Element concentration and isotope ratio measurements by single-collector mass spectrometry often require the detection system to handle ion beams with very large intensity ratios. In order to obtain accurate and reproducible element concentration and isotope ratio data, the detection system must have a linear response with respect to the intensity of the incident ion beam. An extended range scaling pulse counting detector equipped on a Varian 810 quadrupole inductively coupled plasma–mass spectrometer (ICP-MS) was tested for linearity across count rates of ∼2000 to 110,000,000 cps with different concentrations of natural U solutions. We also tested detector linearity by the laser ablation analysis of 206Pb/238U, 207Pb/235U, and 207Pb/206Pb ratios in well-characterized 416–1565 Ma zircon standards. Results indicate that there is no correlation between the measured isotope ratio and ion intensity for the solution tests or the tests of natural zircon standards. The results of these tests confirm the suitability of this instrument for isotope ratio measurements that require a substantial dynamic range without having to switch between pulse counting and analog modes on electron multipliers or switching between electron multiplier to Faraday detectors.
In situ U-Pb geochronology of zircon by laser ablation (LA) mass spectrometry has become a vital tool in the earth and planetary sciences. The measurement of isotope ratios by solid sampling LA systems require a rapid mass scanning capability (1–10 amu ms−1) of single-collector mass spectrometers or simultaneous multicollection because sample signals can rapidly change with time. In addition, the combination of potentially very large ion intensity ratios and the transient nature of the signals during LA analysis of zircon requires that the ion detection systems of the mass spectrometer possess a wide linear dynamic range.
Here we present the results of tests that assess the linearity of an extended range scaling pulse counting detector fitted to a Varian 810 quadrupole inductively coupled plasma–mass spectrometer (ICP-MS) over 103 to 108 counts per second (cps) signal intensities. Our data indicate that this detection system yields a linear response across this range of signal intensities making it well suited for U-Th/Pb dating of zircons by laser ablation mass spectrometry. With a Varian 810 Quadrupole ICP-MS coupled to either a Cetac LSX-213 213 nm wavelength or a Photon Machines Analyte 193 nm wavelength laser ablation system, we also demonstrate that we can consistently measure the U-Pb ages of zircons within 1.5% of their ID-TIMS age and can routinely achieve precisions of pooled ages of ±∼0.5% (2σ) and 1%–7% for individual spot analyses, depending on U concentration, age, and grain size. Many laser ablation laboratories analyzing zircons for geochronology have utilized magnetic sector single- and multiple-collector ICP-MS [Gehrels et al., 2006]. However, quadrupole-based ICP-MS instruments have the potential to yield comparable U-Th/Pb age data of zircon [Horn et al., 2000; Jackson et al., 2004; Košler et al., 2002] as the magnetic sector instruments.
Measurements were carried out on a Varian 810 quadrupole ICP-MS in laser mode coupled to either a Cetac LSX-213 213 nm wavelength or a Photon Machines Analyte 193 nm wavelength laser ablation system and in solution mode coupled to a Peltier-cooled spray chamber with a 200 μL min−1 quartz nebulizer. Nickel sampler and skimmer cones are used for all analyses.
2.1. Varian 810 ICP-MS
The ICP source of the Varian ICP-MS contains a Turner Interlaced load coil which results in low plasma potential [Douglas and Tanner, 1998] and reduces the spread in the kinetic energy of ions entering the ion optics. The ion optic design utilizes a 90° off-axis ion mirror that results in greater than 80% ion transport efficiency through the ion optics. The parabolic field produced by the mirror acts to focus ions to a focal point in front of the quadrupole mass analyzer eliminating the need to scan ion optic voltages during analysis. Potentials across the ion mirror can be optimized for maximum sensitivity of either heavy or light elements. In addition to the high ion transfer through the optics, the off-axis design and hollow geometry of the ion mirror plus the curved fringe rods of the quadrupole mass analyzer significantly limit the number of photons and neutral species that enter the detector. This results in very low backgrounds (0–20 cps for 238U) and correspondingly low (low ppt) detection limits for many elements.
The ICP-MS can be operated in either normal-or high-sensitivity modes. These modes are applied by adjusting the potentials across the extraction lenses and ion mirror (Table 1). In essence, the potential across the first extraction lens is adjusted from “passive” operation (−1 V; soft extraction) to roughly −500 V (hard extraction). The second and third extraction lenses as well as the corner lens, ion mirror, and entrance plates are all adjusted, either manually or through an automated optimization routine, for maximum sensitivity. The increase in ion intensity at the detector from normal-to high-sensitivity modes, all else being equal, is between a factor of 4 to 8. Depending on the application, there are advantages to operating in either normal-or high-sensitivity modes. The high-sensitivity mode offers increased sensitivity and lower detection limits for many elements including Pb, Th and U. The factory specification for this instrument is 3 × 108 counts s−1 ppm−1 for 232Th, but by adjusting the tuning settings for increased signal at high mass, this specification in solution mode can be well exceeded. For LA-ICP-MS of zircon, we operate only in high-sensitivity mode to maximize the signal intensity of masses 204 and 207.
Table 1. Machine Conditions Used During These Experiments
Normal Sensitivity Solution
High Sensitivity Laser
Gas flow parameters (L min−1)
RF power (kW)
Sampling depth (mm)
Pump rate (rpm)
Ion optics (volts)
First extraction lens
Second extraction lens
Third extraction lens
Mirror lens left
Mirror lens right
Mirror lens bottom
Dwell time (ms)
Points per peak
The detector used in the Varian 810 ICP-MS is an all digital, extended range, scaling pulse detector [Stresau and Hunter, 2001] developed by ETP Electron Multipliers. The dynamic range extension is achieved by attenuating the electron beam created within the detector (after the conversion of ions to electrons and before final amplification) under conditions of high signals. In order to “reconstitute” attenuated signals, the attenuation factor of the detector must be calibrated.
The detector design can be divided into three distinct functional sections. The first is an ion-to-electron conversion section (conversion dynode). This is followed by the controllable electron attenuation section. This in turn is followed by the amplification section. The controllable electron attenuation section can vary the ion detection efficiency continuously from 90% down to 0.01% [Stresau and Hunter, 2001], however, the Varian 810 ICP-MS software has reduced this to three discrete attenuation modes, labeled “normal,” “medium,” and “high.” This has the benefit of minimizing detector attenuation calibration requirements while still providing excellent dynamic range and precision.
2.2.1. Detector Attenuation
Control of the attenuation is either automatic or manual. In autoattenuation mode, a prescan of the signal is performed before acquiring the analytical data in order to determine the most appropriate attenuation mode. When a signal has been determined to have exceeded the user specified threshold for attenuation (1 × 106 cps), the electronics automatically set the appropriate attenuation mode for the signal. In manual mode the user can select any one of the three attenuation modes and the detector will measure all signals possible under this mode only. To span the three attenuation modes the user must calibrate the two attenuation factors, the factor spanning normal to medium and the factor spanning medium to high. The thresholds between attenuation modes normal to medium and medium to high are roughly 1 × 106 and 100 × 106 cps, respectively. Calibration of the two attenuation factors is achieved by presenting two solutions containing the elements expected to require the dynamic range extension. The solution used to calibrate the normal to medium attenuation factor has a concentration such that it can produce a count rate just below the normal to medium threshold. The second solution used to calibrate the medium to high attenuation factor has a concentration such that it can produce a count rate just below the medium to high threshold. The concentration difference between these two solutions is typically a factor of 50.
2.2.2. Detector Calibration
Prior to each analytical session we calibrated the attenuation factors for U, Th, and Pb by using an approximately 10 ppb mixed Pb-Th-U solution for the normal to medium attenuation factor and an approximately 500 ppb mixed Pb-Th-U solution to calibrate the medium to high attenuation factor. The applied attenuation factors are the same for both “wet” and “dry” plasma modes. The consistency in attenuation factors determined from 10 ppb and 500 ppb Pb-Th-U solutions measured prior to each of the last 33 analytical sessions over a 2 year period indicates minimal long-term drift of the factors. The standard deviation over those 33 sessions is less than 2% for the medium attenuation and less than 2.5% for the high attenuation (Table 2) indicating some long-period drift (weeks to months), but based on our tests below, there is no significant drift over an analytical session. It is important to note that sudden changes in attenuation factors (e.g., 22–23 April) reflect where we have performed a detector voltage calibration that will affect the measured attenuation factors.
Table 2. Attenuation Factors That Were Measured Prior to Each of 33 Analytical Sessions Between February 2008 and June 2010
26 Sep 2008
16 Oct 2008
21 Oct 2008
2 Dec 2008
13 Dec 2008
24 Feb 2009
25 Feb 2009
26 Feb 2009
9 Mar 2009
10 Mar 2009
22 Apr 2009
23 Apr 2009
28 Apr 2009
6 May 2009
14 May 2009
15 May 2009
19 May 2009
20 May 2009
20 Aug 2009
27 Aug 2009
7 Oct 2009
17 Oct 2009
18 Oct 2009
21 Nov 2009
23 Nov 2009
24 Mar 2010
7 May 2010
11 May 2010
28 May 2010
3 Jun 2010
7 Jun 2010
9 Jun 2010
29 Jun 2010
% Standard deviation
2.2.3. Detector Attenuation Stability
Detector attenuation is quite stable over long periods of time, but it is essential that these be precisely measured, especially for isotope ratio measurements involving extreme differences in isotopic abundances that would vary across different attenuation modes (e.g., 238U/206Pb and 235U/207Pb in zircon). Although the electron attenuation factor is stable, it will subtly vary over a period of days to weeks (Table 2), due to detector aging and the typical voltage adjustments required to maintain stable pulse counting operation. To ensure this was not a significant issue for this project and to obtain the most precise isotope ratio measurements, the detector was calibrated prior to each analytical session. The calibration data also served to provide a measurable level of confidence in this new detector design. Additionally, the design of this detector allows for the measurement of input signals typically spanning at least 0–1 GHz in a single “pulse-counting” mode. Instruments that have two detection modes, “pulse-counting” and “analog,” can also provide a large dynamic range, but by having a single detector providing a large dynamic range in a single counting mode eliminates the need for dual detection systems that required extensive cross calibration on a regular basis. Since this detector technology is relatively new, we undertook extensive tests to ensure that the detector system is suitable for isotope ratio measurements that require a substantial dynamic range.
2.2.4. Detector Dead Time Correction
In the Varian ICP-MS, the pulse generated at the detector output is first passed through a preamplifier, followed by a discriminator; the output is then fed into a paralyzable monostable multivibrator. This circuit design produces a fixed width output pulse which is then accumulated. The fixed pulse width means that the dead time is a constant and defined by the electronics and does not need calibration. The pulse width output by the monostable multivibrator is 20 ns, hence the dead time is set to a constant 20 ns.
2.3. Laser Systems
The CETAC Technologies LSX-213 laser system consists of a frequency quintupled 213 nm Q-switched Nd:YAG laser. The Photon Machines Analyte.193 excimer laser ablation system is powered by an ATL Ex300si ultrashort pulse length (∼4 ns) ArF laser head. The CETAC laser was used to collect data on 27 September 2008 and 28 April 2009 and the Photon Machines laser was used on 13 May 2010. For the experiments shown in this contribution, we used a round 50 μm diameter laser spot at the sample surface, a repetition rate of 15 Hz, and a laser power setting of 3 J cm−2. The carrier gas used to carry ablated material to the detector was He at a flow rate of 600–700 ml min−1.
2.4. Operating Conditions
Typical operating conditions for the ICP-MS are listed in Table 1 for both solution (normal-sensitivity) and laser (high-sensitivity) modes. The dwell time on each mass for both modes is listed in Table 1; isotopes in minor abundance have greater dwell times (e.g., 20 ms for 207Pb). The ICP-MS is tuned for maximum sensitivity and minimal oxide production and there is often a slight trade-off between the two. Typical ion intensities in solution mode are 7–10 × 105 cps ppb−1 and 1–4 × 106 cps ppb−1 for 238U in normal-and high-sensitivity modes, respectively. In solution mode, the CeO+/Ce ratio is tuned to be below 3%; in laser mode, the CeO+/Ce ratio is less than 1%. In line scan laser mode with operating parameters of 50 μm spot size, 20 μm s−1 scan rate, repetition rate of 15 Hz, and laser energy output of 3 J cm−2, typical ion intensities range from 5–9 × 105 cps to 4–8 × 106 cps for 238U in NIST 612 glass in normal- and high-sensitivity modes, respectively. This corresponds to a U sensitivity that is greater than 100,000 cps ppm−1 for NIST 612 glass.
3. Detector Linearity Tests
3.1. Variable Concentration U Solutions
An important factor in zircon geochronology by LA-ICP-MS is ensuring a stable and linear response from the detector over an analytical session. In order to mimic the potentially large difference in 238U and 207Pb-206Pb ion intensities derived from LA-ICP-MS analysis of zircon, we conducted an experiment in which we measured the 238U/235U ratio of a natural U solution, U-114. We analyzed 13 different aliquots with concentrations ranging from 100 ppt to 150 ppb in random order over a single 8 h analytical session. Instrumental drift and mass bias were corrected with the measurement of a 10 ppb U500 standard solution (238U/235U ratio = 1.0003) between every sample of U114. During the experiments the ion intensity of 235U never went above 750,000 cps (no attenuation), but the 238U ion intensity varied from 278,000 to 105,000,000 cps (crossing all attenuation modes) for the variable concentration U114 samples. This allows us to monitor the response of the attenuation factors on the measured 238U/235U ratio. The weighted mean of instrumental drift and mass bias corrected 238U/235U ratios of the variable concentration U114 solutions is 138.05 ± 0.51 (2σ) (Figure 1). Importantly, there is no systematic correlation between low, medium, and high attenuation of the 238U signal and the measured 238U/235U ratio indicating a linear response of the detector over a dynamic range of 2,000 to 100,000,000 cps.
3.2. Zircon Standards
We conducted three separate experiments in order to assess the precision and accuracy of measured zircon ages using the extended range pulse counting detector equipped on the Varian 810 ICP-MS. These tests were performed on 27 September 2008, 28 April 2009, and 13 May 2010. The first test was conducted by analyzing the Temora-2 zircon as an unknown and our in-house zircon from the Stettin syenite as our external standard. During the second and third tests, we analyzed 3 different well-characterized zircon standards: Peixe, FC5z and Stettin. The ages of the zircons standards range from 416 Ma to 1565 Ma and have U concentrations varying from about 80 to 600 ppm. Individual analyses are 60 s in length and are split into 3 parts each lasting approximately 20 s: background measurements, ablation and washout. For the first experiment we alternated between Temora and Stettin grains. During the second and third experiments, we analyzed each standard three times in a row before switching to another standard. This yielded an even distribution of all zircon standards that allowed us to switch between external standards without biasing the data. The external standard used to correct instrumental fractionation and drift was switched between FC5z and Stettin (Table 3).
In practice, we reduce the data in two steps. First, the raw data consisting of signal intensities of 202Hg, 204Pb(Hg), 206Pb, 207Pb, 208Pb, 232Th, and 238U are imported into Wavemetrics Igor Pro™ (v. 6.12A) software with the add-in Iolite (v. 2.11) [Hellstrom et al., 2008]. The program simply allows us to select segments of baseline (∼15 s) and sample signals (∼20 s) of each analysis. We use an automated signal selection protocol that selects all but the first 2 s of the sample signal to avoid surficial Pb contamination. It is important to note that in order to avoid any unintentional biasing of the data, we do not manually select sample signals. After automated signal selection, we can inspect the signals for inclusions and/or cracks that contain high common Pb and/or age zoning and we can modify the signal selection accordingly. Once baseline and signals are selected, the baseline-subtracted mean sample signal intensities and 2 standard error uncertainties are exported into a comma delimited file. The second step of data reduction consists of importing the sample data into our data reduction spreadsheet where we correct for instrumental mass and elemental fractionation, calculate random and systematic errors in addition to the internal errors exported from Iolite, and, if necessary, perform common Pb corrections to the data.
Instrumental drift over an analytical session can be significant and nonlinear. To assess the drift, we plot the fractionation factor (true ratio/measured ratio) as determined for each measured external standard versus time and model the functional form of the drift with the excel add-in Xlxtrfun™ by Advanced Systems Design and Development. The function is then used to calculate the fractionation factor that applies to the unknown sample measured at any given time. We then calculate the standard error of at least two standards run before and after the sample to calculate an uncertainty in the fractionation factor correction and add those errors quadratically to the internal errors exported from Iolite. Error correlation values are calculated with the equations given by Ludwig  and concordia diagrams are made with IsoPlot (v. 3.50) [Ludwig, 2003]. All systematic errors such as uncertainties in the age of the standard, decay constant and common Pb compositions are included after data are pooled. The uncertainties in the data presented here only include internal and fractionation correction uncertainties. The concordia diagrams include errors in the U decay constants as assigned by IsoPlot.
Common Pb corrections are often required where there is measurable 204Pb in the sample signal. Unfortunately, many ICP-MS instruments and He and Ar gasses have small but measurable amounts of Hg which can swamp out any small 204Pb signals. Typically, our 204Pb(+Hg) signal intensity ranges from 800 to 3000 cps and this signal is not accurately subtracted by baseline subtraction methods because of very small changes in instrument sensitivity during the introduction of ablated sample material into the ICP source. Instead we use the 202Hg/204Hg of 4.35 [Zadnik et al., 1989] and the measured 202Hg intensity to subtract the 204Hg component from the 204 signal intensity and calculate a 206Pb/204Pb ratio. Because of the errors in the Hg correction, the measured 206Pb/204Pb ratios can be negative in samples with very little common Pb (Table 3). In samples with measurable 204Pb, we perform a common Pb subtraction with common Pb compositions interpreted from Stacey and Kramers . We show the robustness of this technique with two analyses of the Stettin standard that contained appreciable common Pb (italics in Table 3). The robustness of these corrections is shown in Figure 2 where concordia diagrams show the Stettin data before and after common Pb correction.
4. Description of Zircon Standards
The Temora-2 standard comes from the Middledale Gabbroic Diorite, a high-level stock that outcrops as scattered boulders near the town of Temora, New South Wales, Australia [Black et al., 2003, 2004]. The stock is located in the Paleozoic rocks of Lachlan Orogen of Eastern Australia on the southern end of the Tasman Orogenic Belt [Black et al., 2003, 2004]. The accepted ID-TIMS age for Temora-2 is 416.8 ± 1.3 Ma (n = 9; 2σ) analyzed at the Royal Ontario Museum, Canada [Black et al., 2004].
The FC5z standard is derived from the Duluth Complex in Minnesota, USA and is from a similar sample locality as that of the AS3 and FC1 zircon standards (1099.1 ± 0.5 Ma, 2σ) [Paces and Miller, 1993]. The Duluth Complex is a large suite of intrusive mafic rocks that are related to the 1.1 Ga Midcontinent Rift. The complex is divided into two different series, the anorthositic and troctolitic [Paces and Miller, 1993]. Both the AS3 and FC5z standards come from the anorthositic series.
The AS3 standard is located in the southern most extent of the Duluth Complex and the FC5z standard is from the Forest Center area in the northwestern area of the Complex [Paces and Miller, 1993]. Previously reported ID-TIMS U-Pb zircon age from the Forest Center area (FC1) is 1099.0 ± 0.6 Ma (2σ) [Paces and Miller, 1993] which is the same as the 1099.1 ± 0.5 Ma (2σ) [Paces and Miller, 1993] ID-TIMS age for the AS3 standard. The FC5z standard has a reported chemical abrasion ID-TIMS age of 1096.2 ± 1 Ma (S. A. Bowring, personal communication, 2008).
The source for the Stettin standard zircon is the Stettin Syenite Pluton, Wisconsin, USA which is chemically related to the Wolf River Batholith [LaBerge and Palmer, 1980; van Wyck, 1995]. The zircons used for the Stettin standard were separated from a pyroxene syenite in southern portion of the Stettin Pluton [van Wyck, 1995]. Five fractions of 20–30 zircon grains and sixth fraction with a single zircon grain have a weighted average ID-TIMS age of 1565 ± 8 Ma (2σ) [van Wyck et al., 1994].
5. Results of Zircon Measurements
The zircon standards were chosen for their range of ages, but also their variable U concentrations which results in 238U as well as 206Pb ion intensities that cross the detector attenuation thresholds (Table 4). For Peixe, Stettin, and Temora zircons, the 238U ion intensity crossed the first attenuation threshold and all of the Pb intensities did not (Table 4). It is important to note that some of the grains had relatively low U contents resulting in unattenuated U signals. For FC5z, both the 238U and 206Pb ion intensities of >1.3 and >10 × 106 cps were attenuated. For all zircon analyses, the 207Pb signal was unattenuated (Table 4). Therefore, the accuracy of the 206Pb/238U and 207Pb/235U ratios depend on the robustness of the detector attenuation since the standards used for instrumental fractionation corrections were not attenuated in the same way as all of the standard grains measured as unknown samples.
Table 4. Average Count Rates for 206Pb, 207Pb, and 238U for Each Zircon Standarda
Count rates that crossed the first attenuation threshold are in bold. No count rate crossed the second attenuation threshold.
To assess the accuracy of our measured 206Pb/238U and 207Pb/235U ratios of the zircon standards, we compared our measured values to the accepted values. Since we assume that, for the most part, there was no common Pb in the zircons, the 206Pb/238U and 207Pb/235U ratios are a function of age. Therefore, we can compare measured versus accepted isotope ratios using the measured concordia age as a proxy. This approach is advantageous because the age calculation requires two isotope ratios, 206Pb/238U and 207Pb/235U. We measured concordia ages of 419.1 ± 2.1 Ma for the Temora-2, 558 ± 2.7 Ma for Peixe, 1092.5.4 ± 6.4 Ma for FC5z, and 1565.2 ± 4.3 Ma for Stettin; all errors are presented at the 2σ level (Figure 3). Based on these experiments, we found the ages of the various standards to be within error, with the exception of Peixe which is within 1.5% of the accepted ID TIMS age, of the accepted age of the various zircon standards (Figure 4). This confirms not only the linear response of the detector system over a large range of ion intensities, but also the robustness of the data reduction methods for U-Pb geochronology of zircons by LA-ICP-MS.
By the very nature of isotope ratio measurements, it is imperative that we accurately measure the ion intensities. In many cases, isotope ratios can be extreme, such as 206Pb/238U and 206Pb/204Pb ratios in zircon. As our tests have shown, the new detector developed by ETP and Varian provides a linear response over a large range of ion intensities and is not affected by the attenuation of the ion intensities. Additionally, the stability of the detector during a lengthy analytical session ensures there are no systematic biases in the data related to the linearity of the detector. Couple this with the precision of the instrument, as demonstrated by our zircon standard experiment, we can show that next generation quadrupole ICP-MS instruments can be viable alternative to multicollector ICP-MS instruments for some isotopic studies.
We thank Yongjun Gao for assistance with the maintenance and operation of the ICP-MS. We are extremely grateful for discussions with George Gehrels, University of Arizona, who kindly provided us with zircon standard materials. We also thank Clark Johnson, University of Wisconsin–Madison, for kindly providing us zircons from the Stettin syenite and Stephen Anderson, Varian (Melbourne), Inc., for lending us his technical knowledge of the Varian 810 ICP-MS. We also thank the anonymous reviewers for their comments in helping us make this a better paper. This research was supported by the American Chemical Society Petroleum Research Fund and the State of Texas Norman Hackerman Advanced Research Program, NSF/IF 0824967 and NASA NNX09AC06G (T.J.L.).