Determination of ions in individual fluid inclusions by laser ablation optical emission spectroscopy: development and applications to natural fluid inclusions Ce�cile Fabre,*a Marie-Christine Boiron,a Jean Dubessya and Alain Moissetteb aCREGU–UMR G2R 7566, BP 23, 54501, Vandoeuvre-le`s-Nancy Cedex, France. E-mail: cecile.fabre@g2r.u-nancy.fr; Fax:+33 (0) 3 83 91 38 01 bLASIR, UER de Chimie C8, USTL , 59655 Villeneuve d’Ascq and CREGU–UMR 7566, BP 23, 54501 Vandoeuvre-le`s-Nancy Cedex, France Received 30th November 1998, Accepted 1st April 1999 Chemical data on the composition of individual fluid inclusions are required to model paleofluid–rock interactions, as these inclusions are the direct remains of the former fluid circulation in the earth’s crust.The proposed technique of laser ablation coupled with optical emission spectrometry (LA-OES) determines ion ratios in individual fluid inclusions. The radiations of the elements present in the plasma are analysed with a spectrometer equipped with a pulsed and gated multichannel detector.The lateral resolution is around 6 mm and the depth of the quartz crater obtained for one laser shot is around 1 mm. For fluid inclusions, the laser beam is initially focused on the surface of the sample, to drill until the inclusion is reached, then the liquid is analysed. Plasma temperature studies showed that various kinds of standards (synthetic glasses, fluid inclusions and minerals) can be used for the establishment of calibration curves for Na/K, Na/Ca and Na/Li ratios.As the emission line intensity is a function of the ablated mass, only emission line ratios between two elements are used. In addition, a self-absorption process was considered for the calibration strategy. The relative standard deviations of the calibration curves vary from 5% (glasses) to 25% (fluid inclusions) and the detection limits are those required for the determination of ions in individual inclusions (Na and Li 10, Ca 20 and K 750 ppm).Natural fluid inclusions were analysed using the LA-OES technique and it is now possible to determine the ratios of major elements in individual fluid inclusions. This paper reports recent developments using LA-OES, Introduction previously described by Boiron and co-workers14,15 and Quantitative chemical analyses of fluid inclusions are a Moissette et al.16 and applied to individual fluid inclusions. It prerequisite for understanding and modelling fluid–rock inter- is now possible to determine the ratios of major elements.This actions as only these inclusions contain direct evidence of the work consisted in establishing the analytical procedure and composition of fluids responsible for past geological processes. the calibration curves. First applications to major ion ratios Fluid inclusions with diVerent ages and compositions (size, to natural fluid inclusions are also described. abundance and diversity) within a single crystal complicate the interpretation of the bulk chemical analysis of fluids Principle and experimental procedure extracted using crush–leach techniques.The determination of ion contents in individual fluid inclusions is a new challenge Principle to achieve this aim. During the last 10 years, various techniques have been The eVects of absorption of laser radiation on a material are developed in an attempt to determine ion ratios in individual heating, melting, vaporisation, atomisation, excitation and inclusions. Some modern analytical methods are based on the ionisation, depending on the energy deposited on the sample.analysis of electromagnetic radiation produced by excitation Lines emitted from excited atoms and ions are used in laser with accelerated particles or electrons, e.g., X-ray microanal- microanalysis based on plasma emission spectroscopy.17–19 ysis of frozen inclusions,1 X-ray emission (PIXE) or gamma- The incident laser radiation generates plasma of high temperaray emission (PIGE) spectroscopy2–4 and synchrotron X-ray ture containing atoms and ions in an excited state.When fluorescence spectroscopy.5–8 PIXE, PIGE and SXRF atoms and ions transfer from an excited electronic level to a methods, although non-destructive, are ‘heavy’ methods and lower electronic level, a radiatial transition may occur with the analytical results depend on the shape and depth of the emission of photons.inclusions in the host mineral. Other methods under develop- The intensity of an emission line during such electronic ment are destructive at the scale of a focused laser beam and transition can be written as: are based either on ICP-MS9–13 or optical emission spectroscopy (OES).14–16 Both techniques are coupled with laser Iij= VBcNgiAij 4plijZ(T ) exp A-Ei kT B (1) ablation (LA) at the microscopic scale (<10 mm) to extract or to reach the liquid to be analysed within the inclusions. where I is the intensity of the emission line corresponding to LA-ICP-MS9–13 seems more appropriate for the determination the transition i�j, i and j are the indices of the higher and of trace elements, although LA-OES14–16 gives better results lower energetic quantum state, respectively, V is the collection for major elements, especially Na and Ca (better detection limits).solid angle for the plasma emission, B is Planck constant, c is J. Anal. At. Spectrom., 1999, 14, 913–922 913Fig. 1 Schematic diagram of the experiment set-up (built thanks to EU project program MAT1-CT-93–002924) showing the Nd5YAG laser coupled with an optical emission spectrometer. M, mirror; PM, parabolic mirror; L, lens; D, diaphragm; V-C, video camera; P-D, pulsed detector, X-Y, microscope stage; C-O, Cassegrain objective. the speed of light, N is the number of free atoms of the studied of the elements present in the plasma are analysed directly with a spectrometer (40 mm working distance) (Dilor) element in the plasma, gi is the statistical weight of the quantum state i, Aij is the transition probability for spon- equipped with a pulsed and gated multichannel detector. An adjustable time delay (between 50 and 1000 ns) is used to taneous emission from i to j, lij is the wavelength of the emission line, Z(T ) is the partition function of the quantum remove the continuous emission occurring in the first tens of nanoseconds after the laser shot and to optimise the signal- state, T is the electronic excitation temperature (K) and Ei is the energy level of excited electronic level i.to-background ratio. The best delay is 150 ns for a temporal window of 500 ns. The monochromator has a 280 nm focal length (Spex-280; Spex, Edison, NJ, USA). It is equipped with Instrumentation two gratings with 300 grooves mm-1 blazed at 250 and 600 nm An Nd5YAG laser (Continuum, Minilite, Evry, France) (Jobin-Yvon, Longjumeau, France).A 200 nm spectral range coupled with an optical emission spectrometer (Fig. 1) delivers is simultaneously recorded, utilising the 300 grooves mm-1 a laser pulse (5 ns). The operating conditions and the charac- holographic grating. The whole system was assembling by teristics of the instrument are summarised in Table 1. The laser Dilor. The spectral resolution of the spectrograph is 1.2 nm. beam is focused on to the sample through a Cassegrain-type microscope objective (Dilor, Lille, France). Plasma is created Spatial resolution and ablated mass during the laser–matter interaction.The typical emission lines Lateral resolution was measured for diVerent types of mineral (quartz, halite, sylvite, calcite, fluorite) and metallic matrices. Table 1 LA-OES operating conditions for solids and individual fluid inclusions The lateral resolution (Fig. 2) is around 6 mm for each type of material even if the depth of the crater hole obtained for Laser system — one laser shot can be very diVerent, e.g., from 1 mm for quartz Laser mode Q-switched up to 6 mm for copper.Hence the ablated mass is diVerent for Laser type Nd5YAG (Minilite, Continuum) each matrix and depends on various physical parameters of Energy per pulse Maximum output 2.8 mJ (266 nm) the solid (thermodynamic parameters, optical and transport Wavelength 266 nm (quadrupled frequency) Pulse width 5 ns properties).17,18,20 Thus the type of crater depends on the Acquisition time 500 ns ablated mass and consequently on the heat diVusion in the Crater size 6–10 mm in quartz for one shot matrix, the melting temperature of the material and its Spectroscopic system — crystalline state. Spectrograph Spex 270 M Wavelength 200–800 nm Spectral range 200 nm Experimental procedure Slit width 50 mm Spectral resolution 1.2 nm The laser beam is initially focused on the surface of the sample Grating 300 grooves mm-1 progressively ablated until the inclusion is reached.The pro- Gas Argon gress of the ablation is monitored using the matrix lines, Si Flow rate 0.8 l min-1 for a quartz matrix. Once the inclusion is opened, the Si 914 J. Anal. At. Spectrom., 1999, 14, 913–922Fig. 2 Microphotographs of ablation craters made on (A) metallic copper (10 shots) and (B) quartz (100 shots). emission lines decrease and the characteristic lines of the ions Fig. 3 Optical emission spectra obtained by LA-OES on synthetic fluid inclusion (one laser shot) for the determination of the three present in the liquid phase appear.calibration curves, using the emission lines of sodium (588.9 nm), It is worth noting that at least 10 laser shots are required lithium (670.7 nm), potassium (766.5 nm) and calcium (787.2 nm). to empty an inclusion of 20×20 mm. With larger inclusions, 30–50 shots can readily be made. This eVect is due to the short laser pulse duration (5 ns) and to the supersonic ejection eVect of plasma temperature in order to estimate its influence of plasma (10 mm/10 ns) which prevents heat transfer to on the calibration curves.The two emission lines and their non-ablated liquid and so its vaporisation after one laser shot. electronic transitions22–24 are: 515.32 nm (3d104p1�3d104d1) The concentrations of the ions in the liquid diVer widely; and 510.55 nm (3d94s2�3d104p1) (Fig. 3). sodium is often the main cation. Thus, filters are used to Experiments were carried out under an argon flow on pure reduce the intensity of the Na emission doublet to prevent metallic copper, silicate glass (Si, K, Cu, Na) containing 5% saturation of the intensified photodiode array.Two types of m/m of copper, copper solution (6000 ppm) and copper-rich filters have been used: (i) the first filter cuts the emission lines synthetic fluid inclusions (5000 ppm) to estimate the plasma with wavelength lower than 400 nm (Ca) and reduces the temperature.This temperature is an average temperature over signal intensity of sodium (around 590 nm) to 40%; (ii) the time and space. Temperature is time averaged because the second filter permits the recording of 6% Na intensity, 14% signal is integrated over 500 ns (T decreases slightly with for Li and 92% for K and Ca. time). Temperature is also space averaged because all the Sweeping the sample with an argon flow has been proved plasma is imaged on the entrance slit of the spectrometer and to be very eYcient in enhancing the plasma temperature and on the detector. then the line intensities.21 Thus, a cell was specially adapted A study of the depth eVect was carried out on copper metal.to work under a constant argon atmosphere. The argon flux The results show that the average plasma temperature does was optimised in order to obtain the best reproducibility of not change until a depth of the laser focus of around 80 mm. the signal. It has been demonstrated that low (<0.5 l min-1) Therefore, the fact that the plasma originating from fluid or high (>1.5 l min-1) argon fluxes are not adequate for a inclusions located at depths less than 80 mm is expelled through correct and repeatable signal intensity.The optimised value of a crater of a few tens of micrometres will not have any the argon flux was estimated to be around 0.8 l min-1. influence on the average plasma temperature. Only the first shot creates a plasma with a temperature (9300 K) lower than Calibration standards those following and the temperature RSD calculated on 30 shots performed in the same crater is around 2.4%.Temperature of plasma The calculated plasma temperatures are comparable for the For a constant ablated mass of an element, the excitation diVerent materials and around 11 000±500 K (solution, temperature of the plasma controls the intensity of emission 11 500±320 K; glasses, 10 400±540 K). The small variation lines.19 Consequently, the emission line ratio of two distinct of temperature that can occur between glasses and aqueous elements is a function of plasma temperature.solutions is insuYcient to achieve a variation of a two-element As shown by eqn. (1), the intensity of the optical emission emission lines ratio for the diVerent temperature conditions line is a function of excitation temperature. This temperature [as shown in eqn. (1), the ratio of two emission lines is a is of great importance and can be calculated using the following function of temperature].Using Student’s t-test, it can be equation (considering that the plasma is in thermodynamic demonstrated that the temperature calculated for the metallic equilibrium): copper (11 500±230 K) is the same than those found for the solution. These results show that the calibration of atomic emission kT= E1-E2 lnAg1A1l1 g2A2l2B-lnAI1 I2B (2) lines of the diVerent elements versus their concentration can be performed using synthetic glasses and applied for fluid inclusion analysis.where k is Boltzmann’s constant, T is the plasma temperature (K), E1 and E2 are the excitation energy levels for the two Standards considered emission lines, g1 and g2 are degeneration degrees of the higher level, A1 and A2 are transition probabilities and The determination of the ion content in fluid inclusions requires the calibration of the method with standards that l1 and l2 are the wavelengths of the emission lines.Using eqn. (2) and two Cu emissions lines, it is possible to study the could be analysed under the same conditions as the natural J. Anal. At. Spectrom., 1999, 14, 913–922 915Table 4 Compositions (molal ) of synthetic fluid inclusions in quartz samples. Standards consist of synthetic glasses, minerals and used for the establishment of the calibration curves synthetic fluid inclusions in quartz (Tables 2–4). Synthetic glasses were prepared from carbonates (CaCO3, Synthetic fluid inclusion in quartz Na2CO3, K2CO3) and oxide (Li2O) in an SiO2 matrix (Table 2).After decarbonation, the powders were heated twice Sample Na K Li Ca in platinum crucibles in an oven at around 1300 °C for 10 min. FI 1 0.05 0.025 0.006 0.05 Between the two heatings, the glasses were crushed to homog- FI 2 0.1 0.01 0.005 0.05 enise the standard. The compositions of the synthetic glasses FI 3 0.1 0.2 0.02 0.02 FI 4 1 — — 1 FI 5 0.5 — — 0.5 Table 2 Compositions (in % m/m) of the 30 synthetic glasses used for FI 6 1 — 0.1 — the establishment of the calibration curves FI 7 0.5 — 0.5a — FI 8 2 0.2 — — Sample SiO2 Al2O3 Na2O K2O CaO Li2 O FI 9 0.66 0.33 — — FI 10 2 2 — — SG1 71.25 11.15 5.68 3.82 — — FI 11 0.08 0.02 0.005 0.014 SG2 71.99 11.10 5.69 4.03 — — FI 12 0.10 0.02 0.008 0.03 SG3 69.59 11.28 6.26 3.80 — — FI 13 0.10 — — 1 SG4 65.35 10.81 8.72 3.18 — — FI 14 1 — 0.10 — SG5 65.69 11.30 9.50 3.20 — — aCorresponds to the concentrations of lithium that could not be used SG6 76.96 — 20.47 4.07 — — for the Na/Li calibration curve owing to the self-absorption of the Li SG7 71.93 10.89 4.70 4.05 — — emission line at 670.7 nm. SG8 66.28 — 12.67 7.28 — — SG9 71.04 — 18.51 4.59 3.89 — SG10 72.47 — 15.52 5.29 4.88 — SG11 62.64 17.44 1.19 10.87 4.68 4.13b were then analysed by electron microprobe analysis to check SG12 58.05 15.73 10.06 — — 18.54b the element ratios.SG13 74.33 — 20.78 2.80 3.06 — In addition, homogeneous Li minerals (an Li mica and two SG14 69.64 &md 18.17 3.28 — — SG15 62.00 16.34 11.37 3.49 4.17a — muscovites; Table 3) of known compositions (analysed by SG16 62.20 16.07 11.80 5.53 2.65a — ICP-AES) were used to calibrate the Na/Li and Na/K ratios. SG17 59.60 15.40 11.75 3.99 6.21a — These minerals were chosen for their homogeneity in lithium SG18 73.98 — 15.68 5.08 3.11 — and major elements.SG19 75.97 — 12.75 4.98 4.34 — Synthetic fluid inclusions in quartz (Table 4) were prepared SG20 75.68 — 9.74 4.97 7.59 — by hydrothermal synthesis following the methodology of pre- SG21 78.67 — 8.05 8.29 2.24 — SG22 71.12 — 15.20 4.24 7.82 — vious workers.25 Synthetic quartz devoid of fluid inclusions SG23 78.47 — 12.25 4.92 2.28 — was used as the raw material.After fracturing, prisms of SG24 69.60 10.65 5.52 3.93 — — quartz were loaded into gold tubes with the standard solutions. SG25 70.66 10.96 5.90 4.13 — — The capsules were heated in autoclaves at 650 °C and 250 MPa SG26 72.94 15.57 4.56 4.14 0.57a 0.08 for 1 week.A series of synthetic fluid inclusions were kindly SG27 50.00 14.00 5.50 0.20 7.20a 0.17 provided by K. Schmulovich, who performed the synthesis at SG28 81.98 7.42 0.15 — 0.07a 13.16b SG29 50.00 14.00 5.00 0.20 7.20a 2.09 800 °C and 900 MPa for several days. The compositions of the SG30 68.19 — 16.36 4.85 6.97 — solutions before and after experiments were checked by atomic aCorresponds to concentrations where emission line of calcium could absorption spectrometry.Pieces of the synthetic quartz with not be used for the Na/Ca calibration curve owing to the interference inclusions were analysed by the crush–leach technique to with an emission line of aluminium. bCorresponds to the concen- validate the compositions of the trapped fluids. In addition, trations of lithium which could not be used for the Na/Li calibration microthermometry was performed on the synthetic inclusions curve owing to the self-absorption of the Li emission line at 670.7 nm.to check their salinity using a Chaix–Meca stage.26 Table 3 Compositions (in % m/m) of minerals used for the establishment of the calibration curves Component M1 (Li mica) M2 (muscovite) M3 (muscovite) SiO2 50.3 45.73 73.04 Al2O3 25.74 34.89 15.32 FeO (total ) 0 2.45 0.74 Fe2O3 (total ) — — 0.17 MnO 0.4 0.02 <0.01 MgO 0 0.48 0.2 CaO 0 0 0.85 Na2O 0.32 0.48 3.85 K2O 9.94 10 4.96 TiO2 0 0.05 0.13 P2O5 nd nd 0.14 B2O nd nd 0.005 F 7.36 0.19 0.086 H2O (total ) nd nd nd Rb2O nd — nd Li2O 11.24a 0.23 0.073 Sum 97.94 94.33 99.48 Formula [SiO3AlO10|F2]KLi1.5Al1.5 [SiO3AlO10|(OH)2]KAl2 [SiO3AlO10|(OH)2]KAl2 aCorresponds to the concentrations of lithium which could not be used for the Na/Li calibration curve owing to the self-absorption of the Li emission line at 670.7 nm. 916 J. Anal. At. Spectrom., 1999, 14, 913–922dimensions of the crater is around 2×10-9 g per shot. Ablation Calibration curves of a fluid inclusion of 10×10×10 mm requires five shots to This study was focused on the 580–780 nm spectral range, empty all the inclusion, so the calculated ablated mass (only covering the main lines of major elements present in inclusions liquid ) is around 0.5×10-9 g.This value is based on the (Na, K, Ca, Mg, Li, etc.). Na (588.9 and 589.5 nm non- hypothesis that only liquid is ablated. In the case of sodium, resolved by our spectrometer; for the two electronic transitions self-absorption starts from concentrations above 7000 ppm 3p�4d and 3s�3p; resonance transitions), Li (670.7 nm; (Fig. 4), which represents >0.05 mol kg-1 for aqueous solu- 1s22s�1s22p; resonance transition), K [766.5 nm; tions in fluid inclusions. The concentrations of the diVerent 3p6(1s)4s�3p6(1s) 4p; resonance transition] and Ca standards (Tables 2–4) and those found in the natural fluid [786.6 nm, second order of 393.3 nm; 3p6(1s) 4s�3p6(1s) 4p] inclusions are always above this calculated value.Hence for of ionic calcium (Ca II ) emission lines (Fig. 3) were used to fluid inclusion analyses, the self-absorption process is expected establish the calibration curves for the diVerent ion ratios to occur for Na. It is worth noting that the calibration curve (Na/Li, Na/Ca and Na/K) using synthetic glasses, minerals of the emission line of the Na doublet seems to show linear and synthetic fluid inclusions in quartz (Tables 2–4).behaviour with respect to concentration in the range of concentrations where the self-absorption process occurs. Self-absorption Concerning lithium, the shape of the calibration curve is similar to that of sodium and consists of two branches It must be noted that the three emission lines selected for above and below around 3000 ppm in glasses. This threshold sodium, lithium and potassium are the most intense but are for self-absorption process corresponds to 0.12 mol kg-1 H2O also resonance lines.27 It is necessary to determine the consefor fluid inclusions.The concentration of lithium is usually quence of the resonance eVect on the calibration curves. The below this limit, as shown from the analysis of geothermal major problem which can occur with the resonance lines is the fluids.29 It is worth noting that Li-rich fluids can be easily self-absorption process.28 Self-absorption corresponds to the identified from low eutectic temperatures determined by micro- reabsorption of the emitted photons by atoms at the fundathermometric data.30–33 As for Na, the calibration curve mental electronic level, hence the emission intensity is reduced remains linear with concentration (Fig. 5) when the self- (and no longer proportional to the measured concentration). absorption process occurs. For the usual fluid inclusions, the Consequently, during the self-absorption process, the intensity self-absorption process is expected to occur for Na and not of the emission increases linearly with the concentration of for Li.the element in the matrix and the probability of the electronic For calcium (Fig. 6), the intensity data do not show any transition. Hence it is necessary to quantify the element self-absorption: the calibration curve displays a single straight concentration threshold from which self-absorption is a limitline passing through the origin. This result is in complete ing process before any analytical application. The net intensity agreement with the fact that the emission line of calcium is of the diVerent emission lines (used for the establishment of not a resonance line, but a II line coupled with the the calibration curves) was measured versus the element confundamental state.centration in glasses. Self-absorption can be demonstrated The calibration curve for potassium shows a regression line when the intensity of an emission line versus concentration is passing near the origin (Fig. 7). For potassium, the experimen- not a linear function and when the slope decreases with tal data are more diYcult to interpret because the photoca- increase in concentration. In addition, the part of the calithode of the intensifier has a low quantum eYciency at 766 nm. bration curves established for high concentrations does not It is six times less sensitive than the emission intensity for Na pass through the origin. This kind of variation is illustrated at 589 nm.This can explain the fact that the self-absorption in Fig. 4 for the self-absorption of the Na doublet at 589 nm. eVect for potassium cannot be identified even if it really occurs The estimated ablated mass in glasses determined from the Fig. 4 Calibration of the emission line of sodium at 589 nm, to Fig. 5 Calibration of the emission line of lithium at 670.7 nm. An demonstrate the eVect of self-absorption for this line and the limit of this eVect. The arrow indicates the threshold of the self-absorption expansion for values <800 ppm is also represented.The arrow indicates the threshold of the self-absorption eVect. eVect. An expansion for values <6000 ppm is also shown. J. Anal. At. Spectrom., 1999, 14, 913–922 917Before each study, a test is aled out under the same conditions: use of the same optical filter, no argon flow and the record of the spectrum obtained at the second laser shot. This allows the identification of some variations in the focusing of the laser beam or a change in the optical coupling between the microscope and the spectrometer.In addition, this intensity measurement is used as an intensity reference for any comparison of calibration curves. However, in the case of fluid inclusions the repeatability is very poor because the absolute intensity of the emission lines depends on the ablated mass, which can change from shot to shot. Indeed, the focusing of the laser beam on the liquid phase is diYcult to optimise inside the fluid inclusion.Hence the absolute intensity of an emission line of a given element from a fluid inclusion cannot be calibrated versus its concentration (or with respect to water). In addition, no oxygen or hydrogen emission lines are detected which could be ratioed with the intensity of an element and linked with a direct calibration on the molality scale. Therefore, only intensity line ratios of dissolved elements are relevant and standardisation is made with respect to sodium.It can be noted that RSDs found for emission lines ratios are lower in synthetic glasses (between 5 and 15% depending on the concentration of the elements) owing to the good Fig. 6 Calibration of the second order of the emission line of calcium reproducibility of the emission line intensity from shot to shot at 393.3 nm, showing no self-absorption eVect. and the high intensity peaks. The repeatability of ratios even in fluid inclusions is often close to that obtained with glass samples (15–20% against 10–15%), which is acceptable.Reproducibility and repeatability limitations are strongly associated with signal-tobackground (S/B) values. It was demonstrated on glasses that an emission line with S/B <5 leads systematically to a repeatability worse than 20%. This can explain the high RSD for the intensity ratio when one of the two elements has a weak emission line, either due to its low concentration or because the sensitivity of the detector is poor, especially for the resonance line of potassium at 765 nm.Detection limits have been calculated for major elements in fluid inclusions, considering an optimum signal-to-background ratio of 5. The detection limits are for Na and Li 10, Ca 20 and K 750 ppm. These values are divided by a factor of two in glasses. Such detection limits are those required for the determination of ions in individual fluid inclusions. Calibration strategy for the determination of cation ratios For glasses and minerals, 10 series of five shot accumulations are recorded for each standard.The ratio used for the calibration is the average value corresponding to 50 shots. For each fluid inclusion, an average ratio is obtained after around Fig. 7 Calibration of the emission line of potassium at 766.5 nm. 10 shots. The final ratio for each standard is the mean value obtained on 10–12 fluid inclusions (corresponding to 100 with our apparatus.Therefore, samples with low concenspectra). The self-absorption process has been discussed pre- trations of potassium (<3% m/m) could not be studied. viously. The determination of the calibration curves takes into account the possibility of self-absorption for one element or Reproducibility both elements. The calibration curves for Na/Ca, Na/Li and Na/K are For solid samples (glasses and minerals), the shot-to-shot repeatability of the emission line absolute intensities is around given in Fig. 8, 9 and 10, respectively. For the three calibration curves, data from synthetic glasses, minerals and synthetic 10% for glasses (up to 20% for potassium) and in the range 4–10% for the emission line for a metal (e.g., copper). These fluid inclusions plotted along the same curve, attesting that emission line ratios are independent of the studied material. values are in agreement with those found by Geertsen et al.34 with the LA-OES technique.It is worth noting that the noise The RSDs are in the range 5–25% depending on the standards; high RSDs are found for the lowest concentrations of potass- of the electronic detection system limits the reproducibility to about 3%. Hence the repeatability of this technique is close to ium owing to the low intensity of the signal, especially in fluid inclusion standards. the RSD found using the electron microprobe technique for the analysis of solid matrices. The reproducibility of the ablation In Fig. 8, three curves are presented for Na/Ca ratios. The first (D1) is the regression line established from the standards. conditions is controlled with a standard, in our case copper metal. The intensity of the line at 520 nm was selected for its The second (D2) is calculated for low sodium concentrations for which the self-absorption process does not occur. The shot-to-shot stability (RSD#5%) and its good response. 918 J. Anal. At. Spectrom., 1999, 14, 913–922Fig. 10 Calibration curve obtained for Na/K ratios on synthetic Fig. 8 Calibration curve obtained for Na/Ca ratios on synthetic glasses glasses, synthetic inclusions and minerals. and synthetic inclusions. Therefore, a limit of calibration for Na/K ratios of Na (ppm)/K (ppm) <4 had to be used. Analytical strategy for natural samples Two types of calibration curve were established for the determination of the Na/Li, Na/Ca and Na/K ratios: D1 lines for which self-absorption process occurs for Na and D2 lines with no self-absorption process.Therefore, it is necessary to determine which calibration curve is relevant. In the previous section, the threshold of concentration above which the selfabsorption process occurs was established. It corresponds to a given intensity of the Na doublet emission lines. Consequently, below this intensity limit the relevant calibration curves are D2 lines and above it D1 lines are used. Application to natural fluid inclusions Natural fluid inclusions were analysed using LA-OES to check the validity of the method.Samples well characterised by microthermometric, Raman spectroscopic and crush–leach data were chosen for the first application of this technique (Pierre-Joseph clefts, French Alps). In addition, they were selected for their large size (30–50 mm) so as to be able to make several shots on each fluid inclusion. They are two phase Fig. 9 Calibration curve obtained for Na/Li ratios on synthetic glasses, inclusions, without any of solids (Fig. 11). The selection of synthetic inclusions and minerals. these natural inclusions was also based on moderate salinity (7% m/m equiv. NaCl), which is commonly observed in geological fluids. slope of line D2 is four times higher than that of D1, corresponding to high Na concentrations. This calibration Figure 12 shows spectra obtained by several shots on the same fluid inclusion. Na, K, Li and Ca emission lines were curve could be used for materials with low sodium levels (0.05 mol kg-1 H2O).observed. The emission line intensity varies during the ablation progress. However, it is important to note that the variations In Fig. 9, the calibration curve D1 for Na/Li was established from standards for which the self-absorption process occurs of the line intensity ratios shot to shot are within 25%; this indicates that the phenomenon has no influence on the plasma only for Na. The second line, D2, corresponds to the case of a low sodium concentration, below its self-absorption thresh- temperature.It was not possible for all the inclusions analysed to obtain old. It can be noted that its slope is three times greater than that of line D1. the three element ratios because (i) the detection limit for K is around 750 ppm, and the emission line intensity must be For calibration of Na/K (Fig. 10), the calibration curve shows a strong decrease in slope for Na/K >4. This eVect suYciently high to obtain a good SBR which could be then used for the ion ratio; (ii) filters are necessary to decrease the results mainly from the strong self-absorption process of Na, because the samples used have high Na concentrations (>20% Na emission line intensity to prevent saturation of the detector, but these filters simultaneously prevent Ca emission; hence in m/m Na2O).These high Na concentrations were necessary to have a measurable potassium signal for high Na/K ratios.some cases Ca emission lines could not be recorded; and J. Anal. At. Spectrom., 1999, 14, 913–922 919limit of detection for lithium in fluid inclusions, as the salinity is only 7% m/m equiv. NaCl. The RSD is around 30%. However, the RSD for the Na/Li ratio can be better (around 15–20%) for higher Li concentrations. Such an Na/Li determination is accurate enough since lithium concentration cannot be used as marker of fluid–mineral equilibrium. The Na/Li ratio is generally used as an indicator of fluid sources and only its order of magnitude is geochemically relevant.The Na/Ca ratio calculated for 26 fluid inclusions is 9 with an RSD of 20%. For the Na/Ca ratio we have found in some cases a relatively high RSD (close to 40%), which may be due to contamination of calcium. Despite the high RSD values, the technique gives satisfactory results, taking into account that the salinity of the fluid inclusions is around 7% m/m equiv.NaCl. Although the crush–leach technique analyses populations of thousands of fluid inclusions and LA-OES works on individual inclusions, the ratios for the four elements studied are in good agreement (Table 5). Such a result is encouraging for the validation of the LA-OES technique. Fig. 11 Microphotograph of fluid inclusions from alpine quartz, used for the first application of the LA-OES technique. Conclusions (iii) although Li displays a good signal response, it is not always detected in all the inclusions analysed, owing to the This work has shown that LA-OES can be used for localised analysis and especially for the quantification of ion ratios in too small size of the inclusion (<15 mm).Around 30 fluid inclusions from the same piece of wafer individual fluid inclusions. The laser beam is used to drill up to the inclusion and to produce plasma from the trapped were analysed by LA-OES. For each inclusion, the same procedures as those used for synthetic fluid inclusions were liquid.An optical spectrometer analyses directly the radiation emissions from the plasma. Plasma temperature studies showed carried out. The Na/K, Na/Li and Na/Ca ratios were calculated using the calibration curves (Table 5). that several standards (synthetic glasses, fluid inclusions and minerals) can be used for the establishment of calibration The Na/K value found with the calibration curve is 3 with an RSD of 33% for all the analyses. This result seems to be curves.Calibration curves for three element ratios (Na/K, Na/Li and Na/Ca) were established in the 580–790 nm spectral realistic for this geological context. For Na/K ratios, the reproducibility is satisfactory although the SBR is relatively range. The RSDs calculated from the calibration curves range from 5% (glasses) to 25% (fluid inclusions) and the detection small (often <5) as the potassium concentration is too low or the ablated mass of the liquid is too small.The Na/K ratio limits are those required for the determination of ions in individual fluid inclusions (Na and Li 10, Ca 20 and K allows the estimation of the fluid trapping temperature based on the equilibrium between fluid and K and Na feldspars.35 750 ppm). LA-OES is probably one of the techniques that can be used The RSD of the Na/K analysis results in a temperature uncertainty of around±50 °C. Such a temperature determi- to solve the challenge of determining major ion ratios in an average volume of liquid of mass <10-9 g.It is now possible nation is not accurate enough for geological interpretation. However, it could probably be optimised by using a to determine the ratios of major elements (Na, K, Li, Ca) in individual fluid inclusions. A recently available echelle spec- red-sensitive detector. The Na/Li ratio is close to 55 and it corresponds to the trometer has been used in LA-OES and allowed the collection Fig. 12 Optical emission spectra obtained by LA-OES on one natural individual fluid inclusion (three laser shots).The emission lines of Na, Li, K and Ca are used for the determination of the cation ratios via the calibration curves. 920 J. Anal. At. Spectrom., 1999, 14, 913–922Table 5 LA-OES data for 30 fluid inclusions from natural sample (Alpine clefts), their relative standard deviations (RSDs) and their cation ratios calculated from calibration curves No. of RSD RSD RSD Na/K Na/Li Na/Ca Inclusion shots I(Na)/I(K) (%) I(Na)/I(Li) (%) I(Na)/I(Ca) (%) (ppm) ± (ppm) ± (ppm) ± 71–47A 12 166 12 10 24 3 0.3 — 11 2.5 71–47B 11 120 25 21 21 12 24 2 0.4 65 13.7 — 71–47C 9 163 13 17 10 26 3 0.3 — 10 2.7 71–47D 6 126 24 22 26 9 20 2 0.4 69 17.9 9 1.9 71–47E 14 120 23 17 18 16 29 2 0.4 51 9.4 — 71–47F 15 180 9 18 8 15 3 0.3 — 8 1.3 71–47G 18 147 16 20 18 7 40 2 0.4 62 11.4 7 2.7 71–47H 8 190 9 12 25 4 0.3 — 12 2.9 71–47I 4 16 26 8 38 — 45 11.6 8 2.9 71–47J 1 110 18 18 8 8 — 52 9.1 8 0.6 71–47a 5 183 22 9 29 3 — 69 — 9 2.7 71–47b 14 182 31 21 42 10 31 3 1.0 64 26.4 10 3.0 71–47c 1 7 — —7— 71–47d 15 199 7 16 36 10 32 4 0.3 48 17.0 10 3.1 71–47e 4 18 11 19 — 55 — 11 2.1 71–47f 4 16 2 9 11 — 47 1.1 9 1.0 71–47g 2 27 19 8 7 — 85 15.9 8 0.6 71–47h 9 7 17 — — 7 1.1 71–47i 5 7 11 25 — — 11 2.6 71–47j 5 8 11 8 44 — 19 2.0 8 3.7 71–47k 3 14 49 10 18 — 41 19.9 10 1.8 71–47l 6 9 42 — — 9 3.9 71–47m 4 107 6 24 — — 6 1.4 71–47n 6 11 17 — — 11 1.9 71–47o 5 103 11 16 27 — 30 — — 71–47p 4 19 31 9 21 — 58 17.7 9 2.0 71–47q 3 8 37 — — 8 3.1 71–47r 4 15 20 — — — 71–47s 7 5 30 — — 4 1.3 71–47t 7 231 9 22 5 — — 9 2.1 LA-OES 3 54 9 s 1.0 16.1 2 RSD (%) 32.9 29.9 19 Crush–leach 3 69 7 J.Anal. At. Spectrom., 1999, 14, 913–922 92111 T. J. Shepherd and S. R. Chenery, Geochim. Cosmochim. Acta, of the emission spectrum from 200 to 800 nm.36 Such a spectral 1995, 59, 3997. range permits the emission lines of most chemical elements to 12 A.Aude�tat, D. Gu�nther and C. A. Heinrich, Science, 1998, 279, be obtained and will make LA-OES as a multi-element analyt- 2091. ical technique. Its use for fluid inclusion analysis will be 13 D. Gu� nther, A. Aude�tat, R. Frischknecht and C. A. Heinrich, checked in the near future. J. Anal. At. Spectrom., 1998, 13, 263. 14 M. C. Boiron, J. Dubessy, N. Andre�, A. Briand, J. L. Lacour, P. Mauchien and J. M. Mermet, Geochim. Cosmochim. Acta, 1991, Acknowledgements 55, 917. 15 M.C. Boiron, A. Moissette, C. Fabre, J. Dubessy, D. Banks and The EU program MAT1-CT-93–0029 supported this work B. Yardley, in Proceedings of the XIV ECROFI Conference, ed. and P. Mauchien (CEA, LSLA, Saclay, France) is warmly M. C. Boiron and J. Pironon, Nancy, 1997, p. 44. acknowledged for managing the project and coordinating the 16 A. Moissette, J. Dubessy, M. C. Boiron, C. Fabre, P. Mauchien teams to achieve success. J. L. Lacour (CEA, LSLA) is and J. L. Lacour, in Proceedings of the XIV ECROFI Conference, acknowledged for assistance during the advancement of the ed.M. C. Boiron and J. Pironon, Nancy, 1997, p. 211. 17 L. Moenke-Blankenburg, Laser Microanalysis. Chemical Analysis, experimental procedure. The authors thank C. PeiVert a Series of Monographs on Analytical Chemistry and its (CREGU), S. Decitre and L. Tissandier (CRPG) for their Applications, vol. 105, Wiley-Interscience, New York, 1989. help in the synthesis of the glasses which were used for the 18 L.J. Radziemski and D. A. Cremers, Laser Induced Plasma and calibration of the technique, and K. Schmulovich for providing Its Applications,Marcel Dekker, New York, 1989. a series of synthetic fluid inclusions. T. Lhomme (CREGU) 19 U. Paisch, M. Clara and R. Niessner, Spectrochim. is thanked for her help during the experiments. The authors Acta, Part B, 1998, 53, 1957. 20 C. Chale�ard, J. Anal. At. Spectrom., 1997, 12, 183. thank D.Banks (Leeds University) for the crush–leach analysis 21 CEA, Fr. Pat., No. 93-13855, 1993. of fluid inclusions and M. Cathelineau (CREGU) for fruitful 22 N. Andre�, unpublished thesis, Universite� Paris Sud, 1995. discussions during the progress of this study. 23 C. Geertsen, unpublished thesis, Universite� Paris Sud, 1996. 24 P. Mauchien, A. Bengston, R. DeMarco, E. DaSilva, J. Dubessy, F. Noronha, C. Prieto, G. Tomandl and B. W. Yardley, in Final References Report, European Program MAT1-CT-93–0029, 1996. 1 C. Ayora, J. Garcia-Veigas and J. J. Pueyo, Geochim. Cosmochim. 25 S. Sterner and R. J. Bodnar, Geochim. Cosmochim. Acta, 1984, Acta, 1994, 58, 43. 48, 2659. 2 A. J. Anderson, A. H. Clark, P. Max, G. R. Palmer, 26 B. Poty, J. Leroy and L. Jackimowicz, Bull. Miner., 1976, 99, 182. J. D. Macarthur and E. Roedder, Econ. Geol., 1989, 84, 924. 27 Handbook of Chemistry and Physics, ed. R. C. Weast, CRC Press, 3 C. G. Ryan, D. R. Cousens, C. A. Heinrich, W. L. GriYn, Boca Raton, FL, 60th edn., 1980. S. H. Sie and T. P. Mernagh, Nucl. Instrum. Methods B, 1991, 28 M. Autin, unpublished thesis, Universite� Lyon I, 1990. 54, 292. 29 A. J. Ellis, in Geochemistry of Hydrothermal Ore Deposits, ed., 4 M. Volfinger, C. Ramboz, M. Aissa and C. G. Choý�, in H. L. Barnes, John Wiley, New York, 2nd edn., 1979, p. 632. Proceedings of the XIV ECROFI Conference, ed. M. C. Boiron 30 W. F. Linke, Solubilities of Inorganic and Metal–Organic and J. Pironon, Nancy, 1997, p. 344. Compounds, vols. I and II, American Chemical Society, 5 J. D. Frantz, H. K. Mao, Y. G. Zhang, Y. Wu, A. C. Thomson, Washington, DC, 1965. J. H. Underwood, R. D. Giauque, K. W. Johns and M. L. Rivers, 31 A. S. Borisenko, Geol. Geophys., 1977, 8, 15. Chem. Geol., 1988, 69, 235. 32 In Solubility of Inorganic Substances in Water, ed. 6 D. A. Vanko, S. R. Sutton, M. L. Rivers and R. J. Bodnar, Chem. A. N. Kirguintsev, L. N. Trouchnikova and V. G. Lavrentieva, Geol., 1993, 109, 125. 1972. 7 J. A. Mavrogenes, R. J. Bodnar, A. J. Anderson, S. Bajt, 33 M. Cathelineau, Ch. Marignac, M. C. Boiron, G. Gianelli and S. R. Sutton and M. L. Rivers, Geochim. Cosmochim. Acta, 1995, M. Puxeddu, Geochim. Cosmochim. Acta, 1994, 58, 1083. 59, 3987. 34 C. Geertsen, J. L. Lacour, P. Mauchien and L. Pierrard, 8 P. Philippot, B. Menez, P. Chevallier, F. Gibert, F. Legrand and Spectrochim. Acta, Part B, 1996, 51, 1403. P. Populus, Chem. Geol., 1998, 144, 121. 35 S. Verma and E. Santoyo, J. Volcanol. Geotherm. Res., 1997, 79, 9. 9 D. Gu� nther, A. Aude�tat, R. Frischknecht and C. Heinrich, in 36 C. Haisch, U. Panne and R. Niessner, Spectrochim. Acta, Part B, Proceedings of the XIV ECROFI Conference, ed. M. C. Boiron 1998, 53, 1657. and J. Pironon, Nancy, 1997, p. 144. 10 A. Moissette, T. J. Shepherd and S. R. Chenery, J. Anal. At. Spectrom., 1996, 11, 177. Paper 8/09338E 922 J. Anal. At. Spectrom., 1999, 1