Atomic Absorption Spectroscopy — Alan Walsh Division of Chemical Physics CSIRO P.O. Box 160 Clayton, Vic, 3168, Australia

When Foil Miller invited me to I realize that anyone who reminis­ tained in absorption and atomic spec­ participate in the Silver Anniversary ces is usually so decrepit that his tra in emission. The result of this Symposium on Great Moments in memory is totally unreliable. I shall, musing was quite astonishing: there Analytical Chemistry, he suggested therefore, try to avoid too many er­ appeared to be no good reasons for that I may care to indulge in some rors of fact by restricting my com­ neglecting atomic absorption spectra; personal reminiscences and com­ ments largely to matters which are on the contrary, they appeared to ments regarding the development and documented in reports of CSIRO, in offer many vital advantages over present status of atomic absorption correspondence, or in publications. atomic emission spectra as far as spectroscopy. I hope the ones I have My initial interest in atomic ab­ spectrochemical analysis was con­ selected may illuminate, if not an­ sorption spectroscopy was a result of cerned. There was the attraction that swer, the problem posed by the title two interacting experiences: one of absorption is, at least for atomic va­ of my address. the spectrochemical analysis of met­ pours produced thermally, virtually als over the period 1939-46; the other independent of the temperature of Presented at the Pittsburgh Conference on of molecular spectroscopy over the the atomic vapour and of excitation Analytical Chemistry and Applied Spec­ period from 1946-52. The interaction potential. In addition, atomic absorp­ troscopy, Cleveland, Ohio, March 6, 1974. tion methods offered the possibility of Silver Anniversary Symposium on Great occurred early in 1952, when I began Moments in Analytical Chemistry and to wonder why, as in my experience, avoiding excitation interference, Applied Spectroscopy. molecular spectra were usually ob- which at that time was thought by many to be responsible for some of the interelement interference experi­ enced in emission spectroscopy when - 2 - using an electrical discharge as light source. In addition, one could avoid problems due to self-absorption and self-reversal which often make it dif­ The purpose of this report .is to suggest a new technique for ficult to use the most sensitive lines recording absorption spectra which offers many interesting possibilities. in emission spectroscopy. The method is basically simple and is illustrated in the diagram below. As far as possible experimental problems were concerned, I was par­ ticularly fortunate in one respect. For several years prior to these first thoughts on atomic absorption, I had been regularly using a commercial in­ frared spectrophotometer employing a modulated light source and synchro­ nously tuned detection system. A fea­ ture of this arrangement is that any radiation emitted by the sample pro­ duces no signal at the output of the detection system. This experience Assuming that the sample is vaporised by the usual methods, e . g. had no doubt prevented the forma­ flame, arc, or spark, then the emission spectrum is "removed" by means tion of any possible mental block as­ of the chopper principle . Thus, the absorption spectrum of the vapour sociated with absorption measure­ ments on luminous atomic vapours. is measured by passing through it white light which is chopped. The absorption spectrum and the emission spectrum are then scanned by a In an internal report for the period February-March 1952,1 suggested detector, the output from which is amplified by an amplifier tuned to that the same type of modulated sys­ the same frequency as the chopper. Thus the emission spectrum produces tem (Figure 1) should be considered no output signal and only the absorption spectrum is recorded. for recording atomic absorption spec­ For analytical work it is proposed that the sample is dissolved, tra. The following extracts from that report may be of interest: and then vaporised in a Lundegardh flame . Such flames have a low temperature (2000°K) compared to arcs or sparks (500CTKJ and have the advantage that few atoms would be excited, the great majority being in the ground state . Thus absorption will be restricted to a small number Figure 1. Extract from report for Feb­ ruary-March 1952

698 A • ANALYTICAL CHEMISTRY, VOL. 46, NO. 8, JULY 1974 Report Stagnant or Pregnant?

"Assuming that the sample is va­ was interposed between the sodium ysis. "Look," I said, "that's atomic porised by the usual methods, e.g., lamp and the entrance slit of the absorption." "So what?" was his flame, arc, or spark, then the emis­ spectroscope. When a solution of so­ reply, which was the precursor of sion spectrum is 'removed' by means dium chloride was atomised into the many similar disinterested reactions of the chopper principle. Thus the air supply of the flame the signal at to our atomic absorption project over emission spectrum produces no out­ the oscillograph was reduced to zero. the next few years. put signal and only the absorption The principle of the method is there­ My report for June-July 1952 dis­ spectrum is recorded." fore established." cusses the problems of recording "For analytical work it is proposed In retrospect, such optimistic naiv­ atomic absorption spectra of flames that the sample is dissolved and then ety is quite incredible. with a continuum source and con­ vaporised in a Lundegardh flame. This simple experiment gave me a cludes that a resolution of about 0.02 Such flames have a low temperature great thrill, and I excitedly called in À would be required; this was well (2000°K) compared to arcs and John Willis, who at that time was beyond the best spectrometer avail­ sparks (5000°K) and have the advan­ working on infrared spectroscopy and able in our laboratory at that time. tage that few atoms would be excited, was later to make important contri­ The report concluded as follows: the great majority being in the butions to the development of atomic "One of the main difficulties is due ground state. Thus absorption will be absorption methods of chemical anal- to the fact that the relations between restricted to a small number of tran­ sitions and a simple spectrum would result. In addition, the method is ex­ pected to be sensitive since transi­ tions will be mainly confined to those CHEMICAL PHYSICS SECTION from the ground level to the first ex­ cited state." At this stage I was thinking of elec­ 42nd Bimontly Report. April-May, 1952. trical discharges, as well as flames, as a means of atomization. It will also be apparent that initially I had not C.P. 1/14. Atomic Absorption Spectra appreciated the difficulties which In the previous report the application of atomic absorption spectra would be involved in recording atomic absorption spectra when using a con­ to spectrochemical analysis was suggested. The possibilities of this tinuum source. approach have been explored and the results obtained to date are most The next Bimonthly Report, for the encouraging. In the preliminary work the apparatus shown below was used. period April-May 1952, includes the diagram shown in Figure 2 and de­ scribes our first experiment as follows: "The sodium lamp was operated from 50 cycles/sec. and thus had an alternating output so that it was not necessary to use a chopper. The D lines from this lamp were isolated— but not resolved from each other—by The sodium lamp was operated from 50 cycles/sec. and thus had means of a direct vision spectroscope an alternating output so that it was not necessary to use a chopper. and their intensities were measured by means of a photomultiplier tube, The D lines from this lamp were isolated-but not resolved from each the output from which was recorded other-by means of a direct vision spectroscope and their intensities on a cathode ray oscillograph. Ampli­ were measured by means of a photo-multiplier tube , the output from which fication of the signal was achieved by was recorded on a cathode ray oscillograph. Amplification of the signal the A.C. amplifiers in the oscillo­ was achieved by the A. C . amplifiers in the oscillograph . With the slit- graph. With the slit-width used the signal gave full-scale deflection on the width used the signal gave full-scale deflection on the oscillograph screen. oscillograph screen. A Meker flame A Meker flame was interposed between the sodium lamp and the entrance slit of the spectroscope. When a solution of sodium chloride was atomised into the air supply of the flame the signal at the oscillograph was reduced to zero. The principle of the method is therefore established. No attempt Figure 2. Extract from report for April- May 1952

ANALYTICAL CHEMISTRY, VOL. 46, NO. 8, JULY 1974 • 699 A metallic and semi-metallic elements in plant ash, soil, mineral or similar samples." My reply, dated 27th February 1953, was as follows: "At the moment my work on atom­ ic absorption spectra is still in the development stage and I cannot spec­ ify exactly what equipment will be necessary, but I think it will include the following items :- (a) Monochromator having a reso­ lution of 1 À. (b) Discharge tubes and gas-circu­ lating system. (c) Flame burner assembly. (d) Photomultiplier plus associated power packs. (e) Amplifiers having two homo- dyne rectifiers on the output side. (f ) Ratio-meter, potentiometer or ratio-recorder. (g) One or two choppers. "It is still too early to make any definite claims for the method, but it certainly offers most exciting possi­ Figure 3. Schematic diagram of use of sharp line source bilities and will, I believe, prove par­ to measure peak absorption ticularly valuable in your work, pro­ vided the sample can be taken into solution. It may prove possible to ex­ tend the method to solids. "I should add that we are most absorption and concentration depend resolution required for atomic absorp­ anxious not to divulge any informa­ on the resolution of the spectrograph, tion measurements is, in effect, pro­ tion to people overseas, so please re­ and on whether one measures peak vided by the sharp line source. gard this letter as confidential." absorption or total absorption as At this stage we had arrived at the In view of the almost complete lack given by the area under the absorp­ principle of the technique which, in of interest in our work over the next tion/wavelength curve." due course, became the generally ac­ few years, this request for secrecy was At this juncture the possibilities of cepted method of making the intensi­ quite superfluous. measuring peak absorption were ob­ ty measurements required in atomic Later in 1953 I discussed possible viously coming into consideration. absorption methods of chemical anal­ commercial exploitation of our ideas There is then a gap of four months ysis. with various instrument manufactur­ in my reports, owing to absence from These early experiments, carried ers in the United States and England, the laboratory on sick leave for much out in collaboration with John P. but the only person who viewed our of this period. The next report, for Shelton, were originally confined to work with enthusiasm was A. C. the period December 1952-January hollow-cathode lamps through which Menzies of Hilger and Watts Ltd., 1953, refers to the poor sensitivity ob­ argon was flowed by a closed circulat­ London; CSIRO arrived at a tentative tained in the determination of copper ing system. We did not commence exclusive licence agreement with that by use of a continuum source and a the development of sealed-off hollow- firm, based on the patent which we monochromator obtained by placing a cathode lamps until December 1953- lodged in November 1953. slit and detector on the focal wave of January 1954, when we first became The next significant event was the a Littrow spectrograph. The report aware of the work of Dieke and Cross- first public exhibition of a working states: white (2), which had been published atomic absorption spectrophotometer "It is thought that this (poor sensi­ in 1952. (Figure 4). It was demonstrated in tivity) is due to the low resolution of The first person to express any in­ March 1954 in Melbourne University the Littrow spectrograph and to the terest in the application of the tech­ as part of an Exhibition of Scientific excessive amount of scattered light at nique we had developed was John Instruments, arranged by the Austra­ low wavelengths. It is hoped to over­ David, at the CSIRO Division of lian Branch of the (British) Institute come this difficulty by using a hol­ Plant Industry in Canberra; and on of Physics. low-cathode source (copper cathode) February 24th, 1953, he wrote to me a The apparent complexity of the in­ as source. This will emit sharp lines letter which began as follows: strument was due largely to its being and a low resolution spectrometer "I understand from several sources of the double-beam type, which in will then be sufficient. The first at­ that you have in mind a new tech­ our early experiments we regarded as tempts at producing hollow-cathode nique of spectrochemical analysis in­ essential because of the poor stability sources have not been successful." volving the measurement of the ab­ of many of our hollow-cathode lamps. This use of a sharp line source to sorption by the analysis line of an el­ The viewer was possibly further con­ measure peak absorption is illus­ ement in a vaporized sample rather fused by the optical path being in op­ trated schematically in Figure 3. In than the measurement of its emission. posite directions on the instrument this case, the function of the mono­ "I would very much appreciate any and on the explanatory diagram! chromator is to isolate the required information which you may have and Whatever the reason, the instrument line for measurement from all other are prepared to give me regarding its aroused no interest whatsoever during lines emitted by the source. The high application to analysis for traces of the three days it was on exhibition.

700 A • ANALYTICAL CHEMISTRY, VOL. 46, NO. 8, JULY 1974 However, when Dr. Menzies visited Melbourne shortly afterward to assess its performance, he was sufficiently impressed for his firm to decide to produce, under licence to CSIRO, the first commercial atomic absorption spectrophotometer. As soon as our final patent specifi­ cation was lodged on October 21, 1954 (3), I submitted to Spectrochimica Acta my first paper (4), in which I discussed the factors governing the relationship between atomic absorp­ tion and atomic concentration, and the experimental problems involved in making atomic absorption mea­ surements. The paper was published early in 1955, at about the same time as the paper by Alkemade and Milatz (5), who had independently arrived at the atomic absorption method. Nei­ ther paper created any great impact, and Alkemade and Milatz did not Figure 4. Photograph of atomic absorption spectrophotometer demonstrated pursue their work further, possibly at Institute of Physics Exhibition, Melbourne, March 1954 because they regarded this method merely as one for determining "all metals usually to be determined by flame photometry." In 1956 and 1957 we published pa­ pers (6, 7) describing results obtained with our instrument, but these papers also created little interest. John Shel- ton wrote to me from London on March 5th, 1956, after giving a lec­ ture on our work to the Institute of Physics, and reported that my first paper had given the impression that "the method was a scientific curiosity rather than a practical analytical method." When I gave a series of lectures on the subject at the Louisiana State University Symposium in 1958, the net result was that only one person, Jim Robinson, was stimulated into activity. He became a "hot gospeller" and in due course played an impor­ Figure 5. Photograph of simple atomic absorption spectrophotometer pro­ tant part in stirring up interest in the duced commercially in Australia United States. The surprising thing is that the ap­ pearance in 1958 of the papers by Allan (8) in New Zealand and David Australia. The necessary items were total staff of five. During the period (9) in Canberra did not arouse any manufactured by three small com­ 1958-62, some 30 Australian laborato­ sizable impact, even though they de­ panies in Melbourne and then assem­ ries were equipped by these "do it scribed eminently successful appli­ bled by the user, according to our in­ yourself" units. cations of the technique. structions. As it transpired, for the While knowledge of the technique On the commercial side, Hilger and next few years the members of our re­ spread rapidly throughout Australian Watts had produced an instrument search group were increasingly in­ industry, there was one memorable which did not incorporate a modu­ volved in supporting the commercial exception. I recall the technical direc­ lated source and therefore could not production in Australia of atomic ab­ tor of one of our biggest mining com­ fully exploit the technique. Other in­ sorption equipment. That a new type panies phoning CSIRO Head Office strument manufacturers subsequently of Australian industry was eventually in the early 1960's and stating that he perpetrated the same error. By 1958 created was, of course, cause for had just returned from South Africa there was no sign of any instrument much satisfaction, but it was inevita­ where they were using a brand new manufacturer willing to produce the ble that there was a substantial re­ instrument called an atomic absorp­ type of instrument which we thought duction in our research effort over a tion spectrophotometer. He wanted to desirable. This was most curious period of several years. know if there was anyone in CSIRO since by that time there was some in­ Figure 5 shows a typical instrument who knew anything about it. Our terest by other Australian laborato­ produced in this manner, the elec­ man in Head Office said he didn't ries in possible applications of this tronic units having been produced to know but he would make inquiries! technique. It was at this stage that our specifications by a firm in Mel­ In this period there was also a we decided to arrange for the produc­ bourne called Techtron Appliances slightly increased interest by manu­ tion of appropriate equipment in Pty. Ltd., which at that time had a facturers in other countries. For me,

702 A • ANALYTICAL CHEMISTRY, VOL. 46, NO. 8, JULY 1974 one of the "great moments" was in 1962 when I described to various members of staff of the Perkin-Elmer Corp. in Norwalk the impressive re­ sults which were being obtained by the laboratories in Australia which were by that time successfully using the technique. It was during these discussions that Chester Nimitz asked, rather tersely: "If this goddam tech­ nique is as useful as you say it is, why isn't it being used right here in the United States?" My reply, which my friends in Norwalk have never al­ lowed me to forget, was to the effect that he would have to face up to the fact that, in many ways, the United States was an underdeveloped coun­ try! The Perkin-Elmer decision to embark on a large-scale project relat­ ing to the production of atomic ab­ sorption equipment was made shortly afterward. They were, in fact, guilty of overreacting, as witnessed by their subsequent claim that atomic absorp­ tion was "the greatest invention since Figure 6. World sales the bed." of atomic absorption spectrophotometers It was also in 1962 that Techtron decided to market a complete spec­ trophotometer. Initially, this incorpo­ rated an imported monochromator, Australian production of an appropri­ ate monochromator being delayed od. The situation is particularly de­ means of extending the range of ap­ until 1965. By that time the ruling pressing in view of the well-known plication of atomic absorption meth­ engine constructed in the CSIRO Di­ limitations of existing flame methods, ods of analysis is by developing new vision of Chemical Physics was in full of which the most serious was de­ methods of atomization. operation and has since supplied all scribed as follows in our first paper In some respects I imagine the the master gratings required by Tech­ (3) describing our spectrophotometer: above remarks present a gloomy pic­ tron. Other firms also were becoming "By far the most serious difficulty ture of the present state of atomic ab­ increasingly interested, and from that in the absorption method is due to sorption spectroscopy. Fortunately, time onward I do not think the tech­ the difficulty in atomizing various el­ there is a happier way of assessing nique has suffered any major setback. ements. This problem of complete at- the situation, and I want to mention In 1965 its future was virtually se­ omization of the sample seems to us briefly some aspects which lead me to cured by the development (down- to be the outstanding problem at the think that this stagnation is more il­ under) of the nitrous-oxide flame present time." lusory than real. In the first place, so- which made the technique applicable As Allan (10) stated in his review of called flameless methods of atomiza­ to more than 65 elements. I might the subject in 1962, "It still is." tion, based on developments of the add that the development of this In a review (11) I presented in that L'vov furnance, have in the last two flame by workers in atomic absorp­ same year, I pointed out that when a years become of rapidly increasing tion spectroscopy was most altruistic, flame is used some elements are only importance, and their full potentiali­ since it put back on its feet emission partially atomized, "thus resulting in ties have by no means been fully ex­ flame photometry which, at that loss in sensitivity and the possibility ploited. They now provide a remark­ stage, appeared to be dying rapidly. of chemical interference due to varia­ able ability to analyse extremely The recent remarkable growth in tions of the degree of atomization of small amounts of material, and their the number of atomic absorption one element with the concentration of importance, particularly in biochemi­ spectrophotometers produced is other elements, radicals or com­ cal and clinical work, is already ap­ shown in Figure 6 and is indicative of pounds in the solution .... This type parent. The technique is still in its the increasingly widespread accep­ of interference is present to the same infancy, and we can expect it to de­ tance of atomic absorption methods. extent in emission and absorption velop rapidly over the next few years. In view of such growth, there would methods and is responsible for serious Of particular interest to me was the appear to be little justification for de­ limitations in flame methods." paper presented at the Toronto Con­ ference last year in which Segar and scribing the subject as having been Several years later, exactly the Gonzalez {13) described, I believe for stagnant at any stage during the past same statement was being made, as if the first time, the coupling of a gas decade. But the growth in the num­ it represented the disclosure of a new chromatograph to the graphite cu­ ber of applications has resulted main­ fundamental truth regarding flame vette of an atomic absorption spectro­ ly from fairly straightforward exten­ spectroscopy. sions of a technique which originated photometer. This combination may It is not my intention to decry the well make even more ubiquitous the more than 20 years ago. I believe it is use of a flame as an atomizer, since it no exaggeration to state that more remarkable techniques of chromatog­ is unlikely that any other method will raphy. than 99.99% of all analyses are still match its enormous range of applica­ carried out by that original tech­ tion, its speed of operation, or its re­ The recent development of im­ nique. In this respect the subject has markable convenience. Nevertheless, proved methods for the operation of been stagnant over an extended peri- I remain convinced that the best electrodeless lamps seems likely to

704 A • ANALYTICAL CHEMISTRY, VOL. 46, NO. 8, JULY 1974 produce a new interest in the flame fluorescence methods since, as was first pointed out by Alkemade (14) and demonstrated by Winefordner and MODULATED Vickers (15), they have striking po­ ATOMIC tential advantages over atomic ab­ sorption spectroscopy. In this respect, SPECTRAL it has always surprised me that work­ LAMP ers in flame fluorescence have, in general, failed to exploit the fact that the fluorescence phenomenon pro­ vides its own monochromator. In the arrangement shown in Figure 7, for example, if the illuminating source is "pure," the only need to have any wavelength selection is to avoid ex­ cessive noise owing to radiation from ATOMIC FLUORESCENCE the atomic vapour. I would also like to refer to our re­ cent work on the development of atomic absorption and atomic fluo­ rescence methods for the direct anal­ SYNCHRONOUS ysis of solids (16, 17). The sample is DETECTION made the cathode of a low-pressure SYSTEM discharge, and the atomic vapour is produced by cathodic sputtering. The encouraging results obtained in the analysis of low-alloy steels have re­ OUTPUT cently been published (17). In that SIGNAL paper we reported the difficulties en­ countered when the method was ap­ plied to the analysis of aluminum- Figure 7. Schematic diagram of nondispersive atomic flame fluorescence and zinc-base alloys. These difficul- spectrophotometer New automatic polarimeters make it easier to measure optically active substances with great accuracy.

Perkin-Elmer's new Models 241 and 241 MC Polarim- Four integration times: 1,5,20 and 50 sec. Plus a recorder eters could be just what you need if you're faced with readout unit with four cyclic ranges: ±0.05°, ±0.5°, one or more of the following measurement problems with ±5.0° and ±50.0°. optically active substances: Operation? Extremely simple and versatile. Just push • Quantitative determinations a button. You get an automatic gain and balance control, • Characterization of pure substances test deflection in + and - directions, automatic lamp • Study of kinetic reactions ignition. • End point determination of titrations Like to know more about the Models 241 and 241 MC? • Detection for liquid chromatography Write to the Instrument Division, Perkin-Elmer Corpora­ • ORD measurements tion, Main Avenue, Norwalk, Conn. 06856. These instruments work on the highly-valued principle of the optical null balance. You'll find them faster, surer and more accurate than other instruments. They give you the rotation on a five place digital display with sign. PERKIN-ELMER CIRCLE 185 ON READER SERVICE CARD 706 A • ANALYTICAL CHEMISTRY, VOL. 46, NO. 8, JULY 1974 ties have now been overcome, and it seems that the sputtering technique will be applicable to the analysis of a wide range of metals and alloys. It re­ mains to compare the performance and range of these methods with those of other analytical techniques. In the meantime, I would not ex­ pect the scientific instrument manu­ facturers to be greatly interested in the simple sputtering cell shown in Figure 8.1 would, however, like to think that some of them are musing on possible ways of embellishment to ensure that any commercial version will have an impressive price tag. We are now exploring the extension of our work to the vacuum ultraviolet for the determination of carbon, phosphorus, and sulfur. We also pro­ pose to extend our experiments to the analysis of powders and solutions. We are increasingly conscious of the im­ portant advantages of cathodic sput­ Figure 8. Photograph of atomic absorption spectrophotometer incorporating tering over thermal methods of atom- sputtering chamber for atomization of solid samples ization in isotopic analysis. Finally, I believe our recent work on cathodic sputtering takes us one step nearer the goal I discussed in my first paper on atomic absorption spectroscopy, the development of absolute methods of spectrochemical analysis. It would appear, therefore, that the Alan Walsh is assistant chief of the ble, and inexpensive hollow-cathode subject has not really been stagnant, Division of Chemical Physics, Com­ discharge lamps. He has published but merely pregnant, and has now monwealth Scientific and Industrial over 60 papers in atomic, infrared, given birth to new offspring on which Research Organization (CSIRO), and Raman spectroscopy. Dr. Walsh I trust Bunsen and Kirchhoff will Melbourne, Australia. He received is currently interested in what he look with approval and regard as wor­ his DSc from Manchester Univer­ considers the ultimate goal of spec- thy descendants of their original sity in England. He is considered trochemical analysis, the develop­ brainchild. the "father" of atomic absorption ment of absolute methods. This inter­ spectroscopy and exhibited a com­ est has resulted in the development of References plete apparatus for the technique in methods of atomization by using ca- March 1954. His now classic paper thodic sputtering; and these, in turn, (1) G. Dieke and H. M. Crosswhite, J. Opt. Soc. Amer., 42, 433 (1952). appeared in Spectrochimica Acta in have led to the development of reso­ (2) Australian Patent Application 1955. He holds basic patents in Aus­ nance detection and selective modu­ 23,041/53 (Nov. 17, 1953). tralia, the U.S., and other countries lation techniques. More recently, he (3) Australian Patent Specification on atomic absorption spectroscopy, and his colleagues have adapted these 163,586 (Oct. 21, 1954). (4) A. Walsh, Spectrochim. Acta, 7, 108 multiple monochromators, and im­ techniques to the development of (1955); Erratum, ibid., ρ 252. provements in grating monochroma­ atomic absorption and atomic fluo­ (5) C. T. J. Alkemade and J. M. W. Mi- tors. One of his major contributions rescence methods for the direct anal­ latz, Appl. Sci. Res., B4, 289 (1955). was the development of intense, sta- ysis of solid samples. (6) J. P. Shelton and A. Walsh, Proc. XVth Congress IUPAC, 2, TV-50, Lis­ bon, 1956. (7) B. J. Russell, J. P. Shelton, and A. Walsh, Spectrochim. Acta, 8, 317 (1957). (8) J. E. Allan, Analyst, 83, 433 (1958). (9) D.J.David, ibid., ρ 536. (10) J. E. Allan, Spectrochim. Acta, 18, 605 (1962). (11) A. Walsh, Proc. Xth Colloquium Spectroscopicum Internationale, ρ 127, Spartan Books, Washington, 1962. (12) B. V. L'vov, Spectrochim. Acta, 17, 761(1961). (13) D. A. Segar and J. G. Gonzalez, paper presented at Fourth International Conference on Atomic Spectroscopy, To­ ronto, Canada, 1973. (14) C. T. J. Alkemade, Proc. Xth Collo­ quium Spectroscopicum Internationale, ρ 143, Spartan Books, Washington, 1962. (15) J. D. Winefordner and T. J. Vickers, Anal. Chem., 36, 161 (1964). (16) Β. Μ. Gatehouse and A. Walsh, Spectrochim. Acta, 16, 602 (I960). (17) D. S. Gough, P. Hannaford, and A. Walsh, ibid., B28, 197 (1973).

708 A • ANALYTICAL CHEMISTRY, VOL. 46, NO. 8, JULY 1974