16 Analytical Atomic Spectroscopy 1.3 FLAMES While much of the work reported this year reflects a consolidation of existing knowledge, some significant contributions have been made in three areas. In the study of interference effects, new insights into mechanisms, and original experimental approaches to minimiza- tion have been offered. Several comprehensive investigations of noise and precision have allowed invaluable quantitative comparisons of different flame and instrument types.A great deal of attention has been paid to the study, development and improvement of nebulizers and in particular to micro-sampling devices, which continue to grow in popularity. 1.3.1 Fundamental Studies Few reports of new frame types have appeared this year although an investigation of a N,O/ toluene flame (2063), with a maximum temperature of 2600 K, showed impressive AAS detection limits; Pt 0.03 pug ml-1, Pd 0.04 pg ml-1, Rh 0.05 pg ml-1 and Ir 0.5 pg ml-I.Further study has been made of the O,/C,H,/He flame reported last year (see also ARAAS, 1978, 8, Ref. 207). Its capabilities for AFS have again been compared (1351) with the air/C,H, Aame and its superior quantum efficiency, 2.3-fold for Fe and 5.7-fold for Cr, demonstrated.This arises largely as a consequence of the dilution of quenching species by the He. Improvements in atomization for elements such as Ca were also obtained. At the present stage of development however, both SNR and AF signals are lower. Intriguing possibilities for AES were suggested by an investigation of effective cross-sections for the electron-impact excitation of Cu lines (1 819).Theoretical studies have included an investigation of the chemical equilibria of the species generated in the N,O/C,H, high-temperature flame used in AAS (1258). An important finding was that the equilibrium compositions were profoundly affected by theAtomization and Excitation 17 introduction of aqueous solutions.Rise velocities (1 324) and rise-velocity profiles (1 326) continue to be studied in the air/C,H, flame using the single aerosol droplet technique (see Anal. Chem., 1968, 40, 1860). Additional reference on the preceding topic - 380. The measurement of temperature in small defined regions of a flame has been achieved with a 3-line (Co) spectroscopic method (1726).The measurement zone was defined, in the horizontal plane, by the position at which single aerosol droplets were introduced and, in the vertical plane, by mounting the monochromator on its side with the slits horizontal. The flame itself could be moved in both planes by a stepping-motor to map the temperature distribution. Measurements were in good agreement (within 40-50 K) with published data for the air/C2H, flame. An interesting 3-wavelength pyrometer has been described (1 102) for use in dust explosions and flames, which has potential for analytical application.Temperature was determined by measuring the continuum emission from dust particles at wavelengths of 0.8, 0.9 and 1.0 pm. Additional references on the preceding topic - 1169, 1729, 2062.Several theoretical and experimental studies of noise and its origins in flames have been made. A standardized figure-of-merit for the effect of flame emission noise in FAES, FAAS and FAFS was derived (1650) and used to determine the choice of optimum modulation frequency. A computer-controlled spectrometer system using FAES and FAFS was employed (1653) to determine the predominant type of noise in air/C,H,, N,O/C,H,, N,O/propane, air/H, and iso-octane liquid fuel flames.The precision of single-beam and double-beam instruments was compared and the conclusions drawn that near the detection limit the single-beam was slightly superior to the double-beam instrument, but at higher absorbances analyte absorption noise was predominant and the two instrument types differ little (929, 1 199).Another comparative noise study, between different flame types, produced the conclusion to the great relief of many analysts, that, on the basis of background noise and overall precision, the air/C,H, flame was the best choice for most purposes (602). A comprehensive experimental investigation of the precision of FAES considered the properties of conventional slot and Meker burners i n a commercial AA spectrometer (619). For all types of C2H2 flames, maximum analyte emission signal, SNR, and SBR at high and low analyte concentrations were obtained at near stoicheiometric fuel/ oxidant ratios with the Meker burner.With the slot burner, results were similar except that maximum SBR and maximum SNR for low analyte concentrations required lean-flame conditions.A similar study of FAAS precision with the same instrument (464) presented curves of RSD of absorbance versus absorbance for the elements Al, Co, Cr, Eu, K, Mn, Ni, Pb, Se, Si, Ti and V. An instrumental method for improving the precision of AES (and potentially AAS) measurements in a N20/C,H, flame has been described (1966). This used one channel of a dual-channel instrument for the analyte and the second channel to monitor CN band-head emission.The latter signal was used to stabilize the CN emission by servomechanically controlling the C2H2 flow rate. The improved RSD, demonstrated in the determination of A1 by this method, was superior, by one order of magnitude, to that obtained by using CN as a form of internal standard.While not pcrhaps of immediatc analytical interest, the surprisingly large differences, up to lo%, between the molar absorbances of the isotopes 40Ca, 44Ca and 48Ca in air/C,H, and N,O/C,H, are worthy of note (1746). These were explained in terms of isotope shifts and Lorentz and Doppler broadening effects. Other references of interest - Background absorption correction by Zeeman and other magneto-optic cffccts: 906, 907, 91 3.See also Chapter 2, Section 2.2.1. Oscillator strengths for Cu: 381.18 Analytical Atomic Spectroscopy 1.3.2 Atomization and Interference Studies Research on the mechanisms of atomization and interference continues to make extensive use of the single-droplet generator. In an elegant study (1328), the flame was scanned vertically above an introduced droplet to provide emission intensity versus time information.This was applied in a theoretical model to obtain vaporization and ionization rate constants, which might offer routes to practical methods for reducing vaporization interferences. Another investigation of a single system of Cu and Zn as chloride, nitrate and sulphate employed the single-droplet technique to produce a particle of a controlled size in the flame (98, 459).Vaporization rates and particle trajectories in the flame were studied. An interesting insight into the mechanism of the buffering action of Ba, La and Sr in FAAS has been given by Shcherbakov and Karyakin (1938). They showed that the matrix element interference increased in the order of decreasing crystallographic ionic radius, i.e., Fe (0.74) < A1 (0.51) < Si (0.42).The buffering by Ba, La and Sr was attributed to improved analyte diffusion in the less dense structures formed by the large buffer cation (Sr, 1.12; Ba, 1.34; La, 1.14) compared to the more closely packed structure formed by the matrix element (see also ARAAS, 1978, 8, Ref. 810). The releasing effect of, for example, La on Mg in the presence of phosphate has been studied by using an AAS inhibition- release titration technique (1004). Additional references on the preceding topic - 140, 920, 1542, 1693, 1770. Comprehensive experimental invcstigations of interference in the determination of specific elements include studies on Cr (1897), Ru (1963) and Ta (1049). Fresh examples of specific superpositional spectral interferences have been reported; FeO molecular bands on the Cs 825.1 nm line in FAES (746) and background interference by B on the Pb 405.8 nm line (1078) in FAES and in d.c.Ar arc plasma emission spectrometry. An instrumental selective spectral line modulation approach has been used to reduce interferences in FAES (1327) (see also J. A. Bowman, J. V. Sullivan and A.Walsh, Spectrochim. Acta, 1966, 22, 205). Two flames, one as the sample emission source and the other as the selective modulator were arranged in a dual-beam system. Emission from the source alternately bypassed or passed through the second modulating flame. At wavelengths where emission from the source coincided with absorption by species in the second flame, modulation occurred.Thus an a.c. signal was generated at the spectrometer detector whenever flame background features or element spectral lines were common to both flames. By introducing pure analyte element into the second flame, selective modulation was achieved with a reduction in both narrow-line and broad-band spectral interference effects. Since the modulation occurred only for the narrow absorption-line band-width, a consider- able improvement in the eaective resolving power of the complete instrument resulted.An investigation of the profile of the Ca 422.7 nm resonance line, isolated by a selective modulation technique, has been made using a Fabry-Perot interferometer (9 14). The results demonstrated the sharpening of atomic resonance lines by selective modulation.Interferencc effects by various cations in the detcrmination of Mn (304) have been eliminated by the use of the appropriate flame conditions, i.e., by employing N,O/C,H, in place of air/C,H,, at the expense of a modest (2X) reduction in sensitivity. The perennial problem of Ca interference in the determination of Ba by FAAS was reduced by using a reducing N,O/C,H, (382) or a separated N,O/C,H, flame (751).In the analysis of CU alloys, preheating thc aerosol minimized interferences from the matrix (7 19). Schliercn techniques were used to study (1587) changes in the flames produced by the presence of organic solvents. It was demonstrated that the failure of organic solvents to give the full expected increase in sensitivity from the increased rate of analyte input could be correlated with changes in flame geometry.It has been shown (1715) that organic solvents such as petroleum products, which arc difficult to use with analytical flames, can be accom-Atomization and Excitation 19 modated by the N,O/H, flame. Improvements (10-fold) in detection limits were claimed (1833) for diethyl ether/methanol mixtures in the determination of Mg and Zn in copper and nickel salts by FAAS.The replacement of inorganic acids, such as HC1, by organic acids, such as acetic acid, has been recommended (571), and was reported to minimize the deposition of salts on the burner slot when solutions of high salt content were analysed. This method was investigated with solutions prepared by sodium peroxide fusions with organosilicon compounds in a P a n bomb, followed by dissolution of the melt in acid.Different acids, HI, HCl, HNO, and acetic, were compared. No relationship between salting-up of the burner and solubility of the salt was evident. Noise levels, however, lowest with HI, showed an inverse relationship with the solubility of the sodium salt. Other references of interest - Ca interference suppression, silicate analysis: 796.Enhancement of A1 absorbance by nitrogen compounds: 174. Interference, slag and cement analysis : 922. Mathematical interference correction procedures: 51 1, 1803, 1804. Mineral acid interferences on CaOH emission: 122. Molecular absorption spectra, Na salts: 1799. Suppression of interferences on CaOH emission: 699. 1.3.3 Devices €or Sample Introduction Some of the important parameters of sample transport and atomization have been studied in conventional pneumatic, ultrasonic and high-solids nebulizers using laser Fraunhofer diffraction measurements (1307) to provide data on dynamic particle and droplet size.The apparatus was incorporated into a commercial microcomputer-based AAS instrument and enabled rapid presentation of particle size histograms together with mean and median particle diameter for particles in the size range 2.4-150pm. The fundamentals of aerosol production and transport have been discussd (1319) and used to predict ideal design solutions for practical systems (202). 1.3.3.1 Nebulizers.The improvement of nebulizer design and efficiency has been the subject of a refreshingly systematic new look at the problem employing three different techniques (153, 1295).In the first method the solution was charged electrically, by, for example, applying an electric potential to the nebulizer tip. This produced a 63% decrease in droplet diameter. A second technique made use of the disruptive and exothermic decomposition of thermally unstable compounds, such as hydrazine, dissolved in the solution.In yet another method, by dissolving a highly absorbing dye, such as nigrosin, in the solution thermal dissolution and decomposition was encouraged. Two new pneumatic nebulizer designs of the concentric type (1204, 1365, 1596) with demountable and interchangeable components have been described. A design for a modified Babington-type of nebulizer has been reported (1738) and its operation discussed.The authors pointed out that with this type of nebulizer higher solution flow rates, 50 ml min-1 in this case, were required compared with conventional nebulizers with flow rates of about 5 ml min-1. This design, however, loses some of its attraction for trace-element analysis by using nickel-plated brass in its construc- tion, In a further study of the Babington-type of nebulizcr (546) it was claimed that the formation of a “Babington film” of sample liquid is not essential in recent designs.The same authors advocated the use of homogenates of meat, ~021, etc., with such nebulizers to avoid the use of tedious sample digestion procedures for FAAS or plasma emission analysis. While spectral interferences were found to be low with this preparative method, it would be surprising if chemical and matrix interference effects were not a considerable problem for some elements at least.20 Analytical Atomic Spectroscopy Informative comparisons of nebulizer performance (926, 1998) were carried out for pneumatic (concentric and cross-flow) types, with and without desolvation, and for an ultra- sonic nebulizer with desolvation.Quantitative data presented included comparative analyte delivery rates, pneumatic/concentric, 110 ng min-1; pneumatic/cross-flow, 181 ng min-1; rising to 900 and 81 5 ng min-1, respectively, when used with desolvation. The ultrasonic nebulizer delivered analyte at a rate of 1186 ng min-1. The limitations of desolvation for solutions with high salt contents were illustrated by, for example, a fall from 57.4% absorption, obtained with 0.4ppm Cu in aqueous solution, to 16.6% in 5% sodium chloride solution using the ultrasonic nebulizer.An impact cup interchangeable with a conventional impact bead has been used to reduce the volume of sample reaching the flame and also the droplet size with advantages for solutions containing a high salt content (1 504). 1.3.3.2 Microsampling Devices. Solution microsampling devices for FAAS and FAES continue to proliferate (160, 239, 1942), many with automatic or semi-automatic injection (97, 1763, 1808). They have also been widely reported for flow injection analysis of digests of plant materials for Co and Cu (1458), Cu and Zn (1708), and Cu and Mg (495). 1.3.3.3 Sample Zntroduction by Volatilization. The technique of volatilizing the sample into a flame for atomization has attracted further interest. A graphite furnace atomizer has been used for this purpose (928; see also ARAAS, 1977, 7, Ref, 1062). The apparatus was improved by introducing a slotted quartz T-tube into the flame as the final atomizer. The advantage was a much reduced non-specific light loss by scattering or molecular absorption compared with direct analysis by electrothermal atomizer methods.Results for As, Hg and Se were presented and applications in AFS and ICP-AES considered. Direct current arc volatilization of copper sample from graphite electrodes into an air/C,M, flame (489) has also been described. Cadmium, Pb and Zn were determined by FAAS with detection limits of 0.08, 0.7 and 0.1 1 pg g-1 respectively and the spectral interference from the copper and zinc matrix eliminated.A laser, of the type used in laser microprobe analysers, was applied to volatilize Ag, Au and Ni into an air/C,H, flame (1689) for the surface-analysis of electroplated copper sheets by AAS. The electrically heated graphite chamber-in-flame method (ARAAS, 1977, 7, Ref. 466), which is suitable for solid powder samples, continues to receive attention, mainly by Russian workers. The method has been applied to the determination of elements forming refractory oxides, such as Al, B, Ca, Mo, V and W, and in addition makes use of the reducing properties of a luminous-diffusion flame (303, 1951). The advantages of the technique for solid powder samples were illustrated in the analysis of sulphide ores, which were mixed with graphite powder and placed in the chamber.Cd, Cu, In, Pb, Sb and T1 were determined with detection limits of 7.3X 10-10 g, 3.1 X 10-11 g, 9.3X 10-10 g, 2.1 X 10-9 g and 3.4X 10-9 g, respectively. Further development of the atom-trapping concentration procedure (see also ARAAS, 1976, 6, Ref. 1288) has been reported for Cu (1464). In this technique, sample solution was nebulized in the usual way for fixed periods of time into an air/C,H, in which a water- cooled silica tube was mounted. Copper species collected on the tube were released into the flame and atomized when the cooling water was switched off. Some 50-fold gain in sensitivity was obtained for 0.01 pg ml-1 Cu compared with convcntional FAAS.The principles of the Pt-loop solution microsampler for FAAS have been discussed (1684) and the method applied to the determination of Cd and Pb in drinking water. 1.3.3.4 Injection Methods. The advantages of the direct analysis of solid, powder or slurry samples, in avoiding the problems of a sample digestion and dissolution, still attract con- siderable research despite the obvious difficulties of reproducible sampling, samplc homo-Atomization and Excitation 21 geneity and matrix interferences.Solid metal samples, in the form of rod or wire, etc. were held in a graphite rod and inserted into a flame for analysis by AAS (1751) (see also H. Ramage, Nature, 1930, 126, 279 and K. Govindaraju, J. Morel and N. L’Homel, Geosturidards Newsl., 1977, 1, 137).Aspiration of HCl dramatically increased the signal. Slurries of finely comminuted (<325 mesh) coal in a 0.5% aqueous solution of Triton X-100 were used (215, 530, 1694) for the direct determination of some 23 elements by AAS in air/C,H, or N,O /C,H, flames. Reasonable accuracy (+5-*25% error) and precision (%1-*4%) were claimed. The use of brass-bodied metal screens for sizing the coal powder must leave some doubts, however, about freedom from contamination by Cu and Zn.In comparing this with an XRF method the authors state that the total time for the determina- tion of 12 elements in 6 samples was the same for both methods. I .3.3.5 Chromatographic Sample Introduction. The use of flame spectroscopic detectors iit chromatography continues to grow.The band emission of Se, (principal bands 450-500 nm) was used as the basis for a Se detector (129). The sensitivity for Se and the quenching of emission by carbon compounds were both less than for S. The detection limit for Se was about 2x10-12gs-1. The normally exponential response of S , Se and Te detectors was linearized (329) by providing a high S background.A F-specific detector, using CaF band emission at 529nm, designed for GC (see also 2. Anal. Chem., 1972, 258, 273) has been applied to HPLC (1686). This used an O,/C,H, total-consumption burner with two aspira- tors, one connected to the HPLC output and the other to a solution of Ca(NO,), in methanol / water. A simple low-cost FAFS detector interfaced to a standard liquid chromato- graph (1400) employed a chopped Xe continuum source, a N,-sheathed air/C,H, flame, a 0.1 m monochromator, and synchronous electronic detection. Its performance was discussed and compared with a conventional molecular u.v.-absorption detector.The detection and determination of organic compounds by complexing them with a suitable metal and using a liquid chromatograph with a FAAS metal-specific detector has been discussed and illustrated by the determination of a copper-labelled amino acid, histidine (10 17).A chromatographic support treated with sodium azide and hexamethylphosphorotriamide (HMPT) has been used to concentrate carbon disulphide from air for final analysis by a flame emission method (1 854). 1.3.4 Applications of Lasers The past year has seen continued intense activity in the field of laser applications in analytical chemistry and the separate sub-section initiated in Vol. 8 of ARAAS is repeated in the present volume. A useful book cokering much of the interesting work in this field has appeared in 1979. The book entitled “Analytical Laser Spectroscopy”, contains eight chapters by different authors under the general editorship of Omenetto (1 138).A review of recent developments in tunable lasers for spectroscopy has also been published (2067) and two authors have compared the potential of several laser techniques for individual atom detection (421, 1287). 1.3.4.1 Atomic Fluorescence Spectrometry. A review of the theory and applications of laser AFS has been published (687). Further improvements in limits of detection, using pulsed dye lasers and a specially designed aerosol delivery /flame system, have been described (1 398).This new system was evaluated with respect to long-term stability and to the accuracy and precision achieved in real sample analysis. Another paper from the same group described methods used to overcome problems of molecular fluorescence and scatter observed in the analysis of NBS CRMs (1402).Problems associated with the practical applications of laser AFS have generated more interest this year and Yeung et a!.22 Analytical Atomic Spectroscopy (1349) have described developments aimed at tackling three of the major problems in this field, namely, lack of high intensity tunable U.V. lasers, light scattering, and lack of atom sources with good efficiency and good optical properties.The authors discussed the use of collision-induced non-resonant excitation, wavelength modulation, and introduced the ICP and premixed O,/H, flame as atom cells. The ICP has also been evaluated as an atom cell for laser AFS by Winefordner and co-workers (1 164, 1373). Use of the ICP with a CW dye laser was not recommended as detection limits were worse or not significantly better than those obtained using the ICP in emission mode (1164).The pulsed N,-pumped tunable dye laser was claimed however to be a useful source for gcnerating AFS signals in an ICP and has been evaluated in detail (1373). The measurement of laser AF signals using a graphite furnace as the atom cell was described by Bower et al.(1375). Both conventional furnaces and continuous-flow furnaces were evaluated. Measurements of laser AF in a furnace atomizer have also been reported by Falk and Tilch (934) who claimed a detection limit for Pb of 0.02ppb. Analytical measurements of Na AFS in flames using pulsed dye lasers was reported by Gonchakov et al. (1843) and Daily er al. (345). The former group gave a detection limit of 0.0004ppm and a working range up to 1 ppm using an air/C,H, flame, Experimental observations of the saturation behaviour of Na atoms in a variety of flames have been reported by Omenetto et al.(1348). The intensity ratio of the two AF lines for the yellow Na doublet can be significantly different from the thermal value of 2, depending on the relative values of the quenching and mixing cross-sections of the levels involved.For excitation at 589.0 nm it was predicted that the ratio would be 2 only if the quenching rate and the radia- tive spontaneous rate are much smaller than the mixing rates between the two excited levels. The ratio should decrease on going from a flame with a poor quantum efficiency to a flame with high quantum efficiency.This conclusion was supported by experimental data obtained using six different flames. Saturation using laser excitation has also been evaluated as a method for the determination of the absolute concentration of Na and Mg in air/C2H2 flames (682). The occurrence of energy level saturation using lasers has been confirmed by comparing experimental plots of AF signal versus laser power with theoretically derived expressions (725).Resonance Rayleigh scattering of laser radiation by Na vapour in shielded O2/H,/N,, O,/H,/Ar and O,/C,H,/N, flames has also been described (693). 1.3.4.2 Molecular Fluorescence Spectrometry. The fluorescence spectra of SrOH, CaOH, CrO, MnO and C, (440) and C,, CH, OH, CN in air/C,H, and N,O/C,H, flames (317, 1658) have been characterized.The spectra of flame species extend over a broad wavelength range and contribute significantly to the background signal in AFS mcasuremcnts. These observations indicate the potential of laser AFS for obtaining spatially resolved profiles of species in flames and the E P . Methods for the characterization of the temperature, density of species and quantum efficiencies have already been described (1350).A lascr fluorescence method was used to measure the rotational temperature of the lowest vibrational state of the electronic ground state of OH in an air/CH, flame and this was found to be in equilibrium with the N, vibrational temperature measured by laser Raman scattering (1099). Other references of interest - Effect of thermal variations on the spatial mode structure of a linear flashlamp-pumped dye laser: 2046.Measurement of absolute OH concentrations in low-pressure flames by laser fluorescence: 135. Spin-orbit rclaxation of Pb (SPJ by H after laser excitation: 396.Atomization and Excitation 23 1.3.4.3 Laser-enhanced loni~aiiort Spectrometry (LEI). Considerable advances have again been achieved in the use of optogalvanic detection in trace-element analysis.The principal authors in this field, who now prefer to refer to their technique as laser-enhanced ionization, have produced a stream of original (1551, 1692, 1987) and conference papers (99, 100, 574, 575, 1077, 1352, 1372, 1397). In addition the thesis of Turk is now available (1278) and two US. patent applications (1228, 1277) have appeared, which describe the apparatus and method used for galvanic detection.The method employs a tunable dye laser to promote analyte atoms in a flame or other atom reservoir to a discrete excited atomic level. Thermal ionization from this excited state occurs at an enhanced rate compared to the ground state and the enhanced ionization is detected electrically by applying a voltage across the flame and observing changes in the electrical current through the flame.Both W and Mo rods and plates have been used as electrodes and were placed 12mm apart, with the lower end 6 mm above the 5-cm slot burner head supporting the flame (1692). In Vol. 8 of ARAAS (Section 1.3.4.2, p. 22) preliminary results were reported for five elements indicating that ppb detection limits were achievable and that the magnitude of the difference in energy between the excited atomic state populated by the laser radiation and the ionization energy, referred to as E,*, had a critical effect on sensitivity.In a recent paper, detection limits for 18 elements at 28 wavelengths were reported (1692) that supported the earlier conclusions.In all cases, except for Cu, detection limits were in the ppb range, were comparable to or better than laser AFS and were remarkable for flame measurements. Most of the transitions employed originated from the ground state, but some used the low atom populations in thermally excited states. Acceptable or good detection limits were still achieved as long as the value of Ei* was small.For example Pb at 283.3nm (transition 0-35287 cm-I, Ei* 24533 cm-1) gave a detection limit of 3 ngml-1 whereas at 280.2 nm (transition 10650-46329 cm-1, Ei* 13491 cm-1) a value of 0.6 ng ml-1 was achieved, The detection limit of 100ngml-1 for Cu was ascribed to the large value of Ei* and poor performance was also assumed for elements with high ionization potentials, such as As, Cd, Se and Zn, which have not so far been measured.A mechanism for LEI has been described by means of a model that takes into account the degree of population of the excited state, the exponentially improved ionization rate from the excited state and the depletion of neutral species by ionization ( I 372, 1551). The model was fitted by a least-squares procedure to experimentally measured sensitivities for 14 elements and 23 different transitions.The investigation of ionization interferences in LEI has been reported (575, 1692). Moderately high concentrations of elements such as Na, which are easily ionized thermally, have a serious depressive effect on LET signals. The increased level of electrons in the flame causes an increase in ion recombination rates and hence a depression of signal.It also gives an increased background current, which reduces measurement precision. Serious depression by Na above 10 pg ml-1 was shown for In at 10 ng ml-1 but this was released by application of higher cathode potentials and greater tolerance was found using W plates compared to rods for the cathode. In the presence of easily ionized elements, analysis by standard addition or by using carefully matched standards, or with prior separation using ion-exchange columns (573, was found to be necessary.The application of LEI to the determination of Ni and Mn in alloys was demonstrated (1692). CW dye lasers have been compared with pulsed lasers for LEI using both flame and furnace atomizers (574, 1397). Comparison was also made between LEI and AFS detection.In both cases detection limits were reported to be in the sub-ppb to ppb range with linear dynamic ranges of 4 to 6 orders of magnitude. When using a single laser a maximum photon energy of 4.4eV is available. For elements of high ionization potential, this still leaves an energy gap, Ei*, which is too large for significant thermal ionization to occur.Although two-photon transitions with a24 Analytical Atomic Spectroscopy single wavelength laser have been observed (1692), the sensitivity was poor due to the low probability of such transitions. Attempts have therefore been made to use a second light source to promote the second step in the excitation process. New or improved detection limits were reported for Cd, Cu and Zn by use of a Xe arc continuum source or EDL light sources as supplementary photo-excitation (100).Recent work has included the use of two tunable dye-lasers pumped by a single N, laser (1352, 1987). Improvements in detection limits for 7 elements varying from a factor of 7 to 1600 were reported (1987) compared to the use of a single laser. Other reference of interest - Opto-galvanic spectroscopy in a U hollow-cathode discharge : 726. 1.3.4.4 Other Studies. Resonance ionization spectroscopy determination of Li via a three- photon process involving two separate pulsed lasers tuned to 670.8 and 610.4nm has been demonstrated (179). Li atoms were generated by photodissociation of Li vapour using a third laser. The determination of pg amounts of Na in a flame by photo-ionization has also been reported (1868). The analytical application of acoustic pulses generated in a flame following laser excitation of sodium atoms has been described (1571; see also ARAAS, 1978, 8, Ref. 8). Linear calibration was obtained over the range of 1-100 pg ml-1. A two-photon absorption technique has been proposed for the measurement of atomic 0 in the mesosphere and lower thermosphere (134).Excitation by laser radiation at 225.6 nm using two photons to populate a level equivalent to 130.4 nm resulted in fluorescence at 130.4 and 844.7 nm. It was proposed to use the line at 844.7 nm as the resonance fluorescence is strongly absorbed by ground state 0 atoms. Other references of interest - Coherent optical transient spectroscopy in flames: 132. Redistribution of population among the higher levels of Na in an O,/H,/Ar flame following laser excitation: 1256. 1.3.5 Atomic Fluorescence Spectrometry An instrument designed for multi-element analysis by Xe arc continuum source FAFS and FAES, with which the thirteen elements Ag, Ba, Ca, Cd, Cu, Fey Mg, Mn, Ni, Pb, Si, Sn and Zn were determined in lubricating oil, has been described (1733). The Xe arc source has also been used for automatic correction of the background due to light scattering (1177) in an instrument for the determination of Cd by FAFS. Detection limits of 0.5 and 0.1 pg 1-1 for blood and urine, respectively, and improvements in precision were obtained. Non-dispersive FAFS using an air/LPG flame has been applied to on-line monitoring of Mg in boiler-feed water (938). It has also been used for multi-element analysis of air-filter samples volatilized by a graphite furnace atomizer into an air/C,H, flame. The detection limits for Cd, Fe, Pb and Zn were, respectively, 0.01, 4, 5, and 0.25 pg (1578). 1.3.6 Methods Based on Molecular Emission New developments in molecular emission spectrometric methods of analysis are few. The determination of ammonium-N,, by measurement of the emission intensity of the NH species (bandhead at 336.0 nm) evolved from alkaline solution has becn reported (141). The application of various methods for the determination of S-species in air using flame photo- metric (S,) detectors have been discussed (427, 1269, 1270). Sulphate-S has been determined following reduction of sulphate to H,S using a tin-orthophosphoric acid reagent. The H,S was fed into a MECA cavity where S, emission was measured (580, 1443). The indirect determination of amines and amino acids using MECA has also been reported (1565).