Capabilities of an Argon Fluoride 193 nm Excimer Laser for Laser Ablation Inductively Coupled Plasma Mass Spectometry Microanalysis of Geological Materials† DETLEF GU� NTHER*a , ROLF FRISCHKNECHTa , CHRISTOPH A. HEINRICHa AND HANS-J. KAHLERTb aSwiss Federal Institute of T echnology (ETH) Zurich, Institute of Isotope Geology and Mineral Resources, CH-8092 Zurich, Switzerland bMicroL as L asersystem, D-37079 Go�ttingen, Germany Recent developments in laser ablation inductively coupled for trace element analysis of geological samples and other environmental materials, owing to the increased sensitivity of plasma mass spectrometry (LA-ICP-MS) have demonstrated its potential for in situ microanalysis for major, minor and ICP-MS and the more ecient interaction of UV lasers with solid samples.The use of UV laser beams has led to a more trace elements in solids, such as minerals. With the low backgrounds and high sensitivity of new ICP-MS instruments, controlled ablation process.1 With new generation ICP-MS instruments, modifications of the sample cone geometry and limits of detection of 1–10 ng g-1 in a 40 mm ablation pit for many elements can be reached. Fractionation eects due to changes in the torch box configuration, limits of detection for ten selected rare earth elements (REE) of less than 10 ng g-1 dierent ablation rates of various elements have prevented quantification without matrix-matched standards with 1064 nm in a 35–40 mm pit can now be reached.2,3 Earlier work by Hirata and Nesbitt,4 Fryer et.al.5 and Jeries et al.6 demon- Nd5YAG lasers. These eects have been reduced but not eliminated using shorter UV wavelengths (e.g. a quadrupled strated the limitations of LA-ICP-MS analysis because of significant fractionation eects observed during the ablation Nd5YAG 266 nm). Excimer lasers with wavelengths below 200 nm are expected to reduce fractionation eects further, but for some elements, especially Zn, Pb and U, which are of major interest in geological samples.Stix et al.7 reported matrix they present a serious challenge to the design of optical systems, especially if high-resolution UV ablation needs to be eects for mineral analysis using synthetic glass standards. Various strategies applied to LA-ICP-MS to minimise these combined with high quality visual observation, which is essential for the study of complex materials, such as geological eects have been reported, including moving the stage during ablation,4 spraying water onto the ablation site or the use of samples.An LA system was developed using an homogenized UV laser beam (193 nm, Argon Fluoride excimer) with a several internal standards for specified groups of elements of geochemical interest.8 However, none of these approaches can common UV–visual objective on a modified petrographic microscope with reflected and transmitted light illumination, in be used for routine LA analysis or match the improved ablation characteristics oered by the 193 nm excimer system described.combination with a Perkin-Elmer Elan 6000 ICP-MS instrument. The optical system allows imaging of both visible and UV laser light onto the sample surface at the same time. Laser operating parameters and their influence on the ablation process were investigated using NIST SRM 612/610. EXPERIMENTAL Fractionation eects due to dierential ablation of various ICP-MS Instrumentation elements as a function of time can be reduced to interelement correlation coecients of r=0.9 or better and have become The ICP-MS instrument, an Elan 6000 (Perkin-Elmer, insignificant within the precision of quadrupole ICP-MS using Norwalk, CT, USA), has a sensitivity for La of 100×106 this new optical design.Energy densities and repetition rates counts s-1 per mg g-1 abundance when used with solution need to be kept within limited ranges for accurate and sample introduction using a standard concentric nebuliser, and reproducible determinations of trace elements such as Zn, U 1000–2000 counts s-1 per mg g-1 for a 40 mm (10 Hz) ablation and Pb, which have previously presented strong fractionation pit.Quantitative determination of trace elements in mineral problems. LA-ICP-MS determinations on natural hornblende, grains with LA depends on the determination of at least one augite, and garnet, calibrated against NIST SRM 612 using major element of known concentration which can be used as any major element as an internal standard, agree well with an internal standard element to correct for fluctuations in the independent literature data.These experiments with the Argon rate of sample ablation. The Elan 6000 is supplied with an Fluoride 193 nm excimer system demonstrate a greatly analogue and digital detector system, which eectively allows reduced matrix dependence of the ablation process, which simultaneous detection of concentrations up to 3×106 facilitates in situ analysis of unknown samples.counts s-1 in the pulse-counting mode and at signal intensities above this threshold in the analogue mode. The analogue Keywords: L aser ablation; excimer laser; inductively coupled mode is limited in speed by a detector-settling time of about plasma mass spectrometry; geological material 3 ms, which leads to a 70% measurement eciency (the eective time fraction used for measurement). Ultraviolet laser ablation inductively coupled plasma mass ICP-MS instrument optimisation for laser induced aerosol spectrometry (LA-ICP-MS) is an increasingly important tool introduction is slightly dierent from that used for solution nebulisation. However, no single parameter can be used to explain the dierences between wet and dry plasma conditions.† Presented at the 1997 European Winter Conference on Plasma Spectrochemistry, Gent, Belgium, January 12–17, 1997. An in-house machined Al cone with an orifice diameter of Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 (939–944) 9390.7 mm and two additional rotary pumps in parallel with the illumination with a central observation to comply with the Schwarzschild mirror objective performance. The condenser standard supplied pumps (Edwards 18 Two Stage Vacuum Pump, Edwards High Vacuum, Oakville, Ontario L6K 2H4, lens behind the homogeniser collimates the parallel light into a field of 0.5 x 0.5 cm in front of the field lens.The field lens Canada) were used, to reduce pressure in the expansion chamber between the sampler and skimmer cones.2 The provides the optically processed laser light to the Schwarzschild imaging objective. A masking aperture behind the field lens reduced pressure lowers the background by 1–2 orders of magnitude for most of the elements and increases the peak consists of round holes and was varied from 100 mm to obtain an ablation pit of 4 mm diameter to 2.5 mm to obtain a 100 mm intensity of elements with m/z>85 by a factor of three.pit. A 45° dielectric mirror which reflects UV while transmitting visible light, reflects the UV beam onto the reflecting objective Laser Ablation Sample Introduction System lens (25× magnification), with 1.35 cm focal length and 35–40% transmission, which images the aperture onto the An excimer laser (Compex 110i, Argon Fluoride 193 nm, sample (Fig. 1). The objective thus images the aperture onto Lambda Physik, Go� ttingen, Germany) with a gas mixture the sample surface, which permits even illumination at constant containing 5% F2 in Ar with small amounts of He and Ne energy densities, independent of the size and shape of the was used.The maximum output energy of the laser is sample spot. The mirror objective has a high numerical aper- 200 mJ per pulse with a beam size of 2×1 cm (homogeneous ture of 0.4, to obtain high image resolution and to avoid a beam energy). The repetition rate can be varied from 1 to high energy density on the window of the sample chamber. 10 Hz with 200 mJ per pulse using air cooling, and up to Pulse energy can be varied from 50 mJ to 1.5 mJ using 100 Hz at 145 mJ per pulse using water cooling. The beam dierent laser output energies and by a beam splitter in front path is shown in Fig. 1. The inner maximum of the beam of the first prismpulse energy can be measured after the profile is converted into an outer maximum with prism 1 aperture (Laser Energy Meter EM1, EDL-500A, GMP, and prism 2 (telescope) and homogenised by two crossed Renens, Switzerland) or at the sample site by inserting a small lens arrays (9×9 lenses, 3 mm thick, MicroLas, Go� ttingen, Germany).The prism set up is necessary to produce the active area joulemeter (ED-100A, GMP). The stability of the Fig. 1 Schematic representation of the optical beam path for an excimer laser (193 nm, ArF) in combination with a petrographic microscope. 940 Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12laser pulse energies measured for 1000 pulses showed a relative RESULTS AND DISCUSSION standard deviation of less than 1%. The low energy density of the large excimer beam profile The sample is placed in a Plexiglas cell with a volume of (2×1 cm) was increased for LA using a homogeniser in combi- 20 cm3, from which the ablated material is carried into the nation with a condenser and a field lens to produce energy ICP-MS by an Ar gas stream.A fused silica window 0.4 mm densities of 2–20 J cm-2 corresponding to pulse energies of thick is anti-reflection coated to minimise UV reflection and 50 mJ to 1.5 mJ. In addition, the divergence of the excimer to reduce the formation of multiple laser beam images. The laser beam was limited using the two arrays transferring the sample can be observed through the petrographic microscope beam into a homogeneous field of 0.5×0.5 cm, which is using eye pieces and a range of low magnification objective produced right in front of the aperture.Only small areas lenses. During ablation, the optical image can be continously (2.5 mm in diameter or less) of this homogenised field are monitored through the Schwarzschild objective and a TV imaged on the sample surface for pit sizes of 4–100 mm, hence camera. The microscope is an assembly of a microscope head allowing constant energy densities on the sample surface.with a changeable mirror-holder (Axiotech, Zeiss) and a micro- Owing to the homogeneous beam profile, the intensity is a scope body (Axioplan, Zeiss) to extend the working distance linear function of the energy density or energy per pulse. The between the microscope table and the objective and to allow sensitivity over a small part of the attainable range is shown sample observation with transmitted and reflected light. All for a 20 mm pit [Fig. 2(a)]. The pulse energy per shot can be optical elements are anti-reflection coated and mounted on an increased up to 1.5 mJ per shot. Deviations from a linear optical bench. A Plexiglas box covering the bench protects the function are caused primarily by measurement uncertainties. user from potentially dangerous UV light and shields the The pulse energy is a linear function of the aperture size optical parts from dust. Flushing of the box with N2 further and allows sample ablation within the whole range of pulse reduces laser beam absorption. energies for a given pit size [Fig. 2(b)]. The ablation depth per pulse is a linear function of the pulse energy and can be varied between 0.05 and 0.5 mm per pulse. This low ablation rate and homogenous illumination of the laser beam onto the sample Data Acquisition surface allows a very well defined ablation process independent The 20 elements listed in Table 1 were measured using a 10 ms of the spatial resolution.A signal for ablation through a 30 mm dwell time and 3 ms quadrupole settling time. Backgrounds thin section of quartz mounted on a plate of Na bearing glass were measured for 30 s (laser not firing) and the transient is shown in Fig. 3. The ablation rate per pulse calculated from signals from the analytes were acquired for approximately 30 s. the Si signal is 0.075 mm. Background corrected signals were integrated. The internal Owing to the high numerical aperture of the objective, standard used for glass and mineral analyses was Ca.required for high-resolution imaging, clean cylindrical pits can Calibration for the minerals was carried out with NIST SRM only be drilled to a depth of about 0.8 pit diameters. However, 612 glass as the external standard. Limits of detection were apertures can be switched during the ablation process, which calculated as three times the standard deviation of the back- permits a stepwise ‘zooming-in’ to target, for example, a small ground normalised to the volume of sample ablated (counts s-1 inclusion deep inside the sample, as illustrated by Fig. 4. In per mg g-1). The data acquisition parameters used are listed addition, prior to data acquisition, very low pulse energies of in Table 1. 50 mJ can be used to remove very thin layers of the sample Table 1 LA-ICP-MS working conditions ICP–MS— Instrument Elan 6000 (Perkin Elmer) Intermediate gas 0.75 l min-1 flow Aerosol carrier 1.3 l min-1 gas flow Outer gas flow 16.4 l min-1 Detector mode Dual, pulse counting and analogue mode Rf power 1200W Vacuum pressure 1.45×10-5 Torr* 7.6×10-6 Torr (two additional rotary pumps, Al cone with 0.5 mm diameter) Isotopes 29Si, 42Ca, 66Zn, 88Sr, 90Zr, 93Nb, 137Ba, 139La, 140Ce, 141Pr, 146Nd, 147Sm, 151Eu, 159Tb, 157Gd, 163Dy, 165Ho, 167Er, 169Tm, 173Yb, 175Lu, 178Hf, 181Ta, 208Pb, 232Th, 238U Excimer laser— Compex 110I (Lambda Physik) ArF 193 nm Output energy 200 mJ at 193 nm Pulse duration 15 ns Repetition rate 1–100, 5, 10, 20, 50, 100 Hz Aperture 0.075–2.5 mm (8 dierent hole sizes) Objective Schwarzschild objective (25× magnification), 1.35 cm focal length Ablation cell Plexiglas, 20 cm3 Gas inlet Nozzle (0.1 mm) Fig. 2 (a), Sensitivity and energy per pulse as a function of the energy Window Fused silica, 0.4 mm thick, anti-reflection coated density for a given pit size of 20 mm; (b), linear function between pulse energy and pit size, obtained by the imaging optics of an excimer laser configuration.* 1 Torr=133.322 Pa. Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 941sample uptake visible in SEM pictures taken for a 100 Hz ablation (picture not shown). Dierent pulse energies (2–20 J cm-2) and repetition rates (5–100 Hz) were used for the analysis of the NIST SRM 612. Listed in Table 2 are only two mean concentrations determined, using 10 and 100 Hz. The higher relative standard deviation and the overestimated concentrations (based on Ca as the internal standard) for Zn and Pb (given in bold in Table 2) using repetition rates above 20 Hz demonstrated that fractionation eects are occurring during the ablation.The correlation of the signal intensity of Ca with Pb and Ca with Zn give correlation coecients of about 0.8, even for a 193 nm laser, whereas other elements are not influenced by the high repetition rate. The mean concentrations from analysis using 10 Hz deviate from the certified values by less than 5% (see Table 2).Transient signal correlation for Ca with Zn gave a Fig. 3 Signal (30 s) for drilling through a 30 mm thin section of quartz correlation coecient between these signals of r=0.96, which (approximately 0.1 mm ablation depth per pulse using 10 Hz). is an indication of a controlled ablation process, minimising element fractionation as a function of time to a statistically insignificant value. The pulse energy, responsible for the sample uptake, shows less influence on elemental fractionation, however, the best interelement correlation coecients have been calculated for energy densities between 8.5 and 14 J cm-2.Owing to the greatly reduced fractionation eects, other calibration strategies such as direct solution ablation using the 193 nm can be used in LA, as has been reported elsewhere.9 Limits of detection are a function of the background noise and the sensitivity.10 The dependence of the limits of detection on pit size (constant repetition rate 10 Hz) is shown in Fig. 5.In this experiment, 26 isotopes were measured with a total of 30 s integration of the gas background and 30 s integration for the signal. The linear function of pulse energy and aperture size allows limits of detection below 10 mg g-1, even for 4 mm pits. Uncertaties in the limits of detection pattern for 88Sr Fig. 4 SEM picture of a ‘drill cascade’ into the NIST SRM 612 using sequentially smaller aperture sizes to sample a 10 mm pit about 100 mm below the sample surface.with a larger diameter, in order to remove all contamination from the sample surface. This is important for the determination of the trace element distributions in very small samples. Signal intensities can be changed by using dierent repetition rates. Experiments using 10, 20, 50 and 100 Hz showed increased sensitivity with higher repetition rates, through higher intensities over a shorter time. Unfortunately, the relationship between the repetition rate and signal intensity is not a linear function.Measurements of the output energy show that the laser is not able to maintain the same pulse energy at maximum repetition rates at 100 Hz, as was obtained at 10 Hz. The energy losses above 20 Hz are approximately 10%. The signal intensities for 5–100 Hz dier by more than 10%, which Fig. 5 Relation between limits of detection and pit size (constant frequency and energy density). can be explained by more particle losses owing to the fast Table 2 Improvements in accuracy using selected ablation parameters Dierence Dierence Reference Low frequency (10 Hz) relative to High frequency (100 Hz) relative to value/ measured concentration reference value measured concentration reference value Element m/z mg g-1 mg g-1 (%) mg g-1 (%) Zn 66 49.4 47.9 3.0 60.4 21.2* Sr 88 75.3 74.6 0.9 75.0 0.2 Ba 137 38.9 39.1 0.3 39.4 0.6 Ce 140 37.9 37.8 0.4 38.0 0.2 Tb 159 36.9 37.1 0.4 36.4 1.3 Lu 175 37.2 36.6 1.5 36.1 3.3 Pb 208 36.6 35.3 3.5 44.0 19.1 U 238 36.7 37.1 1.2 37.5 2.6 * See text for explanation of bold figures. 942 Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12and 141Pr (see Fig. 5) are caused by single spikes in the reference material (NIST SRM 612 Trace Elements in Glass) was used for the quantification of trace elements in these background leading to an increased standard deviation. The high background on Pb, caused by ubiquitous Pb contami- minerals, using Ca as the internal standard. The reproducibility of homogeneous samples is for most elements better than 4%.nation in a normal laboratory environment, leads to higher limits of detection compared with all other elements shown in However, variations due to mineral heterogeneity were not investigated in detail. The comparison of the LA-ICP-MS data this figure. The limits of detection calculated for the ablation rate of 0.1 mm per pulse indicate high transportation eciency with literature12,13 values shows satisfactory agreement for most of the elements (Tables 3–5). (90%) from the ablation cell into the plasma.The relatively high count yield per ablated material is probably due to a The amount of ablated material measured as Ca sensitivity (counts s-1 per mg g-1) was in all minerals within 20% of that predominance of smaller particles obtained by ablation with the 193 nm excimer laser, as indicated by particle-counting of the NIST SRM 612, which demonstrates that the ablation rate with excimer system is relatively matrix-insensitive. The studies currently in progress.11 Analyses with optimised laser and ICP-MS conditions have same ablation behaviour was observed for zircon (data not shown).been carried out on the natural hornblende, garnet and augite. For the demonstration of the accuracy an external calibration CONCLUSION Table 3 Trace element determinations of Kakanui hornblende, in The high quality petrographic microscope laser ICP-MS comparison with literature data system, designed for the 193 nm excimer laser, allows sample observation using transmitted and reflected light.The use of a Reference Reference beam homogeniser leads to an improved beam quality and a value/ value/ homogeneous imaging of the laser beam onto the sample Element mg g-1 This work Element mg g-1 This work surface, and thus to a controlled ablation process and a very Rb 15* 21±1.2 Sm 4.4† 4.4±0.2 reproducible pit structure.The energy density is independent Sr 480* 483±9.2 Eu 1.6† 1.56±0.03 of the pit size and of the pit geometry due to the homogeneous Y 10* 9.6±0.5 Gd 4.2† 3.7±0.12 lateral and angular illumination, such that energy density is a Zr 58* 51.8±1.7 Dy — 2.6±0.2 Nb 24* 25.5±0.2 Yb — 0.34±0.06 linear function of pulse energy. The improved beam quality Ba 260* 295±5.1 Hf 1.9† 2.2±0.1 allows pits of less than 4 mm diameter with limits of detection La 5.0† 5.0±0.01 Th — 0.06±0.02 of less than 10 mg g-1 for selected elements between 85Rb and Ce 15.9† 17.6±0.2 Cr 5† 7.6±1.7 238U.The adjustable ablation rate of 0.05–0.5 mm per pulse is Nd 15.2† 15.9±0.2 Sc 18† 18±0.9 suitable for vertical depth profiling and allows controlled removal of surface contamination prior to analysis. As a result * Ref. 12. † Ref. 13. of the improved pulse to pulse stability of the excimer laser, Table 4 Trace and major element determinations in Kakanui augite Reference Reference value/ value/ Element mg g-1 This work Element mg g-1 This work Rb — 0.7±0.2 Eu 0.79† 0.8±0.1 Sr 58* 58.8±0.5 Gd 2.3† 1.7±0.3 Y 8.5* 7.8±0.3 Dy — 1.9±0.1 Zr 24* 20.7±1.1 Yb 0.67† 0.48±0.17 Nb — 0.28±0.09 Hf — 1.1±0.2 Ba — 0.41±0.16 Th — 0.03±0.01 La 1.76† 1.21±0.05 Sc 29† 37±1.5 Ce 5.7† 5.6±0.2 SiO2 (50.73% m/m) (53.4±2.5% m/m) Nd 6.7† 5.7±0.2 — — — Sm 2.14† 2.17±0.2 — — — * Ref. 12. † Ref. 13. Table 5 Trace and major element determinations in Kakanui garnet Reference Reference value/ value/ Element mg g-1 This work Element mg g-1 This work Rb — N.d.* Eu 0.65† 0.62±0.02 Sr — 0.82±0.3 Gd 3.5† 2.9±0.4 Y 55† 44.8±0.7 Dy 8.5† 6.9±0.6 Zr 48‡ 43.8±0.5 Yb 9.1† 7.8±0.9 Nb — 0.62±0.02 Hf 1† 0.8±0.18 Ba — 0.22±0.05 Th — N.d.La 0.02† 0.03±0.01 Sc 132† 169±2 Ce 0.2† 0.12±0.07 SiO2 (41.4% m/m) 43.3±0.7% m/m) Nd 1.14† 0.34±0.23 — — — Sm 1.15† 0.85±0.1 — — — * N.d.=not detected. † Ref. 13. ‡ Ref. 12. Journal of Analytical Atomic Spectrometry, September 1997, Vol. 12 943the precision of analyses for homogeneous materials such as REFERENCES glass, are less than 4% even for concentrations below 1 mg g-1. 1 Jackson, S. E., Longerich, H. P., Dunning, G. R., and Fryer, B. J., Elemental fractionation eects are insignificant, if a carefully Can. Mineral, 1992, 30, 1049. selected combination of intermediate energy density depending 2 Gu� nther, D., Longerich, H. P., Jackson, S. E., and Forsythe, L., on the material and relatively low repetition rates (<20 Hz) is Fresenius’ J.Anal. Chem., 1996, 355, 771. used, which minimise local melting of the sample. Increasing 3 Jackson, S. E., Longerich, H. P., Horn, I., and Dunning, G. R., Abstract, VM Goldschmidt Conference, 1996, p. 283. the pulse repetition rate to between 10 and 100 Hz improves 4 Hirata, T., and Nesbitt, R. W., Geochim. Cosmochim. Acta, 1995, count rate, but the improvement is less linear and leads to 59, 2419. increased element fractionation in silicates above 20 Hz. 5 Fryer, B. J., Jackson, S. E., and Longerich, H. P., Can. Mineral, Ablation rate was found to be largely matrix-independent 1995, 33, 303. amongst the NIST SRM 612, hornblende, augite and garnet, 6 Jeries, T. E., Pearce, N. J. G., Perkins, W. T., and Raith, A., which may in future open the possibility of direct LA-ICP-MS Anal. Commun., 1996, 33, 35. analysis of trace elements without a pre-determined internal 7 Stix, J., Gauthier, G., and Ludden, J. N., Can. Minal, 1995, standard element. However, calibration without an internal 33, 435. 8 Longerich, H. P., Jackson, S. E., and Gu� nther, D., Fresenius’ standard needs more studies on dierent minerals and other J. Anal. Chem., 1996, 355, 538. solid samples to be carried out. A direct comparison between 9 Gu� nther, D., Frischknecht, R., Mu�schenborn, H. -J., and Heinrich, excimer 193 nm and Nd5YAG lasers (266 nm) is not possible C. A., Fresenius’ J. Anal. Chem., 1997, 357, 358. due to the dierences in the beam profile, the wavelength and 10 Longerich, H. P., Jackson, S. E., and Gu� nther, D., J. Anal. At. the optical components necessary to transfer the beam onto Spectrom., 1996, 11, 899. the sample surface. 11 Frischknecht, R., Garbe-Scho� nberg, D., Cousin, H., and Gu� nther, D., unpublished data. The authors acknowledge the financial support of the ETH 12 Czamanske, G. K., Sisson, T. W., Campbell, J. L., and Teesdale, Zurich and the support of the Swiss National Science W. J., Am. Mineral, 1993, 78, 893. 13 Mason, B., and Allen, R. O., J. Geol. Geophys., 1973, 16, 935. Foundation (Project 2100–045548.95/1). GMP (Renens) also provided support of the optical components. Expert technical help by Urs Menet was essential for the successful construction Paper 7/01423F of the laser ablation system, for which we are most grateful. Received February 28, 1997 Critical reading of the manuscript by H. Longerich and two anonymous reviewers is also greatly appreciated. Accepted June 3, 1997 944 Journal of Analytical Atomic Spectrometry, September 1