JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER VOL. 9 1093 Minimization of Spectral Interferences in Inductively Coupled Plasma Mass Spectrometry by Simplex Optimization and Nitrogen Addition to the Aerosol Carrier Gas for Multi-element Environmental Analysis* Trijntje van der Velde-Koerts and Jan L.M. de Boert RIVM P.O. Box 1 3720 BA Bilthoven The Netherlands Low-level trace element determinations by inductively coupled plasma mass spectrometry in environmental samples e.g. groundwater can be seriously hampered by spectral interferences originating from matrix- components and/or the argon plasma. Two approaches to reduce the level of interference were investigated variable step size simplex optimization with a number of response factors (analyte signals precision oxide and other polyatomic interferences) was applied to reduce simultaneously a number of molecular ion interferences; and nitrogen was added to the aerosol carrier gas and the nitrogen flow rate was used as an additional variable in the simplex optimization.Simplex optimization alone resulted in a more consistent choice for the settings of the instrumental variables and in reduction of the levels of polyatomic interferences. By the addition of nitrogen spectral interferences were further reduced by a factor of 1.5 to 3. For each set of optimum instrumental settings the spectral interferences were quantified as apparent analyte concen- trations. Final levels of interference were < 3 pg I-' for Cr Cu Nil Se and Zn and <10 pg I-' for As and V in 0.02 mol I-' Na 0.02 mol I-' CI 0.01 mol I-' Ca 0.01 mol I-' Mg or 0.004 mol I-' SO solutions.Matrix- induced sensitivity changes were also reduced by the addition of nitrogen. Keywords Inductively coupled plasma mass spectrometry; nitrogen addition; polyatomics; simplex optimization In principle inductively coupled plasma mass spectrometry (ICP-MS) is a very suitable technique for the determination of trace elements in solutions. Favourable aspects of the technique are the low detection limits multi-element analysis capability extended linear dynamic range and limited sensi- tivity for matrix-induced sensitivity changes of the analyte signal when an internal standard is used. However low-level determination in real samples for instance groundwater can be seriously hampered by interferences caused by polyatomic ions having the same M / Z ratio as the analyte ions to be determined.'-'' Tan and Horlick' showed that the spectral background in ICP-MS may be quite complex particularly in HC1 and H2S04.The minor isotopic species of the background spectra cannot be ignored when elemental determinations are being made at the ultra-trace level. These isotopic species also involve compounds related to the argon plasma and oxides and hydroxides. Tables of possible polyatomic ion interferences are given in references. '-,J ',12 Various approaches can be followed to reduce the effects of polyatomic ion interferences. These approaches are (i) optimiz- ation of instrumental parameters like gas flow rates power etc. and/or instrumental configurati~n~>~*'~ (ii) mathematical correction procedure^^'^*'' (iii) the introduction of organic solvents like propanol into the p l a ~ m a ~ ? ~ (iu) the application of matrix matching for certain well-defined samples like human serum,' (v) addition of gases other than argon to the nebulizer gas 5*6,13-15 coolant gas,16+17 or all gases," and (vi) reduction of the waterload to the plasma to reduce the level of oxides and hydroxides by spray chamber coolinglg or desolvation of the aeros01.~'-~~ A combination of the various approaches is also possible.Hill et applied simplex optimization of the operating parameters of the spectrometer in combination with the addition of nitrogen to the outer or aerosol carrier gas to reduce the interference of ArCl + on 75As. Mathematical correc- tion of the analyte signal [approach (ii)] can be carried out * Presented at the XXVIII Colloquium Spectroscopicurn Inter- t To whom correspondence should be addressed.nationale (CSI) York UK June 29-July 4 1993. straightforwardly for isobaric overlap and for polyatomic ions having another isotope combination at another m/z value that is free of interference. In principle correction can also be carried out for polyatomic interferences4,'' based on an assumed fixed ratio of a parent ion signal and the polyatomic ion signal. However this ratio is possibly not constant in all circumstances and also the method is inherently limited. Ketterer et al.7 applied multiple linear regression (MLR) and principle components regression (PCR) to determine low levels of cadmium in the presence of Zr Mo Ru In and Sn.However this approach is limited to multi-isotope element systems. Lam and Horlick16 observed modest signal enhance- ments (up to a factor of 4) and an order of magnitude reduction for certain background species by the addition of nitrogen to the outer flow of an argon plasma [approach ( u ) ] . In contrast Beauchemin and Craig17 found a reduction of the sensitivity by a factor of up to 5 upon the addition of nitrogen to the argon outer gas. Yet the detection limit for iron was improved by up to a factor of 4. Addition of xenon to the aerosol carrier argon gas also reduced the sensitivity considerably while the polyatomics were even more reduced.14 Multiple desolvation at -80°C in combination with an ultrasonic nebulizer [approach (ui)] considerably reduced the amount of water and HC1 introduced into an argon inductively coupled plasma.22 Polyatomic ion interferences from ArO' ClO' and ArCl+ were reduced by several orders of magnitude.Oxide levels for a number of metals were also considerably reduced e.g. UO' :U+ ratios were reduced to 0.06% when cryogenic desolv- ation was used in combination with the addition of acetylene to the central channel.22 The approaches mentioned above all suffer from limited (practical) applicability such as applicability to only one or a few elements or insufficient reduction of the level of polyatomic ion interferences to be able to determine all trace elements of interest at the desired low concentration levels in 'real world' samples with sometimes high levels of dissolved solids.None of these approaches have led to a procedure that takes full advantage of the low-level multi-element analysis capabilities of ICP-MS. Moreover quantitative information is still lacking in the literature about the actual level of the interferences in a particular case.1094 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER VOL. 9 This study gives the quantification of spectral interferences by polyatomic ions for a number of environmentally important trace elements in relation to the analysis of groundwater. Elements considered were As Cr Cu Ni Se V and Zn while the matrix components were Na Ca Mg C1 and SO4,-. Quantification was done after variable step size simplex optim- ization of the instrumental settings with a series of criteria to reduce simultaneously a number of spectral interferences.Optimization was carried out both in the presence and absence of Cl (HC1). In the second approach to reduce the effects of polyatomic ion interferences nitrogen was added to the aerosol carrier gas and the nitrogen flow rate was used as an additional variable in the simplex optimization procedure. The choice was made for addition to the carrier gas because this was the most effective one compared with addition to the outer gas and/or auxiliary gas.' After optimization the levels of poly- atomic ion interferences were established again. Experiment a1 Chemicals Working solutions were prepared from 65% m/m HNO (Suprapur Merck Darmstadt Germany); 1.000 g In (Fixanal Riedel-de Haen Seelze Germany) 1.000 g Cr Ni Co Cu Zn As" Se Ba (Titrisol Merck); 1.000 g I-' V (Fluka Buchs Switzerland); NaCl CaC12.4H,0 Na2S0 Mg( NO3),*6H2O Ca( N03),.4H20 (Suprapur Merck).Instrumentation An ICP mass spectrometer (PQ2plus FI Elemental Winsford Cheshire UK) upgraded with a 'high-performance' interface was used equipped with a Pt sample cone a Ni microskimmer cone a Meinhard nebulizer and a water-cooled spray chamber. Nitrogen was added to the aerosol carrier gas via a T-piece before the inlet using a mass flow controller (Hi-Tec F-201C range 0-50 ml min - ' N Bronkhorst Ruurlo The Netherlands). Simplex Optimization of the Instrumental Variables Optimization of the instrumental variables was carried out under four different conditions resulting in four sets of measure- ment conditions optimization condition I (OC I) in the presence of HNO,; optimization condition I1 (OC 11) in the presence of HC1; optimization condition I11 (OC 111) in the presence of HN03 while adding nitrogen; and optimization condition IV (OC IV) in the presence of HCl while adding nitrogen. R.f.power outer gas flow intermediate gas flow Ar carrier gas flow N2 carrier gas flow and sampling depth (i.e. distance between coil and sample cone) were optimized using a variable step size simplex procedure. The initial simplex was constructed according to the scheme of Yarbro and D e m i ~ ~ g . ~ ~ The simplex was moved to the region of optimum response by reflection contraction and expansion as described by Deming and Morgan,,' Deming and Parker26 and Shavers et The "'In signal the relative standard deviation (RSD) of the "'In signal the BaO+:Ba+ ratio and the background signals at rn/z=52 (ArO' Arc' ArN' ClOH') m/z=60 (Ni contribution of the cones) and m/z=78 (Ar,+) were used as optimization criteria for all four optimization conditions.The background signals at rn/z=51 (C10') and m/z=75 (ArCl+ ) were used as extra criteria when chloride was present (OC I1 and OC IV). The simplex optimization procedure was continued until each response reached the limit shown in Table 1. During the simplex optimization a solution of 10 pg 1-' of In + 25 pg 1 -' of Ba in either HNO or HCI (Table 1 ) was continuously nebulized. The spray chamber temperature and the sample uptake rate were kept constant at 2°C and 0.7 ml min-'.At each combination of instrumental settings the lenses were retuned for maximum signal at m/z = 115. The final simplex was repeated on a different day This resulted in the same choice of instrumental settings for each of the four optimization conditions. The four sets of measurement conditions chosen are shown in Table 2. Quantification of Spectral Interferences Spectral interferences were quantified as apparent analyte concentration i.e. the analyte concentration that was found when a matrix solution without analyte was aspirated. The apparent analyte concentration was established for seven matrix solutions at the four sets of optimum measurement conditions (Table 2). Each matrix study was carried out using a set of three solutions in the following order a standard a blank and a matrix solution.The pre-measurement stabiliz- ation time was 3 min for the standard solution (10 pg 1-' of V Cr Ni Cu Zn As Se In in 1% v/v HNO,) and 10 min for the blank (10 pg 1-l of In in 1 % v/v HNO,) and matrix solution (matrix+ 10 pg 1-' of In in 1% v/v HNO,). The 10min period for the blank served as wash-out and stabiliz- ation period. The matrix solutions were aspirated for 10 min prior to the measurement because earlier experiments showed that some of the interferences increased with time and this equilibration period was taken to represent the worst case situation. This increase was attributed to a build-up of deposits on skimmer and/or sample cone during the aspiration of a matrix solution. For at least one type of interference (increase of signal at m/z= 52 and 56 in Mg-containing solutions) it was shown that a deposit on the skimmer cone was responsible for the interference.The interference remained at the same level when a blank solution was aspirated following the aspiration of the Mg-containing solution and only disappeared after cleaning of the skimmer cone. The following matrix solutions were investigated 0.02 moll-' NaCl 0.01 mol I-' CaCl 0.01 ml1-' Ca(NO,) mixed 0.02 moll-' NaCl+0.01 mol I-' Ca(NO,) (final con- centrations) 0.004 moll-' Na,SO 0.01 mol I-' Mg(N03)2 mixed 0.004 mol I-' Na,SO,+0.01 moll-' Mg(N0,)2 (final concentrations). The concentrations reflect the somewhat higher concentrations of the matrix components in Dutch groundwater. Cones were cleaned before each matrix study.Every matrix study was carried out in duplicate (on two different days). Data acquisition was performed in multi-element peak jump mode with 3 runs per sample 50 sweeps dwell time 10240 ps 5 points per peak 6 digital-to-analogue (DAC) steps and pulse counter detector. Isotopes chosen for analysis were 51V ',Cr ',Cr 58Ni 60Ni 63Cu 65Cu 66Zn 68Zn 75A~ 82Se and '''In as internal standard. Results and Discussion Simplex Optimization For routine analysis it is important that as many elements as possible can be determined in one run under the same set of operating conditions. Simplex optimization is a quick and efficient method of optimizing various interrelated variables at the same time. Optimization is usually performed with one optimization criterion as is described by Hill et aLi3 who used the Sb" :ArCl+ ratio as criterion for the reduction of ArC1'.Where different isotopes suffer from different spectral inter- ferences as in routine analysis several optimization criteria are important. Because optimization towards one response can result in a detrimental effect on another response the final simplex is reached when all of the chosen responses have reached a 'compromise' value. When doing so the optimum response area for the instrumental variables could be obtained in one or two days by testing 18-25 different instrument settings.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER VOL. 9 1095 Table 1 Limits for response values in the simplex optimization procedure for the four optimization conditions (OC I-IV) Limits for response values Response "% signal/counts s-' 154( BaO) 138Ba m/z = 5211 15 m/z = 60/115 m/z = 7811 15 m/z = 51/115 m/z = 75/115 RSD 1 1 5 1 ~ (yo) oc I (1% v/v HNO no nitrogen) > 200 000 < 2 < 0.001 < 0.05 <0.1 <0.1 - - oc I1 (0.2 mol 1 - HCI no nitrogen) > 200 000 < 2 < 0.00 1 < 0.5 < 0.1 <0.1 < 20 < 1.0 oc I11 (1% v/v HNO nitrogen added) > 200 000 < 2 < 0.001 < 0.05 < 0.01 <0.1 - - ~ ~ ~~ oc IV (0.2 mol I-' HC1 nitrogen added) > 200 OOO < 2 (0.001 < 0.5 < 0.05 < 0.05 < 10 < 0.5 Table 2 Optimum measurement conditions (MC) after simplex optimization at the four optimization conditions mentioned in Table 1; MC I is the result of simplex optimization at optimization condition I etc.Simplex optimized values for instrument variables Instrument variable Intermediate gas/l min - ' Outer gas/l min-' Ar aerosol carrier gas/l min - N aerosol carrier gas/ml min-' Power/kW Sampling depth/mm Sample uptake*/ml min-' Spray-chamber temperature*/"C MC I 1.7 13.50 0.70 0.0 1.225 12.5 0.70 2 MC I1 1.7 13.75 0.76 0.0 1.250 12.5 0.70 2 MC I11 1.5 14.00 0.7 1 4.0 1.275 12.5 0.70 2 MC IV 0.8 14.00 0.73 6.0 1.325 12.5 0.70 2 * Variable kept constant during simplex optimization. During the simplex optimization it was found that the optimum measurement condition differed for different sets of glassware (torch nebulizer spray chamber elbow). The spec- tral interferences were therefore quantified using one set of glassware. The optimum N carrier gas flow is quite low (4-6 ml min-').At higher N flows (20-40 ml min-') the indium signal was significantly reduced ( 50-90% reduction). The BaO+:Baf ratio and the background signal at rn/z=52 were increased by a factor of 3-4 at a flow of 20 ml min-l. Also the doubly charged ion levels (Ba2+) were increased at higher N flows. In contrast relatively low interference levels were found for 75As at higher N flows at the beginning of the simplex optimization in the presence of HC1. Because of the unfavourable responses mentioned above the corresponding vertices were discarded and the simplex figure developed in another direction i.e. towards lower N flows. However the final interference level for 75As was similar to the levels found in the beginning of the procedure. So the choice of optimization for compromise conditions is the reason for the low optimum N flows compared with the 30-50 ml min-' found by other investigator^.^,^.'^.'^ The background signals of the blank and the net analyte signals in a 1% HN03 solution without matrix component remained similar for all four sets of optimum measurement conditions.Louie and Soo" obtained a small increase in both the background signal of the blank and the analyte sensitivity when nitrogen was added to the main inlet of the argon gas whereas Beauchemin and Craig28 observed an increase in the background signal of the blank and a decrease in analyte sensitivity when nitrogen was added as a sheathing gas to the sample aerosol. Matrix Induced Sensitivity Changes of the Indium Signal Table 3 shows the matrix induced sensitivity changes of the indium signal for the four sets of optimum measurement conditions.For most matrix solutions the sensitivity changes for MC 111 and MC IV where nitrogen was added (see Table 2) were less than for MC I and MC 11 where nitrogen was absent. The reduction of matrix-induced sensitivity changes by the addition of nitrogen was also shown by Hill et a[.,13 Branch et a[." and Beauchemin and Craig17 who found an elimination or reduction of matrix-induced sensitivity changes in a 0.01-0.3 mol 1-' sodium solution by addition of 3-5% nitrogen to the outer or aerosol carrier gas. Spectral Interferences Table4 shows the apparent V Cr Ni Cu Zn As and Se concentrations in seven different matrix solutions for the four Table 3 Sensitivity changes of the internal standard indium if matrix was added to 1% v/v HNO for the four sets of optimum measurement conditions (MC) in Table 2 Relative signal change of "'In (YO) Matrix/mol I-' 0.02 NaCl 0.004 Na,S04 0.01 CaC1 0.01 Mg(N0,)2 0.01 Ca(NO,) mixed 0.02 NaCl + 0.01 Ca( NO,) mixed 0.01 Mg(N03) + 0.004 Na2S04 MC I -51 - 13 - 34 - 18 - 43 -44 - 25 MC I1 - 37 - 15 - 26 -8 -31 - 38 - 26 MC I11 - 18 -21 -11 0 1 - 32 - 14 MC IV - 27 -8 -9 7 5 - 25 - 181096 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER VOL.9 Table 4 Apparent analyte concentrations (pg 1-') in different matrix solutions for the four sets of optimum measurement conditions (MC) in Table 2 0.02 moll-' NaCl in 1% HNO 0.004 moll- Na,SO in 1 YO HNO MC I 2.0 0.3 6.3 0.6 0.5 6.3 0.4 <0.1 <0.1 2.5 1.1 MC I1 MC I11 2.5 <0.1 8.0 <0.1 <0.1 5.0 0.1 <0.1 <0.1 1.8 0.4 1.5 < 0.1 4.3 <0.1 <0.1 2.1 0.1 0.9 1.2 1.1 0.2 MC IV 2.2 <0.1 5.8 t 0 .1 0.2 2.3 0.2 0.6 1 .0 1.9 0.4 MC I <0.1 < 0.1 <0.1 <0.1 t 0 . 1 1.4 1 .o 2.2 <0.1 <0.1 2.9 <0.1 0.2 <0.1 <0.1 <0.1 1.3 0.7 1.1 <0.1 <0.1 2.8 <0.1 <0.1 <0.1 <0.1 <0.1 0.7 1 .o 2.5 0.5 <0.1 2.6 0.01 moll-' Ca(NO,) in 1% HNO 0.01 moll-' Mg(NO,) in 1% HNO ~~ MC I MC I1 MC IT1 MC IV <0.1 0.3 0.2 2.1 13.3 0.1 1.2 <0.1 t 0 . 1 <0.1 0.5 <0.1 0.2 < 0.1 3.1 20.9 0.1 1.3 <0.1 <0.1 <0.1 0.4 <0.1 <0.1 <0.1 1.8 12.4 0.1 1 .o 0.7 0.7 <0.1 0.1 0.02 mol I-'NaCl+0.01 mol I-' Ca(NO,) < 0.1 0.3 <0.1 1.7 10.4 <0.1 0.9 0.5 0.4 <0.1 0.3 MC I 3.5 0.3 11.3 <0.1 5.3 5.9 1 .o <0.1 0.2 6.1 1.1 MC TI MC 111 2.4 0.2 8.4 1.5 10.1 10.5 1.2 <0.1 <0.1 3.2 0.5 1.5 <0.1 4.1 1.5 8.6 11.3 1.2 0.5 0.5 1.7 0.2 MC IV <0.1 <0.1 <0.1 <0.1 <0.1 0.9 0.5 1.1 0.2 <0.1 1.2 MC I MC I1 MC I11 MC IV < 0.1 1.5 0.4 1.5 0.2 0.1 0.4 0.9 0.2 0.1 0.7 <0.1 0.7 0.1 0.5 t 0 .1 <0.1 0.3 0.5 <0.1 <0.1 0.6 <0.1 0.6 <0.1 0.5 <0.1 <0.1 0.2 0.5 0.1 0.2 0.3 <0.1 0.8 <0.1 0.7 <0.1 <0.1 0.2 0.2 <0.1 0.2 0.4 0.004 moll-' Na,SO,+0.01 moll-' Mg(NO,) MC IV 3.7 0.2 8.7 1 .o 6.5 2.0 1 .o 0.3 0.4 3.4 0.3 MC I <0.1 <0.1 0.1 <0.1 <0.1 2.1 0.8 2.0 <0.1 < O . l 2.7 MC I1 <0.1 0.2 <0.1 <0.1 <0.1 3.0 0.8 1.8 <0.1 <0.1 4.6 MC IT1 <0.1 <0.1 <0.1 <0.1 <0.1 0.9 0.4 1 .o 0.3 <0.1 1.6 MC IV < 0.1 <0.1 0.1 <0.1 <0.1 0.8 0.4 0.8 <0.1 < 0.1 1 .o MC I 24.5 1.3 75.4 2.2 12.2 <0.1 1.1 <0.1 <0.1 22.7 0.6 MC I1 15.7 0.9 48.9 3.2 19.9 <0.1 1.2 < 0.1 <0.1 14.0 0.4 MC I11 9.7 0.6 27.0 2.2 14.0 <0.1 1.1 0.8 1 .o 8.6 0.3 MC IV 9.8 0.8 24.4 1.9 10.6 < 0.1 0.9 0.6 0.7 6.6 0.3 sets of optimum measurement conditions.The variations of the measurements (difference between the duplicate measure- ments) are not shown for reasons of readability. In many cases the agreement between the duplicate measurements of the interferences was remarkably good with a difference between the duplicates of <30% of their average. In the case of the CaC1 measurements the difference was < 15% for the signifi- cant/relevant interferences.This makes a fair comparison of the different measurement conditions feasible. The apparent analyte concentrations at MC 111 and MC IV (nitrogen added) were lower than at MC I and MC I1 (no nitrogen added) as is shown in Table 4. Measurement condition I11 gave a reduction of 1.5-3 times (relative to MC I) for chloride interferences [NaCI and mixed NaCl + Ca(NO,) solutions] and sodium interferences (NaCl solution). Measurement condition IV gave a reduction of 1.5-3 times (relative to MC I) for chloride interferences (CaCl solution) sodium interferences [Na2S04 and mixed Na,SO + Mg( NO3) solutions] sulfate interferences (Na,SO solution) calcium interferences [CaCl and Ca( NO3) solutions] and magnesium interferences [ Mg( NO,) and mixed Mg( NO,) + Na,SO solutions].JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER VOL.9 1097 The level of the ClO' and ArCl' interferences at m/z=51 53 75 was influenced by the other matrix components present in the solutions. The apparent V Cr and As concentrations in the NaCl and the mixed NaCl+Ca(NO,) solutions were lower than in the CaCl solution although the chloride concen- tration is the same in all three solutions (Table 4). This effect is not caused by a matrix-induced sensitivity change of the indium signal because a calculation of the apparent analyte concentration without internal standard also gave lower inter- ference levels for the NaCl and mixed NaCl + Ca( NO3) solu- tions. A contamination of the CaCl solution is also not likely because the ratio of the signals at m/z = 5153 and m/z = 75:77 is 3.0-3.1 which is equal to the expected ratio of 3.07 for 35C1:37C1.Possibly sodium acts as an electron donor for the reduction of positively charged polyatomic ion interferences. The level of ArO' interferences at m/z=52 and MgO,+ interferences at m/z= 58 was also matrix dependent. The apparent Cr and Ni concentrations in the Mg(NO,) solution were higher than in the mixed Na,S04 + Mg(N03)2 solution. This effect could again be caused by sodium. The NaC1' interferences at m/z = 58 and 60 can effectively be reduced (below 0.2 pg 1-') if the proper measurement conditions are chosen (Table 4). Interferences at m/z = 66 and 68 were enhanced in the presence of nitrogen (MC I11 and MC IV); this could be caused by enhanced 38Ar14N,+ and 40Ar14N + interferences.The apparent Se concentration at m/z = 82 was more than 0.2 pg 1-' in the matrix solutions investigated. When the apparent Se concentration at m/z = 82 was calculated without an internal standard the spectral interference was less than 0.2 pg 1-' in all matrix solutions except the sulfate- and mag- nesium-containing solutions. For sulfate-containing solutions the interference is ascribed to 34S1603+ and in the magnesium- containing solutions the interference could be caused by an enhancement of 40Ar,'H2+ In the other matrix solutions the apparent Se concentration is an effect of a matrix induced sensitivity change of the indium signal. The apparent Cu concentration at m/z=65 was in most cases more than 0.2 pg I-'.In sulfate- or calcium-containing solutions the signal at m/z=65 increased when the sulfate or calcium concentration was increased (data not shown). The spectral interference in these solutions is therefore assigned to 33S1602+ 32S33S' and 48Ca'601H+. In the NaCl solution the apparent Cu concentration at m/z=65 is an effect of a matrix induced sensitivity change of the indium signal because the apparent Cu concentration was less than 0.2 pg 1-' if indium was omitted from the calculation. The interference in the Mg(NO,) solution could not be attributed to a matrix induced sensitivity change of the indium signal because calculation without internal standard gave the same results. Also spectral interference by ArMg' had to be excluded because of the absence of a concentration-interference relation.Copper con- tamination of the Mg(NO,) solution is also very unlikely because there is no Cu present at m/z=63. So the origin of this interference is not clear up to now. A comparison with measurements without simplex optimiz- ation is hard to make. Pilot studies without optimization using the old standard interface (2 x lo5 counts s-' for 100 pg 1-' of In) showed very variable and sometimes high levels of inter- ferences even when the instrumentation met the BaO' and Ba2+ specifications (0.2 and 2%) of the manufacturer. Reductions of the levels of interference of 5-100 times were then observed when simplex optimization was applied. In principle any factor of improvement can be obtained (within certain boundaries) depending on the starting situation.Simplex optimization after installation of the new high- performance interface with a combined set of optimization criteria (Table 1) resulted in more consistent sets of instrumen- tal settings and more reproducable levels of interference as was shown above. When comparing the levels of interference found in this study to those mentioned in the literature comparable levels were found for "V by Lam and Hor1ick.l6 These authors found an apparent concentration for this element of 78 pg 1-' in 2% HC1. This is equivalent to about 7 pgl-' for the 0.02moll-' C1 concentration used in this study. Lam and Horlick16 investigated the addition of nitrogen to the outer gas of their SCIEX Elan 250 instrument.Hill et ~ 1 . ' ~ found an apparent concentration for arsenic of 12.5 pg 1-' for NASS-2 certified reference material (CRM) Seawater (National Research Council of Canada) (& 0.5 mol - ' NaCl ) when adding nitrogen to the aerosol carrier gas. This is equivalent to about 0.5 pg 1-' for 0.02 mol 1-' NaCl and thus about 2-2.5 times lower than the level of interference found in this study. The instrument they used was a VG PQ2 with a standard interface. However Hill et only optimized for the reduction of the 40Ar35C1 ' interference on 75As while in this paper optimization for a set of responses including the background signal at m/z = 52 was carried out which resulted in a compromise set of measurement conditions. Increase of the N2 addition' would most certainly result in an increase of this background signal and deterioration of the detection limit for chromium at mlz = 52.Conclusion Variable step size simplex optimization proved to be a powerful tool to select optimum measurement conditions in ICP-MS analysis not only for one response factor but also for a combined set of response factors (analyte signals precision oxide and other polyatomic interferences). This is especially important for routine analysis when a suite of elements has to be measured in the same run. Optimization in this way resulted in a more consistent choice for the settings of the instrumental variables in more reproducable levels of inter- ferences and in minimization of the levels of polyatomic interferences. The addition of N to the aerosol carrier gas further reduced the levels of polyatomic interferences.1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 References Tan S. H. and Horlick G. Appl. Spectrosc. 1986 40 445. Vaughan M. A. and Horlick G. Appl. Spectrosc. 1986 40 434. Munro S. Ebdon L. and McWeeny D. J. J. Anal. At. Spectrom. 1986 1 211. Date A. R. Cheung Y. Y. and Stuart M. E. Spectrochim. Acta Part B 1987 42 3. Evans E. H. and Ebdon L. J. Anal. At. Spectrom. 1989 4 299. Evans E. H. and Ebdon L. J. Anal. At. Spectrom. 1990 5 425. Ketterer M. E. Reschl J. J. and Peters M. J. Anal. Chem. 1989 61 2031. Vanhoe H. Vandecasteele C. Versieck J. and Dams R. Anal. Chem. 1989 61 1851. Lam J. W. H. and Horlick G. Spectrochim. Acta. Part B 1990 45 1327. Ashley D. At. Spectrosc 1992 13 169. Vaughan M.-A. and Templeton D. M. Appl. Spectrosc. 1990 44 1685. Shao Y. and Horlick G. Appl. Spectrosc. 1991 45 143. Hill S. J. Ford M. J. and Ebdon L. J. Anal. At. Spectrom. 1992 7 719. Smith F. G. Wiederin D. R. and Houk R. S. Anal. Chem. 1991 63 1458. Branch S. Ebdon L. Ford M. Foulkes M. and O'Neill P. J. Anal. At. Spectrom 1991 6 151. Lam J. W. H. and Horlick G. Spectrochim. Acta. Part B 1990 45 1313. Beauchemin D. and Craig J. M. Spectrochim. Acta Part B 1991 46 603. Louie H. and Soo S. Y.-P. J. Anal. At. Spectrom. 1992 7 557. Hutton R. C. and Eaton A. N. J. Anal. At. Spectrom. 1987,2,595. McLaren J. W. Lam J. W. and Gustavsson A. Spectrochim. Acta Part B 1990 45 1091. Tsukahara R. and Kubota M. Spectrochim. Acta Part B 1990 45 581.1098 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY OCTOBER VOL. 9 22 Alves L. C. Wiederin D. R. and Houk R. S. Anal. Chem. 1992 28 Beauchemin D. and Craig J. M. in Plasma Source Mass 64 1164. Spectrometry ed. Jarvis K. E. Gray A. L. Jarvis J. and 23 Jakubowski N. Feldmann I. and Stuewer D. Spectrochim. Acta. Williams J. Special Publication No. 85 Royal Society of Part B 1992 47 107. 24 Yarbro L. A. and Deming S. N. Anal. Chim. Acta 1974,73 391. 25 Demina S. N. and Morgan S . L. Anal. Chem. 1973 45 278A Chemistry Cambridge 1990 p. 25. 26 Deming; S. N. and Parker Jr. L. R. CRC Crit. Rev. Anal. Chem. 1978 7 187. 27 Shavers C. L. Parsons M. L. and Deming S. N. J. Chem. Ed. 1979 56 307. Paper 3/06899D Received November 28 1993 Accepted April 18 1994