8 SILVER MEDAL ADDRESS Pvoc. Analyt. Div. Chem. SOC. Simultaneous Multi-element Trace Analysis The following is the Silver Medal Address delivered by Dr. G. F. Kirkbright, the first Society for Analytical Chemistry Silver Medallist, at a meeting of the SAC/AD held on November 6th, 1974, and reported in the November issue of Proceedings (p. 284). At the beginning of the meeting the President, Dr.G. W. C. Milner, presented the Silver Medal and a cheque for LlOO to Dr. Kirkbright. It is expected that the other two papers presented at the meeting will appear in a later issue of Proceedings. Dr. G . F . Kirkbright (R) receiving the Silver Medal and cheque from The President, Dr. G . W . C . Milner. Optical Density - Its Rewards and Penalties G. F. Kirkbright Department of Chemistry, Imperial College, London, SW7 2A Y When thinking of a suitable theme for this paper I wished to avoid a simple catalogue for- mat that comprehensively described the results of the activities of my research group during the past few years.The subject of simultaneous multi-element analysis in atomic spectros- copy, and multi-component analysis in molecular absorption and emission spectrometry, has become of increasing importance during the past decade.A number of effects associated with the techniques that might be used to accomplish these analyses, but which conspire against their use in the simultaneous multi-channel mode, have been of Interest to us during this period. These effects are most frequently related to the fundamental processes of absorption of radiation in real sample cells.For the theme of this paper, therefore, I have chosen “optical density” (rather than absorbance) ; I hope to illustrate, from our own research work and that of other workers, the rewards and penalties of its utilisation, particularly in the context of simultaneous multi-species analysis. Introduction Techniques that depend on the absorption or emission of electromagnetic radiation by atoms or molecules are well known to provide sensitive, and frequently highly selective, analytical methods for their determination. The rewards that occur from the utilisation of these phenomena have become widely recognised in the examination of matrices where the analyte species is a major component and, more particularly, when the analyte species is present in trace concentration.The advantages for these purposes of techniques such as ultraviolet - visible and infrared absorption spectrometry, solution spectrofluorimetry and phosphorimetry, X-ray fluorescence, arc and spark emission spectrometry and atomic- absorption and atomic-fluorescence spectrometry have been repeatedly demonstrated. InJanuary, 1975 SILVER MEDAL ADDRESS 9 each of these techniques, the species to be determined is contained within a “cell” whose geometry is known either very accurately (in techniques such as ultraviolet absorption spectrometry in solution) or is reasonably well defined (as in a flame or arc).The analyte absorption or emission signal observed from the cell is related to the optical thickness and particle density.In the general case, we may define “optical density’’ as the product of the absorption coefficient for the analyte species at the wavelength of measurement, Kv, and the effective path length of the cell, L. The magnitude of the absorption coefficient, Kv, will depend on the particular transition and the nature of the energy states involved in the absorption process and will also be proportional to the particle density (Le., concentration) of the absorbing species.This well known relationship, expressed for all techniques of absorption spectrometry via the Beer - Lambert law, permits quantitative analysis and relates the observed signal strength to the particular characteristics of the analyte species. In order to achieve high sensitivity in absorption spectrometry, appreciable optical density (absorbance) should be attainable with low concentrations of analyte in the sample “cell.” This results from the difficulty associated with the precise measurement of extremely small optical densities, i.e., where the intensities of the incident and transmitted radiation from the source are similar. In most absorption techniques, difficulties in making precise optical density measurements begin when less than about 2-3 per cent.of the incident radiation is absorbed by the analyte at the wavelength of measurement. Consequently, as mentioned above, in the development of improved methods utilising absorption spectrometry, an attempt is usually made to produce the highest optical density possible for low analyte concentrations.In fluorescence techniques, where the emission of radiation must be pre- ceded by its absorption, similar considerations apply. While this approach is beneficial when the determination of only one species at a time is required for a particular sample, it is difficult to apply when simultaneous multi-element analysis is required. In this context, it may be preferable deliberately to create low optical density conditions, and to improve the techniques by which the low particle densities associated with these low optical densities are measured, rather than to strive for greater optical density from the components of a par- ticular sample.This approach may be generally applicable in spectrometry but at present it seems that the best methods available for achieving precise monitoring in optically thin cells involve the measurement of emission or fluorescence rather than absorption. I would like to elaborate upon and illustrate these considerations by reference to scme recent work in my own laboratory and elsewhere where techniques utilising this approach have proved successful in simultaneous multi-component analysis for both molecular and atomic species.Molecular Fluorescence Spectrometry It is well known that the detection and determination of both inorganic and organic species by spectrofluorimetry and spectrophosphorimetry may provide considerably higher sensitivity than molecular absorption spectrometry. This usually results from the relative ease with which the luminescence emission from the species determined is monitored at concentrations in the sample cell that produce only low optical density.The conventional “right-angle” illumination and viewing of the sample cell most fre- quently employed in solution luminescence studies is shown in Fig. 1; the radiant power of the luminescence can be obtained from the expression Pp = light absorbed x quantum efficiency = (Po - Pol0-ecZ)+ = Po(l - 10-ec’)+ where = quantum efficiency of the luminescence process for the analyte molecule; = the effective path length of the incident beam through the solution that is accepted = molar absorptivity of the analyte species; = concentration of the analyte species.Po = incident radiant power from the source; I e c at the detector; Equation (1) expresses the total luminescence radiant power.Only a fraction of this It is clear, however, that the luminescence radiant power is collected by the detector optics.10 SILVER MEDAL ADDRESS Proc. Analyt. Div. Chem. SOC. power, PF, can be directly proportional to the analyte concentration, c, for dilute solutions only when the higher terms involved in the expansion of the light absorption factor can be neglected. Under these conditions, equation (1) reduces to Pp = P,,ec+ It is therefore necessary to retain low optical density conditions in the sample cell in order to obtain linear analytical working curves over a wide concentration range. This may be achieved by working with solutions of low concentration and utilising powerful sources (PpccpO) and sensitive detection techniques such as photon counting to detect the weak luminescence signals which result, or by minimising the path length, 1.A method by which this can be accomplished is the adoption of the “front-surface” rather than the “right-angle” method of illumination and viewing (see Fig. 1). -I- PF Fig. 1. Right-angle (a) and frontal (b) viewing in fluorescence measurements.An additional spectrometry is factor that affects the analytical working curves for molecular luminescence the type of “inner-filter” iffect in which radiation emitted by the analyte species at its wavelength of maximum fluorescence intensity is absorbed before it can leave the cell. The absorption may result from “self-absorption” due to pronounced overlap of the absorption spectrum of the analyte with its luminescence spectrum, or may be caused by the presence of other species in the sample that absorb at this wavelength.Again, it is preferable to use low optical density conditions in these circumstances ; additionally, these “inner-filter” effects can be minimised by attempting to decrease the absorption and emission band-widths in order to decrease their overlap.The above considerations can be illustrated by recent studies in our laboratory concerned with the determination of polynuclear aromatic hydrocarbons (PAH) by luminescence techniques1 Although the detection and determination of these compounds can frequently be made with high sensitivity by conventional fluorimetry at room temperature in organic solvents , the broad absorption and emission spectra obtained under these conditions make their determination unselective.Thus, when samples contain complex mixtures of PAH compounds, their spectra overlap; it is necessary to resort to gas or liquid chromatography or other separation procedures before their identification and determination by solution spectrofluorimetry. If the Shpol’skii effect2 is utilised, however, this disadvantage is mini- mised and the presence or absence of individual PAH compounds in relatively complex mixtures can be confirmed.Using this effect, PAH compounds are included into the cry- stalline matrix formed at 77 K or below by selected n-paraffin solvents; extremely well resolved fine structure is then observed in their luminescence emission spectra. The obser- vation under these conditions of line-like structure, in which these individual lines may be less than 0.1 nm in half-width, can be explained by the postulate that the solute analyte molecules become embedded in the crystalline solvent lattice formed on cooling.In contrast to the case that applies in solvents that form transparent glasses at low temperature, where the glass does not show short range order and the electronic transitions are very sensitive to variation in the molecular field, in the crystalline solid solutions produced for PAH com- pounds in n-paraffin solvents the solute molecules experience a well defined molecular field that gives rise to sharp-line (quasi-linear) electronic spectra.Our early studies with this technique indicate that utilisation of these spectra should provide for the sensitive and extremely selective detection and determination of these compounds.January, 1975 SILVER MEDAL ADDRESS 11 Wavelengthhm Fig.2. Fluorescence emission spectrum of a M solution of 3,4- benzpyrene in n-octane at 77 K. To place our work on low-temperature luminescence spectrometry using the Shpol’skii effect into the context of the theme of this paper, a number of advantages in its application can be quoted.The spectra in Figs. 2 and 3 show the improved resolution obtained for 3,4- benzpyrene when the quasi-linear luminescence emission spectrum is produced in n-octane rather than its luminescence emission spectrum at 77 K in EPA (diethyl ether - isopentane - ethanol, 5+5+2), the solvent used frequently for low-temperature fluorimetry.The use of “front-surface” illumination of the crystalline matrix formed from n-octane results in a linear calibration graph over a concentration range of about 103-fold. The problems associated with overlapping excitation and emission spectra are minimised by the well resolved nature of the spectra. Indeed, it appears that the excitation spectra are also quasi-linear in many instances and it is probable that in the future selective excitation of quasi-linear luminescence using a narrow band-width tunable dye laser will result in virtually specific analyses even in complex mixtures.With this type of narrow-line excitation of a narrow “line-like” luminescence, a technique for analysis of molecular species analogous in its operation to the techniques of atomic-absorption and -fluorescence may be visualised.Wavelengthhm Fig. 3. Fluorescence emission spectrum of a M solution of 3,4-benzpyrene in EPA at 77 K. Atomic-absorption Spectrometry Fig. 4 shows the manner in which the absorption or emission line-width for an atomic For an atom cell of given geometry, it can be species varies with increasing optical density.12 SILVER MEDAL ADDRESS PYOC.Analyt. Div. Chem. SOC. expected that at low optical densities the absorption half-width is independent of atomic concentration. At high values, when the absorption (or emission) factor begins to approach unity ( i t ? . , the equivalent at the central wavelength for a black body at the same temperature), little further increase is possible at the central wavelength. The absorption (or emission) factor at the wings of the line can continue to increase, however, and line broadening occurs.When a line source (i.e., a hollow-cathode lamp) of spectral width smaller than the absorption line width of the analyte is employed for atomic-absorption spectrometry in a particular atom cell, this behaviour imposes an upper concentration limit for the analyte element beyond which there is progressive deviation from linearity; eventually, at very high optical density, there is no further dependence of absorbance upon Concentration.Similarly, when a con- tinuum source (e.g., a xenon arc lamp), the spectral band width selected from which is wider than the absorption line width, is employed a linear growth curve with a slope of unity is obtained until the half-width of the absorption line begins to increase; at higher optical density, a growth curve with a slope of Q is obtained (see Fig.5). For both line and continuum sources, this behaviour is predicted from evaluation of the expression for the total absorption factor, AT, for high and low optical density conditions (e.g., see ref.4). As mentioned earlier, the lower limit of optical density at which precise atomic-absorption measurements can be made depends on the instrumentation employed and the signal to noise ratios obtained at these low values. In general, however, difficulties with these measurements would begin when less than 1-2 per cent. absorption is to be measured. These two factors at high and low optical density serve to restrict the linear working range over which precise and accurate measurements can be made in atomic-absorption spectrometry.E -12 -8 -4 0 4 8 12 w 2(X - Xo)/A 1, Fig. 4. Variation of atomic line-width at different values of optical density (K,Z) for line characterised by dispersion profile. Adapted from ref.3. The above considerations of restrictions of the linear working range in atomic-absorption spectrometry become of primary importance when an attempt is made to undertake simul- taneous multi-element analysis rather than single-channel work. Problems arise when the simultaneous determination of a number of elements is required using a single set of flame conditions and a single sample solution.Fig. 6 shows the atomic-absorption calibration graphs that might be obtained for two elements, A and B, when the determination of element A can be made with greater sensitivity than that of B. If the linear working range for each falls between, for example, 3 and 70 per cent. absorption at the wavelengths chosen, then it is apparent that if both elements are to be determined simultaneously in the same sample solu- tion, the ratio of the concentration of B in the solution to that of A must lie within a relatively narrow fixed range.It is therefore not possible to determine simultaneously both elements by atomic-absorption spectrometry when they may be present in samples over a wide range of concentration ratios. This situation naturally becomes of increasing complexity as the number of elements whose simultaneous determination is required increases.Attempts to resolve this difficulty by introduction of the sample solution into the cell at several dilutions result only in slow sequential analysis. It is also not possible to obtain a satisfactory solution to the prob- lem by resorting to the use of atomic lines of different oscillator strength; this procedure simply exchanges the necessity for the concentration of a particular element in the sample toJanuary, 1975 SILVER MEDAL ADDRESS 13 lie within one concentration range if one of its resonance lines is used, for the necessity for its concentration to lie in a different range at an alternative wavelength.In order to ensure that the concentration of an element lies in the linear analytical range at the wavelength chosen, at least some knowledge of its concentration level in the sample is pre-supposed.The problem of simultaneous multi-element analysis by atomic-absorption spectrometry using either flame or non-flame atom cells is further complicated by the fact that different elements may require different atomisation conditions in the cell.Thus, the flame stoicheiometry required for the efficient atomisation of A may be different from that required for element B in the same sample, particularly when the need to minimise interference effects is taken into account. Concentration of A or B Fig. 6. Calibration graphs in atomic- 0.1 1 -0 10 100 and B whose determination is of widely Nfthh cm- different sensitivity (for 1 per cent.absorption) using a single set of conditions in a given atom cell. 0.001. I I I I absorption spectrometry for elements A Fig. 5. Growth curves calculated for atomic absorption with continuum and narrow line sources. Adapted from ref. 4, where the calculation refers to the magnesium 285.2-nm line and a cell tempera- ture of 2400 K. Low optical density High optical density Line source .. Continuum source The situation may be even more difficult if non-flame atomisation is to be employed in the simultaneous multi-element mode. In non-flame cells, such as the graphite filament or furn- ace atomiser, high sensitivity is usually obtained by the deliberate formation of a short- duration optically dense atom population; attempts to control atomisation conditions so as to retain a worthwhile range of concentration ratios in the sample over which simultaneous determinations of a number of elements may be made, while minimising interference effects, will give rise to difliculties.Thus, while a number of workers have reported instrumental systems in which ingenious techniques have been described to facilitate simultaneous multi- element atomic-absorption spectrometry by the use of a number of spectral line sources and detectors with a single atom cell (e.g., see refs.5-49, the fundamental limitation to the general utility of all of these systems lies in the atom cell itself as described above. This is not to say that these systems will not fulfil a need for certain well specified tasks, where a small number of elements must always be determined in a particular sample type whose composition does not vary very.greatly from sample to sample, and where the best compromise conditions can be established carefully for the simultaneous determination of these elements at a single dilution. In view of the difficulties involved in the general application of atomic-absorption spectrometry to simultaneous multi-element work, where each user requires the determination of a different combination of elements at widely different concentrations in samples of different types, it is perhaps not surprising that no commercial multi-element instrumentation based on atomic- absorption spectrometry has appeared.A technique that we have employed, which may assist the application of flame sources in multi-element analysis by atomic-absorption spectrometry, involves the use of a special type14 SILVER MEDAL ADDRESS Proc.Artalyt. Div. Chew. SOC. of capillary burner system shown in Fig. 7. In this burner, the pre-mixed fuel and oxidant carry the nebulised sample to form a stable flame on the longer set of capillaries.The second, shorter capillaries interspaced between those carrying the pre-mixed gases and sample aerosol can be supplied with an ineft gas such as nitrogen or argon without affecting the flame stability. The provision of these auxiliary capillaries enables the flame gases above the primary reaction zones to be diluted with inert gas. Thus, in a situation where two elements, A and B, are to be determined in a single sample solution at one dilution and, for example as in Fig.6, element A gives greater sensitivity than B and is present at a concentration that produces very high absorbance at the resonance line chosen, the auxiliary inert gas supply can be employed to lower the optical density obtained for B to a value that ensures that it is measurable within its linear working range.Thus, as shown in Fig. 8, the absorbance for A is measured within its linear working range without the use of the auxiliary gas and the “off-scale” value for B is ignored. The auxiliary gas is then switched on automatically at a pre-set flow-rate and the absorbance for B is measured within its working range while the very low value for the absorbance of A is ignored.By this technique, which can readily be operated automatically, the particle density in the flame cell is lowered by dilution of the flame rather than by dilution of the sample solution. A subsidiary advantage of the burner system is that when air is used in the auxiliary capillary system, the flame stoicheiometry above the primary reaction zone can be varied so as to provide alternatively fuel-rich and fuel-lean conditions for the atomisa- tion of different elements.Diluent - t Flame gases + sample Fig. 7. Capillary burner with provision for dilution of flame with inert gas. 0 Fig. 8. Time + Use of capillary burner (see Fig. 7) with different diluent gas flows for the determination of two elements A and B in a single sample solution. The above approach, however, produces only a partial solution to the problems described earlier in the use of atomic-absorption spectrometry for multi-element analysis and produces sequential rather than simultaneous analysis.It may find application, however, for a number of specific problems. As mentioned in the introduction, a more general solution to these problems may be to “dilute out” the effects and deliberately create very low optical density conditions simultane- ously for all elements to be determined in a single sample.Thus, low particle densities and/or path length should be used. Naturally, this will produce conditions where only very small absorbance values are to be measured. With the type of absorption measurement system available at present, it is *cult to make these measurements with high precision.Probably the best available alternative is to monitor the absorption indirectly, via that fraction re- emitted as atomic fluorescence, or to abandon absorption and return to the measurement of the thermal emission from atoms at low particle density in a hot source. Thus, it seems that the detection of the low light signals associated with atomic emission from optically thin sources may be more suitable than atomic-absorption spectrometry for the precise measurement of low particle densities. I shall now consider the use of these techniques for simultaneous multi- element analysis, via the use of atomic-emission spectrometry with an inductively coupledJanuary, 1975 SILVER MEDAL ADDRESS 15 high-frequency plasma source and atomic-fluorescence spectrometry with pulsed spectral sources.Radiofrequency Plasma Emission Spectrometry Fig. 9 shows a schematic representation of a plasma emission spectrometer system that we have described earlier.l03l1 The 36-MHz plasma torch system is similar to that described by other w0rkers12-l~; aqueous sample solutions can be introduced into the plasma without the requirement of desolvation of the aerosol.Our early studies confirmed reports from other ,;Q Grating mo nochromator PMT EHT =I4 r Lock- i n - amplifier Recorder laboratories that high detection sensitivity in emission could be obtained with this source for a wide range of elements and that the linear working concentration range attainable was fre- quently considerably greater than that attainable in atomic-absorption spectrometry or flame- emission spectrometry.In a recent study, we predicted the comparative performances of the nitrous oxide - acetylene flame and high-frequency plasma atom cell as emission sources for simultaneous multi-element analysis by the use of simple mathematical models of both cellsu In these models of flame and plasma, reasonable assumptions are made concerning the tem- perature, velocity and flow conditions of the hot gases. A summary of the characteristics of the two atom cells is shown in Table I. - TABLE I TYPICAL OPERATING CONDITIONS FOR THE RADIOFREQUENCY INDUCTION-COUPLED PLASMA AND PRE-MIXED NITROGEN SHIELDED NITROUS OXIDE - ACETYLENE FLAME Parameter Fuel gas flow-rate Injector gas flow-rate Oxidant gas flow-rate Coolant gas flow-rate Shield gas flow-rate Sample uptake rate, QLI Nebuliser efficiency, i+h Path length of cell, L Width of burner, b Height of reaction zone, ht Height of viewing zone Surface area of core Volume of core Mean gas density, p Mean specific heat of gas, Cp Flame expansion factor Ambient temperature, To Flame temperature, T Maximum temperature, Tmax.Plasma Flame 3.3 dm3 min-1 3.0 dm3 min-1 6.6 dm3 min-1 15 dm3 min-I 1.6 cm3 min-l 0.01 20 mm 25 mm 10 mm 22 cm2 8 cma 1.78 g dm” 0.524 J g-l K-l 1.0 300 K 8260 K 9000 K 15 dm3 min-l 4.2 cm3 min-l 0.1 50 mm 0.4 mm 0.1 mm 10 mm 50 mme 2 mm3 0-96 g dm-3 1.35 J g-l K-l 1.667 300 K 2800 K 3000 K16 SILVER MEDAL ADDRESS Proc. Analyt. Div. Chem. SOC. It is instructive to compare the particle density obtained for analyte elements introduced into the atom cells and the effect of the particle density on the degree of self-absorption and the working curves produced.The particle density, NA, of an analyte element in either cell can be calculated from where Qs is the volume of flame or plasma gas entering the atom cell per second at an ambient temperature To, C$ is the nebuliser efficiency, A is Avogadro's number, Qs is the sample uptake rate, MA is the relative atomic mass of the analyte element, T is the mean flame temperature, y is the molar flame expansion factor and CA is the concentration of analyte element A in aqueous solution.For the purpose of comparison of the two atom cells, the line emission observed for the calcium atom and ion at 422.67 and 393.37 nm, respectively, zinc at 213.86 nm and copper at 327.40 nm will be considered.The particle densities of elements calculated from equation (3) give the total particle density of all species of that element, so that in order to determine the particle densities of the indi- vidual species, the free atom fraction and the degree of ionisation must be measured or calcu- lated.Few data are available concerning the free atom fraction of these species in the radio- frequency plasma, although consideration of the dissociation energies of their oxide and hydroxide species suggests that the assumption of a value of unity for the free atom fraction is realistic. With values of the dissociation energy of calcium hydroxide, copper(I1) oxide and zinc oxide of 104, 95 and 65 kcal mol-l, respectively, we calculate a value of unity for the free atom fraction of each element assuming a partial pressure of low6 atm for the oxide species and a temperature of 8000 K.We cannot assume a value of unity for the free atom fractions of these elements in the nitrous oxide - acetylene flame, however, as lower values have been found by several workers.For the purposes of our calculations, we have assumed values of 0.33, 0.33 and 0.50 for the free atom fractions of calcium, copper and zinc, respectively.1Q-" In the radiofrequency plasma, the degrees of ionisation of calcium and zinc were measured to be 0-75 and 0-1, respectively, from the observed intensities of the atom and ion lines, assuming a temperatnre of 8250 K.The calculated degree of ionisation of copper using the Saha equa- tion under these conditions is 0-5. In the plasma, the high free electron density (about 10aom--3) is so much greater than the analyte particle density that the degree of ionisation remains virtually constant at all concentrations of the analyte considered. In the nitrous oxide - acetylene flame, the degree of ionisation of calcium has been calculated to be 0.43 using the Saha equation and assuming a partial pressure of 10" atm for the analyte atoms.The measured degree of ionisation of calcium in this flame has been reported to be 0.43 (ref. 22) and 0.38 (ref. 23) and for this calculation we have assumed a value of 0.4. The calculated degree of ionisation of zinc in this flame is less than 0.01 and can be neglected, while that of copper is only about 0.03, and has also been neglected in this calculation.The degree of ionisation for each element in the flame refers to that for a 1 p.p.m. solution; no allowance has been made for any variation in the degree of ionisation with analyte concentration. TABLE I1 PARTICLE DENSITIES OF ANALYTE ATOMS IN THE PLASMA AND FLAME ATOM CELLS AND THE DOPPLER HALF-WIDTH AND OSCILLATOR STRENGTH OF THE LINES OF THE ATOM CONSIDERED Particle den~itylm-~ Doppler hdf-width/nm t 3 r- Species A/cm M A f Plasma Flame Plasma Flame Ca I 422.67 40-1 1-75 6.8 x 1015 8-2 x 1015 0.0043 0.0025 Ca I1 393.97 40.1 0.69 2.1 x 1015 5.4 x 1015 0-0040 0-0024 Cn I 327.40 63.5 0.16 8.6 x 1 0 x 4 8.6 x 1015 0.0027 0.00 16 Zn I 213-86 65-4 1.2 1.6 x 1015 1.3 x 10ls 0-0017 0~0010 A Table I1 shows the calculated particle densities of these species when a solution containing 1 p.p.m.of analyte element is nebulised into both the plasma and flame operated under the conditions shown in Table I, the wavelength of the line of the species considered, the oscillator strength of these lines and the Doppler half-widths of these lines calculated assuming a plasma temperature of 8250 K and a flame temperature of 2800 K.January, 1975 SILVER MEDAL ADDRESS 17 The absorption coefficient at the line centre, Kv, of the analyte element at the line considered can be calculated from the following equation: In equation (a), e is the electronic charge, m is the mass of the electron, c is the velocity of light, A&, is the Doppler half-width, f i s the oscillator strength, E,, is the permittivity of free space and No is the particle density in the ground state.The values of Kv and KvL calculated for the plasma and flame at the particle densities obtained from equation (4) and assuming that the radiation is viewed from the centre of the atom cell are shown in Table 111.TABLE I11 ABSORPTION COEFFICIENTS OF ANALYTE ATOMS IN THE Plasma PLASMA AND FLAME ATOM CELLS Flame r 1 Species h/nm Kv/rn-l KvL Ca I 422.67 0.408 0.00408 cu I 327.40 0,046 0.00046 Zn I 213.86 0.406 0.00406 Ca I1 393.37 0.460 0.00450 f 1 Kv/m-l K,L 8-41 0.210 2-06 0.051 0-79 0-020 5.80 0-145 From the values of KVL obtained for these lines, it is possible to construct theoretical calibra- tion graphs using the relationship where I,, is defined by the Einstein - Boltzmann equation: where u is the partition function of the atom or ion, h is Planck's constant, gk is the statistical weighting factor of the upper state involved in the transition, Ek is the energy of the upper state and k is the Boltzmann constant.Curves obtained for the four lines considered are shown in Fig.10. The intensity represented for the plasma is the intensity emitted per unit volume relative to the intensity emitted per unit volume in a flame taken as unity at the limit where the absorption coefficient, Kv, becomes equal to that of a black body at the wavelength concerned. It is apparent from these curves that the radiofrequency plasma exhibits an extended linear working range at high solution concentration compared with that attainable with the flame.In the plasma, the linearity of the Ca I line at 422.67 nm, the Ca I1 line at 393.37 nm and the Zn I line at 213.86 nm are predicted to be almost identical, as the value of KvL for each line is similar. Also, as the absolute emitted intensity at any line is very much greater in the plasma than in a flame, the potential sensitivity of the technique is much higher and the linear range should also be extended to lower optical densities (Le., concentra- tions).A second advantage of the plasma system is that as the absolute emitted intensity is very much higher than that obtained with a flame, it is possible to reduce the sample uptake rate of the nebuliser system and maintain adequate signal intensity, which reduces the value of K,.This technique, therefore, offers a very simple means of extending the linear range obtained at high concentrations in the plasma with minimal sacrifice in sensitivity. One of the advantages claimed for the radiofrequency plasma system is that the residence time of an analyte particle in the discharge and tail-flame is long compared with that in a flame or arc excitation cell.For the plasma system employed in the present work, it can be assumed that the linear velocities of analyte particles and carrier gas atoms are the same; the residence time of analyte particles will therefore be the same as that for the gas atoms or molecules. The linear velocity of gas through the entrance port into the plasma is given by the equation VO = 4QG/vLa (7)18 SILVER MEDAL ADDRESS Proc.Artalyt. Div. Chem. SOC. where L is the path length of the cell (Le., the diameter of the tail-flame), Qa is the flow-rate of the gas at ambient temperature and vo is the linear velocity of the gas up the plasma tube. The corresponding equation for the flame is vo = Qa/bL (8) where L is the burner slot length and b the burner slot width.Solving these equations for the operating conditions of Table I gives a linear velocity of gas into the plasma of 0.16 m s-1 and into the flame of 8.25 m s-l. After passing into the plasma core or primary reaction zone of the flame, a volume expansion takes place. Thus the linear velocity of gases after leaving the reaction zone is given by where y is the molar flame expansion factor.respectively. equation where ht is the thickness of the reaction zone and vBV is the average linear velocity through the reaction zone. For the plasma, it is assumed that the average velocity through the plasma core is 4.4 m s-l and for the flame it is assumed that the average velocity through the primary reaction zone is 70 m s-l.The calculated residence times are 5.7 ms for the plasma and 1.5 ps for the flame. VT = voy T/To (9) Solving this equation for the plasma and flame gives linear velocities of 4.4 and 130 m s-1, The residence time of analyte particles in the reaction zone can be calculated from the tr = h t / L ( 10) 0.1 1 10 100 1000 Concentration, p.p.rn. Fig. 10. Calculated growth curves for plasma and long flame cells for elements considered: (A) calcium a t 422-67nm; (B) calcium at 393-37nm; (C) zinc at 213-86 nm; and (D) copper at 327-40 nm.If it is assumed that the emission intensity is viewed over a length of 10 mm above the top of the plasma core or primary reaction zone, the residence time of analyte in the viewing zone can be calculated from the equation The residence time for an analyte particle in the plasma tail-flame is therefore 3.3 ms com- pared with 77 ps in the flame.t, = l/lOoV, (11)January, 1975 SILVER MEDAL ADDRESS 19 The plasma system can readily be seen to retain the analyte atom in the volume viewed for a much longer period than the flame. It can therefore undergo many more collisions to effect excitation during its residence compared with the number experienced in the flame. As the lifetime of the excited atom is typically a few nanoseconds, it is apparent that the atom can be excited and may emit about thirty times in the plasma tail-flame compared with once in the flame.The additional advantage of the longer residence time is that for samples whose rate of vaporisation is low, complete vaporisation may be possible during their residence in the plasma, whereas complete vaporisation cannot be attained in the nitrous oxide - acetylene flame.Fig. 11 shows the effect of variation in the sample uptake rate in the radiofrequency plasma upon the calibration graphs obtained for calcium at 422.67 nm. At low sample uptake rates, the particle density, and hence optical density, is low, and the linear range is therefore extended to higher solution concentration compared with the range obtained with high sample uptake rates. The observed emitted intensity at each line is, of course, reduced at low uptake rates as the particle density is lower.The detection limit does not deteriorate at low sample uptake rates to the same extent as the signal attenuation predicted from the reduction of particle density.This is probably due to the fact that less solvent is transferred to the dis- charge at low sample uptake rates so that there is less impedence mismatch between the resonance circuit of the oscillator valve and the plasma discharge. Also, it is to be expected that the plasma temperature will increase at low sample uptake rates as there is less solvent in the discharge to absorb the applied power.0.001 I I I 1 I I 2 20 200 2000 20 ooo Concentration, p.p.m. Fig. 11. Effects of sample uptake rate on calibration graphs in the R.F. plasma for calcium at 393-37 nm. Uptake rates: A, 4-5; B, 3.2; C, 1-5; and D, 0.8 ml min-'. In the nitrous oxide - acetylene flame, a similar effect of longer linear range at low sample uptake rates i s observed for the calibration graphs, but even at low uptake rates the linear range is still small.The advantage of the radiofrequency plasma is that the observed linear range of the calibra- tion graphs can be made to extend for about five to six orders of magnitude, while for the flame under the conditions employed the maximum observed was three orders of magnitude.Analysis of Aluminium Alloys In order to compare the application of flame emission and plasma emission spectrometry, the determination of six minor elements in aluminium alloys was examined. The plasma and flame were operated under the conditions shown in Table I. Solutions of BCS aluminium samples and standard solutions were introduced into the plasma at a sample uptake rate of 1.5 cms min-l and were determined sequentially from the same solution without dilution for different elements.Titanium was determined at 365.35 nm, iron at 371.97 nm, manganese at 403.07 nm, zinc at 213436 nm, copper at 327.40 nm and magnesium at 285.21 nm. A slit width of 25 pm [a spectral band pass of 0-08 nm] was employed throughout. The acetylene gas flow-rate to the nitrous oxide - acetylene flame was adjusted so as to maximise the signal obtained for titanium.The solutions were introduced directly into the flame and analysed for titanium, manganese and iron at the same lines as those employed for the plasma.20 SILVER MEDAL ADDRESS Proc. Analyt. Div. Chem. SOC. Because of the non-linearity of the calibration graphs for magnesium and copper at these lines, appropriate dilutions were made before these metals were determined.Insufficient sensitivity was obtained for zinc in the flame to allow its detection in three of the aluminium samples and zinc was detectable but could not be determined with adequate precision in the fourth sample. The results of these analyses are shown in Table IV.Each value shown is the mean of six results obtained by each technique; the values are compared with the BCS certificate values in each instance. There appears to be no significant discrepancy between the results except where dilutions of the stock solution were necessary (for magnesium and copper) due to re- stricted linear range when the nitrous oxide - acetylene flame was employed. The results obtained for titanium by flame emission spectrometry are somewhat high; this may be due to the fact that no ionisation suppressant was added to the standard titanium solutions or to the aluminium samples.TABLE IV ANALYSIS OF ALUMINIUM ALLOYS BY PLASMA AND FLAME EMISSION All analyses were performed on a solution containing 1 g of alloy in 100 cm3 of solution by plasma emission These solutions were also used for flame emission spectrometry except for: (a) 0-1 g of alloy spectrometry.per 100 cm3 of solution; (b) 0.01 g of alloy per 100 cm3 of solution. Alloy Element BCS 216/2 BCS 26311 BCS 300 BCS 181/1 cu Mg Fe Mn Ti Zn cu Fe M g Mn Ti Zn c u Fe :: 2 Ti Zn cu Fe Ti Zn Certificate value, per cent. 3.99 f 0.02 0-36 f 0.01 1.42 f 0.03 0.10 & 0.01 0.14 f 0.01 0.02 f 0.01 4.66 f 0.01 0-28 f 0.01 0.75 f 0-01 0.71 f 0.01 0.037 f 0.001 0.20 f 0.01 0.09 f 0.01 0.35 f 0.01 4-92 f 0.05 0.36 f 0.01 0.038 f 0.001 0.05 f 0.01 1-28 f 0.02 0.30 f 0.01 2-76 f 0.03 0.41 f 0.01 0.16 f 0.01 5.98 f 0.04 Plasma value, per cent.3.97 & 0.08 0.36 f 0.01 1.41 f 0.04 0.10 f 0.01 0.14 f 0.01 0.02 f 0.01 4.54 f 0.10 0-28 f 0.01 0.75 f 0.03 0.70 f 0.01 0.03'7 f 0.003 0.20 f 0.02 0-10 f 0.01 0-33 f 0-02 4-94 f 0.11 0-36 f 0.01 0-037 f 0.004 0.05 & 0.01 1-27 f 0.04 0.30 f 0.01 2-78 f 0.09 0.41 f.0-01 0.16 f 0.01 5-94 f 0.15 Flame value, per cent. 3.8 & 0-4(a) 0-34 f 0.04 1-4 f O-l(b) 0.09 f 0.01 0.18 f 0.04 -* 4-3 f 0.3(a) 0.27 f 0.03 0.90 f 0*04(b) 0.71 f 0-04 0.04 f 0-02 0.10 f 0.02 0.34 f 0-03 4.6 f 0-3(b) 0.35 -& 0.05 0.04 & 0-02 1.4 f O.l(a) 0.29 f 0.02 2.9 f O-l(b) 0.41 f 0.02 0-18 f 0-06 -t -* -* * Not detected.t Not determined. The results shown for the analysis of aluminium alloy samples by both plasma and flame emission spectrometry demonstrate the utility of the wide concentration ranges for each ele- ment over which linear calibration is obtained with the high-frequency plasma source.Thus, whereas all of the analytical results shown in Table IV for plasma emission spectrometry were obtained at a single dilution (1 g of alloy per 100 cm3) , in the corresponding analysis by flame emission spectrometry several dilutions were required for each sample in order to effect the determination of five elements. Atomic-fluorescence Spectrometry Atomic-fluorescence spectrometry has been shown by a number of workers to provide for high sensitivity in the determination of a range of elements using either flame or non-flame atomisers.As radiation from a spectral line or continuum source must be absorbed by the analyte atomic population before it can be re-emitted as atomic fluorescence, these effects which influence the total absorption factor, AT, and its variation with the analyte concentra-January, 1975 SILVER MEDAL ADDRESS 21 tion, as described earlier for atomic-absorption spectrometry, also affect the working curves obtained in atomic-fluorescence spectrometry (e.g., see ref.24). The working curves that result in atomic-fluorescence spectrometry with atomic line sources are illustrated in Fig.12. The ideal fluorescence intensity curves are similar to those obtained in atomic-absorption spectrometry owing to the direct proportionality between the integrated fluorescence intensity, IF, and the total absorption factor, AT, through the expression where I0 is the radiant flux that excites the fluorescence under consideration, W is the width of the exciting beam of radiation, i2 is the solid angle over which the excited fluorescence is detected and measured (47r is the total radius over which fluorescence is emitted from the cell) and 4 is the fluorescence yield (the fraction of the absorbed photons that is re-emitted as fluorescence).Log N Fig. 12. Hypothetical growth curve in atomic- fluorescence spectrometry with a narrow line source. IF is intensity of fluorescence signal and N the con- centration of ground-state analyte atoms in cell.Broken lines indicate effect of increasing extent of incomplete illumination and viewing of atoms in cell. In practical atom cells, the assumptions made in the derivation of the simple expressions for AT and Ip are usually not adhered to. Thus, with flames, it is not often that the entire fluorescence cell is within the solid angle excited by the source and viewed by the detector, and where a cell has an appreciable path length some fluorescence emission is invariably lost by re- absorption.These effects become most serious for resonance fluorescence effects and high optical densities (as illustrated in Fig. 12) and provide the high concentration limit to the linearity of the working curves.For non-resonance fluorescence processes, such as in direct- line fluorescence, re-absorption of fluorescence radiation may be less severe because the popula- tion of the intermediate level is small; under these conditions, a longer linear working range may be achieved. At low optical densities, however, the linear relationship between fluores- cence intensity and analyte concentration is easily preserved in most atom cells.As seen from equation (12), the intensity of fluorescence, IF, is directly proportional to the source intensity, Io. It is therefore possible to make measurements in atomic fluorescence when the total absorption factor, AT, (or optical density) is very low, by the use of an intense spectral source and sensitive detection systems.Such measurements are possible at optical densities that would be too low to monitor directly by atornic-absorption spectrometry. It is this factor which accounts for much of the gain in sensitivity attainable by atomic-fluorescence spectrometry compared with atomic-absorption spectrometry for a number of elements. This factor also results in extension of the available linear working range by provision o1 a method of measurement of lower optical densities even though the upper limit to the working range may be similar.The greater available linear working range attainable, together with the minor advantage that it is frequently easier to arrange a number of sources and a single detector around the atom cell employed in atomic-fluorescence spectrometry than in atomic-absorption spectrometry, results in the justification of the use of atomic-fluorescence22 SILVER MEDAL ADDRESS PYOC.Analyt. Div. Ckem. SOC. spectrometry rather than atomic-absorption spectrometry for simultaneous multi- element analysis. Fig. 13 shows the working curves for chromium, zinc and calcium in an air - acetylene flame using pulsed hollow-cathode lamp sources and an atomic-fluorescence spectrometer capable of rapid sequential operation25; the instrumental system employed was based on the multi-channel atomic-fluorescence spectrometer described by Mitchell and Johansson.= The concentration range over which linear working curves can be obtained is typically as great as lo4 or lo6.This may be exploited, in a manner similar to that described above for the high-frequency inductively coupled plasma source, in the simultaneous detennin- ation of a number of elements by atomic-fluorescence spectrometry at a single dilution in samples where widely different concentration range ratios of one element to another are expected in the sample.This has been confirmed for multi-channel atomic-fluorescence spectrometry in the simultaneous determination of six elements in sea water,m lubricating oils28 and aluminium alloys. g9 0901 001 0.1 1 10 100 1000 Concentration, p.p.m.Fig. 13. Atomic-fluorescence spectro- metric working curves for chromium, calcium and zinc obtained with an air- acetylene flame and pulsed hollow-cathode lamp sources. Adapted from ref. 25.Before leaving consideration of the technique of atomic-fluorescence spectrometry , the recent advances in this technique using pulsed laser sources of high intensity should be rnen- tioned. As described above, any method whereby source power delivery to the atom cell used for atomic-fluorescence spectrometry may be increased has a beneficial effect on sensitivity in atomic-fluorescence spectrometry.As pointed out by Fraser and Winefordner,m the use of a stable, repetitively pulsed source of excitation with a small duty cycle, i e . , a small ratio of on- to-off time, might permit an increase in the fluorescence signal to noise ratio owing primarily to decreased noise; during the short on-time, the signal may be of the same magnitude as the average signal obtained using a continuously operated source, whereas the noise would be small because of the small number of random photo-detector pulses from dark current, flame background emission, etc.Thus, the high peak power output from a repetitively pulsed, tunable laser should provide for high power delivery to an analytical atom cell and permit the attainment of high detection sensitivity.This type of system should in most respects be the ideal source for analytical atom$-fluorescence spectrometry. Fraser and W i n e f ~ r d n e r ~ , ~ ~ have reported the use of a fast repetition rate pulsed system utilising a nitrogen laser-pumped tunable dye laser for excitation of atomic fluorescence from a wide range of elements in air - hydrogen, air - acetylene and nitrous oxide - acetylene flames.The peak power output of the dye laser over a spectral band width of 0-1-1 nm is about 10 kW, the average power output of the dye laser is about 0.001 W, the pulse half-width is 2-8 ns and the repetition rate is 1-30Hz depending on the dye used. Ten dyestuffs were used, each covering a spectral range of about 30 nm in order to provide a system tunable over the range 360-650 nm.As the output of the tunable dye laser can be varied over a range of 10-30 nm, depending on the dye, it is possible to wavelength scan the output of the dye laser. This pro- vides a convenient means of correction for any background scatter signals and obtaining spectral information by the use of a fast response photomultiplier tube and a boxcar integrator capable of aperture gate widths of the order of 10 ns.With this laser technique, therefore, atomic fluorescence is effectively excited with a wide-line source with a power output ofJanuary, 1975 SILVER MEDAL ADDRESS 23 10 kW, i.e., during the laser “on-time” fluorescence is excited with a source having an effective black body temperature of 75 OOO K at 400 nm and during the “off-time” no signal or noise is being measured.With a duty cycle of about IO-’, i.e., with the laser “on” for only 1 part in l o 7 parts, random noise effects from flame flicker, electronic measurement, noise and fluctua- tions in sample introduction are negligible at most concentrations. The resonance fluores- cence signal to noise ratio at high concentrations is dependent on shot noise in the signal owing to the randomness in the emission of photons, pulse-to-pulse amplitude variations in the dye laser output and also scatter noise.At low concentrations, i.e., near the limit of detection, the resonance fluorescence signal to noise ratio is primarily dependent on scatter noise from particles and optical inhomogeneities in the flame gases. Linear analytical fluorescence calibration graphs were constructed by Fraser and Wine- f ordner for fluorescence of aluminium, calcium, cobalt, chromium, gallium, indium, iron, manganese, molybdenum, nickel, strontium, titanium and thallium over three or four orders of magnitude concentration range.Resonance and non-resonance fluorescence effects were utilised. A limitation of the present dye laser excitation systems is the lower useful wave- length limit of 360 nm, which precludes the possibility of excitation of intense fluorescence of many elements.In flame atomic-absorption and atomic-fluorescence spectrometry with continuum sources and with dilute atomic vapours, the integral of the atomic-absorption coefficient, AT, is generally dependent only on the wavelength and the optical density of the absorber.This is strictly correct, however, only in the limit of zero incident light flux, i.e., when the incident radiation does not cause significant population of the excited state relative to the ground state. This is a good approximation only for low incident light fluxes available from conven- tional sources such as hollow-cathode or electrodeless discharge lamps.When the source flux is very high, however, as in laser excitation, this expression does not provide an accurate description of the attenuation of the light beam. Under these conditions the absorptiort coeficient becomes also a function of the incident radiant flux density (non-linear absorption) .53 The intense radiation may then induce in the sample a state of near-saturation of the energy levels in which the excited state population becomes substantially equal to that of the ground state.Winefordner and co-workers93 and PiepmeierW have described the theoretical and practical consequences of the occurrence of saturation in atomic-fluorescence spectrometry for a broad-band laser excitation source and a monochromatic laser excitation source, respectively.The reader is referred to the publications of these authors for a comprehensive treatment of the theoretical consequences of working near to saturation conditions. In the context of this paper, however, the effect of working close to saturation conditions, and the important observa- tion that under these conditions the radiated fluorescence flux is influenced very little by collisional quenching, have several consequences of great importance from a practical analytical point of view: (1) The fluorescence signal is not greatly influenced by the source stability.The satura- tion effect above a certain value of the source power ensures that any pulse-to-pulse variation in the source irradiance does not affect the stability of the fluorescence signal. (2) The linearity of the fluorescence working curve is extended to higher concentrations. At saturation, the medium becomes transparent at the wavelength of the transition because the excited and ground-state populations are similar and further absorption in the irradiated volume cannot occur.If the irradiance of the source is such that the atomic system can be kept at saturation for any value of analyte atom concentration, the fluorescence flux will be linearly related to this and the calibration graphs will have a slope of unity even at high apical densities.If the medium is transparent, self-absorp- tion cannot take place, and also the source irradiance is no longer a function of the length along the absorption path. Obviously, this consideration of transparency under saturation conditions applies only for the irradiated volume; the factors of self-absorp- tion and self-reversal must still be considered for the post-filter effect produced in any unilluminated volume between the fluorescing volume and the detector.(3) The proportional dependence of fluorescence signal on the quantum efficiency which is observed at low irradiance is removed under saturation conditiorts.Thus, provided that the atomisation efficiency does not change, the magnitude of the saturated fluores- cence signal would not be expected to be greater in the oxygen - argon - hydrogen flame than in a hydrocarbon flame containing nitrogen in spite of the large difference in24 SILVER MEDAL ADDRESS Proc. Analyt. Div.Chem. SOC. quantum yield for these flames. Thus, the nitrous oxide - acetylene flame is just as effective as other flames from this viewpoint and can be used in atomic-fluorescence spectrometry to maximise atomisation efficiency and minimise chemical interferences without the penalty of lower quantum yield experienced under low irradiance condi- tions. Obviously, the first consequence has a direct bearing on the detection limits obtained in atomic-fluorescence spectrometry using a low source.The extension of the linear range of the analytical working curves is of direct relevance in the context of multi-element analysis as is also the benefit of independence of the fluorescence signal on the flame type employed. Thus, a single flame type and stoicheiometry may be suitable for the determination of a large number of elements; we may expect the successful use of linear working curves over a very wide range of concentrations due to the improved signal to noise ratios at low optical densities and the extension of the upper limit of linearity obtained when working close to saturation conditions.I hope that I have been able to illustrate some of the problems for multi-component analysis that result from the very process, the absorption of electromagnetic radiation by atoms and molecules, which makes the techniques of spectroscopic analysis possible.I also hope I have shown the results of some recent work, both in our own laboratory and elsewhere, directed to- wards solution to these problems. It is a pleasure to thank the students who have worked with me during the past years in our research in analytical atomic and molecular spectroscopy.For our own studies reported in this paper I wish to thank Mr. C. de Lima, Dr. L. Ranson and Dr. A. F. Ward. I wish again to express my thanks to the Council of the Society in honouring me with the award of the first Society for Analytical Chemistry Silver Medal. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. SO. 31. 32. 33. 34. References Kirkbright, G. F., and de Lima, C. G., ANaZyst, 1974, 99, 338. Shpol’skii, E. V., Zh. Prikl. Spektrosk., 1967, 7 , 492. Hooymayers, H. P., Ph.D. Thesis, Utrecht, 1966. Zeegers,P. J. T., Smith, R., and Winefordner, J. D., Analyt. Chem., 1968, 40, 26A. Walsh, A., in Kirkbright, G. F., and Dagnall, R. M., Editors, “Atomic Absorption Spectrophotometry,” Mavrodineanu, R., and Hughes, R. C., Appl. Opt., 1968, 7, 1281. Aldous, K. M., Mitchell, D. G., and Jackson, K. W., PYOC. 4th Int. Conf. Atom. Spectrosc., Toronto, Rawson, R. A. G., Proc. 4th Int. Conf. Atom. Spectrosc., Toronto, 1973. Kirkbright, G. F., and Ranson, L., unpublished work. Kirkbright, G. F., Ward, A. F., and West, T. S.. Analytica Chim. Acta, 1972, 62, 241. Kirkbright, G. F., Ward, A. F., and West, T. S., Andytica Chim. Acta, 1973, 64, 363. Greenfield, S., Jones, I. C.. and Berry, C. T., Analyst, 1964, 89, 713. Wendt, R. H., and Fassel, V. A., Analyt. Chem., 1965, 37, 920. Wendt, R. H., and Fassel, V. A., Analyt. Chem., 1966, 38, 337. Dickinson. G. W., and Fassel, V. A., Analyt. Chem., 1968, 40, 247. Hoare. H. C., and Mostyn, R. A,, Analyt. Chem.. 1967, 39, 1153. Ekmmans, P. W. J. M., and de Boer, F. J.. Spectrochim. Acta, 1972, 27B. 391. Kirkbright, G. F., and Ward, A. F., Talanta, 1974, 21, 1145. Willis, J. B., Spectrochim. Acta, 1970, 25B, 487. Koirtyohann, S. R., and Pickett, E. E., Paper presented a t the XIIIth Colloquium Spectroscopicum de Galan, L., and Samaey, G. F., Spectrochim. Acta, 1970. 25B, 245. Amos, M. D., and Willis, J. B., Spectrochim. Acta, 1966, 22, 1325 and 2128. Manning, D. C., and Capacho-Delgado, L., Analylica Chim. Acta, 1966, 36, 312. Kirkbright, G. F., and West, T. S . . Chemy Britain, 1972, 8, 428. Mitchell, D., “Proceedings of Technicon International Symposium, New York, 1970,” Halos Associ- Mitchell, D. G., and Johansson, A., Spectrochim. Ada, 1971, 26B, 677. Jones, M., Kirkbright, G. F., Ranson, L., and West, T. S., Ancxlytica Chim. Acta. 1073, 63, 210. Gardels, M., Demers, D., and Mitchell, D., “Proceedings of Technicon International Symposium, Dagnall, R. M., Kirkbright, G. F., West, T. S., and Wood, R., Analyst, 1972, 97, 246. Fraser, L. M., and Winefordner, J. D., Analyt. Chem., 1971, 43, 1693. Fraser, L. M., %nd Winefordner, J. D., Analyt. Chem., 1972, 44, 1444. Omenetto, N., Hatch, N. N,, Fraser, L. M., and Winefordner, J. D., AnaZyt. Chem., 1973, 45. 196. Omenetto, N., Benetti, P., Hart, L. P., Winefordner, J. D., and Alkemade, C. Th. J.. Spectrochim. Piepmeier, E. H., Spectrochim. Acta, 1972, 27B, 431 and 445. Butterworths, London, 1970, p. 1. 1973. Internationale, Ottawa, June, 1967. ates, Miami, 1970. New York, 1970,” Halos Associates, Miami, 1970. Acta, 1973,28B, 289.