Images of the Direct Sample Insertion Process in an Inductively Coupled Plasma CAMERON D. SKINNER AND ERIC D. SALIN* Department of Chemistry, McGill University,Montreal, Quebec, Canada, H3A 2K6 An inductively coupled plasma was imaged with a charge deposition11 and the use of vaporization enhancing gases in coupled device camera through interference filters during direct the DSI probe.15 These methods of sample preparation and sample insertion. Images were acquired throughout the treatment do not necessarily require a probe in the form of a insertion and revealed how the analyte behaves in the plasma.cup and may benefit from an examination of probe design. Graphite cups, graphite tubes and wire loops were used as The key to a better understanding of plasma behavior during sample carrying probes. The diameter of the graphite cup has the DSI process is to image the entire plasma simultaneously. a dramatic eVect on the plasma and calcium emission. We define the behavior of the analyte in the plasma, in the Narrower cups disturb the plasma the least and keep the context of this study, as the apparent position of the analyte calcium within the center of the plasma.The use of a carrier in the plasma as measured by the imaging system. The vaporizgas through a graphite tube and hollow stem cup shows that a ation of the sample provides a transient signal so that it is darker central channel is established. The use of a central gas diYcult to observe the changes that take place in the plasma in combination with Freon-12 yields a performance without a high speed imaging system.A spectrometer may be comparable to that of wire loops. In the case of a deep and used to image a small portion of the plasma, and if one is narrow graphite cup the analyte appears to emerge into the persistent enough the entire plasma may be mapped by this plasma along the walls of the cup. method. A more eYcient prospect is to image the plasma with a two-dimensional detector in combination with an imaging Keywords: Inductively coupled plasma; direct sample insertion; spectrometer.The advantages of this method are the wide imaging; Abel inversion choice of wavelengths and high spectral resolution.16 The disadvantages are low light throughput and the potential for Our laboratory has been interested for many years in method- image overlap from adjacent images. A relatively inexpensive ologies that increase the sensitivity of atomic analysis, with alternative to these two methods involves using interference much of the research focusing on direct sample insertion (DSI) filters in conjunction with a high speed imaging detector.This as a method of introducing samples into the inductively method provides high light throughput and allows short coupled plasma (ICP).1–4 The DSI technique was extensively integration times while providing moderate spectral isolation. reviewed in 1990 by Karanassios and Horlick.5 This technique The detector may be based on one of several technologies increases the sensitivity by vaporizing the sample in a short but charge coupled devices (CCDs) are nearly ideal because of time period, producing high analyte concentrations in the their high quantum eYciency and low noise.A more complete plasma. Some types of samples, e.g., geological materials, or description of how CCDs (and charge injection devices) operate large samples can produce visible disturbances in the plasma and their applications to spectroscopy was given by Bilhorn as the sample vaporizes from the probe.6,7 These include and co-workers.17,18 changes in the shape of the plasma, emissions from the entire CCDs may be operated in several modes.One of the modes plasma, bright and variable emissions above the cup and that is of particular interest is the frame transfer mode, in unstable plasma operation. Large disturbances of the plasma which an image is acquired on half of the CCD array.The are undesirable and the eVects of probe type and geometry on charge that has accumulated is then transferred to a masked the plasma are poorly understood. portion of the array and the exposed area begins to collect a DSI probe designs are varied but probes are generally new image. During the relatively long exposure time, the fabricated from graphite or refractory metals.8 The design of charge in the masked region can be digitized with high the probe has a significant eVect on the intensity, duration and precision.This process repeats for each image. noise of the signal that is observed and also memory eVects.9 Once the image has been acquired, it may be examined Wire loops have been used to hold small volumes of liquid directly to indicate the spatial behavior of the analyte and that may be dried and introduced into the plasma. These plasma or it may be processed to extract information about probes heat very quickly and produce signals that may last the analyte by performing background subtraction.Additional for only a few tenths of a second.1 Thin walled and thin information can be obtained by deconvolving the image to stemmed graphite cups produce signals that rival wire loops obtain radial intensity profiles. The deconvolution of lateral for speed of vaporization while being able to hold larger intensity measurements to yield radial intensity profiles can be amounts of sample.4,10 These graphite cups can also be used done by the Abel inversion.19 Several researchers have utilized as a target for spray depositing a volume of sample much this method to determine the radial intensities of the ICP.20–31 larger than the capacity of the cup.11 Others have also found The Abel inversion is a special case of computed tomography that the thickness of the cup walls has an important impact where the symmetry of the plasma can be used to advantage.32 on the signal that is observed.12 The graphite probe can also The Abel inversion makes several assumptions about the be treated as a sacrificial sample holder and the sample and source, namely circular symmetry, an optically thin source and probe can be consumed with oxygen.13 In these cases the a well known center.The first two assumptions have been design of the probe has been based on common sense and partially addressed by modifications to the algorithm.28,33 The experience.Recently, we have begun to investigate other possthird assumption is usually fulfilled by examination of the ible probe designs because of our interest in preconcentration of analyte species directly on to activated charcoal,14 spray data. The Abel inversion has an additional shortcoming of Journal of Analytical Atomic Spectrometry, October 1997, Vol. 12 (1131–1138) 1131being highly sensitive to noise in the input data. The eVect The controller was equipped with a 16 bit A/D converter with a maximum readout rate of 100 kHz.The controller and of the noise can be moderated by smoothing of the data23,28 or fitting the data to one or more polynomials.19,22,29,31 software package provide full control of all camera functions such as thermoelectric cooling of the detector head, exposure The method that was used in this series of experiments is based on the method of Scheeline and Walters.33 Each row of time, shutter control, readout rate and on chip binning.Esco Products (Oak Ridge, NJ, USA) interference filters data in the image is considered to be an image of a small cross-section of the plasma. The plasma is divided into a series (390 mm Part No. S903900; 420 mm Part No. 5904200) were used for spectral isolation. The filters have a 25.4 mm (1 in) of rings where each ring is the width of a pixel from the image. The intensity that is measured for any given pixel is the sum od and 10 nm bandpass (FWHM) at the 390 and 420 nm center wavelengths.With these filters it was possible to observe of the contributions from each ring. The center and the edges of the image are determined and the areas of the segments of the 393.3 nm Ca II and the 422.6 nm Ca I emission lines. All images were acquired using a Nikon AF Micro 60 mm lens. the circle are calculated. The radial intensity is calculated from the outside of the image towards the center. The radial This lens has 50% of maximum transmission at 390 nm and 80% of maximum at 420 nm.The interference filters were intensity of the outermost ring is the measured intensity divided by the area of the segment of the circle from which the attached to the lens with a custom built adapter ring that could also accommodate optical density filters (neutral density radiation originates. The next radial intensity is calculated by subtracting the contribution from the outer radius and again filter set, Part No. 03FSQ 011; Melles Griot, Irving, CA, USA) as needed.For all of the experiments the F/n of the lens was 32. dividing by the area of the segment. This process continues until the center radial intensity is calculated. Each time the The camera was operated with a 5 ms exposure with the shutter locked open during the experiment. The images were measured intensity is corrected for contributions from the outer radii. Fig. 1 illustrates this method for determining the acquired using an active area of 150×256 pixels on the CCD chip at 16 bits resolution. This enabled an image to be acquired intensity at any given radius.The generalized equation is approximately every 0.38 s. Each experiment comprised a sequence of 45 images (a movie) that recorded the entire insertion process. The beginning of the experiment was trig- Ir=ir/2Ar, r-2 . j=R j=r+1 IjAr,j gered manually. Background movies were acquired with empty probes. where Ir is the radial intensity, ir is the measured lateral intensity at radius r, R is the outermost pixel from the centre of the image and Ar, r is half of the area of a particular segment of the circle that represents the plasma cross-section. DSI Probe Designs To gain a greater understanding of how DSI probes behave The probes used for this series of experiments were constructed in the plasma, we imaged the plasma with a CCD camera.from graphite rod using techniques previously described.15 Some of the images were subjected to Abel inversion to reveal Several types were used (Bay Carbon, Bay City, MI, USA, greater information about the characteristics of emission pat- Part No.S-8 HD for 3 mm probes; SGL Carbon Group, Speer terns of calcium as it leaves the DSI probe. Canada, St. Laurent, Canada, Part No. 580-.375 for 5 mm probes, Part No. 580-500 for 8 mm probes) and are illustrated in Fig. 2. The leftmost cup is commonly used in our laboratory EXPERIMENTAL for introducing liquid samples.Three diVerent cup diameters Camera (dimension a, Fig. 2) were investigated; the large diameter was 8.3 mm, the standard diameter was 5.2 mm and the narrow A Princeton Instruments Canada (Stittsville, Canada) frame transfer camera with a 512×1024 element CCD array and a diameter was 3.1 mm. These three cups were all 6.3 mm deep (dimension b). Two cups of longer length were also used (b= ST-138 controller was used for this series of experiments. The camera system was interfaced to a generic EISA-based 486 PC 12 mm) with the standard and narrow diameters.The second type of cup illustrated is a hollow stem cup which has a using the cabling and interface card supplied by Princeton Instruments. Images were acquired using the Windows-based standard diameter but a stem that has been enlarged to allow for the 1.6 mm diameter hollow stem. The third type is a WinView 1.3A software supplied by Princeton Instruments. A6,6 Pixel no. Radial intensity I6 = i6/2 A6,6 6 5 4 3 2 1 A5,5 A4,4 A3,3 A3,4 A3,5 A3,6 A2,2 A1,1 A1,2 A1,3 A1,4 A1,5 A1,6 I3 = i3/2 A3,3–2(I4 A3,4 + I5 A3,5 + I6 A3,6) I r = ir /2 Ar,r –2 S I j Ar,j j = R j = r + 1 Fig. 1 Radial intensities (Ir) for any given radius (r) are calculated from the observed intensity (ir) and the area (Ar,r) of the plasma that generates that measurement. For the inner radii the measured intensity must be corrected for the contribution from the outer radii of the plasma. 1132 Journal of Analytical Atomic Spectrometry, October 1997, Vol. 12Fig. 2 The four types of probes used in the experiments. The three on the left are made from graphite. The fourth is a tungsten wire loop. hollow tube with an internal diameter of 2.4 mm. A small piece Fig. 3 The area of the plasma imaged by the camera and the torch of reticulated vitreous carbon foam (3 mm long, porosity 100; and DSI cup. Energy Research and Generation, Oakland, CA, USA) was placed within the tube to hold the sample.The fourth type of probe used was a 0.5 mm diameter tungsten wire loop of 3 mm the intensity (I) that was obtained from the image at the diameter.1 normal viewing height for a spectrometer [16 mm above top In all cases 5 ml of 100 ppm calcium was introduced on to of the load coil (ATOLC)]. The beginning of each experiment the probes with an Eppendorf pipette. The liquid was dried was triggered manually and consequently synchronization of by heating the probe inductively in the plasma coil. When the the images for background subtraction had to be done manuhollow stem cup was used the carrier gas was introduced ally.In some experiments there is a slight synchronization through the stem at a flow rate of 290 ml min-1. For the error that results in imperfect background correction. The experiments with the graphite tube a reduced flow rate of eVect of this error is most pronounced for the images of the 180 ml min-1 was used, otherwise the foam would be launched actual insertion where the plasma rapidly changes brightness.through the plasma. Freon-12 (1000 ppm in argon) (Matheson For the images prior to and after the insertion the error is Gas Products, Montreal, Canada) was the carrier gas. negligible because the diVerence from frame to frame of the background is small. Our experience with DSI has shown that the signals are Overview of the Method of Calculating Radial Intensities reproducible (3–5% RSD) when simple aqueous standards are The file generated by the camera software was read into used.This was also found to be true with the images collected Matlab 4.0 for Windows 3.1 (The MathWorks, Natick, MA, with the camera. For a given set of experimental conditions USA). An image was chosen for the center finding routine the variation in intensity, from movie to movie, at any particubased on high calcium signals with no or minimal saturation lar time during the insertion is about 10%.The larger variaof the detector. Some images were saturated for a small area bility is due to the problem of manual synchronization of the directly above the cup. The center of mass determined for this beginning of each experiment. small area may not be correct but will not aVect the accuracy Calcium was chosen as an analyte for two reasons: it has of the radial calculations in the unsaturated regions. Each of an atomic and ionic emission line in the spectral range of the the 256 cross-sectional images of the plasma was smoothed camera and it is suYciently refractory that the signal can be using a ten point moving average.The center of the image was observed over several seconds. Initially we believed that the found by finding the center of mass for each of the smoothed two lines would provide complementary information about cross-sections. The center determined from this image was the plasma, but the images are strikingly similar and provide used for all of the images in the movie. The images that the the same information.radial intensities were to be calculated for were then also Fig. 5 shows some of the images that were taken with the smoothed and the radial intensities were calculated using the three diVerent diameter graphite DSI cups. Only a few of algorithm described in the Introduction. Each half of the image the many images in each movie can be presented here owing was calculated separately and the two halves were combined to space constraints.The images in Fig. 5 were taken with the to yield a radial intensity map of the plasma. 390 nm filter at 1.25 kW. All of the images have been corrected for background so the images show net calcium emission. The eVect of the synchronization error for background correction RESULTS AND DISCUSSION can be seen in the first image from the medium diameter cup Fig. 3 illustrates the area of the plasma that was imaged by in Fig. 5. The first image for each size of cup is immediately the camera.Included is the region of the plasma directly above prior to, or during, the insertion of the cup and is assigned the DSI probe to well above the normal analytical viewing time 0. There is some inaccuracy in the timing because the zone. Many of the images that are shown have distortions due beginning of image acquisition was triggered manually. We to the bonnet as well as the top of the torch. All of the images estimate that the maximum error is equal to the image presented in this paper utilize false coloring that scales the acquisition time of 0.4 s.The second image in each sequence intensity to a color and should not be confused with the actual colors emitted by the plasma. The color scale that is used for the images is illustrated in Fig. 4. The numerical values of the minimum and maximum will be presented with the images since the scaling may change from image to image. In the instances where an optical density filter was used the scales Fig. 4 Color scale used for images. Black is the minimum value and white is the maximum. were corrected for the filters’ attenuation. Some figures include Journal of Analytical Atomic Spectrometry, October 1997, Vol. 12 1133Fig. 5 Images from diVerent sizes of DSI cups. Note that the width of the plume is proportional to the size of the cup. 1134 Journal of Analytical Atomic Spectrometry, October 1997, Vol. 12Fig. 6 Images taken from the other types of DSI probes studied. Journal of Analytical Atomic Spectrometry, October 1997, Vol. 12 1135is immediately after insertion (t=0.4 s) and shows the beginning (note that the times for the frames are not the same). It seems that the combination of low probe mass and the use of gas of the calcium emission. The third image is taken at the time of greatest emission intensity during the experiment. The fourth through the tube produces rapid and eYcient vaporization.In the case of the tube the cooling eVect of the gas retards the image is taken prior to the removal of the probe from the plasma. vaporization maximum until 1.4 s. With the hollow stem cup the larger diameter of the graphite cup allows the sample to The narrow cup (3.1 mm) yields the greatest intensity (I) when compared with the medium and large diameter cups (5.2 heat more rapidly and the maximum emission is observed earlier (0.76 s) than with the hollow tube.and 8.3 mm) for the region in the plasma where analytical measurements are routinely made. The dramatic increase in As was the case with the three diVerent diameters of cups, the analyte is largely dispersed by the time it reaches the area signal from the large to the narrow diameter cup is probably due to three sources: the narrower cups heat more quickly, in the plasma where analytical measurements would normally be made; however, the graphite tube is best able to keep the reach a higher temperature and release more calcium into the plasma per unit time;10,34 the smaller cup also places a smaller analyte within a central channel, probably because the absence of a cup decreases turbulences.One interesting point that load on the plasma, allowing greater eYciency of excitation; and the narrower cup keeps the calcium confined to a narrower should be made is that the intensity of emission just above the tip of the tube is much higher than that of any other probe.plume in the center of the plasma. The first two observations are consistent with work reported by Barnett et al.34 regarding This is shown in Fig. 7, where the intensity is 37 000 about 2 mm above the top of the tube. The other types of probes do the rate of heating and heat loss as a function of cup geometry. Many other workers have determined that the thickness of the not display this intense emission directly above the probe. Unfortunately, the intensity drops oV very quickly above the supporting stem is important to the emission versus time profile.9 In general, the thicker the stem or the cup walls the top of the tube.To see if inserting the tip of the tube close to the observation point would yield a significantly larger signal, lower is the rate of vaporization. Umemoto and Kubota10 found that when the internal volume of the cup was held a longer (33 mm) hollow tube probe was constructed. Fig. 8 shows the results that were obtained when the viewing height constant and the wall thickness was increased that the time of maximum emission was retarded and that the maximum cup of the spectrometer was set to 16.5 mm ATOLC and the insertion depth was increased.The results are disappointing temperature was reduced. A narrow plume is advantageous because it increases the since there is only a gradual loss of signal as the probe is inserted deeper into the plasma. There may be two explanations analyte signal intensity in the viewing zone and minimizes the impact of the sample on the plasma.If large amounts of sample for the apparently conflicting results in Figs. 7 and 8: the large emission above the top of the tube observed with the camera are allowed to mingle with the plasma, interferences become possible. With liquid nebulization the majority of the sample may be due to broadband emission that the filter allows through or the vaporization and excitation conditions are is constrained to a narrow central channel, thus limiting the eVect of the sample on plasma behavior.35 Within the torch favorable.The background calcium 393 nm data from the the narrow cup is the most eVective cup design at keeping the calcium in a narrow plume. Additionally, we found that the narrow cups disturb the plasma the least. The medium diameter cup causes the plasma to fluctuate on insertion but the plasma stabilizes quickly. The large cup was diYcult to insert into the plasma without the plasma being extinguished.The images reveal that the plasma is most adversely aVected by the large diameter cup, the emission intensity being very low. The integrated intensity in the region above the torch is 4.3×108, 1.7×108 and 8.8×106 counts for the narrow, medium and large diameter cups, respectively (from the brightest image). The increase in diameter spreads the calcium over a greater volume of the plasma, Fig. 7 Emission intensity as a function of the viewing height above as can be seen from the images.Additionally, the integrated the top of the graphite tube. The emission of calcium is much higher intensity shows that the excitation conditions are poorer for just above the tube and rapidly falls oV. Note the distortions due to cups with larger diameters. the top of the torch and the bonnet. These data were taken from time Fig. 6 shows some of the results that were obtained with 1.52 s in Fig. 6 (the image from time 1.14 s was unusable because of sample probes that may be considered ‘high performance’.detector saturation). These probes heat and vaporize the sample very rapidly. All of the images were acquired through the 420 nm filter at 1.25 kW forward power. The tungsten wire loop produces the shortest transient signals and is shown here as a benchmark. Calcium readily forms refractory carbides on graphite but the use of Freon greatly improves the volatility.15 For the experiments comparing graphite probes with the wire loop, a carrier gas, introduced through the stem of the graphite probe, of argon enriched with Freon-12 (1000 ppm) was used.15 The enhanced volatility of the calcium yields signals that are comparable to those of more volatile species.From Fig. 6, it can be seen that the wire loop has the shortest transient of the three types of probes tested. The emission intensity (I) dropped to 1200 at 1.52 s and 170 at 1.9 s (not shown). The intensity from the hollow stem graphite cup is 480 by 12.2 s, the last frame with the cup in the plasma.Fig. 8 Normalized signal intensity as a function of insertion depth of The hollow tube’s intensity is 130 by 4.2 s, indicating that the hollow stem tube. The expected large emission near the top of the tube was not observed. tube has a performance comparable to that of the wire loop 1136 Journal of Analytical Atomic Spectrometry, October 1997, Vol. 12spectrometer were examined for any evidence of broadband confined to the center of the plasma.It also shows that the plasma changes within the torch. There is a slight decrease in emission, but none was observed. Hence it seems probable that a position of the tube low in the plasma leads to eYcient the emission from the right-hand side of the plasma. This is likely due to poor centering of the hollow tube in the plasma. vaporization and excitation. When the tube is inserted to greater depths, the eYciency of the excitation is reduced One of the primary reasons for embarking on this study was to gain a greater understanding of how the analyte moves because the tip of the tube is no longer in the hot base region of the plasma.Additionally, there is more of the tube in the through the plasma. In many cases direct study of the images has revealed how the probes alter the behavior of the analyte plasma, which probably reduces the amount of energy available for vaporization and excitation.in the plasma. In some cases, calculation of the radial intensities can suggest where the calcium is in the plasma. We have been The use of gas through the DSI probe creates a central channel in the plasma similar to the central channel found curious about where the analyte emerges from the graphite cup. This question has been partially answered by calculating with liquid nebulization. The gas flow needed to establish a central channel is much less in the case of the DSI probe the radial intensities.A deep and narrow diameter cup (12×3.1 mm, dimensions because the probe is inserted into the plasma and does not have to overcome the magnetohydrodynamic resistance that b and a, respectively, in Fig. 2) was inserted such that the base was at a normal position (-6 mm ATOLC) and the top of the is encountered with liquid nebulization.36 Fig. 9 shows a contour plot of the diVerence when gas is and is not used (i.e., an cup was visible to the camera.The top of the cup is most easily seen in the fourth image in Fig. 10. Some of the images image of when the gas was flowing was subtracted from when it was not) with the empty hollow tube probe. The contour (390 nm filter) are shown in Fig. 10. The scaling of the intensities has been adjusted to show the presence of two lobes of plot clearly shows that the central channel is darker and higher intensity directly above the walls of the cup. The radial intensities were calculated for these images and revealed that in the case of the long DSI cup the material that is vaporized from the cup emerges along the walls and not as a plume from the center of the cup (Fig. 11) assuming that the excitation conditions are constant across the plasma in the region above the cup. Similar images were obtained from the calcium 422.6 and aluminium 394.4 nm atom lines. The length of this particular cup may allow the formation of a stable vortex above (and perhaps within) the cup that reduces mixing of the calcium in the plasma.As can be seen from Fig. 10, the annular emission is stable for nearly 8 mm above the cup. When the radial Fig. 9 Contour plot of the diVerence between when gas and no gas is used with the hollow tube probe. The values on the axes are the Fig. 11 Radial intensities calculated 2.8 mm above top of cup from pixel number. The distortions from the torch and bonnet can be seen on the plot (lower 25 pixels); an outline of the torch has been added.Fig. 10. The cup is 3 mm wide and the two maxima are 3.3 mm apart. Fig. 10 Images of a deep and narrow DSI cup inserted to +6 mm ATOLC. The intensity scaling has been adjusted in each image for clarity. The top of the cup is visible in the first image as the blue line of calcium emission. The calcium emerges along the walls of the cup and maintains an annular plume for approximately 7 mm. These images are background corrected and the area within the torch has been enlarged.Journal of Analytical Atomic Spectrometry, October 1997, Vol. 12 113711 Rattray, R., Min� oso, J., and Salin, E. D., J. Anal. At. Spectrom., intensity maps from standard length cups (6.3 mm) were exam- 1993, 8, 1033. ined (both medium and narrow diameter), no evidence of this 12 Umemoto, M., and Kubota, M., Spectrochim. Acta, Part B, 1991, annular emission was found. It seems probable that the shorter 46, 1275. cups disturb the flow of the gases in the plasma so that there 13 Liu, X.R., and Horlick, G., J. Anal. At. Spectrom., 1994, 9, 833. is rapid mixing of the calcium as it emerges from the cup. 14 Cazagou, M., Blaise, J., Skinner, C. D., and Salin, E. D., Appl. Spectrosc., in the press. 15 Skinner, C. D., and Salin, E. D., J. Anal. At. Spectrom., 12, 725. CONCLUSIONS 16 Olesik, J. W., and Hieftje, G. M., Anal. Chem., 1985, 57, 2049. 17 Bilhorn, R. B., Sweedler, J. V., Epperson, P. M., and Denton, Imaging of the plasma during the insertion of DSI probes has M.B., Appl. Spectrosc., 1987, 41, 7, 1114. revealed how the analyte behaves in the plasma. For a given 18 Bilhorn, R. B., Epperson, P. M., Sweedler, J. V., and Denton, amount of calcium narrower graphite cups yield higher emis- M. B., Appl. Spectrosc., 1987, 41, 1125. sion intensities. The narrow cups are better able to keep the 19 Cremers, C. J., and Birkebak, R. C., Appl. Opt., 1966, 5, 1057. 20 Galley, P. J., Glick, M., and Hieftje, G.M., Spectrochim. Acta, analyte in the center of the plasma. This is especially true of Part B, 1993, 48, 769. the region within the torch where extensive mixing may 21 Monnig, C. A., Gebhart, B. D., Marshall, K. A., and Hieftje, increase the likelihood of interferences. The use of gases G. M., Spectrochim. Acta, Part B, 1990, 45, 3 261. through the center of the probe establishes a central channel 22 Babis, J., Pilon, M. J., and Denton, M. B., Appl. Spectrosc., 1990, that reduces the background of the plasma and entrains the 44, 1280.analyte in a central channel. In the case of a long and narrow 23 Olesik, J. W., Den, S., and Bradley, K. R., Appl. Spectrosc., 1989, 43, 924. DSI graphite cup, the calcium emerges along the walls of the 24 Hauser, P. C., and Blades, M. W., Appl. Spectrosc., 1988, 42, 595. cup. There is rapid mixing in the plasma that reduces the 25 Walters, P. E., Gunter, W. H., and Zeeman, P. B., Spectrochim. eVectiveness of narrow cup design and the use of carrier gases.Acta, Part B, 1986, 41, 133. 26 Choot, E. H., and Horlick, G., Spectrochim. Acta, Part B, 1986, The authors acknowledge the generosity of Princeton 41, 935. Instruments Canada and Hugh Garvey for the loan of the 27 Niebergall, K., Brauer, H., and Dittrich, K., Spectrochim. Acta, CCD camera and associated controllers and software. We also Part B, 1984, 39, 1225. 28 Blades, M. W., Appl. Spectrosc., 1983, 37, 371. thank NSERC for funding under the Strategic Grants Program. 29 Choi, B. S., and Kim, H., Appl. Spectrosc., 1982, 36, 71. 30 Blades, M. W., and Horlick, G., Appl. Spectrosc., 1980, 34, 696. 31 Kornblum, G. R., and De Galan, L., Spectrochim. Acta, Part B, REFERENCES 1977, 32, 71. 1 Sing, R. L. A., and Salin, E. D., Anal. Chem., 1989, 61, 163. 32 Monnig, C. A., Marshall, K. A., Rayson, G. D., and Hieftje, 2 Blain, L., Salin, E. D., and Boomer, D. W., J. Anal. At. Spectrom., G. M., Spectrochim. Acta, Part B, 1988, 43, 1217. 1989, 4, 721. 33 Scheeline, A., and Walters, J. P., Anal. Chem., 1976, 48, 1519. 3 Moss, P., and Salin, E. D., Appl. Spectrosc., 1991, 45, 1581. 34 Barnett, N. W., Cope, M. J., Kirkbright, G. F., and Taobi, 4 Rattray, R., and Salin, E. D., J. Anal. At. Spectrom., 1995, 10, 829. A. A. H., Spectrochim. Acta, Part B, 1984, 39, 343. 5 Karanassios, V., and Horlick, G., Spectrochim. Acta Rev., 1990, 35 Inductively Coupled Plasmas in Analytical Atomic Spectrometry, 13, 89. ed. Montaser, A., and Golightly, D. W., VCH, New York, 6 Blain, L., and Salin, E. D., Spectrochim. Acta, Part B, 1992, 47, 399. 1987, p. 141. 7 Zaray, G., Broekaert, J. A. C., and Leis, F., Spectrochim. Acta, 36 Inductively Coupled Plasmas in Analytical Atomic Spectrometry, Part B, 1988, 43, 241. ed. Montaser, A., and Golightly, D. W., VCH, New York, 8 Karanassios, V., and Horlick, G., Spectrochim. Acta, Part B, 1989, 1987, p. 142. 44, 1361. 9 Karanassios, V., Horlick, G., and Abdullah, M., Spectrhim. Paper 7/02038D Acta, Part B, 1989, 45, 105. ReceivedMarch 24, 1997 10 Umemoto, M., and Kubota, M., Spectrochim. Acta, Part B, 1989, 44, 713. Accepted July 14, 1997 1138 Journal of Analytical Atomic Spectrometry, October 1997, Vol. 12