Global and Planetary Change 21 Ž1999. 151–171 www.elsevier.comrlocatergloplacha The application of laser ablation ICP-MS to the analysis of volcanic glass shards from tephra deposits: bulk glass and single shard analysis Nicholas J.G. Pearce a,) , John A. Westgate b, William T. Perkins a , Warren J. Eastwood a , Philip Shane b,c b a Institute of Geography and Earth Sciences, UniÕersity of Wales, Aberystwyth, Wales SY23 3DB, UK Physical Sciences DiÕision, Scarborough College, UniÕersity of Toronto, 1265 Military Trail, Scarborough, Ontario, Canada, M1C 1A4 c School of EnÕironmental and Marine Sciences, UniÕersity of Auckland, Tamaki Campus, PriÕate Bag 92019, Auckland, New Zealand Abstract Laser ablation inductively coupled plasma mass spectrometry ŽLA-ICP-MS. has been applied to the chemical analysis of fine-grained Ž125–250 mm. volcanic glass shards separated from tephra deposits. This has been used both for bulk sample analysis and for the analysis of individual shards. Initial work concentrated on the use of an infra-red ŽIR. laser operating at 1064 nm, which gave craters of the order of 200 mm and was suitable for the analysis of bulk samples. This technique requires of the order of 80 mg of sample to determine a full suite of trace elements. Modification of the laser optics to enable operation in the ultra-violet ŽUV, at 266 nm. enables craters between 5 and 50 mm diameter to be produced, and the UV laser couples better with glass than the IR laser. We have applied this UV laser system to the analysis of single shards from Miocene tephra deposits from the Ruby Range in south-west Montana. Detection limits are below 1 ppm for a wide range of petrogenetically significant elements, but are critically dependent upon operating conditions. Calibration is achieved using synthetic multi-element glasses, with internal standardisation provided from electron probe analyses. Analysis of single shards provides a wide range of data from a single sample, enabling Ži. magmatic evolution to be discerned within one eruption and Žii. the identification of separate populations of shards within one deposit which may not be apparent from the electron probe data. In this paper we will present a summary of the techniques used for both bulk sample and single shard analysis and compare some new bulk analyses with analyses of glass derived from other analytical methods. q 1999 Elsevier Science B.V. All rights reserved. Keywords: laser ablation; ICP-MS; tephra analysis; volcanic glass; tephrochronology 1. Introduction The accurate recognition of individual tephra deposits in the stratigraphic record can provide valu) Corresponding author. Tel.: q44-1970-622599; fax: q441970-622659; e-mail: njp@aber.ac.uk. able information for many branches of geological science. These isochronous markers contain a juvenile volcanic component Žsuch as glass shards or phenocrysts., which, after deposition and reworking, can become heavily contaminated by detrital material. This contamination can make bulk analyses unreliable. The separation of any juvenile component 0921-8181r99r$ - see front matter q 1999 Elsevier Science B.V. All rights reserved. PII: S 0 9 2 1 - 8 1 8 1 Ž 9 9 . 0 0 0 1 2 - 0 152 N.J.G. Pearce et al.r Global and Planetary Change 21 (1999) 151–171 Že.g., glass shards or phenocrysts. from the bulk sample can be difficult and time consuming. Thus, tephra deposits are commonly characterised by grain discrete methods such as electron probe microanalysis ŽEPMA., which requires only small amounts of material Žsee Smith and Westgate, 1969.. Trace element determination in tephra deposits has, to date, generally involved analysis of bulk material Žeither whole sample or a separated juvenile component. by X-ray fluorescence ŽXRF. or instrumental neutron activation ŽINAA. ŽSarna-Wojicicki et al., 1979; Westgate and Gorton, 1981.. Both methods require of the order of grammes of juvenile material to be separated, and to be free from impurities. Recently, inductively coupled plasma mass spectrometry ŽICP-MS. has become a widely used technique for the analysis of geological materials. Solution nebulisation ICP-MS requires between 0.1 and 0.5 g of sample to be digested in acid for analysis, and has detection limits well below the part per million level ŽJarvis et al., 1992.. In addition, solid samples can be vapourised and the volatilised material passed into the plasma of an ICP-MS, in a manner analogous to the nebulisation of solutions, for analysis. Instrumental operating conditions vary between solid and solution analysis but the instrument can be optimised on a daily basis to give the best sensitivity for the particular sample introduction method. Laser ablation ŽLA. is fast becoming the most widely used solid sample introduction method for ICP-MS and has Ži. excellent spatial resolution and Žii. removes the need for sample digestion in bulk analysis. The latter also overcomes Ži. the need for large bulk separates and Žii. volatile or precipitation loss of some elements during the acid digestion process and furthermore Žiii. has the major advantage of being considerably quicker than sample dissolution methods. Bulk sample LA-ICP-MS can thus be used to analyse very small volumes of solid material and will routinely achieve sub-part per million detection limits ŽPerkins and Pearce, 1995, Perkins et al., 1991, Westgate et al., 1994.. LA-ICPMS analysis, however, requires knowledge of at least one element in the unknown to be used as an internal standard for calibration. In our first contribution to the development of this technique, Westgate et al. Ž1994. described the potential of LA-ICP-MS for the analysis of glass shards and the correlation of tephra deposits, using Ce determined by other methods Žeither INAA or solution ICP-MS. for internal standardisation. More recently, Pearce et al. Ž1994, 1996. described the use of a minor isotope of a major element determined by EPMA data for internal standardisation. In this contribution we will briefly summarise our earlier work on bulk samples, presenting some new data, and describe current developments for the trace element analyses of individual glass shards. 2. Instrumentation: general 2.1. InductiÕely coupled plasma-mass spectrometry The basic instrument used in this study is a VG PlasmaQuad II q ICP-MS. This is fitted with the 1992 high sensitivity interface, available as a commercial upgrade on all post-1992 instruments. For single shard analysis it has been necessary to modify the first stage vacuum system to enable more sample to be introduced into the spectrometer, in a manner similar to the currently available commercial ‘S-option’ upgrade. This is described more fully later. For bulk sample analysis, however, the increased pumping of the first stage vacuum is not required. 2.2. Laser The basic laser system employed in this study is a VG Elemental LaserLab, based on a Spectron 500 mJ Nd:YAG infra-red ŽIR. laser operating at 1064 nm. In its standard operating mode Žthe configuration used for bulk analysis of glass shards. this laser produces craters of about 150–200 mm in diameter at a power of about 150 mJ. By inserting frequency doubling crystals into the path of the IR laser beam we can modify the laser light to operate either in green Žat 532 nm. or in ultra-violet ŽUV, at 266 nm.. IR radiation is absorbed strongly by ferromagnesian minerals Žwhere Fe 2q is present. but not by colourless minerals or glass Žsee Jeffries et al., 1995., whilst ultra-violet radiation is strongly absorbed by most materials including glass Žsee Perkins et al., 1997.. The IR laser removes material by erosion of N.J.G. Pearce et al.r Global and Planetary Change 21 (1999) 151–171 the sample by a plasma formed at the point of focus of the laser at sample surface. This is termed plasma erosion. In contrast, the strong absorption of the UV laser light directly ablates material from the sample surface by a process termed photo-ablation. During LA-ICP-MS, whether IR or UV, the sample is placed in a fused silica chamber Žtransparent to both IR and UV. which is flushed with approximately 1.1 L miny1 argon and any ablated material is carried in the argon stream into the core of the plasma. 2.3. Sample preparation Each tephra sample was sieved to obtain the coarsest size fraction. These grains were then washed in 1% HF for 20 s and rinsed thoroughly in water. These grains were transferred to a beaker, immersed in water and cleaned with an ultrasonic probe for 20 s. Glass shards were separated magnetically from the dried sample, the purity of the separate monitored under transmitted or incident light microscopes. Tetrabromethane, appropriately diluted with acetone, was used on occasion to increase the purity of the separate, and in such cases, the final glass separate was thoroughly washed in acetone. 2.4. Electron probe micro-analysis The major element composition of all glass shards referred to in this study was determined by wavelength dispersive spectrometry methods on a Cameca SX-50 microprobe in the Department of Geology, University of Toronto. Shards were mounted in resin blocks and polished prior to analyses. Concentrations were produced using the ‘PAP’ procedure for quantitative analysis, which gives combined corrections for absorption and atomic number effects ŽPouchou and Pichoir, 1985.. Calibration was achieved for various elements as follows: Si, Al, Na, K—obsidian standard, University of Alberta; Fe, Mg—pyrope; Ca, Mn—bustamite; Ti—synthetic TiAl-pyroxene glass; Cl—tugtupite. A defocused beam was used Ž15 mm diameter. accelerated to 15 kV with a 6 nA current to minimise volatilisation loss and migration of alkali metals within the glass. Elements were analysed in the following order simultaneously: Spectrometer 153 1 ŽTAP. —Na, Si, Mg, Al; Spectrometer 2 ŽLiF. —Fe, Mn; Spectrometer 3 ŽPET. —K, Ti, Cl, Ca. Detection limits are about 0.03 wt.% oxide Ž3 sigma.. To remove the effects of post depositional hydration of shards, all analyses were normalised to an anhydrous basis. 3. Calibration Fully quantitative calibration of the LA-ICP-MS system is achieved by erecting a calibration curve using the peak area Žarea counts per second, ACPS. from analyte ACPSrinternal standard ACPS against concentration in reference materials for each element of interest. This requires the knowledge of one element present in the sample for internal standardisation Žsee Perkins and Pearce, 1995.. Westgate et al. Ž1994. used Ce as an internal standard to prove the viability of this method for the analysis of glass separates from tephra deposits. EPMA of glass shards is routine in tephrochronological studies, and data thus obtained may be used for internal standardisation. Minor isotopes of some of the major elements in the glass shards Že.g., 44 Ca, 29 Si, 57 Fe. produce analyte peaks during LA-ICP-MS analysis small enough not to saturate of the pulse counting detector of the ICP-MS and these elements can be used as internal standards. This has been described by Pearce et al. Ž1996., and significantly advances the studies of Westgate et al. Ž1994. by removing the need to use another trace element technique to determine the concentration of the internal standard. It is important to maintain the same operating conditions between analysis of the calibration standard and of the unknown. This overcomes effects of differential element fractionation between standard and sample caused by heating due, in particular, to pre-ablation of the standard Žsee Jeffries et al., 1996.. Glass shards gradually absorb water from their surroundings after deposition, and this is reflected in low analytical totals in EPMA. It is normal practice in tephra studies to normalise major element compositions to a water free basis. This removes the variable effects of hydration in samples of the same material from different areas, and makes comparison easier. We have adopted the same principle for trace N.J.G. Pearce et al.r Global and Planetary Change 21 (1999) 151–171 154 element data, which are all calculated on a water-free basis Žachieved by using normalised major element data for the internal standard.. In old glass shards, which may contain up to about 10% water by weight, the adoption of an anhydrous basis can make a considerable difference in the calculated results. For comparison with other analytical methods, we have normalised solution ICP-MS or INAA data to an anhydrous basis also, using H 2 O calculated by difference from the electron probe data. glass for calibration and used trace element compositions determined only from acid digestion of multiple samples followed by solution ICP-MS at Aberystwyth. These data differ for some elements from the data compiled by Pearce et al. Ž1997.. The calibration data for NIST 612 used by Pearce et al. Ž1996. is compared with the new compilation from Pearce et al. Ž1997. in Table 1. It must be remembered that, until the NIST 612 glass is properly certified, there will always be some uncertainty in analyses calibrated against this reference material. 3.1. Reference materials We routinely use the National Institute of Science and Technology ŽNIST. 612 partially certified reference material for calibration. This is a silicate glass spiked with a range of trace elements at a nominal 50 mg gy1 , but is certified for only eight elements. Pearce et al. Ž1997. have recently compiled all published data for this reference material plus new analyses from four additional laboratories, to provide an interim working value for a wide range of elements. Pearce et al. Ž1996., in a description of bulk glass sample analysis by LA-ICP-MS, used the NIST 612 Table 1 Comparison of REE concentrations used for NIST 612 calibration standard in previous studies and a multiplication factor to apply to correct previously published data. Concentrations in ppm La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu Westgate et al. Ž1994. Compilation data from Pearce et al. Ž1997. Multiplication factor for data in Pearce et al. Ž1996. 33.9 34.7 35.3 34.9 36.5 34.9 37.0 39.1 36.8 39.4 38.7 39.0 42.7 40.4 35.77"2.15 38.35"1.64 37.16"0.93 35.24"2.44 36.72"2.63 34.44"1.59 36.95"1.08 35.92"2.68 35.97"0.82 37.87"1.09 37.43"1.50 37.55"1.25 39.95"2.86 37.71"1.95 1.0552 1.1052 1.0527 1.0097 1.0060 0.9868 0.9986 0.9187 0.9774 0.9612 0.9672 0.9628 0.9356 0.9334 4. Bulk sample analysis Analyses of bulk samples were performed using the IR, large spot configuration of the laser ablation accessory. Westgate et al. Ž1994. described sample preparation methods with picked and cleaned glass shards being mounted as small cones of about 0.01 g weight on card, bonded with cyano-acrylate glue. Samples were ablated for approximately 15 s prior to the acquisition of a spectrum with the laser focused about 5 mm below the sample surface to increase the signal ŽAbell, 1991.. The spectrometer is swept across the massrcharge Ž mrz . range from 6 ŽLi. to 238 ŽU. 100 times in 66 s during a single acquisition, with sections of the mrz range being skipped to avoid peaks for abundant isotopes of the major elements. Typically about 80 mg of material is removed during one acquisition. A full description of instrumental and analytical conditions for bulk analysis is given by Westgate et al. Ž1994., and Pearce et al. Ž1996. describe the calibration strategy using a minor Ca or Si isotope, presenting a wide range of comparative data. Analyses can be calibrated either ‘fully quantitatively’ or by a single internal standard multi-element method Žso called ‘semi-quantitative’ analysis.. In fully quantitative analysis a calibration curve is erected for each element individually and the spectrum from the unknown compared directly with this. In so called ‘semi-quantitative’ analysis, an instrument response curve covering the entire elemental mass range is generated, which, after some data manipulation, produces analytical data for all ele- N.J.G. Pearce et al.r Global and Planetary Change 21 (1999) 151–171 ments in the unknown using this single calibration Žsee Pearce, 1991; Pearce et al., 1992; Pearce et al., 1996; Perkins and Pearce, 1995 for fuller descriptions of the methodology.. In both methods the concentration of an internal standard in the unknown is required. In quantitative analysis the internal standard must be present in the reference material, whilst semi-quantitative analyses can be performed using a reference material which does not contain the internal standard element. Here, we present some new bulk analyses from samples calibrated against the most recent NIST 612 concentrations ŽPearce et al., 1997., and reproduce one analysis from Pearce et al. Ž1996. corrected to the updated NIST 612 concentrations. The new concentrations used for the calibration standard will naturally change all data presented in Pearce et al. Ž1996. but these can be converted simply by using the factor given in Table 1. Here, for bulk analytical methods, we will only consider the rare earth elements ŽREE.. Table 2 shows fully quantitative analyses from a range of glass separates from tephra deposits. All analyses were performed using 44 Ca derived from EPMA data as an internal standard. The Lost Chicken tephra ŽUA771 and UT86. is of late Pliocene age and is probably derived from a vent in the Wrangell Mountains, AK. UT912 is from a series of Upper Cretaceous tephras from the Kanguk Formation, Banks Island, Canada has been described and analysed by Kemp Ž1993.. UT778 is a proximal sample of the Toba tephra from Sumatra, which has been analysed by both solution and laser ablation ICP-MS. This tephra deposit was produced by an eruption centred on Lake Toba, northern Sumatra at 75,000 years B.P. and produced the largest eruption documented of Quaternary age Žsee Ninkovitch et al., 1978.. The LA-ICP-MS data in Table 2 compare exceptionally well with the data derived by solution ICPMS or INAA for the same bulk samples with the exception of the INAA data for UT86 Žfrom Westgate et al., 1985. which is systematically higher than both the solution or laser ablation ICP-MS data for the same sample. In general the LA-ICP-MS data fall within "5–10% of the solution ICP-MS or INAA data. Pearce et al. Ž1996. noted a systematic drift in the REE data for many of their analyses and suggested that this may be due in part to the concentration used for these elements in the standard. The use of the new concentration data for NIST 612 has removed this systematic error in their results. Repeat analyses on bulk samples give a precision of about "5% Žsee Pearce et al., 1996.. Table 2 Fully quantitative LA-ICP-MS analyses for glass separates from tephra deposits. 612 glass. All concentrations in ppm La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu 155 44 Ca used as internal standard. Calibration against NIST UT771 LAICP-MS Ž n s 3. UT771 solution ICP-MS UT86 LAICP-MS Ž n s 1. UT86 solution ICP-MS UT86 INAA UT912 LAICP-MS Ž n s 1. UT912 INAA UT778 LAICP-MS Ž n s 2. UT778 solution ICP-MS 12.46 25.39 2.84 10.60 2.04 0.38 1.82 0.29 1.87 0.42 1.29 0.21 1.43 0.23 11.95 22.95 2.8 9.74 1.87 0.41 2 0.3 1.88 0.39 1.18 0.19 1.41 0.2 11.71 23.50 2.40 9.16 2.02 0.37 1.73 0.28 1.86 0.39 1.11 0.19 1.24 0.22 11.3 22.34 2.66 9.69 1.71 0.38 1.89 0.28 1.77 0.36 1.1 0.17 1.34 0.19 14.95 31.8 27.60 54.49 5.82 19.71 3.10 0.35 2.41 0.38 2.73 0.58 1.89 0.31 2.34 0.39 35.5 53.28 23.22 45.89 5.01 18.37 4.16 0.38 4.08 0.70 5.02 1.08 3.28 0.54 4.24 0.72 23.29 48.64 5.52 18.84 4.54 0.35 4.64 0.89 5.22 1.15 3.75 0.65 4.42 0.76 2.66 0.51 0.43 1.74 0.31 20.6 3.6 0.31 0.41 2.33 0.39 N.J.G. Pearce et al.r Global and Planetary Change 21 (1999) 151–171 156 Table 3 Comparison of ‘semi-quantitative’ LA-ICP-MS data with solution nebulisation ICP-MS for the deposit of Santorini tephra at Golhisar, ¨ Turkey. 57 Fe used as internal standard. Concentrations in ppm La Ce Pr Nd Sm Eu Gd Tb Dy Ho Er Tm Yb Lu GHB3 ) 150 mm LA-ICP-MS Ž n s 3. GHB3 ) 150 mm solution ICP-MS GHB3 85–150 mm LA-ICP-MS Ž n s 3. GHB3 85–150 mm solution ICP-MS GHB3 - 85 mm LA-ICP-MS Ž n s 3. GHB3 - 85 mm solution ICP-MS 29.02 68.74 7.74 29.93 6.27 1.13 7.43 1.02 7.74 1.65 4.71 0.84 5.72 0.84 28.82 55.06 6.55 25.80 5.71 0.97 6.55 1.05 6.89 1.58 4.79 0.74 5.09 0.77 28.94 68.60 7.90 28.97 6.19 1.13 7.27 1.09 7.77 1.62 5.38 0.82 5.45 0.85 28.95 55.33 6.74 25.20 5.44 0.99 6.60 1.06 7.03 1.53 4.58 0.71 5.12 0.82 24.74 60.39 7.10 29.15 6.15 1.07 6.32 1.07 7.59 1.67 5.00 0.79 5.18 0.77 30.06 57.16 6.89 26.16 5.81 1.12 6.68 1.06 6.92 1.60 4.69 0.77 5.20 0.82 Table 3 shows laser ablation ICP-MS analyses compared with solution ICP-MS data for the Santorini ash at Golhisar, Turkey Žsee Eastwood et al. ¨ 1997-this volume.. For these samples the LA-ICPMS data have been produced using a ‘semi-quantitative’ method with the minor 57 Fe isotope ŽFe determined by EPMA. being used as an internal standard. Fe is present at only 51 ppm in the NIST 612 glass, and the minor 57 Fe peak derived from this low concentration is of no use for fully quantitative analysis. However FeO is present in the Golhisar ¨ glass shards at 1.99 wt.%, giving a sizeable signal for the minor 57 Fe isotope. The solution ICP-MS data of Eastwood et al. Ž1999-this volume. has been recalculated to an anhydrous basis with 1.47% H 2 O calculated by difference. In general a good correlation between the two sets of analyses is observed, the LA-ICP-MS data falling within "10% of the solution ICP-MS data. Slight differences exist between the two data sets with two of the three LA-ICP-MS determinations showing slightly higher concentrations than the solution data. However, the agreement between the two data sets is good enough to enable correlation of the Golhisar deposits with other pub¨ lished data cited by Eastwood et al. Ž1999-this volume.. Sample preparation for LA-ICP-MS takes a matter of 1 to 2 min, and analyses can be performed for up to 60 individual elements at the rate of about 20rh. For bulk analysis, only about 80 mg of material is consumed. When compared with the need to separate between 0.1 g and 0.5 g of glass for solution ICP-MS Žrequiring lengthy acid digestion., about 0.25 g for INAA Ža slow analytical process. or ) 1 g for XRF Žlimited in the range of elements which can be detected., it is clear that LA-ICP-MS is an excellent method for the analysis of these small samples. In addition, the quality of the semi-quantitative LA-ICP-MS data is highly acceptable, and the quantitative data, which takes only slightly more off-line data processing, is excellent. 5. Single shard analysis Tephra studies have concentrated on grain-specific techniques for discrimination and correlation of individual deposits, and EPMA has been applied very successfully to this branch of the science Že.g., Smith and Westgate, 1969.. EPMA is a microbeam technique, but is limited by relatively poor sensitivity. EPMA can provide substantial information on magmatic evolution during an eruption Žin terms of major element variation., but the amount of trace element data obtained can be very limited. Many tephra deposits produced from volcanoes erupting N.J.G. Pearce et al.r Global and Planetary Change 21 (1999) 151–171 evolved magmas may be extremely similar in major element composition. This is especially true of tephra beds derived from the same volcano and thus individual deposits may be difficult to identify unequivocally on the basis of major element chemistry alone Že.g., Vreeken et al., 1992, who could only distinguish tephra deposits from Saskatchewan by age determinations and not major element chemistry.. In these cases, it may be necessary to resort to trace element analyses to distinguish individual tephra deposits. Throughout the development of laser ablation techniques for the analysis of glasses from tephra deposits, our aim has been to develop analytical techniques which will allow individual shards to be analysed for a wide range of petrogenetically useful trace elements. This has required certain modifications to the analytical instrumentation we have used to enable the necessary sensitivity to be achieved, and these are described below. 5.1. Instrumental modifications for single shard analysis Analysis of small, individual glass shards has required the modification of the IR laser to UV, giving a maximum output of 3 mJ of UV at 266 nm. The shorter wavelength of the UV, and the use of a reflecting microscope objective coated specifically to reflect at 266 nm, enable a finer focus of the laser to be achieved. This gives craters in glasses of between 5 mm Žat about 0.05 mJ. and about 50 mm Žat 3 mJ, see Jeffries et al., 1995.. The dramatic reduction in signal obtained from these small craters when compared with the large Ž150–200 mm diameter. craters in IR bulk analysis necessitates increasing the sensitivity of the ICP-MS itself. This is achieved by pumping the first stage of the vacuum system at a considerably increased rate, literally ‘sucking’ more of the plasma Žand thus more of the sample. into the spectrometer. The analytical rationale for this ultrahigh sensitivity ŽUHS. mode is no different from that employed in normal operational mode, although it is necessary to re-optimise the instrument for the new vacuum conditions. This UHS mode increases the sensitivity for some elements by up to an order of magnitude, and largely compensates for the loss in signal from the much smaller ablation craters produced by the small spot UV laser. We are currently 157 investigating the effects that varying instrumental conditions and geometries have upon the relative change in sensitivity, and may be able to improve on this order of magnitude increase in sensitivity already achieved with the UHS modification. 5.2. Sample preparation and analysis Cleaned and picked glass shards are mounted in resin blocks and polished for electron probe analysis. EPMA is performed on the samples with a record being made Žusually on a Polaroid photograph. of which shards are analysed. The EPMA provides major element data which can then be used as an internal standard for subsequent UVLA-ICP-MS analysis. For UVLA-ICP-MS analysis, the resin blocks are placed in the ablation chamber of the laser system, and using the annotated Polaroid from the EPMA analysis, the same shards can be relocated. No further sample preparation is required. Laser operating conditions may vary from day-to-day dependent upon the size of material being ablated and the range of elements of interest. The limiting factor in the analysis however is the length of time the shard will survive being ablated: thin walled shards obviously will not last as long as thicker shards at the same laser power. The laser power can thus be varied to maximise acquisition time. In addition the scan parameters of the quadrupole mass spectrometer can be adjusted to allow the maximum time per analyte peak, with the spectrometer being ‘jumped’ between masses Žor mass regions. of interest. By adjusting these parameters, analyses of individual shards can be performed to give the optimum sensitivity. Fig. 1 shows the Periodic Table with the range of elements typically determined by UVLA-ICP-MS in rhyolitic shards. The acquisition of a spectrum using this element menu takes about 20 s. Fig. 2 shows the theoretical lower limits of detection for a range of elements determined at different laser flash lamp voltages using the element menu given in Fig. 1. A laser flash lamp voltage of about 650 V corresponds to about 0.1 mJ of power giving a crater of about 5 mm diameter, at 950 V craters of about 50 mm diameter are produced from about 3 mJ of laser light. We would normally operate at about 800 V, which provides about 1.5 mJ of power and gives craters of around 20 mm diameter. This gives a 158 N.J.G. Pearce et al.r Global and Planetary Change 21 (1999) 151–171 Fig. 1. Periodic Table showing the elements routinely determined during the analysis of single glass shards Žthose in bold italics . alongside those which could potentially be analysed Žin normal typeface.. The underlined elements can be used as internal standards if included in the analysis. strong signal whilst allowing a reasonable length of ablation before the shard is destroyed. The shards are positioned under the laser beam on an x–y motorised stage which can be moved in either 50 mm or 2.5 mm steps. During an acquisition, the sample can be moved Žin 2.5 mm steps. under the laser beam to increase the amount of material which is ablated, thus maximising the signal. Calibration spectra, obtained from the NIST 612 glass were obtained under exactly the same conditions as used to analyse individual shards. 6. Montana tephras—an example of single shard analysis Fig. 2. Lower limit of detection ŽLLD. for UVLA-ICP-MS at varying laser voltages, using modified vacuum interface Žultra-high sensitivity mode.. LLDs calculated from NIST 612 glass at the same operating conditions as analysis of shards. Crater diameters are approximately 5 mm at 650 V, 20 mm at 800 V, 40 mm at 850 V and 50 mm at 950 V. The Six Mile Creek Formation in the Ruby Range of Southwest Montana, USA, contains a series of tephra deposits of Miocene age ŽFritz and Sears, 1993.. Three tephra deposits ŽBT3, MSD2 and WR1. have been sampled from the basal part of the Andersen Ranch Member. The relative age of these three tephra beds in unknown although each is compositionally distinct Žsee Fig. 3.. Glass shards were separated, mounted and analysed by EPMA in Toronto Žsee above.. The exact position of each EPMA analysis was recorded on back-scattered electron ŽBSE. image. The same shards were then analysed by UVLA-ICP-MS at Aberystwyth, relocating each shard from the BSE photograph of the sample and using the EPMA data for internal standardisation. In this instance we used a laser flash lamp voltage of 850 V and calibrated against the NIST 612 glass. 29 Si was used as the internal standard, and the N.J.G. Pearce et al.r Global and Planetary Change 21 (1999) 151–171 159 instrument was recalibrated prior to the analysis of each sample. Major and trace element data for all anlaysed shards from the three samples are presented in Table 4. 6.1. Results and discussion Figs. 3–8 show a variety of the analytical data derived from individual shards in the three tephra beds. Immediately obvious is that, instead of the single data point for each trace element produced by a bulk analysis technique, data plot in fields across a range of trace element compositions. The reproducibility of the LA-ICP-MS technique is typically "5% on repeat analyses of the same, homogeneous material Žsuch as the calibration standard, see Westgate et al., 1994; Pearce et al., 1996, 1997.. Clearly the fields of data from the individual shards within one sample spread across a considerably wider range than "5% of the average Žbulk sample. value. These fields thus represent real within-sample, inter-shard variation in trace element composition. It has not been possible until this study to record the extent of this variation within a single sample. Fig. 3 shows CaO vs. SiO 2 , MgO vs. SiO 2 and FeOt vs. CaO from the EPMA analyses of these shards. The overall trend of the data is consistent with a co-magmatic origin for these samples, with evolution by fractional crystallisation from a common parent. These graphs show fields of data occupied by the individual samples, but none is separated clearly from each other. MSD2 shows two discrete populations of shards from the major element data, clearly picked out in the plot of CaO vs. FeO with one group having approximately 3 wt.% FeO, the other around 2.3 wt.% FeO. For all samples, the general trend of the data within and between samples, is consistent with magmatic evolution by fractionation of alkali feldspar and a ferromagnesian mineral Žcausing both CaO and MgO to decrease with increasing SiO 2 .. Fig. 3. EPMA major element data Žas wt.%, normalised to anhydrous analysis. from Miocene glass shards from the Ruby Range, Montana. Ža. SiO 2 vs. CaO, Žb. SiO 2 vs. MgO, Žc. FeOt vs. CaO. N.J.G. Pearce et al.r Global and Planetary Change 21 (1999) 151–171 160 Table 4 Major and trace element data for individual shards from BT3, MSD2 and WR1, Six Mile Creek Formation, Ruby Range, Montana. Major elements by EPMA, trace elements by UVLA-ICP-MS. All analyses normalised to an anhydrous basis, based on H 2 O by difference from the EPMA analyses BT3-1-01 BT3-1-02 BT3-1-03 BT3-1-04 BT3-1-05 BT3-1-06 BT3-1-07 BT3-1-08 BT3-1-09 BT3-1-10 Major elements, wt.% SiO 2 76.801 TiO 2 0.304 Al 2 O 3 11.856 FeOt 1.805 MnO 0.072 MgO 0.033 CaO 0.601 Na 2 O 2.927 K 2O 5.444 F 0.051 Cl 0.201 O-F, Cl y0.095 Total 100.000 H 2 O by 5.539 difference Trace elements ppm Isotope Li 7 26.6 B 11 26.0 Rb 85 91.2 Sr 88 32.4 Y 89 39.6 Zr 90 256 Nb 93 49.8 Cs 133 7.27 Ba 138 612 La 139 49.0 Ce 140 105 Pr 141 12.1 Nd 143 41.1 Eu 151 1.99 Sm 152 9.0 Eu 153 1.44 Gd 158 14.5 Tb 159 1.73 Dy 161 12.0 Ho 165 2.86 Er 166 7.22 Tm 169 2.21 Yb 172 6.85 Lu 175 1.60 Hf 178 9.41 Ta 181 5.38 Pb 208 64.4 Th 232 25.3 U 238 9.7 77.187 0.221 11.846 1.786 0.086 0.070 0.620 2.837 5.314 0.039 0.000 y0.008 100.000 5.652 76.239 0.270 11.862 2.124 0.082 0.125 0.698 3.234 5.274 0.051 0.089 y0.048 100.000 5.937 77.162 0.187 11.699 1.853 0.090 0.051 0.663 3.033 5.232 0.039 0.000 y0.008 100.000 5.210 76.262 0.382 12.135 2.161 0.026 0.088 0.842 3.251 4.716 0.058 0.156 y0.077 100.000 5.638 76.222 0.370 11.965 2.078 0.074 0.111 0.829 2.980 5.227 0.034 0.202 y0.092 100.000 5.785 76.668 0.223 11.895 2.161 0.058 0.118 0.784 2.979 4.959 0.049 0.201 y0.094 100.000 5.892 76.514 0.249 11.932 2.027 0.014 0.089 0.743 3.137 5.212 0.040 0.088 y0.045 100.000 5.048 76.166 0.397 12.089 2.314 0.000 0.079 0.791 3.133 4.976 0.035 0.044 y0.026 100.000 5.563 76.414 0.254 12.004 2.126 0.067 0.078 0.709 3.192 5.123 0.043 0.000 y0.009 100.000 5.914 37.6 42.5 146.9 31.0 51.4 301 46.3 10.84 727 65.7 125 15.4 58.8 2.15 10.8 3.30 15.4 3.22 13.3 3.48 9.71 3.11 7.70 2.06 9.35 6.21 67.6 33.1 11.1 44.4 49.7 90.5 38.2 49.5 277 55.0 10.45 631 54.8 114 16.1 50.4 2.03 15.3 3.09 19.8 3.82 15.5 4.36 10.51 3.84 9.87 2.04 15.37 6.21 78.1 30.1 8.8 40.0 35.4 161.3 15.1 31.2 155 39.1 9.32 351 42.9 97 12.5 39.1 2.19 10.7 1.68 11.6 2.98 12.6 3.92 5.73 2.46 7.84 1.62 7.79 4.84 60.2 21.3 14.4 35.1 40.3 119.3 21.4 50.0 166 48.2 9.68 452 55.8 126 15.6 54.8 1.43 13.2 1.57 14.6 3.17 12.2 3.84 6.76 3.01 10.97 1.49 9.90 5.75 93.2 30.0 14.6 42.5 50.3 87.8 34.5 45.4 266 49.6 10.28 701 57.4 121 15.1 61.5 2.50 14.0 2.92 16.7 3.23 13.6 5.41 7.97 3.32 11.15 1.94 15.24 7.02 46.0 29.0 13.4 36.1 28.1 103.6 38.9 44.2 242 51.6 9.40 665 56.0 134 15.7 49.1 2.44 14.4 3.68 13.6 2.24 14.3 2.99 7.94 2.73 8.93 1.29 10.97 5.50 57.7 29.2 12.5 41.7 41.3 119.6 39.4 48.5 263 45.4 9.45 684 59.8 127 16.4 53.2 2.10 13.2 2.77 17.6 3.03 11.8 4.84 10.65 3.09 10.08 2.03 15.06 5.63 67.0 34.2 15.1 45.8 57.7 126.5 36.6 46.4 252 45.8 10.83 621 51.0 111 16.1 41.7 3.38 16.0 3.30 16.4 3.33 12.0 4.30 7.89 4.02 6.30 1.65 12.38 6.32 50.4 26.7 11.5 33.4 36.9 119.0 33.4 48.3 251 45.9 8.91 667 56.2 124 16.0 55.6 2.74 9.9 2.72 16.4 3.63 14.2 3.68 9.26 2.88 9.02 1.57 10.10 5.94 46.1 33.2 12.5 N.J.G. Pearce et al.r Global and Planetary Change 21 (1999) 151–171 161 BT3-2-01 BT3-2-02 BT3-2-03 BT3-2-04 BT3-2-05 BT3-2-06 BT3-2-07 BT3-2-08 BT3-2-09 BT3-2-10 76.390 0.380 11.971 2.093 0.105 0.135 0.767 2.973 5.071 0.013 0.181 y0.079 100.000 6.432 77.390 0.193 11.702 1.854 0.052 0.076 0.618 2.850 5.158 0.053 0.113 y0.058 100.000 5.783 77.016 0.232 11.933 1.823 0.062 0.046 0.592 2.944 5.330 0.028 0.000 y0.006 100.000 6.416 77.107 0.268 11.715 1.674 0.000 0.044 0.622 2.911 5.510 0.023 0.226 y0.100 100.000 5.997 77.215 0.182 12.046 1.647 0.015 0.056 0.533 2.994 5.259 0.066 0.000 y0.013 100.000 5.805 76.405 0.398 11.898 2.019 0.108 0.118 0.796 3.027 5.172 0.041 0.045 y0.027 100.000 5.635 76.330 0.351 12.093 2.065 0.022 0.116 0.745 3.033 5.123 0.040 0.157 y0.074 100.000 5.542 76.101 0.243 12.011 2.134 0.144 0.078 0.815 3.033 5.193 0.048 0.361 y0.162 100.000 6.726 76.412 0.180 11.990 2.231 0.103 0.107 0.691 2.988 5.107 0.046 0.267 y0.122 100.000 5.666 76.602 0.329 11.799 2.068 0.000 0.058 0.729 2.921 5.429 0.000 0.112 y0.047 100.000 5.582 41.5 47.5 92.7 41.7 58.5 292 61.8 8.09 784 70.1 132 18.7 66.7 4.30 15.9 2.54 18.4 2.39 16.4 4.09 12.29 3.77 13.26 2.40 12.00 6.25 67.0 45.7 13.7 49.9 64.1 95.0 37.5 46.4 255 40.0 12.47 698 64.5 134 17.6 65.0 3.38 19.8 3.44 17.6 4.21 13.0 5.03 8.84 3.73 12.90 3.07 17.42 7.38 69.7 35.2 13.7 36.5 46.5 116.2 33.3 53.6 221 40.7 9.24 650 60.6 121 17.5 43.2 3.27 14.2 2.79 15.6 3.22 15.8 5.23 10.25 3.46 6.38 2.07 8.81 4.72 39.9 34.2 12.6 38.0 46.0 119.5 40.3 49.1 268 54.7 9.21 743 62.5 139 16.5 61.5 3.93 15.6 3.49 15.0 4.12 19.8 5.30 9.34 3.46 12.70 2.36 16.83 7.41 77.6 36.1 15.2 48.4 58.7 98.0 52.0 52.6 281 61.2 9.61 815 69.0 142 18.9 53.9 3.64 18.4 5.49 19.4 3.96 17.9 4.58 15.01 6.06 14.98 3.73 14.04 9.30 81.0 44.7 17.8 45.9 47.4 99.2 39.3 55.5 315 55.5 9.47 752 70.5 141 20.8 67.2 2.15 19.1 3.69 18.1 4.21 16.5 4.33 14.17 4.75 11.97 2.23 12.98 7.34 72.8 46.9 22.4 51.1 50.7 133.3 46.2 55.7 264 53.8 11.18 722 63.5 146 20.4 60.5 4.18 19.3 2.44 19.4 3.64 17.0 5.24 12.12 4.80 11.29 3.66 16.94 7.52 75.3 42.5 18.6 44.6 52.7 127.2 43.6 66.1 327 52.4 10.34 931 84.2 126 23.4 76.1 3.64 15.4 4.57 16.4 3.78 28.1 6.61 12.50 5.07 12.67 2.44 20.03 9.75 94.4 57.4 12.7 39.9 46.5 111.9 35.0 39.6 227 55.4 11.19 736 58.7 127 17.8 48.9 2.64 14.2 2.54 18.6 4.50 17.4 6.88 12.30 4.03 7.11 2.88 17.16 7.49 70.8 41.6 13.9 24.3 29.5 98.3 55.7 58.4 335 57.6 7.17 1164 77.1 160 18.4 64.3 1.96 18.1 2.16 20.8 3.69 21.0 5.12 9.27 3.36 11.80 2.78 13.22 8.24 115.4 76.0 11.0 (continued on next page) N.J.G. Pearce et al.r Global and Planetary Change 21 (1999) 151–171 162 Table 4 Ž continued . MSD21-01 Major elements, wt.% SiO 2 76.270 TiO 2 0.195 Al 2 O 3 12.031 FeOt 2.304 MnO 0.041 MgO 0.074 CaO 0.835 Na 2 O 3.136 K 2O 5.096 F 0.021 Cl 0.000 O-F, Cl y0.004 Total 100.000 H 2 O by 5.987 difference Trace elements ppm Isotope Li 7 37.5 B 11 30.8 Rb 85 102.3 Sr 88 35.5 Y 89 50.7 Zr 90 213 Nb 93 43.8 Cs 133 8.39 Ba 138 530 La 139 42.1 Ce 140 105 Pr 141 14.3 Nd 143 39.9 Eu 151 2.74 Sm 152 9.3 Eu 153 1.89 Gd 158 10.1 Tb 159 1.94 Dy 161 11.2 Ho 165 1.84 Er 166 5.36 Tm 169 1.67 Yb 172 6.88 Lu 175 1.49 Hf 178 6.54 Ta 181 4.48 Pb 208 18.2 Th 232 14.3 U 238 6.13 MSD21-02 MSD21-03 MSD21-04 MSD21-05 MSD21-06 MSD21-07 MSD21-08 MSD21-09 MSD21-10 76.502 0.311 11.846 2.153 0.015 0.067 0.791 3.054 5.221 0.034 0.022 y0.016 100.000 6.191 74.994 0.347 12.128 2.955 0.064 0.164 1.244 3.133 4.830 0.016 0.222 y0.097 100.000 5.822 74.723 0.399 12.365 3.003 0.086 0.147 1.182 3.171 4.801 0.073 0.111 y0.061 100.000 5.163 75.309 0.408 12.153 2.849 0.017 0.111 1.066 3.282 4.784 0.027 0.000 y0.005 100.000 6.262 75.036 0.362 12.039 3.044 0.086 0.178 1.180 3.170 4.846 0.026 0.067 y0.033 100.000 5.935 76.293 0.305 11.983 2.232 0.038 0.065 0.846 2.974 5.120 0.064 0.157 y0.079 100.000 5.943 74.695 0.422 12.229 2.881 0.163 0.142 1.270 3.194 4.953 0.064 0.000 y0.013 100.000 5.560 75.874 0.383 11.887 2.537 0.111 0.135 0.913 3.026 5.013 0.039 0.155 y0.073 100.000 5.423 76.308 0.264 11.843 2.354 0.060 0.134 0.795 3.121 4.997 0.041 0.156 y0.074 100.000 5.963 33.8 25.4 97.0 54.9 61.2 345 51.2 6.90 777 69.5 145 17.4 53.8 2.21 9.9 2.16 16.3 2.52 12.4 2.97 8.73 2.02 11.83 1.07 9.94 3.52 17.1 21.5 4.82 21.8 14.8 89.1 74.9 56.9 420 45.7 5.08 756 59.4 118 14.3 50.0 2.06 10.0 2.31 13.0 2.01 7.2 2.37 6.38 1.29 7.40 1.12 9.89 3.73 19.7 21.0 3.90 19.1 15.8 96.1 58.0 42.3 291 39.3 4.45 608 44.7 100 12.3 36.5 2.09 8.2 1.68 11.1 1.32 7.9 1.86 3.95 1.54 4.61 0.77 6.22 2.38 16.8 14.6 4.62 29.9 25.0 156.0 66.1 56.7 392 50.9 5.77 717 59.8 125 13.4 45.3 2.60 11.8 1.90 10.3 2.07 10.7 2.28 5.59 1.34 8.40 1.62 8.33 3.46 23.0 19.6 5.23 25.2 14.1 88.4 70.5 61.9 419 49.4 5.95 754 60.4 123 16.6 54.9 2.31 11.1 2.47 14.7 2.47 10.6 3.04 6.21 1.89 6.43 1.52 8.90 3.43 19.2 20.7 4.66 18.8 12.4 91.2 46.4 52.4 301 53.1 4.53 669 55.7 107 12.4 42.8 2.25 8.2 1.48 11.3 1.50 8.2 1.50 5.55 1.27 6.90 1.38 6.71 3.49 24.1 21.3 5.53 45.2 38.8 95.8 64.5 47.1 327 53.6 8.33 592 54.6 104 17.6 38.1 3.57 10.2 3.25 10.6 2.38 10.2 2.50 7.83 2.72 7.50 1.10 9.93 3.94 23.7 18.0 6.01 22.0 16.5 121.4 55.3 50.1 326 58.4 4.60 744 57.3 120 14.4 46.3 1.93 8.0 2.92 12.2 2.02 10.1 2.47 4.79 1.18 7.09 1.15 9.81 3.68 20.5 19.4 6.01 16.5 7.6 88.9 38.9 44.1 260 39.0 3.78 641 50.9 107 12.4 29.4 1.52 6.8 1.90 11.5 1.32 7.2 1.77 4.54 1.14 5.68 0.92 5.55 3.07 16.7 17.6 4.98 Fig. 4a and b show CaO determined by EPMA plotted against Zr and Sr determined by UVLA-ICP- MS. Sr decreases sympathetically with CaO. The Sr data is consistent with evolution by the extraction of N.J.G. Pearce et al.r Global and Planetary Change 21 (1999) 151–171 MSD21-11 MSD21-12 MSD22-01 MSD22-02 MSD22-03 MSD22-04 MSD22-05 MSD22-06 MSD22-07 163 MSD22-08 MSD22-09 MSD22-10 77.072 0.323 11.733 1.990 0.052 0.078 0.630 2.806 5.315 0.000 0.000 0.000 100.000 6.127 76.189 0.238 11.866 2.351 0.085 0.118 0.822 3.197 5.034 0.046 0.110 y0.056 100.000 4.662 76.197 0.301 11.768 2.399 0.066 0.083 0.752 3.086 5.234 0.045 0.134 y0.066 100.000 6.215 76.231 0.271 11.890 2.321 0.101 0.102 0.811 3.077 4.970 0.073 0.289 y0.136 100.000 5.647 76.164 0.346 12.040 2.313 0.058 0.106 0.852 2.939 5.013 0.032 0.247 y0.110 100.000 6.072 76.825 0.263 11.881 1.892 0.000 0.062 0.633 3.100 5.261 0.055 0.067 y0.039 100.000 6.115 75.152 0.341 12.150 2.875 0.054 0.138 1.161 3.301 4.690 0.015 0.222 y0.097 100.000 5.986 76.036 0.303 12.027 2.379 0.120 0.114 0.859 3.085 5.031 0.057 0.000 y0.011 100.000 5.217 76.483 0.222 11.724 2.395 0.000 0.079 0.755 3.123 5.153 0.034 0.067 y0.035 100.000 5.508 74.570 0.371 12.424 3.020 0.055 0.172 1.139 3.194 4.927 0.048 0.156 y0.075 100.000 5.510 76.917 0.200 11.670 2.106 0.042 0.069 0.725 2.867 5.319 0.025 0.112 y0.052 100.000 5.632 76.326 0.349 11.822 2.385 0.041 0.112 0.873 3.167 4.809 0.048 0.133 y0.066 100.000 5.461 23.5 18.5 117.2 37.4 45.2 220 45.4 5.22 673 53.6 120 13.5 39.9 1.82 8.7 1.30 11.1 1.57 7.8 1.82 6.21 1.50 5.33 1.49 5.83 3.51 30.8 20.3 5.51 21.3 17.8 106.8 36.0 43.3 236 52.9 5.50 632 51.8 116 12.6 32.5 1.58 8.0 1.59 13.0 1.90 7.9 2.19 4.35 1.23 7.17 0.91 6.58 3.54 17.3 18.7 5.84 29.0 21.5 99.1 66.9 57.9 336 51.4 5.53 764 61.3 116 15.6 45.9 2.23 10.5 2.21 13.9 2.04 10.9 3.03 8.15 1.44 7.71 1.97 9.16 3.75 22.3 24.0 4.99 58.7 34.8 116.0 55.2 57.5 289 66.2 11.71 757 66.4 137 18.1 56.9 4.04 16.6 3.16 19.7 1.99 13.6 4.82 9.10 2.16 11.35 2.27 10.14 4.25 174.2 24.1 8.84 29.4 22.1 83.0 73.5 71.4 458 52.3 5.96 799 74.0 135 18.8 54.0 2.31 12.8 2.17 17.0 3.47 14.9 3.45 7.65 1.49 7.28 1.58 11.91 5.62 19.2 28.9 4.94 24.3 16.5 110.5 34.1 46.4 225 41.4 6.53 701 55.2 117 13.1 40.2 1.78 10.3 2.13 12.4 1.95 9.7 2.69 6.63 1.63 6.73 1.03 7.15 3.18 22.4 21.4 6.04 21.5 13.1 95.2 57.5 51.9 345 43.1 4.85 767 56.1 126 13.3 43.1 2.36 10.9 3.20 12.1 2.12 9.8 2.61 7.08 1.23 6.71 1.31 9.09 3.54 21.9 23.1 7.02 168.8 101.5 199.5 44.5 42.5 244 82.4 29.82 313 40.2 107 29.3 58.9 8.72 23.4 11.04 17.9 6.89 20.2 4.69 11.35 6.98 16.56 4.14 13.10 13.40 60.5 15.7 7.80 89.6 46.8 145.4 55.1 70.7 290 53.0 15.38 647 61.8 118 22.9 61.2 4.42 18.4 3.46 20.8 4.58 19.6 4.75 9.05 3.55 13.33 2.53 14.23 6.23 314.0 24.6 5.27 29.8 20.8 90.8 69.7 66.3 408 46.6 5.84 854 63.9 121 17.2 63.9 2.42 12.6 2.07 13.4 2.79 16.0 2.91 9.24 2.65 8.03 1.55 11.17 4.54 28.2 27.4 5.60 44.5 30.4 106.4 53.1 67.6 299 54.4 8.49 746 61.8 117 16.9 59.6 3.97 13.3 2.14 15.1 2.81 13.1 2.75 7.80 2.64 9.73 1.60 13.03 5.81 27.2 26.2 6.61 25.8 19.6 126.5 48.1 50.2 276 47.4 5.61 757 58.7 129 14.2 49.9 2.19 12.1 2.09 14.9 2.39 13.2 2.65 5.37 1.73 9.13 1.49 8.27 3.75 30.4 24.3 6.41 (continued on next page) feldspar and allows the same interpretation as the major element data. Zirconium also decreases sym- pathetically with CaO, which would also imply than zircon, or another Zr-rich phase, was also being N.J.G. Pearce et al.r Global and Planetary Change 21 (1999) 151–171 164 Table 4 Ž continued . WR1-1-01 WR1-1-02 WR1-1-03 WR1-1-04 WR1-1-05 WR1-1-06 WR1-1-07 WR1-1-08 WR1-2-01 WR1-2-02 Major elements, wt.% SiO 2 77.407 TiO 2 0.181 Al 2 O 3 11.859 FeOt 1.596 MnO 0.082 MgO 0.019 CaO 0.506 Na 2 O 2.929 K 2O 5.304 F 0.047 Cl 0.135 O-F, Cl y0.066 Total 100.000 H 2 O by 6.069 difference Trace elements ppm Isotope Li 7 36.8 B 11 47.8 Rb 85 116.9 Sr 88 18.8 Y 89 43.8 Zr 90 183 Nb 93 44.7 Cs 133 10.83 Ba 138 425 La 139 57.3 Ce 140 118 Pr 141 15.2 Nd 143 48.6 Eu 151 1.30 Sm 152 11.7 Eu 153 1.24 Gd 158 19.7 Tb 159 1.93 Dy 161 11.6 Ho 165 2.31 Er 166 8.96 Tm 169 2.80 Yb 172 8.48 Lu 175 1.16 Hf 178 6.75 Ta 181 4.05 Pb 208 45.1 Th 232 20.6 U 238 9.7 77.147 0.206 11.755 1.596 0.000 0.075 0.564 3.184 5.279 0.065 0.249 y0.118 100.000 6.642 77.636 0.078 11.856 1.338 0.000 0.063 0.501 3.036 5.290 0.107 0.201 y0.106 100.000 5.679 77.524 0.128 11.819 1.454 0.040 0.079 0.500 3.050 5.219 0.054 0.247 y0.115 100.000 5.965 76.815 0.276 12.082 1.477 0.012 0.072 0.537 3.109 5.395 0.103 0.246 y0.124 100.000 5.863 77.137 0.154 11.960 1.561 0.065 0.067 0.646 2.809 5.299 0.103 0.382 y0.181 100.000 5.948 77.145 0.142 12.102 1.659 0.036 0.049 0.575 2.847 5.298 0.069 0.158 y0.080 100.000 6.099 76.992 0.200 11.981 1.624 0.010 0.050 0.629 2.827 5.509 0.061 0.226 y0.108 100.000 6.349 77.641 0.241 11.880 1.373 0.019 0.057 0.526 2.905 5.275 0.088 0.022 y0.027 100.000 6.369 77.696 0.110 11.747 1.429 0.056 0.060 0.475 2.973 5.384 0.087 0.000 y0.017 100.000 6.013 12.3 23.0 132.2 16.7 23.7 115 35.1 5.98 238 35.8 83 8.7 27.5 0.55 6.7 1.06 11.8 1.42 6.0 1.61 4.73 1.37 5.10 0.73 4.29 8.04 83.6 19.1 9.9 41.3 43.5 215.7 10.0 54.3 162 56.3 11.23 111 74.4 151 18.2 55.0 1.56 14.9 2.00 20.0 3.36 17.5 4.91 11.19 4.12 13.67 2.69 12.54 6.46 80.7 56.0 28.1 43.1 43.1 204.2 10.9 61.7 191 58.5 10.43 173 87.4 200 23.1 58.7 1.99 17.4 2.53 22.5 3.44 18.0 5.17 15.99 4.51 15.44 2.42 11.22 8.68 80.1 60.8 25.2 41.0 38.0 156.9 21.2 42.1 186 52.6 11.34 491 80.7 155 19.4 51.1 2.36 13.3 2.49 23.2 2.87 17.9 4.77 12.81 3.34 10.14 2.56 14.29 6.22 69.6 50.0 17.8 35.1 43.4 152.9 22.5 44.2 183 53.5 11.05 511 71.4 142 19.1 53.1 2.34 17.5 2.40 15.4 2.83 18.6 4.29 9.51 3.38 13.58 3.22 14.08 6.68 50.3 50.1 19.6 51.3 54.2 193.3 25.6 63.1 223 49.9 14.08 610 78.7 161 24.8 59.7 4.43 19.6 2.21 22.4 3.69 15.0 4.81 12.73 5.20 15.18 4.50 15.34 8.93 80.7 61.2 18.7 36.0 40.3 169.5 18.2 44.4 173 51.9 9.50 436 66.3 156 19.4 48.1 1.84 16.0 1.86 21.5 2.06 11.3 4.19 7.66 2.59 10.43 2.11 9.91 6.52 54.3 50.8 22.4 42.6 37.1 238.2 13.9 68.1 209 62.0 12.66 180 103.8 195 23.1 86.3 2.02 20.2 2.30 22.5 3.57 14.9 7.21 16.57 4.30 15.13 3.07 17.42 9.59 75.2 83.1 28.0 35.4 43.4 227.0 10.2 70.7 199 63.9 10.67 140 96.3 198 22.4 71.8 1.46 18.6 1.24 22.0 3.74 20.5 6.70 15.87 4.80 15.88 3.65 15.03 10.90 69.1 87.5 32.1 fractionated from the parent magma. Zircon fractionation is relatively common in evolved magmas of these general compositions. In the same way as major element EPMA data are used to discriminate between individual tephra deposits, the trace element data determined by LA-ICP- N.J.G. Pearce et al.r Global and Planetary Change 21 (1999) 151–171 WR1-2-03 WR1-2-04 WR1-2-05 WR1-2-06 WR1-2-07 77.004 0.275 12.014 1.680 0.000 0.083 0.595 3.004 5.186 0.069 0.180 y0.090 100.000 5.991 77.075 0.200 11.983 1.676 0.000 0.018 0.611 1.520 6.787 0.080 0.114 y0.064 100.000 6.353 77.038 0.106 11.937 1.226 0.056 0.035 0.507 1.646 7.399 0.062 0.000 y0.013 100.000 5.591 77.130 0.247 11.884 1.672 0.011 0.039 0.640 2.990 5.253 0.022 0.201 y0.089 100.000 5.585 76.592 0.144 12.198 1.717 0.068 0.063 0.568 2.942 5.627 0.099 0.000 y0.020 100.000 5.215 77.213 0.257 11.826 1.610 0.055 0.051 0.569 3.064 5.284 0.089 0.000 y0.018 100.000 6.016 77.241 0.170 11.881 1.663 0.012 0.063 0.579 2.827 5.381 0.083 0.201 y0.101 100.000 5.546 32.5 34.2 192.5 30.2 60.9 240 52.6 9.30 696 98.7 178 22.6 65.9 1.72 18.6 2.70 18.9 3.58 17.9 4.69 16.06 3.87 15.91 3.32 16.76 8.00 67.5 79.9 24.1 43.6 37.0 165.9 23.6 53.6 197 51.5 12.35 538 82.1 179 20.1 57.6 1.54 14.4 1.42 16.1 3.19 14.7 4.53 15.15 4.31 12.31 2.04 14.64 9.11 60.9 60.6 21.9 26.6 32.7 221.6 11.5 61.5 175 66.7 11.32 137 87.3 178 22.1 59.8 1.08 13.6 1.89 19.7 2.40 17.7 4.87 14.77 3.79 14.10 2.28 11.86 11.40 101.3 75.8 31.4 48.2 44.1 233.6 23.0 55.5 212 56.8 11.31 620 94.6 207 22.1 68.5 3.13 16.4 2.71 22.9 3.23 13.5 4.69 11.55 4.61 16.75 3.40 19.37 9.50 71.8 69.8 30.0 45.5 46.4 229.9 20.3 53.0 204 57.8 12.89 605 92.3 193 23.9 77.5 1.74 17.7 1.98 23.3 3.99 23.6 5.21 13.16 4.17 15.30 2.10 15.15 9.25 91.0 69.9 31.8 156.9 166.7 156.5 50.0 63.8 219 50.3 32.55 420 69.3 173 30.7 59.6 6.65 28.0 8.40 26.2 9.50 25.9 12.94 17.64 13.84 25.18 11.49 29.58 16.70 107.9 63.7 16.0 33.7 28.6 222.1 32.8 67.9 223 58.5 10.90 702 95.8 200 24.3 72.7 1.58 17.6 2.50 21.2 3.30 20.4 6.73 15.30 3.96 14.77 2.57 14.37 10.29 83.8 80.7 28.8 MS can also be used. Fig. 5a shows Zr vs. Nb for these three units. MSD2 and WR1 occupy distinct and almost entirely separate fields, whilst BT3 over- WR1-2-08 165 WR1-2-09 laps both others. A similar pattern is also observed from Rb vs. Sr ŽFig. 5b., with only one Žhigh Sr. shard from WR1 overlapping the field of data from 166 N.J.G. Pearce et al.r Global and Planetary Change 21 (1999) 151–171 Fig. 4. UVLA-ICP-MS data from individual glass shards Žas ppm, anhydrous. plotted against EPMA major element data Žanhydrous. from Miocene glass shards from the Ruby Range, Montana. Ža. CaO wt.% vs. Zr ppm, Žb. CaO wt.% vs. Sr ppm. MSD2. Fig. 5c shows U vs. Th for the three tephras, with almost complete separation of the three deposits using these elements. The general decrease in Sr between MSD2 and BT3 due to alkali feldspar ex- Fig. 5. UVLA-ICP-MS data from individual glass shards Žas ppm, anhydrous. from Miocene glass shards from the Ruby Range, Montana. Ža. Zr vs. Nb, Žb. Rb vs. Sr, Žc. Th vs. U. N.J.G. Pearce et al.r Global and Planetary Change 21 (1999) 151–171 167 Fig. 6. UVLA-ICP-MS chondrite normalised REE data Žanhydrous. from individual glass shards in WR1 from the Miocene age Six Mile Creek Formation in the Ruby Range, Montana. Chondrite concentrations from Boynton Ž1984.. traction is seen again, whilst Rb is broadly similar in the two tephra beds Žbetween about 90 and 135 ppm., indicative of buffering of Rb in the magma. Relationships such as these can be used to determine the fractional crystallisation history of the magma. Fig. 6 shows a chondrite normalised REE diagram Žspidergram. for sample WR1. The heavy rare earth element ŽHREE. concentrations, particularly for the odd atomic number Ž Z . elements, are very close to theoretical detection limits and this produces a rather ‘spiky’ spidergram in the middle rare earth element ŽMREE. to HREE region. For this reason the odd-Z HREE have been omitted from the plots, and the curve for each element is plotted through the Žapproximately 10 times more abundant. even-Z HREE. This gives a very smooth curve for the REE for each shard, with all but one shard from this sample giving a very consistent REE pattern. The low REE shard in WR1, which has high Sr, appears to have ablated into a feldspar micro-phenocryst within the shard Žlower REE, higher Sr.. The negative Eu anomaly shown by all individual shards is a result of feldspar fractionation at some stage during the evolution of the magma, and is consistent with the SiO 2 –CaO and the CaO–Sr data described above. Fig. 7. UVLA-ICP-MS chondrite normalised CerYb ŽCe n rYb n . vs. EurEu) for individual glass shards from Miocene tephras from the Six Mile Creek Formation, Ruby Range, Montana. Note the two populations of shards in sample MSD2, with three samples showing virtually no negative Eu anomaly. 168 N.J.G. Pearce et al.r Global and Planetary Change 21 (1999) 151–171 Fig. 8. UVLA-ICP-MS chondrite normalised REE data Žanhydrous. for glass shards from MSD2, Six Mile Creek Formation, the Ruby Range, Montana. Ža. Shards with a pronounced negative Eu anomaly Ž18 of the 21 shards analysed., Žb. Shards with a very small negative Eu anomaly Ž3 of the 21 shards analysed.. N.J.G. Pearce et al.r Global and Planetary Change 21 (1999) 151–171 The REE data can be presented in a somewhat different manner ŽFig. 7., which shows a measure of the chondrite normalised slope of the REEs ŽCe nrYb n . plotted against the size of the Eu anomaly ŽEurEu).. Again, WR1 is virtually distinct from MSD2, with BT3 overlapping both samples. BT3 generally has a slightly more pronounced negative Eu anomalies when compared with MSD2. These data are consistent with a similar source evolving by loss of feldspar which concurs with the CaO and Sr data. Notable in sample MSD2 are three shards which show very small negative Eu anomalies ŽEurEu)) 0.96. whilst all other shards from this sample show more distinct negative Eu anomalies ŽEurEu)0.88.. The difference between these two groups of shards is almost the same as the range of data shown by the shards with the larger Eu anomalies, and is not an artefact of the analyses. In MSD2, the presence or absence of a Eu anomaly serves initially to define two populations of shards, and individual REE spidergrams are presented in Fig. 8 for these two populations. Fig. 9 shows FeO vs. EurEu) for shards from MSD2. Whilst the low FeO and high FeO populations of shards identified from the EPMA data are distinct, they have broadly the same size of Eu anomaly. The shards with small to no negative Fig. 9. FeO wt.% ŽEPMA. vs. EurEu) ŽUVLA-ICP-MS. for individual shards from MSD2 from the Six Mile Creek Formation, Ruby Range, Montana. 169 Eu anomaly appear to form a third population of shards in this deposit, with intermediate FeO contents. There is no indication in the major elements of this third population. 7. Conclusions Bulk geochemical data for trace elements derived by LA-ICP-MS compare exceptionally well with bulk data derived by solution nebulisation ICP-MS or INAA analysis for the same samples. Accuracy is typically within "10% and precision is typically around "5%. The major advantage that bulk analysis by LA-ICP-MS has over other methods is that it only consumes about 80 mg of sample rather than the 0.5 g required for solution ICP-MS and INAA. Picking of such small amounts of material is much easier than separating 0.5 g of clean glass shards. There is as yet no analytical technique with the versatility and range of analytes offered by LA-ICPMS which could be used to test the accuracy of the data derived from individual shards. The bulk sample data are so good however, that they imply the single shard data must also be of equal quality. The consistent geochemical interpretations derived from both major and trace element data from individual shards adds to the trust which can be placed in the quality of the data. With detection limits for trace elements at around or below 1 ppm, the application of UVLA-ICP-MS analysis to tephrochronology will add a wealth of new data to the subject. The accuracy of the trace element data may enable tephra beds to be distinguished using trace element data where EPMA data has been unable to separate deposits. The data from single shards not only can be used in addition to EPMA data for the discrimination of individual deposits, but can also be interpreted to give information on magmatic evolution within a single deposit. One of the most powerful applications of this trace element method will be to distinguish between individual populations of shards within one deposit which cannot be discerned in the major element EPMA data. Not only will this give information on the volcanic sources of glass shards, but also add a vast number of elements which can be used in correlation 170 N.J.G. Pearce et al.r Global and Planetary Change 21 (1999) 151–171 studies. Any bulk method, be it solution ICP-MS, INAA, XRF, etc., is unable to produce anything like the quantity of information this new single shard trace element technique can offer, although bulk analytical methods may have slightly better accuracy than LA-ICP-MS. Bulk techniques are also unable to distinguish between separate populations of shards within the same sample. Possibly the only disadvantages of this technique are a spatial resolution of about 20 mm limiting the size of shard we can analyse, and the destructive nature, a problem no different from any solution method. These minor problems are far outweighed by the advantages single shard analysis offers. In this paper, we have summarised our earlier work and presented our most recent analytical data from single glass shards. LA-ICP-MS is still a rapidly developing analytical technique. Over the last 7–8 years laser ablation has entered the realms of quantitative analytical techniques and instrumentation has managed to evolve at roughly the same rate as demands for higher sensitivity. Improvements in sensitivity are likely in the near future, and the potential for better quality data to be achieved for even more elements is great. We are continuing to experiment with analytical conditions, calibration strategies and instrumental configurations in order to explore the full potential of this technique for more highly sensitive and better spatially resolved analyses. Acknowledgements This work has been supported by grants from the Natural Environment Research Council UK, the National Science Foundation, USA, the Natural Science and Engineering Research Council of Canada and the University of Wales, Aberystwyth Žthrough a studentship to WJE and a grant to WTP for modifications to the ICP-MS.. The Royal Society financed modifications to the laser ablation system. We thank Mr. Dave Kelly ŽAberystwyth. for his skills in fixing, maintaining and modifying the ICP-MS electronics. Dr. P.G. Hill and an anonymous referee are thanked for their comments on an earlier version of this manuscript. References Abell, I.D., 1991. Performance benefits of optimisation of laser ablation sampling ICP-MS. In: Holland, J.G., Eaton, A.N. ŽEds.., Applications of Plasma Source Mass Spectrometry. Royal Society of Chemistry, Cambridge, pp. 209–217. Boynton, W.V., 1984. Geochemistry of the rare earth elements: meteorite studies. In: Henderson, P. ŽEd.., Rare Earth Element Geochemistry. Elsevier, pp. 63–114. Eastwood, W.J., Pearce, N.J.G., Perkins, W.T., Lamb, H.F., Westgate, J.A., Roberts, C.N., 1999. Tephrochronology and geochemistry of Santorini tephra in lake sediments from Southwest Turkey. Global and Planetary Change 21, 17–29. Fritz, W.J., Sears, J.W., 1993. Tectonics of the Yellowstone hotspot wake in southern Montana. Geology 21, 427–430. Jarvis, K.E., Gray, A.L., Houk, R.S., 1992. Handbook of Inductively Coupled Plasma Mass Spectrometry. Blackie, Glasgow, 380 pp. Jeffries, T.E., Perkins, W.T., Pearce, N.J.G., 1995. Comparison of infrared and ultraviolet laser probe microanalysis inductively coupled plasma mass spectrometry ŽLPMA-ICP-MS. in mineral analysis. Analyst 120, 1365–1371. Jeffries, T.E., Pearce, N.J.G., Perkins, W.T., Raith, A., 1996. Chemical fractionation during infrared and ultraviolet laser ablation inductively coupled plasma mass spectrometry—implications for mineral microanalysis. Analytical Communications 33, 35–39. Kemp, K.M., 1993. Tephrostratigraphy of the Late Cretaceous Kanguk Formation, Banks Island, Arctic Canada. Unpublished MSc Thesis, University of Toronto, 146 pp. Ninkovitch, D., Sparks, R.S.J., Ledbetter, M.J., 1978. The exceptional magnitude and intensity of the Toba eruption, Sumatra: an example of the use of deep-sea tephra layers as a geological tool. Bulletin of Volcanology 41, 286–298. Pearce, F.M., 1991. The use of ICP-MS for the analysis of natural waters and an evaluation of sampling techniques. Environmental Geochemistry and Health 13, 50–55. Pearce, N.J.G., Perkins, W.T., Fuge, R., 1992. Developments in the quantitative and semi-quantitative analysis of trace elements in carbonates. Journal of Analytical Atomic Spectroscopy 7, 595–598. Pearce, N.J.G., Westgate, J.A., Perkins, W.T., 1994. Trace element analysis of single glass shards in volcanic deposits by laser ablation ICP-MS: applications to tephrochronology. Geological Society of America Annual Meeting 1994. Abstract with Programs, A-483. Pearce, N.J.G., Westgate, J.A., Perkins, W.T., 1996. Developments in the analysis of volcanic glass shards by laser ablation ICP-MS: quantitative and single internal standard multi-element methods. Quaternary International 34–36, 213–227. Pearce, N.J.G., Perkins, W.T., Westgate, J.A., Gorton, M.P., Jackson, S.E., Neal, C.R., 1997. Compilation of published data for NIST 610 and NIST 612 partially certified reference materials including new major and trace element data. Geostandards Newsletter, in press. Perkins, W.T., Pearce, N.J.G., 1995. Mineral microanalysis by laserprobe inductively coupled plasma mass spectrometry. In: N.J.G. Pearce et al.r Global and Planetary Change 21 (1999) 151–171 Potts, P.J., Bowles, J.F.W., Reed, S.J.B., Cave, M.R. ŽEds.., Microprobe Techniques in the Earth Sciences. Chapman & Hall, London, pp. 235–291. Perkins, W.T., Fuge, R., Pearce, N.J.G., 1991. Quantitative analysis of trace elements in carbonates using laser ablation inductively coupled plasma mass spectrometry. Journal of Analytical Atomic Spectroscopy 6, 445–449. Perkins, W.T., Pearce, N.J.G., Westgate, J.A., 1997. Calibration strategies for laser ablation ICP-MS: examples from the analysis of trace elements in volcanic glass shards and sulphide minerals. Geostandards Newsletter, submitted. Pouchou, J.L., Pichoir, F., 1985. ‘PAP’ fŽrZ. procedure for improved quantitative microanalysis. In: Armstrong, J.T. ŽEd.., Microbeam analysis—1985. San Francisco Press, USA, pp. 104–106. Sarna-Wojicicki, A.M., Bowman, H.R., Russell, P.C., 1979. Chemical correlation of some late Cenozoic Tuffs of Northern and Central California by neutron activation analysis of glass and comparison with X-ray fluorescence analysis. U.S. Geological Survey Professional Paper 1147, 15 pp. Smith, D.G.W., Westgate, J.A., 1969. Electron probe technique for characterising pyroclastic deposits. Earth and Planetary Science Letters 5, 313–319. 171 Vreeken, W.J., Westgate, J.A., Alloway, B.V., 1992. Geomorphic significance of Miocene rhyolitic tephra beds from the Cypress Hills, Saskatchewan, Canada. Quaternary International 13–14, 23–28. Westgate, J.A., Christiansen, E.A., Boellstorff, J.D., 1977. Wascana Creek Ash ŽMiddle Pleistocene. in southern Saskatchewan: characterisation, source, fission track age, palaeomagnetism and stratigraphic significance. Canadian Journal of Earth Science 14, 357–374. Westgate, J.A., Gorton, M.P., 1981. Correlation techniques in tephra studies. In: Self, S., Sparks, R.S.J. ŽEds.., Tephra Studies. D. Reidel Publishing, pp. 73–94. Westgate, J.A., Walter, R.C., Pearce, G.W., Gorton, M.P., 1985. Distribution, stratigraphy, petrochemistry and palaeomagnetism of the Late Pleistocene Old Crow tephra in Alaska and Yukon. Canadian Journal of Earth Science 22, 893–906. Westgate, J.A., Perkins, W.T., Fuge, R., Pearce, N.J.G., Wintle, A.G., 1994. Trace element analysis of volcanic glass shards by laser ablation inductively coupled plasma mass spectrometry: application to tephrochronological studies. Applied Geochemistry 9, 323–335.