High Speed Photographic Study of Wet Droplets and Solid Particles in the Inductively Coupled Plasma R. S. HOUK*, ROYCE K. WINGE AND XIAOSHAN CHEN† Ames L aboratory—US Department of Energy, Department of Chemistry, Iowa State University, Ames IA 50011, USA Motion pictures of the ICP were taken at 4000 frames s-1 . eVorts at analysis of solids as slurries using solutions for calibration. An intact wet droplet containing yttrium causes the formation of a pale red cloud juxtaposed on the usual ambient emission structure of the plasma.Most of these droplet clouds are EXPERIMENTAL shaped like an oval or a comet. A few droplets produce small, bright spheres followed by faint, wispy streaks that point ICP Conditions downstream. Such a spherical cloud is caused by some rapid Standard operating conditions are given in Table 1. A event such as explosion of a droplet in the final stages of Meinhard concentric pneumatic nebulizer was used with a solvent evaporation.The faint streaks are some residue, Scott-type double pass spray chamber operated at room perhaps small solid particles. In a particular frame, a number temperature. In some experiments, the wet aerosol from the of these faint streaks protrude from the tip of the initial spray chamber was injected into the plasma directly. In other radiation zone (IRZ) into the normal analytical zone (NAZ). cases, the aerosol from the double-pass spray chamber was When wet droplets are introduced, that portion of the analyte dried into particles with a conventional desolvation system, that travels through the center of the plasma passes through i.e., the aerosol stream was heated to 140 °C and then cooled three distinct regions (i.e., the IRZ, the streaks, then the to 0 °C in a water-cooled condenser.This desolvation system NAZ), rather than directly from the IRZ to the NAZ. Groups was similar to that described by Fassel and Bear.13 The plasma of two or three droplets tend to appear together in the same was operated horizontally in the orientation usually used for time interval (#0.12 ms) in the plasma.These droplet clouds ICP-MS and for ICP-AES with axial viewing. are not seen, and the particle streaks are much less evident, when the solvent is removed before the aerosol is injected into the plasma. Aqueous slurries of Y2O3 in various particle sizes Samples (0.1 or 3 mm mean diameter) produce white streaks along the Yttrium was chosen because the red emission from neutral Y center line of the plasma, which are attributed to individual atoms and YO molecules can be distinguished easily from the solid particles.These observations also support the general blue emission from Y+. The yttrium was introduced in several precept that calibration of the response for slurries using diVerent forms: (a) 5000 ppm Y dissolved in 2.5% aqueous aqueous solution standards is best accomplished by keeping the HNO3; (b) 0.1 mm diameter colloidal Y2O3 suspended in H2O particle loading such that each wet droplet contains no more at 100 000 ppm Y; (c) 3.2 mm diameter Y2O3 suspension in than one solid slurry particle.H2O at 10 000 ppm Y; and (d) 8.5 mm diameter Y2O3 suspended Keywords: Inductively coupled plasma; inductively coupled in H2O at 10 000 ppm Y. plasma mass spectrometry; nebulization; sample introduction; The slurry particle sizes cited above were the mean diameters slurry; noise; particles; droplets given by the supplier.The reader should note that the particles comprising each slurry actually consist of a distribution of sizes. These numerical values are merely used to distinguish The ICP is a very eVective source of atoms and ions for the diVerent slurries in the subsequent discussion. The 3.2 mm analytical atomic spectrometry. To the human eye, it appears diameter slurry was examined under an optical microscope. to be a stable, steady-state source. However, several important dynamic phenomena are evident when the plasma is studied with suYcient time resolution.Along these lines, there is Table 1 Experimental facilities and operating conditions substantial interest in the deleterious eVects of wet droplets Torch Fassel type,12 horizontal, outer tube 18 mm id, and solid particles in the ICP. A variety of optical and jet injector with tapered tip, 1.5 mm id MS studies by Cicerone and Farnsworth1 and Olesik and Nebulizer Meinhard concentric co-workers2–9 have described transient eVects caused by such Type TR–30–C3, droplets and particles passing through the axial channel of the Scott-type double-pass spray chamber12 ICP.Our previous studies emphasized the oscillations in the ICP: Model HFS 2500D size and shape of the plasma that arise when the hot, flowing Plasma Therm (now RF Plasma Products) Power 1.3 kW plasma gas interacts with the surrounding atmosphere.10,11 Frequency 27.12 MHz These earlier photographic studies also showed evidence of Ar flow rates: vapor clouds surrounding intact droplets or particles.The Outer gas 14 l min-1 present work examines these phenomena in more detail and Intermediate gas 0.4 l min-1 illustrates ways to distinguish droplets from particles. This Aerosol gas 1.0 l min-1 paper shows that the droplets can be removed by desolvating Camera: Fastax Model WF3 Film Eastman 7297 color negative, 16 mm the aerosol and also provides insights pertinent to ongoing Framing rate 4000 frames s-1 Lens Wollensak Fastax-Raptar, 75 mm focal length with 10 mm extension † Present address: Metropolitan Water District of Southern tube California, 700 N.Moreno Ave., La Verne, CA 91750–3399, USA. Journal of Analytical Atomic Spectrometry, October 1997, Vol. 12 (1139–1148) 1139Individual particles in a range of sizes from 1 to 4 mm were brighter ‘head.’ In the next frame [Fig. 1(d)], the comet is replaced by a faint white streak that is probably produced by apparent, as were some larger agglomerates of various sizes and shapes.Addition of Triton X-100 at 0.5% v/v did not a solid residue from the wet droplet. The streak is surrounded by a spherical, blue region, the upper boundary of which greatly aVect the size or the number of these agglomerates. The suspensions were agitated during nebulization with a appears as a bulge on the upper side of the blue NAZ. This spherical blue zone represents a high local density of excited magnetic stirrer.A special, high-solids nebulizer, as commonly used for analysis of slurries, was not necessary. The usual Y+ that emit in the blue. The length scale on the photograph is calibrated from the diYculties of plugging of the nebulizer and settling of the particles were not problems in this work, because each photo- width of the plasma at the mouth of the torch, i.e., at the far right edge of the photograph. The width of the plasma at this graphic experiment lasted only a few seconds, as discussed below.For these reasons, a dispersant was not used when the position is approximately 16 mm. Thus, the diameter of the blue cloud in Fig. 1(d) is #3 mm, and its center is about 6 mm ICP photographs were taken. In our opinion, a dispersant was not necessary in this case, in agreement with the observations downstream from the center of the pale red ‘head’ of the comet in Fig. 1(c). Our blue Y+ cloud is slightly wider than the 2 mm of Broekaert and co-workers14,15 for alumina slurries.The slurries were introduced either as wet droplets or as dry, false color image shown by Olesik for a Sr+ cloud that was also #6 mm downstream from the point of initial formation desolvated particles. of Sr+, i.e., 13 mm from the load coil (Fig. 2 of ref. 9). Photographic Conditions Velocity of Droplet Clouds The high-speed camera was operated with rolls of film that were 30 m long. Each roll consisted of 4000 frames.The time The total elapsed time between frames is 1/4000 frames s-1= required to accelerate the camera to the nominal framing rate 0.25 ms. A prominent droplet cloud that is visible during the (4000 frames s-1) consumed approximately half the roll. Each entirety of two successive frames is selected; the cloud shown roll therefore lasted about 1.5 s. The viewing field of the camera in Fig. 1(b) and (c) is suitable. From the beginning of Fig. 1( b) was approximately 45 mm along the torch axis.to the beginning of Fig. 1(c), this droplet travels approximately 7.5 mm, so the flow velocity is (7.5 mm/0.25 ms)=30 m s-1 . Previous measurements of flow velocity in the axial channel Processing of Films and Selection of Prints find a range of 20–30 m s-1, in reasonable agreement with this The prints shown below were selected from viewing of individ- estimate.1,11 Note that the exposure time of a single frame is ual frames from the processed films. The films were also approximately half of the elapsed time between frames because transferred to video cassettes and viewed as full sequences with of the masking of the rotating prism inside the camera.Thus, a video editor. The prints shown in this paper were selected the exposure time of a given frame is approximately 0.12 ms. to represent clearly the various diVerent types of clouds or streaks caused by discrete droplets or particles. Shapes of Droplet Clouds The oval [Fig. 1(b)] and comet [Fig. 1(c)] are the two most RESULTS AND DISCUSSION common shapes for droplet clouds in the ICP. The comet Ambient Emission Structure of ICP shape is caused by a medium-sized droplet that loses a substantial amount of solvent and thus shrinks noticeably The transient events from intact droplets and particles are during the exposure time of a given frame (#0.12 ms). The superimposed on the usual emission structure of the ICP. oval in Fig. 1(b) is simply the cloud caused by the early stages Nebulized solutions containing yttrium produce the familiar of vaporization of a large wet droplet. pale red initial radiation zone (IRZ), blue normal analytical The sequence shown in Fig. 2 illustrates the other general zone (NAZ),16 and deep red tail plume. Presumably, this shape seen for droplet clouds. First, a very large wet droplet emission structure results from vaporization, atomization, exci- ( labelled 1) emerges from the IRZ [Fig. 2(a)]. This droplet is tation and ionization of yttrium from relatively small droplets so large that the corresponding cloud does not shrink notice- and particles that dissociate and atomize properly in the ably in a single frame.In the next frame [Fig. 2(b)], droplet 1 plasma. travels downstream and shrinks considerably, to the point This ambient emission structure is depicted in Fig. 1(a). The where its vapor cloud begins to taper. Another large droplet plasma flows from right to left out of the torch. The torch (2) leaves the IRZ.In Fig. 2(c), droplet 1 moves still further protrudes through a circular hole cut through an aluminum downstream and decomposes into a faint white particle track. shielding box surrounding the load coil. This shielding box The faint white streaks in Figs. 1(d) and 2(c) are probably blocks the bright white emission from the induction region caused by some residue remaining after decomposition of the and greatly facilitates photography of detail downstream in bulk of the solid particle.Alternatively, the streaks could be the analyte zones. emission from neutral Y atoms that have been created from the last solid residue. In either case, these residual streaks are Red Vapor Clouds and Particle Tracks FromWet Droplets of not very wide, which shows that the atoms from them are Solution Aerosols quickly converted into Y+ as they diVuse outwards from the residue into the plasma. The size and shape of the NAZ change with the audio plasma fluctuation, as noted previously.10,11 In roughly 60% of the Next, we return to Fig. 2(c). Droplet 2 shrinks and produces a bright, spherical, pale red cloud that lasts only briefly, hence frames, transient, pale red emission clouds are superimposed on the ambient IRZ and NAZ. Examples are shown in a its image is not elongated. A faint streak trails this small red sphere on the downstream side, i.e., pointing towards the tail sequence of successive frames in Fig. 1( b)–(d). In Fig 1( b), a large droplet leaves the IRZ displaced oV-center. Cooler con- of the plasma. Meanwhile, another large droplet (3) leaves the IRZ on-center, while droplet 4 can be seen inside the axial ditions surrounding this droplet produce the oval-shaped, pale red cloud above and to the left of the tip of the IRZ in channel oV-center. In Fig. 2(d), droplets 1 and 2 are gone, and droplets 3 and 4 both evolve into bright, spherical clouds Fig. 1( b). This droplet moves further downstream and forms a tapered, comet-shaped cloud in Fig. 1(c). This ‘comet’ cloud followed by faint streaks. These pale red spheres with streaks pointing downstream displays a noticeable gap between the diVuse ‘tail’ and the 1140 Journal of Analytical Atomic Spectrometry, October 1997, Vol. 12(a) (b) (c) (d) Fig. 1 Sequence of four consecutive frames during introduction of wet aerosols, 5000 ppm Y. The plasma is horizontal and flows from right to left. The induction region and most of the torch are blocked from view by a shielding box.In this and subsequent photographs, the width of the plasma at the far right is approximately 16 mm, and time begins at the top and progresses to the bottom. (a) Ambient emission structure; (b) a large droplet is seen above the center line; (c) the droplet shrinks into a comet-like shape; (d) the droplet vaporizes and is replaced by a thin white streak. There is a spherical cloud of blue Y+ around the streak in (d).Note also the thin, wispy region to the left of the tip of the IRZ in each frame. are the third and least common general shape for droplet ignition of precipitates in classical gravimetric analytical procedures. 17 In a study with isolated droplets in analytical flames, clouds mentioned in the previous section. This shape is attributed to rapid decomposition, perhaps even explosion, of a Bastiaans and Hieftje18 proposed a similar explosive process as one possible mechanism for the decomposition of analyte particle or a droplet in the final drying stages.For these clouds to be spherical, the time frame of this ‘explosion’ must be fast particles that have dried on the outside and entrapped solvent on the inside. Our observations diVer from those of Bastiaans relative to the exposure time (0.12 ms), otherwise the clouds would be elongated. Gradual atomization of the residue then and Hieftje in that we see evidence of solid particles or residues after the explosion, whereas they saw evidence of a smaller causes the faint white streak.Observation of this succession of events in consecutive frames is unusual; generally, a given wet droplet after explosion of the crust. The isolated droplets used by Bastiaans and Hieftje were #60 mm in diameter, i.e., droplet is seen in only one frame. The precise causes of these ‘explosions’ are unclear. One much larger than any of the droplets used in this study, which could explain why we did not see wet droplets after the possible source is as follows.Suppose a particularly large droplet dries from the outside in. This drying process creates explosions. Childers and Hieftje18 report a morphological investigation a solid crust with water trapped inside. The trapped water overheats, boils suddenly and ruptures the crust. Such of collected particles from a flame that is also pertinent to this work. In their study, samples of KCl were introduced as ‘explosions’ of occluded solvent occur at times during the Journal of Analytical Atomic Spectrometry, October 1997, Vol. 12 1141(a) (b) (c) (d) Fig. 2 Another sequence of four consecutive frames during introduction of wet aerosols, 5000 ppm Y. (a) A very large droplet (1) leaves the IRZ; (b) droplet 1 moves downstream and shrinks into a comet shape, while another large droplet (2) leaves the IRZ; (c) droplet 1 has become a faint white streak, droplet 2 dries and produces a small, spherical pale red cloud followed by a faint streak, and droplets 3 and 4 emerge from the IRZ; (d) droplets 3 and 4 also decompose into small, spherical clouds and streaks.uniform, wet droplets of diameter 67 mm. Solid particles were though all droplets were the same size! A larger variation of possible morphologies probably arises from the polydisperse collected on either a MgO-coated slide or the sample stub of a scanning electron microprobe for examination. Solid particles aerosol produced by a conventional nebulizer, such as that used in this paper and in most analytical work.of a variety of sizes and shapes were found. Some were regular cubic crystals of KCl, especially if the collection point was The ‘explosions’ of some of the Y(NO3)3 particles could also be related to recent observations concerning possible near the onset of vaporization of the droplets. Other particles were agglomerates of many microparticulates or crystals that atomization mechanisms of metal nitrates in electrothermal atomic absorption spectrometry.Holcombe and co- had melted partially. Finally, some of the collected particles were the residues of hollow shells that had been shattered, workers19–21 believe the main mechanism for formation of gaseous metal oxides first involves the formation of a solid either in the flame or on impact with the collection surface.18 These hollow residues are thought to be analogous to the metal oxide: explosions seen in this work and described in this section.Our observations and those of Hieftje and co-workers M(NO3)2(s)�MO(s)+2NO2(g)+DO2(g) (1) illustrate the following important point: for a given sample composition, all droplets and particles need not dry and decom- The solid metal oxide then decomposes into MO (g), often by an explosive shattering of the solid MO. L’vov and pose in precisely the same way. In Hieftje’s work, diVerent morphologies were seen at a particular sampling position even Novichikhin22 argue that gaseous metal oxide can be formed 1142 Journal of Analytical Atomic Spectrometry, October 1997, Vol. 12directly from the nitrate: For example, the shadow stop in Perkin-Elmer SCIEX instruments collects solids, minimizes formation of deposits on the M(NO3)2(s)�MO(g)+2NO2(g)+DO2(g) (2) subsequent ion lenses, and thus helps reduce drift and improve stability. If either mechanism occurs inside an isolated microparticulate in an ICP, our view is that the gas liberated will shatter the These photographs also show that the common practice of specifying the sampling position in ICP-MS, or the observation microparticulate and cause an ‘explosion’ such as those shown in Figs. 2(c) and (d). Neither Holcombe nor L’vov studied position in ICP-AES, relative to the tip of the IRZ is at best a time-averaged approximation. For wet aerosols, the position yttrium nitrate, which is the material used in this work. At any rate, the photographs in Figs. 1 and 2 show that, for of the tip of the IRZ varies both axially and radially on a millisecond time scale, especially when large droplets pass yttrium dissolved in nitrate solutions, there are at least two diVerent processes by which the final stages of drying lead to through the plasma. For dry particles, variations in the particle size and number density also contribute to instabilities in the formation of solid particles and then free atoms in the ICP.Most droplets evaporate smoothly, but a minority undergo position of the tip of the IRZ. these abrupt ‘explosions.’ Lateral Position and ‘Bunching’ of Droplets Overall Depiction of Drying and Vaporization Processes Examination of the entire film shows that the larger droplets often occur together as bunches of several droplets. Such a Fig. 3 shows our interpretation of these photographs pertinent bunch of three droplet clouds is shown in Fig. 2(c). Montaser to the fate of droplets in the ICP.We assume that a tapered, and co-workers23,24 observed a similar bunching of large elongated red vapor cloud represents the path of a spherical droplets from scattering measurements of the droplet stream cloud that shrinks as it moves axially during the exposure. leaving the spray chamber. Four such spherical clouds are shown in Fig. 3; each of these The red clouds are most frequent along the central axis of four clouds is produced by the same droplet. For some droplets, the plasma. Some droplets are occasionally seen oV-center (e.g., a spherical cloud from an explosion forms next, followed by a Fig. 1), but they tend to be smaller and disappear faster than faint streak and a spherical blue cloud of Y+. Most droplets those along the center line. These oV-center droplets have dry and vaporize smoothly without the explosion passed through the axial channel close to the induction region. Inside the load coil, this outer annular region of the axial Analytical Implications of Droplet Clouds and Particle Tracks channel is hotter than the center line.Thus, the large wet droplets dry and vaporize more eVectively if their path through Individual, isolated white streaks such as those shown in the axial channel is oV-center. The greater frequency of large Figs. 1(d) and 2(c) are seen only occasionally when solutions droplets on-center could also mean that the larger droplets are nebulized. However, most of the photographs show a fuzzy are enriched along the center axis of the gas flow out of the region beginning at the tip of the IRZ and extending downtorch injector.stream into the NAZ, e.g., the faint grey zone at the left of the This argument is complicated by the fact that the photo- tip of the IRZ in Figs. 1( b) and 2(d). This region is presumably graphic process integrates the emission through the thickness caused by juxtaposition of a number of such faint streaks, of the plasma.Suppose the plasma is sliced into a collection none of which is suYciently prominent to stand out individuof segments in a manner analogous to the Abel inversion ally. Although the ICP is often considered to have two analyte procedure. A slice from the camera through the plasma along zones, the IRZ and the NAZ,16 in fact there are three distinct the central axis is the thickest slice and thus contains the most regions, as noted in our first photographic study of noise droplets, even if the droplets are actually distributed uniformly behavior.10 The analyte that remains on-center first passes with respect to the radius.through the IRZ, then the grey, transition zone, and finally What are the emitting species responsible for the red vapor the NAZ. clouds? These photographs probably cannot distinguish In ICP-MS, the sampler is usually positioned just downbetween Y (I) atomic lines and YO bands, both of which are stream from the tip of the IRZ, i.e., right in the grey, streaky in the red part of the spectrum.Both neutral atoms and oxides zone. In most ICP-MS experiments, therefore, the solid residues are probably present in the vicinity of wet droplets, and the that cause the streaks pass directly into the sampler, as do the red emission from the vapor clouds shown in Fig. 1 is probably large wet droplets [e.g., Fig. 2(a)]! It is, therefore, not suprising a mixture of both Y (I) lines and YO bands. that solids deposit on the skimmer, photon stop or ion lens.Desolvated Solution Aerosols In a total of at least 20 000 frames from five separate films, large red droplet clouds are never seen when the aerosol is desolvated, i.e., by drying the droplets at 140 °C and then condensing the solvent. This observation proves that the red clouds are indeed from wet droplets, as proposed in our previous work. A variety of adverse eVects such as oxide ions in ICP-MS and noise in ICP-AES and ICP-MS have been blamed on fluctuations in analyte density caused by passage of these droplets.This present work shows that desolvation is a simple way to remove them. Also, the red emission from the IRZ is much less prominent, the tip of the IRZ is more stable with time, and the grey streaky zone is much less evident when the solvent is removed by desolvation. The removal of the streaky region has an additional advantage for ICP-MS. The sampling process in ICP-MS is highly localized spatially, i.e., the analyte ions that pass through the Fig. 3 Drawing depicting evaporation of solvent from droplet, followed by explosion, atomization and ionization of residue. skimmer originate from a small zone just in front of the Journal of Analytical Atomic Spectrometry, October 1997, Vol. 12 sampler. If a solution aerosol is desolvated, very few streaks from gaseous YO vaporized from the Y2O3 particle. This supports our hypothesis that these streaks are caused mainly are present, and the analyte atomizes in a narrower range of axial positions.Several workers report significant improve- by emission from hot, solid particles. ments in sensitivity (i.e., analyte signal per unit concentration) in ICP-MS when the aerosol is dried,25–29 especially for double Behavior of 0.1 mm Slurry Particles focusing mass spectrometers.30,31 Peters and Beauchemin29 also showed that noise can be reduced to some extent by pre- When the 0.1 mm Y2O3 particles are injected as wet droplets, the usual ‘ambient’ emission structure is observed.A single, evaporating the aerosol before it is injected into the ICP. Both these sensitivity enhancements and noise reductions are prob- thin white streak is also seen along the entire length of the plasma (Fig. 5a). Apparently, many more particles are now ably due, at least partly, to the removal of the droplet clouds and the grey streaky region when the aerosol is desolvated present than is the case for the larger 3 mm particles, so the individual tracks appear to be one continuous streak.This or dried. streak is seen only along the central axis; it is never displaced radially. Also, red vapor clouds are seen only near the tip Particle Tracks From Desolvated Y2O3 Slurries of the IRZ, and they occur only briefly and occasionally. Desolvation of these slurry particles does not greatly change The various Y2O3 slurries described in the Experimental section produce diVerent behavior than the yttrium solution the appearance of the plasma.The infrequent red vapor clouds from wet droplets are removed completely when the aerosol is described above. The slurries are desolvated in this experiment so that the dry, solid particles can be distinguished from wet desolvated, as expected. As shown in Fig. 5b, the central streak is slightly wider and droplets. With desolvated slurries, the background IRZ and NAZ are visible but less intense than when an yttrium solution more diVuse when the droplets are desolvated.The reasons why the central streak is narrower when the slurry particles is nebulized. The finer slurries (0.1 and 3.2 mm particle diameter) produce faint, wispy, white streaks through the center of the are introduced in wet droplets are not clear. We suggest the following explanation. Suppose the larger, heavier droplets plasma. For the 0.1 mm slurry, many such streaks combine into a more or less continuous white line down the central axis of remain tightly localized along the center line.It is possible that the gas flow velocity out of the torch injector is suYciently the plasma. For the 3.2 mm slurry, individual streaks can be discerned (Fig. 4). The slurry with the largest particles (8.5 mm) high to enrich large droplets along the center line. The same eVect is used in the jet separator interface for gas produces only ambient yttrium emission, with no streaks. In this last case, the plasma looks much like that from a desolvated chromatography–MS,34 in the particle beam interface for liquid chromatography–MS,35 and in other applications of momen- aqueous solution of yttrium.The white streaks from the particles in Fig. 4 are clearly tum separators. Desolvated particles contain the same amount of analyte but no solvent. They are thus much lighter than evident but are much less prominent than the red vapor clouds discussed previously. The shape and orientation of the white wet droplets and spread out more extensively as they leave the injector of the torch.streaks from the 3 mm slurry are also interesting. They occur at uniform intervals only along the center line of the plasma, Wet aerosols containing these 0.1 mm slurries produce many fewer red emission clouds than the yttrium solution, even they tend to be tilted relative to the long axis of the plasma, and they also curl noticeably at the ends. This tilt and curled though there is more total yttrium in the slurry.Those droplet clouds that are seen exist only in individual frames and do not shape could result if the solid particles swirl slightly with the gas flow in the plasma, as suggested by French.32 Such a swirl persist downstream in the NAZ, even though the nebulizer and spray chamber are the same as those used to produce the eVect could be important in experiments with the monodisperse dried microparticulate injector (MDMI), where the particle droplet clouds shown in Figs. 1 and 2. Each of these droplets contains many Y2O3 particles, as described in the next section. stream is highly localized and must be aimed precisely at the sampling orifice of the MS instrument.6–9,33 Very little yttrium dissolves into the surrounding water, because the slurry is not acidified and Y2O3 is not appreciably soluble In the sequence chosen in Fig. 4, one streak (labelled by the letter ‘A’) stands out as being brighter than the others. This at pH#7 [for Y(OH)3, Ksp#10-23].36 Apparently, the yttrium must be in solution in a wet droplet to produce red vapor clouds brighter streak could be an agglomeration of several slurry particles or two separate particles that happen to pass through that persist in the NAZ.Also, no ‘explosions’ are seen from any of the slurries. the plasma nearly together. This particle streak appears in each frame in Fig. 4 and survives all the way down the central The arguments in the previous paragraph indicate an important facet of the red clouds from wet solution droplets described axis of the plasma.Apparently, at least some of the Y2O3 particles do not atomize completely in the plasma. Brighter previously. Many of these clouds are detached from the IRZ and are surrounded by the blue emission from the NAZ. streaks such as this are unusual; most frames show a succession of uniformly-spaced streaks of similar intensity. It is possible Examples are droplets 1 and 2 in Fig. 2(a) and (b). Such detached clouds are seen only when the yttrium is dissolved that the streaks come solely from agglomerates, not from individual 3 mm particles. However, the observation that most in the droplet. Thus, these isolated red clouds consist primarily of Y (I) and/or YO emission from the yttrium contained within streaks have the same size and intensity argues that the particles either did not agglomerate severely in the liquid phase that droplet, not from Y or YO formed from ‘ambient’ Y+ ions that have been cooled by passage of the droplet.Otherwise, before the sample was nebulized and introduced into the plasma, or that the agglomerates were broken up by the isolated red vapor clouds would also be seen when wet slurries are introduced, as these droplets would cool the ‘ambient’ Y+ nebulization and desolvation processes. Otherwise, the streaks would be of various sizes. into neutral Y and/or YO. Near the central axis all particle streaks are surrounded by a faint transition zone of diameter 2 mm or less.Within a Estimate of Number of Solid Particles per Droplet for Slurries short distance from the axis, bright blue emission from excited Y+ is observed. Apparently, ionization occurs rapidly after Goodall et al.37 provide general guidelines on the subject of slurry analysis that are relevant to this work. For accurate atomization, because the vapor clouds bear the blue color of excited Y+, not the red of neutral Y or YO emission.analysis of slurries using dissolved aqueous standards, the main criterion is that the particle concentration should corre- It is also interesting that the fine streaks shown in Fig. 4 are not red, as would be the case if they were caused by emission spond to no more than one solid particle per wet droplet (on 1144 Journal of Analytical Atomic Spectrometry, October 1997, Vol. 12Fig. 4 Four consecutive frames during introduction of Y2O3 slurry containing 3 mm particles at 10 000 ppm Y.The solvent was removed by desolvation in this experiment. Note the discrete white tracks through the center line of the plasma. The letter ‘A’ marks the passage of a particularly large particle, or possibly two particles in close proximity. average). Generally, the solid should be ground into particles diVerent degrees that depend on the chemical and physical properties of the particles and the slurry medium. Even at the no more than 3 mm in diameter to satisfy this requirement. Denser solids require more thorough grinding into finer relatively high solid load (100 000 ppm Y) used for the colloidal slurry in the present work, the #400 solid particles in a typical particles still.For comparisons with these criteria, calculated numbers of (2.5 mm) droplet would coalesce into a single dry particle only #0.7 mm in diameter, which is still under the maximum size particles per wet droplet are shown in Table 2 for the three slurries used in this work.For the colloidal slurry (0.1 mm of 1.2 mm determined by the criteria of Goodall et al.37 for Y2O3 (density=5.0 g cm-3 39). diameter particles), the calculations show that 10 mm wet droplets each contain #30 000 solid particles. However, only Calculated particle loadings per droplet for the other Y2O3 slurries are also shown in Table 2. These data suggest that few a few of these large wet droplets reach the plasma. Most droplets that leave the spray chamber used in this work are in wet droplets that are small enough to escape the spray chamber (i.e., <5 mm diameter) contain solid particles in the sizes used the size range 0.5–5 mm, with maximum transmission in the 2–3 mm range.38 Even so, these ‘average’ droplets of 2.5 mm (3.2 or 8.5 mm).In other words, only 1% of the 5 mm wet droplets contain a 3.2 mm solid particle, while only 0.05% of diameter still contain #400 of the 0.1 mm diameter particles.The fate of these multi-particle droplets is unclear. Do they such droplets contain an 8.5 mm particle. This low value for occupancy (i.e., low value for solid particles per wet droplet) dry into many discrete 0.1 mm particles, do they agglomerate in the solution, or is an agglomerate formed as the droplets explains why the 8.5 mm slurry produced few visible streaks. The data in Table 2 also show that the probability for occu- dry in the plasma? All three processes probably occur to Journal of Analytical Atomic Spectrometry, October 1997, Vol. 12 1145and experience higher temperatures along the outer edge of the axial channel, i.e., they pass closer to the hot induction region than droplets that stay along the center line. These observations are pertinent to the dictum of Goodall et al.37 that calibration of the analyte response from a slurry using external calibration with an aqueous solution works best if the particle size and total slurry level are adjusted so that each wet droplet contains no more than one slurry particle.In this case, each droplet dries into one slurry particle, whose behavior in the plasma is not greatly diVerent from that of a solution residue. Suppose a spray chamber is used that transmits droplets primarily in the size range 3–10 mm, with a maximum at 5 mm. The estimated solid particle sizes shown in Table 3 indicate that wet droplets (3–10 mm) containing dissolved yttrium nitrate or oxide should dry into solid particles in the range 0.4–2.2 mm.Slurries are usually ground to particle sizes of 1–3 mm (depending on density), which would be expected to vaporize and atomize in a similar fashion as the calibration solution. The particles in the 3 mm Y2O3 slurry are too large for single occupancy. We also suggest that the standard additions protocol should work well when combined with slurry nebulization and the single occupancy criterion. If each droplet contains one particle, the solution phase elements can ‘condense’ onto this particle as the droplet dries.Thus, the elements from the calibration solution are released into the plasma in more or less the same way and time as the atomized elements from the solid slurry particle. Finally, we call the reader’s attention to the observation that the 0.1 and 3 mm slurries, either dry or wet, produce a noticeable line or set of discrete tracks along the axis of the plasma (Figs. 4 and 5).These tracks persist for the entire length of the NAZ, even for the 0.1 mm slurry, which suggests that the particles are not fully atomized, although suYcient yttrium is atomized and ionized to provide the usual ambient a b emission structure. Fig. 5 Drawings contrasting appearance of ICP during introduction This observation supports those of several others concerning of 0.1 mm Y2O3 slurry particles. a, Particles introduced as wet droplets; atomization of slurries of refractory solids.Raemaekers et al. b, particles introduced as dry, desolvated particles. The dried slurries (Fig. 6 of ref. 15) found that the Al emission sensitivity from produce a wider line of particle tracks through the center of the plasma. Also, the wet droplets produce very few large red droplet an Al2O3 slurry (mp=2045, bp=2980 °C)39 was close to but clouds. 2–10% below that from a solution of Al, with better agreement at lower aerosol gas flow rate. Ebdon and co-workers40,41 introduced wet MgO particles (mp=2852, bp=3600 °C)39 with Table 2 Calculated values for number of solid Y2O3 particles per a mean size of 2 mm and a maximum size of 4 mm.They found droplet wet of given size, slurry nebulization that these particles were atomized with an eYciency of 80–85%. No. of solid particles The yttrium oxide particles (mp=2410 °C, bp=?) 39 used in Solid Y per wet droplet of specified size this work are in this same size range, are also refractory, and particle concentration thus would not be expected to atomize fully in an ICP.We Size/mm (ppm) 0.5 mm 2.5 mm 5.0 mm 10mm suggest the un-atomized fraction travels through the plasma 0.1 100 000 3 400 3000 30 000 3.2 10 000 1×10-5 1×10-3 0.01 0.08 8.5 10 000 5×10-7 6×10-5 5×10-4 4×10-3 Table 3 Calculated values for diameter of solid particle of yttrium oxide or nitrate at 10 000 ppm in solution produced by drying wet droplets of particular sizes pancy of a droplet by multiple particles decreases rapidly with Estimated diameter of dry particle increasing particle size.Thus, very few droplets would be (mm) for Diameter of wet expected to contain more than one 3.2 mm particle, so the droplet/mm Y2O3* Y(NO3)3† white streaks in Fig. 4 are probably caused by individual particles rather than agglomerates formed in the plasma. 0.5 0.07 0.11 For the larger slurry particles, most of the droplets that 1 0.14 0.22 leave the spray chamber contain no Y2O3. Nevertheless, the 2 0.27 0.45 ICP still shows the usual background structure of red IRZ 3 0.41 0.68 5 0.68 1.1 followed by blue NAZ.Thus, some yttrium is atomized 10 1.4 2.2 upstream in the plasma to provide the necessary Y atoms and 20 2.7 4.5 ions. Presumably, this ambient yttrium comes from relatively 30 4.1 6.8 small droplets that beat the odds by containing an Y2O3 50 6.8 11.3 particle. These ‘lucky’ droplets that contain yttrium then dry early enough for at least some atomization of yttrium.Some * Density=5.0 g cm-3, ref. 39. † Density=2.7 g cm-3, ref. 39. of these ‘lucky’ droplets are also sprayed outwards radially 1146 Journal of Analytical Atomic Spectrometry, October 1997, Vol. 12inside the white axial streak(s) depicted in Figs. 4 and 5. As CONCLUSIONS noted above, these streaks would pass straight into the The photographs presented here illustrate several points sampling orifice of an ICP-MS device. pertinent to practical analysis with the ICP: 1. Desolvation removes the deleterious eVects of wet droplets.Comparisons With Olesik’s Studies 2. There are two general ways that wet droplets dry into solid particles: (a) by gradual removal of the solvent, and (b) Many of the phenomena from these photographs agree closely by gradual drying followed by rapid explosion of a residue. with observations made by Olesik and co-workers,2–9 who use 3. Wet droplets that are injected oV-axis are dried more time-resolved emission, fluorescence, scattering and MS to eVectively than those that pass through the plasma on-center.study the fate of wet droplets and the subsequent solid particles 4. When wet droplets are introduced, the position of the in the ICP. In the subsequent discussion, we refer mainly to IRZ varies with time. Also, there is a third analyte zone along our observations made during introduction of wet aerosols. the central axis consisting of thin streaks caused by the residues 1.A wet droplet ‘cools’ a zone 1–2 mm in diameter, which of the wet droplets. These faint streaks could be due to small corresponds roughly to the width of the droplet clouds shown solid particles formed as the droplets dry, as suggested by in Fig. 1. Olesik and co-workers. These streaks are most abundant right 2. Large red clouds from wet droplets are much more in the usual sampling position used for ICP-MS. numerous than the fine streaks from particles. 5. Introduction of yttrium as Y2O3 particles in slurries 3.Small droplets desolvate and atomize upstream in the produces the usual ambient emission structure with intact solid axial channel, probably inside the section enclosed by the particles localized along the center line of the plasma. These induction region, and provide the ambient emission structure slurry particles behave similarly to, but not the same as, of the ICP. dissolved solutes. Y2O3 particles of 3 mm diameter can be seen 4. Most droplet clouds or particle tracks are seen on-center to survive intact all the way down the central axis of the near the IRZ at high aerosol gas flow rate.plasma. These particles are larger than the calculated size of 5. Solid particles survive only a short distance after the 1.2 mm for the single occupancy criterion for Y2O3 particles in droplet dries. The free neutral atoms are then excited and 10 mm wet droplets at 1% total slurry loading. ionized rapidly. 6. Blue Y+ emission is formed quickly after the solid par- 6.There is no exact ‘vaporization position’ for solid particles ticles decompose, as if a significant fraction of the excited Y+ from polydisperse droplets. Instead, vaporization occurs over is formed directly from neutral Y in a single elementary step. a finite range of axial positions. 7. The larger wet droplets often appear together in bunches There are two important diVerences in the observations of two or three droplets, as suggested by Montaser and described in this work compared with those of Olesik and co-workers.23,24 co-workers.First are the ‘explosions’ shown in Fig. 1 and Olesik42,43 has recently published two new papers that are discussed above. Second is the basic cause of the red droplet directly relevant to the fates of droplets and particles in the clouds. Olesik’s studies with dual nebulizers indicate that many ICP. The first paper is mainly a review of previous measure- of the droplet clouds represent emission from ytttrium that is ments and describes many of the observations cited above.already present as atoms and ions and is cooled by the water For example, neutral atoms of Sr and Y are very rapidly evaporating from the droplet. If this is the main cause of converted into excited ions, i.e., in 100 to 150 ms. Strontium droplet clouds, introduction of wet slurries would show about atom emission produces a cloud that becomes only #1 mm the same number of isolated red droplet clouds as the nebulized diameter until it disappears as the neutral Sr is converted into solutions, and the number of isolated red clouds would be Sr+ ions.The size of this Sr atom emission cloud is similar to about the same for any of the diVerent size slurries used, since that of the ‘explosions’ shown for yttrium nitrate in Fig. 2. The the same nebulizer and spray chamber are used for either Sr+ emission clouds shown are of similar size as those reported sample.As indicated previously, we see red droplet clouds in the present work.42 from slurries only well upstream in the axial channel near the The second paper shows new results obtained with mono- tip of the IRZ. disperse droplets that have been dried partially to approxi- This diVerence of opinion may be partly due to diVerences mately 13 mm diameter. Such droplets produce fluorescence in the droplet size distribution transmitted by the spray clouds from excited Sr+ in its lowest electronic state that are chambers used.The size measurements for tertiary aerosols of similar size as the emission clouds from excited Sr+ i.e. 3, (i.e., those droplets leaving the spray chamber) reported by to 4 mm diameter depending on the distance from initial Olesik and Fister3 show a substantial number of droplets production of neutral Sr atoms. The fluorescence signal from between 10 and 20 mm, whereas our previous measurements Sr atom falls to zero only 50 ms after it reaches its maximum. show very few droplets above 10 mm diameter.38 Very large The total lifetime of the Sr atom cloud is only #70 ms, which droplets above 10 mm, if present in greater numbers in Olesik’s is fast enough to cause the ‘explosion’ clouds shown above to plasma than in ours, could account for their view that many be spherical rather than elongated.43 It is possible that the droplet clouds represent emission from material already atom- ‘explosions’ shown in the present work are merely small clouds ized but cooled by passage of the droplet through the plasma, of emission from neutral Y atoms.This explanation does not as opposed to our clear evidence that an isolated droplet cloud account for the fact that these clouds are seen in only a small is caused by emission only from yttrium within that particular fraction of the frames. Also, the total lifetime of the Sr atom droplet. However, aerosol size distributions measured with cloud measured by Olesik is #70 ms,43 which should be long diVerent devices may not be closely comparable.enough to cause the ‘explosion’ clouds shown in the present Another possible diVerence between our measurements and work to be elongated rather than spherical. those of Olesik and co-workers may be related to the axial location of the droplet clouds. Our photographs are primarily sensitive to red clouds that are physically separate from the Ames Laboratory is operated by Iowa State University time-averaged red emission of the IRZ.Olesik’s work shows for the US Department of Energy under Contract No. that the droplet clouds are most abundant near the tip of the W-7405-Eng-82. This research was supported by the OYce of IRZ, and it is possible that we miss most of the droplet clouds that are not completely separated from the ambient IRZ. Basic Energy Sciences. The authors are grateful to John Olesik, Journal of Analytical Atomic Spectrometry, October 1997, Vol. 12 114722 L’vov, B. V., and Novichikhin, A. Spectrochim. Acta, Part B, 1995, Barry French and Gary Hieftje for valuable discussions con- 50, 1427. cerning the interpretation of these observations. 23 CliVord, R. H., Sohal, P., Liu, H., and Montaser, A., Spectrochim. Acta, Part B, 1992, 47, 1107. 24 McLean, J. A., HuV, R. A., and Montaser, A., paper presented at REFERENCES the Federation of Anal. Chem. Spectrosc. Socs. Conference, Kansas City, MO, September 1996, Paper No. 494. 1 Cicerone, M. T., and Farnsworth, P. B., Spectrochim. Acta, Part 25 Lam, J. W., and McLaren, J. W., J. Anal. At. Spectrom., 1990, B, 1989, 44, 897. 5, 419. 2 Olesik, J. W., Smith, L. J., and Williamsen, E. J., Anal. Chem., 26 Tsukahara, R., and Kubota, M., Spectrochim. Acta, Part B, 1990, 1989, 61, 2002. 45, 581. 3 Olesik, J. W., and Fister, J. C., III, Spectrochim. Acta, Part B, 27 Jakubowski, N., Feldmann, I., and Stuewer, D., Spectrochim. 1991, 46, 851 and 869.Acta, Part B, 1992, 47, 107. 4 Hobbs, S. E., and Olesik, J. W., Anal. Chem., 1992, 64, 274. 28 Jakubowski, N., Feldmann, I., and Stuewer, D., J. Anal. At. 5 Hobbs, S. E., and Olesik, J. W., Spectrochim. Acta, Part B, 1993, Spectrom., 1993, 8, 969. 48, 817. 29 Peters, G. R., and Beauchemin, D., Spectrochim. Acta, Part B, 6 Olesik, J. W., and Hobbs, S. E., Anal. Chem., 1994, 66, 3371. 1993, 48, 1481. 7 Dziewatkoski, M. P., Daniels, L. B., and Olesik, J. W., Anal. 30 Hutton, R.C., personal communication, 1995. Chem., 1996, 68, 1101. 31 Wiederin, D. R., paper presented at the Federation of Anal. 8 Olesik, J. W., and Dziewatkoski, M. P., J. Am. Soc. Mass Chem. Spectrosc. Socs. Conference, Kansas City, MO, September Spectrom., 1996, 7, 362. 1996, Paper No. 490. 9 Olesik, J. W., Anal. Chem., 1996, 68, 469A. 32 French, J. B., personal communication, 1995. 10 Winge, R. K., Eckels, D. E., DeKalb, E. L., and Fassel, V. A., 33 French, J. B., Etkin, B., and Jong, R., Anal. Chem., 1994, 66, 685. J. Anal. At. Spectrom., 1988, 3, 849. 34 Watson, J. T., Introduction to Mass Spectrometry, Raven, New 11 Winge, R. K., Crain, J. S., and Houk, R. S., J. Anal. At. Spectrom., York, 1985, p. 131. 1991, 6, 601. 35 Willoughby, R. C., and Browner, R. F., Anal. Chem., 1984, 56, 2626. 12 Scott, R. H., Fassel, V. A., Kniseley, R. N., and Nixon, D. E., 36 Vickery, R. C., T he Chemistry of Yttrium and Scandium, Pergamon, Anal. Chem., 1974, 46, 75. New York, 1960, p. 52. 13 Fassel, V. A., and Bear, B. R., Spectrochim. Acta, Part B, 1986, 37 Goodall, P., Foulkes, M. E., and Ebdon, L., Spectrochim. Acta, 41, 1089. Part B, 1993, 48, 1563. 14 Van Borm, W. A. H., Broekaert, J. A. C., Klockenkamper, R., 38 Wiederin, D. R., and Houk, R. S., Appl. Spectrosc., 1991, 45, 1408. Tscho� pel, P., and Adams, F. C., Spectrochim. Acta, Part B, 1991, 39 CRC Handbook of Chemistry and Physics, ed. Weast, R. C., CRC 46, 1033. Press, Cleveland, OH, 52nd edn., 1971, pp. B-64, B-106 and B-153. 15 Raemaekers, B., Graule, T., Broekaert, J. A. C., Adams, F., and 40 Ebdon, L., Foulkes, M. E., and Hill, S., J. Anal. At. Spectrom., Tscho� pel, P., Spectrochim. Acta, Part B, 1988, 43, 923. 1990, 5, 67. 16 Koirtyohann, S. R., Jones, J. S., and Yates, D. A., Anal. Chem., 41 O’Hanlon, K., Ebdon, L., and Foulkes, M., J. Anal. At. Spectrom., 1980, 52, 1965. 1996, 11, 427. 17 Diehl, H., and Smith, G. F., Quantitative Analysis, Wiley, New 42 Olesik, J. W., Appl. Spectrosc., 1997, 51, 158A York, 1952, p. 32. 43 Olesik, J. W., Kinzer, J. A., and McGowan, G. J., Appl. Spectrosc., 18 Bastiaans, G. J., and Hieftje, G. M., Anal. Chem., 1974, 46, 901; 1997, 51, 607. Childers, A. G., and Hieftje, G. M., Appl. Spectrosc., 1986, 40, 939. 19 Bass, D. A., and Holcombe, J. A., Anal. Chem., 1987, 59, 974. 20 Jackson, J. G., Novichkin, A., Fonseca, R. W., and Holcombe, Paper 6/07579G J. A., Spectrochim. Acta, Part B, 1995, 50, 1423. Received November 7, 1996 21 Jackson, J. G., Fonseca, R. W., and Holcombe, J. A., Spectrochim. Acta, Part B, 1995, 50, 1449. Accepted April 30, 1997 1148 Journal of Analytical Atomic Spectrometry, October 1997, Vol