- Email: [email protected]
Simultaneous multi-element analysis in a commercial graphite furnace by diode laser induced fluorescence
S ecmchimica Acfu.Vol. 47B.No. 14,pp. lSlW524, 1992 Pfinsd in GreatBritain.
0
0584-8547/92 $5.00t .oo
1992 Pergsmon RessLtd
TOPICS IN LASER SPECTROSCOPY
Simultaneous multi-element analysis in a commercial graphite furnace by diode laser induced fluorescence A.
ZYBIN,*
C. SCHN~~RER-PATSCHAN and K.
NIEMAX~
Institut fur Spektrochemie und Angewandte Spektroskopie (ISAS), Bunsen-Kirchhoff-Str. 11, D-4600 Dortmund, F.R.G. (Received and accepted 29 July 1992) Abstract-A powerful and compact experimental arrangement for simultaneous multi-element measurement by laser induced fluorescence is presented. The analytes are excited in a commercial graphite furnace atomixer by cw semiconductor diode lasers and the fluorescence is detected by a simple photodiode without interference filters or a polychromator. In preliminary measurements, detection limits of 1 pg ml-’ (10 fg absolute) and 2 pg ml-’ (20 fg) for lithium and rubidium, respectively, have been obtained.
1.
INTR~DUCTIT~N
ALTHOUGH
they are known to be very sensitive, laser spectroscopic techniques are still not accepted by analytical chemists. The reasons are obvious. Tunable lasers, in particular pulsed or cw dye lasers, are expensive instruments to purchase as well as to operate. The dye laser stability and reproducibility is poor and they require a lot of space in the laboratory. While these reasons will hamper the transfer of laser spectroscopic techniques from the research level to laboratories for routine analysis, it is even more unlikely that many tunable dye lasers can be applied simultaneously for multi-species analysis. However, the rapid progress and the new developments in the field of laser diodes opens up the possibility of transferring laser spectroscopic techniques to analytical laboratories for routine analysis [l]. For the last seven years we have been applying diode lasers together with dye lasers in analytical spectroscopy in our laboratories in Dortmund. Some time ago we published the first paper on simultaneous two-element analysis using only diode lasers [2]. In that paper, diode lasers were successfully used to replace hollow-cathode lamps in atomic absorption spectroscopy using a commercial graphite furnace atomizer. With diode lasers there is not only the possibility of simultaneous determination of several analytes in the sample, but, because of the fast tunability of diode lasers, the background can be measured and the dynamic range extended by the measurement of absorption in the line wings. By using the possibility of frequency doubling in non-linear media, laser atomic absorption spectroscopy (LAAS) with diode lasers already allows the determination of a large number of elements [3]. Recently, we have set up a diode laser spectrometer with six laser diodes in our laboratory and determined simultaneously six elements by LAAS [4]. Typical laser spectroscopic techniques, as, for example, laser induced fluorescence (LIF) or the methods of laser ionization spectroscopy, are more sensitive than LAAS. First demonstrations of diode laser induced fluorescence in a graphite furnace [5] and in an analytical flame [63 have been published. In both papers, rubidium was the analyte. LIF was measured by a photomultiplier fixed to the exit of a monochromator. l Instituteof Spectroscopy, Russian Academy of Sciences, Troitxk, Russia. t Author to whom correspondence should be addressed.
1519
A. ZKMN et al.
1520
laser diodes no. 1 no. 2
T out
reference
Fig. 1. Experimental
apparatus.
In this paper, we will demonstrate a very simple, much more sensitive method for analytical multi-element measurement than LAAS: laser induced fluorescence (LIF) in commercial graphite furnaces with a simple photodiode detector. We will show that the application of diode lasers of moderate cw power allows simultaneous detection of elements in the lower pg g-i range without the use of dispersive units like polychromators or interference filters with radiation detectors. Lithium and rubidium were selected as analytes.
2. EXPERIMENT The experimental arrangement is shown in Fig. 1. Two diode lasers were applied, one laser diode for excitation of the lithium resonance lines at 670.776 nm (Hitachi 6712, cw power: 5 mW) and another laser diode for the rubidium D2 line at 780.027 nm (Hitachi 7851, cw power: 50 mW). The laser diodes were tuned to the atomic transitions and stabilized by temperature. The long-term wavelength stability of the diode lasers were better than -CO.4 pm. Since the widths of the rubidium and lithium resonance lines broadened by the argon buffer gas in the graphite furnace atomizer were much larger than the possible drifts of the laser wavelengths, within the experimental uncertainties, no systematic errors due to wavelength drifts of the diode lasers were observed during the measurements. In the first part of the experiment, the laser diodes were operated sequentially. The laser beams were collimated by large aperture lens systems in front of the diode lasers (not shown in Fig.1) and directed through a hole in a plane mirror into a commercial graphite tube furnace (Perkin-Elmer HGA 74). In order to get more effective atomization conditions in the case of lithium, the graphite furnace was plated inside with a tantalum foil. The laser beams were modulated by a high-frequency chopper (3 kHz). The fluorescence radiation was collected by the plane mirror and focused by a lens system onto a photodiode (Hamamatsu S 2386-X) through an aperture (diameter: 1.5 mm). The solid angle of observation was about 15”, the imaging ratio was one to one. The photodiode used had a spectral response in the range 400-1100 nm. The radiant sensitivity was 0.6 A W-r at maximum (908 nm) and the dark current about 5 pA. The photodiode was operated with a 2 kfi load resistor. The signal was fed into a lock-in amplifier (Itaco 3961B). A time constant of 100 ms was chosen. The amplified signals were recorded by a storage oscilloscope (Hameg HM 205-3). The data could be transferred by a serial interface into a personal computer. The resolution of the signals by the computer was moderate (8 bit). In the second part of the experiment, where simultaneous LIF-measurements of rubidium and lithium were demonstrated, the beams of both lasers were alternatively chopped by a lowfrequency chopper (about 10 Hz) shown as a dashed symbol in Fig. 1. Because of the 10 Hz
Simultaneous multi-element analysis
7
6 ._ ce a
1521
100 SO
20 0 0
2
1
4
3
6
5
I
Time (s) Fig. 2. LIF signals of rubidium.
low-frequency modulation, the -time constant of the lock-in amplifier had to be reduced from 100 ms to 10 ms in comparison with the single element measurements. In order to avoid scatter of laser radiation from the windows of the furnace, the graphite tube furnace was operated without windows. The gas flow of the protect’ion gas (argon) was typically 5-10 1 min-‘. Internal gas stop was applied during the atomization phase. Aqueous solutions of 10 ~1 were injected into the furnace. The solutions had been prepared in clean rooms with double-distilled water. The smallest concentration of lithium and rubidium was 10 pg ml-‘, respectively.
3. RESULTS AND DISCUSSION Typical LIF signals for 100, 50 and 20 pg ml-l rubidium are shown in Fig. 2. These are net-signals from the lock-in amplifier stored by the computer. However, it is interesting to show the signals of the photodiode before processing without modulation by the chopper. Figure 3 displays the detector signal plotted against time during the atomization phase. The full curve is the radiation from the furnace including the fluorescence radiation from 20 ng ml-’ lithium, while the dashed curve is the signal from the blank (furnace only). Note that the absolute mass of lithium is 200 pg. The calibration curves of rubidium and lithium, derived from the peak of the LIF
0
12
3
4
3
6
7
8
9
Time (s) Fig. 3. Photodiode signal of the total radiation from the ftirnace. Full curve: furnace radiation plus LIF from 20 ng ml-’ (200 pg absolute) lithium; dashed curve: radiation from furnace only.
1522
A. ZY~IN et al.
1
10
100
1000
Rb-concentration
10000
lOooo0
(pg ml-t)
Fig. 4. LIF-calibration curve of rubidium obtained by laser excitation of the D2 line at 780.027 nm. Two different sets of solutions (0,O) were used for calibration. (m) is the signal of the blank. The 3o-detection limit is 2 pg ml-‘.
signals, are shown in Figs 4 and 5, respectively. Two sets of rubidium solutions were measured. The agreement of the data was good. The detection limit @u-criterion) of rubidium was found to be 2 pg ml-l, which corresponds to 20 fg in the furnace. The blank level of rubidium was of the order of the detection limit. Please note, that our present LIF detection limit for rubidium found with the simple broad-band photodiode is a factor of 200 and 100 better than the detection limits for rubidium in a graphite furnace [5] and in an analytical flame [6], respectively, if a monochromator and a photomultiplier are applied. The detection limit of lithium was about 1 pg ml-l (10 fg lithium). As shown in Fig. 5, the blank was about 3 pg ml-‘. The L&measurement was made with 5 mW of laser power, while the power of the laser diode for the Rb-measurement was about 40 mW. In lithium we could not saturate the transition optically; however, we could observe slight optical saturation of the rubidium signal at maximum power. This is shown in Fig. 6, where the Rb-signal is plotted against the intensity (intensity at 40 mW: 100%). It has to be noted that, in the present experiment, the detection power was limited by intrinsic electronic noise at the detector side (photodiode and lock-in amplifier). Therefore, we expect further improvements of the detection limits with optimized radiation detection. However, laser diodes with higher powers will also improve the detection limits. Better detection limits should be expected, in particular, for lithium. In order to compare the detection power of these simple LIF measurements with absorption measurements, we performed LAAS measurements with the same
0.001
Y’ 0.5
I 5
50
Li-concentration
500
5000
50000
(pg ml-*)
Fig. 5. LIF-calibration curve of lithium obtained by laser excitation of the lithium resonance line at 670.776 nm. (m) is the signal of the blank. The 3u-detection limit is 1 pg ml-r.
Simultaneous multi-element
0
20
60
40
Laser
80
intensity
1523
analysis
106
(%)
Fig. 6. LIF-signal of rubidium plotted against the laser intensity: 100% intensity corresponds to a laser power of 40 mW.
experimental apparatus placing a photodiode behind the graphite furnace (see Fig. 1). The detection limits for rubidium and lithium by LAAS were found to be 200 pg ml-l for rubidium as well as for lithium. These data compare well with the detection limits which were recently found in our laboratory by simultaneous multi-element LAAS in a commercial graphite tube furnace applying a different experimental arrangement [3]. Therefore, we like to stress that in the present experiment, the LIF-detection limits were a factor of 100 and 200 lower for rubidium and lithium, respectively, compared with LAAS measurement under the same experimental conditions. It is obvious that LIF with broad-band detectors is capable of simultaneous multielement analysis. Although the adequate experimental technique for simultaneous multi-element measurement would require a fast wavelength switching of the semiconductor lasers by diode current on and off resonance [4] rather than using mechanical choppers for modulation, we have demonstrated a simultaneous measurement of rubidium and lithium by modulation of the two laser beams with a low-frequency chopper (see Fig. 1). A simultaneous measurement of rubidium (0.3 ng ml-l) and lithium (3 ng ml-l) is shown in Fig. 7. To pass the beams through the lowfrequency chopper, in contrast to single element detection, we had to use additional mirrors and apertures, which are not shown in the schematic figure of the experimental arrangement (Fig. 1). Because this experimental scheme was not an optimum, we lost a large fraction of laser radiation and obtained poorer detection limits in the
z e M z &
100
8o
s ._ 60 L;1 is cz 4o
0
12
3
4
6 Time
6
7
8
9
(s)
Fig. 7. Demonstration of simultaneous LIF detection of rubidium and lithium applying a lowfrequency chopper for alternating excitation. Dotted envelopes for the rubidium and lithium signals are given for better representation. The rubidium and lithium concentrations are 0.3 and 3 ng ml-‘, respectively.
1524
A.
ZYBIN
et al.
demonstration. In particular for the laser beam that excited lithium the loss was significant. The detection limits for simultaneous measurement obtained with our preliminary arrangement were c nt, = 5 pg ml-’ and CLi = 10 pg ml-l.
4. CONCLUSION Our preliminary results have shown that simultaneous multi-element analysis by LIF with semiconductor laser diodes, a commercial graphite furnace atomizer and a broadband photodiode detector is a very simple and promising technique. Detection limits of 1 and 2 pg ml-l were found for lithium and rubidium, respectively. In comparison with LAAS, the detection limits by LIF were found to be better by at least two orders of magnitude. Further improvements can be expected from the application of diode lasers with higher cw power, optimization of the apparatus, in particular, of the detector, and improved signal processing. Acknowledgements-The investigation was performed in Dortmund within the framework of a co-operation project of the Institute of Spectroscopy (Academy of Sciences of Russia, Troitzk) and the Institut fiir Spektrochemie und Angewandte Spektroskopie (Dortmund, F.R.G.). We acknowledge gratefully funding by the Russian and German governments (scientific and technological co-operation project no. x229.11). Furthermore, financial support by the Bundesministerium fiir Forschung und Technologie and the Ministerium fib Wissenschaft und Forschung (Nordrhein-Westfalen) is also gratefully acknowledged.
REFERENCES [l] J. Lawrenz and K. Niemax, Spectrochim. Acta 44B, 155 (1989). [2] R. Hergenroder and K. Niemax, Spectrochim. Acta 43B, 1443 (1988). [3] R. Hergenroder and K. Niemax, Trends Anal. Chem. 8, 333 (1989). [4] H. Groll and K. Niemax, Spectrochim. Acta 48B (1993). [.5] P. A. Johnson, J. A. Vera, B. W. Smith and J. D. Winefordner, Spectrosc. Lett. 21, 607 (1988). [6] P. E. Walters, T. E. Barber, M. W. Wensing and J. D. Winefordner, Spectrochim. Acta 46B, 1015 (1991).
0
0584-8547/92 $5.00t .oo
1992 Pergsmon RessLtd
TOPICS IN LASER SPECTROSCOPY
Simultaneous multi-element analysis in a commercial graphite furnace by diode laser induced fluorescence A.
ZYBIN,*
C. SCHN~~RER-PATSCHAN and K.
NIEMAX~
Institut fur Spektrochemie und Angewandte Spektroskopie (ISAS), Bunsen-Kirchhoff-Str. 11, D-4600 Dortmund, F.R.G. (Received and accepted 29 July 1992) Abstract-A powerful and compact experimental arrangement for simultaneous multi-element measurement by laser induced fluorescence is presented. The analytes are excited in a commercial graphite furnace atomixer by cw semiconductor diode lasers and the fluorescence is detected by a simple photodiode without interference filters or a polychromator. In preliminary measurements, detection limits of 1 pg ml-’ (10 fg absolute) and 2 pg ml-’ (20 fg) for lithium and rubidium, respectively, have been obtained.
1.
INTR~DUCTIT~N
ALTHOUGH
they are known to be very sensitive, laser spectroscopic techniques are still not accepted by analytical chemists. The reasons are obvious. Tunable lasers, in particular pulsed or cw dye lasers, are expensive instruments to purchase as well as to operate. The dye laser stability and reproducibility is poor and they require a lot of space in the laboratory. While these reasons will hamper the transfer of laser spectroscopic techniques from the research level to laboratories for routine analysis, it is even more unlikely that many tunable dye lasers can be applied simultaneously for multi-species analysis. However, the rapid progress and the new developments in the field of laser diodes opens up the possibility of transferring laser spectroscopic techniques to analytical laboratories for routine analysis [l]. For the last seven years we have been applying diode lasers together with dye lasers in analytical spectroscopy in our laboratories in Dortmund. Some time ago we published the first paper on simultaneous two-element analysis using only diode lasers [2]. In that paper, diode lasers were successfully used to replace hollow-cathode lamps in atomic absorption spectroscopy using a commercial graphite furnace atomizer. With diode lasers there is not only the possibility of simultaneous determination of several analytes in the sample, but, because of the fast tunability of diode lasers, the background can be measured and the dynamic range extended by the measurement of absorption in the line wings. By using the possibility of frequency doubling in non-linear media, laser atomic absorption spectroscopy (LAAS) with diode lasers already allows the determination of a large number of elements [3]. Recently, we have set up a diode laser spectrometer with six laser diodes in our laboratory and determined simultaneously six elements by LAAS [4]. Typical laser spectroscopic techniques, as, for example, laser induced fluorescence (LIF) or the methods of laser ionization spectroscopy, are more sensitive than LAAS. First demonstrations of diode laser induced fluorescence in a graphite furnace [5] and in an analytical flame [63 have been published. In both papers, rubidium was the analyte. LIF was measured by a photomultiplier fixed to the exit of a monochromator. l Instituteof Spectroscopy, Russian Academy of Sciences, Troitxk, Russia. t Author to whom correspondence should be addressed.
1519
A. ZKMN et al.
1520
laser diodes no. 1 no. 2
T out
reference
Fig. 1. Experimental
apparatus.
In this paper, we will demonstrate a very simple, much more sensitive method for analytical multi-element measurement than LAAS: laser induced fluorescence (LIF) in commercial graphite furnaces with a simple photodiode detector. We will show that the application of diode lasers of moderate cw power allows simultaneous detection of elements in the lower pg g-i range without the use of dispersive units like polychromators or interference filters with radiation detectors. Lithium and rubidium were selected as analytes.
2. EXPERIMENT The experimental arrangement is shown in Fig. 1. Two diode lasers were applied, one laser diode for excitation of the lithium resonance lines at 670.776 nm (Hitachi 6712, cw power: 5 mW) and another laser diode for the rubidium D2 line at 780.027 nm (Hitachi 7851, cw power: 50 mW). The laser diodes were tuned to the atomic transitions and stabilized by temperature. The long-term wavelength stability of the diode lasers were better than -CO.4 pm. Since the widths of the rubidium and lithium resonance lines broadened by the argon buffer gas in the graphite furnace atomizer were much larger than the possible drifts of the laser wavelengths, within the experimental uncertainties, no systematic errors due to wavelength drifts of the diode lasers were observed during the measurements. In the first part of the experiment, the laser diodes were operated sequentially. The laser beams were collimated by large aperture lens systems in front of the diode lasers (not shown in Fig.1) and directed through a hole in a plane mirror into a commercial graphite tube furnace (Perkin-Elmer HGA 74). In order to get more effective atomization conditions in the case of lithium, the graphite furnace was plated inside with a tantalum foil. The laser beams were modulated by a high-frequency chopper (3 kHz). The fluorescence radiation was collected by the plane mirror and focused by a lens system onto a photodiode (Hamamatsu S 2386-X) through an aperture (diameter: 1.5 mm). The solid angle of observation was about 15”, the imaging ratio was one to one. The photodiode used had a spectral response in the range 400-1100 nm. The radiant sensitivity was 0.6 A W-r at maximum (908 nm) and the dark current about 5 pA. The photodiode was operated with a 2 kfi load resistor. The signal was fed into a lock-in amplifier (Itaco 3961B). A time constant of 100 ms was chosen. The amplified signals were recorded by a storage oscilloscope (Hameg HM 205-3). The data could be transferred by a serial interface into a personal computer. The resolution of the signals by the computer was moderate (8 bit). In the second part of the experiment, where simultaneous LIF-measurements of rubidium and lithium were demonstrated, the beams of both lasers were alternatively chopped by a lowfrequency chopper (about 10 Hz) shown as a dashed symbol in Fig. 1. Because of the 10 Hz
Simultaneous multi-element analysis
7
6 ._ ce a
1521
100 SO
20 0 0
2
1
4
3
6
5
I
Time (s) Fig. 2. LIF signals of rubidium.
low-frequency modulation, the -time constant of the lock-in amplifier had to be reduced from 100 ms to 10 ms in comparison with the single element measurements. In order to avoid scatter of laser radiation from the windows of the furnace, the graphite tube furnace was operated without windows. The gas flow of the protect’ion gas (argon) was typically 5-10 1 min-‘. Internal gas stop was applied during the atomization phase. Aqueous solutions of 10 ~1 were injected into the furnace. The solutions had been prepared in clean rooms with double-distilled water. The smallest concentration of lithium and rubidium was 10 pg ml-‘, respectively.
3. RESULTS AND DISCUSSION Typical LIF signals for 100, 50 and 20 pg ml-l rubidium are shown in Fig. 2. These are net-signals from the lock-in amplifier stored by the computer. However, it is interesting to show the signals of the photodiode before processing without modulation by the chopper. Figure 3 displays the detector signal plotted against time during the atomization phase. The full curve is the radiation from the furnace including the fluorescence radiation from 20 ng ml-’ lithium, while the dashed curve is the signal from the blank (furnace only). Note that the absolute mass of lithium is 200 pg. The calibration curves of rubidium and lithium, derived from the peak of the LIF
0
12
3
4
3
6
7
8
9
Time (s) Fig. 3. Photodiode signal of the total radiation from the ftirnace. Full curve: furnace radiation plus LIF from 20 ng ml-’ (200 pg absolute) lithium; dashed curve: radiation from furnace only.
1522
A. ZY~IN et al.
1
10
100
1000
Rb-concentration
10000
lOooo0
(pg ml-t)
Fig. 4. LIF-calibration curve of rubidium obtained by laser excitation of the D2 line at 780.027 nm. Two different sets of solutions (0,O) were used for calibration. (m) is the signal of the blank. The 3o-detection limit is 2 pg ml-‘.
signals, are shown in Figs 4 and 5, respectively. Two sets of rubidium solutions were measured. The agreement of the data was good. The detection limit @u-criterion) of rubidium was found to be 2 pg ml-l, which corresponds to 20 fg in the furnace. The blank level of rubidium was of the order of the detection limit. Please note, that our present LIF detection limit for rubidium found with the simple broad-band photodiode is a factor of 200 and 100 better than the detection limits for rubidium in a graphite furnace [5] and in an analytical flame [6], respectively, if a monochromator and a photomultiplier are applied. The detection limit of lithium was about 1 pg ml-l (10 fg lithium). As shown in Fig. 5, the blank was about 3 pg ml-‘. The L&measurement was made with 5 mW of laser power, while the power of the laser diode for the Rb-measurement was about 40 mW. In lithium we could not saturate the transition optically; however, we could observe slight optical saturation of the rubidium signal at maximum power. This is shown in Fig. 6, where the Rb-signal is plotted against the intensity (intensity at 40 mW: 100%). It has to be noted that, in the present experiment, the detection power was limited by intrinsic electronic noise at the detector side (photodiode and lock-in amplifier). Therefore, we expect further improvements of the detection limits with optimized radiation detection. However, laser diodes with higher powers will also improve the detection limits. Better detection limits should be expected, in particular, for lithium. In order to compare the detection power of these simple LIF measurements with absorption measurements, we performed LAAS measurements with the same
0.001
Y’ 0.5
I 5
50
Li-concentration
500
5000
50000
(pg ml-*)
Fig. 5. LIF-calibration curve of lithium obtained by laser excitation of the lithium resonance line at 670.776 nm. (m) is the signal of the blank. The 3u-detection limit is 1 pg ml-r.
Simultaneous multi-element
0
20
60
40
Laser
80
intensity
1523
analysis
106
(%)
Fig. 6. LIF-signal of rubidium plotted against the laser intensity: 100% intensity corresponds to a laser power of 40 mW.
experimental apparatus placing a photodiode behind the graphite furnace (see Fig. 1). The detection limits for rubidium and lithium by LAAS were found to be 200 pg ml-l for rubidium as well as for lithium. These data compare well with the detection limits which were recently found in our laboratory by simultaneous multi-element LAAS in a commercial graphite tube furnace applying a different experimental arrangement [3]. Therefore, we like to stress that in the present experiment, the LIF-detection limits were a factor of 100 and 200 lower for rubidium and lithium, respectively, compared with LAAS measurement under the same experimental conditions. It is obvious that LIF with broad-band detectors is capable of simultaneous multielement analysis. Although the adequate experimental technique for simultaneous multi-element measurement would require a fast wavelength switching of the semiconductor lasers by diode current on and off resonance [4] rather than using mechanical choppers for modulation, we have demonstrated a simultaneous measurement of rubidium and lithium by modulation of the two laser beams with a low-frequency chopper (see Fig. 1). A simultaneous measurement of rubidium (0.3 ng ml-l) and lithium (3 ng ml-l) is shown in Fig. 7. To pass the beams through the lowfrequency chopper, in contrast to single element detection, we had to use additional mirrors and apertures, which are not shown in the schematic figure of the experimental arrangement (Fig. 1). Because this experimental scheme was not an optimum, we lost a large fraction of laser radiation and obtained poorer detection limits in the
z e M z &
100
8o
s ._ 60 L;1 is cz 4o
0
12
3
4
6 Time
6
7
8
9
(s)
Fig. 7. Demonstration of simultaneous LIF detection of rubidium and lithium applying a lowfrequency chopper for alternating excitation. Dotted envelopes for the rubidium and lithium signals are given for better representation. The rubidium and lithium concentrations are 0.3 and 3 ng ml-‘, respectively.
1524
A.
ZYBIN
et al.
demonstration. In particular for the laser beam that excited lithium the loss was significant. The detection limits for simultaneous measurement obtained with our preliminary arrangement were c nt, = 5 pg ml-’ and CLi = 10 pg ml-l.
4. CONCLUSION Our preliminary results have shown that simultaneous multi-element analysis by LIF with semiconductor laser diodes, a commercial graphite furnace atomizer and a broadband photodiode detector is a very simple and promising technique. Detection limits of 1 and 2 pg ml-l were found for lithium and rubidium, respectively. In comparison with LAAS, the detection limits by LIF were found to be better by at least two orders of magnitude. Further improvements can be expected from the application of diode lasers with higher cw power, optimization of the apparatus, in particular, of the detector, and improved signal processing. Acknowledgements-The investigation was performed in Dortmund within the framework of a co-operation project of the Institute of Spectroscopy (Academy of Sciences of Russia, Troitzk) and the Institut fiir Spektrochemie und Angewandte Spektroskopie (Dortmund, F.R.G.). We acknowledge gratefully funding by the Russian and German governments (scientific and technological co-operation project no. x229.11). Furthermore, financial support by the Bundesministerium fiir Forschung und Technologie and the Ministerium fib Wissenschaft und Forschung (Nordrhein-Westfalen) is also gratefully acknowledged.
REFERENCES [l] J. Lawrenz and K. Niemax, Spectrochim. Acta 44B, 155 (1989). [2] R. Hergenroder and K. Niemax, Spectrochim. Acta 43B, 1443 (1988). [3] R. Hergenroder and K. Niemax, Trends Anal. Chem. 8, 333 (1989). [4] H. Groll and K. Niemax, Spectrochim. Acta 48B (1993). [.5] P. A. Johnson, J. A. Vera, B. W. Smith and J. D. Winefordner, Spectrosc. Lett. 21, 607 (1988). [6] P. E. Walters, T. E. Barber, M. W. Wensing and J. D. Winefordner, Spectrochim. Acta 46B, 1015 (1991).