899 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1994 VOL. 9 Development of a High Liquid Flow Thermospray Sample Introduction System for Inductively Coupled Plasma Atomic Emission - Spectrometry* Invited Lecture J. A. Koropchak and T. S. Conver Department of Chemistry and Biochemistry Southern lllinois University Carbondale IL 62907 -4409 USA This work describes preliminary results of research devoted to the development of thermospray sample introduction systems that are capable of operating with liquid flow rates of 10 ml min-' or more. This development involved the modification of a thermospray power supply to provide about 1 kW of power and longer vaporizers to increase the residence time of the liquid stream within the heated portion of the system. Use of this system with ICP-AES resulted in sensitivity increases of about a factor of two when sample flow rates were increased from 2 to 5 ml min-'.Prospects for further improvement are also described. Keywords thermospray; sample introduction; inductively coupled plasma atomic emission spectrometry; high sample flow Liquid sample introduction (SI) for detection by inductively coupled plasma atomic emission (ICP-AES) or mass spec- trometry (ICP-MS) is most commonly performed using an aerosol technique typically based upon the use of a pneumatic nebulizer. These SI systems are widely used and provide many practical benefits including simple robust operation and resultant analytical measurements that are accurate and pre- cise. At the same time however it is well known that optimized pneumatic SI systems are inherently sample wasteful.Typically only about 1 % of the analyte input at 1 ml min- will actually reach the plasma; the rest goes to waste within the associated spray chamber.'-3 This aspect reduces substantially the flux of analyte input to the plasma and the signal per unit analyte input that might be achieved. For typical analyte flow rates this proportionally worsens the sensitivity and limits of detec- tion (LODs) that might be achieved. For applications where greater sensitivity is required this limitation presents a signifi- cant disadvantage. High performance alternatives to pneumatic SI for ICP spectrometry are typically based on aerosol generation pro- cesses which do not rely on energy from the argon carrier stream.The two primary examples are SI systems based on ultrasonic ( USN)1,2,4 and thermospray (TSP) nebulizer^.^ Hydraulic high pressure nebulization6 and monodisperse dried microparticulate injection7 are more recent additions to the family of aerosol devices which do not rely on energy from an argon gas jet. With TSP the sample is pumped through an electrothermally heated capillary. If the temperature is high enough to vaporize part of the liquid a jet of solvent vapour along with the remaining liquid will exit the capillary resulting in an aerosol. Primary aerosol droplets from TSP nebulizers decrease in size as the degree of vaporization increases and for optimum analytical operation they have been shown to be smaller than those from pneumatic nebulizers.' Further ana- lyte is preconcentrated due to the nature of the thermospray process and TSP aerosols are hot thereby enhancing solvent evaporation and droplet shrinkage.Coupled with desolvation these factors lead to much higher analyte transport than can be obtained with pneumatic SI and values as high as ~ 5 0 % for sample flows of the order of 1 mlmin-1 have been reported.' One aspect of the thermospray systems which significantly affects performance is the capillary diameter with smaller diameters providing higher transport and lower LODS.'?~' It *Presented at the 1994 Winter Conference on Plasma Spectro- chemistry San Diego CA USA January 10-15 1994. has also been shown that the higher performance of smaller capillaries can be obtained with only the exit of the capillary being of the smallest diameter.' This allows relatively high sample flow rates (1-2 ml min-l) with small apertures (25-50 pm) for high performance.A recent version of this type of system from our laboratory employs fused silica capillaries which have small diameter apertures laser fused to the exit," allowing relatively high sample flows (1-2 ml min-l) without the high pressures exhibited by long capillaries of small internal diameter." With high performance fused silica aperture ther- mospray (FSApT) systems using 50 pm apertures LOD improvements typically between 15-20 times lower than pneu- matic SI are obtained." Fused silica also provides a more chemically robust wetted element than stainless steel and these systems provide stability and low background with sample matrices common to elemental analysis." Furthermore the fact that aerosol characteristics are tuneable with TSP has been shown to allow reduction of matrix interferences.12 Direct comparison of USN and TSP systems has shown TSP to provide somewhat higher sensitivity lower LODs and lower matrix interference^.'^ A commercial system based on FSApT for ICP-AES was introduced at the 1994 Pittsburgh Conference by Leeman Labs.One reason for the inefficiency of pneumatic sample intro- duction is that the flow of gas used to generate the aerosol also carries the aerosol into the plasma and for optimal ICP operation this flow is limited to about 1 1 min-l. However aerosol quality is also determined by this flow rate which provides the energy for aerosol generation.In general aerosol characteristics for SI into atomic spectrometry instruments are poorer (Le. average droplet sizes are larger) as the gas-to- liquid flow ratio is reduced.14,15 The TSP aerosol generation may be thought of as a pneu- matic process involving the use of solvent vapour as the nebulizing gas.5 A difference between a typical pneumatic nebulizer and TSP lies with the fact that the nebulizing gas with TSP is condensable (typically water). In principle large flows of solvent vapour much larger than those employed with pneumatic nebulizers could be used with thermospray systems as long as sufficient energy for evaporation of the solvent and sufficient means for removing the solvent vapour prior to the ICP are provided. As roughly 1 1 min-' of gas would be generated from 1 ml min- of liquid vaporized liquid flows of the order of 10mlmin-I would provide a gas flow rate ten times higher than the argon flow rates at which conventional pneumatic nebulizers typically operate.Vaporization of these high liquid flows could either be used900 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1994 VOL. 9 with a conventional pneumatic approach (thermojetspray) or directly with high sample flows. In the latter case if high transport efficiency can be maintained with the high sample flows direct improvements in sensitivity can be envisioned generally on the basis of the higher analyte mass flux that would result. For applications which are not typically sample limited (e.g. environmental) this presents a feasible approach to improved LODs.However conventional triac-controlled power supplies and vaporizers for thermospray as used in our laboratory and commonly for LC-MS,I6 are limited to aqueous liquid flow rates of only 1-2 ml min-'. In this paper the development of thermospray power sup- plies and vaporizers capable of vaporizing continuous flows of up to 10 ml min-' of water are described. Further preliminary results for optimizing such systems for direct introduction of samples at these high flow rates including analytical perform- ance data are given. In addition prospects for the application of such vaporizer systems with other aerosol generation approaches are also discussed. Experimental The instrument employed for these studies was a Leeman Labs (Lowell MA USA) ICP 2.5 operated at 1 kW with a coolant flow of 17.51min-l an auxiliary flow of 0.51min-' and a nebulizer flow optimized for compromise multi-element analy- sis with a 10 mg ml-' Ni solution to flow rates ranging from 0.56 to 0.65 1 min-' for sample uptake rates ranging from 2 to 5 ml min-' respectively.The carrier gas flow rate was adjusted with a Tylan (Carson CA USA) Model FC-260 mass flow controller. A Leeman Labs Plasma Spec rapid sequential Cchelle spectrometer was used for wavelength selection. The wavelengths employed for this study were Cu I at 324.754 nm Ni I1 at 231.604 nm Pb I1 at 220.353 nm Se I1 at 196.026 nm and T1 I1 at 190.801 nm. Liquid samples were introduced to the thermospray vapor- izer using a flow injection (FI) mode. The FI system flow was established using either a Dupont (Wilmington DE USA) Model 870 HPLC pump or an Autochrome (Berlin Germany) Model M500 HPLC pump.The de-ionized distilled water (DDW) carrier stream was pumped into an SSI (State College PA USA) pulse dampener and then to a Rheodyne (Cotati CA USA) Model 3725 HPLC injector fitted with a 10ml PEEK sample loop. A 2 pm metal free filter was located in the flow stream between the injector and the thermospray vaporizer. Modifications were made to a standard Vestec (Houston TX USA) thermospray power supply which consists of a triac based control circuit and a transformer that outputs 6 V a.c. to heat the vaporizer probe.16 This standard power supply is capable of providing up to about 150 W of power.16 However our goal was to be able to vaporize up to 10mlmin-1 of water.The power required for this is in the order of 440 W and to account for inefficiencies in power utilization power supplies capable of supplying about 1 kW of power were chosen. In order to increase the power output of the Vestec power supply the first modification made was to replace the original 120 V a.c. input transformer having a 6 V a.c. output with a larger transformer (Square D Company Model 1S43F) having a 12 V a.c. output and capable of delivering 1 kW of power. However the vaporizer resistance affected how much current could be drawn limiting the amount of heat dissipated by the vaporizer. The overall result from this first modification was that only 3-4 ml min-' of water could be vaporized. One way to increase the amount of heat dissipated in the vaporizer would be to raise the vaporizer resistance which was con- sidered impractical.Another way is to increase the voltage applied to the probe which would increase the current drawn thereby increasing the amount of power available. To do this the 120 V ax. input to the transformer was increased to 240 V a.c. by placing a second transformer (Square D company Model 2SlF) in between the output of the triac and the existing 120 V a.c.-12 V a.c. transformer. This second trans- former serves to step the output voltage of the triac from I20 V a.c. to 240 V a.c. By inputting 240 V instead of 120 V the output is increased from 12 V to 24 V resulting in a higher current draw by the vaporizer which increases the amount of power drawn and the heat dissipated by the unit.Fig. 1 compares the conventional and final high power TSP systems. 'The result of this modification was that the power output of the probe was increased to 700 W and the probe could vaporize in excess of 10 ml min-' of water. The thermospray vaporizer used in this study was con- structed from stainless steel. The probe tubing was 30 cm long and had dimensions of 1/16 in 0.d. by 100 pm i.d. To serve as additional heating for the liquid a 1 m length of 500 pm stainless-steel tubing was connected to the entrance end of the TSP probe and was also heated with the TSP power supply (see Fig. 2). This additional tubing was typically coiled as was most of the length of the vaporizer resulting in a compact assembly. Temperature feedback was accomplished by spot welding a J-type thermocouple to the pump end of the 500 pm tubing. Aerosols generated by these high flow thermospray vaporiz- (a) Thermocouple (temperature feedback) To vaporizer 1 To vaporizer 1 5 o w - = t l controller Triac Transformer - Potentiometer 120 V input 6 V output Thermocouple (temperature feedback) To vaporizer Triac controller Potentiometer Tra nsf o rmer 240 V input 24 V output 1 kVA Transformer 120 V input 240 V output 2 kVA Fig.1 Schematic diagrams of thermospray power supplies; (a) standard TSP power supply and (b) modified TSP power supply Length =30 crn 0.d. =A in i.d.= 100 pm t 1 & in Union J a T pow:r lead Control $Length = 1 m thermocouple o.d.=$ in Power lead i.d. =500 pm 1 t Sample from Injector Fig.2 High flow rate thermospray vaporizer; all tubing is stainless steel power leads are 0 gauge copper wire and the thermocouple is a J typeJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1994 VOL. 9 90 1 ers were input to our standard desolvation ~ y s t e m ~ 3 ~ consisting of a cylindrical spray chamber maintained at 150 "C by an Omega (Stamford CT USA) Model CN-9000A temperature controller followed by a Friedrich's condenser chilled to - 5 "C by coolant from a Forma Scientific (Marietta OH USA) Model 2095-2 refrigerated recirculating bath. All solutions were prepared by serial dilution of Leeman Labs Plasma Pure Standard stock solutions using DDW. The carrier stream for these studies was DDW. Results and Discussion Essentially all previous studies have shown that the perform- ance of TSP SI systems is strongly dependent on the operating temperature of the vaporizer.' The operating temperature determines the degree of vaporization aerosol characteristics transport efficiency and LODs obtained with the system.Typically increasing temperatures provide increasing signals until a peak' or plateau" is achieved followed by a steady decline in signal and performance at higher temperatures. The provision of either a peak or plateau signal versus temper- ature profile is indicative of sufficient power availability for the sample flow (composition and flow rate) being input. Observation of profiles where the maximum signal is achieved with the power control maximized can be taken as an indi- cation that there may not be sufficient power being input to the sample flow to optimize the vaporization/aerosol character- istics of the system. Previously the effects of sample flow rate on sensitivity have been studied'?" with conventional TSP systems (ie.conventional vaporizers and power supplies limited to about 150 W input power) and optimum signals were obtained for sample flows between 1.5 and 2.0 ml min-'. Also the operation with higher sample flows would not allow the achievement of a peak or plateau signal versus temperature profile. These observations suggest that insufficient power was input to the flow system to optimally vaporize these higher sample flows with the conventional TSP systems. There are at least two possible reasons for this lack of power the power supply limit is insufficient to optimally vaporize the liquid; and/or the available power is not efficiently coupled into the flow stream. The power supplied by conventional LC-MS power supplies (e.g.150 W with Vestec supplies) would be able to vaporize over 3 ml min-' assuming all of the power is coupled into the flow stream. As some of the power is certainly lost during operation of a TSP vaporizer (e.g. to the air and other components in contact with the vaporizer) it is likely that the previous flow limits observed resulted from lack of available power from the power supply. This. is further suggested by the fact that at higher flow rates RSDs degraded," probably because of lack of regulation by the power supply at full power. To compensate for this lack of power new power supplies capable of providing over 1 kW of power were devised as described under Experimental.However full development of a plateau in the signal with this power supply and a standard 30 cm TSP vaporizer was nat-passible fur sample flm rates above 3 ml min-'. Consequently although sufficient power is available this power was not efficiently coupled into the liquid flow system. One reason for this inefficient power coupling is that higher flow rates result in higher flow velocities and lower residence times for the liquid within the heated capillary. As a result the heat transfer rate through the stainless steel becomes a limiting factor. To overcome this limitation the heated portion of the vaporizer was increased by adding a 1 m length of 500 pm stainless-steel tubing at the entrance to the capillary.Power was now input at the inlet of this tube and the control thermocouple was also located at this point. Using this combi- nation of the higher wattage power supply and the longer vaporizer signal versus temperature profiles as depicted in Fig. 3 for a 4ml min-l sample flow rate were obtained. A d I B D 25 30 35 40 45 50 55 60 Control temperat ure/"C Fig. 3 A copper; B selenium C nickel; D thallium; and E cadmium Signal uersus vaporizer control temperature for high flow TSP clear plateau is observed indicating that the energy input is sufficient to optimize the aerosol characteristics with this flow. In general signals for all elements tested provided similar response curves such that a single temperature could be chosen to reasonably optimize signals for all elements.Similar profiles were obtained for flows up to 5 ml min-l. Optimized operating conditions for flow rates ranging from 2 to 5 ml min-l are indicated in Table 1. As indicated optimal temperatures rose with increasing sample flow rate while optimal carrier gas flow rates declined. The latter feature exemplifies the advantage of SI systems where the aerosol formation process is not controlled by the carrier argon flow rate such as TSP or USN with the capability to optimize the carrier argon flow rate independently of the aerosol formation process. At the same time the pump pressure rose with increasing flow rate as expected. With 5 ml min-' this press- ure began to approach the limits of the pump employed consequently higher flow rates were not studied.However this operating pressure can be reduced by decreasing the length of the capillary region of the vaporizer which provides most of the pressure drop with the current design. Previous results have shown that only the exit of the vaporizer need be of small diameter and most designs of high performance TSP systems (for ICP-AES and LC-MS) have employed apertures at the exit of the capillary to allow relatively high flow operation (1-2 ml min-') with small vaporizer exit diameters (25- 75 J A ~ ) . ' ? ' ~ . ~ ~ Vaporizers employing shorter exit capillaries than the 30 cm of 100 pm tubing used in this study which is the primary source of flow resistance and pressure drop with this vaporizer are currently under development. (Note With such a vaporizer we have recently been able to obtain optimized operation at sample flow rates up to 9 ml min-'.Using the conditions in Table 1 calibration data for a series of elements were obtained. Calibration curves for lead obtained a1 flow rates ranging from 2-5 ml min- have slopes of the lines increasing by over a factor of two in going from 2 to 5mlmin-' or close to the increase in mass input for this change in flow rate. Table 2 provides more detailed data for the effects of flow rate. The data in Table 2 are averages of 3 sets of data collected on separate days. Relative standard deviations (RSDs) for the calibration curves and backgrounds for the 3 sets of data averaged to 3% with a range from 0.5 to 10%. A measure of background noise is given as %RSD in Table 2.These values are based on three replicate measure- ments of a blank solution and are also averages of 3 data sets. All correlation coefficients for the linear calibrations were 20.995. Sensitivity is increased by about a factor of 2 in going from 2 to 5 ml min-l for all of the elements except Cu. At the same time background levels also generally increase. On the other hand flow rate seems to have no effect on background902 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1994 VOL. 9 Table 1 High flow thermospray optimum conditions ICP power/kW 1 Argon gas flow rate/l min-' Outer 17.5 Intermediate 0.5 Spray chamber temperature/"C 150 Condenser temperature/"C -5 Sample flow rate/ Argon carrier gas TSP control temperature/"C Pump Pressure/MPa 2 0.645 48 10.4 3 0.600 48 14.0 4 0.570 50 17.6 5 0.555 53 23.5 ml min- ' flow rate/l min-' Table 2 ical parameters High flow thermospray effects of sample flow rate on analyt- Sample flow rate/ml min-' Analytical parameters 2 3 4 5 Pb (220.4 nm)- Slope/counts g pg-' 3.0 x lo5 3.5 x lo5 4.6 x lo5 6.0 x lo5 Background/count s 26620 30751 32736 36409 RSD(%) 1.7 0.6 1.5 0.07 Se (196.0 nm)- Background/counts 9703 10256 10716 11308 RSD(%) 1.5 1.7 1.8 0.8 Slope/counts g pg-' 1.7 x 104 2.0 x 104 2.5 x 104 3.3 x 104 Tl(l90.8 nm)- Slope/counts g pg-l 9.2 x lo3 1.1 x lo4 1.2 x lo4 1.6 x lo4 Background/counts 9070 9511 9696 10590 RSD(%) 1.1 0.5 1.9 1.7 Cu (324.8 nm)- Slope/counts g pg-' 1.7 x lo6 1.8 x lo6 2.0 x lo6 2.1 x lo6 Background/counts 40283 48520 50021 55971 RSD (Yo) 0.5 0.6 0.7 1.1 Ni (231.6 nm)- Background/counts 7443 6849 6674 7282 Slope/counts g pg- ' 1.9 x 105 2.1 x 105 2.6 x 105 3.3 x 105 RSD(%) 1 .o 2.7 2.0 3.3 noise.As indicated earlier higher noise levels with increasing flow can indicate marginal power availability and poor tem- perature regulation. It appears that with this new TSP system this problem is no longer evident. Aerosol plumes produced with the high sample flows were visually larger (e.g. wider and extending further from the vaporizer tip) than those for our previous TSP systems. As a result the spray chamber/desolvation apparatus employed for the current studies may be less than optimum. Future studies will include more detailed investigation of the effects of this portion of the apparatus.Conclusions Signal increases can be obtained which nearly coincide with increases in sample flow rate with the thermospray vaporiz- ation system suggesting that high analyte transport can be obtained with high sample flow rates providing higher analyte mass flux to the ICP. As such this approach may be used to directly increase the sensitivity and likely improve LODs compared to those that are already obtained with TSP SI. Ultimately improvements in LODs of perhaps a factor of 100 compared to pneumatic SI are achievable. Most recently a new high flow vaporizer was designed in this laboratory that allows thermospray generation and optimization at flow rates of up to 9 ml min-'. This new design also shows similar trends as presented in this work and further optimization is continu- ing.Efforts devoted to optimizing the vaporizers (e.g. capillary i.d. capillary length) and desolvation apparatus for this system for best LODs and toward the development of more chemically inert versions comparable to our FSApT systems are also in progress. Furthermore vaporization of these higher liquid flows will also allow the development of pneumatic nebulizers which employ high flows of solvent vapour as the nebulizing gas an approach we call thermojetspray. For example complete vapo- rization of 10 ml min-' of water will result in approximately 10 1 min-' of water vapour. Operation of a pneumatic nebulizer with such a gas flow would provide much higher gas-to-liquid flow ratios than can be obtained with conventional pneumatic nebulizers for ICP-AES.The reason that this ratio and the energy input to aerosol formation can be increased is that with the thermojetspray approach the nebulizing gas is con- densable and can be removed thus avoiding the ICP-imposed gas input limitation. Ideally improved aerosol characteristics and analytical performance would result. Further this approach may allow the use of thermospray-like devices with samples such as slurries that allows self-continuous aspiration and would not require the high sample consumption of the high flow rate approach. The financial support of Chemical Waste Management Inc. and the Hazardous Waste Research and Information Center a Division of the Illinois Department of Energy and Natural Resources (Grant #HWR 92097) and the loan of equipment by Leeman Labs during the completion of this review are greatly appreciated.Assistance from Vestec Inc. in the form of equipment loans and valuable discussions (especially with Marvin Vestal Cal Blakley and John Wilkes) is also greatly appreciated. 1 2 3 4 5 6 7 8 9 10 References Browner R. F. in Inductively Coupled Plasma Emission Spectrometry ed. Boumans P. W. J. M. Wiley New York NY 1987 vol. 2 ch. 8. Gustavsson A. G. T. in Inductively Coupled Plasmas in Analytical Atomic Spectrometry eds. Montaser A. Golightly W. VCH New York 2nd edn. 1992 ch. 15. Koropchak J. A. Spectroscopy 1993. 8 20. Fassel V. A. and Bear B. R. Spectrochim. Acta Part B 1986 41 1089. Koropchak J. A. and Veber M. Crit. Rev. Anal. Chem. 1992 23 113. Jakubowski N. Feldman I. Stuewer D. and Berndt H. Spectrochim. Acta Part B 1992 47 119. French J. B. Etkin B. and Jong R. Anal. Chem. 1994 66 685. 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. de Loos-Vollebregt M. T. C. Tiggelman J. J. Bank P. C. and Degraeuwe C . J. Anal. Atom. Spectrosc. 1989 4 213.JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY SEPTEMBER 1994 VOL. 9 903 11 Koropchak J. A. Veber M. and Herries J. Spectrochim. Acta Part B 1992 47 825. 12 Koropchak J. A. Veber M. Conver T. S. and Herries J. Appl. Spectrosc. 1992 46 1525. 13 Koropchak J. A. Coleman G. N. and Conver T. S. 20th Annual Conference of the Federation of Analytical Chemistry and Spectroscopy Societies (FACSS) Abstract No. 661 Detroit MI USA 1993. Browner R. F. Canals A. and Hernandis V. Spectrochim. Acta Part B 1992 47 659. 14 15 Sharp B. L. J. Anal. At. Spectrom. 1988 3 613. 16 Vestal M. L. and Fergusson G. Anal. Chem. 1985 57 2373. 17 McLean M. A. Vestal M. L. Vestal C. H. Allen M. H. and Field F. A. Proceedings of the 38th American Society for Mass Spectrometry Conference Tucson AZ 1990 p. 1138. Paper 4/00764F Received February 8 1994 Accepted March 29 1994