Time-resolved Signals from Particles Injected into the Inductively Coupled Plasma Journal of Analytical Atomic Spectrometry KEVYN KNIGHT Department of Chemistry Birkbeck College (University of London) Gordon House 29 Gordon Square London UK WCI H OPP SIMON CHENERY British Geological Survey Keyworth Nottingham U K NGl2 5GG STAN W. ZOCHOWSKI MICHAEL THOMPSON AND COLIN D. FLINT Department of Chemistry Birkbeck College (University of London) Gordon House 29 Gordon Square London UK WCl H OPP Atomic emission signals derived from single particles in the 1-10 pm size range have been observed after injection of refractory oxides and silicates into the inductively coupled plasma by the nebulization of dilute suspensions. Detection limits by mass are in the 1 x 10-l3-l x lo-" g range because of the discrete nature of the signal and the low background against which the signals are observed.True calibration is currently impossible because of the unavailability of particles of accurately known mass but element-specific particle counting is easy and a valuable capability in itself for discrimination and identification of uncommon particles in complex mixtures. Keywords Inductively coupled plasma atomic emission spectrometry; particle; time-resolved signal; element-specijic particle detection; discrimination Previous work' has shown that when particulate matter in the 1-10 pm size range is introduced into an inductively coupled plasma (ICP) running under standard operating conditions atomic emission signals are observed in the form of peaks with a duration of about 0.5 ms.Each peak signal results from the discrete cloud of excited atoms derived from a single particle passing through the observation zone of the ICP. The signals are observed over an effectively zero background interrupted only by discrete events of much shorter duration derived from background-stray light photons thermal emission or external ionizing events. Because the atomic emission signals have a high intensity and a low background noise mass detection limits in the range 1 x 10-13-1 x lo-'' are possible depending on the sensitivity of the atomic line used if the signal is sampled at a suitably high frequency. The equipment previously described' was originally designed to study the particulate products of the laser ablation of refractory targets.2 However it was immediately found to be equally applicable to particles introduced into the ICP by the nebulization of aqueous suspensions.The acquisition of such information is potentially a valuable addition to the repertoire of inductively coupled plasma atomic emission spectrometric (ICP-AES) analytical techniques. It could throw light on fundamental aspects of particle-plasma interactions3 and on the efficacy of injection techniques such as laser ablation and the nebulization of slurries. Mass calibration for discrete particles still remains a prob- lem however. Particles of known composition can be easily obtained but in nearly all instances of available material the mass range of the particles is relatively large. The only materials found so far with particles in the correct size range and with a suitably small variability are the suspensions of latex spheres used to calibrate particle size analysers and electron micro- scopes.Unfortunately the latex particles are not useful for time-resolved studies as they are atypically easy to atomize and because carbon is one of the least sensitive elements for detection by ICP-AES. Other commercially available materials claimed to have narrow size cuts in the appropriate range are too variable for the calibration of mass. For example if a powder contained particles between 2 and 4 pm in diameter the mass range would span a factor of eight. The work of Olesik and Hobbs3 describes a device that can repetitively inject particles of uniform mass into the ICP.However the particles are produced by the desolvation of aqueous solutions. In the present study the concern is with the practical impli- cations of injecting inherently particulate matter especially refractory materials. Work described in the present paper therefore covers quali- tative and semiquantitative aspects of the study. Signals often simultaneous multi-element signals have been obtained from a variety of refractory particulate materials of which the size characteristics have been confirmed by scanning electron microscopy (SEM) and Coulter Counter studies. The possi- bility of identifying particular types of particles in complex mixtures and establishing their relative proportions by count- ing was investigated. EXPERIMENTAL Equipment The equipment used has previously been described in detail.' A standard mono-/polychromator ICP-AES system was modi- fied by replacing the normal electronics.A fast multichannel analogue to digital converter (ADC) and data acquisition system was attached via virtual-earth preamplifiers directly to the anodes of the photomultiplier tubes and a stabilized high- voltage supply was provided. Data collected by the system were treated off-line by means of proprietary and in-house software. Each channel selected was sampled in the present study at a frequency of 20 kHz. The ICP was operated under conditions used for conven- tional analysis of aqueous solutions i.e. aerosol injector gas at 1.0 1 min-' intermediate gas at 0.0 1 min-' and outer plasma gas at 12.0 1 min-' of argon. Signals were transmitted to the spectrometers by optic fibres mounted at a height of 12mm above the load coil and sampling a vertical distance of 17 mm in the ICP.No attempt was made to optimize the system for the observation of individual particles. Suspensions were pre- pared by ultrasonic agitation of the powders with water containing a surface active agent (usually sodium dodecyl- benzenesulfonate) and introduced into the plasma by means Journal of Analytical Atomic Spectrometry January 1996 VoE. 11 (53-56) 53of a Spectro Analytical high-solids (Babington-type) nebulizer mounted in a single-pass spray chamber with no impact bead. The SEM studies showed that both before and after nebuliz- ation suspensions of discrete (not aggregated) particles were obtained. Signals were monitored from photomultipliers serv- ing the following wavelengths Si 288.2; A1 308.2; Fe 259.9; Zr 343.8; and Ca 422.7 nm.125 - v) .$ 100- .- r 0. 'is 75 - t 0 5 50- z 25 - 0 Injection of Silica Particles The materials selected for the initial investigations were two grades of chromatographic silica with mean particle diameters of 3.0 and 7.6 pm. The SEM studies showed that the materials consisted of separated spheroids with a size variation of about f 50% by visual estimation. Suspensions of these materials in water containing respectively 3 and 6 mg 1-l were prepared and nebulized into the ICP. The response of silicon was monitored at the wavelength listed above. I I I I 1 I Synthetic Mixtures of Particles The following experiment was conducted to investigate the potential of time-resolved ICP-AES for the recognition of particles of a particular composition in a complex assemblage. A mixture of three minerals zircon (ideal formula ZrSiO,) olivine [( Mg,Fe)2Si04] and albite (NaAlSisO,) was ground to a fine powder.Each of these minerals contains an element absent in the other two and therefore capable of uniquely identifying a particle of the mineral in the mixture. A suspension of 0.3 g 1-l of the mixture was prepared in water and nebulized into the JCP. A mixture of two pyroxenes was also prepared. Orthopyroxene [ideal formula (Mg,Fe)$i&] and clinopyrox- ene [(Ca,Mg,Fe),Si208] were individually crushed and separ- ately injected into the ICP as a suspension (0.3 g 1-I) in water. The two minerals were hand picked from a single rock specimen so a mixture of the minerals would be a genuinely realistic problem.The elements used to discriminate between the minerals were calcium iron and silicon and appropriate wavelengths were monitored while the suspension was being nebulized. The signals were roughly calibrated by using the relative sensitivities obtained by the nebulization of aqueous solutions. Crushed Olivine for Measuring Multi-element Signals A sample of natural olivine ( Fe,Mg),Si04 known to be homo- geneous was crushed and suspended in water. Responses for iron magnesium and silicon were monitored while the suspen- sion was nebulized into the ICP. RESULTS AND DISCUSSION Silica Particles Typical portions of traces obtained for injection of silica particles are shown in Figs.1 and 2 which portray a series of peaks with a width of about 0.5 ms each corresponding to the passage of one particle through the plasma. The peak heights (and areas) are variable as shown in the histograms in Figs. 3 and 4. The variation arises mainly because of the mass differences between the particles. The duration of the signals from single particles was esti- mated by fitting a Voigtian profile to about 120 peaks in the data sets. The distribution of estimated peak widths at half- height are shown in Fig. 5. Both the value of the mean and the narrow dispersion are what would be expected from single particles passing through the cone of acceptance of the optic fibre light guides. The small proportion of high outlying widths 6o 1 -15 I I I I 1 I 0 2 4 6 8 10 Tirne/ms Fig.1 Sample output obtained on the silicon channel when particles of silica of mean diameter 3.0 pm were nebulized as a suspension 1000 1 800 c .- C $ 400 C 8 200 n - - I I I 0 5 10 15 20 Time/ms Fig. 2 Sample output obtained when particles of silica of mean diameter 7.6 pm were nebulized as a suspension. Same response units as Fig. 1 150 1 0 20 40 60 80 100 Peak height (arbitrary units) Fig. 3 Distribution of peak heights when particles of silica of mean diameter 3.0 pm are nebulized. Same response units as Fig. 1 probably represent overlapping signals from particles injected roughly simultaneously into the plasma. A 3 pm particle of the silica would have a mass of about 15 pg assuming a density of 1.0 g cm-3 (the material is porous). The average signal-to-background noise ratio suggested by Fig.1 indicates that a detection limit of about 1 pg could be achieved by the time-resolved ICP-AES system for the silica particles. Silicon is not a sensitive element when determined 54 Journal of Analytical Atomic Spectrometry January 1996 Vol. 1 125 i 0 100 200 300 400 Peak height (arbitrary units) Fig. 4 Distribution of peak heights when particles of silica of mean diameter 7.6 pm were nebulized as a suspension. Same response units as Fig. 1 35 1 0.00 0.15 0.30 0.45 0.60 0.75 Peak width at half-heightlms Fig. 5 Distribution of peak widths at half height when 7 pm particles of silica were nebulized as a suspension by ICP-AES so more sensitive elements could probably be detected with detection limits of 10-100 fg.The transport efficiency of the system (the proportion of the nebulized particles that reach the plasma) based on an assumed density of 1.0 for the silica particles was of the order of 0.5% for the 7.6 pm particles and 4% for the 3 pm particles. The 7.6 pm particles provided signals with higher peaks than the 3.0 pm but not in proportion to the relative masses of the particles. This effect could result if the vaporization of the larger particles were less complete of if there was a change in the excitation conditions owing to greater local cooling of the plasma gas in the vicinity of a larger vaporizing particle. Such effects have been observed in connection with the vaporization of large water droplets in the conventional nebulization of solution^.^ Both of the foregoing observations are relevant to an understanding of the calibration problems associated with the nebulization of slurries in conventional ICP-AES. Signals From Mixtures of Particles The signals from mixtures of particles were monitored simul- taneously at appropriate wavelengths.Typical signals are shown superimposed in Fig. 6. It is clear that each of the channels is responding separately to a different type of particle. There is no perceptible correlation between the traces. The prospects for identifying particles that contain an exclusive element as a major constituent are therefore good. A more demanding requirement would be the discrimination 625 645 665 685 705 725 745 765 785 805 Time (arbitrary units) Fig. 6 Superimposed simultaneous responses from zirconium alu- minium and iron channels when a mixture of three minerals was nebulized as a suspension Si A O ;a 0 0 0 Fe Fig.7 Apparent normalized compositions of particles of orthopyrox- ene (filled triangles) and clinopyroxene (open diamonds) showing the successful discrimination of the two minerals by nebulization of suspensions between particles that do not contain mutually exclusive elements but merely differ in the relative proportions of their constituents. This situation was approached when a mixture of two pyroxenes was examined. The most discriminating elements were found to be iron and calcium with silicon being less useful. Data for each particle reaching the plasma were plotted on a triangular diagram of normalized calibrated responses for three elements. (Signals were normalized by summing the calibrated responses to loo%.) The diagrams for the two minerals are shown super- imposed in Fig.7. There is an almost complete discrimination between the two pyroxenes mostly in the calcium-iron direc- tion as expected. The small proportion of particles outside the main composition zones could be the result of a failure to separate the minerals completely before crushing. This system showed the potential of time-resolved ICP-AES for recognizing specific types of particles among mixtures by means of their multi-element response signatures. Although only three elements were used together in this demonstration a much larger range could be used in principle with the possibility of better discrimination using methods such as the discriminant functions or neural nets.Such discrimination in real time would be within the capabilities of modern personal computers. Care would be needed to avoid confusion owing to two or more particles entering the plasma almost simul- taneously and producing a discriminant score intermediate between those of the different types of particles. Hence it would be necessary to dilute the suspension to a degree that reduces random overlap of particles to a low level. Journal of Analytical Atomic Spectrometry January 1996 Vol. 11 55Si Fig. 8 Mapping of the apparent normalized compositions of particles of olivine obtained by nebulization of a suspension of particles of a homogeneous material Fe Si Fig. 9 Isometric projection of Fig.8 with vertical lines of height proportional to the total signal intensity Multi-element Signals From Crushed Olivine While it is evident that particles of the silica gave rise to signals of poor repeatability it was not clear that normalized signals from several elements would be equally variable. Normalization would be akin to internal standardization and could therefore compensate to a degree for variations in particle mass or excitation temperature. The normalized calibrated responses for crushed olivine were plotted on a triangular diagram (Fig. 8). Relative strengths of the signals are fairly variable showing an arti- factual variation in apparent composition between the individ- ual particles. The source of the variation is not immediately apparent from the plot.However a plot that shows the overall signal strengths in addition to the normalized signals throws light on the question. Fig. 9 is an isometric projection of Fig. 8 with vertical lines superimposed that show the total signal strength. The base of each line stands on the triangular diagram at a point that maps the apparent composition of the particle while the height of the line is proportional to the overall signal strength. It can be seen that the longer lines are distinctly shifted away from the Mg apex of the triangle in comparison with the shorter lines. A plausible explanation of this diagram is as follows. The longer lines (greater signals) represent more massive particles. Therefore the more massive particles produce a vapour which is comparatively depleted in magnesium.As MgO is consider- ably more refractory than either FeO or Si02 the result is consistent with incomplete vaporization of the more massive particles. Such particles would produce a greater total mass of vapour which would be depleted (relative to the solid) in MgO. Therefore the large variation in Fig. 8 is due to selective vaporization of the larger particles. CONCLUSIONS Studies with signals from the ICP produced by the passage of single particles have so far demonstrated the following points. Peak widths are fairly constant for particles of different mass and composition if the injector gas flow rate is constant. This suggests that peak heights or peak areas would be equally useful measures of response. Under standard operating con- ditions for ICP-AES the typical peak width at half-height is about 0.5 ms which is consistent with rough calculations of the time required for a particle to pass through the cone of acceptance of the fibre transfer optics.Average peak heights for particles of silica in the size range 3-7 pm are not proportional to the mass of the particle. This suggests that larger particles are incompletely atomized or the atoms produced therefrom are less effectively excited. Signals from iron magnesium and silicon obtained during the nebulization of a suspension of olivine show relative attenuation of the magnesium signal from larger particles. This is strongly suggestive of partial and selective volatilization of the larger particles. Signals from particles exclusively containing a marker element that are present in a complex assemblage of particles can be readily identified and counted at a high rate i.e. up to a few hundred per second. Such element-specific particle counting could be developed into a valuable facility for search- ing for rare particles of foreign matter in relatively pure feedstocks or other particulate material. Many of these factors are suggestive of applications in industry or studies in the fundamentals of particle-plasma interactions. Equipment used in this study was provided by the Science and Engineering Research Council. REFERENCES Thompson M. Flint C. D. Chenery S. and Knight K. J. Anal. At. Spectrom. 1992 7 1099. Chenery S. Hunt A. and Thompson M. J . Anal. At. Spectrom. 1992 7 647. Olesik J. W. and Hobbs S . E. Anal. Chem. 1994 66 3371. Olesik J. W. and Fister .I. C. Spectrochim. Acta Part B 1991 46 851. Paper 5/05 145 B Received August 2 1995 Accepted September 18 1995 56 Journal of Analytical Atomic Spectrometry January 1996 Vol. 11