The determination of strontium isotope ratios by means of quadrupole-based ICP-mass spectrometry: a geochronological case study Frank Vanhaecke,*a Gu¡ínther De Wannemacker,a Luc Moensa and Jan Hertogenb aLaboratory of Analytical Chemistry, Ghent University, Proeftuinstraat 86, B-9000 Ghent, Belgium bPhysico-chemical Geology, University of Leuven, Celestijnenlaan 200C, B-3001 Leuven, Belgium Received 28th June 1999, Accepted 19th August 1999 Quadrupole-based ICP-mass spectrometry (ICP-QMS) was used for the determination of 87Sr/86Sr isotope ratios in digests of rock samples originating from two magmatic silicate rock formations of the Vosges (the Kagenfels granite and the Nideck rhyolite).The rock formations studied are geographically close to one another and, overall, they show a similar chemical composition. In a preliminary study, the effect of various data acquisition parameters on the isotope ratio precision was systematically studied, permitting optimum conditions to be selected.Rh was used as an internal standard, allowing the blank correction to be made accurately. Cation exchange chromatography was used to avoid isobaric overlap of 87Rbz and 87Srz ion signals to the largest possible extent, while mathematical correction was applied to correct for the remaining interference. The accuracy of the method developed was evaluated by means of isotopic analysis of an oceanic gabbro sample, for which the 87Sr/86Sr isotope ratio was previously characterized by means of thermal ionisation mass spectrometry (TIMS).An excellent agreement between the ICP-QMS and TIMS values was established. On the basis of the isochrons, constructed using the 87Sr/86Sr isotope ratio results and the contents of Rb and Sr (determined by energy-dispersive X-ray �ªuorescence spectrometry) obtained for the Kagenfels and Nideck samples, it could be concluded that for both rock formations, secondary processes (e.g., recrystallisation, high- and low-temperature alteration) have disturbed the Rb¡¾Sr isotopic system.As a consequence, the uncertainties on (i) the initial 87Sr/86Sr isotope ratios and (ii) the ages thus determined are large. Nevertheless, the estimated ages appear to be geologically relevant and provide information on the timing of geological events that affected the rocks several tens of million years after the initial formation. Overall, this case study shows the merits of ICP-QMS for exploratory studies of Sr isotope systematics and geochronology in cases with suf�¡cient variation in the 87Sr/86Sr isotope ratios.Introduction In general, the isotopic composition of the elements is constant in nature.1 However, some exceptions1 exist as a result of (i) mass fractionation,2 (ii) natural radioactivity,3,4 (iii) the interaction of cosmic rays with matter (primarily in the atmosphere)3,4 and (iv) human activity, while (v) deviating isotopic compositions have also been observed for some elements in special classes of meteorites.5 The isotopic composition of Sr in terrestrial material shows variations as a result of the b2-decay of the naturally occurring long-lived radionuclide 87Rb (half-life T1/2~ 48.86109 y) to the stable isotope 87Sr.3,4 As a result, the number of 87Sr atoms actually present after a given time interval t is the sum of those present at the start of the interval (initial 87Sr), plus those produced by the decay of 87Rb during the interval t: 87Srt a87Sri a87RbtOelt ¢§ 1U O1U in which l is the decay constant (~ln 2/T1/2).While it is extremely dif�¡cult to measure the absolute abundance of an isotope, mass spectrometric techniques readily permit accurate and precise determination of isotope ratios. Therefore, it is more convenient to rewrite equation (1) in terms of the 87Sr/86Sr isotope ratio by division through 86Sr (for which the number of atoms present remains unaffected by radioactive decay, i.e., 86Sri~86Srt): O87Sr=86SrUt a O87Sr=86SrUi a O87Rb=86SrUtOelt ¢§ 1U O2U In the geochronological context, the start of the interval can coincide with the formation of a rock or mineral by solidi�¡cation of a silicate melt or with the recrystallisation and isotopic rehomogenisation of an existing rock.Equation (2) is the equation of a straight line in (87Sr/86Sr)¡¾(87Rb/86Sr) coordinates. It is commonly called the `isochron equation', because it expresses the (87Sr/86Sr) ratio of isochronous rocks (same t) that formed from the same, isotopically homogeneous source (same initial ratio), but having different 87Rb/86Sr ratios.3,4 However, this also requires that the system has remained closed with respect to Rb and Sr concentrations during the time interval t.In practice, the age of two or more co-genetic rocks or minerals is derived from the slope of the best �¡t line through the experimentally obtained data points; the intercept yields the initial (87Sr/86Sr) value.Traditionally, thermal ionisation magnetic sector mass spectrometry (TIMS) is used for Rb/Sr geochronology, because most applications require a precision of better than 0.01% relative standard deviation (RSD) on the 87Sr/86Sr ratio. In comparison with TIMS, the isotope ratio precision that can be J. Anal. At. Spectrom., 1999, 14, 1691¡¾1696 1691 This Journal is # The Royal Society of Chemistry 1999obtained with quadrupole-based ICP-mass spectrometry (ICP-QMS) is relatively poor (¢0.1% RSD).6±14 However, for applications for which the ultimate level of precision is not required, ICP-QMS is an attractive alternative to TIMS, owing to its ease of operation (samples can be introduced as aqueous solutions at atmospheric pressure), the widespread availability of ICP-QMS instruments and the much higher sample throughput.Of course, for both ICP-MS and TIMS, the time required for sample preparation exceeds the time required for the actual measurements.Highly favourable cases for ICP-QMS are preliminary dating or isotopic screening studies of rock series in which the expected variation of 87Sr/86Sr ratios exceeds 1% as a result of a combination of relatively great age (w50 million years) and large variations of the Rb/Sr ratios. The present paper consists of two parts. In the Ærst one, the optimisation of the data acquisition parameters for Sr isotope ratio measurements with ICP-QMS is brieØy described.The second part deals with a case study of Rb±Sr dating of two related series of magmatic silicate rocks from the Hercynian (ca. 350 to 250 million years old) Vosges Massif, France. While the application of ICP-MS for Re/Os dating has already been reported in the literature by several authors,5,15,16 to the best of the authors' knowledge this is not the case for Rb/Sr dating, although Chassery et al.17 and Latkoczy et al.18 recently reported on the use of quadrupole-based and sector Æeld ICP-MS, respectively, for the determination of Sr isotope ratios.Experimental Sample preparation A slab weighing 300±500 g was cut with a rock saw with a diamond blade from the interior part of the rock samples collected in the Æeld. Exterior, slightly weathered parts of the slab were removed with a small rock saw. The cleaned slabs were coarsely crushed in a carbon-steel mortar. About 40 g of the homogenised crushed material was ground to Æne powder in a mechanical agate vibrating ball mill.Determination of Rb and Sr concentrations Samples for isotopic measurements were selected on the basis of their Rb and Sr contents, in order to assure a large spread of Rb/Sr ratios. Rb and Sr were determined by energy-dispersive X-ray Øuorescence spectrometry (EDXRF) on `inÆnitely thick' pressed powder pellets. About 3 g of ground silicate powder were pressed into 24 mm diameter pellets, after mixing with a few drops of 2.5% (m/m) aqueous solution of poly(vinyl alcohol) binder.Samples were measured with a Kevex 700 spectrometer (Kevex Instruments Inc., Redwood City, CA, USA), equipped with a Rh-target X-ray tube. Operating conditions were as follows: 40 kV tube voltage, 0.40 mA tube current, Ag secondary target and 30 mm2 SiLi detector collimator. International reference rocks were used as standards. Every sample was measured in duplicate relative to two standards. Full details of the analytical procedure are reported elsewhere.19 Ion-exchange separation of Sr from Rb Self-evidently, for the Sr isotope ratio measurements, the samples of interest had to be taken into solution.Additionally, due to the isobaric overlap of the 87Srz and 87Rbz ion signals, Sr and Rb had to be chemically separated from one another by cation-exchange chromatography prior to the Sr isotope ratio measurements. Approximately 1 g of powdered sample was dissolved in a mixture of 28 M HF and 14 M HNO3 in a PTFE vessel.After evaporation to dryness, the residue was taken up in 10 ml of 2 M HCl. The solution was Æltered through a Whatman 541 Ælter, to retain any gelatinous particles. The Ælter was washed using 5 ml of 2 M HCl. For the cation exchange procedure, a Dowex 50W-X8 column (length, 20 cm; internal diameter, 1 cm) was used (J.T. Baker Chemicals N.V., Deventer, The Netherlands). Na, K, Rb, Mg and the major part of Fe and Ca were eluted from the column using 90 ml of 2 M HCl.Subsequently, Sr was eluted from the column using 40 ml of 8 M HCl. For a gabbroic control sample, a longer column was used because the fairly high Ca content resulted in an accelerated elution of Sr. The Sr-containing fraction was subsequently evaporated to near-dryness and converted into the nitrate form by addition of 1.5 ml of 14 M HNO3 and evaporation to near-dryness. Finally, the residue was taken up in 2.5 ml of Millipore Milli-Q water and diluted appropriately using 0.14 M HNO3.The Sr concentration in the solutions analysed was typically 100±150 mg l21. Sr isotope ratio measurements The instrument used for the Sr isotope ratio measurements is a Perkin Elmer SCIEX ELAN 5000 quadrupole-based ICP-mass spectrometer, in its standard conÆguration (Perkin Elmer, U» berlingen, Germany). A multi-channel peristaltic pump (Minipuls-3), a GemTip cross-Øow nebuliser and a Perkin Elmer Type II spray chamber made of Ryton, drained by the peristaltic pump, were used for sample introduction.This instrument was further equipped with a Perkin Elmer corrosion-resistant torch with standard alumina injector and a Channeltron continuous dynode electron multiplier, operated in the pulse counting mode. Typical operation conditions have been summarised in Table 1. Processing of ICP-QMS data The raw data obtained using the data acquisition parameters shown in Table 2 were processed in the following way. Prior to ratioing, the isotopes of interest were corrected for both (i) signal losses to be attributed to the detector dead time1,20 and (ii) the procedure blank.For the former correction, the detector dead time was determined experimentally on each measuring day according to the procedure described by Russ1 and the value obtained was loaded into the instrument software for automatic correction. In order to improve the reliability of the blank correction, 103Rh was also monitored and used as an internal standard, correcting for potential matrix-induced signal suppression or enhancement, signal drift and instrument instability.The monitoring of 103Rh resulted in a slight deterioration of the isotope ratio precision obtained. It is common knowledge that in ICP-MS, various phenomena cause the relative ion intensities (on a molar basis) to vary as a function of the ion mass. These mass discrimination effects can occur during extraction (nozzle separation effect13), transmission (space charge effects21,22) or detection, and they can amount to several per cent.per mass unit. It is self-evident that for accurate isotope ratio determination, mass discrimina- Table 1 Operation conditions of the Perkin Elmer SCIEX ELAN 5000 ICP-mass spectrometer Rf power 1000W Sampling depth 10 mm from load coil Gas Øow rates Nebulizer gas 0.870 l min21 Auxiliary gas 1.2 l min21 Plasma gas 15 l min21 Sampling cone Nickel, 1.0 mm aperture diameter Skimmer Nickel, 0.75 mm aperture diameter Sample uptake rate 1 ml min21 Lens voltages Tuned for maximum 103Rhz signal intensity 1692 J.Anal. At. Spectrom., 1999, 14, 1691±1696tion has to be appropriately corrected for. In the case of Sr, this correction either (i) involves measurement of an (external) isotope ratio standard with a known isotopic composition or (ii) can be accomplished by monitoring the 86Sr/88Sr isotope ratio, which is constant in nature.23,24 In this work, external standardisation, using NIST SRM 987 (SrCO3) as an isotope ratio standard, was preferred (NIST, Gaithersburg, MD, USA). Comparison of the experimental 87Sr/86Sr isotope ratio result obtained for a Sr standard solution prepared from this reference material and the corresponding certi�¡ed value permitted calculation of a mass discrimination correction factor (CF): CF a O87=86Ucert O87=86Ustd;exp O3U As the experimental result for the 87Sr/86Sr isotope ratio of the standard is the average value of 10 replicate measurements, error propagation, taking into account the standard deviations for both this experimentally determined isotope ratio (sstd,exp) and the corresponding certi�¡ed value (scert), permits the standard deviation on the mass discrimination correction factor (sCF) to be calculated: sCF a CF| AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA sstd;exp O87=86Ustd;exp2 a scert O87=86Ucert 2 s O4U Prior to mass discrimination correction, the results for the samples have also been corrected for the potential bias caused by remaining isobaric overlap of 87Srz and 87Rbz: O87=86Usamp;net a fO87=103Usamp ¢§ O87=103Ubl ¢§ OCF'|0:3857 svp| aO85=103Usamp ¢§ O85=103UblaUg aO86=103Usamp ¢§ O86=103Ubla O5U in which CF'~1z[2(12CF)], assuming that mass discrimination varies linearly as a function of mass, and 0.3857 svp is the 87Rb/85Rb isotope ratio as calculated from IUPAC-tabulated abundances.25 ssamp,net is the standard deviation observed for the results of 10 replicate measurements, calculated according to equation (5).Finally, the `true' isotope ratio for the samples is given by: O87=86Utrue a O87=86Usamp;net|CF O6U and the corresponding standard deviation is given by: strue a AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA sCF CF2 a ssamp;net O87=86Usamp;net2 s |O87=86Utrue O7U On the basis of the ICP-MS (87Sr/86Sr isotope ratios) and EDXRF (Sr and Rb contents) results obtained, the isochrons for both rock formations could be constructed.The content of 87Rb can be calculated from the corresponding Rb elemental content and the isotopic abundance from reference.25 The content of 86Sr, on the other hand, can be calculated from the Sr content and the isotopic abundance (h) of 86Sr, which can be obtained by solving the following set of simultaneous equations (8): hO84U a hO86U a hO87U a hO88U a 1 hO88U hO86U a constant a 82:58 9:86 a 8:375 hO84U hO86U a constant a 0:56 9:86 a 0:057 hO87U hO86U a determined experimentally OICP-MSU a R O8U 8><>: Rearrangement of equation (8) leads to equation (9): hO86U a 1 O9:432 a RU O9U permitting the 87Rb/86Sr ratio to be calculated.Results and discussion Optimisation of data acquisition parameters Preliminary work consisted of a systematic evaluation of the in�ªuence of the data acquisition parameters on the isotope ratio precision and permitted optimum conditions to be selected.For this purpose, the relative standard deviation [RSD(%)] for 10 replicate measurements of the 87Sr/86Sr isotope ratio in a 100 mg L21 Sr standard solution was determined under different conditions. In order to evaluate the in�ªuence of the scanning rate, or more accurately the peak hopping rate, the residence time per acquisition point was varied from 10 to 500 ms (10, 30, 50, 100, 200 and 500 ms) while keeping the total acquisition time per replicate at 2 min.As has already been reported by several research groups, the isotope ratio precision was observed to signi�¡cantly improve with increasing scanning speed:6,9,12,26 forsotope ratio determination, the scanning rate should be suf�¡ciently high to smooth out signal �ªuctuations due to plasma �ªicker and variations in, for example, the sample uptake rate and the nebulisation, ionisation and extraction ef�¡ciencies. Denoyer27 argued that in general, an acquisition time of 30 ms is an optimum compromise between fast hopping through the nuclides monitored and an ef�¡cient use of the total measuring time (high ratio of actual measuring time to mass spectrometer settling time); however, it is interesting to note that in our study a decrease of the acquisition time from 30 to 10 ms resulted in an additional, though small, improvement (by a factor of 1.2 on average) of the isotope ratio precision.Hence, for the actual measurements, a residence time of 10 ms per acquisition point was used.In order to evaluate the in�ªuence of the total measurement time on the isotope ratio precision obtained, the RSD for 10 replicate 87Sr/86Sr measurements was determined using a total measurement time of 5, 15, 30, 60 and 150 s per nuclide monitored and per replicate. As can be expected on the basis of Poisson counting statistics, the isotope ratio precision signi�¡cantly improves when increasing the measurement time and hence the number of counts observed for both nuclides monitored (Fig. 1). In all further work, a compromise setting of 45 s measurement time per nuclide and per replicate was used, enabling a suf�¡cient isotope ratio precision to be obtained while keeping the total acquisition time (for 10 replicates) at an acceptable level. As is to be expected,28 no improvement in the 87Sr/86Sr isotope ratio precision was observed when using 5 instead of 1 acquisition point per spectral peak (in both cases residence time per acquisition point~10 ms), while keeping the total acquisition time per replicate at 2 min.In summary, the aforementioned optimisation experiments Table 2 Data acquisition parameters (after optimization, see text) Data acquisition mode peak hop mode Nuclides monitored 86Sr, 87Sr, 85Rb, 103Rh Number of acquisition points per peak 1 Residence time per acquisition point and per sweep 10 ms Number of sweeps per replicate 3250 Total measurement time per replicate y3 mina Number of replicates 10 aIncluding quadrupole settling time.27 (8) J.Anal. At. Spectrom., 1999, 14, 1691¡¾1696 1693have led to the set of data acquisition parameters shown in Table 2. Finally, the effect of `normalisation' on the isotope ratio precision was evaluated. Although the isotopic composition of Sr varies in nature as a result of the b2-decay of 87Rb to 87Sr, the 86Sr/88Sr isotope ratio is constant and can be used as an `internal standard', correcting for drift or instabilities in the 87Sr/86Sr isotope ratio.For multi-collector ICP-MS instrumentation, successful use of such a normalisation procedure has been reported by Walder and Freedman23 and by Christensen et al.24 As a slight deterioration in the 87Sr/86Sr isotope ratio precision (by a factor of 1.3 on average) was observed when using 86Sr/88Sr as an internal standard, this internal `normalisation' was not used for the actual measurements. As both experiments were carried out using the same residence time per acquisition point and the same total measurement time (only the number of sweeps was varied), this observation can be attributed to a combination of (i) a smaller number of counts observed for each of the nuclides of interest (cf.Poisson counting statistics) and possibly also (ii) a longer time interval between two successive sweeps (monitoring of 3 nuclides instead of 2). Using the optimised conditions, summarised in Table 2, the RSD for 10 replicate measurements of the 87Sr/86Sr isotope ratio in a 100 mg L21 standard solution was observed to be typically 0.1±0.2%.Evaluation of accuracy Before analysing the samples of interest, the accuracy of the procedure developed was evaluated using a rock sample, for which the 87Sr/86Sr isotope ratio was previously characterized by means of multi-collector TIMS by Barling et al.29 This `quality control sample' is an oceanic gabbro drilled from the Mid-Atlantic Ridge.It was subjected to the same sample pretreatment (dissolution and cation exchange chromatography) as the samples. The ICP-MS result obtained, 0.7028 with a standard deviation of 0.0020 (n~10), is in excellent agreement with that obtained by TIMS, 0.702 628 (standard error, 0.000 008). A geochronological case study: granites and rhyolites from the Vosges Massif, France The latest manifestations of Late-Hercynian (i.e., from ca. 320 to 250 million years ago) magmatic activity in the Vosges Massif are the Kagenfels granite and the Nideck rhyolite, both located in the Northern Vosges, France.The Nideck rhyolite is an ignimbrite, i.e., a silicon-rich hot volcanic ash deposited during a powerful volcanic eruption. The Kagenfels granite on the other hand is a plutonic rock formed by slow cooling of a silicon-rich silicate that intruded at rather shallow level in the crust of the earth. The granite is now exposed at the surface due to uplift and erosion of the original cover rocks. The geochemistry and petrology of the two rock formations are currently studied in detail at the University of Leuven and the corresponding results have been30±32 and will be reported on elsewhere.The rocks are very suited for a test case, because several geochronological techniques have already been applied to date the two formations. However, these previous datings have yielded considerably divergent results.33 Hess et al.33 obtained a consistent set of ages for the Kagenfels granite, on the basis of K/Ar and 40Ar/39Ar dating of biotite mineral separates and of Pb-evaporation dating of zircon grains.The formation age of the Kagenfels granite, 331°5 Ma, is apparently 40 Ma older than previously assumed, and older than the biotite 40Ar/39Ar age of 291°4 Ma measured for the Nideck rhyolite. Hence, there is a need to further investigate the reasons for the scattered age data. For the present study, representative samples were selected that showed a very signiÆcant variation of the Rb/Sr ratio.The results obtained for Kagenfels granite and Nideck rhyolite are summarised in Table 3. As can be seen from this table, the typical 87Sr/86Sr isotope ratio precision (expressed as relative standard deviation for 10 replicates) obtained in practice was typically y0.3%, while 2% RSD was used as a realistic estimation of the repeatability of the Rb/Sr results obtained using EDXRF.For one sample, the standard deviation on the 87Sr/86Sr isotope ratio was seen to be atypically high (y1.6%). For this particular sample, the chromatographic separation of Sr from Rb was established not to be very successful, as the Rb concentration in the sample signiÆcantly exceeded that of Sr. Nevertheless, it was decided to take this value into account also, as for isochron construction the statistical weight of an experimental result and its uncertainty are inversely proportional. 34±36 The `Isoplot/Ex' calculation procedure developed by K.R. Ludwig37 was used to analyse and to calculate the parameters of the isochrons. This procedure performs a least squares analysis of the data taking into account the uncertainties of both the X- and Y-parameters. The results are shown in numerical and graphical form in Figs. 2 and 3. Both the Nideck and the Kagenfels data deÆne fairly well correlated linear trends (linear correlation coefÆcients are 0.988 and 0.976, respectively), but the scatter from the best-Æt lines is in both cases larger than what is expected from the instrumental, analytical errors.It thus appears that the Rb±Sr isotopic system has been disturbed by secondary processes (e.g., recrystallisation, high- and low-temperature alteration) since the emplacement. As a consequence, the uncertainty of the estimated initial isotopic ratios and age is quite substantial. Nevertheless, these results can be used. Table 3 Experimentally determined 87Sr/86Sr isotope ratios (ICPQMS) and 87EDXRF).Uncertainties are expressed as standard deviations Kagenfels granite 87Rb/86Sr 87Sr/86Sr Sample 1 5.98 (0.12) 0.7349 (0.0018) Sample 2 19.88 (0.40) 0.8113 (0.0022) Sample 3 38.91 (0.78) 0.9062 (0.0050) Sample 4 57.50 (1.2) 0.8968 (0.0027) Sample 5 97.30 (1.95) 1.1149 (0.0031) Nideck rhyolite 87Rb/86Sr 87Sr/86Sr Sample 1 5.18 (0.10) 0.7263 (0.0020) Sample 2 7.79 (0.16) 0.7323 (0.0019) Sample 3 18.83 (0.38) 0.7607 (0.0021) Sample 4 22.18 (0.44) 0.7910 (0.0035) Sample 5 39.57 (0.79) 0.835 (0.013) Fig. 1 87Sr/86Sr isotope ratio precision [expressed as RSD(%) for n~10] as a function of the measurement time per nuclide. 1694 J. Anal. At. Spectrom., 1999, 14, 1691±1696The calculated, nominal age of 274 Ma (million years) of the Kagenfels is within the range of earlier Rb±Sr whole-rock datings and much younger than the most likely age of formation of 331 Ma.33 The nominal age of 232 Ma derived for the Nideck ignimbrite is deÆnitely too young to correspond to the volcanic eruption age.A geologically reasonable minimum age of formation is about 250 Ma, because the rhyolites are directly overlain by Upper Permian and Lower Triassic sediments. It is interesting that some of the biotite fractions from the Kagenfels granite analysed by Hess et al.33 also yielded apparent K/Ar ages of 260±274 Ma. Hence, it is rather probable that the whole rock Rb±Sr age obtained in the present study reØects a real geological event.A likely process is the partial resetting of the Rb±Sr clock due to alteration of potassic feldspar, a major host phase of Rb and Sr, by hydrothermal Øuids during late-Hercynian uplift and erosion of the Vosges Massif. Fluid circulation is promoted by the development of cracks and faults in uplifted granites due to pressure release. As the Nideck rhyolites were already deposited on the surface at the moment of volcanic eruption, uplift or subsidence should have had a negligible effect on the Rb±Sr system.The young apparent Rb±Sr age of the Nideck rocks (232 Ma) must be attributed to another geological process, such as the recrystallisation and alteration of volcanic glass particles when the permeable volcanic ash became covered by water masses during Late-Permian and Early-Triassic times. Conclusions Although the isotope ratio precision obtained is clearly signiÆcantly poorer than that obtained with TIMS, the results for the granitic and rhyolitic magmatic rock series from the Vosges Massif demonstrate that relevant geochronological data can be derived from ICP-QMS measurements.The case study clearly shows the merits of ICP-MS for exploratory studies of Sr-isotope systematics and geochronology in cases where the expected variation in the 87Sr/86Sr ratio is larger than the analytical precision. The main advantages of quadrupolebased ICP-MS are a high sample throughput, straightforward sample introduction and, especially, the wide availability of ICP-MS instruments in analytical and geochemical laboratories.ICP-MS can be very useful in preliminary studies aiming at a selection of the most interesting specimens out of a larger batch of samples for subsequent TIMS analysis. Another useful application is a quick check of the state of disturbance of the Rb/Sr system in a given rock series. For highly precise isotope ratio determinations, however, ICP-MS cannot compete with TIMS, unless a double focusing magnetic sector ICP-MS instrument, equipped with a multiple collector detection device, is used.23,24,38 References 1 G.P. Russ III, in Applications of Inductively Coupled Plasma Mass Spectrometry, ed. A. R. Date and A. L. Gray, Blackie, Glasgow, UK, 1989, ch. 4, pp. 90±114. 2 T. B. Coplen, J. A. Hopple, S. E. Rieder, H. R. Krouse, J. K. Bo» hlke, R. D. Vocke Jr., K. M. Re�ve�sz, K. J.R. Rosman, A. Lamberty, P. Taylor and P. De Bie¡vre, Pure Appl. Chem., 1999, in the press. 3 G. Faure, Principles of Isotope Geology, John Wiley, New York, USA, 2nd edn., 1986. 4 A. P. Dickin, Radiogenic Isotope Geology, Cambridge University Press, Cambridge, UK, 1995. 5 D. C. Gre�goire, Prog. Anal. Spectrosc., 1989, 12, 433. 6 N. Furuta, J. Anal. At. Spectrom., 1991, 6, 199. 7 M. E. Ketterer, M. J. Peters and P. J. Tisdale, J. Anal. At. Spectrom., 1991, 6, 439. 8 M. E. Ketterer, J.Anal. At. Spectrom., 1992, 7, 1125. 9 S. R. Koirtyohann, Spectrochim. Acta, Part B, 1994, 49, 1305. 10 Q. Xie and R. Kerrich, J. Anal. At. Spectrom., 1995, 10, 99. 11 I. S. Begley and B. L. Sharp, J. Anal. At. Spectrom., 1997, 12, 395. 12 C. R. Que�tel, B. Thomas, O. F. X. Donard and F. E. Grousset, Spectrochim. Acta, Part B, 1997, 52, 177. 13 K. G. Heumann, S. M. Gallus, G. Ra»dlinger and J. Vogl, J. Anal. At. Spectrom., 1998, 13, 1001. 14 F. Vanhaecke, P. Taylor and L.Moens, in ICP Spectrometry and its Applications, ed. S. J. Hill, ShefÆeld Academic Press, ShefÆeld, UK, 1999, ch. 6, pp. 145±207. 15 S. B. Beneteau and J. M. Richardson, At. Spectrosc., 1992, 13, 118. 16 T. Hirata and A. Masuda, Meteoritics, 1992, 27, 568. 17 S. Chassery, F. E. Grousset, G. Lavaux and C. R. Que�tel, Fresenius' J. Anal. Chem., 1998, 360, 230. 18 C. Latkoczy, T. Prohaska, G. Stingeder and M. Teschler-Nicola, J. Anal. At. Spectrom., 1998, 13, 561. 19 L. Vanlerberghe and J.Hertogen, Bull. Soc. Chim. Belg., 1986, 95, 491. 20 F. Vanhaecke, G. De Wannemacker, L. Moens, R. Dams, C. Latkoczy, T. Prohaska and G. Stingeder, J. Anal. At. Spectrom., 1998, 13, 567. 21 G. R. Gillson, D. J. Douglas, J. E. Fulford, K. W. Halligan and S. D. Tanner, Anal. Chem., 1988, 60, 1472. 22 S. D. Tanner, Spectrochim. Acta, Part B, 1992, 47, 809. 23 A. J. Walder and P. A. Freedman, J. Anal. At. Spectrom., 1992, 7, 571. 24 J. N. Christensen, A. N. Halliday, D.-C. Lee and C. M. Hall, Earth Planet. Sci. Lett., 1995, 136, 79. Fig. 3 Rb/Sr isochron for Nideck rhyolite, constructed on the basis of the experimentally determined 87Sr/86Sr isotope ratios (ICP-QMS) and Rb and Sr contents (EDXRF). Fig. 2 Rb/Sr isochron for Kagenfels granite, constructed on the basis of the experimentally determined 87Sr/86Sr isotope ratios (ICP-QMS) and Rb and Sr contents (EDXRF). J. Anal. At. Spectrom., 1999, 14, 1691±1696 169525 K. J. R. Rosman and P. D. P Taylor, J. Anal. At. Spectrom., 1998, 13, 45N. 26 F. Vanhaecke, L. Moens, R. Dams and P. Taylor, Anal. Chem., 1996, 68, 567. 27 E. R. Denoyer, At. Spectrosc., 1994, 15, 7. 28 H. P. Longerich, S. E. Jackson and D. Gu» nther, J. Anal. At. Spectrom., 1996, 11, 899. 29 J. Barling, J. Hertogen and D. Weis, in Proceedings of the Ocean Drilling Program, ScientiÆc Results, ed. J. A. Karson, M. Cannat, D. J. Miller and D. Elthon, Ocean Drilling Program, College Station, TX, USA, 1997, vol. 153, 351. 30 S. Claes, M. Verhaeren and J. Hertogen, Terra Nova, 1995, vol. 7, Abstracts Suppl. 1, p. 301. 31 M. Verhaeren, S. Claes and J. Hertogen, Terra Nova, 1995, vol. 7, Abstracts Suppl. 1, p. 141. 32 J. Mareels and J. Hertogen, unpublished work. 33 J. C. Hess, H. J. Lippolt and B. Kober, Geol. Rundsch., 1995, 84, 568. 34 G. A. McIntyre, C. Brooks, W. Compston and A. Turek, J. Geophys. Res., 1966, 71, 5459. 35 D. York, Earth Planet. Sci. Lett., 1967, 2, 479. 36 D. York, Earth Planet. Sci. Lett., 1969, 5, 320. 37 K. R. Ludwig, Berkeley Geochronology Center, Special Publication No. 1, Rev. November 5, 1998, Berkeley, CA, USA. 38 A. N. Halliday, D.-C. Lee, J. N. Christensen, M. Rehka»mper, W. Yi, X. Luo, C. M. Hall, C. J. Ballentine, T. Pettke and C. Stirling, Geochim. Cosmochim. Acta, 1998, 62, 919. Paper 9/05184H 1696 J. Anal. At. Spectrom., 1999, 14, 1691&p