Analytical Characteristics of an Inductively Coupled Plasma Mass Spectrometer Coupled With a Thermospray Nebulization System* Journal of Analytical Atomic Spectrometry HANS VANHOE STEVEN SAVERWIJNS MAGALI PARENT LUC MOENS AND RICHARD DAMS Laboratory of Analytical Chemistry University of Ghent Institutefor Nuclear Sciences Proeftuinstraat 86 8-9000 Ghent Belgium The analytical characteristics of a thermospray sample introduction system coupled to an inductively coupled plasma mass spectrometer have been evaluated. The results obtained with the thermospray system were compared with those obtained with two other arrangements namely the pneumatic nebulizer coupled with a spray chamber (the conventional arrangement) and the same pneumatic nebulizer coupled with the desolvating unit employed with the thermospray nebulizer.In the presence of salts of Na (NaNO,) or Ca[Ca(NO,),] the non-spectroscopic interferences (analyte ion signal suppression) were more pronounced in the thermospray system than in the conventional arrangement whereas in the presence of mineral acids ( H2S04 H3PO4) both systems gave similar suppressions. The apparent analyte element concentrations due to spectral overlap with MO,' (e.g. SO' SO,' PO' POz+ CaO+) or ArM+ (e.g. ArNa+ ArS' ArP') were lower or similar for the thermospray nebulizer in comparison with those obtained with the conventional arrangement. In order to reduce memory effects a sample flow injection system was used for the analysis of three candidate environmental reference materials [Community Bureau of Reference (BCR) Certified Reference Materials 141R Soil-calcareous loam 144R Sewage sludge-domestic and 146R Sewage sludge- industrial].The accuracy and precision of external calibration with internal standardization standard additions and isotope dilution were compared. Similar results were obtained for the aqua regia soluble content of Cd and Pb in these materials. Keywords Inductively coupled plasma mass spectrometry; thermospray nebulization; memory ejfects; spectroscopic and non-spectroscopic interferences; calibration methods; analysis of environmental samples Although inductively coupled plasma mass spectrometry (ICP-MS) is a well established technique in many analytical laboratories the technique suffers in many instances from some major drawbacks.One of these is the poor efficiency typically 1-2% of the conventional sample introduction system con- sisting of a pneumatic nebulizer and a spray chamber. This feature restricts the sensitivity of the technique. Therefore many researchers have investigated alternative nebulizers with an improved nebulization efficiency. For the introduction of solutions into an ICP-MS system an ultrasonic nebulizer,ls2 a direct injection nebulizer3-' and a hydraulic high pressure nebulizer6*' have been evaluated. All of these nebulizers give an improvement in the sensitivity and the detection limits by one order of magnitude. Recently Montaser et al.' and Vanhoe et a[.' have coupled a thermospray system to an ICP mass spectrometer. In this system the solution is forced to flow through a heated capillary. At the end of the capillary the liquid is partially vaporized and the vapour is expanding * Presented at the 1995 European Winter Conference on Plasma Spectrochemistry Cambridge UK January 8-13.1995. adiabatically producing a heated aerosol. Studies on the funda- mental characteristics of the thermospray have been intensively carried out by coupling the system to an ICP optical emission spectrometer.'-'l As the size of the aerosol droplets (median diameter of 2 pm for the primary aerosol') is lower than for other nebulization systems analyte transport efficiencies of 53% have been reported," resulting in better sensitivities and detection limit^'^^'^ (up to a factor of 2OI4). Coupling of a thermospray system to an ICP mass spectrometer has also led to an improved performance.Montaser et al.' reported detec- tion limits that were improved by a factor of up to 20 for 15 elements whereas Vanhoe et a1.' concluded that the sensitivity and the detection limits were improved by a factor of 10. In addition they reported a decrease in the levels of oxide and doubly charged ions by a factor of 2.5 when an efficient desolvating system is applied. As there is an increased analyte transport to the ICP the solvent transport is also higher so that to avoid solvent overloading of the plasma the thermospray nebulizer must be followed by a desolvating system. Nevertheless from ICP optical emission spectrometry experiments it can be concluded that matrix effects are usually larger for thermospray sample introduction than for pneumatic nebulization even with the use of an efficient desolvating ~ystern.'~.'~ In this work the effects of different matrix salts and mineral acids on analyte element ion signals in ICP-MS were investi- gated.Use was made of a thermospray sample introduction system described in a previous publication' and consisting of a stainless-steel capillary tube with an internal diameter of 180pm and followed by a heated spray chamber and a condenser as desolvating unit. Results are compared with those obtained with a conventional concentric pneumatic nebulizer combined with a spray chamber and with the same pneumatic nebulizer combined with the desolvating unit employed for the thermospray nebulizer. In addition spectroscopic interferences arising from the matrix elements Ca Na P and S were investigated. In order to analyse 'real' samples a sample flow injection system was installed.With this arrangement some environmental samples were analysed and three standardiz- ation procedures namely external calibration in combination with internal standardization standard additions and isotope dilution were compared. EXPERIMENTAL ICP-MS Instrument All measurements were made with a VG PlasmaQuad PQ1 (VG Elemental a division of Fisons Instruments Winsford Cheshire UK). The original interface was replaced by a high- performance interface with a new design of sampling cone and skimmer cone and a better vacuum pumping system in order to improve the sensitivity. The operating conditions for the ICP mass spectrometer are presented in Table 1.Journal of Analytical Atomic Spectrometry September 1995 Vol. 10 575Table 1 ICP-MS operating conditions Conditions Stage Parameter Plasma Pneumatic nebulization system Thermospray nebulization system Ion sampling Vacuum Frequency Rf power Torch Gas flow plasma auxiliary Sample uptake rate Nebulizer gas flow Sample uptake rate Power applied to the capillary Temperature of the aerosol Temperature of the cooling water Carrier gas flow rate Sample cone Skimmer cone Sampling depth Expansion stage Intermediate stage Analyser stage 27.12 MHz 1350 W Fassel-type 14 I min-' 1 1 min-' 0.9 ml min-' 0.720 I min-' 1.33 ml min-' 55 w 120 "C 1 "C 0.820 I min-' Nickel 1.0 mm orifice Nickel 0.75 mm orifice 10 mm (from load coil) 1.6 mbar 1.0 x mbar 2.0 x mbar Pneumatic Nebulization System For comparison with the thermospray nebulization system a Meinhard (TR-30-A3) concentric glass nebulizer and a double pass Scott-type spray chamber with surrounding liquid jacket made of borosilicate glass were used.The spray chamber was thermostated to within 0.1 "C (1 "C) by a recirculating cooling system (Barrington LT6). A peristaltic pump (Gilson Minipuls-2) delivered a constant sample flow rate (uptake rate of 0.9 ml min-l) to the pneumatic nebulizer. Thermospray Nebulization System The thermospray sample introduction system was described in detail in a previous publication.8 The capillary tube (stainless steel with an internal diameter of 180 pm) positioned in a conical flask was followed by a heated L-shaped tube and a modified Friedrichs condenser.The optimum operating con- ditions for the thermospray nebulization system are also given in Table 1. In order to analyse 'real' samples a discrete analyte introduc- tion system was placed between the HPLC pump (Varian 8500 a single-syringe pump) and the capillary tube. It consists of a Valco sample injection valve (six gates) with a 200 p1 sample loop made of PTFE. 'Real' samples were introduced semi-continuously by using a large sample loop of 5 ml. In this way three measurements each lasting 1 min could be made on each sample. Test Solutions For the study of spectroscopic and non-spectroscopic inter- ferences matrix element solutions were prepared using NaNO (Carlo Erba pro analysi) Ca(NO& (UCB pro analysi) H3P04 (Merck pro analysi) and H2S04 (purified by sub- boiling distillation).The respective matrix element solutions- 1 5 50 250 1000 and 5000mg1-' for each matrix element (Na Ca P and S)-were prepared with 0.14moll-' nitric acid. A 10 pg 1-' multi-element solution (Be Al Sc Co In Gd T1 Th and U) was added to all matrix element solutions. This solution was prepared in 0.14moll-' nitric acid from commercially available AAS standard solutions. Water and nitric acid (14 moll-') were purified respectively by a Millipore Milli-Q water purification system (resistivity of 18 MR cm) and by a sub-boiling distillation system. All results were obtained using the scanning mode for data acquisition. The following scanning parameters were chosen mass range between 6 and 256 u 20 channels per u with a dwell time per channel of 320 ps and a total acquisition time of 1 min. The following isotopes were selected 'Be "Al 4'Sc 59C0 "'In 158~d 205T1 232Th and 238U.Sample Preparation and Analysis Procedure for the Reference Materials Three candidate environmental reference materials were ana- lysed namely Community Bureau of Reference (BCR) Certified Reference Materials (CRMs) 141R Soil-calcareous loam 144R Sewage sludge-domestic and 146R Sewage sludge-indus- trial. All materials were treated with aqua regia and the soluble content of Cd and Pb was determined. Therefore approxi- mately 1 g of each candidate reference material was heated under reflux in a mixture of 11.4 ml of aqua regia [9 ml of 10 moll-' hydrochloric acid (purified by sub-boiling distil- lation) and 2.4ml of 14mol1-' nitric acid] and 2-3 ml of Millipore Milli-Q water as described in the DIN 38 41437 procedure," except that the resulting solution was not filtered.The solution obtained was transferred quantitatively into a 100ml calibrated flask and adjusted to volume with 0.14 moll-' nitric acid. A similar procedure was followed for the preparation of a blank solution (without sample). For the determination of the aqua regia soluble content of Cd the standard additions method was applied. All sample solutions were diluted 10-fold with 0.14 mol 1-' nitric acid. For each sample an unspiked solution and a solution spiked with 25 pg 1-' of a Cd standard solution (Cd foil Goodfellow Metals 99.99%) were prepared. Rhodium (100 pg l-' Janssen Chimica AAS standard solution) was added to all solutions as internal standard.Three measurements were made on each solution with the following scanning parameters mass range between 101 and 113 u 20 channels per u with a dwell time per channel of 320ps and a total acquisition time of about 1 min. The isotopes selected were '03Rh '"Cd and "'Cd. For the determination of the aqua regia soluble content of Pb external calibration standard additions and isotope dilution were applied. For each material three sample solutions were prepared an unspiked sample solution (diluted 20-fold for CRM 141R; diluted 40-fold for CRM 144R; diluted 200-fold for CRM 146R) a sample solution spiked with 103 pg 1-' of Pb [National Institute of Standards and Technology (NIST) Standard Reference Material (SRM) 983 Radiogenic Lead Isotopic standard (zosPb206Pb =0.013619)] and a sample solu- tion spiked with 25 pg 1-' of Pb [(Pb(NO,) UCB analytical- reagent grade with a natural isotopic composition for Pb].To all solutions-samples and blank-50 pg 1-' of T1 (Alfa 576 Journal of Analytical Atomic Spectrometry September 19!?5 Vol. 10Products AAS standard solution) was added as internal stan- dard. For external calibration Pb standard solutions with a concentration of 10 25 and 50 pg 1-' of Pb [Pb(NO,) UCB analytical-reagent grade] were prepared to obtain a calibration graph. For external calibration and standard additions the ,08Pb isotope was selected for the quantitative determinations. For the isotope dilution method the mass fractionation factor was determined by measuring a Pb standard solution with a certified 208Pb:206Pb ratio (NIST SRM 981 Common Lead Isotopic standard 208Pb:206Pb = 2.168 1).Three measurements were made on each solution with the following scanning parameters mass range between 200 and 210 u 20 channels per u with a dwell time per channel of 3 2 0 p and a total acquisition time of about 1 min. The isotopes selected were ,05Tl ,06Pb and '08Pb. RESULTS AND DISCUSSION Memory Effects A flow injection system with sample loops of 200 1.11 and 5 ml was used to study the memory effects of the thermospray system. Fig. 1 (a) and (b) gives signal plots as a function of time for respectively a 200 p1 loop (filled with 25 pg I-' of T1) and a 5 ml loop (filled with 25 pg 1-' of Rb). For the 200 11 loop a transient signal is obtained with a time basis of about 40 s indicating that the signal almost immediately drops to a normal background level.For five successive analyte injections a relative standard deviation of 1.6 and 0.7% was found based on respectively the peak height and the peak area. As can be seen using a 5ml loop has the advantage that a constant signal can be obtained for 3-4min so that several measure- ments can be carried out. As already mentioned this semi- continuous arrangement was used to analyse some environ- mental samples. For the 5 ml sample loop it can be noticed that the Rb ion signal decreases by three orders of magnitude 1 x 1 0 ~ 7.5~1 O4 5x1 O4 v) 2 . 5 ~ 1 0 ~ ,o c a sr v) c .- L 5 0 C .- - m 0 ijj 1x106 20 40 60 80 TimelS 85Rb-signal 0 3 6 9 Ti m e h i n Fig.1 '03Tl ion signal as a function of time for a 200 pl sample loop filled with 25 pg I-' T1 and 85Rb ion signal as a function of time for a 5 ml sample loop filled with 25 pg I-' Rb within about 1 min. The Cd La and T1 signals behave in the same way. For elements such as Ta Pt and U a longer period of time is required before an acceptable background level is achieved. These plots are similar to those observed with a concentric nebulizer combined with a spray chamber. However Fig. l(b) illustrates that after the initial drop of the analyte signal the background signal (at the mass of the analyte element) is not stable. The short enhancements of the back- ground signal can be reduced by using the 200 pl sample loop or standard solutions with lower concentration.Additional experiments have demonstrated that this memory effect is located inside the desolvating system and not inside the capillary tube and/or the flow injection system. This phenom- enon is probably due to analyte that is left behind in the desolvating unit which is not surprising because of the large volume of this arrangement. Further experiments will be carried out to reduce the size of this unit. Finally as the solution introduced into the thermospray system is heated in the capillary tube for some elements specific memory effects can be observed. For instance Hf will precipitate as hafnium oxide in the heated capillary tube and cause a considerable memory effect. Lowering the power applied to the capillary tube or adding sufficient hydrofluoric acid to the Hf solution will respectively reduce and even eliminate this precipitation as illustrated in more detail elsewhere." Spectroscopic Interferences The polyatomic ions originating from the matrix elements Ca Na P and S that give rise to spectral overlap with analyte elements were examined.This is of particular interest as on the one hand there is undoubtedly an increase in the matrix transport to the ICP resulting in the presence of larger amounts of Ca Na P or S in the plasma On the other hand a decrease in the solvent transport (in our case mainly water) to the ICP was experimentally observed with our desolvating system8 resulting in lower oxide levels. The extent of formation of these polyatomic ions by the thermospray system was compared with that for two other arrangements namely the pneumatic nebulizer coupled with a spray chamber (conventional arrange- ment) and the same pneumatic nebulizer coupled with the desolvating unit employed with the thermospray nebulizer.An example is given in Fig. 2(a) where the 4oCa'60':1'51n+ ratio (In as internal standard) is plotted as a function of the Ca concentration. As can be seen the pneumatic nebulizer com- bined with the desolvating system (used for the thermospray nebulizer) gives the lowest Ca0:In ratios. These ratios are almost a factor of 10 lower than those obtained with a pneumatic nebulizer combined with a spray chamber. Up to a Ca concentration of 1 g l-' the Ca0:In ratios obtained with the thermospray system are significantly lower than those obtained with the conventional arrangement and are higher than those obtained with the combination of a pneumatic nebulizer and the desolvating system.In general for MO' and MOH' it can be concluded that in most instances (CaO' PO' SO') the normalized MO' and MOH' signals are lower for the thermospray system than those for the conventional arrangement and similar to or higher than those for the pneumatic nebulizer combined with the desolvating unit. A similar conclusion can be made for the formation of MO,' and M02H+ as illustrated in Fig. 2(b) for 31P160160+. From these experiments it can be concluded that the decrease in the normalized MO,' signals observed with the thermo- spray system is almost totally due to the use of the desolvating unit. As can be deduced from Fig.2(c) where the ratio of 40Ar23Na':59Co' is plotted as a function of the Na concen- tration the normalized ArNa' signals obtained with the thermospray system are similar to those obtained with the Journal of Analytical Atomic Spectrometry September 1995 Vol. 10 577I " L 10 100 1000 10000 P concentratiodmg r1 0.001 o'o't.-. 10 100 1000 10000 Na concentratiodrng r' Fig. 2 (a) 4oCa'60' :'"In+ as a function of Ca concentration [Ca(NO,)J (b) 31P160160+.45Sc+ as a function of the P concen- tration (H,PO,) and (c) 40Ari3Na+ :5gCo+ as a function of the Na concentration (NaN03) for three sample introduction systems:l ther- mospray system; 2 pneumatic nebulizer combined with the desolvating unit used for the thennospray nebulizer; and 3 pneumatic nebulizer combined with a spray chamber combination of pneumatic nebulizer and desolvating system and lower than those obtained with the conventional arrange- ment up to an Na concentration of 1 gl-'.For higher Na concentrations higher levels of ArNa+ are obtained with the thermospray system. Similar conclusions can be made for ArP+ and ArS'. Hence although there is a higher loading of matrix elements of the plasma the normalized ArM' signal is slightly lower for the thermospray system in comparison with the conventional arrangement. In general it can be concluded that although in some instances the formation of MO,' and ArM' will be higher for the thermospray system the apparent analyte element concentrations due to spectral overlaps will be lower or similar because the normalized MO,' and ArM' signals are lower or similar.Non-spectroscopic Interferences The influence of the matrix of NaN03 or Ca(NO,) (salts) and of H2S0 or H3P04 (mineral acids) on the ion signal intensities was investigated. Again the matrix effects observed with the three sample introduction systems were examined. The relative intensity of the "'Tl ion signal (the "'Ti intensities obtained for all P concentrations are divided by that obtained for 0.14 moll-' nitric acid) as a function of the P concentration (H,PO,) is presented in Fig. 3(a). When using a pneumatic nebulizer in combination with the desolvating system the T1 ion signal is only slightly suppressed up to a P concentration of 1 g 1-l (15%). For the thermospray system however the 205Tl ion signal is suppressed to a greater extent for all P concentrations for 1 g 1-l of P (H,PO,) a suppression of 50% is observed.It is important to note that the extent of suppression observed with the conventional arrangement is comparable to that observed with the thermospray system. Comparable results were obtained for elements with masses over the whole mass range. Similar conclusions can be made for the matrix effects caused by H2S04 which is illustrated in Fig. 3(b). The only significant difference is that above 1 g 1-' of S the suppression is larger for the thermospray system and that for pneumatic nebulization there is an enhancement of the signal up to a concentration of 200 mg 1-' of S. The latter effect can be explained by assuming that the zone of maximum Mt density in the plasma undergoes a spatial displacement under the introduction of H2S04." The situation is different when the influence of NaN0 and Ca(NO,) is investigated.This is illustrated in Fig. 4(a) and (b) for NaNO and Ca(NO,) respectively. Up to an Na concentration of 1 g l-' the 205Tl signal intensity is only slightly decreased when using the conventional arrangement (for 1 g 1-' of Na a suppression of 15% is observed). This is however not the case for the other two systems. The "'Tl ion signal is suppressed to a greater extent for all Na concen- trations for 1 gl-' of Na a suppression of 65 and 90% for respectively the pneumatic system combined with the desolvat- ing unit and the thermospray system is observed. It is clear that the thermospray system gives the most severe suppression I 40 .- aa 1000 10000 10 I00 $ 0 f O.( P concentration (H3P04)/mg r' eo 4 0 - 20 .01 ' I ' l " " ' 8 ' " ' 1 ' 1 ' ' 3 1 ' " ' " I 3 ' 1 ' 1 ' 1 1 I 1 '11U 0.1 1 10 100 1000 10000 S concentration (H2S04)/mg r' Fig. 3 Normalized '05Tl ion signal (0.14 mol I-' nitric acid is used as reference solution) as a function of (a) P concentration (H,PO,) and (b) S concentration (H,S04). See Fig. 2 for explanation of 1 2 and 3 578 Journal of Analytical Atomic Spectrometry September 1995 Vol. 10Na concentratWmg r' 1 120 = I \ I 40 Ca mentratiodmg r1 Fig. 4 Normalized "'TI ion signal (0.14 mol I-' nitric acid is used as reference solution) as a function of (a) Na concentration (NaNO,) and (b) Ca concentration [Ca(NO,),]. See Fig. 2 for explanation of 1 2 and 3 because for this arrangement the Na matrix loading of the plasma is the highest.As a greater suppression is observed with the pneumatic nebulizer combined with the desolvating unit in comparison with the conventional arrangement it is probable that a better transport efficiency of the matrix (Na) to the ICP takes place when the desolvating unit is employed even in combination with a pneumatic nebulizer. As can be seen in Fig. 4(b) Ca(NO,) does not significantly influence the ion signal when the conventional arrangement is used up to a Ca concentration of 5 g I-'. A relatively important suppres- sion is however observed for Ca concentrations above 100 mg 1-' when the desolvating system is used instead of the spray chamber. When the pneumatic nebulizer is replaced by the thermospray nebulizer the extent of suppression is even larger.A significant loss of sensitivity can already be observed at Ca concentrations above 50 mg 1-I. This phenomenon is undoubtedly due to the deposition of Ca salts on the cone and skimmer which was experimentally verified. No mass depen- dency of the matrix effects caused by NaNO or Ca(NO,) is observed which is similar to the results obtained for the matrix effects caused by the mineral acids. From these experiments it can be concluded that for the thermospray system the matrix effects caused by the mineral acids are similar to those observed with the conventional arrangement whereas for the effects caused by salts they are more severe for the thermospray system. The salt matrix effects are in a few instances due to clogging of the cone and skimmer orifice.Some of the observations described above can be explained by plotting the analyte ion signal intensity as a function of the carrier gas flow rate for several matrices. These measurements were carried out for the conventional arrangement and for the thermospray system with the Na and Ca salts and also with the mineral acids H,SO and H3P04. A summary of the results is given in Figs. 5 and 6. For Na and Ca it can be deduced from Fig. S(a) that for the thermospray system the carrier gas flow rate at which a maximum signal intensity is obtained is ( a 1 I20 - 100 - 80 - 60 - Carrier gas flow ratelm1 min-' z m 0 Nebulizer gas flow rate/ml min-' Fig.5 Ion signal intensity for '05Tl as a function of (a) carrier gas flow rate obtained with the thermospray system and (b) nebulizer gas flow rate obtained with the pneumatic nebulizer combined with a spray chamber for L0.14 moll-' HN03; 2 1 g I-' Na (NaNO,); and 3 1 g I-' Ca [Ca(NO,),] ( a 1 200 - 0 600 800 700 800 900 1000 Carrier aas flow rate/rnl rnin-' 600 800 700 800 800 1000 Nebulizer gas flow rate/rnl rnin-' Fig.6 Ion signal intensity for '"Tl as a function of (a) carrier gas flow rate obtained with the thermospray system and (b) nebulizer gas flow rate obtained with the pneumatic nebulizer combined with a spray chamber for 1 0.14 mol I-' HNO,; 2 1 g I-' S (H,SO,); and 3 1 g 1-' P (H,PO,) Journal of Analytical Atomic Spectrometry September 1995 Vol. 10 579shifted towards lower values in comparison with 0.14 mol I-' nitric acid (a reference solution). This behaviour is in contrast to that of the conventional arrangement which does not show a shift in optimum flow rate [Fig.5(b)]. In addition it can be seen that the absolute signal intensity is also affected. It decreases by a factor of 7 and 3 by adding respectively 1 g 1-' of Na and 1 g 1-l of Ca to the 0.14 moll-' nitric acid solution while such a decrease is not observed for the conventional arrangement [Fig. 5(b)]. Obviously with a carrier gas flow rate optimized for 0.14 moll-' nitric acid (760 ml min-I) the analyte ion signal intensities in an Na or Ca matrix decrease significantly when using the thermospray system. These observations provide an explanation for the strong signal suppressions caused by Na and Ca described above for the thermospray system.For H2S04 and H3P04 the situation is different as illustrated in Fig. 6(a) and (b). The optimum carrier gas flow rate at which a maximum signal intensity is obtained is shifted towards higher values when adding H,S04 or H3P04 to the 0.14mo11-' nitric acid solution. This can also be observed for the conventional arrangement although it is less pronounced. Also the absolute signal intensities are inversely affected by the mineral acids. For H2S04 there is a slight increase whereas for H3P04 an increase of up to 50% can be observed. The same trend can be noted for the conventional arrangement. Again these plots provide an explanation for the matrix effects caused by H2S04 and H3PO4 the suppressions are lower than those caused by Na or Ca and they are similar for both the conventional and thermospray arrangements.From the study of the matrix effects it can be concluded that the signal suppressions caused by mineral acids (e.g. HzSO4 and H3P04) are not higher for the thermospray system. Indeed the thermospray and the pneumatic nebulizer give similar non-spectroscopic interferences. For these matrices the improvement in sensitivity by a factor of 10 reported elsewhere* is not deteriorated by strong matrix effects. For the signal suppressions caused by salts [e.g. NaNO and Ca(NO,),] however it is clear that for highly concentrated salt solutions (matrix element concentration above 1 g 1-') no improvement in the sensitivity will be observed. For these matrices thermos- pray nebulization is not preferred to pneumatic nebulization unless a matrix separation is included in the sample preparation step.Analysis of Environmental Reference Materials As in general the non-spectroscopic interferences will be more severe for the thermospray system it is necessary to eliminate these matrix effects. In addition to a separation or a dilution of the matrix several procedures can be used to correct for or to reduce or eliminate the non-spectroscopic interferences. For the reversible matrix effects use can be made of the conventional methods such as internal standardization stan- dard additions or isotope dilution. Deposition of salts on the cone and skimmer can be reduced by using a sample flow injection system so that only minimal amounts of sample are introduced into the system.In order to evaluate the correction procedures mentioned above three candidate environmental reference materials were analysed namely BCR CRMs 141R 144R and 146R. All materials were treated with aqua regia and the soluble content of Cd and Pb was determined. Table 2 summarizes the results for Cd obtained with the conventional arrangement and with the thermospray system. In order to correct for the signal suppression observed with the thermo- spray system which varied from 25% for CRM 144R to 40% for CRM 141R the single standard additions method was applied. Rhodium was used as internal standard to correct for the signal fluctuations during the analysis. As can be seen the results for the three materials obtained with thermospray nebulization are not significantly different from those obtained Table 2 Results of Cd determination for three BCR CRMs uiz.CRM 141R Soil-calcareous loam CRM 144R Sewage sludge-domestic and CRM 146R Sewage sludge-industrial TN-ICP-MS* PN-ICP-MSt Certified value$ CRM 141R 13.6k0.35 13.6 0.1 13.96k0.33 CRM 144R 1.79k0.06 1.76 k 0.04 1.84 k0.07 CRM 146R 18.4k0.3 18.1 k0.3 18.45 k0.35 ~ ~ ~ ~ * TN-ICP-MS use of the thermospray nebulization system. t PN-ICP-MS use of the pneumatic nebulizer combined with a spray chamber. $ Ref. 20. 0 Results are expressed in pg g-' with 95% confidence limits (number of samples = 5). with pneumatic nebulization. The Cd concentrations range from 1.79 pg 1-' for CRM 144R to 18.4 pg g-' for CRM 146R. In addition a similar precision on the results can be noticed.The relative standard deviation (s,) varies from 1.6 to 2.8% (five samples) for the thermospray system in comparison with an s of 0.74-2.3% for the pneumatic system. These results demonstrate that standard additions accurately corrects for the signal suppression and that rhodium is a suitable internal standard to correct for the signal fluctuations. For the determination of the aqua regia soluble content of Pb in the three materials external calibration with internal standardization standard additions and isotope dilution were used. The Pb concentrations which are given in Table 3 range from about 51 pg g-' for CRM 141R to about 570 pg g-' for CRM 146R. In general the results obtained with the three calibration methods agree well with each other.The results for CRMs 144R and 146R obtained by external calibration in combination with an internal standard are higher indicating that the internal standard (Tl) does not accurately correct for the signal suppression caused by the matrix. The precision on the results is similar for the three calibration methods. In addition no significant differences between the results obtained by thermospray and pneumatic nebulization were observed. From the analysis of the three reference materials it can be concluded that although thermospray nebulizers give rise to more severe matrix effects accurate and precise results can be obtained for heavily loaded samples by using a suitable cali- bration method. In order to correct for matrix effects standard additions and isotope dilution are to be preferred to external calibration combined with internal standardization.CONCLUSION From this work it is clear that a significant enhancement of the non-spectroscopic interferences is observed for the thermo- spray nebulizer owing to an increased transport of the matrix to the plasma. This is particularly true for salts such as NaNO and Ca(N03) partially owing to clogging of the orifice of the cone and skimmer. The latter effect can only be reduced or eliminated by using a flow injection system or by diluting or removing the matrix. Less severe suppressions are observed for mineral acids such as H2S04 and H3P04. In addition for these solutions the pneumatic and thermospray nebulizers give similar effects. It can be concluded that although the thermo- spray system offers better sensitivity and detection limits it can only be succesfully applied for sample solutions with a relatively low salt content.This is similar to the performance of the ultrasonic nebulizer. Nevertheless when the matrix effects are not excessively high suitable calibration techniques such as standard additions and isotope dilution can be succesfully applied to correct for the signal suppressions. Although there is a higher matrix loading of the plasma the spectroscopic interferences due to an overlap with polyatomic . 580 Journal of Analytical Atomic Spectrometry September 1995 Vol. I0Table 3 Results of Pb determination for three BCR CRMs viz. CRM 141R Soil-calcareous loam CRM 144R Sewage sludge-domestic and CRM 146R Sewage sludge-industrial TN-ICP-MS* PN-ICP-MSt Certified value$ CRM 141R- 51.3 k 2.0 external calibrations 53.0k 1.47 51.8k 1.2 51.3 1 1.4 51.9 kO.8 standard additions isotope dilution 50.5 jl 1.0 52.4 & 1.0 external calibrations 105.7 i.1.6 98.42 1.6 standard additions 96.0 2.5 97.1 k 0.9 isotope dilution 96.1 jll.1 98.1 k 1.2 external calibration5 588.9 k4.3 576.7 k 6.0 standard additions 511i.11 585.0 2 6.2 isotope dilution 569111 572.4 k 9.4 CRM 144R- 96.0k 1.5 CRM 146R- 583i17 * TN-ICP-MS use of the thermospray nebulization system. t PN-ICP-MS use of the pneumatic nebulizer combined with a spray chamber. $ Ref. 20. 4 External calibration was used in combination with internal standardization (TI). 1 Results are expressed in pgg-' with 95% confidence limits (number of samples=5).ions are not higher for the thermospray system if an efficient desolvating system is used. This results in lower apparent analyte concentrations. By using a flow injection system the washing time is not longer than with pneumatic nebulizers when changing from one sample to another in the thermospray system. Future work will involve the use of smaller capillaries to improve further the analyte transport to the ICP and the use of fused silica tubes positioned inside a stainless-steel capillary to avoid contamination for elements such as V Cr Zn and Mo when samples in acidic medium are analysed. In order to remove more solvent and therefore to reduce matrix effects a more efficient desolvating system will be evaluated. Applications of the thermospray system are anticipated in interfacing HPLC with ICP-MS for separation (of the matrix) and speciation purposes.We thank the National Fund for Scientific Research (Belgium NFWO) and the Interuniversity Institute for Nuclear Sciences (IIKW) for financial support. M. P. is indebted to the Institute for Scientific and Technological Research (IWT) for a research fellowship. REFERENCES Thompson J. J. and Houk R. S. Anal. Chem. 1986 58 2541. Montaser A. Tan H. Ishii I. Nam S.-H. and Cai M. Anal. Chem. 1991 63,2660. Wiederin D. R. Smith F. G. and Houk R. S. Anal. Chem. 1991 63 219. Powell M. J. Quan E. S. K. Boomer D. W. and Wiederin D. R. Anal. Chem. 1992 64 2253. 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 Shum S. C. K. and Houk R. S. Anal. Chem. 1993 65 2972. Jakubowski N. Feldmann I. Stuewer D. and Berndt H. Spectrochim. Acta Part B 1992 41 119. Jakubowski N. Jepkens B. Stuewer D. and Berndt H. J. Anal. At. Spectrom. 1994 9 193. Vanhoe H. Moens L. and Dams R. J. Anal. At. Spectrom. 1994 9 815. Koropchak J. and Winn D. H. Appl. Spectrosc. 1987 41 1311. Koropchak J. A Aryamanya-Mugisha and Winn D. H. J. Anal. At. Spectrom. 1988 3 799. Peng R. Tiggelman J. J. and de Loos-Vollebregt M. T. C. Spectrochim. Acta Part B 1990 45 189. Vermeiren K. A. Taylor P. D. P. and Dams R. J. Anal. At. Spectrom. 1988 3 571. Koropchak J. A. and Veber M. Crit. Rev. Anal. Chem. 1992 23 113. Koropchak J. A. Veber M. and Herries J. Spectrochim. Acta Part B 1992 41 825. de Loos-Vollebregt M. T. C. Peng R. and Tiggelman J. J. J. Anal. At. Spectrom. 1991 6 165. Veber M. Koropchak J. A. Conver T. S. and Herries J. Appl. Spectrosc. 1992 46 1525. DIN 38 414-S7 procedure Schlamm und Sedimente (Gruppe S) Aufschluss mit Konigswasser zur nachfolgenden Bestimmung des suureldslichen Anteils von Metallen Germany January 1983. Parent M. Vanhoe H. Moens L. and Dams R. unpublished work. Vanhaecke F. Dams R. and Vandecasteele C. J. Anal. At. Spectrom. 1993 8 433. Quevauviller Ph. Muntau H. Fortunat U. and Vercoutene K. EUR Report Luxembourg Brussels 1995 submitted. Paper 5/01233C Received March I 1995 Accepted M a y 9 1995 Journal of Analytical Atomic Spectrometry September 1995 Vol. 10 581