a large spot size with our approach can provide accurate and precise apatite U-Pb ages even in young (<∼75 Ma), low uranium, and high common Pb samples. Initial results of U-Pb dating of detrital apatite grains from a Pan-African (∼500 Ma) source indicate that precise ages (<4% 2σ) can be achieved in most grains using two or three 30 μm spots per grain. The accuracy of our detrital ages is supported by a good match with similar closure temperature 40Ar/39Ar detrital hornblende ages from the same sediment.

Acknowledgments

[45] Facilities at the Arizona LaserChron Center are supported by National Science Foundation awards NSF-EAR 0732436 and 1032156. We thank Scott Johnston who initiated this study and collected the McClure Mountain syenite. The following are thanked for supplying apatite mineral separates: Peter Reiners – Gold Butte apatite; Jim Mattinson – JM71 apatite; Ray Donelick – Forest Center FC5 apatite; Mona Sirbescu – Tin Mountain apatite; Ulrich Glasmacher – Durango apatite; Sidney Hemming – JPC34 apatite; Lynn Peyton – Wind River apatite. Robinson Cecil provided valuable assistance in initial set-up of the Nu-Plasma for apatite U-Pb dating. Mark Baker, David Steinke, and Ben McElhaney are thanked for ALC facility support and machine maintenance. This manuscript and quality of data interpretation was much improved by very thorough reviews from Craig Storey and Matt Horstwood.

Ancillary