Determination of Trace and Ultra-trace Elements in Saline Waters by Inductively Coupled Plasma Mass Spectrometry After Off-line Chromatographic Separation and Preconcentration Journal of Analytical Atomic Spectrometry KYM E. JARVIS AND JOHN G. WILLIAMS NERC ICP-MS Facility Centre for Analytical Research in the Environment Imperial College Silwood Park Ascot Berkshire UK SL5 7TE ELENA ALCANTARA Department of Civil Engineering Imperial College South Kensingron London S W7 UK JULTAN D. WILLS Groupe de Geochimie URA-CNRS Dl 767 Place Eugene Bataillon F-34090 Montpellier CEDEX 5 France A method for the determination of trace and ultra-trace elements in sea-water by ICP-MS after preconcentration and matrix removal using a commercially available ion-chelation system was developed and optimized.Cellulose-immobiliz,ed ethylenediaminetriacetic acid was used as the column material and elution profiles were determined with an on-line configuration. Recovery experiments were carried out firstly in a simple matrix (spiked de-ionized water) and secondly in a complex matrix (synthetic sea-water) employing a preconcentration factor of either 10 or 100. In both types of matrix recoveries of between 90 and 110% were obtained for Cd Co Cu Ni Pb U Y and the 14 REEs. Excess recoveries were obtained for Zn and are thought to result from reagent contamination. Reproducibility was in the range 0.20-7.25% coefficient of variation. Optimized conditions were used to analyse National Research Council of Canada CRMs Estuarine Water SLEW-1 Open Ocean Sea-water NASS-4 and Nearshore Sea-water CASS-2.Accuracy and precision are related to concentration. Cd Co Cu and Ni show good accuracy for concentrations greater than 1 ng m1-I. Both short term instrumental precision and between sample reproducibility are typically better than 10% RSD. Although the rare earth elements are not certified in any of the reference materials a relatively smooth sea-water normalized plot is seen for SLEW-1. Keywords Inductively coupled plasma mass spectrometry; preconcentration techniques; saline waters The accurate determination of trace and ultra-trace elements in sea-water is one of the most important and challenging tasks in analytical chemistry. Heavy metals Cd Cu Ni Pb and Zn and actinide elements Th and U are often required for routine monitoring of marine environmental pollution and for ocean modelling programmes.In particular the REEs are used as tracers in the study of submarine volcanism and also to determine the degree to which riverine water influences the sea-water-estuarine mixing zone.' In reviewing the analytical techniques currently used for inorganic trace analysis ICP-MS initially appears to be one of the most appropriate for ultra-trace determination in saline solutions. The technique offers excellent sensitivity and very low detection limits (ng 1-' and pg 1-I). It is a very rapid multi-element technique (in principle 70 elements can be determined in < 2 ml of sample solution in < 2 min) and offers linear calibration ranges over at least five orders of magnitude. During the last ten years it has become one of the most widely used techniques for the determination of REEs and other trace and ultra-trace elements in a wide variety of matrices.2 However the high salt concentration of sea-water makes it a difficult matrix both physically and chemically for analysis by ICP-MS.The high dissolved-solids loading of sea-water ( M 1-3%) results in matrix deposition on the torch injector spectrometer interface and ion lenses often leading to severe degradation of sen~itivity.~ In addition the analyte response is suppressed by major components of the ample.^-^ A simple remedy is to dilute the sample; however this decreases the already extremely low analyte levels present in naturally occur- ring saline waters in some cases to an undetectable level.The nature of the sea-water matrix also causes a number of polyatomic ion interferences (e.g. 40Ar23Na+ on 63Cu) which can considerably degrade the analytical capabilities of the instrument. Clearly separation of the matrix before analysis by ICP-MS is desirable. In some cases a matrix removal step prior to analysis is sufficient to allow quantitative deter- mination of the analytes. However in others cases such as the analysis of open ocean sea-water the levels of most analytes are below ICP-MS limits of detection. Hence a preconcentration step together with matrix removal is essential. This study deals with the optimization of a method for the simultaneous determination of REEs and transition elements in sea-water by ICP-MS after off-line automated chelation preconcentration.Cellulose-immobilized EDTrA (ethylenedia- minetriacetic acid) was investigated as a chelating agent. The preconcentration method developed is simple and straightfor- ward to carry out and combined with ICP-MS can provide considerable information about the trace and ultra-trace composition of saline solutions. INSTRUMENTATION A commercially available chromatography system (Tracecon Knapp Logistik Automation A-8042 Graz Austria) was used. Fig. 1 shows a schematic of the unit and the relationship between the analytical column pumps and valves switched during the separation and preconcentration cycle. Each of the pinch valves and peristaltic pumps can be individually manipu- lated by the system control software. The chromatography unit was supplied with a solid phase of 0.25mg EDTrA on fibrous cellulose held in a disposable syringe.This functional Journal of Analytical Atomic Spectrometry October 1996 Vol. 11 (91 7-922) 91 7Table 2 ICP-MS instrument operating conditions i M eluent waste Eluent in Sample Fig. 1 Schematic of separation and pre-concentration unit P peristal- tic pump; V valve; B bubble sensor; and C analytical column containing cellulose-immobilized ethylenediaminetriacetic acid ( EDTrA) group has been found to give excellent preconcentration characteristics for the REEs as well as for a range of transition metals.' The control software allowed the daily calibration of all three pumps in order to determine the sample and reagent volumes accurately. Capacitive bubble sensors were employed to prevent the analytical column from running dry.Chelating agent in an immobilized form was packed into a column conditioned and the sample solution loaded. The analytes were complexed by the chelating agent while the sample passed through the solid phase. To make this possible the sample and the chelating agent were buffered to a pH suitable for the formation of complexes between the analytes (divalent or trivalent metal cations e.g. REEs transition elements U and Th) and the chelating functional group (EDTrA). The effect of sample solution pH on the enrichment was studied by Schramel et aL8 and pH 4 was found to be optimum for the quantitative recovery of trace metals. The remaining matrix (monovalent and some divalent ions e.g. Na Mg Ca C1) was either not retained in the column and was sent directly to waste or was weakly chelated and subsequently removed.by a buffer wash. By doing this the necessary matrix removal step was achieved. A small volume of eluent acid was then passed through the column releasing the metal cations from the solid phase and was either sent for direct analysis to the ICP-MS unit (on-line operation) or was collected for later analysis (off-line oper- ation). The preconcentration factor achieved was the ratio of initial sample volume to final eluent volume collected. The time taken to complete each step of the procedure is shown in Table 1. The major time-consuming step was the loading of the sample solution onto the column particularly if a large preconcentration factor was required.Separated samples were analysed using a PlasmaQuad PQ2 STE (VG Elemental Winsford Cheshire UK). Operating conditions for routine analysis were employed and are shown in Table 2. Residual Na Ca and Mg were determined by ICP- AES (JY 24 monochromator system; Instruments SA UK Stanmore Middlesex UK) using the Mg 285.213nm Ca 422.673 nm and Na 589.592 nm spectral lines. Plasma conditions- Instrument Forward power/W Reflected power/W Coolant gas flow/l min-' Nebulizer gas flow/l min-' Auxiliary gas flow11 min-' Znte$ace- Sampling cone Skimmer cone Nebulizer- Type Solution up-take rate/ml min-' Data acquisition parameters- Dwell time per channelfps Number of channels Total measurement time/s Scan range/m/z Skipped mass regions/m/z PlasmaQuad STE 1350 < 5 14 0.80 0.5 Ni 1 mm orifice Ni mini-skimmer 0.7 m de Galan V-groove 0.5 160 20 per mass unit 60 23.6-239.4 27.4-42.6 79.4-80.6 EXPERIMENTAL In order to check the validity of the method two dim of experiments were carried out.First elution prof plotted to determine the necessary volume of eluent to separate the elements of interest and secondly experiments were performed to assess the capabilii column to preconcentrate the trace elements quans The accuracy of the method was tested by analysis National Research Council Canada (NRCC Ottawa CRMs Estuarine Water SLEW-1 Open Ocean 2 NASS-4 and Nearshore Sea-water CASS-2. Reagents Standards and Test Solutions Nitric acid acetic acid and ammonia solution 'Aristar' grade (Merck Lutterworth Leicestershire 1 moll-' ammonium acetate solution was used to samples while a 0.1 moll-' ammonium acetate buffel was used to condition the column prior to sample Solutions were adjusted to pH 4.0k0.1 and were before use by passing them through an EDTrA colu exclusively for reagent clean-up).A 1 moll-' HN03 was used as the eluent. All buffers eluents and test were prepared in high-purity deionized water (18 MC were stored in pre-cleaned (leached with 10% HNO polyethylene bottles. A series of test solutions were prepared from 10 stocks (SPEX Industries Edison NJ USA) to contai 1.0 and 10 ng ml-I of Cd Co Cu Ni Pb Th U Y the 14 REEs. Approximately 1 ml of 1 moll-' an Table 1 Chromatographic unit operating conditions employed for recovery experiments showing the pumps valves time periods an volumes used in each step Step 1 2 3 4 5 6 7 8 Pump P-2 P-2 P- 1 P- 1 P-2 P-3 P-3 P-3 Valve v-2 v-3 v-2 v-3 v-3 v-3 v- 1 v-2 Volume/ ml 2 3 2 300* 3 0.5 3 15 Time/ 18 27 24 3673 27 22 133 667 S Event Buffer solution purges the tubing Column conditioning Sample solution purges the tubing Sample solution is loaded onto the Buffer solution removes the matrix Eluent solution purges the tubing Elution of metals Column cleaning * Volume loaded determines the pre-concentration factor achieved. 91 8 Journal of Analytical Atomic Spectrometry October 1996 Vol.11acetate buffer was added and the pH was adjusted to 4.0k0.1 with 2 mol 1-' HNO and ammonia solution. A stock solution of synthetic sea-water (similar major element content to SLEW-1) was prepared by dissolving 268.82 g of NaC1 101.65 g of MgC12-6H20 and 21.908 g of CaClz.6H20 (AnalaR) in 11 of de-ionized water. The solution was purified prior to spiking by passing it twice through the separation-preconcentration system.The stock solution was diluted by a factor of 10 prior to separation and preccrncen- tration. A working solution was prepared by transferring 100ml of stock solution to a 500ml calibrated flask. Trace multi-element spikes were added followed by 15 ml of 0.1 moll-' ammonium acetate buffer and dilution to volume with deionized water. The pH was adjusted to 4 with 2 11-101 1-' HNO and ammonia solution. BLANK LEVELS AND CONTAMINATION Initial studies indicated a contamination problem for some elements (Co Cu Ni Pb and Zn) as high recoveries were obtained whereas for the less naturally abundant elements such as REEs Th and U good recoveries (typically 100 t 5%) were recorded.The most likely sources of contamination were reagents column material and laboratory environment. Each of the reagents used (buffer eluent and deionized water) was analysed directly by ICP-MS (see Table 3). The concentration of most elements was relatively low and was similar in all the reagents used. However Cu Ni Pb and Zn displayed a significant degree of contamination from the column itself with a Zn concentration of over 50 ng ml-' in the procedural blank solution. Repeated washing of the column did not reduce these levels of contamination. It was also observed that new columns contained appreciable concentrations of these elements. A rigorous regime was implemented in order to minimize possible sources of contamination.Solutions were prepared in a laminar flow hood. When a new column was installed it was thoroughly washed by passing 200ml of eluent then 30ml of de-ionized water 15 ml of eluent and finally 15 ml of de-ionized water through it. Between different recovery experiments the column was washed following the same procedure as above except that only 100ml of eluent were used in the first step. Tubes were washed twice firstly with 2 mol I-' HNO and secondly with de-ionized water. The minimum quantity of reagents was also employed (i.e. the pH was carefully adjusted trying not to add excessive amounts of reagents to the samples and blanks). PERFORMANCE EVALUATION A number of criteria were assessed to evaluate the overall effectiveness of the method.These included the shape of the elution profiles the percentage recovery of each analyte in both simple and complex matrices after passage through the chromatography unit repeatability effectiveness of matrix removal and accuracy of the overall technique for natural samples. Recovery experiments are a vital measure of the efficiency of any separation technique. To assess recovery the unit was operated off-line. The analytical procedure used is shown in Table 1 by loading different sample volumes in step 4 different preconcentration factors were obtained (e.g. when loading 300 ml the preconcentration factor obtained was 100; for 30 ml the pre-concentration factor obtained was 10). The flow rates were maintained at 4.9 6.7 and 1.4 ml min-' respectively for the sample (P-1) buffer (P-2) and elution (P-3) pump.To test the overall recovery of the separation-preconcentration system three 300 ml aliquots of spiked de-ionized water containing 0.1 0.5 and 1.0ngml-' test elements were loaded onto the column. They were stripped off with 3 ml of acid eluent and collected giving a preconcentration factor of 100. To test the effectiveness of the chromatography system to not only preconcentrate the elements of interest but also as a method of matrix removal a complex synthetic matrix was prepared. A 300 ml portion of spiked synthetic sea-water containing 1 ng ml-' of mixed REEs and trace metals was loaded onto the column. Trace elements were stripped off with 3 ml of acid eluent and collected giving a preconcentration factor of 100. Repeatability was assessed using three 30ml Table 3 Concentration (ng ml-') of trace element contaminants in reagents used during preparation and elution of samples.The most significant levels of contamination are seen for Zn in a procedural blank which has been passed through the chromatography column Element c o Ni c u Zn Y Cd La Ce Pr Nd Sm Eu Gd Tb DY Ho Er Tm Yb Lu P Th U De-ionized water 0.183 0.776 0.204 0.326 0.551 0.288 0.192 0.284 0.137 0.196 0.159 0.208 0.189 0.275 0.287 0.23 1 0.288 0.290 0.09 1 0.193 0.072 0.187 0.065 1 moll-' HNO 0.191 0.9 18 0.469 2.13 0.584 0.319 0.195 0.293 0.136 0.198 0.156 0.208 0.184 0.274 0.289 0.23 1 0.288 0.293 0.090 0.193 0.160 0.189 0.08 1 Procedural blank* pre-column 0.178 0.766 0.294 1.12 0.552 0.354 0.196 0.285 0.137 0.198 0.155 0.209 0.190 0.274 0.295 0.232 0.29 1 0.289 0.088 0.193 0.093 0.190 0.069 Procedural blank* post-column 0.261 2.42 4.79 0.617 0.417 0.512 0.317 0.138 0.212 0.158 0.208 0.189 0.275 0.292 0.232 0.300 0.289 0.090 0.193 1.46 0.197 0.371 55.8 0.1 mol 1-' buffer? 0.195 0.823 0.855 1.13 0.556 0.361 0.194 0.287 0.136 0.201 0.160 0.209 0.182 0.274 0.290 0.232 0.290 0.291 0.09 1 0.193 0.164 0.190 0.065 1 mol 1-l buffer? 0.190 0.732 0.334 0.861 0.560 0.398 0.207 0.289 0.142 0.198 0.146 0.209 0.186 0.275 0.293 0.234 0.294 0.294 0.090 0.197 0.158 0.208 0.065 Detection limit1 0.020 0.225 0.404 2.7 1 0.056 0.099 0.006 0.015 0.002 0.003 0.006 0.035 0.010 0.002 0.0 12 0.002 0.005 0.006 0.005 0.005 0.140 0.005 0.025 * De-ionized water-buffer-NH and HNO to adjust to pH 3.? Ammonium acetate solution. 1 Replicate determinations of procedural blank (n = 5 ) 30. Journal of Analytical Atomic Spectrometry October 1996 Vol. 11 91 9aliquots of 10ngml-' mixed REEs and trace metal solution. These were eluted from the column with 3ml of acid eluent and collected giving a preconcentration . factor of 10. The effectiveness of the matrix removal process was evaluated using a 30 ml sample of 10 ng ml-' mixed REEs and trace metal solution. This was eluted with 3 ml of acid eluent and collected (giving a preconcentration factor of 10). A 300ml portion of spiked synthetic sea-water containing 1 ng ml-' of mixed REEs and trace metals was also treated in the same manner. Both collected eluents were analysed by ICP-AES to determine the concentration of residual Na Mg and Ca after the matrix removal step.The accuracy of the procedure was determined by the replicate analysis of three certified reference materials from the NRCC. A 200 ml aliquot of each was taken and 1% v/v of buffer solution was added. The solution was then adjusted to pH4 with 2 mol I-' nitric acid and ammonia solution. The sample was loaded onto the column and then 2ml of eluent acid was used to strip the column and collect the ions. RESULTS AND DISCUSSION The percentage recovery for all elements of interest was assessed in a simple matrix (see Table 4). Percentage recoveries at an initial concentration of 1.0 or 0.5 ng ml-' were better than 100+10% for all elements except Cd Cu Ni and Zn (1 ngml-') and Ni U and Zn (0.5 ngml-').At 0.1 ng ml-' initial concentration recoveries were better than 100 5% with the exception of Cd Ni Th and most notably Zn. Thorium recovery was poor (60+8%) although at higher concentrations Th appeared to be quantitatively recovered. Zinc contamination clearly results in an apparently high recovery. Good precision (n=3) was obtained for all meas- urements with values between 0.20 and 7.25% RSD. Recovery values obtained from a synthetic sea-water matrix are shown in Table 5 and display a slight overall improvement over those from the simple matrix. Excellent recoveries were obtained at a preconcentration factor of 100 at 1 ngml-' for Co REEs Pb Y and U. Thorium recovery tended to be Table 4 Percentage recovery for some transition elements actinides and REEs in spiked deionized water after a preconcentration factor of 100 Percentage recovery Element Cd Ce c o c u DY Er Eu Gd Ho La Lu Nd Ni Pb Pr Sm Tb Th Tm U Y Yb Zn 0.1 ng ml-'* (n=4) 110f6 103 f 3 103 f 5 96f 13 100f2 102f 1 102f 1 103f2 101 &2 98f3 97f 1 100f4 120f7 97f2 102f3 101 f 3 102f 1 60f8 100f2 95f 1 100f2 99*2 669 f 134 0.5 ng ml-'* (n= 1) 108 107 101 102 110 104 108 108 109 106 105 104 89 104 106 102 107 100 108 112 106 106 117 1.0 ng ml-'* (n= 1) 115 101 105 128 99 97 99 98 99 102 95 103 115 101 103 103 97 90 98 91 110 98 111 * Concentration prior to preconcentration. Table 5 Percentage recovery for some transition elements actinides and REEs in synthetic sea-water after x 10 and x 100 preconcen- tration factor Percentage recovery Element Cd Ce c o c u DY Er Eu Gd Ho La Lu Nd Ni Pb Pr Sm Tb Th Tm U Y Yb Zn Preconcentration factor x loo* 1.0 ng ml-'* 94 100 99 105 99 102 99 99 99 97 99 99 83 97 98 98 98 89 99 95 105 98 119 Preconcentration factor x lot 10.0 ng ml-'* 95+ 1 102 f 1 117+3 104+ 8 99f 1 101 f 1 101 f 1 102f0 99f1 96+ 1 96f 1 99fl 112f5 96f2 103 + 1 102f 1 99f0 91 +2 98+ 1 96.5 f 1 109+2 97+ 1 184 + 73 * Starting concentration 1.0 ng m1-I.Starting concentration 10 ng ml-'. slightly low at 89%. Recovery for Ni was also a little low at 83% in the x 100 experiment. Zinc the element most prone to contamination showed an anomalously high value (120%). Data for a times 10 preconcentration factor at 10 ng ml-' was a little more variable. Excellent precision (n=3) was achieved in this experiment .with an RSD of 7-26%.In general recovery values were in the same range and were independent of the preconcentration factor (10 or loo) the test solutions (spiked water or synthetic sea-water) and the concentration of the starting solution (0.1 0.5 1.0 or 10 ng ml-I). For both types of matrices good recoveries were obtained for Ni Pb U Y and the 14 REEs while recoveries for Cu and Zn tended to be high due to contamination from reagents. The effectiveness of matrix removal is illustrated in Table6. In both experiments the amount of total dissolved solids (TDS) in the final solution was well below the limit tolerated by the ICP-MS instrument used in this work i.e. <0.1% TDS. An assessment of accuracy was made by analysing three certified reference materials (Tables 7-9).Although reference values are only available for a limited suite of elements (Cd Co Cu Mo Ni Pb U and Zn) they serve to demonstrate the application of the method. Measurements for SLEW-1 (Table 7) were made over a period of 12 months. Co Cu and Ni displayed good repeatability over the analysis period. The certified Cd concentration is low (0.018 ng ml-') and poor repeatability is expected at this level while Pb and Zn are Table 6 Concentrations (pg ml-') of some matrix elements determined by ICP-AES before and after matrix separation Solution Na Mi? Ca Total Before matrix removal 27 880 10650 2910 41 440 After matrix removal* 105 17 11 133 After matrix removalt 203 41 22 266 * Solution preconcentrated by a factor of 10. t Solution preconcentrated by a factor of 100.920 Journal of Analytical Atomic Spectrometry October 1996 Vol. 11Table7 Analysis of Estuarine Water SLEW-1 after separation and preconcentration by a factor of 100. Precision is expressed at the 95% confidence interval Measured value/ ng ml-' Element Cd c o c u Ni Pb Zn Experiment 1 * (n= 1) 0.083 0.084 2.04 0.831 0.148 1.07 Experiment 2 t (n= 1) 0.026 0.057 1.60 0.695 0.070 1.91 Experiment 3$ (n=4) 0.01 6 f 0.001 0.053 f 0.007 1.47 f 0.16 0.562 f 0.058 0.109+0.006 1.42 f 0.5 1 Reference value/ ng ml- ' 0.01 8 f 0.003 0.046 f 0.007 1.76 f 0.09 0.743 f 0.078 0.028 f 0.007 0.860 & 0.150 * Single aliquot month 1. t Single aliquot month 8. $ Four separate aliquots triplicate scans month 12. Table 8 Analysis of Nearshore Seawater CASS-2 after separation and preconcentration by a factor of times 100.Precision is expressed at the 95% confidence interval Measured value*/ ng ml-' Element (n=4) Cd 1.83f0.10 c o 2.7 f 0.3 c u 62.0 f 4.6 Ni 24.3 f 4.6 Pb 20.2 f 0.7 Zn 161 f 67 Reference value/ ng ml'-' 1.9 f 0.4 2.5 f 0.6 67.3 f 3.9 29.8 f 3.6 1.9 f 0.6 197f :I2 ~~ ~~~ * Four separate aliquots triplicate scans. Table 9 Analysis of Open Ocean Seawater NASS-4 after separation and preconcentration by a factor of 100. Precision is expressed at the 95% confidence interval Measured value*/ ng ml-' Reference value/ Element (n=4) ng ml'-' Mo 8.84f0.60 8.88 f 1.37 U 2.68 & 0.12 2.68 f (1.08 * Four separate aliquots; Cd Co Cu Ni Pb and Zn were not determined. clearly prone to contamination.Method reproducibility (Table7 experiment 3) was better than 15% RSD with the exception of Zn where reproducibility was poor at the lppb level. CASS-2 (Table 8) contains generally higher concen- trations of trace elements than SLEW-1 and accordingly both accuracy and intra-sample precision are improved. Lead dis- played a significant degree of contamination and inaccurate results. At x200 ng rnl-' Zn contamination was probably masked by the naturally elevated concentrations of this element in SLEW-1. Open Ocean Sea-water NASS-4 contains very low concentrations of most trace elements with the exception of Mo and U. Agreement between measured and reference values was excellent (< 0.5%) with an intra-sample precision of < 6% RSD where n=4 (Table 9). Unfortunately there are no saline water reference materials available which are certified for the REEs.An assessment of the internal consistency of the data set may be made by normalizing the REE data to chondritic abundance? and plotting the results against atomic number. Due to the coherent behaviour of the group a smooth curve should result for all elements except Ce and Eu which because of their difrerent oxidation states may display anomalous behaviour.'" The REE data for both CASS-4 and SLEW-1 (Table 10) displayed smooth chondrite-normalized curves with the exception of Sm which had anomously high results for both reference materials Table 10 Measured REE concentrations (ng ml-') in SLEW-1 and CASS-4 after separation and preconcentration by a factor of 100 SLEW-1 Element La Ce Pr Nd Sm Eu Gd Tb DY Ho Er Tm Yb Lu Batch 1 0.05 1 48 0.052 13 0.003 19 0.01 1 58 0.01 1 62 0.000 45 0.002 59 0.000 33 0.002 11 0.000 47 0.001 41 0.OOO 24 0.001 26 0.00021 Batch 2 0.014 08 0.016 01 0.002 86 0.011 66 0.012 69 O.OO0 41 0.002 41 O.Oo0 37 0.002 22 0.000 55 0.001 55 0.OOO 27 0.001 62 O.OO0 28 CASS-4 0.005 10 0.001 59 O.OO0 52 0.002 22 0.001 91 O.OO0 07 O.OO0 46 O.OO0 06 o.oO0 50 0.OOO 15 o.OO0 40 O.OO0 06 O.OO0 48 0.000 07 and between batches of SLEW-1.When REE concentrations in SLEW-1 were normalized to those of CASS-4 (i.e. a sample prepared using the same method) the anomalous result was eliminated (Fig. 2). This suggests that Sm contamination has occured at some point in the preparation procedure. P v) 0 I 52100 .E I I La Ce Pr Nd Sm Eu Gd T b Dy Ho Er Trn Yb Lu !! 8 Fig.2 Measured concentrations of the REEs in SLEW-1 normalized to measured concentrations in CASS-4. Very close agreement is seen between A Batch 1 and B Batch 2 of SLEW-1 from Pr to Lu. Poor agreement for La and Ce may be due to batch differences or recovery problems. Normalization to a second sample prepared using the same method eliminates the anomalously high Sm value (easily identified if concentrations are normalized to chondrite values) in CASS-4 and SLEW-1 suggesting that Sm contamination has occurred Journal of Analytical Atomic Spectrometry October 1996 Vol. 11 921CONCLUSIONS The separation procedure described here is rather time- consuming with a single column in operation; it does offer the potential for simultaneous determination of 23 elements includ- ing the REEs and U. This is in contrast to some chelation procedures which tend to be limited to a relatively restricted range of elements." The NERC ICP-MS Facility is in part supported by the Natural Environment Research Council and this support is gratefully acknowledged. One of us (E.A.) was in receipt of a EC COMMET bursary during the period of this work. Technical advice and assistance was provided by Bridget C.H. Gibson. REFERENCES 1 Elderfield H. Upstill-Goddard R. and Sholkovitz E. R. Geochim. Cosmochim. Acta 1990 54 971. 2 Jarvis I. and Jarvis K. E. Chem. Geol. 1992 95 1. 3 4 5 6 7 8 9 10 11 Williams J. G. and Gray A. L. Anal. Proc. 1988 25 385. Beauchemin D. Mclaren J. W. and Berman S . S. Spectrochim. Acta Part B 1987 42 467. Hutton R. C. and Eaton A. N. J. Anal. At. Spectrom. 1988,3,547. Gregoire D. C. Spectrochim. Acta Part B 1987 42 895. Garbe-Schonberg C.-D. Bruhn R. and Michaelis M. presented at the Conference on Plasma Spectrochemistry Granada Spain 1993. Schramel P. Xu L.-Q. Knapp G. and Michaelis M. Mikrochim. Acta 1992 106 191. Nakamura N. Geochim. Cosmochim. Acta 1974 38 575. Jarvis K. E. Chem. Geol. 1988 68 31. McLaren J. W. Lam J. W. H. Berman S. S. Akatsuka K. and Azeredo M. A. J. Anal. At. Spectrom. 1993 8 279. Paper 6/03822K Received May 3 I 1996 Accepted August 12 1996 922 Journal of Analytical Atomic Spectrometry October 1996 Vol. 1 1