JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1992 VOL. 7 105 Determination of Lead in Natural and Waste Waters Using a Non-dispersive Atomic Fluorescence Spectrometer With a Tungsten Spiral Atomizer* Svetlana S. Grazhulene Vladimir A. Khvostikov Nina N. Vykhristenko and Mikhail V. Sorokin Institute of Microelectronics Technology and High Purity Materials USSR Academy of Sciences 742432 Chernogolovka Moscow District Russia A laboratory-made atomic fluorescence spectrometer with a capacitively heated tungsten spiral atomizer was applied to the determination of lead in waters. The optical system for fluorescence collection and the electronics for registration have been improved which made it possible to suppress the emission of the hot spiral and to measure the fluorescence directly from the inner volume of the spiral.The detection limit of lead (3.5 x lo-' g drn-3) is reduced by a factor of 30 compared with measuring the fluorescence above the spiral. The relative standard deviation is 0.05O/0 (n=lO). An acidity of <1% HN03 a magnesium content of (0.01 g dm-3 and a calcium content of (0.02 g dm-3 do not affect the lead fluorescence signal. Keywords Non-dispersive atomic fluorescence spectrorne ter; tungsten spiral atomizer; lead determination; waters Lead is a major toxicant and its ultimate concentration must be no more than 3 x g dm-3 in drinking water in Russia. The content of lead in USSR rivers may vary from 2 x lo-' to 4.5 x g dm-3 and highly sensitive methods are required for its determination. Moreover rapid tech- niques using simple instrumentation are required for such environmental analyses.There are numerous publications on the determination of lead in waters,'-' and methods involving spectrophoto- metry or stripping voltammetryd have most commonly been employed. However even with the use of time-consuming preconcentration proceduress the spectrophotometric detection limits (DL) are only 1 x 10-5-2 x g dm'3. Stripping voltammetry6 permits the DL of lead to be decreased by an order of magnitude. However this method is also time consuming. The lowest theoretical DL ( 1 x lo-* g dm-3) has been obtained using laser methods such as laser atomic ioniza- tion spectrometry' and laser flame atomic fluorescence spectrometry (AFS).' The real DL is about 1 x lo-' g dm-3.4 The same DL has also been achieved in the non- dispersive determination of lead by AFS using a hydride generation te~hnique.~ The detection limits of commer- cially available instruments for the determination of trace amounts of lead in waters are 9 x g dm-3 [flame atomic absorption spectrometry (AAS)) 7 x g dm-3 (electro- thermal AAS) and 3 x dm-3 [inductively coupled plasma atomic emission spectrometry (ICP-AES)]. Atomic fluorescence spectrometry is often superior to other atomic spectrometric methods as far as linear dynamic ranges and DLs are concerned.The use of a nondispersive optical registration system (ND-AFS) permits the design to be simplified and the cost of the device to be reduced. The potential of AFS using a spiral tungsten atomizer has been reported This paper presents a new model of the ND-AFS based on a tungsten spiral atomizer.The potential of the spectrometer was studied on aqueous solutions of different elements attention being focused on the determination of lead in waters. Experimental Apparatus A diagram of the non-dispersive atomic fluorescence * Presented at the XXVII Colloquium Spectroscopicurn Interna- tionale (CSI) Bergen Norway June 9-14 1991. spectrometer with a tungsten spiral atomizer is shown in Fig. 1. An electrodeless discharge lamp (A) excited by an r.f. generator (B) (frequency of excitation 100 MHz) serves as a fluorescence excitation source. Its radiation is collected by a lens (C) and directed to the tungsten spiral atomizer (D). The atomizer is a tungsten spiral 1.5 mm in diameter and 2 mm long; the wire diameter is 0.1 mm and the number of wire loops is ten.The resonance fluorescence signal is observed at 90" to the incident angle. Fig. 2 shows the design of the analytical unit which includes the tungsten spiral atomizer with a gas flow- forming system a fluorescence excitation source a photo- multiplier and optics for excitation irradiation and fluor- esence collection. If necessary an interference filter with a spectral bandpass corresponding to the wavelength of the element to be determined can be placed in front of the photomultiplier to suppress the emission of light from the hot spiral. The design differs from the previous one,1° in which the fluorescence signal was measured above the spiral by measuring the fluorescence directly from the inner volume of the spiral. In this instance the concentration and I I L To computer Fig.1 Schematic diagram of the ND-AFS instrument with a tungsten spiral atomizer A EDL; B excitation r.f. generator; C lens; D tungsten spiral atomizer; E objective; F diaphragm; G photomultiplier; H optical filter; I narrow-band lock-in amplifier; J pulse generator K peak detector L gated integrator; M indicators; N laboratory-made power source; and 0 control unit106 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1992 VOL. 7 Fig. 2 Design of the analytical unit of the spectrometer A electrodeless discharge lamp; B excitation r.f. generator; C lens; D tungsten spiral atomizer E objective; F diaphragm; G photomul- tiplier; H optical filter; I gas flow-forming system; J light traps; and K sample injector temperature of the atomic vapour are higher.However the emision of light from the hot spiral increases by 2-3 orders of magnitude compared with measuring the fluorescence above the spiral. In order to suppress this effect and also excitation radiation scattered on the spiral loops a special optical system for fluorescence collection with an objective (E) and diaphragm (F) in its focal plane is used. The additional suppression of the emission from the spiral is made at the expense of modulation of the exciting radiation of the electrodeless discharge lamp (modulation frequency 10 kHz depth of modulation 80%) and is followed by the separation of a valid signal from the total signal at the output of the photomultiplier using a narrow-band lock-in amplifier (I) at the modulation frequency.Further the fluorescence signal the duration of which is 20-50 ms is separated by means of the high frequency filter from the scattered signal which is also modulated but with a constant amplitude. The resulting fluorescence signal is measured by the gated peak detector (K) and an integrator (L). The information on amplitude and integrated absor- bance is output to indicators (M). Output to a computer or other external device is also provided. The tungsten spiral atomizer is heated by means of a laboratory-made power source (N) which includes the electron current stabilizer voltage stabilizer capacitor commutator and discharge key. All of this is controlled by the control unit (0) according to the heating programme.The programme includes four cycles annealing drying ashing and atomization. The first three cycles involve heating the spiral in a direct current (at temperatures of up to 1900 K and time up to 100 s) and for the fourth cycle impulsive discharge of the condenser batteries through the spiral is applied (the maximum temperature is 3300 K with ;atomization time up to 100 ms and heating rate up to 1 x lo6 K s-l). The control unit involves the on-line memory which allows eight different heating programmes to be recorded. The possibility of working with an outside memory device is provided. Moreover the control unit synchronizes all the other units of the spectrometer. Reagents and Procedures !Standard and test solutions were prepared from high-purity metal nitrates and de-mineralized water.The stock solu- tions had a metal content of 1 g dm-3 and were diluted as required. Samples with concentrations less than 1 x lo+ g dm-3 were prepared directly before measurement to prevent precipitation on the walls. The solution may be applied to the spiral either by using a micropipette or by dipping the spiral in a glass tube containing the solution to be analysed in which the liquid is held in the vessel by capillary forces. The latter improves the precision from 10% for the former to 2%. A sample volume of 2 p1 was used in both instances. Coincidentally with the fluorescence measurement the emission from the spiral was measured by a photodiode placed near the spiral. The photodiode was calibrated against temperature which allowed the dependence of the lemperature of the spiral on time during heating to be obtained and comparison of the fluorescence signal with atomization temperature conditions.Parameters of the femperature conditions during heating of the spiral (ashing drying time and temperature heating rate and maximum atomization temperature) were chosen for each element to give the best signal-to-noise ( S / N ) ratio. Argon was used as a shielding gas. Results and Discussion The typical fluorescence signal shape for lead under optimum conditions of heating the spiral (the lead content in the solution is 1 x g dm-3) is shown in Fig. 3. The atomization temperature curve under the same conditions is also given. It can be seen that the majority of the lead atoms are atomized at 1200-1 300 "C.It is likely that the further decrease in fluorescence is associated with removal of the main part of the analyte from the spiral. The spiral heating rate is initially 1 x lo5 K s-' at the atomization temperature. As can be seen the fluorescence peak widens at lower spiral heating rates and its amplitude decreases whereas increasing the heating rate above the optimum 1300 c E !i 800 E z 30C I 1 0 0.050 0.100 Tim& Fig. 3 Shape of temperature curve (A) and fluorescence signal (B) of lead in aqueous solution. Conditions of determination lead concentration 1 x g dm-'; drying temperature 360 K; drying time 60 s; temperature of atomization 1200 K; and spiral heating rite 1 x lo4 K s-IJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1992 VOL.7 107 10 .- a .- c 1 v) 9) c a3 C 3 $ 1x10-' E 3 = .- + - a3 K 1x10-2 I 1 1 ixio-' DL 1 x 1 0 ~ 1x10" iX10-' Lead concentration/g I-' Fig. 4 Calibration graph and detection limit for lead in aqueous solution 0.4 0.3 n v) 0.2 a 0.1 0 lX10-' l X l O d 1 ~ 1 0 - ~ iX10-' Lead concentration/g I-' Fig. 5 Concentration dependence of the RSD of lead in aqueous solution does not lead to an increase in fluorescence intensity. The atomization temperature is reached in about 10 ms under optimum conditions. In this event the fluorescence peak is sufficiently narrow (20 ms) and more intense. The calibration graph for the atomic fluorescence deter- mination of lead in aqueous solutions (the amplitude of fluorescence is measured) is shown in Fig. 4. About 10-20 measurements were camed out for each sample and the mean value of the fluorescence was determined.The calibration graph is linear over a concentration range of 2.5 orders of magnitude. The horizontal line in Fig. 4 corre- sponds to a three-fold background fluctuation. The DL obtained is 3 . 5 ~ lo-' g dm-3 (the absolute DL is 0 . 7 ~ g in a probe volume of 2 ~ 1 ) . Corresponding 1 1 1 I I I . 0 1 2 3 4 5 6 Concentration of HNO (% v/v) Fig. 6 Influence of the solution acidity on fluorescence of lead values of the relative and absolute DLs of lead determined previously with fluorescence measurements above the spiral are 1 x g dm-3 and 2 x g respectively.1° Hence the improvement in DL is 30-fold. Moreover additional measurements showed that the fluorescence signal is inde- pendent of the shielding gas consumption whereas an appreciable consumption of inert gas (7 dm3 min-I) to form the atomic vapour is required with fluorescence measure- ments above the spiral.The dependence of the relative standard deviation (RSD) of lead determination on concentration is shown in Fig. 5. The RSD is 0.3 at concentration levels near to the DL and 0.04 for the concentrations > IODL. The analysis of real water samples requires the investiga- tion of the influence of acidity and of other elements particularly calcium and magnesium. These elements are always present in natural and waste waters in amounts that usually exceed that of lead by some orders of magnitude and may affect significantly the analytical results. The investigations were performed on aqueous lead nitrite solutions containing 1 x g dm-3 of lead in the presence of magnesium nitrate calcium nitrate and nitric acid.It was found that magnesium and calcium at concen- trations less than 0.2 g dm-3 and acidities less than 1% do not affect the fluorescence signal. Higher acid concentra- tions give rise to depression of the fluorescence signal (Fig. 6). The real samples of waste water analysed for lead contain many different elements which were determined by ICP-AES (Table 1). The accuracy of the determination of lead in a waste water sample using the standard additions method is shown in Table 2. Table 2 Accuracy of the determination of lead in a waste water sample Lead added/ Lead determined/ g dm-3 g dm-3 RSD n 1 .o 0.82 f 0.07 0.10 7 2.0 1.8 f 0.06 0.03 5 5.0 4.4 f 0.6 0.1 I 5 8.0 7.1 f 0.4 0.02 4 Table 1 Total composition of the waste water sample analysed Concentration/ Concentration/ Concentration/ Concen trationl Element 1 0-6 g dm-3 Element g dm-3 Element g dm-3 Element g dm-3 A1 22 B 112 Ba 23 Ca 32000 c o 3.5 c u 8 .7 Mn 10 SC 1 1 Fe 33 Na 960 Si 3800 K 840 Pb 36 Sr 180 Li 12 Pt 62 Ta 21 Mg 1400 S 4900 Zn 2.6108 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1992 VOL. 7 Table 3 Detection limits of various elements in aqueous solution Parameter Element Ag Bi Cd c u In Mn Te Zn DU 1 0-6 g dm-3 1.5 2.5 35 250 50 100 1000 0.5 Absolute DU10-'4 g 3.0 5.0 70 500 100 200 2000 1 .o Analogous measurements on aqueous solutions of other elements were also performed. In order to determine other elements one must replace the source electrodeless dis- charge lamp and change the temperature programme of heating the spiral. The DLs obtained are given in Table 3.It should be noted that the DL is limited mainly by the emission from the spiral. As the resonance lines of many elements lie in the ultraviolet region and the maximum emission of the spiral is in the infrared and visible regions the insertion of interference filters (filtering is 1 nm wide) makes it possible to suppress additionally this emission effect and may decrease the DL of the elements. Conclusion The non-dispersive atomic fluorescence spectrometer with a tungsten spiral atomizer has been used successfully in the determination of toxic elements especially lead in waters. The DLs of the elements investigated are at the same level as with electrothermal atomizers.The DL of lead is 3.5 x lo-' g dm-3. The RSD does not exceed 0.04 for concentrations 3 1 ODL. The content of calcium and magnesium present in waste water does not significantly decrease the fluorescence signal. The low cost small size and good analytical character- istics of the atomizer make it suitable for use in environ- mental analysis under stationary and field conditions. References 1 Holliday M. C. Houghton C. and Ottaway J. M. Anal. Chim. Acta 1980 119 67. 2 Sthapit P. R. Ottaway J. M. and Fell G. S. Analyst 1984 109 1061. 3 D'Ulivo A. and Papoff P. Talanta 1985 32 383. 4 Marunkov A. G. Reutova I. B. and Chekalin N. V. Zh. Anal. Khim. 1986 41 68 1. 5 Petrova T. V. Dgerajan T. G. and Sawin S. B. Zh. Anal. Khim. 1990 45 579. 6 Fedorina L. I. Risev A. P. and Solomonov V. A. Zh. Anal. Khim. 1989,44 2088. 7 Omenetto N. Human H. G. C. Cavalli P. and Rossi G. Analyst 1984 109 1067. 8 Arkhangelskii B. V. Gonchakov A. S. and Grazhulene S. S. J. Anal. At. Spectrom. 1987 2 829. 9 Gonchakov A. S. Arkhangelskii B. V. and Grazhulene S. S. Vysokochist. Veschestva (Ultrapure Substances) 1988 No. 6 153. 10 Grazhulene S. S. Khvostikov V. A. Sorokin M. V. Vykhris- tenko N. M. Korovyatnikov G. F. and Gonchakov A. S. in Proceedings of the X i Conference on Analytical Atomic Spec- troscopy. Nauka Moscow 1990 p. 37. Paper 1 /034 I5 D Received July 8 1991 Accepted October 21 1991