1. Introduction

Cosmogenic neon is a relatively little used tool for studying Earth surface processes. It is powerful for four reasons. First, it is produced and retained in quartz (Niedermann et al., 1993, 1994; Shuster and Farley, 2005) as well as most other silicates, such as pyroxene (Schäfer et al., 1999), olivine (Poreda and Cerling, 1992), sanidine (Kober et al., 2005), hornblende and biotite (Amidon and Farley, 2012). Therefore, it is applicable to most rock types found on the Earth’s surface. Second, cosmogenic ²¹Ne is a stable nuclide. This gives it an age range limited essentially only by the erosion rate and allows exceptionally old landscapes to be dated (Schäfer et al., 1999; Dunai et al., 2005). Third, neon has three isotopes (²⁰Ne, ²¹Ne, and ²²Ne), each of which have different abundances in the various reservoirs (atmospheric, nucleogenic, or magmatic) that may contribute to the natural ²¹Ne background (Niedermann, 2002). By simultaneously analyzing all three isotopes and verifying whether they plot on a mixing line between atmospheric and spallogenic components, the cosmogenic neon method provides an internal ‘reliability check’ which is absent from other commonly used nuclides. Fourth, neon can be measured using a standard sector field noble gas mass spectrometer. Sample requirements are modest (typically 100-200 mg) and sample preparation is relatively straightforward as it does not require extensive chemical purification or chromatography. This greatly increases sample throughput, which in turn opens up exciting opportunities for detrital work (Dunai et al., 2005; Codilean et al., 2008).

Cosmogenic ²¹Ne is even more useful when combined with one or more cosmogenic radionuclides such as ¹⁰Be or ²⁶Al. Such double- or triple-dating may be used for burial dating (Balco and Shuster, 2009a; Vermeesch et al., 2010), for catchment-wide erosion studies with complex exposure histories (Kober et al., 2009), or to measure the exposure age of old and very slowly eroding surfaces (Fujioka et al., 2005). An implicit assumption of many of these studies is that the accuracy of the ²¹Ne method equals its analytical precision. Violation of this assumption may lead to erroneous results such as samples plotting in the ‘forbidden zone’ of the ²¹Ne/¹⁰Be two-nuclide diagram (Lal, 1991; Kober et al., 2011). An interlaboratory comparison study was set up in the framework of the CRONUS-EU initiative (Stuart and Dunai, 2009) with the aim to address this issue and provide the noble gas community with a well-characterized reference standard for the analysis of cosmogenic ²¹Ne in quartz. The CREU-1 standard is a mixture of natural quartz pebbles, rich in cosmogenic ²¹Ne, which were crushed and thoroughly homogenized to ensure optimal reproducibility (Section 2). Two size-fractions of CREU-1 were analyzed by five prominent cosmogenic noble gas laboratories, each of which used different experimental setups and data reduction protocols (Section 3). In total, 50 aliquots of CREU-1 were analyzed, with reported analytical precisions of 2-6%, but an external reproducibility of 7.1% (Section 4). These analyses were tied to a further 10 measurements of CRONUS-A, which is a second reference material prepared from an Antarctic quartzite analysed by two of the participating labs (Section 5).

2. Standard material

The CREU-1 standard material is pure quartz prepared from exposed vein-quartz clasts of a Miocene erosion surface (19^∘33’53.4”S, 70^∘7’1.5”W, 930 m) in the Atacama desert near Pisagua, Chile (between sites B and C of Dunai et al., 2005). The clasts were shed onto the surface from local sources after the main sedimentation episode at ~12-14 Ma (Dunai et al., 2005). Approximately 400g of material was mixed from five clasts (sample name CH04/5, pebbles 5, 6, 7, 8 and 13, weighing 81g, 104g, 77g, 109g and 55g respectively) that had ²¹Ne excess concentrations within 5% of their mean value. After crushing in a W-carbide disk mill, five size fractions were prepared using stainless steel sieves:

• 40-62μm: 10.45g, wet sieved and dried overnight at 50^∘C
• 63-125μm: 40.62g, wet sieved and dried overnight at 50^∘C
• 125-250μm: 74.4g, dry sieved
• 250-500μm: 188g, dry sieved
• >500μm: 15.8g, dry sieved

Of these five fractions, the 125-250μm and 250-500μm fractions were taken to produce the standard material, while the remaining fractions were preserved, but not processed any further. The 125-250μm and 250-500μm fractions were soaked in concentrated sulfuric acid at 120^∘C overnight, to remove all iron coatings and accessory minerals such as rutile, sphene and fluorite. After the acid treatment, the material was rinsed ten times in cold de-ionized water, followed by five times one hour ultrasonic rinsing in de-ionized water at 80^∘C. Next, the quartz was dried overnight at 110^∘C. Although the preparation steps outlined above probably already ensured a thoroughly mixed quartz sand, a FRITSCH^© rotary cone sample divider laborette 27 was used to split the material into 16 equal fractions. Different aliquots of CREU-1 have been analyzed by five noble gas laboratories, at BGC (Berkeley), CRPG (Nancy), ETH (Zürich), GFZ (Potsdam) and SUERC (Glasgow).

3. Analytical methods

The five participating laboratories employed a variety of noble gas mass spectrometers and analytical procedures for cosmogenic ²¹Ne analysis. Rather than forcing all the participants to use the same heating schedules, gettering times and so forth, they were allowed to use their own measurement routines, so that the calibration exercise fully captured the diversity of approaches used for ²¹Ne analysis. The temperature steps and amount of material used are reported in Tables 1-3.

3.1. BGC

Neon extraction from quartz at BGC employed a 14-sample vacuum chamber with a 3 inch diameter sapphire viewport. Samples of up to 150 mg quartz were encapsulated in a Ta packet and heated through the viewport by a 150 W diode laser (λ = 810 nm) using a feedback control system in which the temperature of the packet was continuously monitored by an optical pyrometer coaxial with the laser delivery optic. Calibration of the pyrometer for the emissivity of the Ta packets was accomplished by placing a thermocouple in the same apparatus. Collateral heating of adjacent samples was prevented by completing one heating step for all samples before beginning the next heating step. This procedure was tested by interspersing blanks consisting of an empty Ta packet. After heating, sample gas was reacted with a SAES^®; getter and adsorbed to a cryogenic trap at 20 K. Neon was then released into the mass spectrometer at 70 K. All sample heating, gas processing, and measurement operations were automatically controlled. Analyses were done with a MAP-215 mass spectrometer updated with modern ion-counting electronics. Under normal operating conditions, this machine had a relatively low Ar⁺/Ar⁺⁺ ratio (2.5-5, depending on source tuning) and inadequate mass resolution to fully resolve ²⁰Ne⁺ from ⁴⁰Ar⁺⁺, so a correction for background ⁴⁰Ar⁺⁺ was required. As described in Balco and Shuster (2009b), this was accomplished by introducing a ³⁹Ar spike and monitoring the Ar charge ratio as well as the ⁴⁰Ar⁺ signal throughout each analysis. The resulting correction on mass 20 varied between analyses, but was typically equivalent to 5.00±0.02×10⁸ atoms ²⁰Ne. Similarly, a correction for ¹²C¹⁶O ₂⁺⁺ on mass 22 was made by establishing a relationship between the Ar and CO₂ charge ratios. Absolute calibration of Ne abundance was made by peak height comparison against an air standard processed in the same way as the samples and analyzed several times daily. Linearity of machine response was verified by varying the volume of the air standard. The pressure of the air standard reservoir was measured during loading with an MKS Baratron manometer, and corrected for atmospheric water vapor using three separate hygrometers at the time of air sample collection. Absolute volumes of the reservoir and pipette were determined by differential pressure measurements, again using the Baratron, against two separate reference glass ampules whose volumes were independently measured by Hg weighing. The amount of cosmogenic ²¹Ne was calculated by assuming two-component mixing of atmospheric and cosmogenic neon. Reported uncertainties include i) counting uncertainties on all masses, including those used to generate corrections for ⁴⁰Ar⁺⁺ and CO ₂⁺⁺; ii) uncertainty in blank subtraction (the ²¹Ne process blank was ~ 0.5 Hz or ~ 90,000 atoms, which was < 1% of typical signals on mass 21 for these measurements); and iii) the reproducibility of the air standards (~1% for ²⁰Ne, ~3% for ²¹Ne).

3.2. CRPG

After 10 minutes cleaning in an acetone ultrasonic bath, quartz aliquots were wrapped in copper foils (Alfa Aesar^®;, 0.025 mm thick, 99.8%). Samples were then loaded under high vacuum in a stainless steel carousel that had been baked during 10 h at 80^∘C. Gas extraction from the quartz was realized by 25 minutes heating in a home-designed single vacuum resistance furnace with a boron nitride crucible (Zimmermann et al., in press). Sequential purification with charcoals in liquid nitrogen, titanium sponges (JohnsonMatthey^®;, mesh m3N8 t2N8) and SAES^®; getters (ST172/HI/20-10/650C) permitted gas cleaning by removal of H₂O, Ar, Kr, Xe and hydrocarbons. Ne was not separated from He. The purified gas was finally analyzed using a VG5400 mass spectrometer. Corrections for isobaric interferences of ⁴⁰Ar⁺⁺ at m/e = 20 and ¹²C¹⁶O ₂⁺⁺ at m/e = 22 were negligible compared to the amount of analyzed neon. The mass spectrometer sensitivity was determined by peak height comparison against a 0.2 cm³ (~1.6×10¹⁰ atoms of ²⁰Ne) pipette of a gas standard having an atmospheric composition. Typical furnace blanks at 1000-1300^∘C (25 min) were 1.0±0.2×10⁸, 3±1×10⁵ and 1.63±0.06×10⁷ atoms of ²⁰Ne, ²¹Ne and ²²Ne, respectively. Excess ²¹Ne (²¹Ne^*) concentrations were calculated following:

(1)

where ²⁰Ne _m is the measured ²⁰Ne, R _c is the cosmogenic ²¹Ne/²⁰Ne-ratio (R _c = 0.8; Niedermann, 2002), R_m is the measured ²¹Ne/²⁰Ne-ratio, and R _a is the atmospheric ²¹Ne/²⁰Ne-ratio (R _a = 0.00296).

3.3. ETH

Noble gases were extracted by heating in a molybdenum crucible. Released gases were cleaned in a stainless steel extraction line equipped with Al/Zr-getters (SAES^®;) and activated charcoal held at the temperature of liquid nitrogen before He and Ne were expanded to a cryogenic pump. Helium and neon were separated by adsorbing neon at 14 K on stainless steel frits and analyzing helium first. After pumping away the helium, neon was released from the cryotrap at 50 K. Noble gas analyses were performed in a custom-made, all-metal magnetic sector mass-spectrometer (90^∘, 210 mm radius) equipped with a modified Baur-Signer ion source with essentially constant sensitivity over the pressure range relevant for this work (Baur, 1980). The ion source was equipped with a compressor device increasing the sensitivity by factors of 120 and 200 for ³He and ²¹Ne, respectively (Baur, 1999) compared to the sensitivities of the same spectrometer with the compressor turned off. The absolute sensitivity and mass discrimination of the mass spectrometer were determined by analysing known amounts of standard noble gas mixtures prepared from commercially available pure gases. The Ne isotopic composition of the standard gas was cross calibrated against two air standards (Heber et al., 2009). Similarly, the Ne amounts delivered by the standard pipette were cross calibrated with air standards as well as with other independently filled standard gas bottles. The uncertainty of the Ne standard gas amounts is estimated to be 2% (Heber et al., 2009). Full procedural blanks (45’ at 600^oC + 20’ at 800^∘ + 15’ at 1750^∘C) were 1.211±0.006×10⁸, 3.5±0.2×10⁵, and 1.17±0.01×10⁷ atoms of ²⁰Ne, ²¹Ne and ²²Ne, respectively. Corrections for isobaric interferences on mass 20 have been applied for ⁴⁰Ar⁺⁺ and H₂¹⁸O⁺ but were always less than 2%. No correction for CO ₂⁺⁺ on ²²Ne was necessary. The low correction factors for doubly charged species were the results of a low electron acceleration voltage of 45V in the ion source. Excess ²¹Ne (²¹Ne^*) concentrations were calculated with Equation 1.

3.4. GFZ

CREU-1 quartz samples were wrapped in aluminium foil and loaded in a sample carousel without further treatment, except for two aliquots of the 250-500μm fraction (GFZ-6-7) which were crushed to ~50μm grain size in an agate mortar before loading. Noble gases were extracted in a resistance-heated furnace equipped with a tantalum crucible and molybdenum liner and analyzed in either of two VG5400 noble gas mass spectrometers, with measurements GFZ1-7 being measured on one machine, and GFZ8-11 on the other (Tables 1 and 2). GFZ-8 was not heated, but instead crushed in vacuo between two hard metal jaws in order to test whether Ne trapped in fluid inclusions of CREU-1 has an atmospheric isotopic composition. Gas purification involved a dry ice trap, two titanium sponge and foil getters, and two SAES^®; (Zr-Al) getters. The noble gases were trapped on stainless steel frits and/or activated charcoal in cryogenic adsorbers and sequentially released for He, Ne, and Ar-Kr-Xe analysis. Isobaric interferences of ⁴⁰Ar⁺⁺ at m/e=20 (up to 20% at 400^∘C) and ¹²C¹⁶O ₂⁺⁺ at m/e=22 (up to 10% at 400^∘C) were corrected according to the method described by Niedermann et al. (1993, 1997). A correction for H₂¹⁸O⁺ at m/e=20 was not necessary due to the mass resolution of ≥600. Blanks had an atmospheric composition and contained 1-3×10⁷ atoms of ²⁰Ne, depending on temperature. Excess ²¹Ne was calculated without applying a blank correction, assuming an atmospheric origin of all the measured ²⁰Ne:

(2)

with all abbreviations as in Equation 1. In some cases a high atmospheric Ne memory (i.e., rapid decay of non-atmospheric Ne isotope ratios) required the application of a special procedure to derive the Ne concentration and isotopic composition at the time of gas admission to the mass spectrometer (see Goethals et al., 2009). Absolute noble gas concentrations were obtained by peak height comparison against a 0.1 cm³ pipette of calibration gas (an artificial mixture of the five noble gases in nitrogen provided by Linde company; Niedermann et al., 1997), which was cross-calibrated in the 1990s against glass ampoule noble gas standards made available by O. Eugster (University of Bern) and whose noble gas concentrations are judged accurate to ~3% at 95% confidence level, and have been propagated into the overall uncertainty.

3.5. SUERC

The clean quartz was thoroughly rinsed in ultra-pure acetone and packed into aluminium foil cylinders. Cosmogenic Ne was extracted by heating each sample packet for 20 minutes. The active gases were removed by exposure to two hot SAES^®; (Zr-Al) getters during heating, and for a further 20 minutes as the furnace cooled. The heavy noble gases and residual active gases were subsequently adsorbed on liquid nitrogen cooled activated charcoal for 10 minutes and exposed to a getter at room temperature to adsorb hydrogen. Neon was then adsorbed on activated charcoal in a cryostatic cold head at 30K. The helium was pumped for 1 minute, then the Ne was desorbed from the charcoal trap at 100K. Neon isotopes were analyzed statically in a MAP-215 magnetic sector mass spectrometer equipped with a modified Nier-type ion source, an axial electron multiplier (Burle Channeltron) operated in pulse-counting mode and a Faraday detector. A room temperature SAES^®; G50 getter and a liquid nitrogen-cooled activated charcoal trap were used to minimize the contribution of interfering species during analysis. The data presented here were taken over a period of two years. Consequently source conditions changed to a small degree. Typically the source was tuned for Ne sensitivity prior to analytical periods; electron voltage of 88 V, trap current of 500 μA and an acceleration voltage of 3 kV. A slit in front of the electron multiplier was used to achieve a resolving power (m/Δm) of approximately 400. For all samples and calibrations the abundances of masses 18, 19, 20, 21, 22, 40 and 44 were determined by integrating counts recorded in 40-100 blocks of 5 seconds each. Peak heights of masses 2 and 16 were measured on the Faraday detector. Instrumental sensitivity was calculated from repeated analysis of aliquots of 2.2×10¹⁰ atoms ²⁰Ne in air sampled from a 5 liter reservoir. Isotopic mass discrimination was approximately 0.50 ± 0.03 %/amu. The average high temperature ²⁰Ne blank was 1×10⁸ atoms. There was no observed increase when empty Al foil was heated. The Ne isotopic composition of blank measurements after correction for interfering species (see below) was indistinguishable from air ratios. Since it is likely that a significant amount of air-derived Ne is released from the quartz during heating, no blank correction has been made to the data. Excess ²¹Ne concentrations were calculated assuming an atmospheric origin of all the measured ²⁰Ne according to Equation 2. Interference at m/e = 20 from H₂¹⁸O⁺ was calculated from measurement of H₂¹⁶O⁺ at mass 18. The contribution never exceeded 0.03%. No H¹⁹F⁺ signal was observed in blanks and mass spectrometer backgrounds. The dominant interference at m/e = 20 came from ⁴⁰Ar⁺⁺. The charge state ratio ⁴⁰Ar⁺/⁴⁰Ar⁺⁺ is governed by the partial pressure of H in the mass spectrometer ionization region. A first-order relationship between ⁴⁰Ar⁺/⁴⁰Ar⁺⁺ and H⁺ beam size was recorded. The partial pressure of H remained constant resulting in ⁴⁰Ar⁺/⁴⁰Ar⁺⁺ = 2.30-2.32. The contribution of ⁴⁰Ar⁺⁺ to the measured ²⁰Ne signal in CREU quartz samples was <1%. Correction for ¹²C¹⁶O ₂⁺⁺ at m/e = 22 was calculated from measured mass 44 (¹²C¹⁶O ₂⁺) using a CO₂⁺/CO ₂⁺⁺ = 50 to 58 (determined by repeated measurements interspersed with sample measurements). No pressure dependence on the CO₂⁺/CO ₂⁺⁺ ratio was recorded for a 50-fold variation in the partial pressure of H and CO₂. Correction for interfering ¹²C¹⁶O ₂⁺⁺ never exceeded 1%.

4. Results

All five labs reported data for the coarse fraction, while three labs measured the fine fraction as well. The results for both sets of analyses are reported in Tables 1 and 2. The ²¹Ne/²⁰Ne and ²²Ne/²⁰Ne compositions of the individual heating steps and their sums are consistent with a predominantly spallogenic origin of the released ²¹Ne (Figure 1). The pooled analyses comprise 16-30% atmospheric and 70-84% excess ²¹Ne, with individual heating steps containing up to 98% excess ²¹Ne. Linear regression of the spallation line yields a slope of 1.108 ± 0.014 (2σ), which is in statistical agreement with previously published values (Table 8 of Niedermann, 2002).

The total excess ²¹Ne contents of all the aliquots are shown in Figure 2. The reported 2σ analytical uncertainties are between 2 and 6%. The MSWD (Mean Square of the Weighted Deviates, a.k.a. ‘reduced Chi-square’, McIntyre et al., 1966) is reasonably close to unity for ETH, GFZ and BGC, indicating good agreement of the observed scatter with the measurement errors. The extremely low MSWD of 0.005 for CRPG may indicate overestimated analytical uncertainties, but could also be due to chance, as only two aliquots were analyzed. Finally, the coarse fraction of SUERC is characterized by an MSWD of 4.1, which may indicate underestimated analytical uncertainties. However, measurements of the fine fraction by the same lab have an MSWD of 1.3. There is no systematic difference between the fine and the coarse grain size fractions of ETH, SUERC and GFZ. Measurements GFZ-6-7 were performed on material from the coarse fraction that was crushed for 5 minutes in air with an agate mortar to ~50μm, resulting in some loss of excess ²¹Ne. GFZ-8 was crushed in vacuo, and the data shown are for that crushing extraction. The ²¹Ne excess of GFZ-8 is considerably less than the ²¹Ne deficit in GFZ-6-7, probably because the in-vacuo crusher was much less efficient than the mortar. Measurements GFZ-6-8 were not included in subsequent calculations and figures. Total ³He concentrations measured at GFZ were 204±10×10⁶at/g for the coarse fraction (seven measurements) and 109±7×10⁶at/g (a single analysis) for the fine fraction. The resulting ²¹Ne∕³He-ratios are significantly greater than the production-rate ratio. This is likely caused by a combination of helium loss due to hot acid etching during sample preparation, and the fact that helium is not quantitatively retained in quartz at surface temperatures (Shuster and Farley, 2005).

BGC analyzed material from two different vials of CREU-1, thus presenting an opportunity to verify the homogeneity of the standard. Measurements BGC-1-4 were performed on material from the same vial as ETH, whereas measurements BGC-5-8 were done on the same vial as CRPG. The observed difference between the two vials analyzed by BGC falls within the analytical uncertainty. The difference between the results of BGC and ETH/CRPG, however, falls well outside the statistically acceptable range. The error-weighted means of all the labs do not agree with each other within the analytical uncertainties, defined as the standard errors of those means. Therefore, in order to calculate a global average of all the data (using both the fine and the coarse grain fractions), we used a random effects model with two sources of uncertainty. We assume that the intra-laboratory averages x_i (where i is an identifier for each participating lab) come from a normal distribution of the form:

(3)

where μ is the global mean, σ_i² the analytical uncertainty (variance) of the i^th lab, and ζ² is the amount of overdispersion, i.e. the excess scatter (variance) that cannot be explained by the analytical uncertainty alone. To understand this formula, consider the following two special cases. If σ_i = 0 (perfect reproducibility within each lab) then μ is the arithmetic mean of the laboratory averages. And if ζ = 0 (perfect reproducibility between all labs) then μ is the error-weighted mean of those same laboratory averages. In order to simultaneously take into account the finite analytical precision of each lab and the variance between the labs, Equation 3 was iteratively solved for both μ and ζ, yielding an average ²¹Ne concentration of 348±10×10⁶at/g and an overdispersion (defined as 2ζ/μ) of 7.1%.

5. Comparison with CRONUS-A

In addition to CREU-1, two of the participating labs also analyzed CRONUS-A as a second reference material. CRONUS-A was collected in Antarctica’s Arena Valley (77^∘ 52’ 58.9”S, 160^∘ 56’ 35.1”E, 1666m elevation), from a large (40kg) yet thin (~2cm) slab of Beacon sandstone. Quartz was purified at the University of Vermont by crushing, sieving and repeated etching in dilute HF, using procedures designed for cosmogenic ¹⁰Be-²⁶Al analysis. CRPG reported one and BGC a further nine analyses of CRONUS-A, using the same protocols that were used for the CREU-1 measurements (Table 3). The average cosmogenic ²¹Ne content of the nine CRONUS-A samples measured by BGC was 338.9±3.8×10⁶at/g, i.e. 7.6±3.7% lower than that of CREU-1. The single CRONUS-A analysis of CRPG is lower than its CREU-1 measurements by a similar amount (4±17%), although the reported analytical precision of the latter estimate is much poorer. Additionally, published CRONUS-A values have been reported by two laboratories which did not participate in the interlaboratory comparison, at Harvard University (330±3×10⁶at/g, Middleton et al., 2012) and the California Institute of Technology (338±10×10⁶at/g, Amidon and Farley, 2012). Normalizing the average CRONUS-A value reported by BGC to the CREU-1 reference value results in a ²¹Ne concentration of 320±11×10⁶at/g. We propose that when this value is used as a reference, CRONUS-A can serve as an alternative to CREU-1.

6. Discussion

It is interesting to note that significant amounts of excess ²¹Ne remained trapped in the quartz after the second highest heating step, at temperatures of up to 820^∘C. Total degassing was not achieved until the final temperature step at 1140^∘C and more. This is significantly higher than the 800^∘C release temperature for cosmogenic neon reported by Niedermann (2002). Nevertheless, for all samples of all labs, the data points of the higher temperature steps plot on the mixing line between atmospheric and cosmogenic neon (Figure 1), which strongly suggests that the non-atmospheric neon in all samples is essentially purely cosmogenic, although quartz occasionally also contains a nucleogenic neon component released at high temperature with a ²¹Ne/²²Ne ratio of approximately unity (Ne_HT , Niedermann et al., 1994; Niedermann, 2002). However, in view of the position of all data points in Figure 1 it seems very improbable that a sizeable fraction of the non-atmospheric ²¹Ne in our samples could be nucleogenic Ne_HT . Even in this unlikely case this would be largely irrelevant for the purpose of interlaboratory comparison, because for all samples we sum the non-atmospheric ²¹Ne from all temperature steps.

Despite the fact that CREU-1 is pure and highly enriched in spallogenic neon, the ²¹Ne concentrations reported by the participating labs are significantly overdispersed with respect to the formal analytical uncertainty. In theory, this overdispersion could be due to inhomogeneity of the standard material itself, as different labs analyzed aliquots from different vials of CREU-1. However, the analysis of two of these vials by BGC, and comparison with measurements of those same vials by ETH and CRPG, shows that this is not the case. Therefore, CREU-1 is homogenous. If the overdispersion cannot be attributed to the standard material itself, then it must be due to biases introduced by the different standard calibration bottles used (Heber et al., 2009), or to differences in the neon sensitivity between samples and standards introduced by sample processing or tuning conditions.

7. Conclusion

Our calibration experiment has shown that, although the reported analytical precision of cosmogenic noble gas measurements may be as low as 2%, the accuracy is not quite as good. We suggest that the 7.1% dispersion observed in our study be used as a more realistic estimate of the accuracy of the ²¹Ne method at the present time. It should be borne in mind that this may even be an optimistic value, for a highly enriched and well behaved standard material. Using realistic and conservative analytical uncertainties is especially important for studies combining ²¹Ne with other (radio)nuclides, and to assess the resolving power of such studies. For single nuclide studies, CREU-1 or CRONUS-A measurements can be used to normalize ²¹Ne to the reference values reported in this paper, so that measurements from different labs can be compared on an equal footing and relative differences in ²¹Ne can be compared on the level of the analytical precision (Dunai and Stuart, 2009). Those interested in obtaining aliquots of these standards may contact T. Dunai (tdunai@uni-koeln.de) for CREU-1 or T. Jull (jull@email.arizona.edu) for CRONUS-A.

Acknowledgments

This research was funded by CRONUS-EU (Marie Curie RTN project 511927). We would like to thank Mark Kurz and an anonymous reviewer for positive and constructive feedback on the submitted manuscript.

References