Elemental analysis of micro-samples of liquids or slurries by coiled- filament in-torch vaporization-inductively coupled plasma atomic emission spectrometry (ITV-ICP-AES)† Vassili Karanassios,* Victor Grishko and Gregory G. Reynolds Guelph-Waterloo Center for Graduate Work in Chemistry, University ofWaterloo, Department of Chemistry, Waterloo, Ontario, Canada N2L 3G1 Received 8th September 1998, Accepted 17th December 1998 In coiled-filament in-torch vaporization (ITV) sample introduction for inductively coupled plasma atomic emission spectrometry (ICP-AES), a sample is placed onto a coiled filament (e.g., a Re wire) and the sample carrying filament is inserted into a vaporization chamber which is attached to an ICP torch.In this study, six vaporization chambers were tested using in-situ generated smoke. Of these, a 6.5 cm3 (internal volume) chamber was chosen. An Ar–H2 (#3% v/v) carrier gas was used and the eVect of flow rate and filament insertion position was determined for several elements (e.g., Zn, Pb, Cd, Mn, Mg, V, Be and Sr) that covered a range of wavelengths and vaporization characteristics.Analytical performance characteristics were also obtained and calibration graphs were linear over several orders of magnitude. Detection limits were in the pg to sub-pg range. Coiled-filament ITV-ICP-AES was also briefly tested with powdered solids as slurries using a biological standard reference material and Mn and Zn as test elements.photodiode array spectrometer1,2,19 and was briefly evaluated Introduction with a photomultiplier tube (PMT)-based direct reading spec- In-torch vaporization (ITV)1–3 is an attractive alternative to trometer and with ICP-MS using Pb, Cd, Zn and Sr as test electrothermal vaporization (ETV)4–6 and to direct sample elements.3 In these feasibility studies, the vaporization chamber insertion (DSI)7–9 sample introduction for the analysis of was part of a de-mountable torch and, overall, ITV sample limited size samples by inductively coupled plasma atomic introduction was largely un-optimized.In this work, several emission spectrometry (ICP-AES) or ICP mass spectrometry vaporization chambers were designed and tested. In addition, (ICP-MS). In coiled-filament ITV, a metallic filament is the ITV was further characterized with liquid micro-samples and sample carrying probe. A sample is pipetted onto the probe was briefly tested with slurries of a powdered sample.which is inserted into a vaporization chamber and is positioned about 10 cm from the plasma. Electrical power is applied to the probe for drying, charring/ashing and vaporization and Instrumentation the vaporized sample is carried to the plasma with the aid of The ITV sample introduction system (Fig. 1) consists of a a carrier gas. In a way, ITV can be thought of as an in-situ, vaporization chamber, an electrically heated filament (e.g., Re mini-electrothermal vaporization device which is positioned wire) attached to a thermocouple ceramic insulator1 and a close to the plasma and is driven by a modified DSImechanism.modified drive mechanism which has been used for direct Unlike DSI-ICP, ITV uses an external power supply and sample insertion.8 Initial focus was on the design of a vaporiz- this facilitates independent optimization of vaporization and ation chamber. The objective was to maintain the relatively plasma operating conditions.In addition, rapid heating of the small volume of the previous chamber1–3 but with some sample carrying probe, the relatively small volume of consideration to gas flow dynamics. the vaporization chamber and the short distance analyte vapor Six chambers with diVerent volumes, shapes and carrier gas travels to reach the plasma result in a fast, transient atomic inlet configurations and inlet diameters were designed empiri- population with a high concentration of analyte vapor per cally.For example, the internal volume ranged from 3.4 to unit time. Also, the short distance analyte vapor travels to 18.0 cm3, the shape from conical to oval, the carrier gas inlet reach the plasma reduces or eliminates transport eVects10–18 from single- to dual-tangential and the internal diameter of and the short duration of analyte emission (e.g., tens to the carrier gas inlet from 4.0 to 1.5 mm. Volume is an important hundreds of milliseconds of full-width at half-maximum3) consideration because larger volumes tend to increase disper- results in sharp and narrow signals. In ITV-ICP-AES, such sion and dilution of vaporized samples whereas smaller vol- signals typically translate to improved detection limits.3 umes are thought to give rise to condensation and to analyte Furthermore, owing to the use of metallic rather than graphite loss on the cold walls of the transport system4,5,11–13,20–22 (in sample carrying probes, carbide formation (a key chemical this case, the vaporization chamber and the central tube of limitation of graphite tubes or cups used with the typical the torch).In addition, tangential introduction of the carrier ETV-ICP4–6 or DSI-ICP7,8 systems) is no longer an issue. gas with a high linear velocity was expected to help form a ITV-ICP was developed and characterized with a centralized vortex on top of the filament. The vortex was expected to provide thorough mixing of the hot sample vapors with the carrier gas, to keep the gas containing the vaporized †Presented at the 8th Solid Sampling Spectrometry Colloquium, Budapest, Hungary, September 1–4, 1998.sample away from the walls of the vaporization chamber and J. Anal. At. Spectrom., 1999, 14, 565–570 565Fig. 1 Cross-section of ITV sample introduction (only the vaporization chamber is shown to scale). to rapidly cool them by the time they reached the central tube of the torch. Rapid mixing and cooling have been reported to Fig. 2 Photographs of ITV sample introduction. (a) Vaporization help vaporized samples condense into aggregates or micro- chamber as attached to an ICP torch, (b) ceramic and seal mechanism and (c) ITV and seal as mounted on the DSI-drive mechanism. particles that can be transported to the ICP with high eYciency.4,11,13,22 To obtain an indication of gas flow dynamics inside the vaporization chambers and to help visualize gas flow patterns, Light emission from the ICP was monitored using a direct reading spectrometer (JY-48, Instruments SA, Edison, NJ, the chambers were tested using in-situ generated smoke by placing mineral oil mixed with an inert binder onto the filament USA) with 32 PMT channels.Current from the photomultiplier tubes was amplified using a low noise current amplifier/ and by applying power to the filament. Each chamber was evaluated visually for smoke-generated black deposits on the low-pass filter (SR570, Stanford Research Systems, Sunnyvale, CA, USA) and the resultant voltage was digitized using a chamber, on the ceramic or both and by examining videotaped recordings of smoke distribution under diVerent gas 12-bit analog-to-digital converter, a data acquisition rate of 500 Hz, LabView software and an Apple Macintosh microcom- flow rates and filament insertion positions (Fig. 1). When the top of the filament was positioned at about 15 mm into the puter.3 This data acquisition sub-system allowed the fast, transient emission signals of ITV sample introduction to be chamber shown in Fig. 1 and the carrier gas flow rate was set at 0.6 l min-1, a swirling stream of smoke which was confined measured.within the center of the chamber was observed. Also, black deposits on the ceramic were not noticed. As a consequence, Experimental this chamber was used throughout and the 15 mm insertion position and 0.6 l min-1 flow rate provided an initial starting For system characterization with liquids, Zn (I, 213.856 nm), Pb (II, 220.353 nm), Cd (I, 228.802 nm), Mn (II, 257.610 nm), point for optimization.An additional advantage of this chamber is that it clips onto any ICP torch with a ball joint, Mg (II, 279.079 nm), V (II, 292.402 nm), Be (II, 313.042 nm) and Sr (II, 407.771 nm) were used. These elements were chosen thus facilitating rapid sample introduction system changeover. Photographs of the vaporization chamber, the seal and part because they cover a wide range of wavelengths and have diVerent vaporization characteristics. Standards were pur- of the drive mechanism are shown in Fig. 2. The 0.25 mm diameter Re filament was 6 cm long and it chased from Leco (St. Joseph, MI, USA) and single element solutions were prepared by serial dilution with distilled, formed a three-coil loop with a diameter of 3 mm. Microliter volumes of a liquid sample or slurry were deposited onto the de-ionized water (18MV) of the respective 1000 mg ml-1 stock standard solution.The system was also briefly tested with filament, the coiled filament attached to the ceramic was manually driven into the vaporization chamber, a seal was slurries. These were prepared by accurately weighing up to 1 g of a powder, adding 10–100 ml of distilled, de-ionized water formed, the sample was dried, ashed/charred (if necessary) and vaporized by applying progressively higher power levels. (18MV) to the powder, adding a thixotropic agent (TritonA X-100, 0.05%) and magnetically stirring the resultant suspen- Power applied to the filament was supplied by a laboratory variac.sion. Unless otherwise stated, 10 ml volumes were placed onto 566 J. Anal. At. Spectrom., 1999, 14, 565–570the Re filament (Iso-Mass Scientific, Calgary, Alberta, Canada) using an Eppendorf micropipette (Cole-Parmer, Vernon Hills, IL, USA). To reduce the potential for contamination from particles landing on the filament,1 solutions were dried and slurries were dried and ashed/charred inside the vaporization chamber.The dried/ashed residues on the filament were vaporized by applying about 35 W. At this power level, the filament was glowing white hot. The vaporized sample was carried to the ICP using an Ar or an Ar–H2 carrier gas. Argon and hydrogen were purchased from Linde (Toronto, Ontario, Canada). Argon was mixed with H2 using a rotameter (Linde) and the carrier gas flow rate was controlled using a mass flow controller (Brooks Instruments, Hatfield, PA, USA).Plasma observation height was set at 14.5 mm above the load coil and, with the exception of Sr (900 W) and Mn (1200 W), a 1.5 kW forward power was used throughout. Results and discussion Similar to ETV-ICP-AES23 and DSI-ICP-AES7,8 sharper signals in ITV-ICP-AES generally translate to improved detection limits. Sharper signals were expected by increasing the heating rate and, depending on elemental volatility, the final temperature of the filament.To use the fastest possible heating rate once the desired power level was set, power was applied to the filament using the on/oV switch of the variac. However, as power increased over that used previously,3 a pressure pulse became more evident, in particular, at longer wavelengths. In addition, a spectral interference from Re (II, 213.904 nm line) on Zn (I, 213.856 nm line) was observed [Fig. 3(a) and (b)] Fig. 4 Rhenium emission observed by monitoring the Re 213.904 nm line.(a) Regular power (#35 W, insert: scale expansion 10 times) and and a black deposit was noted on the top part of the (b) high power (#70 W, insert: scale expansion 20 times). vaporization chamber. A possible explanation is that the hot Re filament is attacked by O2 and H2O which are present in the Ar carrier gas, to form black hydrated oxide(s).24 either at the onset of the rise in background [Fig. 3(b)] due to Re emission from vaporizing Re oxide(s) in the plasma Depending on power levels, analyte emission was observed (thus complicating background correction) or on top of it (thus also causing a carrier eVect).14,23 The problem of spectral interference was solved by mixing #3% v/v of H2 with the carrier gas and examples are shown in Fig. 3(c) and (d). Most likely, H2 acted as an oxygen scavenger. The small rise in background and the diVerence in the amplitude of the signals shown in Fig. 3(b) and (d) is attributed to a beneficial eVect of mixing hydrogen through the central channel.25 The amount of Re vaporizing from the filament at diVerent power levels was tested by monitoring the Re 213.904 nm spectral line and examples are shown in Fig. 4. Even at high power levels, for example 70 W, only a low intensity signal was observed by monitoring the Re line, thus indicating that Re vaporizing from the filament is not a problem. With power fixed at about 35 W, optimization of insertion position and carrier gas flow rate was attempted.The eVect of carrier gas flow rate was examined with the insertion position initially set by smoke experiments (i.e., using in-situ generated smoke as described under Instrumentation) and the results are shown in Fig. 5. The signal-to-background ratio (SBR) was used to establish optimum flow rates because plasma background was also reduced as the carrier gas flow rate increased. These results indicate that compromise conditions would be required for simultaneous, multi-element determinations.With the carrier gas flow rate set to the optimum for each element (Fig. 5), the eVect of insertion position was studied next. Significant loss of signal intensity was observed when the top of the filament was positioned below the zero insertion position (Fig. 1); signals became broad and, in some instances, Fig. 3 Interference of the Re 213.904 nm spectral line on the Zn nearly disappeared. In addition, loss of signal intensity was 213.856 nm line.(a) Water blank obtained using Ar carrier gas, observed when the filament was inserted above the top inser- (b) dried solution residue of a stock standard solution of Zn, (c) water tion position (Fig. 1). As a consequence, the insertion position blank obtained using Ar–H2 carrier gas and (d) dried solution residue of a stock standard solution of Zn. was varied between 0 and 22 mm or zero and top insertion J. Anal. At. Spectrom., 1999, 14, 565–570 567Fig. 7 Reproducibility for (a) Cd (300 pg) and (b) Be (3 pg). and 20 mm (Fig. 6), compromise conditions would be required for simultaneous, multi-element determinations. With the insertion position set to the optimum for each element, the eVect of carrier gas was tested again using the optimum insertion position for each element (Fig. 6) and SBRs did not vary by more than about 15% of those shown Fig. 5 EVect of carrier gas flow rate. in Fig. 5. As a consequence, the ‘optimum’ flow rates (Fig. 5) and insertion position values (Fig. 6) were maintained throughout detection limit and precision determinations. position in Fig. 1. Analyte signal intensities increased with an The relative standard deviation (%RSD) obtained for peak increase in insertion position, passed through a maximum and height and peak area measurements (ten replicates) was 4.2% then decreased (Fig. 6). The reduced peak heights at low and (2.1% peak area) for Cd and 6.1% (2.5% peak area) for Be.high insertion positions are attributed to a destruction of the Examples are shown in Fig. 7. All other elements had %RSDs vortex that caused analyte loss either on the ceramic and/or between these two values. These %RSDs reflect not only the the cold walls of the vaporization chamber. Although there is precision of the technique but also that of manually pipetting relatively little loss of signal intensity in the region between 15 a sample onto the filament. Automation should further improve the overall precision of ITV-ICP.Calibration graphs (not shown for brevity) were linear over several orders of magnitude. To further increase their dynamic range and to improve concentration detection limits of environmentally important elements, such as Pb, Cd and Zn, preconcentration using multiple drops was tested. For example, 10 ml of a sub-ppb stock standard solution of Zn were deposited onto the filament, the solution was dried and the experiment was repeated ten times.Although the approach allowed subppb (or low pg level ) determinations of Zn, it also proved to be time consuming. Is there a way of increasing the capacity of the filament and thus reducing drying time? This question was addressed using a 16 cm long filament that was coiled into five loops with a diameter of 4.1 mm and a maximum capacity of 50 ml. Solution residues were vaporized by applying 50W to the filament and examples are shown in Fig. 8. The approach may find applicability, in particular, in relatively clean water samples.Detection limits (3s) were obtained using 10 ml of single element standards and were estimated by setting one-fifth of the peak-to-peak value for the noise between 1 and 3 s equal to 1s. The values were: Zn (10 pg), Pb (15 pg), Cd (10 pg), Mn (0.3 pg), Mg (3 pg), V (10 pg), Be (0.1 pg) and Sr (0.08 pg). These values compare favorably with those obtained by ETV sample introduction,4,26 with ITV oVering significant improvements, especially for carbide-forming elements.In addition, detection limits improved over those reported previously,1,3 for example, for Be by one order of magnitude. Can ITV be used for the analysis of solid micro-samples as well? This question was briefly addressed using slurries27 of a Fig. 6 EVect of insertion position (as per Fig. 1) on signal intensity. powdered biological standard reference material (NIST SRM 568 J. Anal. At. Spectrom., 1999, 14, 565–570Fig. 8 Water blanks and emission signals for Zn for coiled filaments with 10 and 50 ml capacities. obtained with this chamber and liquid micro-samples is superior to that reported previously.1,3 Also, Ar–H2 carrier gas mixtures reduced or eliminated spectral interference and carrier eVects from Re vaporizing from the filament, and allowed the use of higher ITV power levels, thus helping to improve detection limits. In addition, the feasibility of direct elemental analysis of powders as slurries has been demonstrated.A new holder that replaces the ceramic and allows either coiled filaments or metal cups to be used with ITV has been designed and tested,27,28 thus further expanding the scope and application of ITV-ICP. It is clear from the results reported here and elsewhere1–3,19,28,29 that ITV has the potential to become a useful sample introduction system for the ICP, in particular, when only micro-samples are available for analysis. Examples include samples of clinical, biological or forensic origin.Also, the short duration of ITV signals may prove to be beneficial to ICP-AES with area sensor detectors and to ICP-TOF-MS (ICP-time-of-flight-mass spectrometry)30 for simultaneous elemental determinations from micro-samples. Acknowledgements Financial assistance from the National Science and Engineering Council of Canada (NSERC) is gratefully acknowledged. References 1 V. Karanassios, K. P. Bateman and G. A.Spiers, Spectrochim. Fig. 9 Slurry of NIST SRM 1577a Bovine Liver. (a) Calibration graph Acta, Part B, 1994, 49, 847. for Mn (slope: 0.98) and Zn (slope: 0.93) and (b) reproducibility for 2 V. Karanassios, K. P. Bateman and G. A. Spiers, Spectrochim. Mn (100 pg in 10 ml of slurry). Acta, Part B, 1994, 49, 867. 3 V. Karanassios, P. Drouin and G. G. Reynolds, Spectrochim. Acta, Part B, 1995, 50, 415. 1577a, Bovine Liver). Examples of calibration graphs are 4 J. M. Carey and J.A. Caruso, Crit. Rev. Anal. Chem., 1992, shown in Fig. 9(a). Reproducibility (peak height) was 6.2% 23, 397. for Mn and 7.3% for Zn and examples are shown in Fig. 9(b). 5 J. M. Carey, F. A. Byrdy and J. A. Caruso, J. Chromatogr. Sci., These results indicate that ITV has the potential to be used 1993, 31, 330. 6 D.C.Gre� goire, Can. J. Anal. Sci. Spectrosc., 1997, 42, 1. for the analysis of powders (as slurries). To accommodate 7 V. Karanassios and G. Horlick, Spectrochim. Acta Rev., 1990, solid micro-samples that cannot be converted easily to a 13, 89.powder or to permit analytical determinations when only solid 8 V. Karanassios and T. J.Wood, Appl. Spectrosc., 1999, 53, 197. micro-samples are available for analysis, the coiled filament 9 C. D. Skinner and E. D. Salin, J. Anal. At. Spectrom., 1997, was replaced by a Re cup; an application of Re-cup ITV-ICP- 12, 1131. AES for direct elemental analysis of solid micro-samples is 10 D. L. Millard, H.C. Shan and G. F. Kirkbright, Analyst, 1980, 105, 502. described elsewhere.28 11 S. E. Long, R. D. Snook and R. F. Browner, Spectrochim. Acta, Part B, 1985, 40, 553. Conclusions 12 S. M. Schmertmann, S. E. Long and R. F. Browner, J. Anal. At. Spectrom., 1987, 2, 687. A new vaporization chamber that clips onto an ICP torch has 13 T. Kantor, Spectrochim. Acta, Part B, 1988, 43, 1299. been designed and tested. The chamber facilitates rapid sample 14 R. D. Ediger and S. A. Beres, Spectrochim.Acta, Part B, 1992, 47, 907. introduction system changeover and, overall, the sensitivity J. Anal. At. Spectrom., 1999, 14, 565–570 56915 D. C. Gregoire, S. Al-Maawali and C. L. Chakrabarti, 25 A. Montaser, K. D. Ohls and D. W. Golightly, in Inductively Spectrochim. Acta, Part B, 1992, 47, 1123. Coupled Plasmas in Analytical Atomic Spectrometry, ed. 16 C. M. Sparks, J. Holcombe and T. L. Pinkston, Spectrochim. Acta, A. Montaser and D. W. Golightly, VCH, New York, 1992, ch. 19. Part B, 1993, 48, 1607. 26 H. Matusiewicz, J. Anal. At. Spectrom., 1986, 1, 171. 17 G. Zaray and T. Kantor, Spectrochim. Acta, Part B, 1995, 50, 489. 27 L. Ebdon, M. Foulkes and K. Sutton, J. Anal. At. Spectrom., 18 R. W. Fonseca and N. J. Miller-Ihli, Appl. Spectrosc., 1995, 49, 1997, 12, 213. 1403. 28 H. R. Badiei and V. Karanassios, J. Anal. At. Spectrom., 1999, 19 V. Karanassios, K. P. Bateman and G. A. Spiers, Spectrochim. 14, 603. Acta, Part B, 1994, 49, 989. 29 V. Grishko and V. Karanassios, in Proceedings, Second Biennial 20 H. Matusiewicz and R. M. Barnes, Appl. Spectrosc., 1984, 38, 745. International Conference on Chemical Measurement and 21 H. Matusiewicz and R. M. Barnes, Spectrochim. Acta, Part B, Monitoring of the Environment, ed. R. Clement and B. Burke, 1985, 40, 29. Ottawa, ON, Canada, 1998, vol. 2, p. 507. 22 C. J. Park, J. C. Van Loon, P. Arrowsmith and J. B. French, Can. 30 P. P. Mahoney, S. J. Ray and G. M. Hieftje, Appl. Spectrosc., J. Spectrosc., 1987, 32, 29. 1997, 51, 16A. 23 D. C. Gregoire, M. Lamoureux, C. L. Chakrabarti, S. Al-Maawali and J. P. Byrne, J. Anal. At. Spectrom., 1992, 7, 579. 24 F. A. Cotton and G. Wilkinson, Advanced Inorganic Chemistry, Interscience, New York, 3rd edn., 1972. Paper 8/07032F 570 J. Anal. At. Spectrom., 1999, 14, 565–5