Optimization and Application of ICPMS with Dynamic Reaction Cell for

Sep 20, 2007 - Applied Life Sciences, Muthgasse 18, 1190 Vienna, Austria, and Department of Lithospheric Research, University of Vienna,. Althanstrass...
1 downloads 0 Views 236KB Size
Download PDF
Anal. Chem. 2007, 79, 7753-7760

Optimization and Application of ICPMS with Dynamic Reaction Cell for Precise Determination of 44Ca/40Ca Isotope Ratios Sergei F. Boulyga,*,† Urs Klo 1 tzli,‡ Gerhard Stingeder,† and Thomas Prohaska†

Department of Chemistry, Division of Analytical ChemistrysVIRIS Laboratory, University of Natural Resources and Applied Life Sciences, Muthgasse 18, 1190 Vienna, Austria, and Department of Lithospheric Research, University of Vienna, Althanstrasse 14, A-1090, Vienna, Austria

An inductively coupled plasma mass spectrometer with dynamic reaction cell (ICP-DRC-MS) was optimized for determining 44Ca/40Ca isotope ratios in aqueous solutions with respect to (i) repeatability, (ii) robustness, and (iii) stability. Ammonia as reaction gas allowed both the removal of 40Ar+ interference on 40Ca+ and collisional damping of ion density fluctuations of an ion beam extracted from an ICP. The effect of laboratory conditions as well as ICP-DRC-MS parameters such a nebulizer gas flow rate, rf power, lens potential, dwell time, or DRC parameters on precision and mass bias was studied. Precision (calculated using the “unbiased” or “n - 1” method) of a single isotope ratio measurement of a 60 ng g-1 calcium solution (analysis time of 6 min) is routinely achievable in the range of 0.03-0.05%, which corresponded to the standard error of the mean value (n ) 6) of 0.012-0.020%. These experimentally observed RSDs were close to theoretical precision values given by counting statistics. Accuracy of measured isotope ratios was assessed by comparative measurements of the same samples by ICP-DRC-MS and thermal ionization mass spectrometry (TIMS) by using isotope dilution with a 43Ca-48Ca double spike. The analysis time in both cases was 1 h per analysis (10 blocks, each 6 min). The δ44Ca values measured by TIMS and ICP-DRC-MS with doublespike calibration in two samples (Ca ICP standard solution and digested NIST 1486 bone meal) coincided within the obtained precision. Although the applied isotope dilution with 43Ca-48Ca double-spike compensates for time-dependent deviations of mass bias and allows achieving accurate results, this approach makes it necessary to measure an additional isotope pair, reducing the overall analysis time per isotope or increasing the total analysis time. Further development of external calibration by using a bracketing method would allow a wider use of ICP-DRCMS for routine calcium isotopic measurements, but it still requires particular software or hardware improvements aimed at reliable control of environmental effects, which might influence signal stability in ICP-DRC-MS and serve as potential uncertainty sources in isotope ratio measurements. Natural variation of the isotopic composition of calcium of rocks, minerals, and biological samples represents potential interest for geochronology, climate change studies, archaeometry, * To whom correspondence should be addressed. E-mail: sergei.boulyga@ boku.ac.at. Fax: 43 1 360066059. Tel: 43 1 427753453. † University of Natural Resources and Applied Life Sciences. ‡ University of Vienna. 10.1021/ac0711790 CCC: $37.00 Published on Web 09/20/2007

© 2007 American Chemical Society

and other scientific disciplines. Ca isotope variation is basically caused by the β-decay of 40K (natural abundance 0.0117%, halflife 1.277 × 109 years) and by isotopic fractionation during metabolic processes and chemical-exchange reactions. According to an IUPAC report,1 natural variation of 40Ca abundance ranges between 96.880 and 96.982%, and therefore, highly precise methods are required to resolve isotopic differences. Usually the measured isotopic ratio (e.g., 44Ca/40Ca) is normalized to a standard (δ44Ca ) [(44Ca/40Ca)sample/(44Ca/40Ca)standard -1] × 1000). Heumann et al.2 stated that fractionations are proportional to the mass difference. δ48Ca is therefore about two times larger than δ44Ca. However, interferences are present as, for example, 46Ca and 48Ca overlap with titanium isotopes and 40Ca overlaps with both 40K and 40Ar. The latter represents a particularly severe problem for mass spectrometric methods employing Ar plasma as an ion source. Therefore, even after introduction of highly precise inductively coupled plasma mass spectrometers (ICPMS) with multiple ion collectors (MC-ICPMS), only a few 44Ca/40Ca isotopic measurements by ICPMS have been reported by using either cold plasma conditions3 or a collision cell4 for reduction of 40Ar+ ion intensities. Application of collision cell and dynamic reaction cell (DRC) technologies in ICP mass spectrometry is particularly attractive for the determination of traditionally “difficult-to-analyze” isotopes like, for example, 40Ca, 56Fe, or 80Se, which suffer from interference by argon-containing ions. During almost one decade of being available to the analysts, ICPMS with collision and reaction cells has been found advantageous for a number of applications including precise isotopic analysis of Fe, Se, Pb,5,6 and other elements since it has been shown that the effect of collisional damping in the DRC in combination with fast scanning allows one to obtain isotope ratio precision close to counting statistics.5 The most significant factor affecting the precision of isotope ratio measurements in ICPMS is fluctuation of the ion density in the beam extracted from the (1) Rosman, K. J. R.; Taylor, P. D. P. Pure Appl. Chem. 1998, 70, 217-235. (2) Heumann, K.; Schiefer, H.-P.; Spiess, W. In Stable Isotopes; Schmidt, H.-L., Fo ¨rstel, H., Heinzinger K., Eds.; Elsevier: Amsterdam, 1982; pp 711-718. (3) Fietzke, J.; Eisenhauer, A.; Gussone, N.; Bock, B.; Liebetrau, V.; Na¨gler, Th.F.; Spero, H.J.; Bijma, J.; Dullo, C. Chem. Geol. 2004, 206, 11 -20. (4) Palacz, Z.; Shuttleworth, S.; Meffan-Main, S.; Turner, P. 2004 Winter Conference on Plasma Spectrochemistry. Fort Lauderdale, FL, January 5-10, 2004; Book of abstracts p 365. (5) Bandura, D. R.; Baranov, V. I.; Tanner, S. D. J. Anal. At. Spectrom., 2000, 15, 921-928. (6) Boulyga, S. F.; Becker J. S. Fresenius J. Anal. Chem. 2001, 370, 618-623.

Analytical Chemistry, Vol. 79, No. 20, October 15, 2007 7753

ICP. The effect of these fluctuations can be reduced by appropriate selection of dwell time (time for integrating the data for a particular m/z during a single measurement period). For instance, the measurement period should be short enough to allow a scan cycle to be completed in a faster time than the period of the beam fluctuation, along with that the dwell time must be long enough to average the high-frequency noise. Reducing kinetic energy of ions by collisions with the reaction gas increases ion residence time in the pressurized quadrupole and results in broadening and overlapping ion packets extracted from the ICP leading to temporal homogenization of the ion beam. By employing this effect for isotope ratio measurements, Bandura et al.5 achieved a precision close to counting statistics for Fe, Pb, and Ag isotopes. However, precise determination of the 44Ca/40Ca isotope ratio still remains a challenge especially due to prominent interferences. ICPMS equipped with a hexapole collision cell allowed determining 44Ca/40Ca isotope ratios with a RSD of 0.26%.6 However, the employed mass spectrometer did not allow a measurement time per mass (dwell time + settling time) shorter than 20 ms because of specific constructive features of the collision cell and the quadrupole mass analyzer. Several attempts have been made since then to measure 44Ca/40Ca isotope ratios by ICP-DRC-MS. The reported results had significantly higher relative standard deviations compared to those obtained by Bandura et al.5 for Fe or Pb isotope ratios. Hattendorf et al.7 measured 44Ca/40Ca isotope ratios by DRC-ICPMS by using ammonia as a reaction gas with a precision of 0.3-0.9%, which was good enough for tracer experiments with enriched 44Ca. In a recent paper by Stu¨rup et al.,8 calcium isotope ratios were determined in urine samples after oxalate precipitation by using ICP-DRC-MS with methane as a reaction gas. The authors stated that, despite the successful removal of 40Ar+ interference by the reaction with methane, it was not possible to measure 44Ca/40Ca and 42Ca/40Ca ratios with a precision better than 0.5% RSD, most likely due to uncertainties associated with the detection system when measuring high count rates. The present work is aimed at the systematic optimization of ICP-DRC-MS for precise measurement of 44Ca/40Ca isotope ratios and the evaluation of experimental factors affecting measurement precision and accuracy. The paper compares results obtained with ICP-DRC-MS and thermal ionization mass spectrometry (TIMS) by using internal mass bias correction with a 43Ca-48Ca doublespike. EXPERIMENTAL SECTION Sample Preparation. All sample preparation steps, including microwave-assisted (MLS1200 mega, Milestone-MLS GmbH, Leutkirch Germany) digestion of the bone samples, have been conducted in a class 10000 clean room. Precleaned PE bottles (Semadeni, Ostermundigen, Switzerland) were used for sample storage at 4 °C. HNO3 was prepared by double subboiling distillation (Milestone-MLS GmbH) of analytical reagent grade acid (Merck KGaA, Darmstadt, Germany). H2O, pretreated by reverse osmosis, was further purified by a laboratory-reagent (7) Hattendorf, B.; Wanner, H.; Gu, H.; Dorn, S.; Gu ¨ nther, D. In Plasma Source Mass Spectrometry: Current Trends and Future Development; Holland G., Bandura D.R., Eds.; RSC Publishing: Cambridge, 2005; pp 91-98. (8) Stu ¨ rup, S.; Bendahl, L.; Gammelgaard, B. J. Anal. At. Spectrom. 2006, 21, 297-304.

7754

Analytical Chemistry, Vol. 79, No. 20, October 15, 2007

grade water system (F+L GmbH, Vienna, Austria) followed by subboiling distillation (Milestone-MLS GmbH). 43Ca-48Ca doublespike was kindly provided by the team of Prof. Eisenhauer from GEOMAR (Forschungszentrum fu¨r Marine Geowissenschaften, Kiel, Germany). Calcium carbonate powder (NIST SRM915a) was digested in subboiled concentrated nitric acid and diluted with purified water. A calcium ICP standard was obtained from Merck KGa (Darmstadt, Germany). About 0.3 g of NIST SRM 1486 bone meal was digested in a microwave oven in a mixture of 4 mL of HNO3 and 0.5 mL of H2O2. Since Ca represents the matrix element in all samples including SRM 1486, no any additional chemical extraction of Ca was applied. Samples were mixed for double-spike measurements with the 43Ca-48Ca spike in a concentration ratio of ∼1:25, evaporated to dryness, and redissolved in 50 µL of subboiled 3 M HCl in order to guarantee complete chemical equilibrium. The final concentration of Ca was ∼300 ng µL-1. In order to avoid any possible contamination with environmental calcium, the same sample solutions were used both for TIMS and for ICPMS analyses. ICPMS Instrumentation and Measurement Procedure. An Elan DRC-II inductively coupled plasma mass spectrometer (Perkin-Elmer Sciex, Concord, ON, Canada) was used for calcium isotopic ratio measurements with NH3 as a reaction gas in the DRC. A PFA-100 nebulizer with a Cinnabar spray chamber was used for solution introduction. The optimization of the instrumental parameters was accomplished by investigation of the effects of single ICP-DRC-MS parameters on mass bias and precision of calcium isotopic ratios measurements and is described in more detail in the Results section. The optimized instrumental settings are summarized in Table 1. Table 1 uses the terminology of the Sciex software to define the data acquisition procedure. According to this terminology, each single measurement contains a definite number of “sweeps”, “readings”, and “replicates”, where the “reading” represents the mean of the sweeps and the “replicate” represents the mean of a set of “readings”. The DRC flush was set off during all isotopic measurements in this work to avoid perturbation of the ion beam, and the settling time in DRC mode was set to 200 µs instead of the usual value of 4000 µs. The chosen regime with the turned off flush pulse and the axial field voltage set to 0 V is a standard “isotope ratio regime” supplied by the manufacturer and can be selected in the software by selecting the appropriate method. The Ca concentration for ICP-DRC-MS analyses was chosen to provide high enough counting rates for the low-abundance 44Ca, while avoiding saturation of the pulse stage of the detector when measuring 40Ca. Accuracy of ICP-DRCMS can be affected by the uncertainty of the dead time of employed SEM (i.e., uncertainty introduced by dead time error can be larger that the measurement precision); hence, adjustment of 40Ca ion intensities to the same level for both samples and standards is required if relative measurements are performed. All sample and standard solutions were diluted therefore to the same final calcium concentration (∼60 ng mL-1 Ca) in order to achieve similar count rates for 40Ca+ (∼1.3 × 106 counts/s) in order to compensate for possible error introduced by dead time uncertainty onto the δ44Ca value. Possible interferences on 40Ca by 40K were controlled in a separate run by measuring the 41K+ intensity under the same experimental conditions.

Table 1. Experimental Parameters of DRC-ICPMS rf power, W plasma gas flow rate, L min-1 auxiliary gas flow rate, L min-1 nebulizer gas flow rate, L min-1 solution uptake rate, mL min-1 axial field voltage, V DRC pause Lens voltage, V RPa (Mathieu parameter a of the DRC) RPq (Mathieu parameter q of the DRC) cell gas flow, mL min-1 QRO, V CRO, V cell pass voltage (CPV), V dwell time (two isotopes), ms dwell time (four isotopes), ms detection mode settling time, ms number of sweepsa number of readingsa number of replicatesa analysis time per one measurement (block), min number of measurements (blocks), nb

1080 15 0.9 0.97 0.1 0 Off 3.6 0 0.33 (40Ca); 0.3 (44Ca) 1.05 -4 2.0 -25 1 (40Ca); 5 (44Ca) 1 (40Ca); 2 (43Ca); 2 (44Ca); 2 (48Ca) ion counting 0.2 400 16 9 ∼6.0 6 (optimization) 10 (determination of Ca isotope ratios in samples)

a According to terminology of the Sciex software, each single measurement contained 400 sweeps, 16 readings, and 9 replicates (except experiments shown in Figures S-3, S-4, and S-5, where the number of sweeps, readings, and replicates were adjusted depending on the used dwell time in order to keep the overall analysis time per one measurement of ∼6 min). b n is the number of sequential measurements (each single measurement of a 6-min duration).

Figure 1. Cup configuration for Ca isotope measurement by a Finnigan Triton mass spectrometer. 40Ca+, 41K+, 42Ca+, 43Ca+, and 44Ca+ are measured in the first sequence. 43Ca+ and 48Ca+ are measured in the second sequence. Cup C is fixed.

TIMS Instrumentation and Measurement Procedure. A thermal ionization mass spectrometer (Finnigan Triton, Bremen, Germany) was used for comparative Ca isotopic measurements with the detector configuration shown in Figure 1. A 1-µL sample (containing ∼300 ng of Ca) of calcium solution was placed on a rhenium evaporation filament and heated close to dryness at 0.7 A for TIMS measurements. Then the filament was heated at 1.5 A for 3 min. Finally, the electrical current was slowly increased to ∼2.2 A until a weak red glow was visible followed by an immediate shut down of the electrical current. The ionization rhenium filament was held at 2800 mA, and the current of the evaporation filament was increased to ∼1700 mA until a stable

signal of ∼10 V for 40Ca+ was achieved. The signal was focused during heating, and a peak centering procedure was performed. The baseline was recorded before each measurement block with the beam valve closed. Total measurement time for one analysis was 1 h. RESULTS Effects of the ICP-DRC-MS Parameters on the Mass Bias and Precision of Isotopic Measurements. After warming up the ICP-DRC-MS for ∼15-20 min, the plasma torch position and the nebulizer gas flow were optimized every time before performing isotopic measurements in order to achieve stable ion signals for the isotopes of interest (see details in Supporting Information (SI)). Because the parameters of the ion source and the ion optics (including the dynamic reaction cell) in an ICPMS can potentially influence both precision and accuracy of isotope ratio measurements, an optimization of these parameters was performed with account to their possible effects on the measured 44Ca/40Ca isotope ratio. The aims of this optimization were as follows: (i) achieving minimal standard deviation of the measured 44Ca/40Ca isotope ratio; (ii) finding such conditions where a possible deviation of particular experimental parameters would result in the smallest alteration of the measured isotope ratio; (iii) excluding any effect of uncontrolled variation of experimental conditions during the analysis on the measurement result. Panels a and b in Figure 2 represent an example of the measured 44Ca/40Ca isotope ratio and its precision (in the further text, the RSD is referred to as the relative standard deviation of a single measurement value calculated using the “unbiased” or “n - 1” method, n ) 6) as a function of the nebulizer gas flow rate and the rf power, respectively. The maximum sensitivity for 40Ca+ ions was observed at a nebulizer gas flow rate of 1.02 L min-1. The precision of the isotope ratio measurement was better than 0.05% RSD at nebulizer gas flow rates between 0.95 and 0.99 L min-1, but it becomes worse both at lower and at higher gas flow rates. A nebulizer gas flow of 0.95-0.99 L min-1 corresponded also to the most stable ion signal of 40Ca+. The observed mass bias changes with changing nebulizer gas flow rate. A similar effect was observed by Heilmann et al.9 in a study of sulfur isotopic measurements with a high-resolution sector-field ICPMS Element 2. Thus, changing parameters of the ion source resulted in a mass-dependent redistribution of ion densities in the ICPMS with a following change of extraction efficiencies for ions with different m/z ratios. Although the intermediate effects leading to changing mass bias may be different in a quadrupole and in a sector-field ICPMS due to significant differences in the extraction potentials and ion optics, the basic reason for changing mass bias is the effect of plasma temperature on ion energy. Thus, with increasing nebulizer gas flow, the central channel in an ICP expands and the temperature within this channel decreases resulting both in altering ion densities of Ar+ and of analyte ions and in reduction of the mean energy of ions. Ion lenses, serving as differential energy filters, will translate this effect into altering ion focusing efficiency. According to Bandura10 the measured 44Ca/40Ca isotope ratio will decrease with increasing plasma (9) Heilmann, J.; Boulyga, S.F.; Heumann, K. G. Anal. Bioanal. Chem. 2004, 380, 190-197. (10) Bandura, D. Institute for Biomaterials and Biomedical Engineering University of Toronto, Toronto, Canada, personal communication, 19.03.2007.

Analytical Chemistry, Vol. 79, No. 20, October 15, 2007

7755

Figure 2. 44Ca/40Ca isotope ratios and corresponding precision (RSD, n ) 6) as a function of nebulizer gas flow (a), rf power (b), lens voltage (c), and quadrupole rod offset (d). Each group of 6 experimental isotopic ratios represents one particular value of the nebulizer gas, rf power, lens voltage, and QRO. Other parameters are summarized in Table 1.

temperature at a fixed lens voltage optimized at a lower plasma temperature. On the other hand, the 44Ca/40Ca ratio will increase with lens voltage if all other parameters are fixed. Although the temperature of the plasma was not controlled in the present work, the observed behavior of 44Ca/40Ca isotope ratio seems to correspond to these considerations. Thus, an increasing rf power results in reduced mass discrimination, which decreases nearly linearly with increasing rf power (Figure 2b). However, the precision of measured 44Ca/40Ca isotope ratios deteriorated at higher rf power. In general, increasing the rf power is undesirable because this increases ionization efficiency of Ar atoms interfering with Ca. Figure 2c presents the mass bias as a function of the focus lens potential. The ions of different m/z, which enter the vacuum region of the mass spectrometer, are accelerated to the same final velocity due to supersonic beam expansion.11 As the plasma potential is low under the condition of a balanced rf generator of the ICPMS Elan 6100 DRC, the kinetic energy of these ions is different, which results in different focusing efficiency and in a dependency of the mass discrimination on the focus lens setting. Changing the quadrupole rod offset (QRO) relative to the cell rod offset (CRO) alters both ion trajectories within the DRC (trajectories may only partially coincide with the quadrupole aperture at certain ion energy) and the transmission efficiency of ions into the mass analyzer. Because ion trajectories in the quadrupole are mass dependent, changing the QRO introduces (11) Fulford, J. E.; Douglas D. J. Appl. Spectrosc. 1986, 40, 971-974.

7756 Analytical Chemistry, Vol. 79, No. 20, October 15, 2007

mass discrimination and may alter the precision of isotope ratio measurements (Figure 2d). According to Bandura et al.,5 the mass analyzer rod offset potential should be negative relative to the DRC rod offset potential, since the ions that are almost thermalized in the DRC are rejected otherwise. However, not only is the relative difference between CRO and QRO crucial but also their absolute values. Thus, the absolute values of the CRO and CPV define the depth of the potential well at the beginning of the cell and the potential barrier in the middle of the cell. Bandura et al.5 set CRO to +3 V and QRO to -3.5 V in order to increase the “time lag” of the cell response for effective ion beam homogenization. In the present work, a slightly positive CRO voltage of 2 V was also found optimal, as it creates some positive threshold potential enhancing thermalization of ions, which are then extracted into the quadrupole analyzer with a QRO set to -4 V. Pressurizing the dynamic reaction cell with ammonia not only leads to a reduction of the 40Ar+ intensity but also temporally homogenizes the ion beam and results in an improved precision of the measured 44Ca/40Ca isotope ratio (Figure 3). As can be expected from previous studies summarized by Tanner et al.,12 the observed mass bias is also altered due to a change of space charge density within the reaction cell as well as due to massdependent collisional focusing and scattering of ions in a pres(12) Tanner, S. D.; Baranov, V. I.; Bandura, D. R. Spectrochim. Acta, Part B 2002, 57, 1361-1452.

Figure 3. 44Ca/40Ca isotope ratio and corresponding precision (RSD, n ) 6) as a function of cell gas flow. Each group of 6 experimental isotopic ratios represents one particular value of the cell gas flow. Other parameters are summarized in Table 1.

surized rf-only quadrupole. Vanhaecke et al.13 demonstrated that various DRC parametersscollision gas flow rate, reaction gas flow rate, and band-pass settingssaffect the mass discrimination. Thus, other DRC parameters, such as axial field voltage and cell pass voltage, alter the ion transmission time within the reaction cell, which results in changing the space charge distribution. In the present work, the effect of those parameters on the mass bias has also been investigated. Finally, the axial field voltage was set to 0 in order to allow longer residence time of ions in the DRC and a collisional damping of ion density fluctuations of an ion beam extracted from an ICP. A slightly negative axial voltage can also be applied to decelerate ion motion in the reaction cell, but a similar effect was observed when the CRO was set to +2 V. In general, particular DRC parameters produce a similar effect on precision and observed mass bias in isotopic measurements, so that also optimal conditions other than those given in Table 1 can be found. When the DRC is operated with no axial field voltage applied, the effect of space charge on ion transmission through the cell can be significant. Hattendorf and Gu¨nther14 observed that large mass jumps (from U to Mg) require extended settling time for the DRC operated without an axial field; otherwise, mass discrimination is introduced. This effect is attributed to the fact that the space charge field, which is changing due to the change in the mass band-pass of the cell during the peak hopping, requires a certain settling time to be re-established. Since the axial field voltage was set to 0 in the present work, similar effectssalthough of much lower magnitude due to much smaller mass jumpswere observed: the 44Ca/40Ca ratio changed by up to 6% for very fast mass hopping (at settling + dwell time 20%) integral ion count rate, if keeping the total analysis time constant, and (ii) only a slightly shorter measurement cycle, which includes settling time and dwell time. As the abundance of 40Ca and 44Ca isotopes is significantly different, using the same dwell time for both isotopes is not an optimal condition for achieving good precision. According to Koirtyohann,17 the optimum ratio of dwell times follows the ratio

tm/tM ) xAM/Am where tM and tm are the dwell times used for the major and minor isotopes, respectively, and AM and Am are the relative isotopic abundances of the major and minor isotopes, respectively. Figure 4 compares the measured RSDs with corresponding theoretical precision values acquired at different dwell time ratios. (15) Bandura, D. Institute for Biomaterials and Biomedical Engineering University of Toronto, Toronto, Canada, personal communication, 12.05.2006. (16) Appelblad, P. K.; Rodushkin, I.; Baxter, D. C. Anal. Chem. 2001, 73, 29112919. (17) Koirtyohann, S. R. Spectrochim. Acta, Part B 1994, 49, 1305-1311.

Analytical Chemistry, Vol. 79, No. 20, October 15, 2007

7757

Figure 4. Measured (RSD, n ) 6) and theoretical precisions as a function of dwell time ratio. The sum of dwell times was kept constant (t(40Ca) + t(44Ca) ) 6 ms). Other parameters are summarized in Table 1.

In general, the experimental values are only slightly higher than the calculated and a dwell time ratio of 1:5 was found optimal. Even keeping all DRC-ICPMS parameters unchanged did not completely exclude the fluctuation of mass bias during 44Ca/40Ca isotope ratio measurements. The measured calcium isotope ratios changed with time if isotopic analysis with ICP-DRC-MS was started immediately after turning on the mass spectrometer. The shifting isotope ratio during measurement significantly affected the precision (Figure S-6 in SI). The isotope ratio stabilized only after 3-4 h of operating the ICPMS in DRC mode, and a good precision could be achieved. In addition, spontaneous changes of the 44Ca/40Ca isotope ratio occurred from time to time, which could be finally attributed to changes in laboratory conditions. Opening or closing the mass spectrometer hood during the measurements altered the temperature within the plasma box by ∼6 °C, leading to an immediate change of the measured 44Ca/40Ca ratio worsening isotope ratio precision (Figure S-7 in SI). Similar effects were observed by an abrupt alteration the capacity of the plasma exhaust connected to the central exhaust system (Figure S-8 in SI). We employed double-spiking of samples with 43Ca-48Ca in order to investigate whether such changes concern all isotopes of Ca. Both 44Ca/40Ca and 48Ca/43Ca isotope ratios could be measured with a satisfactory precision. Application of double-spiked samples for long-term stability study showed that 44Ca/40Ca and 48Ca/43Ca ratios behave the same. This fact can be interpreted as change of environmental laboratory conditions, which were later attributed to exhaust instability (see SI). In general, such deviations of mass bias can create problems when applying external standardization by bracketing samples with standards, and they should therefore be controlled during measurement. The precision of 44Ca/40Ca isotope measurements (the relative standard deviation of a single measurement value calculated using the unbiased or n - 1 method) was in the range of 0.03-0.05% under optimized conditions applying the following precautions: (1) The measurement was started after at least 3 h of constantly running the ICPMS in DRC mode without changing reaction gas flow. (2) Continuous aspiration of calcium solution was performed. (i.e., the ICPMS ran under wet plasma conditions only.) 7758 Analytical Chemistry, Vol. 79, No. 20, October 15, 2007

Determination of 44Ca/40Ca Isotopic Ratios in Ca ICP Standard and Digested SRM 1486 Bone Meal against SRM 915a. 44Ca/40Ca isotope ratios were analyzed in samples and evaluated relative to the isotope ratio in calcium carbonate NIST 915a (δ44Casample (‰) ) [(44Ca/40Casample)/(44Ca/40CaNIST 915a) - 1] × 1000). As possible inevitable change of environmental conditions can affect both the precision and accuracy of isotope ratio measurements, such effects were controlled and corresponding outliers were rejected. The criteria for the rejecting measured data were any changes in the ICP exhaust capacity, torch box temperature, and cell gas pressure (see SI for details) observed during the measurement sequence. Furthermore, isotope dilution with 43Ca-48Ca double spike was applied in order to compensate for time-dependent variations of mass bias. Isotope dilution with a spike (42Ca/48Ca) was introduced by Russell and Papanastassiou18 for more accurate measurements of the calcium isotopic composition by using TIMS. Later on, a 43Ca-48Ca double spike was prepared by Heuser et al.19 The latter spike was more preferable for TIMS measurements as the relative mass difference between spike isotopes is closer to the mass difference between 44Ca and 40Ca isotopes. Furthermore, the Faraday cup configuration of the Triton TIMS allows simultaneous recording of 43Ca and 48Ca isotopes. When using a mass spectrometer with a single ion detector in ICP-DRC-MS, four isotopes (40Ca, 43Ca, 44Ca, 48Ca) must be measured sequentially, and therefore, such an approach implicates additional losses of analyte ions due to a shorter measurement time per isotope. Nevertheless, despite the shorter integral measurement time per isotope, a good precision can be achieved (Figure S-8 in SI) at dwell times set to 2 ms for lower abundant 43Ca, 44Ca, and 48Ca isotopes and to 1 ms for 40Ca. Potential interferences originating from matrix elements can only be avoided by separation of the analyte from the matrix prior to isotopic measurement, or the isotopic ratios should be controlled by other independent methods. Thermal ionization mass spectrometry was applied therefore to assess the accuracy of the (18) Russell, W. A.; Papanastassiou, D. A. Anal. Chem. 1978, 50, 1151-1154. (19) Heuser, A.; Eisenhauer, A.; Gussone, N.; Bock, B.; Hansen, B. T.; Na¨gler, Th.F. Int. J. Mass Spectrom. 2002, 220, 385-397.

Figure 5. δ44Ca values measured in Ca ICP Standard and in SRM 1486 relative to SRM 915a and by using TIMS and ICPMS with doublespike calibration.

measured isotope ratios. After an aliquot of the isotope-diluted solutions was taken for TIMS analysis, the same solutions were analyzed by ICP-DRC-MS. A typical 44Ca/40Ca isotope ratio acquired by TIMS is shown in Figure S-9 in SI. An iterative approach for calculation of the isotope ratio in the spiked sample was adapted from Heuser et al.19 with account to detector configuration used in the present work. It should be mentioned that the measured values do not represent the true isotopic composition of the NIST SRM 915a calcium carbonate standard because the 43Ca/48Ca double-spike is not calibrated for the determination of the absolute 44Ca/40Ca ratio.19 In ICP-DRC-MS analyses the possible interference on 40Ca by 40K was controlled in a separate run by measuring the 41K+ intensity. The intensity of 41K was between 105 and 1.5 × 105 counts/s in all samples, corresponding to ∼2 × 102 counts/s of 40K+ (or ∼0.15 ‰ of 40Ca+ intensity). The results of TIMS and ICP-DRC-MS measurements with double-spike calibration are presented in Figure 5. The analysis time in both cases was 1 h per analysis (10 blocks, each 6 min). δ44Ca values (relatively to SRM 915a) measured by TIMS and ICPDRC-MS in SRM 1486 were 3.1 and 2.9‰, respectively. The δ44Ca values measured by TIMS and ICP-DRC-MS with double-spike calibration in the Ca ICP standard (3.2 and 4.1‰, respectively) coincided within the obtained reproducibility. It should be mentioned that the matrix of the Ca ICP standard and SRM 915a solutions was similar, whereas digested SRM 1486 contained along with Ca other elements like phosphorus and trace elements, which are characteristic for bone tissue. This matrix, for instance, severely affected the evaporation process of SRM 1486 in TIMS and deteriorated the signal stability and precision of Ca isotope ratio measurements (compare TIMS results for Ca ICP standard and SRM 1486 in Figure 5). The reproducibility of δ44Ca values measured both by TIMS and by ICP-DRC-MS in digested bone meal (SRM 1486) was worse when comparing to the pure Ca standard solution. Thus, concomitant elements might deteriorate reproducibility of Ca isotopic analysis even if calcium represents the major constituent part of the sample matrix.

CONCLUSIONS Collisional damping of the ICP noise in the dynamic reaction cell allows precise measurement of 44Ca/40Ca isotope ratio measured by peak hopping with a single ion detector. Collisional homogenization of the ion beam occurs with NH3 as a reaction gas, which is used at the same time for the removal of Ar+ interference on 40Ca. Precision of individual isotope ratio measurements of a 60 ng g-1 calcium solution (analysis time of 6 min) is routinely achievable in the range of 0.03-0.05%, which corresponds to a standard error of the mean value (n ) 6) of 0.0120.020%. The present work reports a principal possibility for achieving precision and reproducibility by using ICP-DRC-MS, which are close to that obtained by ICPMS and TIMS instruments equipped with multiple ion detectors provided all potential environmental effects influencing signal stability in ICP-DRC-MS are under control. The up-to-date experimental equipment does not allow, however, satisfactory recording of all parameters and control of their stability, which is required for achieving such a precision in routine analyses. Possible variations of experimental and environmental parameters result in drifts of mass bias in isotopic measurements. Therefore, the only way to correct for such drifts at present is applying internal standardization by measuring another known isotope ratio of the same element. This makes necessary measurement of an additional isotope pair reducing the overall analysis time per isotope or increasing the total analysis time. Although the applied isotope dilution with 43Ca-48Ca doublespike compensates for time-dependent deviations of mass bias and allows achieving accurate results, this approach is not optimal with respect to its higher cost and lower precision of ICP-DRC-MS analysis. Further development of external calibration by using a bracketing method would allow a wider use of ICP-DRC-MS for routine calcium isotopic measurements, but it still requires additional investigations to control all potential uncertainty sources influencing the measured isotope ratio. For that a detailed control and recording of the stability of experimental parameters by the ICPMS software or hardware is necessary. This would allow offAnalytical Chemistry, Vol. 79, No. 20, October 15, 2007

7759

line detection of accidental drifts and reliable subsequent rejection of outliers. The achievable good precision of calcium isotopic ratios measured by ICP-DRC-MS together with its relatively low analytical cost and straightforward sample preparation procedure will open better possibilities for a wider use of ICPMS for studying calcium isotopic composition, and this method can find potential use in geological, archaeometric, and biological applications as well as in authentication studies of goods and artifacts. ACKNOWLEDGMENT Dmitry Bandura is greatly acknowledged for the discussion of fundamental aspects of ICP-DRC-MS and for a number of useful comments and recommendations, which significantly improved

7760

Analytical Chemistry, Vol. 79, No. 20, October 15, 2007

this study. We thank Anton Eisenhauer and Ana Kolevica for supplying calcium double-spike. The authors acknowledge financial support by the Austrian Science Fund FWF (START project 267-N11). SUPPORTING INFORMATION AVAILABLE Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.

Received for review June 4, 2007. Accepted August 7, 2007. AC0711790