The Application of Separated Flames in Analytical Atomic Spectrometry M. S. CRESSER Department of Soil Sciences, University of A berdeen, Scotland P. N. KELIHER De9artment of Chemistry, Villanova University, J’illanova, Penn., U .S.A . and G. F. KIRKBRIGHT Department of Chemistry, Imfierial College, London S . W.7 Contents Introduction Use of the mechanically separated flame for atomic emission and fluorescence Use of the mechanically separated flame in atomic absorption spectrometry Use of the inert-gas separated flame in atomic emission spectrometry spectrometry Comparison of long path and separated air - acetylene flames in atomic emission Use of the inert-gas separated nitrous oxide - acetylene flame in atomic emission Use of other inert-gas separated flames in atomic emission spectrometry spectrometry spectrometry Use of the inert-gas separated flame in atomic absorption spectrometry Use of inert-gas separated air-acetylene flames in atomic fluorescence spectrometry Use of the inert-gas separated nitrous oxide - acetylene flame in atomic fluorescence Use of other inert-gas separated flames in atomic fluorescence spectrometry Conclusion References spectrometry140 CRESSER, KELIHER AND KIRKBRIGHT Introduction The most widely used atom cells in analytical atomic emission (AES), absorp- tion (AAS) and fluorescence (AFS) spectrometry are the pre-mixed flames of hydro- carbons or hydrogen with air, oxygen or other oxidant, supported at slot or Meker-type burners.When used in conjunction with an indirect pneumatic nebu- liser to provide for the introduction of liquid samples in aerosol form, these flames permit a high sensitivity for many elements to be attained by each technique, with considerable freedom from chemical and physical interference effects.Two of the most important criteria for a flame that is to be successfully applied for practical analytical flame spectrometry are that it should exhibit a high atomisation efficiency for a wide range of elements, and a low background emission or absorp- tion and noise near to the wavelength of the atomic lines used for analysis. Pre- mixed flames exhibit the familiar and well defined structure typified by the flame at a Bunsen burner. The primary zone, where the initial oxidation reaction occurs at the flame front, shows intense emission from species such as CH, C,, OH, CN and NO, and poor signal-to-noise ratios are usually obtained when this part of the flame is viewed; it is therefore seldom useful for practical analytical work.The secondary reaction (or diffusion) zone, where the further oxidation of the products of the primary reaction (typically CO and H2) occurs with atmospheric oxygen, similarly exhibits background emission from OH and the chemiluminescence associated with the oxidation of carbon monoxide. This part of pre-mixed flames does not generally provide a stable environment of constant temperature and composition for utilisa- tion in analysis. The volume between the primary and secondary reaction zones, referred to as the interconal zone, however, frequently shows the highest tempera- ture in the flame volume and may attain approximate local thermal equilibrium for hydrocarbon flames; in addition, the environment in this flame may be relatively reducing in character, particularly in fuel-rich flames, and favour the atomisation of elements that would otherwise form thermally stable oxide or hydroxide species in the flame gases.Even a t the high temperature attained in this zone the radiative background emission from the interconal zone may be very low compared to that observed from the primary and secondary reaction zones. This volume of the flame gases is therefore most suitable for the spectroscopic examination of samples introduced into the flame. Unfortunately, in conventional pre-mixed flames it is difficult to ‘view’ this zone (in emission, absorption or fluorescence) without also sampling the secondary reaction (diffusion) zone, which surrounds it.As long ago its 1891, however, Teclul and Smithells and Ingle2 independently demonstrated the existence of the two different reaction zones in a pre-mixed flame supported at a Bunsen burner by their physical separation by using a flame ‘separator’ tube. When this device is used to ‘lift-off’ the secondary reaction zone from around the interconal zone, a technique is provided by which the latter zone can be exposed for sampling without interference from the secondary zone. The separated flames that result, obtained either with a Smithells separator or by lifting off the secondaryAPPLICATION OF SEPARATED FLAMES IN ANALYTICAL ATOMIC SPECTROMETRY 141 reaction zone with an inert-gas shield, have potential advantages in each of the three techniques of quantitative analytical flame spectrometry.Although only 5 years have elapsed since the quantitative analytical use of separated flames was first ~uggested,~ the widespread interest shown in the per- formance of the inert-gas separated flames has led to the availability of suitable burners and to their use with both commercial (Technicon Instrument Corp., Model AFS-6 ; Varian-Techtron Pty. Ltd.) and e~perimental*-~ purpose-built atomic absorption and fluorescence spectrometers. Several of the important con- siderations that govern the potential of this technique for reducing flame back- ground and noise have been previously virtually ignored in both the extensive literature and earlier reviews', of the subject; this has sometimes resulted in conflicting reports concerning the value of the use of separated flames in analytical spectrometry. One of the objects of this review is, therefore, to give a clear indica- tion of the relative merits of the use of separated flames, with emphasis on the effects of instrument a1 parameters, particularly monochromat or field of view, spectral band-pass and burner design.A short account of the evolution of the analytical usage of both mechanical flame separation and separation by inert-gas shielding is included, as the interpretation of the early published data leads to a clearer appreciation of the applicability of separated flames for analysis. An excellent account of the use of separated flames in fundamental studies of flame structure and processes prior to their analytical use has recently been given elsewhere9 and is therefore not included here.Use of the Mechanically Separated Flame for Atomic Emission and Fluorescence Spectrometry Kirkbright, Semb and West3 realised the potential advantage of the use of the Smithells-type flame separator for the virtually total elimination of the radiation and noise originating from the secondary reaction zone. These authors investi- gated the use of a fused silica tube (internal diameter 20 mm) to lift the secondary diffusion flame to a height of 60 to 80 mm above the primary reaction zone of a pre-mixed air - acetylene flame burning at a Meker-type burner head (Fig. 1). A decrease in OH and CO chemiluminescence intensity of at least two orders of magnitude was reported for the interconal zone of the separated flame relative to the unseparated flame (see Fig.2). The atomic emission intensity from elements introduced into the flame also decreased (by as much as one order of magnitude in the worst case that was investigated, k., calcium) on separation of the flame, which was probably caused by the slightly lower temperature in the interconal zone of the flame. Oxygen enrichment of the air used to support the flame (up to 15 per cent. v/v) was reported to restore this intensity to its value in the un- separated flame without deleterious effect on the background emission. External coating of the tube with magnesium oxide was shown to double the useful signal for the elements investigated.A similar arrangement was later described by the same authors in a preliminary report of the examination of the analytical usage of the142 CRESSER, KELIHER AND KIRKBRIGHT I/*\, I \ \ 1 I I Secondary\ I diffusion \ flame -’ zone +J .- *!I *Primary reaction zones :$I E LLI Q) co m (D I Wavelength, nm Fig. 1. Silica separator for a mechanically separated Fig. 2. Emission spectra of ( A ) normal and (B) separated flame on a Meker-type air - acetylene flame. (Ordinate of A must be multiplied by burner head 15 for comparison of absolute intensities) separated nitrous oxide - acetylene flame.1° For this flame, the Meker-type burner head was replaced by a circular-slot burner. An increase in the volume of the analytically useful ‘red-feather, zone of the slightly fuel-rich flame gave rise to a corresponding increase in the intensity of the CN band emission, but the OH band emission intensity and the CO oxidation chemiluminescence were again virtually quantitatively suppressed.Detection limits for atomic emission spectrometry were quoted for aluminium, beryllium and molybdenum to illustrate the analytical use of this flame. The authors observed that the prevention of air entrainment also led to a virtual elimination of the green continuum produced by higher oxides of molybdenum, and noted that the intense emission produced in the red could be assigned to the lower oxide of molybdenum, MOO, not normally observed in flames. The formation of this species is no doubt related to the appreciable degree of atomisation of molybdenum in the interconal zone of the hot pre-mixed flame.In a later publication,ll the same authors used a modified silica separator device; this consisted of a separator tube of similar diameter to that used previously, but fitted with a side-arm and silica observation window to permit the undistorted transmission of radiation from the interconal zone. This separator device was used with the circular-slot burner in conjunction with an f / 5 prism monochromatorAPPLICATION OF SEPARATED FLAMES IN ANALYTICAL ATOMIC SPECTROMETRY 143 (Unicam SP 900) and also an f / l O grating instrument (Varian-Techtron AA4 spectrometer). The detection limits in emission spectrometry obtained with these instruments were compared for aluminium, beryllium, molybdenum, titanium and vanadium; tungsten emission was examined only with the latter instrument.This work demonstrated the beneficial effect of the higher available resolution of the grating instrument at longer wavelengths. In the most extreme case quoted, the close proximity of the Mo 390-3 nm line to the CN band head at 388.34 nm led to a 50-fold improvement in detection limit in a separated flame when the grating rather than the prism instrument was used. The authors commented that with the experi- mental arrangement used, only a relatively small region of the interconal zone was observed by the AA4 grating monochromator; the narrower field of view of this instrument may also contribute significantly to the improvement in detection limit.Although flame separation produces an enhancement in the intensity of the CN band emission, no data on relative detection limits in separated and conventional flames were included in this paper for the two instrumental assemblies. The pre-mixed nitrous oxide - acetylene flame fitted with a similar silica separator to those described above was used with a grating monochromator (Varian-Techtron AA4 spectrometer) to investigate the atomic emission spectra of the rare-earth elements,12 and a modified version, fitted with a second side-arm and silica end-window, was used to investigate the determination of beryllium by both thermal emission and atomic fluorescence spectrometry.13 It is of interest to note that although these early papers concerned with the application of the use of mechanically separated flames in emission and fluorescence spectrometry indicated that the interconal zone of a separated, pre-mixed nitrous oxide - acetylene flame was indeed a useful atom cell for analytical flame spectrometry, no direct compari- sons were reported between detection limits obtained in separated and unseparated flames.No evidence was presented to suggest that instrumental parameters consti- tute an important factor that must be considered in the evaluation of the applic- ability of separated flames for reasons other than the simple direct effect of monochromator resolution. Use of the Mechanically Separated Flame in Atomic Absorption Spectrometry The simplest mechanical method of flame separation for use in AAS is that described by Ure.14-16 This consists of parallel silica plates placed on each side of either a nitrous oxide - acetylene flamel42l5 burning at a conventional flat-top 50-mm slot burner, or an air - acetylene flame16 at a 100-mm slot burner.Ure used this flame for the determination of copper, zinc, manganese and molybdenum in soil extracts, and also applied the slot burner for emission studies, and commented that the relatively slight improvement in detection limit observed on separation was probably attributable to the fact that the monochromator used did not ‘see’ the secondary diffusion zone under normal circumstances, so that separation of this zone resulted in little advantage. This aspect of flame emission is considered in more144 CRESSER, KELIHER AND KIRKBRIGHT detail in a later section of this review.Hingle, Kirkbright and West1' described a more elaborate long-path assembly for AAS in which the interconal zone of a pre- mixed air - acetylene flame was drawn through a silica tube enclosed by a furnace. Facilityaas provided for the introduction of hydrogen into the interconal zone to stabilise fuel-lean flames, and absorption sensitivities were reported for eight elements. The results compared favourably with those reported for other flame-in- tube methods in which total-consumption nebuliser - burners were used; the authors claimed, however, that their system was more suitable than others for use with pre- mixed hydrocarbon flames and was less susceptible to background absorption effects. Use of the Inert-gas Separated Flame in Atomic Emission Spectrometry An alternative technique for the exclusion of entrainment of atmospheric oxygen above the primary reaction zone of pre-mixed flames is to surround the flame with a 'wall' of inert gas such as argon or nitrogen (see Fig.3). A simple Secondary diffusion flame ii c 0 E lnterconal zone 03 ti L 0 - s H Primaryreaction p Fig. 3. Inert-gas separator for Meker-type burner headAPPLICATION OF SEPARATED FLAMES IN ANALYTICAL ATOMIC SPECTROMETRY 146; method that can provide for a laminar flow of gas around a burner head to produce this effect involves surrounding the burner stem with a concentric tube or box of suitable dimensions packed with alternate layers of plain and corrugated (crimped) metal strip^.^^^^^ Close-packed capillary tubes20 and metal plates into which slots21 or concentric rows of holes22 are cut, have also been used for this purpose. The importance of providing a stiff wall of inert gas with laminar flow as it emerges from the shielding device, and maintaining this to a height in the flame well above that which is viewed by the detector, cannot be over-emphasised. The use of slots or concentric rows of holes cut into a metal plate is only satisfactory provided that the plate is sufficiently thick to allow laminar flow to be established and the slot or orifice dimensions are small enough for a suitable gas velocity to be reached with- out an excessive gas consumption. The principal advantages of this method of flame separation are the greater range of fuel - oxidant mixture strengths over which stable separated flames can be supported and the ability to nebulise a wide range of organic solvents into the flame without inducing flame instability or carbon deposition; reflection losses that occur at the walls of the silica tube are also eliminated, The use of inert-gas separation results in a similar reduction in flame back- ground emission intensity for the air - acetylene flame to that observed with a silica separator; the decrease in flame temperature on separation, however, may be somewhat greater.l8 The utility of the inert-gas separated air - acetylene flame in emission spectrometry has been illustrated by a detailed investigation of the determination of bismuth a t 306.8 nm ( i e ., within the strongest OH band emis- sion) .la Various authors have applied this type of inert-gas separation to the determination of twenty-four elements by flame emission spectrometry in the air - acetylene flame.21#23-25 Although improvements in the detection limit on separa- tion were reported for seventeen elements to be about one order of magnitude when an f/5 prism monochromator (Unicam SP 900) was used,23 other authors21 found that by using an f / l O grating monochromator (Varian-Techtron AA4) of greater resolving power, only small improvements in emission detection limits were obtained when the flame was separated by inert gas. The same authors used a 50-pm slit (a nominal spectral band-pass of 0.17 nm) throughout their investigation, however, and commented that the best results were obtained with a narrow band- pass.Although the presence of emission from concomitant elements in samples can sometimes render the use of a narrow spectral band-pass desirable, or even essential, in most instances spectral slit-width is one of the parameters that must be optimised for the attainment of the best detection limit. Moreover, the optimum signal-to- noise ratio may be obtained a t a different spectral band-width for separated and conventional flames. Under optimised conditions, an improvement by a factor of two or three may still be obtained with a grating monochromator of relatively high resolution25 on separation of flames burning a t circular Meker-type burner heads. Nevertheless it is true that the separation of air - acetylene flames sup- ported at long-slot burners offers only minor advantages when these are used for emission studies in conjunction with a monochromator with a narrow field of view.146 CRESSER, KELIHER AND KIRKBRIGHT 0 1 I 0 1 01 ' ' ' I ' 1 ' 1 4 3 2 1 0 1 2 3 4 Distance from flame centre, mm 4 3 2 1 0 1 2 3 4 Distance from flame centre, mm (b) O l ~ 1 t l I ! l 4 3 2 1 0 1 2 3 4 Distance from flame centre, mm (C) Fig.4. Emission intensity profiles of (a) OH band emission, (b) CuOH band emission and (c) Mn atomic emissionAPPLICATION OF SEPARATED FLAMES IN ANALYTICAL ATOMIC SPECTROMETRY 147 Comparison of long-path and separated air - acetylene flames in atomic emission spectrometry Several authors26-30 have indicated from theoretical considerations that, for low analyte concentrations, the dependence of signal intensity upon flame length should be similar in flame atomic emission and flame atomic absorption spectroscopy.Koirtyohann and Pickett2 have also reported an appreciable improvement in the barium line - calcium hydroxide background emission intensity ratio by using a long-path burner rather than a circular-slot or Meker burner for the determination of barium in the presence of excess calcium by atomic emission spectrometry. In this respect, the use of a long-path burner in emission studies resembles to some extent the use of a separated flame, where the emission interference from many molecular species that occurs in the secondary diffusion zone is avoided. The analytical exploitation of this effect, however, depends on a number of factors, the most important of which are the monochromator aperture and the burner configuration. Fig.4(a) shows the emission intensity profiles of a conventional 100-mm path length air - acetylene flame at the 306.4-nm OH band head viewed ‘end-on’ by using an EEL 240 flame spectrophotometer under conventional operating condi- tions. Although suitable light stops can be used to produce an even more pro- nounced low background intensity flame volume, which is viewed by the mono- chromator, this stratagem may be of only limited analytical use because of the substantial loss in emission signal strength that is incurred, and this may then re- quire the use of higher amplification or an increased spectral band-pass; either or both of these factors can have a sufficiently deleterious effect on the line signal - background noise ratio to offset any real advantage gained.A similar set of emis- sion profiles for the CuOH band emission is shown in Fig. 4(b). As expected, a region of low molecular background emission exists immediately above the primary reaction zone. The superimposition of a typical set of atomic emission profiles for a relatively easily atomised element, such as that shown in Fig. 4 (c) for manganese at 403.1 nm, can be used to illustrate that for instrumentation having a relatively narrow aperture, a slight advantage can be gained by the use of a long- path burner in flame emission studies; this advantage, however, is less significant than that obtained by the use of a separated flame even when a narrow spectral band-pass is used.Many spectrometers are designed primarily for flame emission spectrometry and use a relatively wide field of view; with instrumentation of this type there is often little to be gained by using long-path burners rather than circular Meker-type emission burners as the increase in atomic line intensity obtained is usually then associated with a corresponding increase in flame background. This situation is illustrated in Fig. 5 , which shows the influence of burner configuration on the spectral line intensities and line-to-background intensity ratios obtained for four elements by using a range of burners and an f / 5 prism monochromator (Unicam SP 900). It is evident that an increase in line emission intensity is always accom- panied by an increase in background intensity unless additional slits are added K148 CRESSER, KELIHER AND KIRKBRIGHT Relative signal-to-background ratios imp .......... ... Mn Re tat ive sig nal-to- bac kground ratios A B C D E F A B C D E F Relative signal intensities Relative signal-to-background ratios A Mg C D E D F Relative signal intensities Relative signal intensities Relative signal-to-background ratios Fe €3 A B C D E F Relative signal intensities Fig. 6. Histograms showing the effect of burner configuration on relative signal intensities (lower blocks) and relative line-to-background ratios (upper blocks) for manganese, nickel, magnesium and iron. A, 60-mm slot burner with triangular cross-section; B, 50-mm triple-row capillary burner; C, normal Meker-type burner; D, 50-mm slot burner (flat top) ; E, 50-mm triple-row capillary burner used with an additional light-guide to restrict the area of the flame viewed to a central region; and F, circular capillary burner head with argon sheathAPPLICATION OF SEPARATED FLAMES IN ANALYTICAL ATOMIC SPECTROMETRY 149 between the burner and monochromator to limit the region of the flame viewed to a narrow central region.This was achieved here with a triple-row capillary burner, which produces a flame that is much broader than that obtained with a conventional slot burner. The wider flame is necessary if an intense signal is to be obtained and the field of view is to be limited to view only a central region of the flame. Even under these most favourable conditions, the resulting line-to-background intensity ratio does not compare favourably with that obtained by using a separated flame.With instruments of this type the use of a long-path separated flame for emission spectrometry probably represents the optimum configuration. The increased curvature that results from the use of a long-path burner for emission work is not significant at the low analytical concentrations of interest.26 The results shown in Fig. 5 also demonstrate why the use of separated flames in emission spectrometry with monochromators of restricted field of view, such as those used in many atomic absorption instruments, only gives rise to little advantage. Use of the inert-gas separated nitrous oxide - acetylene flame in atomic emission spectrometry The inert-gas separated nitrous oxide - acetylene flame possesses similar advantages over the mechanically separated flame to those already described for the corresponding air - acetylene flames, namely maintenance of a stable flame without carbon deposition when organic solvents are nebulised, and stability over an extended range of fuel - oxidant concentration ratios.The immediate effect of nitrogen or argon separation is an elongation of the highly reducing interconal zone,*1 accompanied by a corresponding increase in the emission intensity from the CN molecular band emission from this zone. The decrease in flame temperature in the interconal zone on separation depends on whether argon or nitrogen is used (see Fig. S), but is generally small (about 100 "C) immediately above the primary Helght, mm Fig.6. Effect of inert-gas sheathing upon temperature - height profiles for the nitrous oxide - acetylene flame : A , unshielded flame ; B, argon shielded flame ; and C, nitrogen shielded flame150 CRESSER, KELIHER AND KIRKBRIGHT reaction zone.32 A study of the influence of acetylene flow-rate on the emission intensity observed for elements that form refractory elements in conventional and inert-gas separated flames was also reported in this paper; for the nine elements investigated, an appreciable reduction in fuel flow-rate was invariably required on separation so as to obtain maximal atomic line emission intensity, and improve- ments of an order of magnitude or greater were reported for the detection limits obtained with an f / 5 aperture monochromator.Other workers21 have suggested that because the flame background intensity from the reducing species, par- ticularly CN, actually increases on separation, the detectability of some elements might deteriorate. If, however, the fuel flow-rate is re-optimised to the required lower value after separation, then appreciable improvements in line -background intensity ratio and signal-to-noise ratio are readily obtainable for elements such as aluminium (396.1 nm). Amos, Bennett and Brodie21 concluded that little could be gained by shielding, provided that a monochromator with a resolution better than about 0.1 nm was used. An over-all increase in, for example, CN band emission intensity from the background would have an even more deletorious effect when an instru- ment of only moderate resolution and wider field of view was used.However, the inadequate optimisation of experimental parameters rather than resolution cliff erences may explain this apparent discrepancy between the results obtained by the two groups of workers who used different instruments. Detection limits have been reported for some thirty elements by using the separated nitrous oxide- acetylene flame for emission spectrometry.*F, 31 In a recent paper concerned with the choice of a high-temperature flame excitation source for simultaneous multi-element analysis by flame emission spectrometry, Boumans and de Boer33 have described an examination of the pre- mixed nitrous oxide - acetylene flame supported at five different long-path slot burners. For simultaneous multi-element analysis with flame sources, a difficulty exists in obtaining a single set of conditions (fuel flow-rate, height of observation), which are the optimum parameters for the atomisation and excitation of different elements with widely differing spectrochemical properties.These authors pointed out that while the detection limits attainable by atomic emission spectrometry were similar for the elements investigated, no matter what burner was used, the same optimum conditions for observation were obtainable for elements of widely different properties (manganese and vanadium, for example) by using an argon- separated nitrous oxide - acetylene flame. Boumans and de Boer attributed this property of the separated flame to the relatively rapid variation in temperature with height above the primary reaction zone when inert-gas shielding is used; this results in the existence of favourable conditions for the excitation of elements with quite different properties within a relatively small geometrical region of the flame.Thus the steep temperature profile with height in the separated flame, which has been quoted as a possible disadvantage of these flames compared to conventional flames, may be an advantage for the purposes of simultaneous multi-element analysis.APPLICATION OF SEPARATED FLAMES IN ANALYTICAL ATOMIC SPECTROMETRY 151 Use of other Inert-gas separated flames in atomic emission spectrometry The superiority of acetylene over other commercially available fuels in analyti- cal flame spectrometry for the provision of hot, reducing flames is so well established that little work has been reported concerning the potential applications of other separated flames of hydrocarbons or hydrogen, supported by pre-mixing with nitrous oxide.The inert gas separation of nitrous oxide - hydrogen and nitrous oxide - MAPP Gas@(methyl acetylene - propadiene) flames has been shown25 to produce only minor improvements in the detection limits for virtually all of the twenty-four elements investigated. Although these flames have been shown to be useful as atom cells for elements that show little tendency to form refractory oxides, for elements such as aluminium and vanadium, the detection sensitivity is several orders of magnitude greater when acetylene rather than hydrogen or MAPP Gas is used as the fuel.Use of the Inert-gas Separated Flame in Atomic Absorption Spectrometry Inert-gas separated flames possess two fundamental properties that render them of interest for analytical AAS. The most signficant of these is the substantial reduction in the background absorbance of the flame gases when an inert-gas separation is used. This effect has permitted the analytically useful wavelength range of pre-mixed flames to be extended to regions that were previously considered to be inaccessible for analytical flame absorption spectrometry; thus with these flames, elements such as sulphur, phosphorus and iodine, which have their principal absorption lines at wavelengths less than 200 nm, can be determined directly by AAS. The second useful property of the inert-gas separated flame is the protected reducing environment in the interconal zone with fuel-rich mixtures.It is now well established that the ability of the fuel-rich nitrous oxide - acetylene flame to atomise refractory oxide-forming elements is the result of the reducing nature of the zone immediately above the primary reaction zone, as well as the fact that the flame temperature is high. This is illustrated by the fact that in the fuel-lean flame, whose interconal temperature is somewhat higher than that of the fuel-rich flame, the degree of atomisation of elements such as aluminium, vanadium and titanium is extremely low and their sensitive determination by AAS under these conditions is not possible. For elements normally determined by AAS with conventional flames, any reduction in flame background absorption that can be effected results in an increase in the intensity of the source radiation that reaches the detector, and a decrease in absorption flicker noise.Both of these factors can give rise to an improvement in the detection limit for particular elements. The AAS sensitivity (for 1 per cent. absorption), however, will only change substantially if the flame geometry, temperature or reducing properties are affected when an inert-gas separation is 8 MAPP Gas is a registered trade mark of Air Reduction Company, Inc.152 CRESSER, KELIHER AND KIRKBRIGHT used. For the argon or nitrogen-separated nitrous oxide - acetylene flame, changes in these factors result in an improvement in AAS sensitivity of between 15 and 60 per cent.for elements such as aluminium, beryllium, germanium, molybdenum, silicon, titanium, vanadium and zirconium, but an improvement in the detection limits for these elements of up to f o ~ r - f o l d . ~ ~ The improvement in detection limits for refractory-oxide-forming elements when flame separation is used has been found by other workers2' to be generally slightly less than that reported above. This difference may be attributable to the use of a somewhat less efficient separation device, or to the effect of the water cooling of the burner used by the former and demonstrates the danger of comparing results obtained with different burner systems (the spectrometer used by both sets of workers was similar). Two and three-fold improvements in detection limit on separation have been reported for the determination of arsenic and selenium, respectively, in the separated air - acetylene flame.36 This improvement results principally from the greater transparency of the separated flame in the far ultraviolet region of the spectrum below 200 nm.More recently, the inert-gas separated nitrous oxide - acetylene flame has been shown to be virtually completely transparent at the wavelengths used for the AAS determination of these two elements.36 This high 1 3 I I I I I 1 I I I I 1 I 1 1 2 1 Fig. 7. Emission spectra from an iodine electrodeless discharge lamp source : (1) nitrogen purge gas to burner and monochromator, no flame burning; (2) as 1, but with nitrous oxide - acetylene flame ignited; and (3) as 2, but with a 1000 p.p.m.iodine solution being nebulisedAPPLICATION OF SEPARATED FLAMES IN ANALYTICAL ATOMIC SPECTROMETRY 163 Nitrogen Fig. 8. Instrumental assembly for the determination of phosphorus, sulphur and iodine: EDL = electrodeless discharge lamp, C = chopper to modulate source radiation, F = nitrogen- separated nitrous oxide - acetylene flame transparency permits a triple-pass optical arrangement to be used so as to improve the overall sensitivity in this flame for both arsenic and selenium.s7 Subsequent work has been described with this flame being used for the direct determination of sulphur,88 iodine39 and ph0sphorus4~ by AAS at 180.7, 183.0 and 1'78.3 nm, respec- tively. Typical absorption measurement at the iodine 183.0 nm line from an electrodeless discharge lamp source is shown in Fig.7. It can readily be demon- strated that a 50-mm path length inert-gas separated nitrous oxide - acetylene flame is actually considerably more transparent than the atmosphere at these short wavelengths; thus at the iodine 183.0 nm line in this flame the transmission is about 75 per cent. of that of nitrogen alone, i.e., approximately two and a half times more transparent than the equivalent path length through air. The trans- parency appears to be a direct result of (a), the separation of the oxidising atmos- phere of the secondary reaction zone, and (b), the low partial pressure of oxidising absorbing species in the interconal zone of the fuel-rich flame. Because of the high atmospheric absorption at these wavelengths, the air between the source and flame, and flame and monochromator must be displaced by nitrogen and a nitrogen- purged or vacuum monochromator must be used (see Fig.8). It seems probable that the remarkable transparency of the inert-gas separated nitrous oxide - acetylene flame makes it primarily useful in flame AAS for analysis at wavelengths below 200 nm.154 CRESSER, KELIHER AND KIRKBRIGHT Use of Inert-gas Separated Air - Acetylene Flames in Atomic Fluorescence Spectrometry In atomic fluorescence spectrometry, the use of separated flames offers the same two principal advantages as in AAS, i.e., a potential extension of the reducing environment of the atom cell and a decrease in noise by a reduction in both the absorption and emission background intensity resulting from the secondary reaction zone.The degree of improvement attainable is critically dependent upon a number of instrumental factors; a most important consideration is the different effect of flame separation when dispersive (monochromator) and non-dispersive atomic fluorescence spectrometers are used. Any improvement in atomisation efficiency on separation would, if not accompanied by a change in other flame characteristics, be expected to lead to a similar improvement in sensitivity with both dispersive and non-dispersive spectrometers. When non-dispersive AFS instrumentation is used, however, the AFS sensitivity is limited by flame flicker and ‘shot’ noise; intense fluorescence signals are readily attainable, and the use of photomultiplier detectors at high gain can be avoided to minimise dark-current and electronic (amplifier) noise.Under these conditions, the use of inert-gas separation of the flame used as the atom cell should lead to a similar improvement in sensitivity for single fluorescence transitions, irrespective of the useful intensity of the source used. For example, Larkins4 has utilised a non-dispersive AFS spectrometer and reported that, for cadmium, gallium, mercury, indium, magnesium and zinc, where a solar-blind photomultiplier was used to view predominantly a single fluorescence line together with the flame background, the ratio of the detection limit for each element obtained in a conventional air - acetylene flame to that in an inert-gas separated flame was relatively independent of the type of source used to excite fluorescence (see Table I).The detection limits themselves, however, vary for the two different sources by two orders of magnitude or more. It is also of interest to note here that, although the noise from the flame should be independent of the analyte element for a non-dispersive spectrometer, the detection limit ratio (un- separated to separated flame) is not itself independent of the nature of the element analysed; indeed, for some elements the ratio differs appreciably from the 10 to 13-fold improvement generally observed. Such discrepancies can be expected for elements with complex fluorescence spectra, and, for example, when processes such as thermally-assisted fluorescence contribute appreciably to the observed signal intensity.The relative intensities of lines emitted by the source can vary consider- ably from one source to another. This in turn can result in different contributions from processes such as thermally-assisted fluorescence at the different temper- atures of the separated and unseparated flames. In some cases, however, variation in the atomisation efficiency between the two flames is a more probable cause of the apparent variation of the detection limit ratio for different elements. When dispersive (monochromator) instruments are used, dark-current noise and electronic noise may only be negligible compared to flame noise when the latter is particularly high and the fluorescence signal is weak. Under these condi-APPLICATION OF SEPARATED FLAMES IN ANALYTICAL ATOMIC SPECTROMETRY 155 TABLE I COMPARISON OF RATIOS OF DETECTION LIMITS OBTAINED IN INERT-GAS SEPARATED AND CONVENTIONAL AIR - ACETYLENE FLAMES FOR DIFFERENT EXCITATION SOURCES FOR ELEMENTS WITH SIMPLE FLUORESCENCE SPECTRA ~~~ ~~~ ~~ Detection limit ratio (conventional/separated) , 1 Non- Wave- dispersive Dispersive length, Element Source system system nm ~~~ ~~~ Cadmium HCL VDL Gallium HCL VDL Mercury HCL Indium HCL 02 VDL Magnesium HCL Zinc HCL HIL(B) HIL(B) 10 7.5 3.8 3.3 10 10 16.7 15 6 4.5 10 10 ~~ 3 1.3 10 2 2 1.3 8 4 12 2 2.6 2.3 6.2 4.0 2 1.7 ~~~~ 228.8 228.8 294.4 417.2 294.4 417.2 253-6 253.6 303.9 451-1 303.9 451.1 285.2 285.2 213.8 213-8 HCL = Hollow-cathode lamp.VDL = Metal vapour discharge lamp. HIL(B) = High intensity hollow-cathode lamp (Sullivan and Walsh type 6 * ) .02 = Mercury germicidal lamp (Philips type 02 4W). tions the advantage of flame separation is similar to that for non-dispersive instru- ments. When sources of very high intensity are available, dark-current limitation of the sensitivity becomes predominant and this has the effect that inert-gas separation of the flame has less effect on improvement in the attainable detection limits. Larkins’ results* clearly illustrate that with more intense sources less benefit is obtained from flame separation used with dispersive intrumentation. If methods of reducing the contribution from dark-current, such as cooling of photo- multiplier tubes, are adopted, however, the use of flame separation may still prove advantageous even where dispersive spectrometers are used.It can be concluded that the optimum instrumental arrangement for high AFS sensitivity at present appears to be based on the use of an intense line source, an inert-gas separated pre-mixed flame and a non-dispersive detection system. Chemical and physical interference effects, including those of scattering of source radiation, can be minimised by the use of a hot flame in conjunction with an efficient indirect nebuliser, which produces a small mean droplet diameter in the nebulised aerosol.166 CRESSER, KELIHER AND KIRKBRIGHT The detection limits currently attained with an inert-gas separated air - acetylene flame by using a purpose-built non-dispersive fluorescence spectrometer and also a dispersive instrument have been listed by Larkins4 for a wide range of TABLE I1 SOME ATOMIC FLUORESCENCE DETECTION LIMITS (SIGNAL: NOISE = 2) OBTAINED BY USING A NON-DISPERSIVE ATOMIC FLUORESCENCE SPECTROMETER Wavelength of main useful I fluorescence, Detection limit (p.p.m.) L Element Source Separated Conventional nm Silver Gold Cadmium Cobalt Iron Gallium Mercury Indium Iridium Magnesium Manganese Nickel Lead Palladium Rubidium Antimony Selenium Tin Tellurium Thallium Zinc HCL HIL(A) HCL VDL HCL HIL(A) HCL HIL(A) VDL HCL 02 VDL HCL HCL HIL(B) HCL HCL HIL(A) HCL HCL HCL HCL HCL HCL HCL HCL VDL HCL HIL(B) 6 0.005 0.004 0.0002 0-03 0*0015 0.05 0.003 6 30 0.07 1 4 0.0015 0~0002 0.01 0.03 0,002 1 10 80 2 6 3 3 10 0.3 0.02 0.0003 150 0.06 0.04 0.0015 0.4 0.015 0.5 0.04 20 300 15 30 0.7 0.009 0.0009 0.09 0.4 0.01 15 100 ND 11 90 40 40 100 1.5 0.2 0.003 193.7 242.8 228.8 228.8 240.7 240.7 248-3 248.3 294.4 253.7 253.7 303.9 254.4 285.2 285.2 279.5 232.0 232-0 283.3 244-8 257.5 217.6 196.0 284.0 214.3 276.8 276.8 213-9 213.9 HCL = Hollow-cathode lamp.HIL = High intensity hollow-cathode lamp of (A) the Lowe typeeo or VDL = Metal vapour discharge lamp. (B) Sullivan and Walsh types0. 0 2 = Philips type 0 2 4W mercury germicidal lamp. elements. Browner and Manning4' have also reported detection limits for twenty- four elements in separated and unseparated flames by using a modified atomic absorption spectrometer and hollow-cathode lamp excitation sources. It is inter- esting to note that, although different instrumentation was used by these workers,APPLICATION OF SEPARATED FLAMES IN ANALYTICAL ATOMIC SPECTROMETRY 167 the degree of improvement in fluorescence sensitivity on separation of the flame has generally been in quite good agreement when the results obtained with hollow- cathode lamp sources are compared, and that where the detection limits reported differ significantly, the instrumentation giving the lower detection limit generally showed flame separation to be less advantageous.Inert-gas separated air - acetylene flames have also been used for the AFS determination of a range of elements in several other investigations. Amos, Bennett and Brodiezl studied the effect of flame separation on the sensitivities attainable by AFS for silver, gold, cobalt, iron, molybdenum, nickel, lead and platinum, and other workers have studied gold,42943 zinc and cadmium,l* palladium,44 mangan- e ~ e , ~ ~ and mercury.46 In this latter study, the relatively high transparency of the separated flame at short wavelengths was shown to permit the observation of mercury stepwise fluorescence at 253.7 nm, excited by radiation from a mercury line source at 184.9 nm.46 AFS applications work involving the separated air - acetylene flame has been reported by Sychra and M a t ~ u s e k ~ ~ for the determination of nickel in gas oils and petroleum distillation residues; Hobbs, Kirkbright and WestP7 have described the use of this flame for the determination by AFS of bismuth in aluminium alloys.Norris and West4* have utilised a separated air - acetylene flame for the determination of cobalt and nickel in steels by AFS.With a multi-channel atomic fluorescence spectrometer utilising pulsed hollow- cathode sources and a nitrogen-separated air - acetylene flame, the determination of calcium, copper, magnesium, manganese and zinc in soil copper, iron, magnesium, manganese, nickel and zinc in aluminium alloys50 and cobalt, chro- mium, copper, iron, manganese and zinc in sea-water61 has been reported. Use of the Inert-gas Separated Nitrous Oxide - Acetylene Flame in Atomic Fluorescence Spectrometry The merits of the nitrous oxide - acetylene flame as a high-temperature atom cell for analytical atomic spectroscopy have already been described. The high flame temperature and reducing characteristics of the fuel-rich flame result in greater atomisation efficiency in this flame than in most other commonly used flames for those elements that form refractory oxides.For analytical AFS, how- ever, this advantage can be offset by the greater quenching efficiency of the flame gases in this hot flame than in flames such as the pre-mixed oxygen - hydrogen - argon flame. The atomic fluorescence characteristics of several refractory-oxide- forming elements were first described in argon or nitrogen-separated flames by Dagnall, Kirkbright, West and Wood;52-54 atomic fluorescence signals were not observed for titanium and zirconium, and a two to three-fold improvement in detection limits for silicon (251.6 nm), germanium (265.1 nm), aluminium (396.1 nm) and vandium (318.4 nm) was observed on flame separation by using a dispersive spectrometer. In the same studies, however, the AFS detection limit for molyb- denum at 313.2 nm, which lies within the OH band system whose principal head is at 306.4 nm, was improved by a factor of twenty when argon separation was used.158 CRESSER, KELIHER AND KIRKBRIGHT Electronically modulated microwave electrodeless discharge lamp sources, a grating monochromator with a reciprocal linear dispersion of 3.3 nm mm-1 and an a.c.detection system were used in this study to improve the discrimination between the fluorescence signal and flame background; even under these conditions the use of a separated flame is therefore worthwhile to prevent radiation from the secondary reaction zone from reaching the detector to increase ‘shot’ noise. Larkins and Willis5 have reported the use of the inert-gas separated nitrous oxide - acetylene flame for the AFS of several elements by using a non-dispersive spectrometer.These workers compared the results obtained by using a solar-blind photomultiplier detector, an interference filter - photomultiplier tube combination and a conventional grating monochromator - photomultiplier detection assembly. The high flame background intensity of the nitrous oxide flame, compared to the air - acetylene flame, led to inferior results with the non-dispersive system compared to the dispersive instru- ment ; when an interference filter - photomultiplier tube combination was used, the results were comparable with those using the dispersive instrumentation. Although the background from the secondary zone is suppressed on separation, the emission from the reducing species in the interconal zone of the fuel-rich flame is not de- creased under re-optimised conditions and may actually increase slightly in inten- sity.Thus Larkins and Willis5 confirmed earlier observations5* and obtained a two- fold improvement in the AFS detection limit on inert-gas separation for aluminium at 396.1 nm (near to CN band emission) but actually observed a fifteen-fold en- hancement in the detection limit for aluminium at 309.3 nm (within OH band emission). The inert-gas separated flame has been applied to the determination of silicon in steels by AFS.55 Use of Other Inert-gas Separated Flames in Atomic Fluorescence Spectrometry Inert-gas separation of the pre-mixed oxygen - hydrogen - argon,42*56*57 air - hydrogen20 and air - propane58 flames for use in atomic fluorescence spectrometry has been reported. The principal benefit of separation with the hydrogen flames is to remove background radiation from OH species in the secondary reaction zone.Although from a fundamental point of view these flames can give rise to greater fluorescence quantum efficiencies due to the low quenching cross-sections of the flame gas species, in practice the flame temperatures obtained may be insufficiently high to ensure freedom from interference and radiation scattering effects. Conclusion Although early work with both mechanically separated and inert-gas separated pre-mixed flames of hydrocarbons with air or nitrous oxide demonstrated their advantages in each of the three main techniques of analytical atomic spectrometry with dispersive systems, the most valuable application of these flames lies in atomic fluorescence spectrometry with non-dispersive systems, in atomic absorp-APPLICATION OF SEPARATED FLAMES IN ANALYTICAL ATOMIC SPECTROMETRY 159 tion spectrometry for work a t wavelengths less than 200 nm, and possibly in atomic emission spectrometry when simultaneous multi-element analysis is to be under- taken.1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 36 36 37 38 39 40 41 42 43 44 45 46 47 48 49 References Teclu, N., J . Prakt. Chern., 1891, 44, 246. Smithells, A., and Ingle, H., Trans. Chem. Soc., 1892, 61, 204. Kirkbright, G. F., Semb, A., and West, T. S., Talanta, 1967, 14, 1011.Larkins, P. L., Spectrochim. A d a , 1971, 26B, 477. Larkins, P. L., and Willis, J. B., ibid., 1971, 26B, 491. Marshall, G. B., and Smith, A. C., ‘Proceedings of the 3rd International Conference on Atomic Absorption Fluorescence Spectrometry,’ Paris, 197 1. Kirkrbight, G. F., and West, T. S., A p p l . Optics, 1968, 7, 1305. Cresser, M. S., and Keliher, P. N., Amer. Lab., 1970, 21. Sargent, M., Ph.D. Thesis, University of London, 1970. Kirkbright, G. F., Semb, A., and West, T. S., Spectrosc. Lett., 1968, 1, 7. -- , and -, Talanta, 1968, 15, 441. Hing’le, D. N., Kirkbright, G. F., and West, T. S., Analyst, 1969, 94, 864. Ure, A. M., International Atomic Absorption Conference, Sheffield, U.K., 1969. -, Proc. SOC. Analyt. Chern., 1970, 7, 192. Ure, A. M., and Berrow, M.L., Analytica Chim. Acta, 1970, 52, 247. Hingle, D. N., Kirkbright, G. F., and West, T. S., Talanta, 1968, 15, 199. Hobbs, R. S., Kirkbright, G. F., Sargent. M.. and West, T. S., ibid., 1968, 15, 997. Hingle, D. N., Kirkbright, G. F., Sargent, M., and West, T. S., Lab. Pract., 1970, 1069. Aldous, K. M., Ph.D. Thesis, University of London, 1970. Amos, M. D., Bennett, P. A., and Brodie, K. G., Resonance Lines, 1970,2(1), 3. Sychra, V., and Matousek, J., Talanta, 1970, 17, 363. Hobbs, R. S., Kirkbright, G. F., Sargent, M., and West, T. S., Analyst, 1969, 94, 554. Kirkbright, G. F., International Atomic Absorption Conference, Sheffield, U.K., 1969. Cresser, M. S., Keliher, P. N., and Joshipura, P. B., Spectrosc. Lett., 1970, 3, 267. Koirtyohann, S. R., and Pickett, E. E., Appl. Spectrosc. 1969, 23, 597. Boiteux, H., and Mavrodineanu, R., ‘Flame Spectroscopy,’ John Wiley, New York, 1964. Zeegers, P. J . T., Smith, R., and Winefordner, J. D., Analyt. Chem., 1968, 40, 2 6 ~ . Winefordner, J. D., McGee, W. W., Mansfield, J. M., Parsons, M. L., and Zacha, K. E., Vickers, T. J., Remington, L. D., and Winefordner, J. D., ibid., 1966, 36, 42 Kirkbright, G. F., Sargent, M., and West, T. S., Talanta, 1969, 16, 245. Kirkbright, G. F., and Vetter, S., Spectrochim. Acta, 1971, 26B, 505. Boumans, P. W. J. M., and de Boer, F. J., ibid., 1972, 27B, 351. Kirkbright, G. F., Sargent, M., and West, T. S., Talanta, 1969, 16, 1467. Kirkbright, G. F., and Ranson, L., Analyt. Chem., 1971, 43, 1238. Kirkbright, G. F., Ranson, L., and West, T. S., Spectrosc. Lett., 1972, 5, 25. Kirkbright, G. F., and Marshall, M., Analyt. Chem., 1972, 44, 1288. Kirkbright, G. F., West, T. S., and Wilson, P. J., Atom. Absorp. Newsl., 1972, 11, Kirkbright, G. F., and Marshall, M., Analyt. Chem., 1973, 45, 1610. Browder, R. F., and Manning, D. C., ibid., 1972, 44, 843. Sychra, V., and Matousek, J., Analytica Chim. Acta, 1970, 52, 376. Matousek, J., and Sychra, V., ibid., 1970, 49, 175. Sychra, V., Slevin, P. J., Matousek, J., and Bek, F., ibid., 1970, 52, 259. Ebdon, L., Kirkbright, G. F., and West, T. S., Talanta, 1970, 17, 965. Kirkbright, G. F., West, T. S., and Wilson, P. J., Analytica Chim. A d a , 1973, 66, 130. Hobbs, R. S., Kirkbright, G. F., and West, T. S., Talanta, 1971, 18, 859. Norris, J. D., and West, T. S., Analytica Chim. Acta, 1971, 55, 359. Dagnall, R. M., Kirkbright, G. F., West, T. S., and Wood, R., Analyt. Chem., 1971, 43, . , and -, ibid., 1968, 93, 522. -- Analytica Chirn. Acta, 1966, 36, 25. 3 , and -, Atom. Absorp. Newsl., 1969, 8, 34. -- 53 and 113. 1765.160 CRESSER, KELIHER AND KIRKBRIGHT 60 51 62 53 64 55 56 67 68 69 60 --- , and -, Analyst, 1972, 97, 245. Jane's. M., Kirkbright, G. F., Ranson, L., and West, T. S., Analytica Chzm. Ada, 1973, 63, Dagnall, R. M., Kirkbright, G. F., West, T. S., and Wood, R., ibid., 1969, 47, 407. --- , and -, Analyt. Chem., 1970, 42, 1029. Kirlcbrighi, G. F., Rao, A. P., and West, T. S., Analyt. Lett., 1969, 2, 466. Browner, R. F., Dagnall, R. M., and West, T. S., Analytica Chim. Ada, 1969, 46, 206. -- , and -, ibid., 1970, 50, 375. Daghall, R. M., Taylor, M. R. G., and West, T. S., Lab. Pvacl., 1970, 20, 209. Sullivan, J. V., and Walsh, A., Spectvochim. Acta, 1965, 21, 721. Lawe, R. M., ibid., 1971, 26B, 201. 210. D , , and -, Analyst, 1970, 95, 425. ---