Calibration Strategies for the Elemental Analysis of Individual Aqueous Fluid Inclusions by Laser Ablation Inductively Coupled Plasma Mass Spectrometry Journal of Analytical Atomic Spectrometry A. MOISSETTE British Geological Survey Kingsley Dunham Centre Keyworth Nottingham UK NG12 5GG and CREGU Centre de Recherches sur la Giologie des Matibres Premibres iMinCrales et Energitiques BP 23 54501 Vandoeuvre les Nancy Cedex France T. J . SHEPHERD AND S. R. CHENERY British Geological Survey Kingsley Dunham Centre Keyworth Nottingham UK NG12 5GG Using a combination of synthetic fluid inclusions in halite microvolume aqueous solutions and National Institute of Standards and Technology (NIST ) Standard Reference Material (SRM) 611 Glass calibration graphs were established for the determination of elemental ratios in natural fluid inclusions by laser ablation-inductively coupled plasma mass spectrometry (ICP-MS).For simultaneous multi-element analysis optimization studies demonstrate the necessity to adopt a compromise set of operating conditions since ICP-MS sensitivity (signal backgrouiid) may differ from element to element as a function of argon flow radiofrequency power and spray chamber temperature. Synthetic fluid inclusions were prepared by crystallization from saturated sodium chloride solutions containing up to 13 major and minor cations. The microvolume calibration standards 'microwells' consisted of small holes (3 x 3 x 2 mm3) drilled into plastic sheet filled with a standard solution and hermetically sealed. In order to allow direct comparison between the different test materials all the elements (Li Na Mg K Ca Mn Cu Zn Rb Cs Ba Pb B Cl Br) were ratioed to strontium.The relative standard deviations for the element ratios were generally better than 25% indicating that the nature of the sample (salt glass and aqueous solution) does not markedly affect the consistency of ablation or the efficiency of transfer between the ablation chamber and ICP torch. Element ratios for the synthetic fluid inclusions were linear over several orders of magnitude and in close agreement with those for the NIST SRM 611 Glass and microwell solutions irrespective of inclusion size (20-100 pm) and depth in the sample (up to 80 pm). Statistical t-tests on the mean element ratios confirm that microwells and glasses constitute suitable alternatives to synthetic fluid inclusions for the calibration and routine analysis of natural fluid inclusions.Keywords Laser ablation; inductively coupled plasma mass spectrometry; Juid inclusion; laser microanalysis; calibration For the accurate modelling of palaeofluid-mineral equilibria the limiting factor is often a lack of information on the composition of the fluid phase. Fluid inclusions are known to hold this information but their size abundance and diversity often within a single crystal complicates the acquisition of unequivocal bulk chemical data for fluid(s) in equilibrium with the mineral phase. Analysis of individual fluid inclusions is the most satisfactory solution to this problem. Several techniques have been dewloped for this purpose.' Microthermometry is useful for determining the bulk composition and density of fluid inclusions but detailed interpretation is limited by the need to refer highly complex multi-component geological fluids to very simple (two- three- or four-component) exper- imental sy~tems.~?~ Various spectroscopic techniques have been adapted for single inclusion analysis ultraviolet (UV)-visible or fluorescence spectroscopy for the determination of organic compo~nds;~*~ micro-Raman and infrared (IR) spectroscopy for the determination of polyatomic For the ele- mental analysis of inclusion fluids techniques have included X-ray microanalysis of frozen inclusions;" proton-induced X-ray emission (PIXE) and gamma-ray emission (PIGE) spectro~copy;'~~~~ and synchrotron X-ray fluorescence spectro~copy.'~-~~.The latter techniques although non- destructive and capable of achieving micrometre resolution have detection limits that are highly dependent on the shape and depth of the inclusion in the host material. In order to circumvent these problems significant research is now being directed to the use of laser ablation microprobes interfaced to inductively coupled plasma mass spectrometry ( ICP-MS),17 inductively coupled plasma atomic emission spectrometry ( ICP-AES)14,'8 and direct AES19320 instruments. Although wholly destructive with respect to the inclusion fluid the above techniques combine high element sensitivity high spatial reso- lution and relative simplicity of operation. This paper extends the initial work of Shepherd and Chenery17 and describes in detail the procedures developed for the calibration of laser ablation microprobe-inductively coupled plasma mass spectrometry (LAMP-ICP-MS) for the optimum analysis of single inclusions.One of the principal objectives was to determine if synthetic fluid inclusions in halite glass reference materials and aqueous solutions would be suitable as calibration standards for the determination of elemental ratios in natural fluid inclusions. EXPERIMENTAL Calibration Standards A prime requirement for the quantification of fluid inclusion analyses is the existence of calibration standards that can be analysed under the same or similar conditions as natural inclusions. Synthetic fluid inclusions are the obvious choice. Using quartz as the host matrix inclusions of diverse chemical composition can be synthesized under a wide range of exper- imental pressure and temperature conditions.21 However the procedures are time-consuming and for routine LAMP- ICP-MS analysis prohibitively wasteful of low blank high value calibration material.In order to evaluate the use of aqueous solutions or silicate glasses as alternative calibration standards comparative tests were carried out against synthetic inclusions in halite. Whilst not the ideal matrix for aqueous inclusion analysis the ease of synthesis and abundance of large inclusions in halite more than compensate for the disadvantage of Na and C1 interferences. All three forms of calibration Journal of Analytical Atomic Spectrometry March 1996 Vol. 1 I (1 77-1 85) 177material (synthetic fluid inclusions solutions and glasses) were used to derive elemental ratio working curves.Syntheticfluid inclusions in halite Following the methodology of Pironon,22 synthetic brine inclusions of different composition were prepared from high- purity reagent salts. Up to a maximum of 14 salts were used. For the system NaC1-LiCl-MgCl,-CaCl,-SrCl,-KCl (Mg Li Ca Sr K = 1000 pg g-I) the resultant inclusions were extremely abundant up to lo6 pm3 in volume and distributed in well-defined growth zones. For more complex brines (con- taining additionally Rb Ba Cs Pb Cu Zn Mn B= 100 pg g-') the cation chemistry dramatically influenced the growth kinetics of the halite crystals resulting in anhedral crystal clusters containing fewer smaller inclusions.Brine inclusions with various C1 Br ratios were also prepared C1 Br = 50 100,250 the solutions being saturated with respect to NaCl at room temperature (i.e. M 166000 pg g-' C1 and z 100000 pg g-' Na). Microwells containing aqueous solutions In order to assess the efficiency and performance of direct laser ablation of aqueous solutions use was made of microvolume amounts of aqueous solutions. This approach differs from that of Krishna et aLZ3 in not requiring a continuous flow through of solution. The solutions were pipetted into small 3mm diameter wells drilled into a 2mm thick perspex sheet the base and top being sealed with Sellotape or 100pm thick glass cover-slips using low melting-point wax or a silicone rubber adhe~ive.'~ Such microwells are easy to prepare simu- late very closely the ablation of macro-fluid inclusions and provide excellent matrix-matching of natural inclusions of varied chemical composition.Glass reference materials The third type of calibration standard consisted of the National Institute of Standards and Technology (NIST) Standard Reference Material (SRM) 61 1 Glass. This is a certified mater- ial and contains z 500 pg g-' of most elements except for Ca (85 700 pg g-') and Na (140OOO pg g-').25 This was used extensively to optimize the ICP-MS operating conditions described below and as a reference baseline for comparison with the synthetic fluid inclusions and microwells. By adopting a multiple calibration standard approach it was possible to assess variation in ablation efficiency due to material state (liquid versus solid) form of liquid containment (inclusions versus microwells) transfer losses between chamber and plasma torch and possible matrix effects.Instrumentation Fig. 1 shows a schematic layout of the LAMP-ICP-MS analy- sis system. Apart from minor modifications the configuration is as originally described by Shepherd and Chenery.17 UV laser Ablation was carried out using a frequency quadrupled Spectron Nd YAG pulsed laser operating in the far-UV region (266 nm) at 10 Hz. The optical delivery system consists of a high-quality Leitz microscope and Cassegrain x 25 or x 36 reflecting objective lens. Previously Geertsen et a1.26 have suggested that a UV laser-produced plasma is superior in every respect to that of an IR plasma for LAMP-ICP-MS studies.For an IR-induced plasma the plasma is opaque to the incident laser beam. This results in increased heating of INDUCTIVELY COUPLED MASS I PLASMA I MIXER I m Fig. 1 flow sample introduction system used for ICP-MS analysis Schematic diagram of the UV laser microprobe and dual gas the plasma indirect laser-matter interaction uncontrolled ablation selective vaporization of the material and the forma- tion of large ablation craters. By comparison a UV plasma is partially transparent to the laser beam. Laser-matter inter- action is thus more direct resulting in controlled and continu- ous ablation and the formation of craters which are defined by the diameter of the laser spot and not by the size of the plasma. Such mechanisms are still open to investigation and interpretation but for inclusion analysis the most significant advantage is that many inclusion-bearing minerals that are transparent in the IR region (e.g.quartz fluorite) absorb in the UV region. Ablation chamber For this study use was made of a new ablation chamber,27 which has an internal aerodynamic profile and allows smoother and faster transfer of ablated material to the torch. Before reaching the torch the argon flow from the ablation chamber is merged with argon carrier gas from the nebulizer/spray chamber in a glass mixer. Optimization of the different param- eters controlling the dual gas flow ~ystem'~.~*-~' is described below. ICP-MS The ablated material was analysed using a VG PlasmaQuad 2 + ICP-mass spectrometer in peak jumping acquisition mode where only responses at selected masses are analysed.Data processing utilized proprietory VG time-resolved acquisition software that permits identification and integration of the response at each selected mass over time.29*31 Calculation and quantification were performed using custom computer pro- grams. Analyte mass positions were selected to avoid known or suspected polyatomic or isobaric interferences (see Table 1). Optimization of Laser Ablation Conditions In order to optimize the laser ablation conditions for the analysis of inclusions in halite preliminary tests were carried out on inclusion-free halite. These demonstrated that to avoid excessive fracturing of the matrix and potential loss of inclusion fluid (prior to or during inclusion breakthrough) hole forma- tion should be initiated using minimum laser power.Once initiated the power could then be progressively increased to achieve optimum conditions for drilling a clean hole down to the level of the sub-surface inclusion (Fig. 2). These conditions vary from mineral to mineral depending on the position of the absorption edge with respect to the UV wavelength and the Vaporization temperature of the mineral phase. For synthetic inclusions the laser on entering the inclusion was fired for a further 2-10 s according to the size and volume of the inclusion; 178 Journal of Analytical Atomic Spectrometry March 1996 Vol. 11Table 1 analysis. Data taken from ref. 32 Isotope masses (u) and relative abundances used for ICP-MS Relative Isotope Mass/g mol-' abundance (YO) Li 7 92.58 B 11 80.22 Na 23 100 24 78.7 Mg c1 35 Ar 40 99.6 K 39 93.1 Ca 44 2.08 Mn 55 100 c u 63 69.09 Zn 64 48.89 cu 65 30.91 Zn 66 27.81 Br 79 50.54 Rb 85 72.15 Sr 88 82.56 cs 133 100 Ba 138 71.66 Pb 208 52.3 75.53 Fig.2 SEM photomicrograph of a typical laser ablation hole in halite the smaller the inclusion the faster the rate of fluid vaporization. Optimization of ICP Analytical Conditions Since argon flow rates water vapour content of the nebulizer flow and torch power have a marked influence on the tempera- ture and ionization characteristics of the plasma and formation of polyatomic interferences tests were carried out using NIST SRM 611 Glass to determine the optimum dual flow plasma conditions for the determination of different elements.The results expressed as the signal-to-background (S B) ratio are summarized in Fig. 3 and 4. ArgonJEow rates The effects of varying the ablation chamber and nebulizer flow rates are shown in Fig. 3. Wet conditions refer to the injection of a blank solution (1% HN03 solution) from the nebulizer whereas dry conditions refer to zero injection of solution. The best S B ratio under wet conditions corresponds to a nebulizer flow rate of 0.4 1 min-' and a corresponding ablation cell flow WET I DRY I I m 3i I . . 1 2 3 4 Nebuliser and cell argon flow rates Fig. 3 General variation in ICP-MS response [signal background ( S B)] for a range of elements as a function of ablation chamber (cell) and nebulizer (neb) argon flow. Types 1 to 3 refer to wet conditions (1 neb.0.65 min-' and cell 0.3 1 min-'; 2 neb. 0.4 1 min-' and cell 0.6 1 min-l; 3 neb. 0.25 1 min-' and cell 0.9 1 min-'). Type 4 corre- sponds to dry conditions (4 neb. 0.25 1 min-' and cell 0.9 1 min-') Li 1000 1100 1200 1300 1400 Ca Ba $ m v) 80 t 60 20 1000 1100 1200 1300 1400 Plasma power/W Fig.4 Effect of the ICP rf forward power on element intensities [signal backgrounds (S:B)] (a) LiE! and Na +; (b) Cam and Ba+ rate of 0.6 1 min-'. Comparison between wet and dry plasma conditions indicates that wet conditions favour Co Mn and the alkali and alkaline earth elements (except for Sr and Li) whereas dry conditions are more favourable for Cu and Zn. Allowing for a slight reduction in Cu and Zn intensities wet conditions were selected for subsequent calibration tests since they allowed for the introduction of aqueous standard solutions.Spray chamber temperature The spray chamber acts as a dynamic filter for solution droplets produced by the nebulizer. It retains the larger droplets that would otherwise fail to dissociate totally in the plasma and which would increase the noise level and hence the background signal. By varying the temperature of the water-cooled jacket around the spray chamber one can alter the water vapour content of the argon flow and size distribution of the dr0plets.3~ Journal of Analytical Atomic Spectrometry March 1996 Vol. 1 I 179In order to establish the optimum cooling conditions back- ground values were measured for a wide range of elements as a function of temperature from 2.5 to 193°C.Most of the alkali and alkaline earth elements (Be Na K Ca Rb Cs) have their lower background values at around 8.1 "C whilst others appear temperature-independent (Li Mg Sr Ba). The halogens C1 and Br are clearly temperature-sensitive and have their lower background values in the range 2.5-9.4 "C. The background is also lower for B Mn and Fe between 2.5 and 8.1 "C whereas Cu Zn Ce and U show no significant back- ground change with temperature. Thus for most elements 8.1 "C represents a good working compromise. Radiofrequency power to the ICP The radiofrequency (rf) forward power (watts) to the ICP source has been a well-known variable influencing elemental response and singly charged ion to oxide ratios. Gray and Date34 proposed that this was a function of a contraction of the plasma and central channel as power is increased.In order to assess the influence of forward power on those elements of interest S:B ratios were obtained for various power settings (1 100-1325 W). For the alkali and alkaline earth elements S B ratios were optimum at 1100 W (Fig. 4). No significant trend could be discerned for other elements and hence for this study the torch power was set at 1100 W. As demonstrated by these experiments no one set of con- ditions is optimum for all elements. Thus to obtain the maximum chemical information from a single fluid inclusion a compromise set of average instrumental conditions was adopted (Table 2). RESULTS AND DISCUSSION Element intensities were ratioed to Sr since it was not possible to determine quantitatively the concentration of each element in the inclusion fluid.Conventionally when presenting inclusion analyses elements are ratioed to Na the major cation in solution. However in view of the difficulties experienced in measuring Na Sr was selected instead. Sr is considered a suitable alternative because of its generally high concentration Table 2 Operating conditions of the laser ablation microprobe and of the ICP-MS instrument LAMP operating conditions- Laser Wavelength/nm 266 (frequency quadrupled from Maximum energy/mJ Spectron SL803 Nd YAG 1064 nm) 70 (amount of energy used depends on size of crater required) Mode Fully Q-switched TEMOO Laser repetition frequency/Hz 10 Microscope Leitz Aristomet Laser focusing objective Spectrometer VG PlasmaQuad 2+ Gas flow rate/l min-' Pulse length/ns 10 x 25 or x 36 ICP mass spectrometer operating conditions- Forward power/W 1100 Nebulizer 0.4 Cell 0.6 Coolant 13 Auxiliary 0.8 Data acquisition software VG TRA Data acquisition mode Peak jumping Points per peak 3 Dwell time per peak/ms 10 Time unit/s Typically 1-5 in geological fluids17 and in having a low ICP-MS background at m/z 88.Figures of Merit The accuracy of the alternative calibration strategies was evaluated by applying a t-test to the elemental ratios obtained from synthetic inclusions against those from aqueous solutions contained in microwells and glass reference materials (Table 3). The precision of the ratios expressed as the relative standard deviation (&) was evaluated both as within-run repeatability as shown for Li Mg Ca Ba and Br (Table 4) and between- run reproducibility for all elements (Table 5).35 Additionally correlation coefficients (R) were calculated as a test of the coherence of the relationship between elements.In order to display and compare the data for all three calibration materials glass concentrations were normalized to synthetic fluid inclusion and microwell concentrations. In the following sec- tions only the t-tests and between-run S and R values are discussed. NIST SRM 611 Glass The S values for most element Sr ratios range between 5 and 23%. Potassium gives very scattered values (S = 50%) suggest- ing that its concentration in the matrix (500 ppm) is close to the theoretical detection limit (Table 5). Element co-variation plots using raw count data are linear.For most elements R is greater than 0.98 (see Fig. 5) except for K. The corresponding R value for K (0.667) is low reflecting the poor detection limit. Despite having to use the minor Ca isotope 44Ca (2%) because of isobaric 40Ca-40Ar+ interference the Sr and R values are very good. The only problem relates to Na (Fig. 6). At high count rates the Na response exhibits severe non-linearity owing to detector saturation. (N.B. Owing to variations in relative sensitivity resulting from ionization efficiency and mass bias in the quadrupole mass spectrometer the reported signal ratios given in Tables 4 and 5 and Fig. 5 differ from the true mass ratios.) Synthetic Fluid Inclusions in Halite The same procedures were applied to the analysis of synthetic fluid inclusions in halite which range from 20 to 100 pm in diameter and were located up to 80pm beneath the surface (Tables 4 and 5).This is equivalent to a 100-fold variation in inclusion volume. Time-resolved profiles indicate excellent fluid release on laser breakthrough (Fig. 7). Bi-variate plots for Li and Mg against Sr [Fig. 8(a) and (b)] demonstrate very good linearity Table3 Validity of the null hypothesis that assumes mean ratio equivalence (element Sr). Statistical t-test verification; the significance level is indicated by P (i.e. probability)=0.05 (yes) and 0.02 (yes*) Element/Sr Li Mg Ca Rb c s Ba Mn 63cu T U 64Zn 66Zn Pb Br Fluid inclusions- microwells Yes Yes Yes Yes* Yes Yes Yes Yes* Yes Yes Yes Yes Yes ~ Fluid inclusions- glasses Yes Yes Yes* No Yes No Yes Yes Yes Yes Yes Yes Microwells- glasses Yes No No No Yes Yes * Yes No No Yes Yes No 180 Journal of Analytical Atomic Spectrometry March 1996 Vol.11Table 4 LAMP-ICP-MS analyses showing the within-run repeatability for several series of experiments for each type of material (fluid inclusions microwells and NIST SRM 611 Glass)" Elements No. of analyses Li Sr R Mg Sr s (Yo) sr (%) R Ca Sr R Ba Sr R Br Sr s (%) sr (Oh) Fluid inclusions F.I. 1 F.I. 2 F.I. 3 11 16 5 3.57 3.36 3.2 13 14.8 8.7 0.965 0.965 0.996 1.12 1.08 1.19 0.97 0.961 0.98 1.25 1.07 1.26 0.936 0.9 0.944 0.095 0.12 16 7 0.986 0.997 0.036 0.036 9.1 26.5 0.966 0.937 16.2 17.3 21.5 26.3 33.6 16.4 Micro wells NIST SRM 611 Glass Mw. 1 14 3.26 16.1 0.947 1.05 14.3 0.969 1.3 24.2 0.873 0.108 0.93 1 0.037 0.94 16.6 14.5 Mw.2 5 3.22 17.3 0.984 0.9 1 0.986 1.26 0.91 13.7 16.6 Mw. 3 12 3.5 10.1 0.937 1.06 0.9 18 1.31 0.907 10.9 17.1 Gls. 1 11 3.16 7.3 1.25 6.6 1.08 5.1 0.086 6.4 Gls. 2 7 3.24 11.1 0.995 1.02 16.5 0.988 1 8.6 0.992 0.102 8.2 0.987 Gls. 3 10 3.18 5.9 1.1 8.2 0.93 8.1 0.106 8.7 0.034 0.038 0.995 0.883 11.8 14.8 Mass ratio Li:Sr=l Mg Sr = 1 Ca:Sr=l Ba Sr = 0.1 Br Sr = 1.66 * Atomic ratios given in the table correspond to ICP-MS signal ratios and not to mass ratios as described by mass ratio column. over several orders of magnitude. The correlation coefficients are better than 0.96 and the S values of the ratios < 17% across the entire range. As a guide to the applicability to natural fluid inclusions Fig. 8(a) also shows the relationship between inclusion diameter and ICP-MS signal intensities.Some of the deviant points seen in Fig. 8(b) can be attributed to relatively flat inclusions which although large in cross- section had disproportionately low volumes. Other outlying points cannot be so readily explained as there is no apparent correlation with inclusion size or depth in the sample. The reduced precision noted for Ca (S = 28.8%; R = 0.92) is mainly due to the use of the minor isotope 44Ca and the lower concentration of Ca in the synthetic inclusions (1000 ppm). For these experiments 1OOOppm Ca is close to the limit of detection. The S values are poorest for K. Values are 329% and similar to those obtained for the NIST SRM 611 Glass. In both cases there is a high variable background at m/z 39 which is probably due to a combination of the 38Ar-H+ polyatomic interference and K contamination in the spec- trometer.It is estimated that the synthetic inclusions need to contain at least 2000ppm K to provide reliable working calibration graphs for K. For Rb Cs and Ba (100ppm of each) the correlation coefficients are very high (R > 98%). Corresponding S values are also good (Rb 11.05%; Ba 12.5%; Cs 20%). The S and R values for Mn Cu and Zn are slightly higher than for the alkali and alkaline earth elements (25-33% and 3 0.9 respectively). Boron measurements proved unsuccessful. A concentration of 100 ppm B in the synthetic inclusions appears to be close to the detection limit for the instrumental conditions selected. As expected the measurement of Na Sr ratios in an NaCl matrix proved difficult; the average S is > 57%.For a more complete assessment of the Na Sr data see below. For Br Ca Li and Mg a clearer comparison of their respective S values can be seen in Fig. 9 which shows the range in S values for a typical working day and the over- all poorer repeatability obtained for Ca. Data for the halogens are provisional but indicate tremen- dous potential for the direct measurement of Cl Br ratios by ICP-MS. The S and R values for Br are 14.7% and 0.96 respectively those for C1 being poorer ( ~ 3 0 % and 0.89). As with B conditions were not optimized for the determination of the halogens and further improvements in precision and detection limits might be possible. The main limitation is matrix contamination from the host mineral.A notable feature of Fig. 10 is the co-linearity for both 64Zn+ and 66Znf (corrected for isotopic abundance). Similar agree- ment is noted for 63Cu+-88Sr+ and 65C~f-88Sr+. No indi- cation was observed of 23Na40Ar + polyatomic interference on 63Cu+ as is normally the case for analysis of saline solutions by conventional nebulization. This suggests minimal spectral interference and the suitability of these m/z positions for the measurement of pg g-' concentrations of Cu and Zn in NaC1- rich inclusion fluids. Coherent element Sr ratios for inclusions of variable size testify to the precision and efficiency of laser ablation. Reproducible ratios between inclusions and between glass and inclusions show that sampling and removal of material for ICP-MS analysis is not significantly constrained by the depth/ diameter profile of the laser ablation crater.High spatial resolution laser ablation (drilling) to a depth of approximately 60 pm is straightforward. Below 60 ym however the hole can deviate unpredictably from the vertical and it becomes increas- ingly difficult to target 20 ym diameter inclusions. The advan- tages of high spatial resolution and accurate alignment of the laser beam are also relevant to larger inclusions (> 100 pm diameter). During the analysis of large inclusions air bubbles aggregate at the base of the entry hole which shield the liquid phase and give rise to highly erratic vaporization. In order to overcome this problem it is often necessary to re-position the laser and drill a second hole into the inclusion.Unlike the ablation of glass or microwells where the mass of material released is a function of laser power absolute detection limits for inclusions are directly proportional to the absolute mass of the analyte in the inclusion. Thus for a single population of inclusions of various sizes the concentration detection limits are approximately proportional to the cube of the inclusion diameter. Microwells Containing Aqueous Solution In order to facilitate comparison and evaluation of calibration procedures using glass and synthetic fluid inclusions a third approach to test material introduction was adopted namely the laser ablation of microvolume amounts of aqueous solution. A distinct advantage in using microwells is that it is relatively easy to prepare solutions of different composition and molarity especially standards with different C1 Br ratios.For all the elements except for Na (i.e. Ba Br Cs Cu K Li Mg Mn Pb Rb Sr Zn) the microwell data are in excellent agreement Journal of Analytical Atomic Spectrometry March 1996 Voi. I 1 181Table 5 LAMP-ICP-MS analyses showing the between-run reproducibility for each type of sample material* Sample Li Sr Fluid inclusions 3.26 0.972 1.11 0.96 1.16 0.925 1.66 0.743 14.4 17.3 28.8 29.7 212 57.6 0.64 0.89 0.038 0.96 0.108 0.986 0.07 1 0.992 0.064 11.1 0.996 0.1 1 20.4 0.932 0.062 0.904 0.059 0.895 0.02 1 27.9 0.917 0.021 0.878 0.0005 32.3 14.7 12.5 20.3 26.8 31 25.7 157.5 0.08 33.4 0.909 16.9 34.9 0.8 99 Micro wells 3.36 0.966 1.03 13.4 0.963 1.3 19.8 0.906 1.14 15.8 0.9 83 30.2 0.8 0.44 19.5 0.91 5 0.04 0.966 0.108 0.931 0.072 0.972 0.075 0.975 0.08 1 0.927 0.04 9.95 0.986 0.042 0.944 0.019 0.953 0.0 17 0.943 0.024 0.911 0.085 0.872 14.2 14.4 16.6 12.9 11.9 21.5 13.1 17.8 14.6 39.5 23.9 12.5 22.9 0.901 NIST SRM 611 Glass 3.2 7.7 0.995 1.15 0.988 1.01 9.3 0.992 2.58 0.667 12.7 50 117 18.5 0.914 0.096 0.987 0.072 0.978 0.105 0.993 0.105 0.875 0.046 0.979 0.05 18 0.984 0.019 15.3 0.99 0.01 8 0.95 0.025 0.976 0.08 1 0.99 11.6 12.1 22.7 22.7 20.8 17.2 23.8 13.7 Mass ratio Li Sr= 1 Mg:Sr=l Ca:Sr=l K:Sr=1 Na Sr = 100 Cl:Sr=166 Br Sr = 1.66 Ba Sr =0.1 Cs:Sr=O.l Rb Sr = 0.1 Pb:Sr=O.l 63Cu Sr = 0.1 "Cu Sr = 0.1 64Zn Sr = 0.1 "Zn Sr = 0.1 B Sr = 0.1 Mn Sr =0.1 C1 Br = 100 * Atomic ratios given in the table correspond to ICP-MS signal ratios and not to mass ratios as described by mass ratio column.with the glass and synthetic fluid inclusion data. Reproducibilities are generally better than those calculated for the synthetic inclusions owing to the larger masses of test material introduced into the ICP-MS system and for Na and C1 the absence of a solid matrix (Table 5). An unusual and unexplained phenomenon is the difference in slope between the Na-Sr calibration graph for the wells and that for the glass and synthetic inclusions (Fig. 11). For glass the matrix effect increases with laser power (Fig. 6) suggesting preferential vaporization of Na (i.e. chemical frac- tionation). Whilst the synthetic fluid inclusion graph appears superficially similar to that of the glass the enhanced Na signal at high laser power is probably due to contamination from ablation of the matrix.In Fig. 12 the Cl-Sr array for the microwells plots below that of the synthetic fluid inclusions. Since both fluids have identical C1 Sr ratios and allowing for the lower sensitivity of Cl with respect to Na the correlated high Na and C1 responses for the synthetic fluid inclusions would tend to support the matrix effect interpretation. Thus synthetic fluid inclusions in halite are not entirely appropriate for the calibration of Na and C1 in natural fluid inclusions. Further studies are in progress to address the NaCl matrix limitation. Overall microwells present few analytical problems. They confirm the calibration graphs obtained using glass and syn- thetic fluid inclusions and because of the larger solution volumes available for ablation provided data for K (S,= 15.8%; R = 0.90).Several experiments were performed to test the possible virtues of single laser pulse analysis of microwells. For all ratios the S values were >50% indicating the need for continuous pulsed mode ablation to achieve acceptable signal intensities. 182 Journal of Analytical Atomic Spectrometry March 1996 Vol. 11A A 0 2 4 6 8 10 Sr response (a.u.) Fig. 5 Li-Sr and Mg-Sr covariations for NIST SRM 611 Glass 3 1Na A A A Sr 0 2 4 6 8 10 al v) r 0 P v) CT Li I Sr _o L Na L Time (a-u.) Fig.7 Time-resolved ICP-MS spectra for Li Na Mg and Sr for a discrete pulse of fluid released from a 40 pm diameter aqueous inclusion in halite (depth beneath the surface 10 pm) Response ( a d Fig.6 Na-Sr covariation for NIST SRM 611 Glass Statistical Tests In Fig. 13 the data sets of Li versus Sr have been combined for all three calibration materials and procedures. Global correlation coefficients (2 0.9) are statistically very significant for all the elements except for Na. In order to allow rigorous comparison between the synthetic fluid inclusions microwells and NIST SRM 611 Glass a statistical t-test was carried out on the mean ratios for different combinations of the three calibration materials.35 This test measures the truth of a null hypothesis that there is 'no significant difference between the means of two different samples'. The results are given in Table 3 ('yes' and 'no' correspond to adoption and rejection of the null hypothesis respectively).Calculated t-values demon- strate excellent agreement between synthetic fluid inclusions and microwells for all elements. Except for Ba and Rb the NIST SRM 611 Glass also shows good agreement with the synthetic fluid inclusions. The differences in the mean values for Ba and Rb are probably due to a slight change in the ICP-MS responses for two different periods of analysis. However the tests prove conclusively that microwells may be used as calibration standards for the routine analysis of fluid inclusions and that allowing for the poorer results for Ba and Rb NIST SRM 611 Glass may be used accordingly. The degree of statistical agreement between all three calibration materials is shown graphically in Fig.14. This demonstrates that for Mg the data have similar absolute dispersions and approximate very closely to normal distributions the closest match being between the synthetic fluid inclusions and solution microwells. 1Li 70 pm 0 4 8 12 IbMg ' 1 A A A A A A A 0 5 10 1 5 Response (a.u.) Fig. 8 Li-Sr (a) and Mg-Sr (b) covariations for synthetic fluid inclusions in halite. Note in (a) the relationship between inclusion diameter and ICP-MS signal intensities Journal of Analytical Atomic Spectrometry March 1996 Vol. 11 183(Br/Sr)x30 RSD=9O/o 1.5 1 A 1 A A $lA A A A Sr I ' . . . 0 1 2 3 4 Response (a.u.) M&r RSD=16% Fig. 12 Comparison between Cl-Sr covariations for synthetic fluid inclusions and microwells Fig. 9 repeatability of these elements (ratioed to Sr) Spider diagram for Br Ca Li and Mg showing the within-day lLi d 3 W 2 4 a izn A 0 1 2 3 4 Response (a.u.1 0 0 0.5 1 1.5 2 Response (a.u.) Fig.13 Combined element covariation for all three types of cali- bration material (fluid inclusions microwells and NIST SRM 611 Glass) for Li-Sr Fig. 10 Comparison between 64Zn and 66Zn signal intensities. Good agreement indicates absence of significant polyatomic interference. A @Zn Fluid inclusion; A 64Zn fluid inclusion; 64Zn microwells; 0 66Zn microwells; + 64Zn glass; 0 glass SYNTHETIC F.I. MICROWELLS n 4 0 A I I I 1 Sr 0 0 1 2 3 4 Response (a.u.1 0 0.5 1 1.5 2 Mg/Sr Fig. 11 Comparison between Na-Sr covariations for synthetic fluid inclusions microwells and NIST SRM 611 Glass. A Fluid inclusions; A glass; 0 microwells Fig.14 Histogram of measured Mg Sr ratios showing the degree of agreement between all three calibration materials CONCLUSIONS For most purposes microwells containing aqueous solutions and NIST glass reference materials can replace synthetic fluid inclusions as working calibration standards for the elemental analysis of single fluid inclusions. NIST reference glasses are readily available chemically stable and can be used almost indefinitely. Theoretically microwells have less long-term stab- ility (not tested). Their principal advantage however is that they approximate much more closely to the laser ablation of fluid inclusions and can be designed to cover a wider range of chemical compositions and elemental concentrations than MIST glasses. They are also easy to prepare.Synthetic fluid inclusions in halite simulate even more closely the laser response of natural fluid inclusions and for the study of evaporate minerals are the perfect match. Their principal disadvantage is that they cannot be readily used for the calibration of Na and C1 because of host matrix contamination. Techniques for minimizing this problem for the direct measure- ment of C1 Br ratios in inclusions in halite are currently being investigated but for the moment synthetic fluid inclusions in halite are more limited as calibration standards than either 184 Journal of Analytical Atomic Spectrometry March 1996 Vol. 1 1microwells or NIST glasses. Nevertheless all three types of calibration standard yield similar element Sr ratio S values and excellent elemental ratio working curves.Average S values for Li Na K Rb Cs Mg Ca Sr Ba Mn Cu Zn Pb C1 and Br range between 7 and 32% including data for those elements that are close to the detection limit or include a component of matrix contamination. Until the absolute mass of analyte in the inclusion can be determined analyses are best reported as element ratios. However reference to microthermometric data should allow realistic approximation of inclusion fluid concen- trations for geochemical modelling. It is concluded that no one set of instrumental ICP-MS conditions is optimum for all elements and that analysis of more than one inclusion is needed to derive the maximum amount of chemical information from trapped fluid. Nevertheless further studies are in progress to improve the efficiency of fluid release and to optimize the analysis for groups of related elements (e.g.the halogens). A. M. thanks J. Pironon for his help during the synthesis of fluid inclusions. This study was financed in part by the European Union Human Capital and Mobility Programme. Publication is by permission of the Directors of the British Geological Survey (NERC) UK and the Centre de Recherches sur la Geologie des Matikres Premikres Minerales et Energetiques (CREGU) France. REFERENCES 1 2 3 4 5 6 7 8 9 10 11 12 13 Boiron M. C. and Dubessy J. in Fluid inclusions and minerals method and applications eds. de Vivo B. and Frezzotti M. L. Pontiniano Sienna 1994 pp. 45-71. Shepherd T. J. Econ. Geol. 1981 76 1244. Roedder E. in Fluid Inclusions ed. Ribbe P.H. Mineralogical Society of America 1984 p. 12. Bodnar R. J. Mineral. Mag. 1990 54 295. Kihle J. Org. Geochem. 1993 784. Dubessy J. Boiron M. C. Moissette A. Monnin C. and Sretenskaya N. Eur. J. Mineral. 1992 4 885. Dubessy J. Poty B. and Ramboz C. Eur. J. Mineral. 1989 1 517. McMillan P. F. M. Dubessy J. and Hemley R. in Micro- Raman Spectroscopy ed. Corset J. Elsevier Amsterdam 1994 in the press. Pironon J. and Barres O. Geochim. Cosmochim. Acta 1990 54 509. Pironon J. and Barres O. Geochim. Cosmochim. Acta 1992 56 169. Ayora C. Garcia-Vergas J. and Pueyo J. J. Geochim. Cosmochim. Acta 1994 54 811. Anderson A. Clark A. H. Max P. Palmer G. R. Macarthur J. D. and Roedder E. Econ. Geol. 1989 84 924. Heinrich C. A. and Ryan C. G. in Water Rock Interaction eds.14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 Kharaka Y. K. and Maest A. S. Balkema Rotterdam 1992 Rankin A. H. Ramsey M. H. Coles B. van Langevelde F. and Thomas C. R. Geochim. Cosmochim. Acta 1992 56 67. Mavrogenes J. A. Bodnar R. J. Anderson A. J. Bajt S. Sutton S. R. and Rivers M. L. in Proceedings of the PACROFI V Conference Cuernauaca (Mexico) 1994 eds. Izquierdo G. Suarez M. Guevara M. Vanko D. and Viggiano J. C. Philippot P. Chevalier P. Gibert F. and Legrand F. in Terra Abstr. 9 (Abstract supplement No. 1 to Terra Nova 1995,7) 344. Shepherd T. J. and Chenery S. R. Geochim. Cosmochim. Acta 1995 in the press. Wilkinson J. J. Rankin A. H. Mulshaw S. C. Nolan J. and Ramsey M. H. Geochim. Cosmochim. Acta 1994 58 1133. Boiron M.C. Dubessy J. Andre N. Briand A. Lacour J. L. Mauchien P. and Mermet J. M. Geochim. Cosmochim. Acta 1991 55 917. Boiron M. C. Dubessy J. Briand A. Mauchien P. and Alle P. in Proceedings of the PACROFI I V Conference Lake Arrowhead 1992 eds. MacKibben M. A. Montanez I. P. and Hall D. L. p. 17. Bodnar R. J. and Sterner S. M. in Hydrothermal Experimental Techniques eds. Barnes H. L. and Ulmer G. C. Wiley New York 1987 pp. 423-451. Pironon J. Am. Mineral. 1990 75 226. Krishna R. Vijayalakshmi S. Mahalingam T. R. Viswanathan K. S. and Mathews C. K. J. Anal. A-t. Spectrom. 1993 8 565. Shepherd T. J. Fluid Inclusion Laser Ablation ICPMS Analysis. I. Novel Liquid Calibration Standards British Geological Survey short report MPSR/95/38 Nottingham 1995. NIST SRM 611 Glass Certificate of Analysis (1982) from NIST (previously National Bureau of Standards) Washington DC Standard Reference Materials 61 1 Glass 1 mm thick wafers. Nominal trace element concentration 500 ppm. Geertsen C. Briand A. Chartier F. Lacour J. L. Mauchien P. Sjostrom S. and Mermet J. M. J. Anal. At. Spectrom. 1994,9 17. Chenery S. R. Poitrasson F. and Cook J. M. unpublished work. Moenke-Blankenburg L. Schumann T. Gunther D. Kuss H. M. and Paul M. J. Anal. At. Spectrom. 1992 7 251. Chenery S. R. and Cook J. M. J. Anal. At. Spectrom. 1993,8,299. Querol X. and Chenery S. R. in European Coal Geology eds. Whateley M. K. G. and Spears D. A. Geological Society Special Publication London 1995 vol. 82 pp. 147-155. Abell I. D. in Applications of Plasma Source Mass Spectrometry eds. Holland G. and Eaton A. N. Royal Society of Chemistry Cambridge 1991 pp. 209-2 17. CRC Handbook of Chemistry and Physics CRC Press Boca Raton FL USA 60th edn. 1979-80 B236-320. Jarvis I. and Jarvis K. E. Chem. Geol. 1992 95 1. Gray A. L. and Date A. R. Analyst 1983 108 1033. Miller J. C. and Miller J. N. in Statistics for Analytical Chemistry eds. Chalmers R. A. and Masson M. Ellis Horwood Series in Analytical Chemistry Ellis Horwood Chichester 1988. VO~. 2 pp. 1583-1588. pp. 58-59. Paper 51055 221 Received August 21 1995 Accepted November 10 1995 Journal of Analytical Atomic Spectrometry March 1996 Vol. 11 185