In situ location and U-Pb dating of small zircon grains in igneous rocks using laser ablation–inductively coupled plasma–quadrupole mass spectrometry



[1] A new U-Pb zircon dating protocol for small (10–50 μm) zircons has been developed using an automated searching method to locate zircon grains in a polished rock mount. The scanning electron microscope-energy-dispersive X ray spectrum-based automated searching method can routinely find in situ zircon grains larger than 5 μm across. A selection of these grains was ablated using a 10 μm laser spot and analyzed in an inductively coupled plasma-quadrupole mass spectrometer (ICP-QMS). The technique has lower precision (∼6% uncertainty at 95% confidence on individual spot analyses) than typical laser ablation ICP-MS (∼2%), secondary ion mass spectrometry (<1%), and isotope dilution-thermal ionization mass spectrometry (∼0.4%) methods. However, it is accurate and has been used successfully on fine-grained lithologies, including mafic rocks from island arcs, ocean basins, and ophiolites, which have traditionally been considered devoid of dateable zircons. This technique is particularly well suited for medium- to fine-grained mafic volcanic rocks where zircon separation is challenging and can also be used to date rocks where only small amounts of sample are available (clasts, xenoliths, dredge rocks). The most significant problem with dating small in situ zircon grains is Pb loss. In our study, many of the small zircons analyzed have high U contents, and the isotopic compositions of these grains are consistent with Pb loss resulting from internal α radiation damage. This problem is not significant in very young rocks and can be minimized in older rocks by avoiding high-U zircon grains.

1. Introduction

[2] Mafic rocks are not often dated using U-Pb zircon techniques because of their primitive nature (low degree of fractional crystallization) and the low Zr contents in the magma [Watson and Harrison, 1983] which restricts zircon formation. This paper describes a protocol that has been developed to locate small, in situ zircon grains and measure their U-Pb age. We define ‘small’ zircon grains as having a surface area of between 25 μm2 (i.e., 5 μm by 5 μm) and 1000 μm2 (i.e., 20 μm by 50 μm) on a polished surface. In situ studies of small zircon grains face two significant challenges: location and analysis of the grains. To locate zircons we used a scanning electron microscope (SEM)–based mineral mapping software package that allows automated location and identification of zircon grains as small as 2 μm in diameter. In the analysis stage of our study, we use a 10 μm laser spot size with a maximum sampling depth of 15 μm (maximum ablation volume of 1180 μm3) to sample an order of magnitude less zircon, under similar conditions, than ablation at a 35 μm spot size (maximum ablation volume of 14,400 μm3).

[3] The smallest spot size in routine use for laser ablation-inductively coupled plasma-mass spectrometer (LA-ICP-MS) zircon dating is 20 μm although 10 μm spots are reported [i.e., Gehrels et al., 2008; Johnston et al., 2009] and spots as small as 5 μm are used for monazite dating [Simonetti et al., 2006; Zong et al., 2010]. As spot size decreases, the amount of material ablated also decreases, the particle size distribution changes and U-Pb downhole fractionation increases [Guillong and Gunther, 2002; Horn et al., 2000; Tiepolo, 2003]. To minimize these effects and achieve the highest quality data and therefore the most precise dates, ICP-quadrupole mass spectrometer (QMS) studies typically use spot sizes >35 μm [i.e., Feng et al., 1993; Fryer et al., 1993; Horn et al., 2000; Jackson et al., 2004; Meffre et al., 2007]. Sector field studies such as those by Tiepolo [2003] and Frei and Gerdes [2009] use smaller spot sizes, down to 20 μm ablating for 60 s; multicollector studies such as those by Simonetti et al. [2008, 2006], Gehrels et al. [2008], and Johnston et al. [2009] report 10 μm to 40 μm laser spot sizes and ablation times down to approximately 10 s.

[4] Where large zircon grains are available they are technically easier to date and they tend to provide more precise dates. However, in many rock types only small zircons are available for age dating. This paper describes a new, automated, in situ zircon location method and a 10 μm spot size LA-ICP-QMS analysis method that measures U-Pb ages in small zircon grains. The method is rigorously tested on a well characterized rhyolite and is then demonstrated on dolerites and gabbros from Alaska and the Tonga fore arc.

2. Methods

2.1. Finding Zircons

[5] The first step of any zircon dating study is to find zircon grains. Conventionally, this is done by crushing the sample and isolating the nonmagnetic heavy minerals. The minimum size for a zircon grain to be effectively recovered using this process is 30 μm by 30 μm. Methods such as acid extraction [i.e., Bindeman et al., 2002] can recover smaller zircons (10–20 μm). However, grains smaller than 30 μm are best analyzed in situ which eliminates the potential for laboratory contamination, preserves the original crystal shapes and provides a geological context for the growth of the crystal [Simonetti et al., 2006]. Both SIMS and LA-ICP-MS methods can analyze zircons in situ, but finding small in situ grains can be difficult. With the development of commercial SEM-based mineral mapping software, pioneered for applied mineralogy, it has recently become feasible to automatically find small mineral grains within a polished rock on a large number of samples [Fandrich et al., 2007; Gu, 2003]. A similar method has been developed for in situ mapping of baddeleyite but using an electron microprobe [Chamberlain et al., 2010; Schmitt et al., 2010].

[6] For the in situ LA-ICP-QMS protocol described in this study a 2 cm rock chip was encased in a 2.5 cm round epoxy mount, polished and carbon coated. The zircon grains were found in these samples using a FEI Quanta 600 SEM controlled by an automated software package (Mineral Liberation Analyzer or MLA) developed by JKTech©. The specific software option used in this study was the Sparse Phase Liberation-Lite (SPL-LT) method. This method collects high-resolution backscattered electron (BSE) images and then searches them on a frame by frame basis for grains with a BSE brightness close to that of a reference zircon. The software then collects an energy-dispersive X ray spectrum (EDS) for each potential zircon grain and uses a full spectrum pattern match to identify the zircon grains [Fandrich et al., 2007]. We used a probability value of 10−30 or greater to indicate a positive match. The similarity between the reference spectra and the unknown is calculated using a pattern matching algorithm based on a modified Pearson's chi-square test.

[7] The restrictive probability limit of 10−30 worked well for all samples except one which contained small baddeleyite crystals that were misidentified because of a mixed EDS spectra of baddeleyite and adjacent silica. In this sample, the problem was resolved during subsequent CL imaging where the baddeleyite was identified by its crystal shape and EDS spectra. Due to the lack of baddeleyite reference material in our lab, at the time of this investigation, and the low geologic importance of this sample, we did not investigate this sample past the mapping stage. The misidentification of baddeleyite was later remedied by adding baddeleyite to the mineral reference list so that it is identified and distinguished from zircon in more recent analyses. However, in this study we have no data as to how many of our analyses contained baddeleyite; in most samples within this study, if baddeleyite was present, it would have been identified as an “unknown mineral.”

[8] Locations of the zircon grains identified during SEM-EDS mapping were recorded relative to a set of reference points permanently scribed on the sample (Figure 1a). The reference points were registered on the LA-ICP-QMS and used to relocate the grains. The accuracy of this location system was better than 50 μm, allowing us to quickly locate the zircon grains for ablation. Cathodoluminescence (CL) images of the zircon grains were collected after the SEM-EDS mapping using a Gatan PanaCLF detector with a blue filter. Secondary electron (SE) images were also acquired for all grains. Figure 1 shows the BSE image of the entire mount which is stitched together from individual frames (∼500), each with a resolution of 1024 × 800 pixels as well as a CL image of two of the larger in situ grains and a portion of the classified zircon map.

Figure 1.

Examples, from PSA05-004, of various image types generated during the zircon location phase of the dating process. Note that shallow ablation pits are visible in Figures 1b, 1c, and 1d because this image was acquired after these two zircons were shallowly ablated to confirm their identification. All other samples in this study, and most analyses in PSA05-004, were imaged prior to ablation. (a) High-resolution composite BSE image with location of zircon grains analyzed in this study marked by white circles. (b) CL image (top) and SE image (bottom) of zircon grains CL2.1 and CL2.2. These are two of the larger zircon grains found in situ. (c) Enlarged portion of BSE image. (d) Enlarged portion of classified zircon map.

[9] The automated SEM-EDS system can check thousands of grains per hour. With this method a single sample (4 cm2) can be completely scanned, with zircon grains larger than 5 μm located, in less than 3 h. Zircon grains smaller than 5 μm are not mapped because of the low likelihood of producing meaningful ages with a 10 μm spot size. Typically up to 13 samples are run unattended over a weekend. Problems arise in samples where a mineral with a similar mean atomic weight (and therefore BSE brightness) to zircon is abundant. For example, pyrite has a similar BSE brightness to zircon and a pyrite-bearing sample may have thousands of pyrite grains for every zircon making the search much less efficient. Though technically still possible to identify zircon within samples having a high abundance of pyrite, it could take over 10 h to collect EDS for all the pyrite grains in a single sample. In our study, to maximize the number of samples that could be mapped we deliberately chose low-pyrite examples where available and automatically stopped the SEM search if the process had not been completed in 4 h.


[10] The LA-ICP-QMS analyses were carried out using a New Wave 193 nm solid state Nd:YAG Q-switched Laser Ablation System. Between 2002 and 2005, barrel-type ablation cells were used for U-Pb zircon work at the University of Tasmania but systematic differences occurred in the U/Pb ratios of reference zircons measured in different parts of the ablation chamber. To minimize this problem, we initially mounted samples and reference zircons in the center of the cell, however, this made analysis slow and small changes in the topography of the mounts affected the gas flow and therefore accuracy of the results. The New Wave ablation chamber was then modified to provide greater sensitivity, faster washout and more controlled gas flow by suspending a small inner chamber (20 mm diameter) over the ablation site, inside the standard New Wave laser ablation cell (Figure 2). This chamber contained the He outlet to the ICP-MS and was attached to the laser through the glass using magnets. The ablated material was then aspirated into the smaller chamber at the ablation site and carried to the ICP-MS. This design worked similarly to the designs of Eggins et al. [2005] and Autrique et al. [2008] in that the gas flow over the ablation site remained constant throughout the cell. The chamber is also similar in concept to the Large Format cell sold by New Wave (ESI). The higher sensitivity of the smaller chamber over the original New Wave cell is due to the close proximity of the exit hole to the ablation site, leading to better transport of aerosols and less deposition on the sample. All ablations for the present study were performed in a He atmosphere and the He gas was blended with Ar immediately outside the sample chamber [i.e., Eggins et al., 1998; Jackson et al., 2004]. Laser conditions and ICP-MS operating conditions are outlined in Table 1. The Agilent 7500cs ICP-MS was tuned daily to optimize counts and minimize oxide and doubly charged ion production (ThO/Th < 0.2%, Ce2+/Ce < 0.5%) after warming up for 2 h. The typical count rate is in the range 40–50 × 104 cps for 238U, on the international standard glass NIST612 (100 μm line rastering at 3 μms−1, 10 Hz and 4 Jcm−2).

Figure 2.

Schematic of custom ablation chamber designed after a similar model used by the Australian National University [Eggins et al., 2005].

Table 1. LA-ICP-QMS Operating Conditions and Data Acquisition Parameters
ICP-MSAnalysis and Tuning 
ModelAgilent 7500cs 
Forward power1200 W 
Gas flows  
   Plasma (Ar)15 l/min 
   Auxillary (Ar)1.0 l/min 
   He0.8 l/min 
   Ar0.95 l/min 
LaserSeparated GrainsIn Situ
Wavelength193 nm193 nm
Repetition rate5 Hz5 Hz
Pre-ablation laser warm-up30 s30 s
Pulse duration (FWHM)4 ns4 ns
Focusing objective
Degree of defocusing00
Spot size35 and 10 μm10 μm
Irradiance0.75 GWcm−20.5 GW cm−2
Fluence3 J cm−21.8 J cm−2
Drilling rate∼0.5 μ ms−1∼0.5 μ ms−1
Useful yield (atoms detected/atoms ablated)0.00012% 238U, 0.00026% 206Pb0.00012% 238U, 0.00026% 206Pb
Data Acquisition ParametersExplanation of Parameters for Separated Grains and In Situ 
Data acquisition protocolTime resolved 
Scanning modePeak hopping, 1 point per peak 
Detector modePulse mode, dead time correction applied 
Isotopes determined49Ti, 96Zr, 178Hf, 202Hg, 204Pb, 206Pb, 207Pb, 208Pb, 232Th, 238U 
Dwell time per isotope49Ti, 10 ms; 96Zr, 5 ms; 178Hf, 5 ms; 202Hg, 10 ms; 204Pb, 10 ms; 206Pb, 30 ms; 207Pb, 30 ms; 208Pb, 15 ms; 232Th, 5 ms; 238U, 10 ms 
Quadrupole settling time∼4 ms 
Time/scan0.1512 s 
Data acquisition (s)60 s (30 s gas blank plus 30 s ablation) 
Typical dead time20–30 ns 
Typical background count rate2 cps 238U, 16 cps 206Pb 
Sample mountsa25 mm diameter polished grain mounts; 25 mm diameter polished rock mounts 

2.3. LA-ICP-QMS Data Reduction, Precision, and Accuracy

[11] All data reduction calculations were done with Microsoft Excel using macros designed at the University of Tasmania. The methods employed are similar to those used at the Australian National University [Harris et al., 2004] and the University of Melbourne [Jackson et al., 2004; Paton et al., 2010]. A sequence of four primary reference zircons, two secondary reference zircons, 12 unknowns, then four primary reference zircons and two secondary reference zircons comprises a single run; each run takes approximately 1 h. The primary reference zircon used to correct for mass bias, machine drift and downhole fractionation on the Pb/U and Pb/Th ratios was TEMORA1 [Black et al., 2003] and the secondary reference zircon was 91500 [Wiedenbeck et al., 1995]. International standard NIST612 was ablated at the beginning of the analytical day for tuning purposes as well as at the end of the analytical day to correct for mass bias on the 207Pb/206Pb ratio. The 207Pb/206Pb mass bias is not determined using the TEMORA1 zircons because the counting statistics on 207Pb are very poor and these grains sometimes have a detectable amount of common Pb (either in apatite inclusions or due to surface contamination). Individual analyses consisted of a 30 s gas blank followed by 30 s of ablation giving approximately 200 mass scans over a maximum penetration depth of approximately 15 μm.

[12] After triggering the laser, the mass spectrometer took approximately two seconds (s) (∼15 mass scans) to achieve a steady signal. The average of the background count rates (first 30 s of analysis) was subtracted from each isotope. Pb and U isotopic ratios were then calculated for each 0.15 s of measurement. These ratios were filtered to exclude the top and bottom 1%, eliminating spikes and spurious data. The resulting filtered ratios were corrected for machine drift, downhole fractionation and mass bias. Machine drift was calculated by fitting a line to the average Pb/U and Pb/Th ratios on the primary and secondary reference data throughout the day. The downhole fractionation and mass bias correction factors [Jackson et al., 2004; Paton et al., 2010] were calculated by averaging the drift corrected ratios on each quadrupole sweep (0.15 s) on the primary reference zircon measurements (relative to the start of the analysis) and fitting a curve to the data for each of the Pb/U and Pb/Th ratios. A set of correction factors was thus generated for each sweep and for each isotopic ratio. The corrected age data for each quadrupole sweep for each analysis was plotted against the analysis time to examine the stability of the calculated ages.

[13] All uncertainties in this paper are quoted at the 95% confidence or 2σ level. Concordia diagrams were constructed and U-Pb concordia ages calculated with Isoplot 3.6 [Ludwig, 2008]. The uncertainties described in this study are based on the measurements within the integration intervals and include the measurement uncertainties of the reference zircons. When strict error propagation is undertaken on the data scatter (collected with a 35 μm spot size), the ages on both primary and secondary reference zircons can be shown to be greater than that predicted from the standard deviation on the signal [Jackson et al., 2004; Paton et al., 2010]. This is most obvious when using the compilation of replicate analyses of reference zircons over a long analysis period. For our secondary reference zircon (91500) [Wiedenbeck et al., 1995] over 12 months (January 2007–December 2007), using the same sample chamber and LA-ICP-QMS parameters (i.e., 5 Hz, 35 μm spot size and 30 s ablation time). The weighted mean of the 207Pb corrected 206Pb/238U ages from repeated analyses of our secondary reference zircon 91500 (n = 465) is 1065.6 ± 1.7 Ma with a MSWD of 2.5 prior to the addition of the excess uncertainty factors [Jackson et al., 2004; Paton et al., 2010]. This result is close to TIMS age (1065 ± 0.4 Ma) [Wiedenbeck et al., 1995] but suggests an excess uncertainty of 1.4% (i.e., an extra 1.4% uncertainty needs to be added to each analysis in quadrature to get MSWD = 1). Results were similar on the TEMORA1 (n = 946, 415.92 ± 0.42 Ma MSWD 2.1, excess uncertainty 1.1%) and Mud Tank zircons (n = 53, 736.7 ± 4.7 Ma, MSWD 2.1, excess uncertainty 1.4%). All three reference zircons indicate that systematic errors on the accuracy of the measurements are likely to be small (0.1–0.4%). Based on these results a 1.4% excess uncertainty was added in quadrature to each of the analyses in this paper.

[14] Accuracy problems with a 10 μm spot size are likely to be larger than at 35 μm spot size because small differences in the ablation rates of different zircons (matrix effects) can have a larger variation in the U/Pb ratios due to increased downhole U/Pb fractionation [Paton et al., 2010]. Black et al. [2004] and Paton et al. [2010] ascribe both accuracy and precision problems to matrix effects on U/Pb fractionation. In our study, the low MSWDs observed on ages calculated from analyses with a 10 μm spot size suggest precision problems are masked by the larger scatter of the time resolved signal. Comparing the ages calculated for the zircons in sample PSA05-004 we estimate that at a 10 μm spot size the error in our ages is up to 0.8%. To correct for this we have added 0.8% systematic uncertainty after calculating the weighted averages.

[15] The short-term precision (∼1 h) of our LA-ICP-QMS zircon dating method was evaluated by repeated analyses of our two reference zircons (TEMORA1, 91500). Typically the uncertainty of the 207Pb corrected 206Pb/238U age of each of the 12 analyses at 35 μm spot size on separated grains of either reference zircon is 1.0% to 1.5%, comparable to that found in other studies [i.e., Gehrels et al., 2008; Simonetti et al., 2006]. The analyses with a 10 μm spot size are less precise than those with a 35 μm spot size and a 206Pb/238U uncertainty of 2.4% to 3.5% was recorded on individual analyses of either reference zircon with the 10 μm spot size.

2.4. Common Pb Corrections

[16] Most of the zircon grains reported in this study plot slightly above concordia on the Tera and Wasserburg [1972] concordia diagram, indicating that they have a small amount of common Pb [Cocherie and Robert, 2008]. The possible sources of this common Pb are the surface of the grains, the carrier gas and from within the zircon grains. To minimize surface Pb contamination during polishing we use fresh sandpaper sheets and clean polishing laps between each sample so any common Pb introduced into the cracks during polishing belongs to that sample with little or no cross contamination. We also double wash the samples in an ultrasonic bath after polishing. Washing the samples in acids is not recommended for in situ work due to the potential for the dissolution of galena and other high-Pb crystals which could increase the Pb contamination. To minimize common Pb contamination in the carrier gas the average of the background count rates (first 30 s of analysis) was subtracted from each isotope.

[17] The 207Pb correction method was used to correct for common Pb from within the zircon [Compston, 1999]. The crustal growth model of Stacey and Kramers [1975] was used to estimate 207Pb/206Pb for common Pb at the inferred crystallization age. The applicability of Stacey and Kramers' [1975] model was tested on 3 Alaskan samples by analyzing a fine-grained portion of the rock matrix with a large spot size (100 μm, 10 Hz and 1.4 Jcm−2). In all the samples, minor Pb (1 to 2.5 ppm) and U (3 to 6 ppm) with variable Th (7 to 68 ppm) were detected with a Pb isotopic composition similar to that of Stacey and Kramers' [1975] model. In samples with unusual whole rock Pb isotopic compositions care must be taken to make sure that the common Pb composition used in the calculations is appropriate. The alternative 204Pb correction which is widely used in thermal ionization and secondary ion mass spectrometry is not used because of significant isobaric interference from 204Hg present in the He and Ar gases [i.e., Compston, 1999; Jackson et al., 2004]. Some laboratories with multicollector instruments dedicated to U-Pb analysis have managed to reduce the Hg interference to the point where the 204Pb correction can be used [i.e., Gehrels et al., 2008]. However, our instrument, and most quadrupole instruments around the world, are used for a variety of purposes and as a result have a relatively high Hg background (500–1000 cps) so that the 207Pb correction method of Compston [1999] remains the most precise method of correcting for common Pb.

2.5. Choosing Integration Intervals

[18] The integration interval is the portion of the analysis, for both reference material and unknowns, which is used to calculate the age of the zircon grain. In reference zircon grains, the integration interval is typically the middle 25 s of a 30 s analysis. The integration interval for the unknowns in this study was chosen based on four criteria: (1) the analyses are examined for evidence of inclusions (e.g., high 49Ti, 204Pb and 207Pb/206Pb), portions of the analyses interpreted as being inclusion free were selected, (2) the analyses are examined for the stability of the measured Pb ratios with portions having stable ratios over at least 20 mass scans (>3 s) selected [Harris et al., 2004], (3) grains or zones were rejected where the calculated α radiation loading at the inferred age exceeded the critical value of Geisler et al. [2003] (see discussion), and (4) based on the shape of the probability plots, analyses which were significantly different (much greater than two standard deviations from the mean) were rejected as having either inherited (xenocrystic) components or having suffered partial loss of Pb [Harris et al., 2004].

3. Small Grain Systematics

[19] Our protocol uses a 10 μm spot size which is toward the smaller end of what is currently accepted for LA-ICP-MS zircon analysis. It was therefore imperative to test that our method worked effectively. The most robust way to carry out such testing was to take a sample of known age and compare the results from small spot size analysis on both large zircons and in situ small zircons with conventional 35 μm spot size LA-ICP-MS analysis. To provide the standard age a TIMS analysis was carried out on sample PSA05-004. This sample is from a coherent Triassic rhyolite located in the Alexander terrane, 30 km south of Juneau, Alaska. The sample was chosen because both small and large zircons are abundant and no evidence of zircon inheritance was detected during conventional LA-ICP-QMS analysis.

[20] For both the TIMS and the conventional LA-ICP-QMS dating the zircon grains were prepared by crushing 300 g of rock in a Cr steel ring mill and sieving to a grain size of <425 μm. The heavy minerals were separated using a gold pan and the magnetic fraction was removed using a Fe-B-Nd hand magnet. Zircon grains were hand picked from the concentrate. The zircon grains for conventional LA-ICP-QMS analysis were mounted in a 2.5 cm diameter epoxy mount and grains free of inclusions and fractures were selected for analysis. The polished rock fragments are 1 cm thick and it was not possible to view the in situ zircon grains with transmitted light microscopy. In situ zircon grains were selected on the basis of the CL images.

[21] Large zircon grains, 50 μm to 100 μm in length, for TIMS analysis were selected based on grain morphology, quality, size and low magnetic susceptibility. The grains were clean and unaltered with no evidence of older, inherited cores. The detailed analytical procedures for U-Pb geochronology, including the chemical abrasion isotope dilution-thermal ionization mass spectrometry (ID-TIMS) technique applied to single grains of zircon, are described in a digital supplement to the study by Scoates and Friedman [2008]. All mineral treatment protocols and TIMS analyses were carried out in the Pacific Centre for Isotopic and Geochemical Research (PCIGR) at the University of British Columbia, Vancouver, Canada. All analytical uncertainties were propagated through the age calculation using the technique of Roddick [1987]. The 206Pb/238U dates for five grains are overlapping and are concordant to within uncertainty (Table 2). The weighted mean of the 204Pb corrected 206Pb/238U dates from all five grains is 226.86 ± 0.24 Ma with a MSWD of 0.49.

Table 2. ID-TIMS Technique Ages of Five Zircon Grains
GrainaWtb (μg)Uc (ppm)Pb*d (ppm)206Pb/204PbePb*/PbcfPbg (pg)Th/UhIsotopic Ratios ±1σ, %iRhoDiscordantj (%)Apparent Ages ±2σ, Mak
  • a

    Number of grains listed after fraction name.

  • b

    Grain mass determined after leaching on Sartorious SE2 ultramicrobalance to ±0.1 μg.

  • c

    U is corrected for spike, blank (0.2 pg ± 50%, 2 sigma), and mass fractionation, which is directly determined with 233U-235U spike.

  • d

    Radiogenic Pb corrected for spike, fractionation, blank, and initial common Pb. Mass fractionation correction of 0.23%/amu ± 40% (2σ) is based on analysis of NBS-982 throughout course of study. Blank Pb correction of 1.0–2.5 pg ± 40% (2σ) with composition of 206Pb/204Pb = 18.5 ± 2%, 207Pb/204Pb = 15.5 ± 2%, and 208Pb/204Pb = 36.4 ± 2%, all at 2σ. Initial common Pb compositions are based on Stacey and Kramers' [1975] model Pb at 227 Ma.

  • e

    Measured ratio corrected for spike and fractionation.

  • f

    Ratio of radiogenic to common Pb.

  • g

    Total weight of common Pb calculated with blank isotopic composition.

  • h

    Model Th/U ratio calculated from radiogenic 208Pb/206Pb ratio and 207Pb/206Pb age.

  • i

    Corrected for spike, fractionation, blank, and initial common Pb.

  • j

    Percent discordance to origin.

  • k

    Calculations are based on decay constants of Jaffey et al. [1971].

    CA1 11.5576.221.167410.82.90.4280.03581 ± 0.090.2496 ± 0.50.05055 ± 0.480.450−3.0226.8 ± 0.4226.3 ± 2.1220.4 ± 22.1/22.4
    CA2 24.0478.717.2181029.02.40.3690.03582 ± 0.100.2509 ± 0.20.05081 ± 0.170.6262.3226.9 ± 0.4227.3 ± 0.9232.2 ± 7.6/7.7
    CA3 11.7502.018.179612.52.40.3700.03577 ± 0.140.2513 ± 0.40.05094 ± 0.390.4625.0226.6 ± 0.6227.6 ± 1.8238.2 ± 17.9/18.1
    CA4 13.8424.515.4261442.51.40.3910.03587 ± 0.180.2500 ± 0.30.05054 ± 0.170.743−3.3227.2 ± 0.8226.6 ± 1.0220.1 ± 7.8/7.9
    CA5 13.1287.410.41448. ± 0.170.2511 ± 0.330.05079 ± 0.270.6041.9227.1 ± 0.8227.5 ± 1.4231.3 ± 12.3/12.4

[22] Twenty four zircon grains from the rhyolite were analyzed using a conventional LA-ICP-QMS technique (spot size of 35 μm) outlined by Meffre et al. [2007, 2008]. These grains had U contents between 230 ppm and 1570 ppm (Table 3) and the radiogenic Pb contents were between 7 ppm and 48 ppm. The CL images (Figure 3) of the zircon grains indicate two growth stages: (1) an oscillatory zoned core with a weakly resorbed margin and (2) an oscillatory zoned outer rim with CL banding offset from that in the core. No age difference was detected between these two zones. All but three analyses lie on or close to Concordia (Figure 4). Two zircon analyses (asterisked in Table 3) were rejected from all calculations based on the shape of the probability plot (Figure 4). Close inspection of the CL and SE images of these grains identified a hairline fracture in the area where each 35 μm spot was ablated. The weighted average of the 207Pb corrected 206Pb/238U ages is 224.6 ± 1.1 Ma with a MSWD of 0.94 prior to the application of the excess uncertainty factors; this is 0.96% lower than the TIMS estimate. The addition of the additional uncertainty factors (1.4% excess and 0.8% systematic) documented earlier decrease the precision to ±3.5 Ma with an MSWD of 0.40.

Figure 3.

Selection of representative zircon grains from PSA05-004 analyzed with the conventional LA-ICP-QMS technique. CL images are on top, and corresponding SE images are on the bottom of each photo pair. The top three zircon grains (CL13, CL28, and CL43) show evidence of two stages of growth. Bottom three zircon grains (CL1, CL6 and CL23) show only one stage of growth. Large circles show approximate location of 35 μm spots (cores) and small circles 10 μm spots (cores and rims).

Figure 4.

(a) Tera and Wasserburg [1972] concordia diagram for 24 zircon grains from PSA05-004 analyzed with the conventional method of Meffre et al. [2007] using a 35 μm spot size. (b) Probability plot of 207Pb corrected 206Pb/238U ages. Closed circles represent analyses used in final age estimate while open circles show analyses rejected, based on the shape of the inset probability plot, from final age estimate. Data point uncertainty crosses are 2σ.

Table 3. Conventional LA-ICP-QMS Ages (35 μm Spots) of 24 Zircon Grains From Sample PSA05-004a
ZirconAnalysis207Pb/206PbRSE206Pb/238URSE207Pb/235URSE208Pb/232ThRSERadiogenic 206Pb* (ppm)Fraction of 206Pb Common (%)Approximate Surface Areab (μm2)Integration Durationc (s)U±1σTh±1σTh/UHf±1σZrd/Hfα Decay Dosee (α Events/mg)206Pb/238U Age (207 Corrected)±2σPb204f (cps)Pb206 (cps)Pb207 (cps)Pb208 (cps)Th232 (cps)U238 (cps)
  • a

    TIMS age 226.86 ± 0.24 Ma (2σ). Sample PSA05-004 is located at 58.059°N, 134.647°W. RSE, relative standard error.

  • b

    Approximate surface area (μm2) of zircon exposed on polished surface.

  • c

    Duration of integration interval used in all calculations.

  • d

    Zr = 477,600 ppm [Claiborne et al., 2006].

  • e

    Radiation load calculated using the formula of Murakami et al. [1991].

  • f

    Counts per second.


[23] The same 24 grains were analyzed by LA-ICP-QMS using a 10 μm spot size (Figure 5 and Table 4). Seven laser spots were targeted at zircon rims identified using the CL images. The weighted means for the core, rim and single growth zone zircon grains are all within uncertainty of each other (Table 5) though the grains with a single growth zone are slightly younger with a smaller uncertainty than the ages from core and rim zones. The weighted mean 206Pb/238U age (corrected for 207Pb) of the single zone zircons is 221.4 ± 5.1 Ma with a MSWD of 1.04, from cores only is 224.2 ± 7.4 Ma with a MSWD of 0.43, from rims only is 223.7 ± 6.6 Ma with a MSWD of 0.18 and from all 30 combined is 222.5 ± 4.2 Ma with a MSWD of 0.74. All of the calculated ages are within uncertainty of each other with the exception of the age of the single growth zircons which is 0.1% younger than the ID-TIMS result. The analyses carried out with the 10 μm spot size have up to 6% uncertainty on individual spots and are therefore less precise than the conventional methods.

Figure 5.

(a) Tera and Wasserburg [1972] concordia diagram for 30 zircon grains from PSA05-004 analyzed with all parameters the same as the conventional LA-ICP-QMS method of Meffre et al. [2007], except we used a 10 μm spot size. (b) Probability plot of 207Pb corrected 206Pb/238U ages. Black circles represent analyses from single growth zircons, gray from rims only, and white from cores only. All analyses were combined in the final age estimate. Data point uncertainty crosses are 2σ.

Table 4. Small Spot LA-ICP-QMS Ages (10 μm Spots) From 30 Analyses of 24 Zircon Grains From PSA05-004a
ZirconAnalysis207Pb/206PbRSE206Pb/238URSE207Pb/235URSE208Pb/232ThRSERadiogenic 206Pb* (ppm)Fraction of 206Pb Common (%)Approximate Surface Areab (μm2)Integration Durationc (s)U±1σTh±1σTh/UHf±1σZr/Hfdα Decay Dosee (α Events/mg)206Pb/238U Age (207 Corrected)±2σPb204f (cps)Pb206 (cps)Pb207 (cps)Pb208 (cps)Th232 (cps)U238 (cps)
  • a

    TIMS age 226.86 ± 0.24 Ma (2σ). RSE, relative standard error. Additional footnotes are the same as in Table 3.

Weighted mean                      222.5 ± 4.2 Ma       
MSWD                      0.74       
Table 5. Summary of Results From the Test (PSA05-04)
Method/Spot SizeZoneaCrystalsbRejected AnalysescAnalyses UseddAgee (Ma)2SDe (Ma)2SDe (%)MSWDe
  • a

    Crystal growth zone targeted during analysis.

  • b

    Number of individual grains analyzed.

  • c

    Analyses were rejected based on the criteria stated in section 2.5.

  • d

    Number of analyses used in the estimation of the age of the sample.

  • e

    Mean weighted average and uncertainties calculated using the 207Pb corrected 206Pb/238U data of the most coincident analyses when plotted on a Tera and Wasserburg [1972] concordia diagram.

TIMS/NAsingle growth505226.860.240.110.49
LA-ICPMS/35 μmall growth zones24222224.
LA-ICPMS/10 μmsingle growth18018221.
LA-ICPMS/10 μmcore only505224.
LA-ICPMS/10 μmrim only607223.
LA-ICPMS/10 μmcombined24030222.
LA-ICPMS/10 μm (in situ)all growth zones482919223.

[24] A 2.5 cm diameter mount of the rhyolite contained 200 exposed zircon grains ranging in surface area from 20 μm2 to >10,000 μm2. The zircon grains were divided into four groups based on surface area (<100; 100–150; 150–1000; and >1000 μm2). Twelve grains were selected from each of these size ranges ( Table 6). The CL images of the in situ zircon grains show the same variable zoning as the large separated zircon grains in the grain mounts. However, there is a higher proportion of single growth zircon grains in situ compared to that in the grain mounts.

Table 6. In Situ LA-ICP-QMS Ages From Analyses of 48 Zircons in PSA05-004a
ZirconAnalysis207Pb/206PbRSE206Pb/238URSE207Pb/235URSE208Pb/232ThRSERadiogenic 206Pb* (ppm)Fraction of 206Pb Common (%)Approximate Surface Areab (μm2)Integration Durationc (s)U±1σTh±1σTh/UHf±1σZr/Hfdα Decay Dosee (α Events/mg)206Pb/238U Age (207 Corrected)±2σPb204f (cps)Pb206 (cps)Pb207 (cps)Pb208 (cps)Th232 (cps)U238 (cps)
  • a

    Analyses with an asterisk are rejected from the weighted mean on the basis of criteria stated in section 2.5. TIMS age 226.86 ± 0.24 Ma (2σ). RSE, relative standard error. Additional footnotes are the same as in Table 3.

Weighted mean                     223.2 ± 7.4 Ma        
MSWD                     0.21        

[25] When analyzed with a 10 μm spot size, the zircons show a very strong increase in U and Hf content at smaller grain sizes (Figure 6). When all 10 μm analyses from PSA05-004 (both in situ and separated grains) are considered, Hf and U content ranged from 7700 ppm to 15,000 ppm and 300 ppm to 10,000 ppm, respectively, and generally increased as grain size decreased. The Zr/Hf ratios ranged from 32 to 62 with lower values associated with smaller zircon (Figure 7). The high-U grains appear to have exchanged Pb with their surroundings due to radiation damage. In these zircons, the Pb isotope data suggest that the higher-U zircons not only have lost radiogenic Pb they also appear to have gained common Pb. This Pb exchange explains the slight upward trend of the analyses of the high-U zircons on the right side of Figure 8a. All grains with less than 2600 ppm U are nearly concordant and show no evidence of Pb loss; the smallest grain with <2600 ppm U was 100 μm2. The weighted mean 207Pb corrected 206Pb/238U age of the 19 points with U below 2600 ppm is 223.2 ± 7.4 Ma with a MSWD of 0.21 (Figure 8), statistically identical to the conventional LA-ICP-QMS and ID-TIMS estimates for this sample. These results demonstrate there are no significant systematic errors introduced by using the smaller spot size but they do highlight the problem that small zircon grains can have higher U contents and are more likely to be metamict. In our experience the negative correlation of U with grain size is common but not universal.

Figure 6.

U content of small zircon grains in sample PSA05-004 measured in situ using our Agilent 7500cs ICP-QMS with a 10 μm laser spot size. U content uncertainty at 1σ are smaller than the symbols.

Figure 7.

Zr/Hf versus exposed area of zircon from sample PSA05-004. Zr/Hf is calculated using Zr = 477600 ppm [Claiborne et al., 2006], and Hf is measured using our Agilent 7500cs ICP-QMS with a 10 μm laser spot size; see Table 6 for 1σ uncertainty on Hf measurements.

Figure 8.

(a) Tera and Wasserburg [1972] concordia diagram for 48 zircon grains from PSA05-004 analyzed with the new in situ method using a 10 μm spot size. Data point uncertainty crosses are 2σ. Inset in top right only includes analyses used in final weighted mean. (b) Probability plot of 207Pb corrected 206Pb/238U ages. Closed circles represent analyses used in final age estimate, while open circles show analyses rejected from final age estimate. Data point uncertainty crosses are smaller than the symbols.

[26] The age estimates from the ID-TIMS technique and the various LA-ICP-QMS methods used on the test sample in this study are generally in good agreement (Table 5). The 10 μm spots on large grains have twice the analytical uncertainty but return the same age. This indicates that the small volumes associated with the 10 μm spots can be successfully analyzed using our LA-ICP-QMS method.

4. Mafic Rocks

[27] The most useful application of the protocol developed here is for dating medium to fine-grained mafic volcanic rocks. Preliminary studies have been undertaken with identification of zircons in Phanerozoic basaltic rocks from arcs and ophiolites collected from the SW Pacific (Vanuatu, Tonga fore arc), Alaska (Alexander Terrane) and SE Asia (Nan Fault Zone, Ailao Shan ophiolites, Song Ma Fault Zone, Tamky Fault Zone). Details of the mafic rocks analyzed and the success rate at finding and dating zircons are summarized in Table 7.

Table 7. Summary of Characteristics of 63 Samples From the Alexander Terrane in Southeast Alaska, USA
Rock TypeaGrain SizeZr/Ti RangebSamplescSamples With ZrdSamples Datede
  • a

    The rock type of the volcanic rocks is generic because clear-cut textural-based identification of more specific genetic classifications are not possible for this area. Similarly, physical dimensions are quoted for the grain size of the volcanic rocks to avoid genetic inferences.

  • b

    In hydrothermally altered volcanic rocks, immobile elements such as Zr and Ti can be used to subdivide rocks into mafic, intermediate and felsic [Pearce, 1996]. All of the rocks in our study plot in the subalkaline field of Zr/Ti versus Nb/Y space and therefore ultramafic and mafic rocks have Zr/Ti ratios below 0.30, and felsic rocks have ratios above 0.08; intermediate rock are between felsic and mafic. This classification obviously does not apply to the sedimentary rock, but the ratios are given for comparison.

  • c

    Total number of samples within each rock type.

  • d

    Number of samples with zircon identified by SEM-EDS.

  • e

    Number of samples successfully dated using our 10 μm method.

  • f

    One of these samples contained baddeleyite and nine contained zircon; the baddeleyite was not dated.

Ultramafic intrusive>1 mm0.001–0.050300
Mafic volcanic<1 mm0.003–0.0302210f5
Amphibolite>1 mmno data111
Felsic volcanic<1 mm0.160–0.163221
Mafic volcaniclastic<1 mm0.002–0.1432265
 conglomerateno data100
Totals  632315

[28] In our mafic samples U content ranged from 260 ppm to 11,800 ppm and Hf ranged from 8,100 ppm to 17,000 ppm and the Zr/Hf ratios from 28 to 58. Detailed examination of the zircon grain size distributions, identified in this study using the SEM-EDS method, in 19 samples of gabbro and dolerite show no significant difference in zircon abundance and size between the two lithologies. The measured area of zircon grains in these mafic rocks have a strongly skewed distribution and are close to a lognormal distribution. The area of exposed zircons in the Triassic rhyolite lies close to lognormal across three orders of magnitude and the dolerite example is close to lognormal across two orders of magnitude (Figure 9). Below 5 μm diameter the probability distribution dips, suggesting that some grains are not being recognized at the smaller grain sizes; no grains smaller than 2 μm in diameter were recognized. The lognormal parameters estimated from the 19 mafic rocks in this study with more than five zircon grains detected are shown in Figure 10. Based on these statistics, two of these samples (∼10% of gabbro and dolerite samples) have a significant proportion of large zircons and are potentially suitable for conventional zircon analysis.

Figure 9.

Probability plot showing Ln(area) in μm2 of exposed zircon in (a) rhyolite and (b) dolerite. Both samples are from the Alexander Terrane, southeast Alaska. Probability plots were drawn using Isoplot 3.6 [Ludwig, 2008].

Figure 10.

Geometric mean of zircon area in μm2 plotted against volume percent zircon recognized in SEM-EDS analysis. Most grains plot in a discrete trend of increasing size as zircon abundance increases. Samples plotting above the green line have a substantial amount of zircon, and conventional gravimetric methods may recover large zircons from these samples.

[29] Two examples of the usefulness of our U-Pb in situ dating technique for small zircons are given here. The first is a dolerite with poor age constraints in a Mesozoic arc setting from the Alexander terrane of southeast Alaska. The second example applies the technique to dating the eruption age of young mafic igneous rocks dredged from the Tonga fore arc seafloor.

[30] Two samples, PS271-207 and PSA08-001, were collected from a dolerite unit that intrudes the Late Triassic stratigraphic hanging wall to the Greens Creek deposit in southeast Alaska. Prior to our study, the most precise age constraint for this unit came from an apparent U-Pb whole-rock isochron with an age of 206 ± 35 Ma [Premo et al., 2010]. The dolerite is weakly layered to massive, has a granular texture and subhedral plagioclase grains visible to the naked eye. The phenocryst assemblage is plagioclase and clinopyroxene within a groundmass of plagioclase, chlorite, biotite and minor quartz. Both the phenocrysts and the groundmass are altered. The dolerite intruded as sills and dikes [Taylor et al., 2008].

[31] A total of 32 zircon grains were analyzed from the Greens Creek dolerite; 13 from PS271-207 and 19 from PSA08-001 (Table 8). Five analyses were rejected from PS271-207 because the exceeded the α damage critical level (Dc) for a zircon at 226 Ma (see section 5 for further discussion); six analyses were also rejected from PSA08-001 for this reason. Ten other analyses were rejected from PSA08-001 on the basis of never achieving steady Pb or U signals. The weighted mean of the 207Pb corrected 206Pb/238U ages from the eight best analyses from PS271-207 is 222.6 ± 10.5 Ma with a MSWD of 0.62 (Figure 11). The three best analyses from PSA08-001 have a weighted mean 207Pb corrected 206Pb/238U age of 214.3 ± 10.8 Ma with a MSWD of 0.14 (Figure 11). The samples are interpreted to be from the same sill, the Zr/Hf ratios are consistent and the best estimate for the age of each sample agrees within uncertainty. Combining the 11 accepted analyses from both samples gives a weighted mean 207Pb corrected 206Pb/238U age of 218.7 ± 8.0 Ma, with a MSWD of 0.64, for this dolerite sill.

Figure 11.

(a) Tera and Wasserburg [1972] concordia diagram of analyses from zircon grains in samples PS271-207 (black) and from PSA08-001 (gray). White circles are rejected analyses (from both samples). Data point uncertainty crosses are 2σ. (b) Probability plot of 207Pb corrected 206Pb/238U ages. Data point uncertainty crosses are smaller than the symbols.

Table 8. In Situ LA-ICP-QMS Ages From the Greens Creek Doleritea
SampleZirconAnalysis207Pb/206PbRSE206Pb/238URSE207Pb/235URSE208Pb/232ThRSERadiogenic 206Pb* (ppm)Fraction of 206Pb Common (%)Approximate Surface Areab (μm2)Integration Durationc (s)U±1σTh±1σTh/UHf±1σZr/Hfdα Decay Dosee (α Events/mg)206Pb/238U Age (207 Corrected)±2σPb204f (cps)Pb206 (cps)Pb207 (cps)Pb208 (cps)Th232 (cps)U238 (cps)
  • a

    Sample PS271-207 is located at 58.062°N, 134.635°W and PSA08-001 at 58.062°N, 134.646°W. RSE, relative standard error. Additional footnotes are the same as in Table 3.

Weighted mean                       222.6 ± 10.5 Ma       
MSWD                       0.62       
Weighted mean                       214.3 ± 10.8 Ma       
MSWD                       0.14       

[32] Our second example comes from dredge samples from the Tonga fore arc, a young intraoceanic island arc in the southwestern Pacific. The Tonga Trench samples in this study were chosen from a suite of basaltic rocks collected by the Boomerang Cruise of the U.S. research vessel Melville between 7 May 7 to 8 June 1996 between 8000 and 5000 m of water depth in the Tonga Trench wall between 19°S and 15°S (see Table 9 for exact locations). Three of the 13 dolerite and gabbro samples examined by SEM mapping contained small zircon grains. Sample 107-1-7 also contained large zircons and allows another point of comparison between the in situ technique (10 μm spot size) and the more conventional zircon LA-ICP-QMS dating method using a 35 μm spot size.

Table 9. In Situ LA-ICP-QMS Ages of Three Samples From the Tonga Trencha
SampleAnalysis207Pb/206PbRSE206Pb/238URSE207Pb/235URSE208Pb/232ThRSERadiogenic 206Pb* (ppm)Fraction of 206Pb Common (%)U±1σTh±1σTh/UHf±1σZr/Hfbα Decay Dosec (α Events/mg)206Pb/238U Age (207 Corrected)±2σPb204d (cps)Pb206 (cps)Pb207 (cps)Pb208 (cps)Th232 (cps)U238 (cps)
  • a

    Note sample 107-1-7 was analyzed with a 10 μm and 35 μm spot size. Sample 107-1-7 is located at 16.4123°S, 172.311°E, 111-4-6 at 15.419°S, 172.353°E, and 108-3-10 at 16.488°S, 172.454°E. RSE, relative standard error.

  • b

    Zr = 477,600 ppm [Claiborne et al., 2006].

  • c

    Radiation load is calculated using the formula of Murakami et al. [1991].

  • d

    Counts per second.

35 μm Spot Size
Weighted mean                    12.2 ± 0.3 Ma       
MSWD                    0.55       
10 μm Spot Size
Weighted mean                    11.6 ± 1.7 Ma       
MSWD                    0.13       
Arithmetic mean                    10.1 ± 1.6 Ma       
Arithmetic mean                    12.1 ± 6.4 Ma       

[33] Three dredge samples from the Tonga Trench were analyzed using the in situ zircon protocol. The dolerite sample (107-1-7) with both large and small zircons was analyzed using both a standard 35 μm spot size and a 10 μm spot size (Table 9). The weighted mean 207Pb corrected 206Pb/238U age of 12 zircons analyzed with a 35 μm spot size is 12.2 ± 0.3 Ma with a MSWD of 0.55 and for 11 smaller zircons analyzed with a 10 μm spot size is 11.6 ± 1.7 Ma with a MSWD of 0.13 (Figure 12). The similarity of the results show that even at this young age both techniques provide similar ages, though the uncertainty on the in situ data is much higher (15% versus 2.5%).

Figure 12.

(a) Tera and Wasserburg [1972] concordia diagram of analyses using both a 35 μm spot size (gray) and a 10 μm spot size (white) from sample 101-1-7. (b) Probability plot of 207Pb corrected 206Pb/238U ages. Data point error crosses are 2σ both plots.

[34] Only three very small zircons (<100 μm2) were analyzed from sample 111-4-6 (gabbro) and 108-3-10 (dolerite) and a significant amount of the surrounding minerals (chlorite, pyroxene and plagioclase) was ablated during each analysis. The arithmetic mean 207Pb corrected 206Pb/238U age of three zircons analyzed with a 10 μm spot size in sample 111-4-6 is 10.1 ± 1.6 Ma and in sample 108-3-10 is 12.1 ± 6.4 Ma (Table 9). Despite other phases being included in the analysis, the results are similar to those of 107-1-7 (Figure 13) indicating that the samples are close in age.

Figure 13.

(a) Tera and Wasserburg [1972] concordia diagram of analyses using a 10 μm spot size from 11 zircon grains in sample 101-1-7 (black), three from 108 to 3–10 (gray) and three from 111 to 4–6 (white). (b) Probability plot of 207Pb corrected 206Pb/238U ages. Data point error crosses are 2σ in both plots.

5. Discussion

[35] Locating zircon grains using an automated SEM-EDS search is a viable way to find small in situ zircon grains for U-Pb geochronology. The main advantages of this technique are as follows: (1) it requires little sample preparation, (2) it is very good at finding small zircon grains, (3) it provides geochronology for samples which only have small zircon grains, and (4) it limits the potential for contamination from other samples in the laboratory. The technique is not suitable for coarse-grained rocks with large zircon grains as the expected number of grains exposed on one 2.5 cm surface decreases by the square of the grain size. A rock with one 200 μm grain exposed on a typical 4 cm2 rock surface contains 0.01 vol% zircon. If a similar rock had only 20 μm zircon grains, 100 grains would be exposed for the same zircon content.

[36] In our rhyolite sample, PSA05-004, Hf and U content generally increased as grain size decreased and Zr/Hf ratios generally decreased with zircon grain size suggesting these zircons crystallized late from fractionated melts [Claiborne et al., 2006; Wang et al., 2010]. Previous studies have shown that Zr is highly soluble in mafic melts to temperatures well below the solidus of most of these magmas (i.e., 1300 ppm at 930°C and 500 ppm at 860°C) [Harrison and Watson, 1983]. In our mafic samples the correlation of zircon abundance with grain size and the Hf contents and Zr/Hf ratios similar to those expected in felsic rocks [i.e., Claiborne et al., 2006] also suggests that the small zircons form during the final stages of crystallization from small high-silica melt pools strongly enriched in incompatible elements. In certain situations, it is possible to estimate crystallization temperatures from the Ti content in zircon [Watson and Harrison, 2005] but this estimation requires high-resolution characterization of the zircon surfaces, careful consideration of CL banding and consideration of submicron-scale distribution of trace elements [Hiess et al., 2008; Hofmann et al., 2009]. We do not have adequately controlled data to use the Ti content of our zircons to estimate temperature of crystallization. However, based on the Hf content and Zr/Hf ratios of the zircons in this study, we surmise that small nonxenocrystic zircons most likely form, in mafic magma, in small melt pools. This mechanism should also enrich U and Th and may be responsible for the high U content in some of the small zircons in our study [Darling et al., 2009].

[37] Pb loss is a problem which was more prevalent in the small grains analyzed in this study, likely due to the high U content of these grains (Figure 6). The higher U content of the small zircon grains versus the large zircon grains could be caused by small scale differences in the U content of the melt as it cools or may be the result of a hydrothermal overprint in these rocks [Harley et al., 2007]. All Triassic zircon grains analyzed in this study with more than 2600 ppm U content showed features typical of Pb loss. These zircon grains can easily be identified. Not only are they are high in U, but for most grains there is a negative correlation between the apparent 206Pb/238U and U content. There is also a concurrent increase in 207Pb/206Pb with increasing U contents suggesting an exchange of Pb with the environment, not just loss of radiogenic Pb.

[38] There is a general consensus that Pb loss in zircon is controlled by volume diffusion [Hodges, 2004], recrystallization [Hoskin and Black, 2000; Schaltegger et al., 1999] and radiation damage [Geisler et al., 2007; Kramers et al., 2009]. Small zircon grains will be more sensitive to Pb loss by volume diffusion and will have a lower effective closure temperature than larger grains. However, Cherniak and Watson [2000] calculated that the closure temperature for 10 μm zircon grains is over 800°C. The post crystallization temperatures of all samples in this study are lower greenschist facies or less and we saw no evidence that the smaller grains of zircon have lost Pb by normal volume diffusion. The selected samples were at most weakly foliated and only two zircon grains showed evidence of strain. Geisler et al. [2003] argued that Pb loss and open system behavior occurred in zircon grains when the α damage level reached a critical point (Dc). They showed for natural samples this point can be as low as 2.5 × 1015α events/mg. For a zircon grain at 226 Ma this is equivalent to a U content of 3000 ppm at a Th/U of 0.5 (Figure 14). The recognition of Pb loss in zircon grains from the Gallagher rhyolite is consistent with the formation of metamict zircon by α radiation damage.

Figure 14.

Threshold for α radiation reaching 2.5 × 1015α events/mg at the age of rock. Assumes no annealing events and Th/U = 0.5. Radiation load is calculated using the formula of Murakami et al. [1991].

[39] Results from the mafic samples corroborate those found in the rhyolite. Small zircon grains were more likely to have high U content and where the α radiation load has reached 2.5 × 1015α events/mg, open system behavior is ubiquitous. Careful inspection of the data is crucial where the age and U content combine to reach this critical threshold.

[40] The results from the mafic rocks from the Tonga Trench show that this technique can be extended to young magmatic rocks with Zr whole-rock contents as low as 36 ppm. The U contents of the zircons in the Tonga rocks were high (average 950 ppm) but no Pb loss was detected. One of the zircons contained 1.1 wt% U but Pb loss was not identified, probably because the α radiation dose (for a 12 Ma zircon, 1.1%U = 5 × 1014α events/mg) was still well below the critical threshold (Figure 14).

[41] The results of our study illustrate two points regarding Pb loss in the zircon grains in these samples: (1) small zircon grains commonly have higher U contents and therefore Pb loss is more common in these grains, and (2) independent of size or rock type, all zircon grains analyzed in this study which have exceeded the α radiation threshold of 2.5 × 1015α events/mg have undergone Pb loss. The high U of the small zircons is potentially a major limiting factor for the application of this technique in old rocks where radiation damage has more time to accumulate. Our data from mafic rocks suggest that while high-U zircons are common at smaller zircon grain sizes, there are still some low-U zircon grains in this population. In Cenozoic rocks, the high U content of small zircons is an advantage as it generates more radiogenic Pb without the zircon reaching the radiation damage threshold.

[42] Despite its relatively poor precision, our method can produce accurate dates and has already proven useful in areas where chronological data is sparse or nonexistent. The case studies outlined earlier are examples of this. At Greens Creek in the Alexander terrane of southeast Alaska, a number of attempts have been made to find uranium-bearing geochronometers (i.e., zircon, baddeleyite, sphene, monazite) but none could be recovered [Premo et al., 2010]. Using our in situ method on small zircons we were able to refine the whole-rock isochron age of 206 ± 35 Ma [Premo et al., 2010] to 218.7 ± 8.0 Ma. In basalts from the Tonga fore arc zircons dated with our in situ technique confirmed some, but not all, of the Ar-Ar geochronology on low-K, hydrothermally altered rocks indicating that some of the Ar-Ar ages were probably reset during metamorphism.

[43] Our in situ method can provide geologically meaningful ages that could enable new tectonic and geological interpretation in many rocks worldwide. However, there are some limitations: (1) it will not work well in rocks older than 1000 Ma due to Pb loss issues, (2) it will not work well for detrital ages on sedimentary rocks as the poor precision forces reliance on the average of multiple analyses from a single sample, (3) the common 207Pb and 206Pb ratios and concentrations must be well understood, and (4) the technique is relatively time consuming and the data is more difficult to interpret than data from separated zircons with larger spot sizes.

6. Conclusion

[44] Small zircon grains can be routinely recognized in polished rock fragments using automated SEM-EDS mineral mapping software. This search method allows accurate selection of grains and these grains can be efficiently relocated for LA-ICP-QMS analysis. U-Pb geochronology can then be performed, with reasonable uncertainties (5% to 10%) in the 207Pb corrected 206Pb/238U age, on individual small zircon grains using a 10 μm spot size. The protocol outlined here is particularly powerful when used to obtain dates on samples where zircon grains are too small to be easily separated. Small zircons were present in up to 45% of fine-grained lithologies investigated despite these rocks traditionally being considered devoid of dateable zircons and rock types such as andesite, gabbro and dolerite from mid-ocean ridges, island arcs and ophiolites have been dated using this protocol. This technique could also be used to date clasts, xenoliths and dredge rocks where the sample size is restricted as only five grams of rock is required.

[45] The main concern with the technique, aside from the higher uncertainty in measured age, is that in our samples many small zircon grains have high U content. For Late Triassic zircon grains, all zircons with greater than 2600 ppm U showed evidence of Pb loss; for older rocks this problem will be exacerbated. Ongoing studies on other samples have shown that the high U content of small zircons is not ubiquitous and many of these mafic to intermediate igneous rocks with only small zircon grains can be dated using this in situ technique.


[46] We would like to thank the Greens Creek Mining Company, specifically A. West and B. Erickson. This company provided the opportunity and support for this project as well as the permission for the publication of these results. Discussions with J. Mortensen early in the project greatly focused our efforts. Consulting geologist John Proffett first mapped the Gallagher rhyolite without which this experiment could not have been performed at Greens Creek. Many thanks to Karsten Goemann and Maya Kamenetsky at the University of Tasmania Central Science Laboratory, where the SEM-MLA zircon search technique was perfected. Sarah Gilbert, Chris Hollitt, and Leonid Danyushevsky provided superb guidance in all things LA-ICP-MS. A Society of Economic Geologists (SEG) foundation student research grant supported this research. Formal reviews by Axel K. Schmitt and one anonymous reviewer greatly improved this paper.