93 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1991 VOL. 6 Viewpoint Analytical Atomic Spectroscopy Learning From its Past* John B. Dawson* Department of Instrumentation and Analytical Science University of Manchester Institute of Science and Technology P. 0. Box 88 Manchester M60 I QD UK Progress in analytical science is driven by environmental need made possible by technological development and is based on progress in fundamental scientific un- derstanding which it in turn promotes by generating more accurate information. It is more ‘directed’ than the progress in basic science where chance observations and discoveries play a major part and ser- endipity is of the essence. Selected aspects of the development of analytical atomic spectroscopy with a view to iden- tifying its limitations and potential for further development will be discussed in this article.Methods of chemical analysis based on physical principles have in the past fre- quently had a long gestation period between the discovery of the physical phenomenon and the analytical applica- tion therefore if the present situation is to be understood some appreciation of the historical perspective is required. From such knowledge the reasons for the emergence of a particular technique at a given moment the timeliness can be identified. By comparing a number of related analytical procedures it might be possible to identify the fundamental factors that limit their performance and thereby either devise techniques to cir- cumvent the problems or abandon a hopeless pursuit! Arising from the author’s long association with the tech- nique particular reference will be made from time to time to developments in atomic absorption spectrometry (AAS).These examples will be presented with a view to determining how in specific in- stances accumulated knowledge can be applied to resolve current questions. While it might be possible to predict de- velopments in existing methods and such an exercise will be attempted later it is impossible to predict what new method will emerge in 10-20 years time after all who in 1940 could have pre- dicted the emergence of atomic absorp- tion spectrometry and X-ray fluorescence as major analytical tools in the mid- 1950s? * Dr. Dawson presented this review on the emergence of atomic spectroscopy as an Invited Lecture at the 5th Biennial National Atomic Spectroscopy Symposium Loughborough UK 18th-20th July 1990.Historical Perspective The development of physical methods of analysis has taken place over many centu- ries and like chemistry has been both helped and hindered by the alchemist and skilled artisan who through the ages has sought to develop his craft while conceal- ing his trade secrets behind a screen of ‘magic’. More complete accounts of the history of analytical atomic spectroscopy can be found elsewhere.1-5 For the purpose of this discourse only a limited selection of milestones in that history need be considered with a view to estab- lishing general patterns rather than a comprehensive record. There are three stages in the emergence of a new analytical method firstly the discovery of an effect secondly the in- vestigation of the effect leading to an ex- planation and finally exploitation by the analyst.Table 1 presents a summary of the history of several branches of analyti- cal atomic spectroscopy resolved into those three phases of development. Though individuals are credited with par- ticular achievements it is the date of the achievement that is probably most significant as the achievement itself might well have been presaged by the work of others and thus was an inevitable consequence of the knowledge techno- logy needs and culture of the society in which the individual lived. Conversely if all those four conditions are not fulfilled the development does not take place. These criteria for progress are graphically illustrated by the 2000 year long gestation periods of spectacles and of the spectro- scope compared with 150 years for AAS.In these examples the scientific and tech- nical expertise was accumulating during the gestation period. For spectacles the delay in exploiting a long established technology to provide a benefit to many people is somewhat surprising. The reasons for the delay can only be sur- mised but could include inadequate un- dertstanding of optical systems and cultural factors. The development of the spectroscope,6 however is an example of the positive interaction between scientific observation and technical development for though the technical skill required to construct it existed from at least 500 BC the instrument was not built until there was a demand for an improvement in the quality of the optical observations made using prisms.By facilitating precise measurements of spectra the spectro- scope promoted research into the new fields of spectrochemistry and atomic structure. The fundamental knowledge of the nature of the atom and the origin of atomic spectra so acquired made possible the development of the analytical atomic spectrometric techniques of today. On the other hand the emergence of AAS as an analytical tool was delayed by technical limitations and a lack of demand prior to the late 1940s. The independent inventor and entrepreneur might determine the precise moment when exploitation of a physical principle occurs but the general success will depend upon the timeliness. For an idea whose time has come dis- covery and exploitation are virtually assured while the converse is also almost certainly true. In general it appears that the more recent the scientific discovery the shorter the gestation period.This effect is illustrated by the interval between discovery and the analytical ex- ploitation of X-rays in X-ray fluorescence analysis (1 895-1955) artificial radioac- tivity in neutron activation analysis (1934-1955) and the Mossbauer effect for studies of iron in haemoglobin (1957- 1 960). In the early applications of analytical atomic spectroscopy visual detection must have played an essential part e.g. the characteristic colour of a flame is readily recognized. It is therefore reason- able to assume that from earliest times (3000 BC) the colour of the ‘fumes’ (flames) as described by Agricola in 1550 has been used to control the smelt- ing of ores in a primitive application of atomic emission spectroscopy (AES).In the early nineteenth century Wollaston (1802) saw the dark lines in the sun’s spectrum and Talbot (1826) and Wheat- stone ( 1833 respectively visually ob- served flame and spark spectra as being characteristic of specific substances. The next step quantitative analysis was difficult to make using the eye as the measure of light intensities. It also re- quired the development of a sound theo- retical basis of spectrum analysis by Kirchhoff and Bunsen (1860) the photo- graphic recording of spectra (Herschel,94 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1991 VOL. 6 ~ ~~ Table 1 mation from references 1-5 Emergence of atomic spectroscopy as an established analytical technique.Based on infor- Technique Discovery Explanat ion Exploitation Lenses mirrors Ancient artefacts Euclid 300 BC Spectacles prisms 10oO BC Ptolemy 100 AD Amarti 1285 Snell 1621 AD Spectra Colours of rainbow/ Nature of white light Spectroscope prism Seneca 40 AD Newton 1672 Sims 1830 Emission Flame Coloured ‘fumes’ Agricola 1550 Melvill 1752 Salted flames Flame photometer Lundegardh 1930 Electrical Arc discharge Atomic origin of Quantitative analysis Davy 1802 emission/absorption Lockyer 1873 Spark spectra lines Internal standard Wheatstone 1835 Gerlach 1925 Electrodeless K i rc h ho ff/ ICP discharges Bunsen 1860 Greenfield et al. 1964 Babat 1947 I I Absorption Dark lines in sun’s spectrum Wollaston 1802 - Flame atomic absorption spectrometry Walsh 1953 Fluorescence Sodium in Molecular Flame atomic vapour cell fluorescence fluorescence Wood 1905 Stokes 1852 Winefordner 1964 1840) and before it could be fully ex- ploited the development of standardiza- tion procedures which culminated in the enunciation of the internal standard prin- ciple by Gerlach (1925).The replacement of visual detection with photoelectric devices for the measurement of spectral line intensities led to the manufacture in the late 1930s of direct reading spec- trometers using arcs sparks or flames for atomization and excitation and thus rapid and accurate quantitative analysis became possible a century after the pioneering work of Talbot and Wheatstone. This achievement was also a direct response to the industrial need for improved quality control and increasing recognition of the importance of minor and trace elements in many fields.Thus the coincidence of a sound scientific basis with technical de- velopment and ‘customer’ demand trig- gered the wider exploitation of AES that we practise today. The historical examples presented above serve to illustrate the vital roles of timeliness and fundamental understand- ing in the emergence of a new analytical technique. A more detailed examination of factors contributing to the timeliness of the exploitation of atomic absorption and atomic fluorescence spectrometry (AFS) will be presented in the next section and later an attempt will be made to identify some of the fundamental factors that limit the performance of analytical atomic spectrometric techniques.Tim el i n ess One of the most interesting examples of timeliness in recent developments in ana- lytical science is found in the period 1945-1965. At the beginning spectro- chemical analysis was a limited activity largely in the hands of the professional spectroscopist commonly a physicist though some indication of future trends could be discerned in the increasing use of flame photometry and other forms of instrumental analysis by the analytical chemist. By the end of that period AAS was established in the analytical labora- tory the merits of the inductively coupled plasma (ICP) as an emission source had been demonstrated and the potential of AFS as an analytical technique examined. The varying fortunes of these three tech- niques (i.e.AAS ICP-AES and AFS) over the years clearly demonstrate the im- portance of timeliness and might serve to identify the key factors that determine the rate of progress in the exploitation of a technique. The phenomena of atomic emission absorption and fluorescence were recog- nized and understood for many years but only emission was utilized in the analyti- cal laboratory. Earlier in this article it was proposed that lack of demand and technical limitation delayed the analytical exploitation of atomic absorption. The factors favouring the emergence of AAS as a major analytical development of the late 1950s will now be considered. They fall into two categories first general ex- ternal factors that could have applied equally to other techniques e.,q.atomic fluorescence and second the specific ad- vantages of atomic absorption. By the early 1950s all the scientific and technical knowledge and equipment necessary to exploit atomic absorption existed. In particular it was an extension of two actively developing fields absorp- tion spectrometry and flame photometry. One of the successes of those fields was to demonstrate the importance of minor and trace elements in a wide range of cir- cumstances ranging from metallurgy to clinical chemistry. This success in turn created an increased demand for such de- terminations which AAS was well suited to meet all that was lacking was the in- ventor! In 1953 Australian patent No. 23141/1953 was granted to the Common- wealth Scientific and Industrial Research Organization (CSIRO) for the application of atomic absorption to chemical analysis with A.Walsh as the inventor; his definitive paper7 on the subject was pub- lished in 1955. The environment in which the initial development of AAS took place played an important part in its dis- semination. As a government funded re- search organization it was part of the remit of CSIRO to support Australian in- dustry therefore the manufacture and use of atomic absorption equipment was ac- tively promoted both in Australia particu- larly in the mining industry and world- wide. It is doubtful whether any previous analytical technique has enjoyed such a positive launch even so the growth until 1963 was relatively slow.8 For a new ana- lytical method to be successful it should have demonstrable advantages over its predecessors.Many of the advantages attributable to AAS are also requisite for any improved analytical method. The technique combines great sensitivity with specificity by virtue of being based on resonant transitions of a large population of ground state atoms. It is applicable to most elements and samples and in most laboratories and can provide rapid analy- sis with simple equipment. As the tech- nique is particularly suited to processing solutions the necessary sample prepara- tion procedures were established labora- tory practice long before AAS came along. However probably the single most important technical factor contributing di- rectly to its success was the choice of the hollow cathode lamp as the light source their manufacture as sealed lamps in place of the usually continually pumped sources and their operation in an a.c.mode which reduced the effect of back- ground emission. Following the above review of the cir- cumstances leading to the success of AAS it is instructive to examine the for- tunes of another technique AFS,’ that emerged almost ten years later. Many of the factors favouring AAS also apply to AFS and were present in the early 1950s. Atomic fluorescence is however more difficult to generate and detect than either emission or absorption as is illustrated by the fact that it was not observed experi- mentally by Wood (1905) until almost a century after the other two even though the effect had been expected and sought for many years.Futher its exploitation asJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 199 1 VOL. 6 95 an analytical method did not occur as a natural extension of an existing established method of the 1950s e.g. fluorimetry but rather it needed the estab- lishment of AAS to draw attention to the possibilities of AFS. The analytical ad- vantages of atomic fluorescence over ab- sorption include the opportunity to use non-resonant atomic transitions the feasi- bility of multi-element analysis and under favourable circumstances lower detection limits. However these advan- tages have not been sufficient to prompt many researchers to attempt to overcome the technical limitations of the method which include the absence of a univer- sally applicable light source matrix inter- ference by scattering of the exciting radiation and by the quenching of fluorescence and the absence of generally available inexpensive commercial instru- mentation.Furthermore ICP-AES which enjoys many of the advantages attributed above to AFS plus the additional one of a high temperature for sample dissociation was first reported at about the same time (1964) and thus diluted the impact of AFS on the analytical community. Another consideration militating against expan- sion in the use of AFS is the fact that as the light source and sample atomizer are key components of both atomic absorp- tion and fluorescence systems it is to be expected that improvements designed to improve performance for one technique will also benefit the other. Hence it is un- likely that the early lead established by AAS will ever be overtaken by AFS as a general purpose analytical method.The interaction of AAS with the ex- ploitation of the ICP is an interesting one. With the development of the atmospheric pressure electrodeless discharge culmi- nating with Reed's publication on the ICP in 1961,IO the stage was set for further de- velopments in AES. This occurred almost instantly and in 1964 the first paper by Greenfield et al." appeared but it was a further ten years before the number of active laboratories had reached double figures.'? The delay in the widespread ac- ceptance of the technique must be due in part at least to the preoccupation of the spectrochemical community with the ex- ploitation of AAS. However the success of AAS in determining elements and in educating the analytical chemist in the ways of atomic spectrometry along with the recognition that AAS can suffer from chemical interferences that it is primarily a single element technique and that the instruments were becoming increasingly complex and expensive all served to create a climate in which ICP spectro- metry became an attractive proposition.According to Fassel,I2 between 1973 and 1978 'the number of laboratories engaged in analytical investigations and routine applications increased from 10 to about 200 and the number of commercial suppliers of instruments from none to at least nine'. The current usage of the ICP is now widespread and it in turn has spawned a new technique inductively coupled plasma mass spectrometry (ICP- MS).I3 The nature of developments in analyti- cal atomic spectrometry is very much a consequence of the timeliness of ideas and of their interactions. Mutations competi- tive pressures and the environment all play their part in the evolution of an analytical technique just as in the biological world. Fundamental Limitations The fundamental limitations of an analyt- ical method are frequently manifested in the sensitivity and baseline stability.Both these factors are included when the detec- tion limit of a method is calculated. This composite parameter is one of the most useful for comparing the performance of one analytical method with that of another and in an attempt to 'learn from the past' can also be used as a basis for measuring progress and for identifying 'dead ends' in analytical development.Relative Detection Limits in Emission and Absorption Atomic Spectrometry AlkemadeI4 examined theoretically the relationship between sensitivity in AES and AAS at concentration levels ap- proaching the detection limit. He showed that based on the Kirchhoff law the ratio of the amount of radiation absorbed to that emitted by an atomic vapour when ir- radiated by a sharp line source was pro- portional to the ratio of the spectral radiance of a black body at the effective temperature of the radiation source to that of a body at the temperature of the atomic vapour. Based on this premise the theore- tical wavelength dependence of the ratio of the detection limits in absorption to those in emission can be deduced.15 The limiting background noise was attributed to the shot noise of the light source ato- mizing flame and the dark current of the photomultiplier.The theoretical curve in Fig. 1 was calculated on the assumptions that the effective temperature of the illu- minating source for atomic absorption was 6OOO K and that of the atomic vapour 2500 K. The alignment of the experimen- tal data points and the theoretical curve was achieved by the use of a logarithmic scale for the detection limit ratio. Vertical transposition of the theoretical curve to coincide with the data points was equiva- lent to the use of a constant multiplying factor originating in atomic constants the geometry of the apparatus and detector amplification. The general agreement between the theoretical curve and experi- mental results justifies the conclusion that Alkemade's proposed fundamental rela- tionship between emission and absorption signals is valid.Thus at wavelengths greater than 400 nm when an air- acetylene flame is used detection limits can be lower in emission than in absorp- tion. However for AES to achieve detec- tion limits comparable to AAS over the full spectral range the excitation temper- ature of the atomic vapour must be at least 6000 K. When such temperatures are present as in plasmas and provided the spectrometer has adequate resolution atomic emission is the preferred method. Thus from a knowledge of the funda- mental limitations decisions affecting practical analysis can be made. Relative Sensitivities in Flame and Electrothermal Atomic Absorption Spectrometry As absolute sensitivity in analytical atomic spectroscopy is proportional to the number of atoms contributing to the 200 300 400 500 600 700 8oo Wavelengthlnm Fig.1 Theoretical and experimental wavelength dependence of the ratios of the detection limits ob- tained by atomic absorption spectrometry to those obtained by flame atomic emission spectrometry. The theoretical curve is shown as a full line and the experimental points as individual elements with standard error bars96 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1991 VOL. 6 analytical signal at a given instant the generation of the densest possible atomic vapour up to the point at which self- absorption occurs is generally desirable. Transient response systems are frequently reported in order to provide greater ana- lytical sensitivity.The reason for this im- provement can be clearly demonstrated by comparison of estimates of the instan- taneous amount of analyte in the asborp- tion path of an atomizing flame and a graphite tube furnace. If it is assumed that all the atomized sample passes through the observation zone of the flame then the instantaneous absorption signal A is given by wheref= sample flow-rate to the nebulizer approximately 5 ml min-1; e = efficiency of sample transport and atomization about 10%; a = width and height of absorption zone about 0.5 cm; u = flame speed approximately 200 cm s-I; c’ = analyte concentration in g ml-1; K = atomic ab- sorption coefficient per g of analyte. For the furnace the maximum signal A2 is obtained when all the analyte atoms are contained in the furnace at the same instant and can be calcualted from where 1’ = sample volume 20 p1; and Y = internal radius of the furnace 0.25 cm.The theoretical ratio R of the peak ab- sorbance signal of a furnace to the contin- uous signal from a flame atomizer for the same solution is given by 0.1 Kc = 1200 (3) A2 A1 8 . 3 x IO-’Kc R = - = Table 2 presents experimental values of the characteristic concentrations for 33 elements taken from published reports. Over the period 1970-1983 there was no significant changes in the sensitivities re- ported for flame AAS. On the other hand sensitivities in electrothermal AAS in- creased on average 5-fold between 1976 [Column (B)] and 1984 [Column ( C ) ] . This improvement was a result of the use of faster response electronics rapid furnace heating gas stop during atomiza- tion the L’vov platform and chemical modifiers.The 1984 data were used for calculating the sensitivity ratios. Seventy- five per cent. of the ratios are evenly spread between values of 100 and 600 with a mean of 360 the remainder are widely spread between 960 and 2400. The predicted ratio of equation (3) is Table 2 Comparison of sensitivities of flame and electrothermal atomic absorption spectrometry. Sources of data (A) Fuller,1h Welz” and Varian Techtron;lH (B) Fuller;I6 and (C) Grosser,IY SIavin et u/.~‘’.~’ and Sperling.” Where literature values were expressed as A s peak absorbance was assumed to have the same numerical value Element Ag Al As AU Ba Be Bi Ca Cd c o Cr c u Fe Li Mn Mo Ni Pb Pd Pt Rb Sb Se Si Sn Sr Te Ti T1 V Zn Hg Mi? Excitation wavelength 328.I 309.3 193.7 242.8 553.6 234.9 223.1 422.7 228.8 240.7 357.9 324.7 248.3 253.7 670.8 285.2 279.5 3 13.3 232.0 283.3 247.6 265.9 780.0 217.6 196.0 25 1.6 286.4 460.7 214.3 365.3 276.8 3 18.4 2 13.7 Flame/ mg 1-1 per 0.0044 A (A) 0.033 0.88 0.77 0.22 0.33 0.023 0.36 0.047 0.020 0.10 0.065 0.049 0.065 1.6 0.026 0.0044 0.036 0.54 0.065 0.29 0.15 1.3 0.043 0.50 0.58 I .so 0.90 0.09 I 0.36 I .9 0.3 1 1.3 0.0 12 ETA/ pg I-’ per 0.0044 A* (B) 0.25 2.5 1.25 I .o 7.5 0.10 2.0 0.20 0.05 2.0 I .o 1 .5 1.25 10.0 0.50 0.0 1 0. I0 1 .o 5 .o 1 .o 1 .o 25.0 I .o ! .O 10.0 2.5 5 .o I .o 5.0 25.0 2.5 10.0 0.05 (C) 0.080 0.63 0.80 0.60 0.33 0.04 1 .o 0.040 0.043 0.38 0.18 0.34 0.25 9.0 0.070 0.0 I6 0.1 1 0.45 0.64 0.54 1.15 5.3 0.12 1.9 I .4 2.6 0.98 0.07 0.75 3.7 0.73 2.2 0.005 Improvement ratio (ETA flame) (A:C) 410 1400 960 370 lo00 580 360 I200 470 260 360 I40 260 180 370 280 330 1200 100 540 I30 250 360 260 410 580 920 1300 480 5 10 430 5 90 2400 *Derived from estimated peak absorbance and an assumed sample volume of 20 PI.1200. A major factor in the discrepancy between theory and practice is the as- sumption in the theoretical model that in the furnace all the sample atoms are in the absorption path simultaneously. This cannot be so as atoms begin to leave the furnace with a characteristic time con- stant of the order of 0.1 s as soon as atomization takes place. It follows from the simple model of the electrothermal atomization process developed by Fuller,Ih that even if the rate constant of free atom formation is twice that of atom loss the maximum number of sample atoms in the vapour phase in the furnace at any instant will be only half of the maximum assumed in equation (2).If the rate constant of generation is half that of the loss process the peak height is then but a quarter of the theoretical maximum. Thus the effect of atom loss from the furnace could account for much of the difference between predicted and ob- served sensitivity ratios. The higher values of some of the ratios e.,?. Al As Ba Ca Mo Sn and Sr might be more a result of inefficient atom production in the flame than efficient electrothermal atomization! The ratio for Zn appears to be anomalously high for which there is no obvious explanation other than that it is the element with the highest sensitivity determined by furnace AAS.’(’ The above results clearly demonstrate the gain in sensitivity resulting from car- rying out the analytical measurement on a transient confined atomic vapour.Other systems showing varying degrees of improved sensitivity and incorporating some or all of the above principles include the Delves’ cup furnace non- thermal atomic emission spectrometry glow discharge lamps slotted tube atom retarder discrete nebulization and abla- tion techniques. On the basis of the simple models and experience to date it appears unlikely that substantial improve- ment in the production of analytical atomic vapours can be achieved. However for individual elements there appears to be improvement factors of up to 3 or 4 available by the optimization of either the flame or furnace operating con- ditions.If this view is correct then orders of magnitude improvements in an- alytical performance will have to be sought from improved detection of the analytical signal or the generation of aJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1991. VOL. 6 97 different type of analytical signal e.g. ions rather than photons. Signal Processing The noise or instability of a signal deter- mines the precision of an analytical meas- urement and the smallest amount of the analyte that can be detected with a specified probability. At analyte levels above the detection limit (approximately lox) the reproducibility of the measure- ment is determined by fluctuations in the production of the atomic vapour in the excitation of emission either in the atomic vapour or in the radiation source for ab- sorption and in the radiation detection and signal processing electronics.At the detection limit the most impor- tant factor is the stability of the back- ground (baseline) signal. Signal noise is of two kinds ‘white’ (random or stochas- tic) noise when all frequencies are equally present in the noise power spec- trum and ‘coloured’ (flicker or structured) noise when some frequencies are present at a greater power level than others. Some improvement in analytical performance can be achieved by signal processing de- signed to reduce the effects of noise. Signal modulation is widely used to dis- criminate against the mean level of the background and by means of a suitable integration time to reduce the effect of high frequency noise.It is commonly acheived by ‘chopping’ the optical beam. Modulation of the atom vapour produc- tion is rare. However pulse atomization can be exploited as a signal modulation technique by matching the response of the detection system to the temporal profile of the atomization peak. It has been reported2j that by integrating the output signal of the instrument either as a linear running mean or as a time constant with a sampling interval equal to approxi- mately one half of the full width of the signal peak at half the maximum value the precision and detection limits could be improved by factors of 2-3. Correla- tion with a signal processing function identical to the atomization signal peak did not lead to detectably better perfor- mance than the optimized linear running mean.When the noise on the analytical signal is not random and when one component of the analytical system is identified as the major source of analytical imprecision modification of the instrument or proce- dure can be used to reduce the standard de- viation of the results. The introduction of the internal standard principle by Gerlach in 1925 reduced the effects of variability of atomization and excitation in AES. In the early days of AAS a two-channel instru- ment described by Menzies in I958,’-‘ was designed to overcome the effect of the in- stability of the hollow cathode lamps in use at that time. Both these techniques for reducing the effects of system instability employed pairs of spectral lines conse- quently both are limited by the differences in behaviour of the lines when excitation conditions in the source change.Other methods of overcoming the effects of system instability depend on time sharing the signal e.,?. double-beam AAS. In these circumstances noise reduction can be achieved only for changes that occur more slowly than the modulation frequen- cy. In all instances the added complexity of the instrumentation inevitably makes its own contribution to the signal noise. It appears therefore that in the light of general experience to date further attempts to reduce system noise are unlikely to lead to the general lowering of detection limits in conventional emission and absorption analysis.However there may be special circumstances where it is possible to gen- erate highly correlated reference and ana- lytical signals. For example it may be feasible to utilize the properties of polar- ized light and Zeeman effect in the atomic vapour to produce a reference signal corre- lated with the analytical signal in wave- length time and space. Such an approach could be developed based on atomic magneto-optical rotation (AMOR). The theoretical and experimental basis for this technique is established’5 and therefore theoretical predictions on the performance of the system are possible nevertheless it is only by carrying out the experiment that the true potential and limitations of the concept will be revealed. Over the years many studies of noise in analytical atomic spectroscopy have been carried out and though there is now a general understanding of the origin and effect of noise on analytical systems that understanding has not led to substantially improved performance.This somewhat disappointing conclusion stems from the predominantly random nature of the noise in the analytical system whose effect can only be overcome by repeated measure- ment and/or increasing the observation time. Prospects for Future Development in Analytical Atomic Spectroscopy Analytical atomic spectroscopy has a long and distinguished lineage going back several millenia. From history it is clear that the timeliness of innovation is all im- portant and requires the confluence of scienti tic knowledge technical ski 11 env i- ronmental demand and entrepreneurial initiative. Once established knowledge of the fundamental characteristics of a tech- nique is sought so that the performance can be optimized and limitations defined. New techniques emerge either by the application of a previously unexploited physical principle or by extension or hy- bridization of established procedures. It is impossible to predict when the former will occur but for it to occur at all it requires that the worker in analytical science should have a profound knowledge of fun- damental physics and chemistry.The extension of existing practice is more predictable as is illustrated by Alke- made’s prophesy of ‘flame ionic mass spectroscopy’ in 1973j and its realization albeit in a different form by Gray,I3 as ICP-MS in 1978.In this instance by changing the detection system from one for photons to one for ions orders of magnitude improvement in detection limit were achieved. The combination of chromatography with spectroscopy to meet the demand for speciation was in- evitable as was the incorporation of auto- mation and computers into analytical instruments to promote rapid accurate analysis and reduce the dependence on the operator. The potential applications of both these developments is far from fully exploited. Other developments equally predictable but likely to be less successful will be the continued elaboration of ana- lytical systems which are of little rele- vance to the general analytical laboratory. This is because they are too complex and expensive offer little advance over what can be acheived by a combination of stan- dard instrumentation with appropriate preparative chemistry and not infrequent- ly have difficulty in processing real samples! Also there will be continued efforts to exploit fundamentally sound and attractive ideas that are technically unlikely to be realizable in an acceptable manner e.g.simultaneous multi-element atomic absorption. For the more distant future one of the most exciting prospects must be the pos- sibility of using solid-state tunable lasers as high intensity light sources over a wide spectral range.2h How soon will such sources be available? What will they cost? How stable will they be? Can they be used in such a way that the limiting factor in the analysis is the behaviour of the atomic vapour alone independent of the generating process? As the answers to these questions emerge over the next 5-1 0 years perhaps some new technique will also appear and be competing for a place in the analytical laboratory. Who can say? Whatever the techniques of the future the analytical specification for improved methods will include accuracies of the order of 0.1% detection limits in the sub-ng I-’ or fg mass range generation of information on the state or species of the analyte element and rapid automatic op- eration. The achievement of these goals requires an environment in which analyti- cal scientists can deploy some time and resource away from the regular demands of a service load and engage in research and development of existing and new methods. Today however the very success of the present analytical atomic98 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1991 VOL.6 spectroscopic techniques might for a time inhibit the search for the new methods of tomorrow because they cur- rently generally fulfil and indeed in many instances are ahead of the demands placed on them. However if historical precedent can be relied upon when the need for a new method arises then Micawber-like ‘something will turn up’ and though it is unrecognized today it is likely to be based on an already predicted or observed phenome- non! The author thanks Dr. W. J. Price for helpful comments on this manuscript. References EnLyclopedia of Spectroscopy eds. Weise E. K. and Clark G. L. Reinhold. New York 1960 pp. 188-199. Hermann R. and Alkemade C.Th. J.. Flame Photometrv. Wilev-Interscience. New 3 4 5 6 7 8 9 10 1 1 12 13 14 15 16 York 1963. Alkemade C. Th. J. Proc. Soc. Anal. Chem. 1973,10 130. Thorburn Bums D. Proc. Anal. Dil.. Chem. Soc. 1975 12 155. West T. S. Proc.. Anal. Div. Chem. SOL.. 1977. 14 177. Thorburn Bums D. J. Aiiai. At. Spectrom. 1988,3,285. Walsh A. Spectrochim. Acta. 1955 7 108. Walsh A. Appl. Opt. 1968,7 1259. Winefordner J. D. and Vickers T. J. Anal. Chem. 1964,36 161. Reed. T. B. J . Appl. Phjs. 1961,32,821. Greenfield S . Jones I. Ll. and Berry C. T. Analyst 1964,89,7 13. Fassel V. A. Science 1978,202 183. Gray A. L. J. Anal. At. Spectrom. 1986 1 403. Alkemade C. Th. J. Appl. Opt. 1968 7 1261. Dawson J. B. Proceedings of the XVI Col- loquium Spectroscopicurn Internationale Heidelberg October 4th-9th. 1971 I Adam Hilger London 197 1 pp. 347-35 1 . Fuller C. W. Electrothermal Atomization for Atomic Absorption Spectrometry Chemi- 17 18 19 20 21 22 23 24 25 26 cal Society London 1977. Welz B. Atomic Absorption Spectrometry 2nd edn. VCH Weinheim 1985. Varian Techtron Hollonr Cathode Lamp Data Varian Zug 1970. Techniques in Graphite Furnace Atomic Ab- sorption Spec.trophotometr:v ed. Grosser Z. Perkin Elmer Ridgefield 1985. Slavin W. Carnrick G. R. Manning D. C. and Pruszkowska E. At. Spectrosc. 1983 4 69. Slavin W. and Camrick G. R.. Spectra- chim. Acta Part B 1984,39,27 I . Sperling K. R. Spectrochim. Acta Part B 1984,39,37 1 . Dawson J. B. Duffield R. J. King P. R. Hajizadeh-Saffar M. and Fisher G. W. Specwochim. Acta. Part B . 1988. 43 1133. Menzies A. C. Actas Congi,. XV Int. Quim. Para ApI.. Lisbon. 1958,2,2. Stephens R. J. Anal. At. Spemom. 1988 3 227. Hergenroeder R. and Niemax K. TrAC. Trends in Anal. Chem. (Pet-s. Ed.) 1989 8 333.