Analyst December 1983 Vol. 108 PP. 1459-1465 1459 Furnace Atomisation with Non-thermal Excitation- Experimental Evaluation of Detection Based on a High-resolution Echelle Monochromator Incorporating Automatic Background Correction H. Falk E. Hoffmann and Ch. Ludke Central Institute for Optics and Spectroscopy Academy of Sciences Rudower Chaussee 5 1199 Berlin GDR John M. Ottaway and S. K. Giri Department of Pure and Applied Chemistry University of Strathclyde Cathedral Street Glasgow G1 1XL The analytical potential that could be derived from the combination of a FANES source with a high-resolution wavelength-modulated kchelle mono-chromator has been investigated. Detection limits for elements of high excitation potential are improved using the non-thermal excitation process compared with the thermal excitation processes available in carbon furnace atomic-emission spectrometry.For some other elements of lower excitation potential detection limits are impaired owing to the increased complexity of the background spectrum in the FANES source. Keywords ; A tomic-emission spectrometry ; bchelle monochromator ; electro-thermal atomisation ; low-pressure discharge ; wavelength modulation Recently a new excitation source has been described for use in atomic-emission spectrometry, which couples high excitation energy with an efficient atomisation process. The method has been termed FANES,1-3 or furnace atomic non-thermal excitation spectrometry and involves conventional electrothermal atomisation of samples in a tube atomiser in which a low-pressure gas discharge is simultaneously generated using the graphite tube itself as the cathode.The combination of the highly efficient electrothermal atomisation system with the low-noise excitation dischargeJ4 offers high sensitivity analysis for a wide range of elements using small sample volumes. The usual attractive features of emission sources e.g. high dynamic range (5-6 orders) and the ease of operation in a simultaneous multi-element mode, have also been demonstrated. Compared with the use of electrothermal atomisation as an atomic-absorption atom cell the possibility of determining elements such as halogens with resonance lines in the vacuum ultraviolet using FANES is also exciting. The measurement of atomic-emission signals during electrothermal atomisation without supplementary excitation has also been extensively studied and recently re~iewed.~ Under these conditions the existence of local thermal equilibrium (LTE) has been demonstrated6 and parts per billion (parts per log) (p.p.b.) detection limits have been achieved for a wide range of elements.Using this technique carbon furnace atomic-emission spectrometry or CFAES the best detection limits reported to date have been achieved with a spectrometric system based on a 0.75 m 6chelle monochromator incorporating automatic background correction.' The dchelle spectrometer offers the high resolution favourable for atomic-emission measurements with a high optical conductance (f/13 aperture). Background correction is achieved using square-wave wavelength modulation generated by means of a rotating quartz chopper with four separate quartz quadrants of different thicknes~es.~~~ The chopper is mounted either near the entrance or exit slits of the monochromator and the modulation frequency of 40 Hz is adequate for most atomic-emission signals and allows efficient background correction to be achieved for both the continuum background from the graphite furnace and matrix scatter signals from for example clinical material^.^ Although very low detection limits have been obtained for many elements,7~10 the determination of volatile elements with high excitation potentials (e.g, cadmium zinc and selenium) a t sub 1460 FALK et at?.FURNACE ATOMISATION WITH NON-THERMAL Analyst VOZ. 108 parts per billion levels has still not been found possible despite the introduction of platform7J0 and probell atomisation techniques.For the high sensitivity detection of these elements by atomic emission during electrothermal atomisation the FANES approach appears to be attractive if not essential. Measurements of FANES reported to date1-3 have been made with a PG S2 grating spectro-graph (VEB Carl Zeiss Jena). This has a two-channel photomultiplier detection system in place of the photographic plate and was operated in the d.c. registration mode without background correction during signal measurement. Under these conditions detection limits are determined by the fluctuation of the background signal at the position of the atomic line being measured. The 6chelle spectrometer system developed for CFAES obviously offers characteristics highly suited to measurements with the FANES excitation source.The high resolution and background-correction system should both improve detection limits of FANES and in addition the range of elements available by CFAES should be extended by the high excitation energy in the FANES source. To examine these possibilities a short-term collaborative project was set up between our laboratories which allowed the transfer of a FANES source to the University of Strathclyde for coupling with the dchelle spectrometer system. The results of this study which allowed some interesting conclusions to be reached, are reported in this paper. Experimental The dchelle spectrometer system is based on a Spectrametrics dchelle monochromator, modified for wavelength modulation background correction.For experiments using wave-length modulation (WM) the instrumental configuration was exactly as described previ-ously.7 To provide comparison with earlier FANES results obtained using a d.c. system, measurements were also made using intensity modulation (IM). In this instance a rotating chopper disc was mounted between the FANES source and the spectrometer entrance slit. A square-wave modulation waveform with a frequency of 130 Hz was generated and the chopper incorporated a reference signal that was used to synchronise the lock-in amplifier. Hollow-cathode lamps were used for wavelength adjustment by focusing the lamp through the FANES source on to the entrance slit of the spectrometer. Hollow-cathode lamps were also used to measure the linearity range of the detection system and to confirm the linear relationship between the slit aperture (entrance slit width = exit slit width) and the signal amplitude.A schematic diagram of the cross-section of the FANES source used in this study is shown in Fig. 1 which shows only minor modifications to that described earlier.2 This FANES furnace tube is identical in size with that used in the Perkin-Elmer HGA 500 heated graphite atomiser ( i e . 28 mm long and 5.9 mm i.d.) and is connected to electrical power supplies for heating the furnace tube (Uh in Fig. 1) and for exciting the low-pressure gas discharge. In the latter the graphite tube acts as the cathode and a separate anode is introduced as shown in Fig. 1. A mechanical pump is used to reduce the pressure of inert gas (helium or argon) to 1-5 Torr during the discharge process.The discharge current was 30 mA at a voltage of 600 V. The operation of the FANES source is analogous to that used in conventional electro-thermal atomisation. A 20-pl sample aliquot is injected into the furnace and dried and/or ashed as required at atmospheric pressure. During this time the cover over the sample injection hole is removed to allow the released vapours to be cleared from the furnace. The furnace chamber is then sealed and pumped down to 3 Torr of helium (unless otherwise mentioned) and the low-pressure discharge is initiated. This latter process takes approxi-mately 30 s. The atomisation stage is then initiated using the optimised atomisation temperature.The transient atomic-emission signals were recorded on a Servoscribe RE 541 20 strip-chart recorder. After allowing the furnace to cool to ambient temperature, the pressure is raised to atmospheric pressure and the cover removed for introduction of the subsequent sample. The FANES furnace chamber and power supplies are all home built in the laboratories of the Central Institute for Optics and Spectroscopy but are based on con-ventional electrothermal atomisation and low-pressure discharge generating circuits. The heating rate of the electrothermal atomiser can reach a maximum of 2000 "C s-l December 1983 EXCITATION-EXPERIMENTAL EVALUATION OF DETECTION 1461 D Fig. 1. Cross-section of the FANES source. Ua Anode; Uh connections for electrothermal heating of the graphite tube; A graphite tube; B graphite con-tact cylinders ; C removable lid for sample injection; D quartz windows; E rotation arm for changing the graphite tube.The furnace is also water cooled (not shown) to allow rapid introduction of successive samples. Results and Discussion In the investigations reported a range of elements were selected to provide a comparison with detection limits achieved previously with FANES (chromium silver and lead) and CFAES (chromium silver lead zinc and cadmium). Some elements were also chosen with high excitation potentials i.e. with resonance lines in the low ultraviolet wavelengths (zinc, cadmium and selenium) to provide information on potential improvements compared with CFAES and the over-all limitations of the present system.All analyte solutions were pre-pared from AnalaR reagents with 1 0 - 2 ~ nitric acid added and dilution with high-purity distilled water. The effect of sample volume is illustrated in Fig. 2 for cadmium in a helium atmosphere and is typical of electrothermal atomisation. Signal response increases in a more or less linear fashion up to 4 0 ~ 1 after which the signal and reproducibility both deteriorate owing to the greater spreading of the sample in the graphite tube. 20 40 60 80 100 Sample volume/pI Fig. 2. Effect of sample volume on the FANES signals Carrier for 100 pg 1-1 cadmium solution a t 228.8 nm. gas helium; discharge current 20 mA 1462 FALK et a,!. FURNACE ATOMISATION WITH NON-THERMAL Analyst VOZ. 108 Results for the instrument system examined in this study and those for CFAES and previous FANES studies are given in Table I.Detection limits in this work were calculated as the concentration equivalent to three times the standard deviation of the background noise. These results allow the following conclusions to be drawn. 1. The use of the 6chelle spectrometer system with wavelength modulation does not lead to the anticipated dramatic improvement in FANES detection limits. A factor of 2-3 at the most was achieved for silver and chromium. 2. Wavelength modulation does not give an improvement in detection limits compared with intensity modulation under identical conditions of resolution (see cadmium chromium, zinc and lead where WM results are worse than IM). 3. A spectral resolution of 10-20 x lo3 appears to be sufficient for FANES in order to achieve the best detection limits.4. Whilst significant improvements in detection limits compared with CFAES are achieved for cadmium selenium and zinc detection limits are actually poorer than those reported for CFAES for chromium. The results were found to be entirely related to the nature of the background signal generated in the FANES source. Despite the use of the automatic background correction device a small residual background signal was generated in all instances as indicated in TABLE I DETECTION LIMITS ACHIEVED WITH THE FANES ~CHELLE SPECTROMETER SYSTEM USING INTENSITY MODULATION' (IM) OR WAVELENGTH MODULATION (WM) Element Silver. . Cadmium Chromium Zinc . . Lead Lead Selenium a .Wavelength/ nm 328.7 228.8 425.4 213.9 405.7 283.8 196.0 Measure-ment mode IM IM WM WM IM IM IM IM WM WM WM WM IM IM IM IM IM WM IM W M IM WM IM IM IM WM WM WM IM R* x 103 15 33 15 33 6.5 16 25 33 16 25 33 6.5 6.5 16 31 45 27 16 16 16 7 17 16 31 40 16 31 40 16 Dt 0.38 0.11 0.38 0.18 1 .a 0.38 0.18 0.07 1.0 0.38 0.18 0.07 1 .o 0.38 0.18 0.07 0.02 0.38 0.38 0.38 0.4 0.38 0.38 0.11 0.04 0.38 0.11 0.04 0.38 Detection limit /pg IBSl mV 1.5 0.8 co.1 0.1 1 .o 5.9 0.3 0.06 1 .o 0.24 0.1 0.02 23 8 3.5 1.1 0.7 <0.02 1.1 0.05 44 0.18 1.64 0.19 0.10 0.5 0.2 0.1 0.05 This work 0.4 2.3 1.1 2.6 1.1 1.1 2.4 10.2 4.1 3.4 3.5 9.1 46 7.5 6.2 9.6 9.2 4.0 5.3 14.2 17.4 12 56 60 20 30 55 800 67 FANES previous work 1.03 CFAES 2.610 3OO1l 10 (357.9 nm)2 --14 (368.0 nm)* 1.21' 2 400" 46;" * R Practical resolving power of the spectrometer.t D Relative optical conductance of the spectrometer. $ IB Relative intensity of background December 1983 EXCITATION-EXPERIMENTAL EVALUATION OF DETECTION 1463 Table I and Figs. 3 and 4. It is clear that wavelength modulation substantially reduces the background signal. The rise in the signal when the low-pressure discharge is switched on is equivalent to the background signal from the FANES source under IM conditions and this is substantially smaller under WM conditions.The detector noise was only comparable to the source background intensity for the narrowest slit widths used. The FANES source back-ground signal was measured independently and was found to consist mainly of a multi-line spectrum. The background intensity as a function of slit width could be described by a power law the exponent of which varied between 1 and 2 depending on the wavelength used. D D C J -t U t Time + Fig. 3. Background and analyte signals measured a t the cadmium (228.8 nm) line for 10 pl of (a) A Background a t (b) A Background a t 20 p g 1-' cadmium solution using (a) wavelength and (b) intensity modulation. 0.2 mV; B FANES switched off; C background a t 0.4 mV; D analyte peaks.0.1 mV chart recorder voltage; B FANES switched off; C background a t 1 mV; D analyte peaks. From these measurements and the observation of a considerable enhancement of the back-ground signal when small amounts of air leaked into the carrier gas it was concluded that the background consists mainly of molecular bands of the plasma gas. These bands con-tained spectral ranges with continuous as well as line characteristics. It is clear from this information why WM does not give improved detection limits compared with IM. The back-ground intensities at the analyte line and background measurement wavelengths selected by the wavelength modulation device are not equal owing to the complex structured background of the FANES source.Consequently detection limits remain dependent on the signal to background ratio rather than the signal to noise ratio. Although wavelength modulation did not allow exact automatic background correction to be achieved a manual correction could easily be made using the measurements illustrated in Figs. 3 and 4. The background (IM) or residual background (WM) from the source can be measured when the gas discharge is in operation and before the sample atomisation sequence is started. Such a procedure would however be inadequate for structured background signals from the analyte matrix. When argon was used as the carrier gas instead of helium the background intensity was remarkably higher at all analyte wavelengths used whereas comparable analyte intensities were obtained 1464 FALK et al.FURNACE ATOMISATION WITH NON-THERMAL - T m rn E O iij t- 7-T F1 m m I C D \ C -f Analyst Vol. 108 Time -b Fig. 4. Background and analyte signals measured at the lead 405.8-nm line for 20 pl of 50 pg 1-1 lead solution using (a) wave-length and (b) intensity modulation. (a) A Background at 5 mV (note it is negative because the FANES background is larger at the analyte wavelength than at the one used for background correction) ; B FANES source off; C background a t 25 mV; D analyte peaks. (b) A Background at 2.5 mV; B FANES source off; C background at 10 mV; D analyte peaks. The best detection limits for cadmium and zinc of 0.03 and 0.1 pg l-l respectively (assuming a volume of 40p1) indicate that the non-thermal excitation process of FANES does allow as p r e d i ~ t e d ~ ~ ~ a higher population of energy levels with larger excitation potentials than the thermal excitation mechanism available in CFAES.These values are commensurate with electrothermal atomisation atomic-absorption spectrometric (ETA-AAS) detection limits and indicate that FANES combined with CFAES would probably be competitive with current ETA-AAS systems. The relatively high detection limit for selenium of 40 pg 1-1 is explained by the very low transmittance of the optical system at the 196.0-nm wavelength. For chromium the FANES detection limit is significantly poorer than the CFAES detection limits and this would also be expected for other less volatile elements with relatively low excitation potentials owing to the greatly increased structured background from the FANES source.Whilst very high resolution does not appear to produce lower FANES detection limits it may be useful with particular sample types to overcome spectral interferences. Generally, wavelength modulation would be more useful if the positions used for background measure-ment can be chosen to give a signal more exactly correlated with the background intensity at the analyte line. With this system this is difficult and could only be achieved by alteration of the angle of incidence of the light beam at the rotating chopper. A more easily adjustable wavelength modulation device would be useful but might be limiting if matrix spectral interferences varied from sample to sample. In order to improve the performance of FANES and obtain lower detection limits and improved background correction it would seem preferable to attempt to reduce the complexity and magnitude of the background itself by using a more perfect vacuum system and a purer gas supply.This work was made possible by the Cultural Exchange Agreement between the Royal Society of the UK and the Academy of Sciences of the GDR and the authors are very gratefu December 1983 EXCITATION-EXPERIMENTAL EVALUATION OF DETECTION 1465 for the opportunity for collaborative study and the financial support provided through this scheme. Financial support from the SRC for the purchase of the kchelle spectrometer and from the British Council (for S.K.G.) is also gratefully acknowledged. 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. References Falk H. Hoffmann E. Jaeckel I. and Ludke Ch. Spectrochim. Acta Part B 1979 34 333. Falk H. Hoffmann E. and Ludke Ch. Spectrochim. Acta Part B 1981 36 767. Falk H. Hoffmann E. and Ludke Ch. Fresenius 2. Anal. Chem. 1981 307 362. Falk H. Spectrochim. Acta Part B 1977 32 437. Ottaway J. M. Hutton R. C. Littlejohn D. and Shaw F. Wiss. 2. Karl-Mum Univ. Leipzig, Littlejohn D. and Ottaway J. M. Analyst 1979 104 208. Ottaway J . M. Bezur L. and Marshall J. Analyst 1980 105 1130. Michel R. G. Sneddon J. Hunter J. K. Ottaway J. M. and Fell G. S. Analyst 1981 106 288. Ottaway J. M. Bezur L. Fakhrul-Aldeen R. Frech W. and Marshall J. in Bratter P. and Schramel P. Editors “Trace Element Analytical Chemistry in Medicine and Biology,” Walter de Gruyter Berlin 1980 p. 575. Bezur L. Marshall J. Ottaway J. M. and Fakhrul-Aldeen R. Analyst 1983 108 553. Giri S. K. Littlejohn D. and Ottaway J. M. Analyst 1982 107 1095. 1979 28 357. Received February 25th 1983 Accepted July 29th 198