Here we provide an overview of the procedures used in our lab, as of January 2007. More detail, especially on the U-Th-Sm part of the measurements, can be found here. These methods are constantly changing, as are the configurations and specific details outlined here. Please contact Peter Reiners if you have comments or suggestions.
There are essentially three steps to conventional (U-Th)/He dating:
1) selection, characterization, and preparation of grains for analysis,
2) He extraction and measurement, and
3) U-Th (and in some cases Sm) measurement. Go here for a more detailed description of U-Th-Sm measurement.
Grain selection, characterization, preparation
In our lab, selection of grains from mineral separates is done with one of two Leica MZ16 stereozoom microscopes. Samples can be examined, picked, and checked for inclusions under either cross-polarization with a rotating stage at 160x, or in plane-polarized darkfield illumination at 240x. Darkfield illumination at high power provides a much more robust screening for inclusions in apatite grains. Even apatites with no visually detectable inclusions by normal (crossed-polars) methods almost always show small inclusions in darkfield illumination at high power. In fact it is virtually impossible to find truly inclusion-free grains with darkfield/high-power because minute inclusions are nearly ubiquitous and easy to detect. Still, the abundance and size of inclusions can most easily be reduced using this approach. Selection of zircon and other minerals is not as time-consuming, because inclusions are less of a problem given relative U-Th concentrations and more aggressive dissolution procedures. Both single-crystal and multi-crystal aliquots are routinely dated in our lab; there are advantages and disadvantages to each, and it is not yet clear which provides more reliable results. Multiple replicate aliquot analyses of samples is critical to assessing reliability of (U-Th)/He ages, however, and if single-grain analyses are done, more replicates are usually performed.
Precise grain dimension measurements are made primarily in order to make precise alpha-ejection corrections. However, a second reason is that dimensions can be translated to grain volume and mass, permitting estimates of parent and daughter nuclide concentrations in the dated aliquot. Concentrations are not necessary for age determinations, as these are made by molar parent-daughter ratios. But concentrations can be useful for contextural and quality-control reasons (e.g., radiation damage, anomalous results, etc.).
Grain dimensions are measured from digital photomicrographs taken in at least two different orientations, usually perpendicular to the c-axis. Imaging software that is manually calibrated and recorded using a stage micrometer at each operator sitting is used to measure several features of the grain. For zircon we use the approach outlined in Hourigan et al. (2005), including the bipyramidal tip heights and two different c-axis parallel widths; for apatite we use a similar approach but assume pinacoidal terminations (Farleyet al., 1996; Farley, 2002). In both cases, fractures and other features are noted and, if appropriate, accounted for by decreasing the surface-area-to-volume ratio of the grain by a geometric factor. Alpha-ejection correction factors are calculated using surface-area-to-volume ratios and the approach of Farley (2002) for apatite, Hourigan et al. (2005) for zircon, and other similar approaches for other minerals (e.g., Min et al., 2006).
Because direct lasing of minerals volatiilizes parent nuclides, grains are wrapped in metallic foil “microfurnaces” for laser heating (House et al., 2002). Apatite is placed in 1-mm Pt tubing and the ends are pinched closed; zircon and other minerals requiring HF-HNO3 -HCl dissolution are placed in 1-mm Nb foil envelopes. We are currently testing the purity of Nb tubing we recently purchased. If it is sufficiently low in He, U, Th, and Sm, we will be able to avoid time-intensive folding of Nb envelopes from foil sheet.
He extraction and measurement
Approximately 30 crystal-bearing foil packets are placed in a Cu or SS planchet, under a KBr coverslip, inside a ~7-cm laser cell pumped to <10 -9 torr. For Nd:YAG laser degassing we use a sapphire window; for CO 2 laser degassing we use a Cleartran ® (ZnS) window (and no coverslip). Samples are heated for 3-15 minutes by a focused beam of a 1-2 W (Nd:YAG) or 5-15 W (CO2 ) laser. In our current procedures, neither temperature nor wavelength of radiation emitted from the foils are quantified. However experiments with materials of known melting temperatures in Pt foils, with both qualitative visual correlation of these conditions with CRT images of incandescent foils and quantitative pyrometric measurements (albeit with a poorly constrained Pt emissivity parameter) constrain our routine laser heating to temperatures of ~900-1000 °C for apatite and ~1000-1250 °C for zircon. Most importantly, routine reheating and analyses (“re-extracts”) are performed on most apatites and all zircons (sometimes multiple times), to confirm that 4He has been quantitatively extracted (at least to below blank level) in apatite, and zircons yield less than 1-2% in subsequent re-extracts.
Gas released from heated samples is then spiked with 0.1-0.2 pmol 3He, and condensed onto activated charcoal the cold head of a cryogenic trap at 16 K. Helium is then released from the cold head at 37 K into a small volume (~50 cc) with an activated Zr-Ti alloy getter and the source of a Balzers quadrupole mass spectrometer (QMS) with a Channeltron electron multiplier. Peak-centered masses at approximately m/z (“masses,” though the presence of multiply charged species is recognized) of 1, 3, 4, and 5.2 are measured. Mass 5.2 establishes background, and mass 1 is used to correct mass 3 for HD and H3+ . Experiments relating masses 1, 2, and 3 with no He in the system generally showed better correlations between masses 1 and 3 (non- 3He-mass 3 = 0.005 ± 0.002 times mass 1, though this breaks down at higher gas pressures) than between masses 2 and 3, possibly because of other, multiply charged, species at mass 2. Corrected ratios of masses 4 to 3 are regressed through ten measurement cycles over ~15 s to derive an intercept value, which has an uncertainty of 0.05-0.5% over a 4/3 range of ~10^3 (approximately 10x the ratio of blank to typical Durango apatite standard ratios), and compared with the mean corrected ratio to check for significant anomalous changes in the ratio during analysis.
Helium contents of unknown samples are calculated by first subtracting the average mass-1-corrected 4/3 measured on multiple procedural blanks analyzed by the same method (a “hotblank”), from the mass-1-corrected 4/3 measured on the unknown. This is then ratioed to the the mass-1-corrected 4/3 measured on a shot of an online reference 4He standard analyzed with the same procedure [minus the mass-1-corrected 4/3 measured on a 3He-only spike shot analyzed using the same procedure as the reference 4He standard (a “lineblank”)]. The resulting ratio of measured 4/3's is then multiplied by the moles of 4He delivered in the reference shot.
This procedure assumes linearity between measured 4/3 and 4He pressure, which has been confirmed over the the vast majority of the range of 4He contents we analyze by performing multiple replicate analyses of known-age standards with masses and therefore 4He yields ranging over three orders of magnitude. This procedure also relies on the accuracy of the 4He delivery from the reference standard and the precision of its measurement with the 3He spiking procedure. The delivery and its depletion with time are calibrated by multiple capacitance manometry measurements of the volumes of the reference tank and pipette, and the final filling pressure of the tank. One of our two He lines has a 4He tank and pipette with volumes of 15920 ± 7.8 cc (0.05%, 1s ) and 0.9439 ± 0.0016 cc (0.16%, 1s ), respectively; uncertainties reported as standard errors on multiple (n=6) manometric volume determinations. The other He line has tank and pipette volumes of 3655 ± 8.2 cc (0.2%, 1s ) and 0.09675 ± 0.0021 cc (2%, 1s ). We also have a similarly calibrated detachable portable tank that can be moved between lines for cross-calibration.
Between ~2-6 (depending on the number of unknowns) 4/3 measurements of spiked 4 He reference standards are made each measurement day. Although the long-term, day-to-day change in these measurements can be large due to drift of the QMS, in a single measurement day the corrected 4/3 measurements on reference standards vary by less than 0.5% (1s ). This uncertainty can be reduced for reducing unknown data by monitoring intraday secular trends. Average measured 4/3 of lineblanks ( 3 He spike only) are nearly indistinguishable from that predicted by the purity of the 3 He spike (99.75% 3 He). Hotblanks, or procedural blanks measured by lasing/heating empty Pt or Nb foil packets are typically 0.05-0.1 fmol 4He.
Following He measurement, foil packets are retrieved, transferred to Teflon vials, and spiked with a 50 ml shot of a mixed spiked containing 7.55 ± 0.10 ng/ml 233 U and 12.3 ± 0.10 ng/ml 229Th and, if apatite, 50 ml of a 97%-enriched 147Sm spike with 10.8 ± 0.10 ng/ml Sm . Apatite is dissolved directly from the foil in dilute (~20%) warm HNO3 , then diluted with 2.5 ml of 18 MO H2O to a final concentrations of ~0.1-0.2 ppb 229Th and 233U. Our current routine method for zircons and most other minerals uses high-pressure digestion vessels for dissolution of the entire grain-bearing Nb foil packet. Although this process is time-consuming, it insures quantitative recovery of the grain (or, for some minerals, the powder or glass remaining), has low U-Th blanks, and has been proven analytically reliable.
Natural-to-spike isotope ratios are measured on a high-resolution (single-collector) Element2 ICP-MS with all-PFA Teflon sample introduction equipment and sample preparation/analytical equipment. Careful monitoring of procedural blanks and spike and normal concentrations (including evaporative effects) and isotopic compositions is essential for He dating. Our routine U-Th procedural blanks for apatite are 0.6 ± 0.4 pg U and 1.0 ± 0.2 pg Th (uncertainties as 1s standard error). For zircon procedures blanks are 2.6 ± 0.5 pg U and 5.5 ± 1.0 pg Th . For reference a single apatite crystal with a common size (c-axis perpendicular half-width of 75 m m and aspect ratio of 2.5) and U concentration of 0.5 ppm would have a U content of 0.01 ng, ~17 times higher than the apatite U blank. Similarly, routine blanks for zircons are orders of magnitude lower than U-Th contents of even extremely small zircons with very anomalously low U-Th concentrations., given the much higher U-Th concentrations in typical zircons. Long-term (~4-yr) reproducibility of our spiked normal solutions (used to calibrate spike concentrations) is 0.8-1.0%, and new, fresh normals are periodically purchased to mitigate against evaporative effects. For the vast majority of apatites we measure, which have U and Th contents greater than 0.01 ng, precision on measured U-Th ratios is better than 0.5%. Zircons yield similar precision, even though concentrations are about ten times higher, probably due to ionization suppression by high Nb contents. Go here for more information on our U-Th-Sm analytical procedures and summaries of a large number of analyses, including measurement precision and U-Th contents and concentrations for several thousand apatites and zircons, and discussions of the importance of Sm in apatite, and other important aspects of parent nuclide measurement.
Percent relative standard deviation (RSD) on measured 238U/ 233U and 232Th/ 229Th (natural to spike isotope ratios) in apatite (left 2 panels) and zircon (right 2 panels). Apatite: approximately 2600 igneous and metamorphic apatite specimens (blue), 136 meteoritic phosphates (apatite and whitlockite; red), 169 Durango apatite specimens (green), and 173 biogenic apatite specimens (conodonts and fossil tooth enamel) (yellow). Highly linear trends at low %RSD are artifacts caused by rounding of RSD digits by ICP-MS software. Zircon: approximately 1200 igneous zircon specimens (blue); 114 Fish Canyon Tuff zircons (red).
Propagated analytical uncertainties for most typical apatite and zircon samples lead to an estimated analytical uncertainty on (U-Th)/He ages of approximately 1-3% (1s ). In some cases, reproducibility of multiple aliquots approaches analytical uncertainty. Eight analyses of single crystal zircons from a xenolith erupted from a basaltic vent average 157 ka with one standard deviation of 4 ka (2.6%) (Blondes et al., 2007), and fourteen chips of large gem quality crystals of Sri Lanka zircon average 442 Ma with one standard deviation of 10 Ma (2.3%) (Nasdala et al., 2004). In general, however, reproducibility of repeat analyses of (U-Th)/He ages is significantly worse than analytical precision. Cooling ages of apatite and zircon from igneous rocks typically show scatter on the order of one standard deviation of at least 6%, and in many cases more than 10%. This has several possible origins, including variable He diffusion characteristics among grains, unidentified intracrystalline inclusions that prevent complete U-Th-Sm recovery following degassing, or petrographic siting effects such as He implantation from adjacent high-U-Th phases or varying He retention due to varying diffusivity or partitioning of surrounding phases. As discussed above, however, it is likely that a major origin of the observed poor reproducibility comes through uncertainty in the alpha-ejection correction (Farley et al., 1996)—not from uncertainty in actual dimensions of dated grains, but uncertainties in relating observed grain boundaries of dated aliquots to original boundaries in the host rock, implantation from other phases, and inhomogeneous distribution of parent nuclides in dated grains (Farley et al., 1996; Reiners et al., 2004; Hourigan et al., 2005). It is extremely difficult to estimate, a priori, the expected magnitude of error arising from any of these potential sources. Thus He ages typically show a much greater scatter and higher MSWD than expected based on analytical precision alone. In any case, an important take-home message of this is that multiple replicate analyses of (U-Th)/He ages on several aliquots is necessary for confidence in a particular sample age.
In our lab we use several routine standards for checking analytical procedures and calibrations, and all visitors pick and analyze at least several standards as part of their first sample batch. These include apatite from the Coast Mountains of British Columbia (House et al., 2000) and our own sample from the Sierra Nevada, and zircon from the Peach Springs Tuff, and gem-quality detrital samples from Sri Lankan (Nasdala et al., 2004). A Figure below shows results of age determinations on aliquots of two other commonly used standards: those of small (~20-200 um) fragments of much larger (~2-3 cm) gem-quality crystals of Durango apatite, and whole single crystals of Fish Canyon Tuff (FCT) zircon.
Durango apatite, while useful as an analytical standard, is atypical of normal apatites, especially in its high Th/U (~15-22). Replicate aliquots of this standard yield an average age of 31.9 Ma, with two standard deviations of 2.2 Ma (6.6%), and a weighted mean age and error of 31.94 ± 0.17 Ma (95% confidence interval, with a 2 s required external error of 1.9 Ma or 5.9%, MSWD = 5.4). This mean age is within error of, but slightly older than, the 31.4 ± 0.5 (2s ) weighted mean of K/Ar ages of feldspars from volcanic rocks bracketing the Durango apatite deposit (Naeser and Fleischer, 1975) and recalculated using Steiger and Jager's (1977) decay constants (Green, 1985). Green (1985) favored an even younger (but still within error) fission-track age for Durango of 30.8 ± 1.7 (2s ) Ma. More recent measurements by McDowell et al. (2004) yielded a bracket age from 40Ar/39Ar analyses of 31.44 ± 0.18 (2s), and 24 (U-Th)/He ages with a mean age of 31.02 with two standard deviations of 2.0 Ma. The reason for the slightly older age obtained from our compilation is unknown, but lack of Sm measurements in about one-third of the analyses may contribute. Sm typically has the effect of reducing the (U-Th)/He age by about 0.3-0.5% in Durango apatite (although two aliquots appear to have Sm age contributions of ~1% and 5%, with reasonable ages). The observed external error on Durango apatite cannot have an origin in alpha-ejection correction, as these grains are small chips of much larger crystals, but U-Th heterogeneity probably still contributes to He heterogeneity and separation of parent and daughters, and therefore age irreproducibility (Boyce and Hodges, 2005).
The other standard is whole single crystals from the Fish Canyon Tuff (FCT), a commonly used volcanic age standard for a variety of systems, including U/Pb and 40Ar/39Ar. Schmitz et al. (2001) obtained high precision U/Pb zircon ages on the FCT of 28.48 ± 0.06 (2s ) Ma, and other high precision zircon U/Pb studies yield similar results (e.g., Bachman et al., 2007). There is evidence suggesting that some zircon growth significantly preceded eruption, however, and Bachman et al. (2007) suggest an eruption age of 28.04 ± 0.18 (2s ) Ma, based on Renne et al.'s (1998) sanidine 40Ar/39 Ar ages, based on the fact that this system would remain open at magmatic, pre-eruption temperatures. One hundred fourteen single grain (U-Th)/He zircon ages average 28.3 Ma, with two standard deviations of 2.8 Ma, and mean age and errors of 28.29 ± 0.26 Ma (95%; 2s external error of 2.6 Ma or 9.3%, MSWD = 20). The observed external error probably has a component in errors associated with the effects of heterogeneous intracrystalline U-Th distribution on the a -ejection correction. Laser ablation depth profiling of FCT zircons show a range of U-Th zonation styles and extents that should lead to an approximately 10% scatter in ages from this effect alone (Hourigan et al., 2005).
Probability density and individual aliquot (U-Th)/He ages (arranged in order of increasing age) of 169 aliquots of Durango apatite (left) and 114 aliquots of Fish Canyon Tuff (FCT) zircon (right) processed in our lab at Yale (left two panels), and 18 Durango apatites just processed in our new lab at Arizona (right panel). Bold green vertical line (Durango apatite panel) and bold red vertical line (FCT zircon panel) are mean ages of replicate aliquots. Other vertical lines represent ranges of two standard deviations of age replicates determined in other studies. For Durango apatite these are: 1) laser-heating (U-Th)/He (red; House et al., 2000); 2) laser-heating (U-Th-Sm)/He (blue; McDowell et al., 2005); 3) apatite fission-track (grey; Jonckheere et al., 1993). For FCT zircon these are: represent 2 s or 95% confidence intervals for U/Pb TIMS ages of FCT zircons (Schmitz and Bowring, 2001) and 40Ar/39Ar ages of FCT sanidine (Renne et al., 1998). The higher standard deviation of FCT zircon (and all typical crystals, as opposed to fragments of much larger gem-quality crystals) relative to Durango apatite likely originates in errors associated with a -ejection correction, including the effects of heterogeneous intracrystalline distribution of U and Th. About two-thirds of the Durango ages from Yale include Sm measurements, which typically decreases ages by 0.3-0.5%. Statistics are from Isoplot 3.0 (Ludwig, 2003) and references therein.
Back to the Arizona Radiogenic Helium Dating Lab page.