Analyst, June, 1979, Vol. 104, fie. 505-515 505 Effect of Stray Light in Monochromators on Detection Limits of Flame Atomic-fluorescence Spectrometric Measurements R. G. Michel, M. L. Hail, S. A. K. Rowland, J. Sneddon and J. M. Ottaway Department of Pure and Applied Chemistry, University of Strathclyde, Cathedral Street, Glasgow, G1 1XL and G. S. Fell Department of Clinical Biochemistry, Royal Injirmary, Glasgow, G4 OSF Quantitative results are described that demonstrate that the use of a double monochromator to reduce stray light originating from strong thermal emission in the flame gives significant reductions in noise on the background of the fluorescence measurement. This leads to worthwhile improvements in detection limit for all elements with analytically useful resonance lines at wavelengths shorter than approximately 250 nm.The degree of improve- ment depends upon whether water or a real sample is being aspirated. For the determination of cadmium in urine the detection limit is improved by a factor of three and for the determination of selenium in water the improve- ment is a factor of 5-6. Scatter of excitation source radiation is also shown to have a small but significant effect on detection limits when using electrode- less discharge lamps as source. Scatter is more serious in the air - hydrogen flame than the air - acetylene flame. Keywords Flame atomic-fluorescence spectrometry ; stray light ; double monochromator The presence and effects of stray light in monochromators have been described for atomic emission,l atomic2 and molecular-3 absorption and Rarnan4s5 spectroscopy.Larson et aZ.1 have discussed the effect of stray radiation on background levels and hence on accuracy in atomic-emission spectroscopy. They also discussed the causes of stray radiation in mono- chromators and cited publications on the grating imperfections which are the primary cause.6 Here, the one quoted by Goode and Crouch2 is suitable, i.e., the stray radiation is expressed as the ratio of the spectrometer’s response to radiation within the monochromator’s band pass to its response to radiation outside the band pass. The most effective method of reducing stray radiation to negligible levels is to use a double monochromator. This method is almost universally used in Raman spectroscopy and has been suggested as a solution for atomic-emission spectroscopy1 and atomic-fluorescence spectroscopy.7~8 Barnett and Kahn7 found that high concentrations of sodium appeared to interfere with the flame atomic fluorescence of iron.They concluded that the intense sodium emission in the flame resulted in stray light in the monochromator. The intensity of the stray light was such that it saturated the electronics and caused an apparent interference. Haarsma et aZ.8 found that the noise associated with the flame background at the cadmium resonance line (228.8nm) could be reduced by use of a double monochromator. It was clear therefore that stray light originated from the flame background itself as well as from intense sodium emission in the flame. Flame atomic-fluorescence spectrometry has potential as a rapid means for the determina- tion of low levels of toxic metals in blood and urine.g This type of sample contains essential elements such as sodium, magnesium and calcium, which emit strongly in flames and which can cause stray light in monochromators.Stray light originating from the flame does not affect the accuracy of atomic-fluorescence measurements because modulation of the excitation source allows the fluorescence signal to be distinguished from all other signals originating in the flame. However, the noise associated with the stray light signal does degrade precision and detection limits. This can be contrasted with the modulated excitation source radiation, which, when scattered by incompletely dissociated matrix particles present in the flame, A number of definitions of stray radiation have been used in the past.506 MICHEL et nl.: EFFECT OF STRAY LIGHT IN MONOCHROMATORS Analyst, VoZ. 104 does affect the accuracy of the measurement. Scatter signals can be corrected9 but cannot be prevented from reaching the detector if excitation and fluorescence take place at the same wavelength (resonance fluorescence). Noise on scatter signals therefore also degrades detection limits. An attractive means of avoiding scatter noise is to use direct line fluorescence10 or similar transitions where excitation and fluorescence occur at different wavelengths. With microwave-excited electrodeless discharge lamp (EDL) excitation the use of direct line fluorescence often leads to an unacceptable loss in sensitivity, which is not usually balanced by the reduction in scatter noise.For some elements laser excitation does provide adequate sensitivity for direct line fluorescence measurements.11 In summary, it can be seen that, with modulated EDL excitation, accuracy can be ensured by using a background correction system but it is not practical to reduce the noise associated with the scatter signal. However, it is possible to improve precision and detection limits by minimising the contribution of stray light to background noise. Quantitative results are described here that demonstrate that the use of a double mono- chromator to reduce stray light gives significant reductions in the noise on the background of the fluorescence measurement. This leads to worthwhile improvements in detection limit for all elements with analytically useful resonance lines at wavelengths shorter than approximately 250 nm.The double monochromator is particularly useful for the reduction of stray light originating from strong thermal emission in flames caused by the aspiration of biological samples such as blood and urine. Experimental Instrumentation A double monochromator is two similar or identical monochromators which are arranged in tandem and which disperse the incident radiation twice. The second of the two mono- chromators is therefore able to re-disperse the stray light that is present at the exit slit of the first monochromator with the result that there is a negligible amount of stray light remaining at the exit slit of the second monochromator.As a consequence of the double dispersion the spectral band pass of the double monochromator is normally half the spectral band pass of the equivalent single monochromator if the same slit width is used on both. The same band pass can be achieved by doubling the slit width on the double monochromator. The exit slit of the first monochromator and entrance slit of the second monochromator of the tandem are normally physically the same slit (middle slit) in a true double mono- chromator. There are twice the usual number of mirrors and two gratings in a double monochromator, and therefore some light losses are expected relative to the equivalent single monochromator. The instrumentation that was used here has been more fully described in a previous publicati~n.~ Microwave-excited electrodeless discharge lamps were used as excitation sources.Cadmium lamps were made in the laboratory using the method published else- ~here.129~3 Selenium lamps were prepared using the same method but optimised specially for this element.14 The instrument included a two-source background correction system that employed a 300-W high-pressure xenon arc as the second source. All flames were supported on a circular capillary burner with facilities for flame separation. In order to determine the effect of using a double monochromator, it was necessary to compare all measurements with those obtained. using an equivalent single monochromator. The two monochromators chosen were those manufactured by Spex Industries Inc. (Metuchen, N.J., USA). These were the Doublemate double monochromator and Minimate single monochromator. The Minimate is in every respect identical with half of the Doublemate and hence clear comparisons could be made. The detailed specifications of these mono- chromators, taken from the manufacturer’s literature, are shown in Table I. Fluorescence, scatter, flame background and stray light signals were all measured using a photon counter in the digital lock-in mode (the Ortec-Brookdeal 5 C1/5C21 photon counting system was used routinely with a 1 s count time). Simultaneous measurements of flame background on the double monochromator and the single monochromator were made by placing them on opposite sides of the same flame andJune, 1979 ON DETECTION LIMITS OF FLAME AFS MEASUREMENTS 507 taking the signals from the two photomultiplier tubes to the two separate channels on the photon counter.The photomultiplier tubes used were as near identical as possible and both EM1 9789QB tubes were operated at 1100 V. TABLE I SPECIFICATIONS OF MONOCHROMATORS Manufacturer, Spex Industries, Inc., Metuchen, N. J , , USA. Monochromator Focal length . . .. . . .. Aperture . . .. .. .. .. Grating . . .. . . . . .. Blaze . . .. .. . . .. Dispersion .. . . . . .. Resolution .. . . . . .. Near stray light (slits 5 nm band pass, within 3.0 band widths) . . .. Far stray light (> 10 band widths) Configuration . . .. .. .. Slits .. .. .. .. .. Band pass .. .. .. .. . . 1670 Minimate single monochromator 220 mm 1200 grooves mm-1 300 nm 4 nm mm-l 1 nm f14 1672 Dou blemate double monochromator 220 mm fI4 (2) 1 200 grooves mm-1 300 nm 2 nm mm-1 0.5 nm 10-s 10-4 5 x 10-4 10-9 In-line Czerny - Turner Double in-line Czerny - Turner Five pairs 20 nm high, 0.25, 0.5, 1.25, 2.5 and 5 mm 1-20 nm 0.5-10 nm Reagents High-purity de-ionised water acidified to 0.04 M with hydrochloric acid was used for bIank measurements and is referred to as water throughout this paper.Urine samples were obtained as discussed elsewhere9 and were normally from persons exposed to cadmium in their workplace. Urine was also acidified to 0 . 0 4 ~ with hydro- chloric acid. The acid was of AnalaR grade. New batches of acid were checked for cadmium and selenium contents. Undiluted rather than diluted urine was aspirated into the flame. Results and Discussion When using a single monochromator (SM) it is not possible to distinguish easily between true flame background at a particular wavelength and the stray light that is a result of flame background at wavelengths removed from the monochromator wavelength setting.However, it can be assumed that the double monochromator (DM) can discriminate against stray light. Simultaneous measurements of flame background using an SM and a DM then reveal differences that are due to stray light alone. Variation of Flame Background and Stray Light with Wavelength Fig. 1 shows the variation in magnitude of the flame background with wavelength for a nitrogen-separated air-acetylene flame when measured on the DM and the SM. The difference between the two spectra shows that stray light was a significant component of the flame background at short wavelengths.De-ionised water acidified to 0.04 M in hydro- chloric acid was aspirated into the flame to obtain Fig. 1. In Fig. 2 urine was aspirated instead of water and again the stray light difference between the double and single mono- chromators can be seen. The results in Fig. 3 are taken from Figs. 1 and 2 and are the flame background as seen by the double monochromator, L e . , stray light was not present. Accordingly, the difference in background between water and urine being aspirated was a real change in flame background. At wavelengths longer than 300 nm this change leads to increased noise and degraded detection limits when aspirating urine rather than water. However, there was no shift in flame background below 300 nm.This result indicates that the only increase in noise to be expected upon aspiration of urine is caused by scatter of excitation source radiation if a double monochromator is used and elements with resonance lines below 300 nm are being determined.508 MICHEL et d. : EFFECT OF STRAY LIGHT IN MONOCHROMATORS Analyst, VOl. 104 300 400 500 600 - u 200 Wavelength/nm Fig. 1. Flame background with water. Flame background was measured a t 10-nm intervals while aspirating water into a stoicheio- metric nitrogen-separated air - acetylene flame. Spectral band pass on both monochromators was 1 nm. 0, Single monochromator; and @, double monochromator. On this and subsequent figures, Na, K and Ca indicate the position of atomic lines and OH the band spectrum of this radical, which are superimposed on the flame background .1 200 360 400 500 600 Wavelengthhm Fig. 2. Flame background with urine. Flame background was measured a t 10-nm intervals while aspirating urine into the separated air - acetylene flame. Spectral band pass on both monochromators was 1 nm. 0, Single monochromator; and @, double monochromator. Improvement in Flame Noise Levels by Use of the Double Monochromator It was possible to demonstrate the improvement in flame background noise levels by using the results in Figs. 1 and 2. It was assumed that the noise on the flame background was white or shot noise and therefore the noise could be calculated by taking the square root of the flame background measured in counts per second.Prior measurements of various parameterss had indicated that this assumption was correct a t 228.8nm when using both monochromators. Shot noise was probably also dominant throughout the low background region of the flame. Fig. 4 shows the noise improvement with wavelength. The noise improvement factor is the ratio of the noise as measured in the single monochromator to the noise on the double monochromator, where noise is measured as the square root of the flame backgrounds taken from Figs. 1 and 2 at each wavelength plotted. Improvements in the shot noise as a result of using the double monochromator are the smallest improvements that could be expected from the large reductions in flame background that were observed. 200 300 400 500 600 Wavelengthhm Fig.3. Flame background water and urine. Flame background was measured a t 10-nm intervals while aspirating either urine or water into a stoicheiometnc :nitrogen-separated air - acetylene flame. Detection using the double monochromator with a spectral band pass of 1 nm. 0, Water; and a, urine.June, 1979 ON DETECTION LIMITS OF FLAME AFS MEASUREMENTS 509 If proportional noise15 had made important contributions to flame background noise it would have decreased linearly with the flame background. Therefore, the improvements in proportional noise were the largest that could have been expected. From Fig. 4 it can be seen that the improvement in shot noise is much greater for urine than for water. This was a result of the intense emission from the flame when aspirating urine and which causes greater stray light levels.The improvement was only significant, ie., greater than a factor of 2, at wavelengths shorter than 250 nm and varied from a factor of 25 at 200nm to a factor of 2 at 250nm for urine. When aspirating water the corre- sponding figures were 8 at 200 nm and 2 at 225 nm. From Fig. 3 it is clear that the major portion of the stray urine emission from the flame probably originates from increases in flame background over a broad band of wavelengths above 300 nm, together with contributions from sodium, potassium and calcium atomic line emission. b c, m + c, I! E 3 P .- .- 0 z Wavelengthhrn Fig. 4. Noise improvement, single to double mono- chromator. Ratio of noise in single monochromator to noise in double monochromator using results from Figs.1 and 2 (see text). Stoicheiometric nitrogen-separated air - acetylene flame. Spectral band pass 1 nm. 0, Water; and 0, urine. Improvement in Flame Background Noise at Particular Wavelengths Atomic fluorescence and flame background measurements were made at the selenium and cadmium resonance lines at 204.0 and 228.8 nm, respectively. Slit width and hence spectral band pass were varied in order to determine differences in flame background more accurately. Measurements of flame background at only one spectral band pass (Fig. 4) are biased by relative differences in slit width and other physical parameters between the two mono- chromat ors. Figs. 5 and 6 show the variation of flame background with spectral band pass for the two elements and for both monochromators while aspirating urine or water. As predicted in Fig.3 negligible background shift was observed with the DM in changing from water to urine. However, large shifts caused by stray light were observed when using the SM. In Fig. 5 the background at 228.8 nm (cadmium) while aspirating urine was 17 times greater in the SM than in the DM, which indicated a reduction in flame noise of 1/17 or 4.1 when using the double monochromator. When aspirating water the background was 3.8 times greater in the SM than the DM, i.e., a reduction in flame noise of 1.9 when using the double monochromator. Fig. 6 shows similar results for selenium, which indicate a 1/562 or 24 times reduction in flame noise when using the double monochromator and aspirating urine.A 1/31 or 5.6 reduction in flame noise was obtained when aspirating water, Shifts in flame background caused by scatter of excitation source (EDL) radiation have been corrected in Figs. 5 and 6. The magnitudes of the scatter signals are discussed in relation to detection limits in a later section.510 MICHEL et aZ. : EFFECT OF STRAY LIGHT IN MONOCHROMATORS Analyst, VoZ. 104 Spectral band pass/nm rn + C : Y lo4- 2 l o 3 - D 3 m Y n Fig. 5. Background versus spectral band pass for cadmium. Flame back- ground measured at the cadmium resonance line (228.8 nm) using both single and double monochromators. Separated air - acetylene flame. Ratios of flame backgrounds in single mono- chromators to double monochromators are 17 for urine and 3.8 for water. A, Single monochromator with urine; B, single monochromator with water ; and C, double monochromator for urine and for water.Slope of each line = 2. Spectra I band pass/nm Fig. 6. Background vevsus spectral band pass for selenium. Flame back- ground measured at the selenium line (204.0 nm) using both single and double monochromators. Separated air - acet- ylene flame. Ratios of flame back- grounds in single monochromators to double monochromators are 562 for urine and 31 for water. A, Single monochromator with urine; B, single monochromator with water; and C, double monochromator with urine and with water. Light Losses in the Double Monochromatar The cadmium fluorescence signals obtained simultaneously with the background measure- ments of Fig. 4 are shown in Fig. 7. These results show that at the same spectral band pass the fluorescence signals were the same in both monochromators.If there had been no light losses as a result of using the DM rather than the SM then the signals obtained in the DM would have been twice those obtained in the SM. This is because at the same spectral band pass the slit width of the DM is twice the slit width of the SM. As the signals turned out to be equal in both monochromators then light losses were close to 50%. Light losses did not affect detection limits because at the same spectral band pass there were no effective light losses between the two monochromators. There was an approximately 6% difference in sensitivity between the two photomultiplier tubes attached to the two monochromators. A correction for this difference has been incorporated into Figs.1-3 and 7. Scatter of line source radiation, when detected, is effectively similar in nature to the atomic fluorescence signal. One consequence was that when urine was aspirated into the flame scatter signals were of the same magnitude in both monochromators at the same spectral band pass. The effect of the scatter signals on detection limits, however, was different in the two monochromators because of the differences in the flame background. Effect of the Double Monochromator on Detection Limit At the detection limit the noise on the fluorescence signal approaches in magnitude the noise on the background.16 Therefore, the decreases in flame noise brought about by using the double monochromator should improve detection limits in proportion, if flame noise is the limiting noise.For example, the reduction in flame noise by a factor of 4.1, which was obtained for the aspiration of urine (cadmium, Fig. 5) should improve detection limits by aJune, 1979 ON DETECTION LIMITS OF FLAME AFS MEASUREMENTS Io4 [- 51 1 I " 0.5 1.0 2.0 5.0 10.0 Spectral band p a d n rn Fig. 7. Cadmium fluorescence vey- sus spectral band pass. Cadmium fluorescence signals from 4 pg 1-1 of cadmium in water (228.8 nm). Stoi- cheiometric separated air - acetylene flame measured using both single (0) and double (0) monochromators. factor of 4.1.8,However, the scatter of excitation source radiation had a significant effect on detection limits for determinations of both cadmium and selenium in urine.Table I1 shows some typical figures for the magnitude of flame background and scatter at a l-nm spectral band pass on both DM and SM. The scatter signals given are average figures TABLE I1 EFFECT OF FLAME BACKGROUND AND SCATTER ON DETECTION LIMITS USING THE SINGLE OR DOUBLE MONOCHROMATORS Element Parameter Cadmium (228.8 nm) . . Double monochromator, DMS Single monochromator, SMS Improvement factors, SM/DM- in background in noise in detection limit Selenium (204.0 nm) . . Double monochromator, DM Single monochromator, SM Improvement factors, SM/DM- in background in noise in detection limit Flame back- ground*/ Scatter/ counts s-l counts s-1 500 500 50 460 1900 8500 50 460 - 3.8 17 - 1.95 4.1 - - 40 40 4 30 1240 22480 4 NDY - 31 562 - 5.6 24 - - Detection limitt/ PLg 1-1 0.2 0.26 0.37 0.81 - - - 1.9 3.1 160 210 890 3800 - - 5.6 19 * Flame background for stoicheiometric separated air - acetylene flame.Flame background increases for leaner or richer fuel conditions and affects detection limits appreciably. t Detection limit for signal to noise ratio of 2 where noise is expressed as the square root of the total background (flame background plus scatter). : Urine samples were those of workers exposed to cadmium in their work place. Scatter signals of persons unexposed to cadmium tend to be lower by about half. 5 Spectral band pass 1 nm for both monochromators. 7 ND = not detectable.512 MICHEL et al. : EFFECT OF STRAY LIGHT IN MONOCHROMATORS Analyst, Vol. 104 which varied from urine to urineg and which were measured using the two-source background correction facility which was incorporated into the instr~ment.~ Detection limits (signal to noise ratio = 2) in Table I1 were estimated by measuring signals close to the detection limit and calculating noise by taking the square root of the total background, i.e., flame back- ground plus scatter signal.It can be seen in Table I1 that the improvement in detection limit in going from the SM to the DM was not as great as the improvement in flame background noise. This was a result of the addition of the scatter noise into the total noise. The scatter noise was signifi- cant only for urine. The aspiration of water caused little scatter off water droplets and flame gases as shown in Table 11. Despite the scatter noise component the double mono- chromator improved detection limits for cadmium in urine by a factor of 3 and for selenium in urine by a factor of 19.As a result, the determination of cadmium in urine became feasible, because although the concentration of cadmium in the urine of persons unexposed to cadmium is close to 0.2 pg l-l, it was possible to determine cadmium in the urine of persons exposed to cadmium (cadmium present at concentrations above 2 pg 1-l) by direct aspiration of un- diluted urine into the flame.9 It was not possible to determine selenium in this way because this element is present at levels below 1 pug 1-l. Selenium is discussed in this paper simply to demonstrate the utility of the double monochromator for the analysis of real samples, with urine taken as a convenient example.It is apparent from these results that the considerable lowering of the flame background by discrimination against stray light in the DM caused the scatter noise to have an approxi- mately equal effect, with flame noise, on the total noise and hence on the detection limit. In contrast, when using the SM the scatter noise was negligible relative to flame noise, which was then the limiting noise at the detection limit. With the double monochromator scatter was important relative to the flame background for both cadmium at 228.8 nm and selenium at 204.0 nm. It appears, therefore, that scatter will probably be important for all elements with resonance lines at wavelengths shorter than 250 nm where flame background is low and where the corresponding EDLs tend to perform well in terms of atomic-fluorescence detection limits.For elements with resonance lines at I 200 300 400 500 600 Wavelength/nrn Fig. 8. Background in three flaines with water. Using the double monochromator, flame backgrounds were measured at 10-nm intervals while aspirating water into either a stoicheiometric separated air - acetylene flame (0) or the hydrogen flames operated under fuel-lean conditions ; 0, air - hydrogen flame ; and 0, separated air - hydrogen flame. (Optimised for cadmium atomic fluorescence measurements.) Spectral band pass 1 nm.Jane, 1979 ON DETECTION LIMITS OF FLAME AFS MEASUREMENTS 513 wavelengths longer than 250 nm, where the flame background is high and not improved by the DM, scatter will probably not determine the detection limit even with EDLs of high radiant output. This is primarily because cadmium EDLs are normally superior to EDLs of other metals and therefore scatter signals caused by a cadmium EDL are the largest likely to be experienced.Furthermore, scatter of ultraviolet radiation is known to be of greater magnitude than scatter of longer wavelength radiation. If continuum or line excitation sources of greater intensity than EDLs are used then scatter noise will amost always domi- nate. This effect has been discussed elsewhere for laser excitation.17 Choice of Flame The variation in flame background with wavelength for three flames was obtained when using the double monochromator. These backgrounds are shown in Fig. 8 (aspiration of water) and Fig. 9 (aspiration of urine).As would normally be expected the nitrogen- separated air - hydrogen flame had the smallest background at all wavelengths. The separated air - acetylene flame had a background that was lower than the unseparated air - hydrogen flame when aspirating water (Fig. 8) but higher than the same flame when aspirating urine. It appeared, therefore, that the low background, and hence low noise, separated air - hydrogen flame would give the best detection limits when the fluorescence signals are similar in all flames. However, the hydrogen flames have a low temperature and probably do not dissociate matrix particles as efficiently as the air - acetylene flame, which leads then t o increased scatter of excitation source radiation. 200 300 400 500 600 Wavelength/nm Fig.9. Background in three flames with urine. Using the double monochromator flame backgrounds were measured a t 10-nm intervals while aspirating urine into the same flames as in Fig. 8. Spectral band pass 1 nm. 0, Separated air - acetylene flame; 0, air - hydrogen flame; and 0, separated air - hydrogen flame. The increased scatter in hydrogen flames was demonstrated by the results shown in Table I11 for which total background, i e . , flame background plus scatter, was recorded in all three flames at the wavelength of the cadmium resonance line (228.8nm). The detection limits that resulted from these backgrounds are also shown in Table 111. There was little scatter when aspirating water and therefore the separated air - hydrogen flame gave the best detec- tion limit.When urine was aspirated the detection limits deteriorated more in hydrogen flames than in the acetylene flame as a result of the noise on the larger scatter signals in the hydrogen flames. The nitrogen-separated air - acetylene flame is therefore the best choice when analysing urine and will also be a good choice for most other matrices.514 MICHEL et al. : EFFECT OF STRAY LIGHT I N MONOCHROMATORS Analyst, VOl. 10 By changing the fuel flow-rate it was possible to improve cadmium atomic-fluorescence signals and hence detection limits in the hydrogen flames by up to a factor of two. How- ever, the flame was then lean and the burner became less resistant to the formation of salt deposits in the orifices of the burner head. These deposits rapidly caused instability of the fluorescence signals.The cadmium fluorescence signal changed little with changes in fuel conditions in the air - acetylene flame. The optimum signal to noise ratio was obtained in the stoicheiometric flame, where the flame background was at a minimum. TABLE: I11 RELATIVE DETECTION LIMITS OF CA:DMIUM BY ATOMIC-FLUORESCENCE SPECTROSCOPY I N THREE FLAMES Detection limits defined as in Table I1 and the text. Signals in all three flames were approximately the same. Fuel conditions as described in caption to Fig. 8 and in the text. Flame background (228.8 nm*)/ counts s-l Detection limitlpg 1-1 A 3 Separated - 3 d (Separated Separated Sample air - C,H, Air - H, air - H, air - C,H, Air - C,H, air - H, Water . . .. . . 660 1250 70 0.2 0.36 0.09 Urine .. .. . . 960 1950 990 0.26 0.45 0.35 * Double monochromator, spectral band pass 1 nm. Conclusions The use of a double monochromator to reduce stray light originating from strong thermal emission in the flame gives significant improvements in detection limits for all elements with resonance lines at wavelengths shorter than approximately 250 nm. The degree of improve- ment depends upon whether water or a real sarnple such as urine is being aspirated. When water is aspirated the flame background at all wavelengths causes stray light, which is reduced by using the double monochromator. The improvements obtained with urine are much greater than for water because urine in the flame causes intense emission over a broad band of wavelengths, as well as line emission from sodium, potassium and calcium, which all contribute to stray light at analytical wavelengths shorter than 250 nm.It is possible to estimate the performance required of a monochromator in discriminating against stray light by considering the total amount of light passing through the entrance slit. With urine in the flame and thej/4 double monochromator this is approximately 108 counts s-1. This figure was obtained by integrating the flame background over all wavelengths using the data in Fig. 2. A double monochromator with a far stray light specification of is therefore probably more than adequate to reduce stray light to insignificant levels. A normal ruled grating single monochromator has a typical specification of 10-4 and a holographic grating single monochromator has a typical specification of The latter will clearly reduce stray light to lower levels but will not be as effective as the double monochromator.With a double monochromator scatter of (EDL) excitation source radiation has a small but significant effect on detection limits at wavelengths shorter than 250 nm. With a single monochromator scatter is not important because the flame background, which includes the stray light component, is high and the noise on the scatter signal is small in comparison. However, the same inaccuracies caused by the scatter signal will occur when using either monochromator and, to compensate, a scatter correction must be applied. Scatter is more serious in the air - hydrogen flame than in the air - acetylene flame. This degrades detection limits more in air - hydrogen flames than in the air - acetylene flame and favours the latter for routine use.The authors acknowledge the support of the Scottish Home and Health Department for the purchase of the major items of equipment used in this project and for the award of a Postdoctoral Fellowship (to support R.G.M.). We also thank the Eastern District, GreaterJune, 19 79 ON DETECTION LIMITS OF FLAME AFS MEASUREMENTS 515 Glasgow Health Board, for a maintenance grant in support of J.S. and the HSE, Employ- ment Medical Advisory Service, for a maintenance grant in support of M.L.H. Practical and equipment support of this work has also been given by Pye Unicam Ltd. and this is p a t ef ully acknowledged. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. References Larson, G. F., Fassel, V. A., Winge, R. K., and Kniseley, R. N., Appl. Spectrosc., 1976, 30, 384. Goode, S. R., and Crouch, S. R., Analyt. Chem., 1974, 46, 181. Slavin, W., Analyt. Chem., 1963, 35, 561. Stamm, R. F., and Salzman, C. F., Jr., J . Opt. SOC. Am., 1953, 43, 126. Leite, R. C. C., and Porto, S. P. S., J . Opt. SOC. Am., 1964, 54, 981. Sharpe, M. R., and Irish, D., Optica Acta, 1978, 25, 861. Barnett, W. B., and Kahn, H. L., Analyt. Chem., 1972, 44, 935. Haarsma, J. P. S., Vlogtman, J., and Agterdenbos, J., Spectrochim. Acta, 1976, 31B, 129. Michel, R. G., Hall, M. L., Ottaway, J. M., and Fell, G. S., Analyst, 1979, 104, 491. Kirkbright, G. F., and Sargent, M., “Atomic Absorption and Fluorescence Spectroscopy,” Weeks, S. J., Haraguchi, H., and Winefordner, J. D., Analyt. Chem., 1978, 50, 360. Michel, R. G., Coleman, J., and Winefordner, J . D., Spectrochim. Acta, 1978, 33B, 196. Michel, R. G., Ottaway, J. M., Sneddon, J., and Fell, G. S., Analyst, 1978, 103, 1204. Michel, R. G., Ottaway, J. M., Sneddon, J., and Fell, G. S., Analyst, in the press. Chester, T. L., and Winefornder, J. D., Analyt. Chem., 1977, 49, 119. Winefordner, J. D., Schulman, S. G., and O’Haver, T. C., “Luminescence Spectrometry in Analytical Green, R. B., Travis, J. C., and Keller, R. A., Analyt. Chem., 1976, 48, 1954. Academic Press, New York, London, 1974. Chemistry,” Wiley-Interscience, New York, 1973, p. 149. Received September 26th, 1978 Accepted November 21st, 1978