JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 1994 VOL. 9 815 Thermospray Nebulization as Sample Introduction for Inductively Coupled Plasma Mass Spectrometry Hans Vanhoe Luc Moens and Richard Dams Laboratory of Analytical Chemistry University of Ghent Institute for Nuclear Sciences Proefiuinstraat 86 B-9000 Ghent Belgium A sample introduction system consisting of a thermospray nebulizer and a desolvating unit was coupled to an inductively coupled plasma mass spectrometer. Several parameters were optimized including the sample uptake rate the power delivered to the capillary tube the temperature of the aerosol and of the cooling-water the carrier gas flow rate and the r.f. power. Under optimized conditions the sensitivity obtained with the thermospray system is a factor of 10 higher than with a pneumatic nebulizer combined with a spray chamber. The increase was observed for the following elements over the whole mass range Be Al Sc Co In Gd TI Th and U.Since similar background levels 10-40 counts s-' were observed better detection limits could be achieved (from 0.19 ng I-' for U to 1.3 ng I-' for Be). The relative standard deviation (RSD) on analyte signals for a short-term stability test (10 min) is between 3 and 8%. These values can be improved to less than 4% by the use of internal standardization. No drift in analyte signal during several hours was observed. The levels of oxide (MO+:M+) and doubly charged (M2+:M+) ions obtained with the thermospray system are about a factor of 2.5 lower than those obtained with the pneumatic system. Consequently the oxide ion levels for elements with the highest MO bond strength and the doubly charged ion levels for elements with the lowest second ionization energy are not above 1%.Keywords Inductively coupled plasma mass spectrometry; sample introduction; thermospray nebulization The analytical performance of an inductively coupled plasma (ICP) mass spectrometer is strongly related to the sample introduction system. As expressed by Browner and Boorn,' sample introduction is the Achilles' heel of atomic spectroscopy. Most of the ICP mass spectrometers are equipped with a sample introduction system consisting of a nebulizer and a spray chamber. Most often pneumatic nebulizers are chosen because of their simple construction. The poor nebulization efficiency inherent in pneumatic nebulizers (typically 1-2%) is however a major drawback and restricts the sensitivity of ICP Mass Spectrometry (ICP-MS).Consequently alternative nebu- lizers with an improved nebulization efficiency have been evaluated for introduction of solutions into an ICP-MS system e.g. ultrasonic neb~lizer~.~ direct injection nebulizer4 and hydraulic high pressure nebulizer.' In the past few years the use of thermospray nebulization to increase the efficiency of the sample introduction system was investigated. Originally used as an interface between liquid chromatography and mass spectrometry,6 thermospray nebul- ization was introduced as a sample introduction system for ICP atomic emission spectrometry (ICP-AES) in 1985 by Meyer et aL7 The liquid sample is forced through an electro- thermally heated capillary resulting in partial vaporization of the solvent and production of a fine spray.The aerosol droplet size produced by the thermospray system is much lower than that of a pneumatic system; Koropchak and Winn' reported median diameters for primary thermospray aerosols of typically 2 pm compared with median diameters produced by pneumatic nebulizers of more than 10 pm. This diameter is further reduced as the aerosol moves away from the tip of the capillary because of the high temperature of the aerosol (self-desolvating effect). Small droplets are more efficiently transported to the ICP since they are less prone to impaction and other loss processes. Koropchak et d9 reported an analyte transport efficiency of 53% using a capillary tube with an internal diameter of 50 pm.In addition a more rapid and efficient ionization occurs; small droplets more easily undergo desolvation volatilization and atomization. These features lead to an improved performance. Several authors reported signal-to-noise ratio enhancements in ICP-AES9," resulting in lower detection limits. Vermeiren et a!." reported detection limits for Ag Al Cd Co Cu Pb and Zn that are 12-18 times lower compared with those for a pneumatic nebulizer whereas Peng et a1." observed an improvement of the detection limit for 19 elements by a factor of 3 compared with sample introduction with a V-groove nebulizer. The aim of this study was to optimize and to evaluate the coupling of a thermospray nebulization system with an ICP mass spectrometer. In a preliminary study Meyer et d7 reported the use of a thermospray nebulizer for ICP-MS.They observed 15 times more counts for Ce and Tb compared with a concentric nebulizer. More recently Montaser et aL3 studied several nebulization systems for ICP-MS including thermo- spray nebulizers. The thermospray system used in this work consisted of an LC pump and a stainless-steel capillary tube with an internal diameter of 180 pm. Because solvent transport efficiencies are also high (Schwartz and MeyerI2 measured efficiencies above 50%) a desolvating system must be applied. Therefore the capillary tube was followed by a heated spray chamber and a condenser. The optimization procedure included the power applied to the capillary the sample-uptake rate the temperature of the aerosol and of the cooling-water the carrier gas flow rate and the r.f. power.Important analytical features such as stability detection limits background level and sensitivity are discussed and compared with those obtained with a pneumatic system. Also the levels of oxide and doubly charged ions are given. Experimental ICP-MS Instrument A VG PlasmaQuad PQ1 (VG Elemental Ltd. a division of Fisons Instruments Winsford UK) has been used in all experiments. The original interface was replaced by a high performance interface in order to improve the sensitivity. Details of the operating conditions are given in Table 1. Pneumatic Nebulization System The pneumatic nebulization system with which the thermo- spray system is compared consists of a Meinhard (TR-30-A3) concentric glass nebulizer and a double pass Scott-type spray chamber with surrounding liquid jacket made of borosilicate glass.The water flowing through the spray chamber is supplied816 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 1994 VOL. 9 Table 1 ICP-MS operating conditions Stage Parameter Plasma Frequency Torch Pneumatic Sample uptake rate nebulization Gas flow system Plasma A u x i 1 i a r y Nebulizer Ion sampling Sampling cone Skimmer cone Sampling depth Vacuum Expansion stage Intermediate stage Analyser stage Conditions 27.12 MHz Fassel-type 0.9 ml min - 14 1 min-' 1 1 min-' 0.820 1 min - Nickel 1.0 mm orifice Nickel 0.75 mm orifice 10 mm (from load coil) 2.2 mbar 1.0 x mbar 2.0 x mbar by a recirculating cooling system (Barrington LT6 -20 to 100°C) and is thermostated to within 0.1 "C.The sample is delivered to the pneumatic nebulizer with a peristaltic pump (Gilson Minipuls-2). Under optimum conditions a sample uptake rate and a nebulizer gas flow rate of 0.9 ml min-' and 0.820 1 min- ' respectively are applied. Thermospray Nebulization System The thermospray nebulizer consists of an electrically heated stainless-steel capillary tube of length 30 cm and i.d. 180 pm. Direct electrical heating was used to heat the capillary. This implies that a current in this work an a.c. current (50 Hz) is passed through the capillary. The power delivered to the capillary could be varied continuously between 0 and 200 W (a voltage variable between 0 and 6V) by using a variable transformer in combination with a second main transformer. The aqueous solution is delivered to the thermospray nebulizer by a Varian 8500 LC pump which is a single-syringe pump (volume 250ml) with a pump piston which is moved with a stepping motor via a speed reducer and a sprocket-chain driver assembly.The sample uptake rate can be controlled within 0.017 ml min-' (range 0-16 ml min-'). To avoid solvent overloading of the plasma the thermospray nebulizer must be followed by a desolvating system. Peng et al." employed a conventional cooled spray chamber whereas Montaser et aL3 used a membrane separator. In this work a heated spray chamber followed by a condenser was chosen. This approach is similar to the one used by Koropchak et ~ l . ~ Schwartz and MeyerI2 andElgersma et ~ 1 .' ~ A schematic overview of the desolvating system is given in Fig. 1. It consists of a conical flask in which the thermospray nebulizer is mounted using a PTFE piece. Large droplets which impact on the surface of the flask are removed through a drain at this stage. The conical flask is followed by an L-shaped tube (length 25 cm and i.d. 3.5 cm) which is heated by a heating tape. The use of a variable transformer allows the power to be changed from 0 to 100 W. The temperature which can be raised to about 200"C could be maintained at a constant value (within 1 "C) by insulating the heated tube with asbestos tape. The temperature of the heated aerosol was controlled by a mercury thermometer which was mounted at the end of the heated tube. Finally a modified Friedrichs condenser was used to condense and remove most of the solvent leaving the heated tube.The cooling-water which is supplied by a recirculating cooling system (Barrington LT6) was thermostated to within 0.1 "C. An argon carrier gas stream entering the thermospray system at the stage of the conical flask and controlled by a mass flow controller is used to carry the secondary aerosol towards the ICP. An evaluation using ICP-AES of the stability Thermometer Carrier gas 0-220 v Conical flask Thermospray probe f/ Towards ICP Fig. 1 Schematic overview of the desolvating system and the mass transport efficiency of this thermospray nebul- ization system was described previously in Test Solutions and Optimization Procedure For the optimization procedure a 10 pg 1-' multi-element solution (Be Al Sc Co In Gd T1 Th and U) was used. This solution was prepared in 0.14 mol 1-1 nitric acid from commer- cially available AAS standard solutions.For the short- and long-term stability experiments Li B Cu Cd Cs and Bi (from AAS-standard solutions) were also added to the multi- element solution. To reduce the level of impurities both the water and the nitric acid (14 moll-') were purified respect- ively by a Millipore Milli-Q water purification system (resis- tivity of 18 MQcm) and by a sub-boiling distillation system. These purification procedures give the lowest levels of metal impurities as shown in a previous publication." All results described in this work were obtained using the scanning mode for data acquisition.Normally scan conditions were chosen so that one measurement lasted about 1 min. The following isotopes were selected 7Li 9Be "B 27Al 45Sc 63Cu "Co '14Cd "'In 133Cs ls8Gd 205Tl 209Bi 232Th and 238U. Results and Discussion Optimization of Instrumental Parameters Power applied to the capillary tube and sample uptake rate Since the power applied to the capillary tube determines the vaporized fraction and in this way the signal intensity it is important to optimize the applied power. For the capillary used in this study a thermospray could be produced applying a power between 20 and 90 W. The influence of the applied power and of the sample uptake rate on the analyte signal is illustrated in Fig. 2 for 238U. As can be seen the intensity of the analyte signal is initially proportional to the applied power reaches a maximum at a certain power and decreases at higher powers.The power at which a maximum analyte signal intensity is obtained depends on the sample uptake rate; 46 W for 1 mI min-l 54 W for 1.33 ml min-' and 76 W for 1.67 ml min-l. The sensitivity is only slightly dependent on the sample uptake rate with a higher sensitivity for higher rates. The power settings of 20 and 65 W (sample uptake rate of 1.3 ml min-l) correspond to tip temperatures of the vapor- izer of 55 and 113 "C re~pective1y.l~ The observations described for U were analogous to those obtained for the other elements investigated with masses over the whole mass range. The levels of oxide (MO+:M+) and doubly charged ions ( M2+:M+) are not significantly dependent on the applied power as can be seen in Fig.3 for Th U and Ba (for a sample817 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 1994 VOL. 9 ' \ 2 I 15 30 45 60 75 90 Applied power/W Fig. 2 Ion signal intensity for 238U as a function of the power applied to the capillary for three sample uptake rates A 1; B 1.33; and C. 1.67 ml min-' 2.5 1 3.0 0.5 t 4 0.5 0 ' I I I I 10 15 30 45 60 75 90 Applied powerw Fig.3 The ratios of MO+:M+ for A Th; and B U and C of M2+:M+ for Ba as a function of the power applied to the capillary for a sample uptake rate of 1.33 ml min-' uptake rate of 1.33 ml min-I). Only at high powers (above 70 W) can a small increase be observed owing to the fact that at these powers the signal of M + decreases more rapidly compared with that of MO+ and M2+.The ratios of YO+:Y+ GdO+:Gd+ Th2+:Thf and U2+:U+ (not given in Fig. 3) behave in the same way. Since at very low applied powers (< 35 W) the spray obtained was visibly unstable and at high powers (>80 W) a lot of turbulences in the spray could be observed the sample uptake rate and the applied power were maintained at 1.33 ml min-' and 55 W respectively. Aerosol temperature Heating the produced aerosol will decrease the droplet size and improve the analyte transport. Therefore the influence of the aerosol temperature on the intensity of the analyte signal was studied. The thermospray nebulizer itself produces a warm aerosol; an aerosol temperature of 68°C is obtained at a sample uptake rate of 1.33 ml min-'. The aerosol leaving the spray chamber can be further heated in the L-shaped tube (see Fig.1). As can be seen in Fig. 4 for 59C0 "'In and 238U the analyte signal intensity is strongly dependent on the aerosol temperature measured at the end of the heating system (Fig. 1). The sensitivity at a temperature of 118 "C is a factor of 60 higher than that without external heating. Further heating of the aerosol is not advantageous because the signal intensity decreases slightly. In addition a relatively unstable analyte signal is observed at high aerosol temperatures (above 130 "C). Since by heating the aerosol the amount of solvent reaching the plasma is reduced the levels of oxide and doubly charged ions were studied in detail (Fig. 5). Both levels are only moderately influenced by the aerosol temperature.The ratio of MO+:M+ for Th and U [also for Y and Gd (not given in Fig. 5)] is a factor of 1.6 lower than when no external heating 1200 r 1 I A I 60 90 120 150 180 Temperature (aerosolV'C Fig. 4 Ion signal intensities for C 59C0; B 1'51n; and A 238U as a function of the aerosol temperature (applied power 55 W; sample uptake rate 1.33 ml min-') 2.0 I 1 1.5 I x I 0 ' I I I I ' 0 60 90 120 150 180 Temperature (aerosol)/"C Fig.5 Ratios of MO+:Mf for A Th; and B U and C of M2+:M+ for Ba as a function of the aerosol temperature (applied power 55 W; sample uptake rate 1.33 ml min-') is applied owing to a lower solvent content of the plasma when the aerosol is heated. These observations indicate that most of the solvent is already removed without additional heating of the aerosol.The ratio of M2+:M+ for Ba [also for Th and U (not given in Fig. 5)] increases by about 50% with higher aerosol temperatures probably owing to a higher plasma potential as suggested by Gray et a1.I6 who found that increas- ing the water loading in the plasma induces an increase in the plasma potential. Therefore an aerosol temperature of 120°C was chosen because it combines a high sensitivity with low oxide and doubly charged ion levels. Cooling-water temperature Since the temperature of the cooling-water affects the water- loading of the plasma the influence of the cooling-water temperature on the analyte signal intensity and on the oxide and doubly charged ion levels was investigated. The results are summarized in Fig. 6 and 7. As can be seen the analyte signal intensity is strongly dependent on the temperature of the cooling-water; the sensitivity at a temperature of 1 "C is a factor of 7 higher than that at ambient temperature.These observations are similar to those of Jakubowski et who realized an intensity gain by a factor of 2-5 using a GMK nebulizer in combination with a desolvating system consisting of a heating and cooling unit. Higher levels of oxide (MO+:M+) and doubly charged ( M2+:M +) ions were observed when increasing the temperature of the cooling-water (Fig. 7). By varying the cooling-water temperature from 1 to 30"C the ratios of MO+:M+ for Th and U [also for Y and Gd (not given in Fig. 7)] and M2+:M+ for Ba [also for Th and U (not given in Fig. 7)] are increased by a factor of 2 and 1.4 respectively.Similar results were also found by Tsukahara and Kubota" for a concentric nebulizer818 1200 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 1994 VOL. 9 C /"\ - ?? 900- 2 e "I 600 t. .- m C - $ 300 n 0 10 20 30 Temperature (cooling water)/"(= B' \ I 3 I +\+ I - A - - Fig.6 Ion signal intensities for A 59C0; B '"In; and C 238U as a function of the cooling-water temperature (applied power 55 W; sample uptake rate 1.33 ml min-l; aerosol temperature 120 "C) 2.5 I I + 1.0 0 2 0.5 0 0 10 20 30 Temperature (cooling water)/"C + 0.5 % 0 Fig.7 Ratios of MO+:M+ for A Th; and B U and C of MZ+:M+ for Ba as a function of the cooling-water temperature (applied power 55 W; sample uptake rate 1.33 ml min-'; aerosol temperature 120 "C) combined with a heated glass tube and a modified Liebig condenser.They observed a relation between the increase of the oxide and doubly charged ion levels and the increase of the amount of water vapour reaching the plasma. Vermeiren et al." demonstrated that with the desolvating system described in this work the water-loading increases from 8.8 to 15.0 mg min-' by increasing the cooling-water temperature from 5 to 20°C. In order to get a maximum sensitivity and the lowest oxide and doubly charged ion levels a cooling-water temperature of 1 "C was used. Experiments are now underway using cooling- water temperatures below 0°C in order to improve the sensi- tivity and the oxide and doubly charged ion levels further. Results will be reported in the near future. Carrier gasjow rate In contrast with pneumatic nebulizers thermospray nebulizers have the advantage that the primary aerosol production is independent of the gas flow rate used.In this way the argon flow can be optimized without affecting the aerosol production. The signal response behaviour as a function of the carrier gas flow rate is similar to that obtained with pneumatic nebulizers." Fig. 8 shows that for 9Be '151n and 238U the carrier gas flow rate at which a maximum signal for M+ is obtained is nearly the same namely around 0.820 1 min-'. In this way one particular flow rate can be used for all the elements. The cooling-water and the aerosol temperatures have a great influence on the value of the flow rate at which a maximum signal intensity is obtained (Fig. 9 and 10).These maxima shift towards a lower carrier flow rate with increasing temperature of the cooling-water and with decreasing temperature of the aerosol. These observations can be explained by assuming that both temperatures influence the position of the zone in the plasma where a maximum density of singly charged ions M+ 580 680 780 880 980 Carrier gas flow rate/ml min ' Fig. 8 Ion signal intensities for A 9Be(x5); B "'In; and C 238U as a function of the carrier gas flow rate (under optimized conditions see Table 2) 1200 I I 580 700 820 9 40 Carrier gas flow rate/ml min ' Fig.9 Ion signal intensity for "'In as a function of the carrier gas flow rate for four cooling water temperatures A 1; B 10; C 20; and D 30 "C (under optimized conditions see Table 2) 600 I 580 700 820 940 ' Carrier gas flow rate/ml min Fig.10 Ion signal intensity for "'In as a function of the carrier gas flow rate for three aerosol temperatures A 68; B 120; and C 200°C (under optimized conditions see Table 2) O C C U ~ S . ~ ~ * ~ ~ A decrease of the solvent-loading of the plasma resulting from lowering the cooling-water temperature and increasing the aerosol temperature will move this zone towards the induction coil so that a higher carrier flow rate must be used to make sure that this zone is present in the sampling region of the sampling cone. The absolute signal intensities are also affected they decrease with higher cooling-water temperatures and with lower aerosol temperatures. R f . power The analyte signal intensity is moderately influenced by the r.f.power (between 1200 and 1500 W) as illustrated in Fig. 11 for 59C0 '151n and 238U. A maximum signal intensity is reached819 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 1994 VOL. 9 S I I 0 1150 1250 1350 1450 1550 Ref. powerw Fig. 11 Ion signal intensities for A 59C0; B "'In; and C 238U as a function of the r.f. power (under optimized conditions see Table 2) at an r.f. power of about 1400 W. Higher r.f. powers result in a slight decrease of the signal intensity. The optimum operating conditions of the thermospray nebu- lization system are summarized in Table 2. It can be mentioned that the auxiliary and the plasma gas flow rates did not significantly influence the sensitivity and were set at 1 and 14 1 min-' respectively. Sensitivity Since a thermospray nebulizer produces an aerosol with smaller droplet diameter as compared with the pneumatic nebulizer and therefore more particles are transported to the ICP a higher sensitivity should be obtained.The analyte signal inten- sities for a number of elements using the optimum operating conditions given in Table 2 were compared with those obtained with a concentric nebulizer combined with a spray chamber. A summary of the results obtained for a multi-element solution (each element present at a concentration of 10 pg 1-') is given in Fig. 12. As can be seen the sensitivity obtained with thermospray nebulization is on average a factor of 10 higher compared with that obtained with pneumatic nebuliz- ation. The increase is similar for all elements studied (Be Al Sc Co In Gd TI Th and U) over the whole mass range.The signal intensities obtained with a concentric nebul- izer combined with a spray chamber and with the same Table 2 Optimum operating conditions for the thermospray nebul- ization system Sample uptake rate Power applied to the capillary Temperature of the aerosol Temperature of the cooling-water Carrier gas flow rate Auxiliary gas flow rate Plasma gas flow rate R.f. power 1.33 ml min-' 55 w 120 "C 1 "C 0.820 1 min - ' 11 min-' 14 1 min-' 1350 W 1 XIO" I v) v) C 3 0 > c -. .= 1 x 105 .- CI .- v) c 0) v) I i n 4 I h I" Be A1 Sc Co In Gd TI Th U Element Fig. 12 Comparison of the sensitivity for a number of elements (at a concentration of 10 pg 1-I) obtained with the thermospray system and with the pneumatic system nebulizer combined with the desolvating system used for the thermospray nebulizer did not differ strongly from each other.There was a slight increase in intensity (10%) using the desolvating system probably owing to a more efficient sol- vent removal. Background and Blank Level The background level was constant over the whole mass range and varied between 10 and 40 counts s-' as illustrated in Fig. 13 showing the mass spectrum between 202 and 239 m/z of a solution containing 1Opg I-' of T1 Th and U. These count rates are similar to those obtained with a concentric nebulizer combined with a spray chamber. Experiments have however shown that the blank level is elevated for some elements owing to contamination from the stainless-steel capil- lary tube and/or other components of the thermospray system.Trace amounts of Cr Mn Fe Ni Zn Mo Pb (see Fig. 13) and Bi (see Fig. 13) could be noticed in a mass spectrum of a blank solution containing 0.14 moll-' nitric acid. Further studies will be carried out to reduce or eliminate this blank problem by using a fused silica capillary positioned inside a stainless-steel capillary. Such a capillary was successfully employed by Peng et al.." Detection Limits Table 3 lists the detection limits for Be Al Sc Co In Gd T1 Th and U obtained with the thermospray nebulization system and with the pneumatic nebulizer combined with a spray chamber. These detection limits (3s definition) were obtained by measuring ten times a blank solution (0.14moll-' nitric acid). For each element a small mass range (z 10 m/z) was scanned for about 1 min yielding an integration time of 6 s (m/z)-'.From Table3 it can be concluded that for the 9 elements studied the detection limits measured with the thermo- spray nebulizer are on the average a factor of 10 lower than the corresponding data measured with a pneumatic nebulizer. The reduction of detection limits can be attributed to the 1 o5 10' lo3 - m 1 al Q c 3 v) 102 10 w 0 204 208 212 216 220 224 228 232 236 m/z Fig.13 Mass spectrum between 202 and 239m/z of a solution containing 10 pg 1-' of T1 Th and U Table 3 Detection limits (ng 1-') obtained with the thermospray system and with the pneumatic nebulizer combined with a spray chamber Element Be A1 sc c o In Gd T1 Th U Isotope 9 27 45 59 115 158 205 232 238 Thermos pray nebulization 1.3 0.35 0.29 0.34 0.24 0.83 0.3 1 0.19 0.19 Pneumatic nebulization 12 3.9 2.1 3.7 1.3 9.1 3.9 2.1 1.2820 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 1994 VOL.improved sensitivity which results from the higher nebulization efficiency of the thermospray system. Montaser et a1.,3 who used a thermospray nebulizer combined with a membrane separator reported detection limits that were improved by a factor of up to 20 (an average of 4.6) for 15 elements. The values reported range from 0.5 ng 1-' for Pb up to 8 ng 1- for Cd using an integration time of 3 s (m/z)-'. Stability The stability of the thermospray nebulization system combined with the desolvating system was studied both on short- and long-term basis. The percentage relative standard deviation (Yo RSD) obtained for the short-term stability was calculated for 10 successive scans recorded in 10 min for a 10 pg l-' multi-element solution containing Li B Al Sc Cu Cd Cs Gd Bi Th and U.As can be deduced from the results presented in Table4 the RSDs vary between 3 and 8%. These values are significantly higher than those obtained with a pneumatic nebulizer combined with a spray chamber. The latter values are not above 3%. These observations are similar to those reported by Montaser et aL3 using a thermospray nebulizer with a membrane separator. They measured RSDs which were roughly twice the values measured with a pneumatic system. These short-term fluctuations can however be compensated for by the use of internal standardization. As shown in Table 4 the RSD obtained is less than 4% by the use of a suitable internal standard.The choice of the internal standard depends on the element to be determined.2' No drift in the ion signal intensity during several hours could be observed if the thermo- spray system was first stabilized for about 30 min. Oxide and Doubly Charged Ion Levels A comparison of the MO+:M+ and M2+:M+ ratios of various elements for thermospray and pneumatic nebulization was made. The results are given in Fig. 14 and 15. For both systems they were obtained under optimized conditions (with a maxi- mum ion signal intensity). As can be seen the levels of oxide (MO+:M+) and doubly charged (M2+:M+) ions obtained with the thermospray system are lower than those obtained with the pneumatic system; both are improved with a factor of about 2.5.The ratios (YO) of MO+:Mf range from 0.1 (0.32) for Y to 1.1 (2.5) for U whereas those of M2+:M+ range from 0.3 (0.6) for U to 1.0 (3.0) for Ba (the values given in parentheses are those obtained for pneumatic nebulization). The levels of oxide ions are similar to those obtained with a thermospray-membrane separator system by Montaser et aL3 They reported MOf:Mf values (YO) for Y and Ce of 0.1 and 1.3 respectively. The levels of doubly charged ions reported Table 4 Percentage relative standard deviations (O/O RSD) for a short- term stability experiment (ten successive measurements of a 10 pg I - ' multi-element solution recorded in 10 min) obtained with thermospray nebulization (TN) system and with the pneumatic nebulizer (PN) combined with a spray chamber TN without internal standard 5.3 4.7 6.7 4.1 5.7 7.9 3.1 5.9 3.9 4.2 4.6 PN 2.9 2.1 2.9 1.5 2.9 2.9 2.1 2.9 I .9 2.2 2.1 3.0 Thermospray nebulizer 0.5 Y Gd Th U Element Fig.14 Comparison of the MO+:M+-ratios for Y Gd Th and U obtained with the thermospray system and with the pneumatic system 3'5 I 3.0 t " I I Ba Thermospray nebulizer 0 Pneumatic nebulizer n Th U Element Fig. 15 with the thermospray system and with the pneumatic system Comparison of the M2+:M+-ratios for.Ba Th and U obtained by Montaser et aL3 are higher than those obtained in this work; the Ce2+:Ce+ measured was 6.0%. They attributed these higher values to the use of a 40.68 MHz ICP (instead of a 27.1 MHz plasma). Although the thermospray system gives higher solvent transport efficiencies the levels of oxide (MO +:M+) and doubly charged ( M2+:M +) ions are still lower compared with those obtained with a pneumatic system owing to the more efficient solvent removal by the desolvating system described in this work than in a conventional spray chamber.Studies are underway to improve the desolvating unit further. Conclusion It was demonstrated that under optimized conditions the thermospray nebulizer with a desolvating system gives a better performance in comparison with the pneumatic nebulizer. Sensitivity and detection limits are on average a factor of 10 improved when thermospray nebulization is employed because of the higher nebulization efficiency of the thermospray nebulizer. Also the levels of oxide and doubly charged ions relative to the levels of the singly charged ions are reduced by a factor of 2.5 so that the oxide ion levels even for elements with the highest MO bond strength and the doubly charged ion levels for elements with the lowest second ionization energy do not exceed 1 YO.Although thermospray nebulization is more susceptible to short-term fluctuations than pneumatic nebuliz- ation it was demonstrated that internal standardization could compensate for these variations. In this way on a short-term basis RSDs of less than 4% could be obtained. The stability on a long-term basis was similar for both systems if a warming- up period of 30 min was employed for the thermospray nebulizer. The performance of thermospray nebulization for the analy-JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY AUGUST 1994 VOL.9 82 1 sis of solutions with a high salt content (serum sea-water) and the alleviation of memory effects by flow injection will be studied. Future work will also involve the use of other non- metallic capillary tubes to avoid blank problems when samples in acid medium are analysed. Studies will be undertaken to improve the desolvating system in order to reduce spectral interferences caused by oxide and doubly charged ions further. 1 2 3 4 5 6 7 8 9 10 References Browner R. F. and Boom A. W. Anal. Chem. 1984 56 786A. 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. Houk R. S. Anal. Chem. 1991 63 219. Jakubowski N. Feldmann I. Stuewer D. and Berndt H. Spectrochim. Acta Part B 1992 47 119. Blakely C. R. and Vestal M. L. Anal. Chem. 1983 55 750. Meyer G. A. Roeck J. S. and Vestal M. L. XCP Inf. Newsl. 1985 10 955. Koropchak J. A. and Winn D. H. Appl. Spectrosc. 1987,41,1311. Koropchak J. A. Aryamanya-Mugisha H. and Winn D. H. J. Anal. At. Spectrom. 1988 3 799. Peng R. Tiggelman J. J. de Loos-Vollebregt M. T. C. Spectrochim. Acta Part B 1990 45 189. 11 12 13 14 15 16 17 18 19 20 21 Vermeiren K. A. Taylor P. D. P. and Dams R. J. Anal. At. Spectrom. 1988 3 571. Schwartz S. A. and Meyer G. A. Spectrochim. Acta Part B 1986 41 1287. Elgersma J. W. Balke J. and Maessen F. J. M. J. Spectrochim. Acta Part B 1991 46 1973. Vermeiren K. A. Taylor P. D. P. and Dams R. J. Anal. At. Spectrom. 1987 2 383. Vanhoe H. Dams R. and Versieck J. J. Anal. At. Spectrom. 1994 9 23. Gray A. L. Houk R. S. and Williams J. G. J. Anal. At. Spectrom. 1987 2 13. Jakubowski N. Feldmann T. Stuewer D. Spectrochim. Acta Part B 1992,47 107. Tsukahara R. and Kubota M. Spectrochim. Acta Part B 1990 45 581. Vanhaecke F. Vandecasteele C. Vanhoe H. and Dams R. Mikrochim. Acta 1992 108 41. Vanhaecke F. Dams R. and Vandecasteele C. J. Anal. At. Spectrom. 1993 8 433. Vanhaecke F. Vanhoe H. Dams R. Vandecasteele C. Talanta 1992 39 737. Paper 4100873A Received February 2 1994 Accepted April 13 1994