JOURNAL O F ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL. 9 415 Multi-element Detection Using Second Surface Trapping With Electrothermal Vaporization Mass Spectrometry* Invited Lecture Andrew J. Scheie and James A. Holcombet Department of Chemistry and Biochemistry The University of Texas at Austin Austin TX 78772 USA A new technique for electrothermal vaporization mass spectrometry using second surface trapping is presented. The method involves atmospheric pressure vaporization of the analyte from a graphite cup and condensation of the vapour onto a cooled tantalum surface. The tantalum surface is introduced into a vacuum chamber through a series of differentially pumped seals and is positioned under a quadrupole mass analyser and radiatively heated using a 250 W filament. An electron impact ionizer (70 eV 3 mA) is employed for ionization.The system allows thermal pre-treatment of the sample in the cup prior to vaporization and trapping thereby enabling the analytes studied (Ag Cd and Pb) to be trapped and simultaneously detected in their elemental forms. Detection limits for the current system for the elements studied are in the low nanogram range. Keywords Electrothermal vaporization mass spectrometry; second surface trapping; silver; cadmium; lead Inductively coupled plasma combined with detection by mass spectrometry (ICP-MS) relies upon solution nebulization for sample introduction and often requires several millilitres of sample for analysis. While electrothermal vaporizers have been interfaced to mass spectrometer^'-^ and commercial units are now available studies have shown that interferences exist even with simple alkali metal salt mat rice^.^.^ Both instruments use the relatively expensive ICP as an ionization source and because of the attendant need of atmospheric pressure sampling through sampling and skimmer cones they also require sub- stantial pumping stations.Coupling the electrothermal vaporizer to the mass spec- trometer without the ICP ionizer could retain the advantage of handling microsamples while minimizing instrument cost and operating expenses. Also if properly designed it could minimize analyte transport problems and reduce pumping speed require- ments. Several studies”16 have used an electrothermal vaporizer heated within a vacuum under a quadrupole as a diagnostic tool for studying atomization processes in electrothermal atom- izers typically used for electrothermal atomic absorption spec- trometry (ETAAS).Likewise Styris and Redfield17y18 have interfaced an electrothermal vaporizer operated at 1 atm ( 1 atm = 101.325 kPa) to a mass spectrometer through skimmer cones to produce molecular beam sample introduction. For diagnostic purposes both approaches have been very successful. However direct vaporization in vacuum fails to take advantage of the excellent atomization efficiencies shown for ETAAS because it lacks the high temperature collision gas that ensures formation of analyte in the elemental form. Even the molecular beam sampling displays a number of analyte- containing molecular fragment^.^',^^ Hence a variety of mol- ecular species are typically detected whose signal intensities are strongly matrix dependent which is detrimental when considering this approach as a quantitative analytical tool.Nonetheless quantitative results are possible for simple ~amp1es.l~ With the molecular beam approach the skimmer cone samples only a small fraction of the sample vaporized from the surface and even this amount is mixed with a much larger amount of sheath gas; therefore sensitivity may not be optimal. A means of vaporizing the sample directly under the entrance to the mass spectrometer without an atmosphere-to- vacuum interface is desirable. * Presented at the XXVTII Colloquium Spectroscopicum Inter- nationale (CSI) Post-Symposium on Graphite Atomizer Techniques in Analytical Spectroscopy Durham UK July 4-7 1993.To whom correspondence should be addressed. Falklg addressed the possibility of quantitative analysis using atomic beam time-of-flight mass spectrometry (AB-TOFMS). While Falk’s research presented preliminary results with no verification of quantitative results it succinctly discussed the possibilities of improving detection limits by minimizing losses in each stage. It was not suggested however that AB-TOFMS can provide the reproducible atomization capabilities demon- strated for more conventional ETAAS or that it can minimize or eliminate the possibility of interferences or matrix effects which depending on the sample analysed produce variations in the species vaporized. Holcombe and c o - w ~ r k e r s ~ ~ ~ ~ have shown that the use of a second surface atomizer (with AA detection) has the capability of efficiently trapping the analyte on a cooled insert placed within the furnace and that this trapped material could be easily released with a very high atomization efficiency upon re-heating of this surface inside a pre-heated furnace operated at 1 atm.A modification of this trapping method has also been successfully employed by H o c q ~ e l l e t . ~ ~ This general trapping approach also has the benefit of eliminating chemical and spectroscopic interferences since many of the gaseous matrix components and decomposition products of common samples are not significantly trapped on the probe. Furthermore it has been shown that the trapping of a more volatile but condensable matrix component could be prevented2’ and that the analyte captured is dispersed on the second surface21 rather than being present as large crystals or droplets that can be found after simple desolvation of a sample placed on a graphite surface (see for example ref.26). This preliminary decomposition and ana- lyte dispersal should eliminate many of the gaseous products observed which were a result of sample decomposition in earlier electrothermal atomization M S studies. A system has been constructed which incorporates electro- thermal vaporization (ETV) MS and second surface trapping. It contains a rapid atmosphere-to-vacuum sample introduction system a software program which will permit data acquisition under two operational modes (full mass scan and ion hopping) a trapping surface which can be cooled efficiently for trapping and heated rapidly for vacuum desorption and a means of heating the trapping surface within the vacuum.Experimental Apparatus The apparatus used for these experiments is designed to provide sample vaporization in graphite cups at 1 atm conden-416 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL. 9 G Tan ta I u m second Controlled temperature power supply Fig. 1 Schematic diagram of tantalum second surface trap positioned for analyte trapping sation of the analyte onto a cooled tantalum second surface rapid introduction of the trapped analyte into high vacuum and thermal desorption of the trapped analyte under a quadru- pole mass analyser for mass spectral analysis. Details of the 1 atm vaporization and trapping are shown in Fig.1. With the tantalum trap still in vacuum the sample solution is deposited into the graphite cup within the atomiz- ation chamber. Sample drying and thermal pre-treatment occurs without the second surface over the furnace. The translational rod is then retracted and the furnace is raised to meet the trapping surface made from 99.95% pure tantalum foil (0.127 mm thick Johnson Matthey Seabrook NH USA). The cooling probe delivers N gas at ambient temperature to the inside of the trap at 17 1 min-l to ensure efficient conden- sation of the analyte vapour onto the tantalum trapping surface while the furnace is at vaporization temperature. A Varian (Springvale Australia) CRA-90 is used to supply power to the graphite cup (2.9 mm id.) in a modified CRA-90 workhead which is in a 1 atm Ar environment.After the sample is trapped the N coolant gas is turned off the furnace is lowered and the trapped sample is then introduced directly into vacuum without exposure to air. Details of the vacuum chamber are shown in Fig.2. The tantalum trap on the translational rod is inserted into the lo-* Torr (1 Torr= 133.322 Pa) main chamber via two diflerentially pumped chambers sealed by fluorocarbon O-ring mechanical seals at 10-3-10-7 Torr respectively. The analyte- containing surface is positioned directly beneath the electron- impact ionizer of a quadrupole mass filter (Model C50 Extrel Pittsburgh PA USA) with a channeltron electron multiplier (Galileo Electro-Optics Sturbridge MA USA). A 250 W tungsten filament radiatively heats the surface to desorb the analyte thermally for filtering and detection.Quadrupole mass spectrometer v > Heater '*\kw filament Fig. 2 Schematic diagram of tantalum second surface trap positioned within the ultra-high vacuum chamber for analyte detection by MS The quadrupole was operated in pulse-counting mode using an F-100T Amplifier-Discriminator (Modern Instrumentation Technology Boulder CO USA) and Keithley Series 500 data acquisition hardware (Keithley MetraByte Taunton MD USA). The quadrupole and Series 500 were controlled by a 386 microcomputer using a data acquisition and analysis program written in ASYST (Asyst Technologies Rochester NY USA). The data were finally manipulated and plotted using Quattro Pro 4.0 (Borland International Scotts Valley CA USA).Solutions All metal solutions (Ag Cd and Pb) were prepared from 1000 ppm stock solutions by dissolution of the appropriate nitrate salt in distilled de-ionized water. Multi-element solu- tions were also prepared directly from stock solutions; 2-4 pl aliquots were used for analyses. Procedure The experimental run consisted of several steps. First the graphite cup was cleaned at a high temperature (>2500 K) under 1 atm of Ar in the atomization chamber with the tantalum trap still in the vacuum chamber. The sample was then deposited into the cup followed by a gentle drying step and charring at approximately 600K. The rod was then retracted the cup raised to meet the trap and the cooling probe lowered into the rod and trap. The N2 cooling gas (17 1 min-') was turned on and the furnace was heated at 800 K s-' to 1500 K (held for 1 s).Immediately after this the furnace was lowered the coolant gas turned off the N2 probe removed and the rod translated into high vacuum. When the tantalum surface was directly beneath the quadru- pole the heating filament was raised into position underneath the tubular trap. The filament was first pulse-heated (approxi- mately 50% power for 1 s) to desorb any residual gases that might have adsorbed onto the filament and then held at full power to heat the tantalum radiatively and desorb the sample. Throughout the heating pulse-counting data were collected for each mass by the acquisition hardware and software. Analysis time is approximately 10 min per sample.Results and Discussion Multi-element Solution Several masses were monitored for separate samples containing 80 ng of each of the three metals studied in order to determine the chemical form of the analyte leaving the surface. For the elements studied the M+ dominated with no detectable signal noted for species such as MO+. This is in contrast to earlier electrothermal atomization MS mechanistic studies where MO + was commonly observed during decomposition. For example when a Cd sample solution was deposited directly on tantalum dried and charred both Cd+ and CdO+ were detected (Fig. 3). This suggests that graphite cup vaporization produces the free metal as in conventional ETAAS which then condenses onto the tantalum surface. While preliminary results are encouraging this may not always be the case depending on the complexity of the matrix.As more interferents are present a greater possibility exists that the analyte could be trapped and re-released in more than one chemical form. However with the knowledge of available chemical modifiers as well as the flexibility provided by the trapping and heating cycles it could be possible to atomize in the cup before trapping. Thus the elemental form would be trapped thereby simplifying detection and quantification. A profile of the time-dependent signals for multi-element determination of Ag Cd and Pb is shown in Fig.4. With a data collection rate of 0.6Hz for each element and three elements monitored the profiles shown consist of approxi-JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL. 9 v) w c 4.0 0 3 3.5 c.’ .- v) 5 3.0 Y .- - w 2.5 J 417 - - A g - Cd Pb - I 1 1 2 3 4 5 6 Time/s Fig.3 Mass spectrometer signals for 2 p1 of a 4.9 ppm Cd solution directly deposited on a Ta surface and desorbed in vacuo 3000 I v) 2500 1 cdh~g 2 2000 3 8 1500 \ + > .- gJ 1000 e 4- - 500 0 6 12 18 24 30 36 42 Time/s Fig. 4 Sequential multi-element ETV-MS with trapping for a 4 p1 aliquot of 20 ppm Ag Cd and Pb 5.0 I 7 4.5 1 2.0 ’ I I I 1 .o 1.5 2.0 2.5 3.0 Log (amounthg) Fig.5 Calibration curves for mixed element standards (Ag Cd and Pb) mately 20 points defining the shape. The metals range in volatility from high to moderate yet they were all successfully atomized trapped and re-desorbed in the same run. While raising the graphite cup to a temperature sufficient to vaporize the most refractory of the analytes the tantalum must remain cool enough to prevent loss of the most volatile analyte when trapping is occurring. This was successfully accomplished with these metals and N2 cooling yet more effective cooling may be required as the range of vaporization temperatures increases. Also the indirect heating in vacuum of this sample with a 250 W filament was adequate but modifications (such as electrothermal heating of the trap) are required to ensure thermal desorption of the more refractory metals.With the Ag+ signal monitored at m/z 107 Cd+ at m/z 114 and Pb’ at m/z 208 a multi-element working curve was generated (Fig. 5). of detection were extrapolated. They were 4.2 2.4 and 5.3 ng respectively assuming a signal-to-noise ratio of 3 where the noise is counting uncertainty in the background.The inherent background noise with the electron impact ionizer off is less than 1 count s-l but system background gases with the ionizer on increases this value significantly. The dynamic range is presently limited on the high end by pulse coincidence and on the low end by the background noise. For the current system the graphite cup vaporization and trapping process is efficient (approximately 90”/0) so mass transport to the ionizer region in the vacuum system more efficient ionization and reduction of the background are likely areas on which to focus future studies. Conclusions Preliminary results have shown that ETV-MS using second surface trapping is not only capable of sequential multi- elemental detection but that the trapping step permits vaporiz- ation of the analytes in one chemical form; for Ag Cd and Pb this form is the gaseous metal.Improvements on the current system include the need for higher thermal desorption tempera- tures in vacuum for analysis of the refractory metals as well as the previously mentioned possibilities for increasing the signal-to-noise ratio. This research was supported in part by The National Science Foundation grant No. CHE902059 1. We also acknowledge Grady Rollins for his help in construction of the system. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 References Gregoire D. C. J. Anal. At. Spectrom. 1988 3 309. Park C. J. and Hall G. E. M. J. Anal. At. Spectrom.1988,3,355. Newman R. A. Osborn S. and Siddik Z. H. Clin. Chim. Acta 1989 179 1991. Park C. J. and Hall G. E. M. J. Anal. At. Spectrom. 1987,2,473. Ediger R. D. and Beres S . A. Spectrochim. Acta Part B 1992 47 907. Wang P. Majidi V. and Holcombe J. A. Anal. Chem. 1989 61 2652. Bass D. A. and Holcombe J. A. Anal. Chem. 1987 59 974. Bass D. A. and Holcombe J. A. Anal. Chem. 1988 60 578. Droessler M. S. and Holcombe J. A. Spectrochim. Acta Part B 1987 42 981. Styris D. L. Prell L. J. Redfield D. A. Holcombe J. A. Bass D. A. and Majidi V. Anal. Chem. 1991 63 508. Styris D. L. and Kaye J. H. Spectrochim. Acta Part B 1981 36 41. Styris D. L. and Kaye J. A. Anal. Chem. 1982 54 864. Styris D. L. Anal. Chem. 1984 56 1070. Styris D. L. Fresenius’ 2. Anal. Chem. 1986 323 710. Sturgeon R. E. Mitchell D. F. and Berman S . S. Anal. Chem. 1983 55 1059. Ham N. S. and McAllister T. Spectrochim. Acta Part B 1988 43 789. Styris D. L. and Redfield D. A. Anal. Chem. 1987 59 2891. Styris D. L. and Redfield D. A. Anal. Chem. 1987 59 2897. Falk H. J. Anal. At. Spectrom. 1992 7 255. Holcombe J. A. and Sheehan M. T. Appl. Spectrosc. 1982 36 631. Rettberg T. M. and Holcombe J. A. Spectrochim. Acta Part B 1984 39 249. Rettberg T. M. and Holcombe J. A. Spectrochim. Acta Part B 1986 41 377. Rettberg T. M. and Holcombe J. A. Anal. Chem. 1986,58 1462. Rettberg T. M. and Holcombe J. A. Anal. Chem. 1988 60 600. Hocquellet P. Spectrochim. Acta Part B 1992 47 719. Churella D. J. and Copeland R. T. Anal. Chem. 1978 62 309. Paper 3/058401 Received September 28 1993 Accepted November 26 1993 Limits of Detection Using single-ion monitoring for the signals for Ag Cd and Pb and the background counts over the same time period limits