Analyst, October, 1980 981 SHORT PAPERS A Simple Non-dispersive Atomic-fluorescence Spectrometer for Mercury Determination, Using Cold-vapour Generation R. C. Hutton and B. Preston Tioxide Intervtational Limited, Central Laboratories, Stockton-on- Tees, Cleveland, TS18 2iVQ Keywords; Mercury determination; uapour generation; non-dispersive atomic- fluorescence spectrometry; estuarine samples The expansion of interest in environmental pollution has directed considerable effort towards obtaining lower limits of detection for many toxic elements. Analysis for mercury in par- ticular has placed considerable demands on currently available instrumentation and a recent review1 reflects the great analytical interest in this element. Both atomic-absorption and atomic-fluorescence techniques have been commonly employed to determine mercury and both have their advantages and disadvantage^.^-^ In our laboratory, we have for many years used an atomic-fluorescence technique.How- ever, demands for lower limits of detection coupled with the increased volume of samples presented for analysis have rendered our apparatus inadequate for present requirements. When deciding on a replacement, it was obvious from the number of samples analysed that one instrument would have t o be dedicated to mercury analysis. Rather than use an atomic- absorption procedure it was decided, after a review of the literature, that a purpose-built non-dispersive fluorescence instrument could achieve the levels required. The increased light-gathering power of a non-dispersive system is often off set by background scatter from the atom cell, usually a flame. However, cold-vapour generation gives minimum background scatter and considerable advantages could be obtained by using cold-vapour generation with a non-dispersive fluorescence system.Non-dispersive mercury fluorescence has been reported in the l i t e r a t ~ r e , ~ - ~ but in some instances5 the detection limits obtained required extreme conditions, e g . , solar-blind photomultipliers or sodium-lit rooms. This work describes the performance of a simple non-dispersive fluorescence spectrometer and its application to the routine analysis of marine samples. Non-dispersive fluorescence has not gained the popularity of dispersive techniques. Experimental Apparatus The instrument employed in this work was a purpose-built non-dispersive fluorescence spectrometer, illustrated schematically in Fig.1. The instrument was housed in a light-tight box with a partition in the centre separating the light source and fluorescence cell from the detection system. The fluorescence cell was a 10-mm 0.d. silica tube and fluorescence was measured at a height of 3 mm above the top of the tube. Source. Philips 02 4-W mercury-discharge bulb, powered by a laboratory-constructed 0.35-A constant-current power supply. Mercury cell. 10-mm 0.d. silica tubing, blackened both internally and externally to reduce reflected light. Slits. Photomultiplier. Photomultiplier power s u p p l y . Ampli$er. Chart recorder. 2 mm wide x 1 cm high. EM1 9781B (EM1 Electronics Ltd., Hayes, Middlesex) operated at 600 V.Bentham 215 high-voltage supply (Bentham Instruments Ltd., Reading, Berkshire). Bent ham 2 1 OE current-sensitive amplifier. Chessell flat-bed recorder (Chessell Ltd., Worthing, Sussex).982 SHORTPAPERS Analyst, VoZ. 105 Reagents Dissolve 100 g of tin(I1) chloride in 500 ml of con- centrated hydrochloric acid and dilute to 1 1 with distilled water. Add 25 ml of concentrated sulphuric acid and 110 ml of concentrated hydrochloric acid to 50 ml of distilled water. Dissolve 1.08 g of mercury(I1) oxide in the minimum volume of 1 + 1 V/V hydrochloric acid and dilute to 1 1 with distilled water. Procedure Tin(I1) chloride solution (5 ml) and 5 ml of acid solution were placed in the reduction cell. Argon was bubbled through the solution a t 1.5 1 min-l to remove residual mercury.A suitable aliquot of sample solution (usually 1 or 5 ml) was placed in the reactor cell using a micropipette and the fluorescence signal was read off on a chart recorder. The tin(I1) chloride solution was renewed after every 5-10 determinations, depending on mercury levels. Tin(11) chloride solution, 10% m/V. Acid solution. Mercury standard, 1000 pg ml-1 of mevcury. Cool and dilute to 100 ml. A procedure similar to that of Hatch and OttlO was employed. To recorder Fig. 1. Schematic layout of non-dispersive fluorescence system. A, Rotameter; B, 125-ml capacity reaction cell; C, fluorescence cell; D, mercury discharge bulb; E, slits on central paIti- tion; F, optical guide; G, photomultiplier; H, 0.35-A constant current power supply; I, exit to vacuum line.Results and Discussion The instrumental layout illustrated in Fig. 1 was the result of careful optical optimisation, which led to the lowest possible background light levels. It was found preferable, for example, to use an optical guide (a 15-mm diameter brass tube) rather than a lens between the slits and the photomultiplier tube. Reflections from lenses were found to degrade the signal to background ratio. The central partition, separating the source and the fluorescence cell from the photomultiplier, was necessary to reduce stray light from the source. A reduction in background from the source of approximately 300-fold was achieved with the partition in place, giving an approximate source background equivalent t o 0.005 pg 1-l.Possible molecular background from ozone, generated by the mercury discharge bulb, was eliminated by slight suction, which also removed residual mercury vapour from the instrument. The calibration graphs obtained are illustrated in Fig. 2. These sliow a linear response from the detection limit to over 100 pg l-l, over three orders of magnitude. A practical detection limit of 0.04 pg 1-1 was determined by monitoring continuous day-to-day variations on a 1-ml sample containing 1 pg 1-1 of mercury. A typical precision run using 1 ml of a sampleOctober, 1980 SHORT PAPERS 983 containing 1 p g 1-1 of mercury with an amplifier time constant of 1 s gave the following results (chart divisions): 43, 43, 43, 45, 43, 42, 44, 42, 44. These results give a mean of 43.2 chart divisions with 26 = 0.045 pg 1-1 and a coefficient of variation = 2.25%.Precisions of 2-3% can be achieved. Greater sensitivity could be obtained by the use of up to 5 ml of sample solution, but in general this was found not to be necessary. The detection limit obtained statistically also represents approximately twice the total blank signal obtained practically. 1. i- .- to3 i- C W t .- 8 ; 102 0 - ‘c 5: -I 10 *‘ I x’ 0 0 1 1 10 100 Mercury concentration/pg I-’ Fig. 2. Calibration graphs for mercury fluorescence on logarithmic co-ordinates. A, 1-ml sample volume; B, 5-ml sample volume. The performance of this instrument compares favourably with that of other vapour genera- tion systems previously reported, without requiring complex instrumentation.The instru- ment could possibly be improved further by the inclusion of a solar-blind response photomultiplier, but the practical problems associated with working at such low mercury levels may outweigh the expense involved in the purchase of such a tube. The instrument has been routinely used to determine mercury levels in acid extracts of estuarine sediments and shrimps. Comparison of results obtained by dispersive fluorescence show that no disadvantages result from the use of a non-dispersive system and that lower blank levels can be quoted with confidence, owing to the increased sensitivity. This is illustrated in Table I, which gives results obtained from a typical sample batch. The in- strument can typically analyse 30-40 samples per hour (including standards and blanks).TABLE I COivlPARISON OF DISPERSIVE AND NON-DISPERSIVE DETERMINATION OF MERCURY IN EXTRACTS FROM ESTUARINE SEDIMENTS AXD SEA SHRIMPS Sample volume = 1 ml. Mercury concentration/pg 1-1 Blank .. .. 1 . . . . 2 . . . . 3 . . . . 4 . . .. 5 . . . . 6 . . . . 7 . . . . 8 . . . . e . . . . 10 .. .. Sediments Dispersive hTon-dispersive <1 <0.05 14 10 16 7 12 11 6 6 9 10 9 12 Shrimps r---_h ---_ Dispersive Non-dispersive <1 t0.05 3 4 3 3 3 3 3 3 8 6 4 3 3 4 3 4 3 4 7 - -984 SHORT PAPERS Conclusion Analyst, Vol. 105 It has been demonstrated that by careful optical optimisation, a simple inexpensive non- dispersive mercury fluorescence spectrometer can achieve detection limits comparable to those obtained with more complex systems. The instrument is both sensitive and precise and allows a rapid throughput of samples without the need to alter the operating conditions frequently. The authors acknowledge the help of Mr. L. Best with the fabrication of the spectrometer. This work is published by permission of the Directors of Tioxide Inteirnational Limited. References 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. Ure, A. M., Anal. Chim. Acta, 1975, 76, 1. Thompson, K. C., and Reynolds, G . D., Analyst, 1971, 96, 271. Muscat. V. I.. and Vickers. T. T.. Anal. Chim. Acta. 1971, 57, 23. Thomerson, D. R., Int. Lab., 1377 (January), 57. Caupeil, J . E., Hendrikse, P. W., and Bongers, J. S., Anal. Chim. Acta, 1976, 81, 53. Rigin, V. I., 21. Anal. Khim., 1979, 34, 261. Shimomura, S., and Hiroto, R., Anal. Lett., 1973, 6, 613. Thompson, K. C., Lab. Pract., 1972, 21, 645. Nakahara, T., Tanaka, T., and Musha, S., B d l . Chem. Soc. Jpn., 1978, 51, 2020. Hatch, W. R., and Ott, W. L., Anal. Chem., 1968, 40, 2085. Received March 25th, 1980 Accepted May 16th, 1980