FEBRUARY, 1966 THE ANALYST Vol. 91, No. 1079 Some Sensitive and Selective Reactions in Inorganic Spectroscopic Analysis* BY T. S. WEST (Chemistry Department, Imperial College, London, S. W.7) THIS paper reviews some recent work in inorganic trace analysis by the author and his colleagues within the area of spectroscopic analysis. The work is concerned with analytical applications of the phenomena of absorption and emission of light by atoms or molecules in solution or flame media. The subject matter is, therefore, conveniently subdivided under four main headings, uiz.- Absorption Spectrophotometry in Solution (Absorptiometry). Absorption Spectrophotometry in Flames (Atomic-absorption Spectroscopy). Emission Spectrophotometry in Solution (Spectrofluorimetry) . Emission Spectrophotometry in Flames (Atomic-fluorescence Spectroscopy).In this paper, the philosophical aspects of each topic are stressed in relation to the achievement of maximum sensitivity and selectivity of determination, and examples are quoted from hitherto unpublished or recently published work. The main theme running through this work is the analytical application of chelation or co-ordination reactions. ABSORPTION TECHNIQUES SPECTROPHOTOMETRY- There are two main avenues worthy of exploration in search of selectivity of reaction for this technique. These are the design of highly selective reagents, and the use of mask- ing agents to achieve selectivity without recourse to separation. Most attempts to obtain selectivity have been directed towards the introduction of steric hindrance into reagent moleculcs close to the chelating centre.Some of the most commonly cited examples of selectivity do not, however, stand up to closer examination. For example, Merritt and Walker’s1 8-hydroxyquinaldine reagent is said to be unreactive towards aluminium, whilst retaining its activity towards other cations that react with 8-hydroxyquinoline. This is certainly true as far as non-precipitation of aluminium is concerned and the absence of any visible reaction] but we have found that, when attempts are made to extract other ions away from aluminium at pH 5.8 or above, significant amounts of aluminium are also extracted. The presence of the extracted complex may also be demonstrated in the chloroform extract by virtue of its fluorescence.2 Similarly, we have found that, although the literature describes neo-cuproine as a specific reagent for copper(I), it is only so within the context of spectrophoto- metry; many other ions, e.g., cadmium, cobalt, nickel and zinc, extract as colourless com- plexes.3 These examples illustrate that many selective reactions are only conditionally so, and that it is not always possible to transpose selectivity of reaction from one technique to another.* Presented at a meeting of the Society on Wednesday, October 6th, 1965. 6970 WEST: SOME SENSITIVE AND SELECTIVE [Analyst, Vol. 91 The principle of the clathrate or inclusion compound is now a familiar one. We have attempted to use this principle to construct a chelate cage molecule into which only ions of a certain size would fit. For example, the compound cycZo-tris-7-( 1-azo-8-hydroxynaphtha- lene-3,6-disulphonic acid) or Calcichrome, (I), at pH 12 reacts only with calcium (0.9 L&) and (1) not with strontium (1.1 A) or barium (1.3 A).4 Other ions that would normally be expected to fit into the cage are insoluble at pH 12 or are bound anionically in unreactive forms.This chromogenic reaction is, therefore, conditionally selective for calcium, and may be used to determine this ion spectrophotometrically in the presence of several thousandfold excesses of barium and strontium.6 The value of the molecular extinction coefficient is 7600, which compares with about 10,000 for murexide. The colour is, however, very stable and the selectivity is much greater. The compound Acid alizarin black SN, (11), resembles an opened out Calcichrome molecule and also possesses a good reactivity towards calcium, but its reaction with thorium, uranium and vanadium is of more interest.6 In the presence of masking agents such as cyanide, salicylate and 1,2-dihydroxybenzene-5-sulphonic acid at pH 4 to 4.2, thorium forms a 1 to 2 complex with Acid alizarin black SN,7 and gives a molecular extinction coefficient of 28,000 with only uranium and vanadium interfering.It is possible to determine all three metals sequentially by adding selective masking agents. Thus an excess of acetate masks thorium and uranium, whilst an excess of nitriloacetic acid masks thorium and vanadium. Both vanadium(v) and (IV) form complexes, of which the latter offers the better sensitivity.The molecular extinction coefficients for uranium and vanadium(1v) are 17,000 and 22,000. It occurred to us, however, that a better analytical sensitivity for thorium would be obtained by furnishing this molecule with additional hydroxyl groups so that it could form a 1 to 1 complex.6 This was done in the molecule (111), which behaves as expected and forms a (111) 1 to 1 complex with an extinction coefficient of 50,000 for thorium, 25,000 for uranium(v1) and 41,000 for vanadium(1v). This range of sensitivities compares favourably with the most sensitive reagents for these metals, e.g., morin ( E h 41,000) for thorium, dibenzoyl- meth;me (E fi 18,000) for uranium and diphenylbenzidene (E 23,000) for vanadium.February, 19661 REACTIONS IN INORGANIC SPECTROSCOPIC ANALYSIS 71 Some of the reagents that react with a great number of cations and can, therefore, be used as mass masking agents, have some gaps in their coverage.Thus EDTA may be said to be selectively unreactive towards silver, niobium, antimony and a very few other metals. We have found that the reagent bromopyrogallol red forms a particularly intense gelatin- sensitised colour reaction with niobium in a tartrate medium at pH 5 ~ 8 . ~ The 3 to 1 bromo- pyrogallol red - niobium complex has a molecular extinction coefficient of 60,000, which compares favourably with the standard thiocyanate method for niobium ( E = 35,000). In the presence of EDTA a 2 to 1 complex is formed, but the value of E falls to only 53,000. In this medium only tungsten, molybdenum, titanium and antimony interfere, and when a large excess of tartrate is added, the molecular extinction coefficient falls to only 47,500 whilst these ions no longer interfere.This sensitive and reproducible reagent system can be modified to allow the determination of niobium in steel down to 0.001 per cent. without recourse to the usual hydrolytic separation process for ni~bium.~ The reagent may also be applied to the determination of antimony(m), with which bromopyrogallol red forms a 1 to 1 complex in the presence of EDTA, cyanide and fluoride ions to give a molecular extinction coefficient of 39,000. The procedure appears to be superior in many respects to the standard Rhodamine B method ( E = 34,000) and is easier to operate.1° The sensitivity of absorbance measurements in solution is limited by two factors, both of which suggest that there is a lower limit of detection of about lo-* M for any complex in aqueous solution.The first factor is the inability of even a good spectrophotometer to measure to less than 0.001 absorbance units. Absorbance is defined as the logarithmic ratio of the intensity of the incident radiation (Io) to the intensity of the transmitted radiation (It). For trace amounts I , fi It and log I = 0. The second limiting factor is that molecules have a limited capacity to absorb light. Before light is absorbed, an electronic transition must occur within the molecule. The probability of such a transition is limited by several con- siderations that cannot be discussed in this paper, but any molecule can be regarded as having a light-capture cross-section. One method of increasing this is to spread a mesh of closely packed 7r orbitals in the molecule to capture photons and secure a transition.Braudell has discussed this extensively, and has shown that the maximum molecular extinction coefficient for any organic molecule is likely to be of the order of 100,000. It is, of course, possible to synthesise complex organic ligands possessing extensively conjugated-bond systems, but the task is a difficult one and the capability of such molecules to react efficiently with a metal cation is frequently rather small because of steric effects. For this reason we have paid very serious attention to the idea of utilising the formation of ternary complexes in which the cation reacts, not with one ligand species only, but rather with two.In this way it is possible for a much more complex absorbing organic envelope to be put around an ion than is normally possible, and consequently the sensitivity of such ternary systems is likely to be considerably superior. What is perhaps even more important is that the selectivity of ternary-complex formation is likely to be much superior to that of binary-complex formation. If we have a series of divalent metals, lM, 2rul, 3M, of similar chemical habit, they are likely to react to form complexes with a ligand, H2L of thenature of lML, 2ML, 3ML, etc., or lML2,-, 2MLi-, etc. As a generalisation, it is often permissible to say that the absorption characteristics of most metal complexes of a reagent tend to resemble closely the next higher ionisation stage of the ligand molecule considered in its reactive form as an acid.Hence, if a reagent reacts in the form HL-, its metal complexes frequently tend to have absorption spectra closely resembling that of L2-. Consequently, most metals form similarly coloured complexes so that selectivity is low. However, when two ligands are involved, e.g., H,L and H2R, the chances of duplication of ternary complexes of the nature M.L.R are much smaller, and this makes for selectivity of reaction. There are two chief routes to the formation of ternary complexes. In the first, before a metal ion can form a ternary complex of analytical utility, the situation must arise that one ligand does not fully satisfy all the co-ordinative requirements of the ion, so that the second ligand species can still react, i.e., neither ligand alone must form a co-ordination- saturated complex with the ion.Another alternative is that the first ligand, on entering the co-ordination sphere of the cation, fully satisfies it, but does so in a purely dative fashion, so that this primary complex ion still bears its over-all positive charge of the original central ion, and is free to ion-associate with a second ligand of suitable anionic charge to form a ternary complex.72 WEST: SOME SENSITIVE AND SELECTIVE [AnaZyst, Vol. 91 An example of the first type is the extremely useful ternary complex formed between alizarin complexan, cerium(II1) (or lanthanum) and fluoride ion.12J3,14,15716 This provides a sensitive and virtually specific method for the determination of the fluoride ion, which is probably the first instance recorded of the development of a positive colour reaction for the fluoride ion.The molecular extinction coefficient of this complex is of the order of 30,000, and we are currently developing similarly selective ternary complex systems for fluoride ions which yield molecular extinction coefficients of the order of 90,000.17 The second type of complex is well illustrated by a method we have recently described for the determination of the silver ion. A ternary complex formed between silver, 1,lO-phenan- throline and bromopyrogallol red, (AgPhen,) ,BPR, yields an extinction coefficient of 51,000, and in the presence of EDTA, fluoride and peroxide, only gold(II1) interferes.18 The stability of the colour system is vastly superior to that of silver - dithizone, and it is more sensitive (E -rr 30,000 for silver - dithizone).More recently we have developed3 this principle for the determination of the metals shown in Table I. The values quoted for the extinction coefficients TABLE I TENTATIVE SENSITIVITIES OF TERNARY ION-ASSOCIATION SYSTEMS Ion determined Cadmium Cobalt Copper (11) Manganese Nickel Lead Zinc Molecular absorptivity 92,000 (Ethyl acetate) 92,000 (Ethyl acetate) 75,000 (Ethyl acetate) 65,000 (Ethyl acetate) 50,000 (Chloroform) 70,000 (Nitrobenzene) 95,000 (Ethyl acetate) Molecular absorptivity (dithizone) 85,000 59,000 45,000 32,000 34,000 72,000 94,000 of these complexes extracted into chloroform, ethyl acetate or nitrobenzene, are tentative, and are capable of improvement by optimisation of conditions.In this series tetra-iodotetra- chlorofluorescein (Rose Bengal extra, C.I. 45440) is used as the counter ion for the central metal - phenanthrolinium cation. The mechanism of the colour reaction is of considerable interest, but this factor cannot be discussed in this paper. It should be added, however, that inter-element selectivity for the ions shown in the table can readily be achieved and we have so far devised specific procedures for copper and lead within the group itself.3 Another principle that appeared attractive in order to overcome the “sensitivity barrier” in spectrophotometry is the application of an amplification procedure. The Leipert amplification of iodine is, of course, well known.It results in a 6-fold yield of iodine per original iodide ion subjected to analysis. We have recently applied this principle to the determination of phosphorus.lg The phosphate ion is converted to phosphomolybdate, in which 12 molybdate ions are associated with each phosphate ion, and the complex is extracted away from the excess of molybdate and any other heteropoly acids formed from antimony, arsenic, germanium and silicon by means of butanol- chloroform. The extract is then put in contact with a pH 9 buffer, which re-extracts the phosphate ions and the associated 12 molybdate ions into the buffer. In this medium these are no longer chemically combined, so that the molybdate ions can now be made to react with the sensitive reagent, 4-chloro- 2-aminobenzenethiol, to give an easily measured complex.This amplification procedure results in an effective molecular extinction coefficient of 360,000 for phosphate ion, so that solutions as dilute as 0.008 p.p.m. of phosphorus may easily be determined. By contrast, the standard molybdenum-blue procedures have values of about 27,000. A similar, but considerably less sensitive and selective, procedure for phosphate has recently been described by Umland and Wunsch.20 ATOMIC-ABSORPTION SPECTROSCOPY- Atomic-absorption spectroscopy is, of course, concerned with the light-absorption charac- teristics of atoms as opposed to molecules, and solution and cuvette are replaced by a steady- state flame of suitable physical characteristics, to maintain a population of free ground-state atoms.A conventional monochromated light source is not acceptable for this technique because of the narrow profile of absorption bands due to free atoms. Consequently, sources such as hollow-cathode lamps, capable of emitting even narrower bands, must be used. The technique is more sensitive than flame (thermal emission) photometry because, for the majority of elements in the majority of flames, the overwhelming bulk of free atoms remainFebruary, 19661 REACTIONS IN INORGANIC SPECTROSCOPIC ANALYSIS 73 in the non-emitting ground state. It is also a more selective technique than flame photometry because it is very free from inter-element interference. It is not very dependent on flame temperature, whereas flame photometry exhibits an exponential dependence. However, like solution absorptiometry, it follows the same physical laws, and has a maximum sensitivity which is controlled by log Io/It fi 0 for traces (a signal that is independent of amplifier gain) and by the limiting laws of the probability of an electronic transition.Most instruments for atomic-absorption analysis of traces are, therefore, operated under conditions of maximum sensitivity. Devices such as long flames, heated tubes, multiple traverse of flames, are being examined on the one hand, and other substrates such as plasma jets and sheathed flames on the other, to obtain higher atomic populations for such “difficult” metals as niobium, molybdenum and titanium. Since selectivity of reaction is an inherent property of atomic-absorption spectroscopy, we have turned our attention principally to the use of organic complexing reactions as a means of increasing sensitivity.This can be achieved by virtue of the fact that most metals can be persuaded to partition into water-immiscible solvents as metal chelate complexes. After the solution has been sprayed into the flame, the, droplets evaporate to solid particles, which then dissociate to free atoms. When a solvent such as an ester or ketone is sprayed, the efficiency of aspiration increases greatly. Thus the throughput ratio of ethyl methyl ketone to water, in terms of ml per minute of liquid fed into the flame is about 4.19 When the ratio of absorbance signals of a metal, extracted into an equal volume of this solvent, is compared with that of an equal concentration of the metal in aqueous solution, a 4-fold enhancement of signal is generally obtained. Other advantageous factors inherent in the spraying of solutions of metal chelates are that the droplets evaporate more quickly and are smaller, and that the solids so formed are more volatile and the majority dissociate exothermally. In addition, ions that would formerly have been associated with anions such as phosphate (which are relatively difficult to break down) are no longer chemically bound to them.The relative importance of all these factors will vary from system to system, but undoubtedly one of the chief advantages is the improvement in rate of aspiration into the flame. Sensitivities can yet again be increased by control of phase ratios.Thus for an apparatus that gives a calibration-curve range of 1 to 10 p.p.m. of silver in aqueous solution (absorbance from about 0.04 to about 0.3), a calibration curve of 1 to 5 p.p.m. (absorbance from about 0.1 to about 0.5) is obtained after extraction into an equal phase volume of hexone, whilst extraction from a large volume of aqueous solution into a smaller volume of the same solvent in a simple separating funnel gives a calibration curve of 0.01 to 0.10 p.p.m. of silver in the original aqueous solution (absorbance from about 0.03 to 0.25).21 In some instances organo-metallic complexes break down too easily in a flame, e.g., tetraethyl-lead, so that calibration curves can only be constructed against standards prepared from tetraethyl-lead and measured at the lowest possible point in the flame.22 In other instances, e.g., the formation of a co-ordination complex of platinum with the thiocyanate ion or dithizone, can lead to the formation of stable complexes that do not break down to atomic species in the flame.23 Both these results were obtained with air - propane flames; variation of flame composition could alter these observations substantially.In other instances, such as the determination of copper in niobium and tantalum,24 the use of solvent extraction permits the separation and concentration of traces of copper from preponderant amounts of matrix material, which might otherwise lock up the ion being determined in a refractory oxide in the flame, or which might block up burner heads and other orifices.Within certain limits the type of extractant used is immaterial, because the specificity of the atomic-absorption technique circumvents interference from other co-extracted trace impurities. EMISSION TECHNIQUES Because of the inherent limitations of absorbance techniques with respect to lower limits of detection and determination, the author and his colleagues have been concerned with investigating simiIar techniques that might be suitable for adaption to analytical pro- cedures, with detection or determination limits several orders of magnitude beyond those of absorptiometric procedures. The answer in respect of both solution and flame techniques appeared to lie in the development of emission techniques in both media. There are three principal emission phenomena observed in solution.These are fluorescence, phosphorescence and chemi-luminescence. This paper is concerned only with the first74 WEST : SOME SENSITIVE AND SELECTIVE [Analyst, Vol. 91 mentioned. Phosphorescence in solution is rare, and may indeed be considered as a special type of long-lived fluorescence induced by the existence of triplet-excited electronic states. Chemi-luminescence is also rather rare in solution, and arises chiefly as a result of free-radical reactions involving the luminescent species. The stability of luminescent species is very small, so that the technique does not readily lend itself to direct analytical manipulation.26 The chief emission phenomena in flame media are fluorescence and thermal emission, although flame luminescence is also not unknown.The technique of flame photometry, which is based on thermal excitation, is well known, and chemi-luminescence in flames is relatively rare. Accordingly, this paper is once more concerned only with fluorescence phenomena in flames. SPECTROFLUORIMETRY- The fundamental aspects of analytical spectrofluorimetry in solution have recently been described elsewhere by the author,26 as have the basic requirements with respect to instru- mentation of the te~hnique,~’ and will not, therefore, be described here. A molecule in solution absorbs light $7 reacting with the photons that are in contact with it for about second (ie., the period of oscillation of the electromagnetic wave). The molecule will only absorb light that is keyed to the energy difference between its ground state and an excited electronic state (ie., the absorption is quantised).The excited-state molecule is unstable and will tend to get rid of its surplus energy. Those stable substances that we normally think of as coloured shed their energy by various intermolecular processes that may be called radiationless transfers and which amount to radiation of the energy as heat. Some absorbing species, however, have abnormally stable excited states, so that these can hold on to their acquired energy for up to lo-* second. During this relatively long period, part of the absorbed energy will be degraded by intramolecular vibrations that occur within 10-l2 second until the molecule reaches its lowest vibrational level. These excited- state molecules then have a very high probability of releasing their surplus energy radiatively as a single photon corresponding to the difference between the energy levels of the lowest vibrational state of the excited level and one of the vibrational states of the ground state, i.e., some of the absorbed light is re-emitted as light of a longer wavelength.The emitted light is characteristic of the molecule and invariably bears a mirror-image relationship to the longest-wavelength absorption band of the molecule. For dilute solutions that absorb only a small fraction of the exciting radiation, I,, the basic equation which relates the intensity of fluorescent light, F , to the concentration of the fluorescing species, C, may be simply stated as- where K is a proportionality constant made up of the quantum efficiency of the system, 4, the molecular extinction coefficient at the wavelength of excitation, E , and the absorbing pathlength in solution, 1. Comparison of this basic equation with that for absorbance, viz.- reveals that here no logarithmic dependence on a ratio of intensities is involved.Secondly, the law includes an I,, term so that the greater the intensity of the exciting radiation, the greater the analytical signal, F . Yet again, since F normally represents the output from a photomultiplier tube, it can be electronically amplified within reasonable limits imposed by the “noise” of the detector and the circuit. Furthermore, two sets of spectra, the excitation and emission spectra, become available for detection and determination.This also is in marked contrast to the availability of analytical information from absorbance measurements. Unquestionably, the technique is a far more sensitive one than absorption spectro- photometry, and it is surprising that analytical chemists have paid so little attention to it. The lack of suitable commercial equipment may be partly responsible, but there is also a dearth of reagents for inorganic analysis. The last-mentioned factor can only be set right by analytical chemists themselves. F = K . I , . C A = log Io/It = E . L . C We are currently investigating spectrofluorimetry from two angles- (a) the development of a range of spectrofluorimetric reagents for a wide range of inorganic ions; and (b) the tracing of relationships between spectrofluorimetric activity and reagent structure.February, 19661 REACTIONS IN INORGANIC SPECTROSCOPIC ANALYSIS 75 Generally, in absorption work, the more complicated and massive the molecule, the greater its molecular extinction coefficient.The converse, happily, appears to be true in spectrofluorimetric analysis. For example, we have found it possible to determine traces of thallium(1) down to 0.01 p.p.m. in a medium simply made 3.3 M with respect to hydrochloric acid and 0-8 M with respect to potassium chloride.28 Irradiation of this solution at 250 mp produces blue fluorescence emission peaking at 430 mp. Out of 42 cations and 11 anions examined, only relatively large amounts of lead, copper(II), tin and cerium(m) produced a fluorescence.Many of the other ions interfered by precipitating or by inner-filter effects, but all these except for gold, bismuth, platinum and antimony, were eliminated by a simple separation process. The thallium(1) was oxidised to thallium(m) by hydrogen peroxide and extracted from 1.5 M hydrochloric acid by diethyl ether. It was then back-titrated into an aqueous phase and concomitantly reduced to thallium(1) by aqueous sulphur dioxide, after which the latter was boiled out, the acidity adjusted with hydrochloric acid and measurements made as already described. Similarly, we have applied the reagent 2-hydroxy-3-naphthoic acid to the spectrofluori- metric determination of beryllium down to 0-0002 p.p.m. (20 nanograms) by using the calcium salt of CDTA as the masking agent29 at pH 5.8 in an acetate buffer.Only those ions interfere that cannot be held in solution as soluble salts under these conditions, i e . , bismuth, cerium(m), chromium, iron, tin, titanium and thorium. Similarly, the reagent salicylidene-o-aminophenol may be used to determine aluminium down to 27 nanograms.30 When the procedure is used in combination with a diethyldithio- carbamate extraction procedure, out of 46 cations examined, only chromium(11r) , scandium and thorium showed interference. We have similarly developed methods for the deter- mination of scandium with salicylaldehyde semicarba~one,~~ for molybdenum and tungsten with carminic gallium and aluminium with salicylidene-o-aminophenol,33 and are developing procedures for others such as gold, copper and phosphorus.There are few manipulative difficulties in spectrofluorimetry, and we have encountered no serious problems from quenching effects due to oxygen or from steep temperature gradients on calibration curves. Indeed, there is no real reason why this technique should not be developed as extensively as absorption spectrophotometry. Even with ordinary instru- mentation it appears to be capable of yielding limits of determination three orders of magnitude lower. ATOMIC-FLUORESCENCE SPECTROSCOPY- It is a logical sequence to proceed from emission spectrophotometry in solution to emission spectrophotometry in flames. Thermal excitation of atoms in flames is an inefficient process. Except for a very few metals, even in the hottest flames the majority of atoms present in the flame plasma remain in the ground state, and the small number of atoms in the emitting excited state varies exponentially with changes in flame temperature. The emission is also very subject to inter-element effects, so that extensive use has to be made of radiation buffers and similar devices.However, atoms in the ground state can readily be persuaded to fluoresce by irradiating them with light of the correct wavelength required to produce an electronic excitation. Usually this light corresponds to that used for measurements of atomic absorbance. The atomic state in a gaseous phase exhibits the phenomenon of resonance re-radiation, i.e, the excited atoms fluoresce light of the same wavelength as they have absorbed, but with diminished intensity owing to the quantum efficiency of the process.Normal fluorescence of longer wavelengths may also be observed in some instances, although it is a good deal less common than in (molecular) fluorescence in solution. The basic principles of this technique are so much a blend of the previously discussed techniques of spectrofluorimetry in solution and atomic absorbance in flames that little need be said of them. As before, the analytical fluorescence signal, F , may be expressed as- F = K . I o . C . Again, there is a linear signal that may be multiplied or augmented as necessary by control of I,,, and also the possibility of electronic amplification exists where it does not in atomic absorbance.76 WEST: SOME SENSITIVE AND SELECTIVE [Analyst, Vol. 91 The experimental arrangements of atomic-absorption spectroscopy and of atomic- fluorescence spectroscopy may best be expressed diagrammatically- I Absorbance = Log 2 I T Signal = I F IF = K.1, [Atom,] In some respects the instrumental demands of atomic-fluorescence spectroscopy are simpler than those of atomic-absorption spectroscopy.Incident light of a very narrow spectral profile is required for atomic-absorbance measurements because of the narrow band- width of the absorption bands. This is best obtained from a hollow-cathode lamp, with the cathode made from the metal concerned or one of its alloys. On the other hand, for atomic- fluorescence spectroscopy, much cheaper and more readily obtained non-reversed sources such as spectral-discharge lamps may be used. The freedom from interference, characteristic of atomic-absorption spectroscopy, should be almost as forthcoming in atomic fluorescence so that isotopic analysis should be possible, and the sensitivity of the method may well be several thousand times that of atomic absorption. Some factors such as self-absorption and scattering, which are of little or no account in atomic absorption, may produce additional problems in this newer technique, and others, such as quenching, which are not observed in absorption, require to be considered.Scattering may not be eliminated by light modulation in the same way as thermal emission is eliminated in atomic absorption, but the avoidance of total-consumption burners and the use of the fine sprays that are obtained when organic solvents are used, would greatly minimise such problems.Thermal emission in the flame may be rendered harmless by modulating the source and a.c. amplification. In preliminary experiments we have readily been able to detect cadmium down to M solutions, and obtain measurable responses with as little as 0-2 nanograms of cadmium ion.34 Using the 2288 A line, which promotes the 5 ‘ ~ ’ ~ --f 5’$1 transition, and measuring the resonance emission obtained with solutions of cadmium extracted into ethyl acetate as the [CdIZ-] complex from an acid solution, we have obtained results accurate to within + 1 per cent. at the level, and have found that cadmium suffers no interference from 100-fold amounts of NH,, Ag, Al, As, Au, Ba, Be, Bi, Ca, Ce, Co, Cr, Cu, Fe, Ga, Hg, In, K, La, Li, Mg, Mn, Mo, Na, Ni, Pb, Sb, Sc, Se, Sn, Sr, Te, Th, T1, U, V, W, Y, Zn, Zr or from acetate, borate, bromide, citrate, chloride, perchlorate, cyanide, fluoride, oxalate, phosphate, silicate, sulphite, sulphate or tartrate.When the following complexing agents were present : peroxide, EDTA, 1 ,lo-phenanthroline or reducing agents such as ascorbic acid or hydroxyl- ammonium chloride, interference was observed only from the phenanthroline. This was because of the formation of a precipitate. When sufficient acid was added to dissolve it, no interference was encountered. These results were obtained with rather crude and inefficient apparatus, and are capable of considerable improvement by modifications to obtain greater efficiency of irradiation and light collection. Undoubtedly, this is a technique that will prove to be invaluable for inorganic trace analysis in the nanogram range in the near future.Practically the only papers to have appeared on analytical aspects of this technique are those by Winefordne9~36,3~ ~ 3 8 and his co-workers. The author wishes to express his indebtedness to students cited in the references, and in particular to his colleagues, R. M. Dagnall and G. F. Kirkbright. Sincere thanks are also given to many industrial firms who have contributed apparatus and grants in aid of research, and especially to the Science Research Council, without whose support none of the work on atomic-absorption spectroscopy or spectrofluorimetry would have been possible.February, 19661 REACTIONS I N INORGANIC SPECTROSCOPIC ANALYSIS REFERENCES 77 1.2. 3. 4. 6. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 1 7 . 1s. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. Merritt, L. L., and Walker, J. K., I n d . Engng Chem., Analyt. Edn, 1944, 16, 387. Dagnall, R. M., West, T. S., and Young, P., Analyst, 1965, 90, 1066. Bailey, B., Dagnall, R. M., and West, T. S., Talanta, in the press. Close, R. A4., and West, T. S., Ibid., 1960, 5, 221. Herrero Lancina, M., and West, T. S., Analyt. Chem., 1963, 35, 2131. Kusakul, P., and West, T. S., in preparation. Belcher, R., Ramakrishna, T. V., and West, T. S., Talanta, 1965, 12, 681. Ramakrishna, T. \-., and West, T. S., in preparation. Christopher, D. H., and West, T. S., TaZanta, in the press. Uraude, E. A,, in Braude, E. A., and Nachod, F. C., Editors, “Determination of Organic Structures Belcher, R., Leonard, M. h., and West, T. S., Talanta, 1959, 2, 92. , , Analytica Chim. Acta, 1965, 32, 301. -- by Physical Methods,” Academic Press Inc., N.Y., 1955, L-olume I, p. 131 et seq. , , J.. Chem. Soc., 1958, 2390. --- 9 , , Ibzd., 1959, 3577. Leonard, M. A,, and )Vest, T. S., Ibid., 1960, 4477. Belcher, R., and West, T. S., Talanta, 1961, 8, 853 and 863. Cabello-Tomas, L., and West, T. S., in preparation. Dagnall, R. M., and West, T. S., Talanta, 1964, 11, 1533. Djurkin, V., Kirkbright, G. F., and West, T. S., Analyst, 1966, 91, 89. Umland, F., and Wiinsch, G., Z. analyt. Chem., 1965, 213, 186. Belcher, R., Dagnall, R. M., and West, T. S., TaEanta, 1964, 11, 1257. Dagnall, R. M., and West, T. S., Ibid., 1964, 11, 1553. Cabrera, A. M., and West, T. S., unpublished work. Kirkbright, G. F., Peters, M. I<., and West, T. S., unpublished work. Howard, P., and Wcst, T. S., unpublished work. West, T. S., Lab. Pract., 1965, 922. Kirkbright, G. F., West, T. S., and Woodward, C., Talanta, 1965, 12, 677. Dagnall, R. M., Smith, R., and West, T. S., Talanta, in the press. Kirkbright, G. F., West, T. S., and Woodward, C., ‘4naZyst, 1966, 91, 23. - ~ - , in ereparation. Dagiall, R. M. Smith, R., and West, T. S., Chem. & Ind., 1965, 1499. Dagnall, R. M., West, T. S., and Young, P., Talanta, in the press. Winefordner, J . D., and Vickers, T. J.. Analyt. Chem., 1964, 36, 161. Winefordner, J. D., and Staab, R. A., Ibid., 1964, 36, 165. -___ , Ibid., 1964, 36, 1369. Mankeld, J. M., Winefordner, J. D., and Veillon, C., Ibid., 1965, 37, 1061. ~-~ -, Ibid., 1965, 1030. , I , Analyt. Chem., 1965, 37, 137. ~ - - Received September loth, 1965