We describe a new high-sensitivity multicollector noble gas mass spectrometer (ARGUS) that has been specifically designed for simultaneous collection of all Ar isotopes and hence is ideally suited for 40Ar/39Ar geochronology. The instrument uses a detector housing that holds five Faraday collectors. One collector is equipped with a 1011 ohm resistor (for 40Ar), and the remaining four are fitted with 1012 ohm resistors (for 36Ar to 39Ar). ARGUS has a mass resolution of 225–250 and a measured sensitivity of 7 × 10−14 mol/V at 200 μA trap current. During the course of a 1 month run cycle 145 air calibrations yielded a weighted average 40Ar/36Ar of 300.67 ± 0.07 (2σ, eight data points rejected, and mean square weighted deviation = 1.18). The ages of three mineral standards (Taylor Creek Rhyolite sanidine, Heidelberg biotite, and Alder Creek sanidine) are within error of the accepted ages, and uncertainties for two are improvements on previously published data. ARGUS decreases analytical time while allowing more measurements to be made at higher precision compared to standard peak-jumping single-collector mass spectrometers.
The noble gas isotopes have found a multitude of uses as tracers and chronometers of geological and hydrological processes [e.g., Ozima and Podosek, 2002; McDougall and Harrison, 1999; Porcelli et al., 2002]. The majority of the instruments developed for precise noble gas isotope ratio determinations have been magnetic sector mass spectrometers [e.g., Reynolds, 1956; Hohenberg, 1980; Lynch and Kay, 1981]. Data collection using these instruments has typically been performed in “peak-jumping” mode; the magnetic field strength or accelerating potential are changed in order to sequentially focus ion beams of different masses onto a single collector. The ability to produce stable, flat-topped peaks with high resolution allows multiple collectors to be positioned to allow ion beams of different isotopes to be separated in the focal plane of a mass analyzer and measured simultaneously [e.g., Wieser and Schwieters, 2005]. Multicollection should improve the precision of isotope ratio determinations as small fluctuations in ion production and external factors (e.g., temperature) affect all ion currents simultaneously. Additionally ion current intensities are measured for a greater proportion of the total analysis time. Consequently sample size can be reduced, analytical time is decreased, allowing more measurements to be made, and memory effects are minimized.
Multicollection was first applied to stable isotope mass spectrometry in the 1940s [Nier, 1947]. However, the relatively low concentration of noble gases in rocks and minerals has restricted their use in static vacuum mass spectrometry. Stacey et al.  described the first multicollector mass spectrometer designed for argon isotope measurement. It remained one of only a handful of custom-made instruments until the commercial development of the ARGUS multicollector mass spectrometer. ARGUS is the first mass-produced multicollector designed specifically for the simultaneous collection of all Ar isotopes. In this paper we describe the mass spectrometer and review the performance of the instrument that is located in Natural Environmental Research Council (NERC) Argon Isotope Facility at the Scottish Universities Environmental Research Centre (SUERC).
2. Mass Spectrometer Description
ARGUS (Figure 1) is an all-metal single focusing, 13 cm radius 90° extended geometry magnetic sector mass spectrometer designed for operation in static mode. The source, detector and getter housings are machined from solid 316LN stainless steel to minimize the residual magnetic properties and limit outgassing. The flight tube is tapered to minimize the volume to approximately 490 cc. The analyzer background measured in static mode and regressed back to inlet time (time zero, closing of pumps) contains 9.9 × 106 atoms 40Ar and 3.1 × 106 atoms 36Ar, and static rates of rise are 2.4 × 107 atoms 40Ar/min and 1.1 × 106 atoms 36Ar/min.
It is equipped with a modified Nier-type electron bombardment source with x and z focusing. External source magnets are used for collimating the electron beam. All source parameters, including the 5 kV acceleration potential, are computer-controlled, monitored by electronic read-backs and can be optimized using an autofocusing routine. A sensitivity of 7.0 × 10−14 mol/V at 200-μA trap current for Ar has been measured using a known amount of the international mineral standard Heidelberg biotite (HD-B1 [Fuhrmann et al., 1987]). The ion beam does not come into contact with the ion exit slit. The resolution of the system (225–250) is defined by the collector apertures. The abundance sensitivity at the Ar mass range is typically less than 5 ppm of 40Ar (1011 ohm circuit) at 39Ar (1012 ohm circuit) and less than 3 ppm of 39Ar (1012 ohm circuit) at 38Ar (1012 ohm circuit).
The electromagnet is manufactured from highly refined Lowmoor soft pure iron. This material is unique in its purity, with an iron content of at minimum 99.9%. This allows for the production of a homogeneous magnetic field and low residual field memory. The control loop for the magnetic field comprises a temperature-stabilized Hall probe that controls the magnet power supply via a twin serial 18-bit DAC feedback circuit. This control system is capable of more than 5,000 individual magnetic field steps across a peak with a resolution in excess of 1,500.
The ARGUS at the NERC Argon Isotope Facility is equipped with five Faraday detectors mounted perpendicular to the flight tube. The resolving slits and 1 mm entry apertures are positioned to allow true simultaneous collection of masses 36 to 40 (Figure 1). The 40Ar+ beam is measured in the high mass-2 (H2) collector. The amplifier on this collector is fitted with a 1011 ohm resistor. The 39Ar+ to 36Ar+ beams are collected in Faraday cups equipped with high gain transimpedance amplifiers that are fitted with 1012 ohm resistors. The 1011 ohm amplifier output reaches base line ±20 ppm after measuring an 8 V (5.60 × 10−13 moles) signal within 2 s. The 1012 ohm amplifier typically reach baseline within 4 s. The Faraday acquisition control unit has the capability to simultaneously measure the output from up to 10 detectors and controls the Tau correction for each. The electronics for all the Faraday amplifiers are kept in a housing evacuated to 1 to 2 mbar and maintained at 16 ± 0.1°C by Peltier cooling system in order to reduce electronic noise. Direct measurement of collector noise for ten separate runs gave averages of 4 × 10−6 V on the 1011 ohm circuit and 1 × 10−5 V on the 1012 ohm circuit. Collector gain calibration is performed by the computer-controlled application of predetermined voltages to each collector. Over the course of 1 year the relative gain between collectors did not exceed 19 ppm.
An embedded personal computer runs a real time operating system that controls all the electronics via a fiber optic loop. This unit also “time stamps” all data before they are passed back to the Windows XP PC running the primary interface software.
Ion beam intensities for all Ar isotopes are typically measured in a 15-cycle run. Baselines (i.e., valleys between the masses) are measured over 30 s with an integration time of 1 s immediately after sample gas inlet during equilibration prior to analysis. Each data collection cycle has a period of 20 s and an integration period of 1 s. The total time for data collection is typically 327 s. Figure 2 shows the peak shape during continuous scanning of our air standard. Peak sides are stable to less than 30 ppm and peak tops are stable to less than 10 ppm.
4. Air Standards
Figure 3 shows typical data collection output. The beam intensities of the larger peaks (e.g., 40Ar) decrease with time, showing a slight concave upward form (gas consumption within the small volume dictates an exponential regression). This effect is largely removed when considering ratios which show a linear fit (Figure 3). Figure 4 shows a crossplot of 40Ar (1011 ohm resistor) and 36Ar (1012 ohm resistor) intensities versus measurement precision. For the 1011 ohm circuit beam intensities of 1.6 × 10−12 moles are measured with a precision of 0.002% whereas measurement of 1.1 × 10−14 moles has a precision of 0.19%. For the 1012 ohm circuits, collector beam intensities of 2.9 × 10−15 moles are measured with a precision of 0.07% whereas 2.3 × 10−17 moles have a precision of 9.61%. Air calibration shots of 5.3 × 10−13 moles 40Ar produces a 7.5 V signal on the 1011 ohm circuit with a precision of 0.003%. The 1.8 × 10−15 moles 36Ar produces a 0.025 V signal on the 1012 ohm circuit and is measured with a precision of 0.065%. In a monthly series of 145 air calibrations (April 2009, see auxiliary material for raw data) the variation of 40Ar and 36Ar signals is 0.09% and 0.18%, respectively. The 40Ar/36Ar data show a normal distribution and have a weighted average of 300.67 ± 0.07 (2σ, 8 out of 145 rejected, mean square weighted deviation [MSWD] = 1.18; see auxiliary material).
The degree to which source sensitivity and mass fractionation is affected by Ar partial pressure has been assessed by the analysis of different amounts of air. Figure 5 shows that there is no significant difference in 40Ar/36Ar for 40Ar signal size 23 V (1.6 × 10−12 moles) to 0.10 V (7.0 × 10−15 moles; n = 74). During analysis of 40Ar/39Ar mineral age standards there appears to be no significant variation in the 40Ar/39Ar with signal sizes (see auxiliary material). Uncertainty in the determination of 40Ar/36Ar are governed by the mass spectrometer's ability to precisely measure the small ion beam, i.e., 36Ar. Figure 4 shows that for 36Ar+ beams that are smaller than 1.0 × 10−15 moles uncertainty increases rapidly with decreasing ion beam size.
5. Determinations of 36Ar
The precise measurement of 36Ar is critically important in 40Ar/39Ar geochronology for two reasons. First, the atmospheric 40Ar/36Ar is utilized as the basis for correcting the relative abundances of Ar isotopes for mass-dependent measurement bias. Second, 36Ar is used to resolve atmospheric 40Ar from radiogenic 40Ar which is the limitation of accuracy and precision in dating relatively young and/or K-poor materials.
As multicollection does not rely on interpolation the ratio of simultaneously measured isotopes (online ratios) collected by ARGUS essentially smooth out small fluctuations in ion beams due to ion production and environmental noise. Hence it is possible to use the online 40Ar/36Ar and the measurement of 40Ar to precisely determine the amounts of 36Ar. Uncertainty can be propagated using the online 40Ar/36Ar in both the sample and associated blank analyses. However, whether the calculation of 36Ar using the 40Ar/36Ar and a precise measurement of 40Ar yield a smaller 36Ar uncertainty than direct measurement of 36Ar is a factor to consider? Figure 6 shows the uncertainty in 36Ar when determined using (1) 36Ar peak intensities and (2) 40Ar peak intensity and online 40Ar/36Ar. For 36Ar ion beam sizes of between 5.3 × 10−15 and 2.8 × 10−18 moles it is as equally precise to determine 36Ar by measuring either peak heights or using multicollector 40Ar/36Ar. Evidently the noise that is introduced by environmental factors (e.g., temperature fluctuations) are not detectable and there is no benefit to using the online ratios. Consequently we use peak heights to measure isotope abundance and subsequently calculate the ratios and uncertainties using quadratics.
6. Age Standards
In order to assess the performance of ARGUS we have undertaken studies of three well characterized 40Ar/39Ar age standards. We present data from analysis of HD-B1 (24.7 ± 0.3 Ma, 1σ [Fuhrmann et al., 1987]), Taylor Creek Rhyolite sanidine (TCR-2s; 28.34 ± 0.16, 1σ [Renne et al., 1998]) and Alder Creek sanidine (ACs, 1.19 ± 0.007 Ma [Renne et al., 1998] and 1.19 ± 0.001 Ma [Nomade et al., 2005]).
Purified separates were crushed gently to decrepitate any fluid or melt inclusions and sieved to produce a separate of grain size 20–63 μm. Samples were loaded into Cu packets, placed into quartz vials and then positioned in an Al can for irradiation. Adjacent to samples we placed Al packets of Fish Canyon Tuff sanidine (FCs; 28.02 ± 0.16 Ma, 1σ [Renne et al., 1998]) to permit characterization of the irradiation flux to the samples. The samples were irradiated in the McMaster reactor for 12 h.
Samples were step-heated for 5 min in a fully automated all-metal, double vacuum, resistively heated furnace over a temperature range from 500 to 1750°C. Extracted gases were cleaned for 10 min using 3 SAES GP50 getters (two operated at 450°C and one at room temperature). Furnace blanks were stable at less than 1.5 × 10−15 moles 40Ar and 8.8 × 10−18 moles 36 Ar. Gas extraction, cleanup and mass spectrometer inlet were fully automated.
ArArCalc [Koppers, 2002] was used for regressing raw data and determination of 40Ar/39Ar ages. Isotope data are corrected for blanks, radioactive decay, mass discrimination and interfering reactions. 40Ar/39Ar ages also include a 0.5% error assigned to the J parameter. Raw data, correction factors, 40Ar/39Ar age spectra and isochron plots (normal and inverse) are presented in auxiliary material. Each auxiliary material data file also contains a summary of how blanks were handled. The criteria for fitting of plateaus are that they must include at least 60% of 39Ar in three or more contiguous steps. Finally, the probability of fit of plateau to the data is >0.05.
40Ar/39Ar age spectra and inverse isochrons are all within error of each other for HD-B1, TCR-2s and ACs (Figure 7). We quote ages and uncertainties from the 40Ar/39Ar age spectra for each standard and total fusion ages (TFA) to allow direct comparison to previous K-Ar dating of standards. ARGUS dates HD-B1 as 24.59 ± 0.13 Ma (1σ, 0.51%, MSWD = 0.36, TFA = 24.52 ± 0.13 Ma), TCR-2s as 28.26 ± 0.14 Ma (1σ, 0.51%, MSWD = 0.52, TFA = 28.27 ± 0.14 Ma) and ACs as 1.191 ± 0.007 Ma (1σ, 0.56%, MSWD = 0.66, TFA = 1.194 ± 0.007 Ma). All ages (plateau, inverse isochron and TFA) fall within error of the accepted ages for each standard. The uncertainties for HD-B1 and TCR are improvements on previously published data. Low MSWDs are a function of multicollection.
Ongoing initiatives aimed at intercalibrating geological timescales, such as EarthTime [Kuiper et al., 2008], highlight the importance of precise quantification of isotopic ratios for geochronological purposes. The ARGUS multicollector static gas mass spectrometer has been designed specifically to improve isotope ratio measurements associated with 40Ar/39Ar geochronology. This is perhaps best exemplified by the precision of isotope ratio determinations. The reproducibility of 40Ar/36Ar in repeated air standards measured by ARGUS is approximately 5 times better than the equivalent from single-collector mass spectrometers in peak-jumping mode (typically ±0.1% based on data from 3 other 40Ar/39Ar laboratories). This improvement in precision is a result of rapid, simultaneous collection of beam intensities from all Ar isotopes using low-noise, high-sensitivity Faraday collectors. The use of Faraday cups with different amplifier gains provides a flexibility that is unavailable in conventional single-collector instruments. For example, the combination of 1011 ohm and 1012 ohm resistors in the SUERC instrument allows irradiation durations to be decreased by a factor of 10 owing to sensitivity differences for the collection of 40Ar compared to the collection of 39Ar to 36Ar. This minimizes the effects of interfering reactions and decreases the production of radioactive waste.
The major current limitation of the ARGUS is its inability to precisely measure small ion beams. The 1012 ohm resistor Faraday used to measure the small ion beams (e.g., 36Ar) is not as sensitive as an electron multiplier. One possible way to solve this issue is to deploy an electron multiplier in the mass 36 position. However, problems of intercalibrating with Faraday collectors may have a detrimental effect on precision, the major benefit of multicollection (i.e., will relative gain between the collectors match the 19 ppm of the current ARGUS collector configuration?). The availability of higher-sensitivity amplifiers (1013 and 1014 ohm circuits) suggests that soon Faraday detectors may approach electron multiplier sensitivity, which has the potential to revolutionize static gas mass spectrometry.
NERC is thanked for continued funding of the Argon Isotope Facility, and SUERC is financially supported by the Scottish Universities. We would like to thank Paul Renne (Berkeley Geochronology Centre), Matt Heizler (New Mexico Tech.), and Klaudia Kuiper (Vrije Universiteit) for providing single-collector mass spectrometer air calibration data for comparison. We are also grateful to Andrew Tait (SUERC) for technical support and Doug Hamilton (ThermoFisher) for advice. Two reviewers, Andrew Calvert and Anthony Koppers, are both acknowledged for constructive reviews.