437 JOURNAL O F ANALYTICAL ATOMIC SPECTROMETRY MARCH 1993 VOL. 9 High-sensitivity Detection of Selenium and Arsenic by Laser-excited Atomic Fluorescence Spectrometry Using Electrothermal Atomization* U. Heitmann T. Sy and A. Hese lnstitut fur Strahlungs- und Kernphysik Technische Universitat Berlin HardenbergstraRe 36 D- 70623 Berlin Germany G. Schoknecht Bundesgesundheitsamt lnstitut fur Sozialmedizin und Epidemiologie General-Pape-StraRe 62 0-72107 Berlin Germany High-sensitivity detection of the trace elements selenium and arsenic is reported. The method applied is laser-excited atomic fluorescence spectrometry using electrothermal atomization within a graphite furnace atomizer. For the production of tunable laser radiation in the vacuum ultraviolet (VUV) spectral region a laser system was developed that consists of two laboratory-built dye lasers pumped by a Nd:YAG laser.The laser radiations are subsequently frequency doubled and sum frequency mixed by non-linear optical KDP or BBO crystals respectively. The system works with a repetition rate of 20 Hz and provides output energies of up to 100 pJ in the VUV at a pulse duration of 5 ns. The investigations focused on the detection of selenium and arsenic in aqueous solutions and in samples of human whole blood. From measurements on aqueous standards detection limits of 1.5 ng I-' for selenium and 5.4 ng I-' for arsenic were obtained with corre- sponding absolute detected masses of only 15 or 54 fg respectively. The linear dynamic ranges spanned six orders of magnitude and good precision was achieved.In the case of human whole blood samples the recovery was found to be within the range 96-104O/0. The determination of the selenium content yielded medians of 119.5k17.3 pg I-' for 200 frozen blood samples taken in 1988 and 109.1 k15.6 pg I-' for 103 fresh blood samples. Keywords Laser-excited atomic fluorescence spectrometry; electrothermal atomization; detection of selenium and arsenic; vacuum ultraviolet spectral region; human whole blood Within recent years enormous interest has arisen in high- sensitivity detection methods in the medical and biological sector. In particular the investigation of trace elements and their influence on the human organism has attained increasing importance.',' Many methods for example atomic absorption (AAS) or fluorescence spectrometry (AFS) inductively coupled plasma (ICP) techniques neutron activity analysis (NAA) and related techniques have become established in the analytical They make the detection of numerous elements poss- ible but some difficulties often occur when biological samples have to be analysed.This is mostly caused by spectral or chemical interferences during the meas~rements.l'-'~ The introduction of a laser as the excitation source in contrast to conventional discharge lamps leads to almost complete elimination of such problems. This technique is called laser-excited atomic fluorescence spectrometry ( It offers double selectivity because of the efficient excitation of specific atomic transitions with spectrally narrow laser radi- ation followed by the spectrally selective detection of the fluorescence of the element.Therefore even in the most complex samples background-free analytical signals can be obtained without supplementary compensation being neces- sary. Moreover the high laser intensities of pulsed systems lead to saturation of the fluorescence of the element which reduces the influence of laser fluctuations and implies lower detection limits. The combination of LEAFS with the well studied electrother- mal atomization within a graphite furnace (ET-LEAFS) yields further improvements because previous experience with regard to sample preparation and handling can be u ~ e d . ' ~ ' ~ In addition only low sample volumes are required typically 10 pl are used in the present experiments. A recent overview on the * Presented at the XXVIII Colloquium Spectroscopicurn Internationale (CSI) Post-Symposium on Graphite Atomizer Techniques in Analytical Spectroscopy Durham UK July 4-7 1993.topic of ET-LEAFS and the superiority compared with conven- tional analytical methods has been given by Sjo~trom,'~ and in an article by Smith et a1.,20 who summarized the detection limits for several elements obtained by different methods. The best detection limit reported of 0.005 ng I-' was achieved for thallium which corresponds to a mass of 0.1 fg.21 The present investigations were focused on the high- sensitivity detection of the trace elements selenium and arsenic. Selenium is an essential trace element for humans. It is responsible for circulatory disturbances and diseases of the heart,22-24 and received worldwide attention in connection with Keshan disease that appeared in a province of China in 1979 and which was explained by selenium defi~iency.~',~~ Former investigations have also shown a protective role of selenium against c a n ~ e r ~ ~ ' ~ ~ and it also has some de-toxifying effects on heavy metals.29 On the other hand in high concentrations selenium shows toxic properties.The average selenium content in human blood of German citizens is normally in the region of about 100 pg 1-'. Although numerous investigations have been made in the past the influence of selenium on the human organism has not been completely clarified. Arsenic is a non-essential trace element but it is also of enormous interest in medi~ine,~'.~~ for example high concen- trations are related to a chronic liver disease.32 An important factor is the high toxicity of arsenic even in moderate concen- trations because it is used extensively in agriculture and industry.33 The extremely short excitation wavelengths of selenium and arsenic in the vacuum ultraviolet (VUV) spectral region are a real challenge for an analytical detection system.The lines most often used for these elements are 196.0 and 193.7 nm respectively. Therefore the aim of the work was the develop- ment of a laser system that enables the efficient generation of tunable radiation in the VUV and the application of the system to the high-sensitivity detection of trace elements by ET-LEAFS. The laser radiation is produced by sum frequency438 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1993 VOL.9 mixing (SFM) in a non-linear BBO crystal. An alternative method is stimulated Raman shifting (SRS),34,35 but the appli- cation of SFM yields higher conversion efficiencies in this spectral region.36 For wavelengths below 190 nm where BBO shows an increasing absorption the situation is changed and SRS would be the preferred method. In the present paper results obtained for the detection of selenium and arsenic in aqueous solution are presented. The selenium contents in 303 samples of human whole blood were determined in co-operation with the German Bundes- gesundheitsamt (BGA) in Berlin who study the effects of trace elements on the human organism. To our knowledge this is the first application of ET-LEAFS to the detection of both of these elements.Experimental VUV Laser System For the production of tunable VUV laser radiation the system which is depicted schematically in Fig. 1 was developed. It consists of two laboratory-built and symmetrically arranged dye lasers; each laser has one oscillator and one amplifier stage. For the oscillators grazing incidence arrangements with retroreflectors and double prism beam expansions were chosen. This is the best compromise to get high output energies in combination with small bandwidths. A commercial Nd:YAG laser (Lumonics HY 500 special version) was used as the pump source for the two sychronously pumped dye lasers. This is internally frequency doubled and provides output energies of up to 100 mJ with a pulse duration of 8 ns at a repetition rate of 20 Hz.The system also enables repetition rates of 50 and 100Hz. One dye laser operates in the red spectral region around 641 nm and the other runs in the yellow region at 565 nm. The laser radiation of the second laser is subsequently frequency doubled within a KDP crystal. The crystal is followed by a quartz compensator which rotates in the opposite direction and eliminates the beam shift during angle tuning to obtain optimal type-I phase matching. The laser beams at 641 and 283nm are then combined and sum frequency mixed within a BBO crystal. This yields the required VUV laser radiation at 196 nm which is finally separated from the remaining fundamental waves by a Pellin-Broca prism. The laser radiation in the VUV is guided under an atmosphere of pure nitrogen because of the absorption of the oxygen present in the laboratory air which is about 5% per meter.The laser system produces output energies of up to 100 pJ at a pulse duration of 5 ns in the VUV which means a peak power of 20kW. The corresponding spectral bandwidth is 13 GHz and the relative standard deviation (RSD) of the pulse fluctuations was determined to be 25%. A wavelength scan is achieved under control of a personal computer which scans both dye lasers by stepping motor drives and also rotates the KDP crystal. The BBO crystal is not moved. To obtain long- term stability of the laser system both dye laser oscillators as well as the non-linear optical crystals are temperature stabilized to within k0.05 "C at about 33 "C. In this way stable con- ditions over several days are achieved.196 nm A 641 nrn Dye laser 532 nm .I- ll Fig. 1 Schematic arrangement of the tunable VUV laser system A detailed description of the VUV laser system with corre- sponding system performances can be found in ref. 36. ET-LEAFS System 'The analytical arrangement for the ET-LEAFS measurements is depicted schematically in Fig. 2. The VUV laser radiation is focused through a pierced hole in the mirror into a commercial graphite furnace atomizer ( Perkin-Elmer HGA-500) equipped with an autosampler (Perkin-Elmer AS-40). Pyrolytic graphite coated graphite tubes with L'vov platforms were used. The analytical signal is observed as the backward fluorescence (i.e. backward in relation to the direction of the laser beam used for excitation) of the excited atoms collected via the mirror and focused by a lens onto the entrance slit of a monochroma- tor (B & M Spektronik d=0.1 mm f=25 cm).For signal detection a solar blind photomultiplier (Hamamatsu R 166 UH) is used to avoid background signals caused by thermal radiation of the atomizer. The signal is subsequently preampli- fied (Ortec 9301) and worked up by a fast boxcar integrator (Stanford Research SR 250). A personal computer finally handles the signal recording and also controls the laser system and the atomizer. Sample Preparation Aqueous solutions are obtained by dilution of standard solu- tions of the elements Titrisol (Merck 1 g 1-I) with multiply distilled and de-ionized water (Millipore). The acid level is adjusted to be 0.2% v/v in nitric acid.Because of the high volatility of selenium and arsenic a commercial palladium nitrate modifier (Merck 10 g 1-I) is used for chemical modifi- cation Depending on the element the modifier is further diluted to concentrations of 0.2-0.5 g 1-1 of palladium. For blood samples a reference blood standard Seronorm (Nycomed Batch 010010) with a certified selenium content of 93 & 4 pg l-' was used. Real blood samples from the BGA are transferred directly into ethylenediaminetetraacetic acid (H,EDTA) coated tubes. Owing to the high sensitivity of the system the blood samples were further diluted normally by a factor of 10 or 20. The samples should then contain 0.2% v/v nitric acid and 0.01% v/v Triton X-100. The addition of Triton X-100 causes a reduction in the surface tension enabling better sample deposition onto the L'vov platform and also reduces residues within the vessel of the autosampler.During an analytical cycle 10 p1 of the sample are first transferred into the graphite tube followed by the same amount of modifier. Thermal treatment then takes place using a standard graphite furnace programme as first proposed by Welz et a1.37 and only slightly modified for the present experi- ments. The ashing temperature is raised to 1000°C and an atomization temperature of 2200 "C was chosen. This pro- cedure enables about 250 successive measurements to be made before the graphite tube has to be changed. Results Aqueous Selenium Samples The first investigations focused on the detection of selenium.A simplified energy-level scheme for selenium is given in Fig. 3. As already mentioned the main absorption line lies in the VUV spectral range at 196nm. From the excited level there are three possibilities for observation of fluorescence either resonant detection of the transition back to the ground state or of the cross transitions at 204 or 206 nm. The last two cases have the advantage of being free from laser stray light. To test the spectral behaviour of selenium the exit slit of the monochromator and the photomultiplier was replaced by an optical multichannel analyser (0-SMA Princeton Instruments). As seen from the spectrally and time resolved fluorescence after excitation at 196 nm (cJ Fig. 4) only the\ Boxcar 4 integrator __ I 2 Signal Preamplifier Printer Computer Photomultiplier controller ;” \ \ filter Trigger \ \ High vo I tag e Control unit u \ Fig. 2 Schematic set-up of the ET-LEAFS system Nd YAG pump laser Fig.3 Simplified energy-level scheme of selenium \ Dye lasersT - andSFM three expected fluorescence lines at 196 204 and 206 nm appear. No secondary transitions or spectral interferences of other elements can be observed. Only some laser stray light can be seen in front of and behind the main peak. However the 0-SMA system used was not optimized for applications in the VUV spectral range. This explains the high sample concentration used in this example. For high-sensitivity measurements the photomultiplier is therefore used which provides a gain of about 500. A further interesting point is the dependence of the fluor- escence signal on the excitation energy.With increasing laser energy the element fluorescence shows a saturation effect. This can be seen in Fig. 5 for excitation at 196 nm and observation on the cross transition at 204nm. The two curves represent different optical adjustments on different days. Nevertheless both curves show similar behaviour and yield the same satu- ration energy of 1.0f0.2 pJ in the presence of the palladium modifier. With a cross-section of the laser beam within the atomizer of 0.2 mm’ this corresponds to a saturation intensity 439 190 195 200 205 21 0 Wavelengthhm Fig.4 Time and spectrally resolved fluorescence of an aqueous selenium sample over a time period of 3.2 s; excitation at 196 nm of 100+20 kW cm-’. The saturation level is defined as the level where half of the maximum excitable fluorescence is obtained.Some measurements were also made without the presence of the modifier and even this led to a higher saturation energy of 2.1 f 0.4 @. These results were considered again on different days with modified optical parameters (e.g. beam cross-section). The calibration curve for aqueous selenium samples that is given in Fig. 6 shows excellent linear behaviour. The detection limit follows from extrapolation of the calibration curve for three times the standard deviation (SD) of a blank signal to 1.5ng1-I. With a sample volume of lop1 this leads to an absolute detected mass of only 15 fg which means an enormous increase in sensitivity compared with conventional systems. From this detection limit a linear dynamic range over six440 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1993 VOL.9 80 500 9 *O - 0 2 4 6 8 10 12 14 16 18 E nergytpJ Fig.5 Dependence of the fluorescence signal on the laser energy (with Pd modifier); excitation at 196 nm; and observation at 204 nm. See text for explanation of the two curves 100000 1 1 L c. .- f 1000 - same blood sample led to slightly higher values of 6.5% for the RSD and 1.1% for the precision of the mean. As a first practical application a co-operative study with the BGA was undertaken. During an epidemiological study some thousands of whole blood samples were taken from the population of Berlin and the selenium content was deter- mined.38 Firstly 200 blood samples that were taken in 1988 and subsequently frozen were analysed.This yielded the distribution of the selenium content given in Fig. 7. The mean value was found to be 122.0 pg 1-I. From the Gaussian fit a median of 119.5 pg 1-1 was obtained with a corresponding SD of 17.3 pg 1-l. In Fig. 8 where the selenium content is depicted as a function of the age of the test person no tendency can be seen either with regard to the age or to the sex of the test person. Some extremely high values of a few samples can only be explained by an additional intake of selenium preparations. In addition to the measurements mentioned above 103 fresh whole blood samples were also investigated which led to a mean value of 114.2 pg 1-I. The median was determined to be 109.1 pg l-' with a corresponding SD of 15.6 pg 1-l.Within statistical uncertainties this is the same result as was obtained from the frozen samples. The comparative investigation of analytical samples with different detection methods was an important aspect. An overview of the expected or certified values is given in Table 1 and the measured values obtained with the present system and with a flow injection atomic spectrometry system (Perkin- Elmer FIAS-200) at the Freie Universitat (FU) of Berlin. Investigations with neutron activitation analysis (NAA) at the Hahn-Meitner-Institut (HMI) in Berlin were also proposed. I0 ' I I I I I 0.1 1 10 100 1000 10000 Selenium concentratiordpg I -' 40 I Fig. 6 Calibration graph for aqueous selenium samples detection limit 1.5 ng 1-' orders of magnitude is obtained. The calibration curve bends over with concentrations above 5000 pg 1-1 and unfortunately concentrations above 1000 pg 1-1 cause contamination of the graphite tube.Therefore measurements were restricted up to this practical limit. Normally an unknown sample is analysed with three success- ive measurements and the mean value taken. In this way standard deviations of between 2 and 3% were obtained. The long-term behaviour of the system was investigated by 34 successive measurements of the same 5 pg 1-1 aqueous selenium sample which represents a time period of nearly 2 h. An RSD for an individual measurement of 5.1 % was achieved which resulted in a precision of 0.9% for the mean value. Whole Blood Samples To calibrate the system for use with whole blood samples the reference blood standard Seronorm was used.When determin- ing the selenium concentrations in unknown samples a cali- bration curve was first recorded using five standard solutions. Each individual sample was then analysed three times and the mean value was compared with the calibration curve to give the real selenium content. In all cases the area of the fluor- escence signal i.e. integration of the fluorescence of the element at the observation wavelength over time was evaluated. This enables the alternative procedure of aqueous selenium stan- dards to be used because equal element concentrations yield equal signal areas independent of the sample matrix. From investigations with the standard additions method the recovery of a known selenium blood concentration was determined to be within the range 96-104%.Owing to the more complex sample matrix successive measurements of the 70 84 98 112 126 140 154 168 182 196 Selenium contenttpg I - ' Fig.7 Distribution of the selenium content of frozen whole blood samples mean 122.0; median 119.5; and SD 17.3 pg 1-' 200 C cn Q Q) 100 a 0 .O 0. 0 Fig. 8 Selenium content of whole blood samples as a function of the age of the test subjects 0 female 121 subjects; and 0 male 79 subjectsJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1993 VOL. 9 441 250 I v) C 3 - .- 200 f? e +- .- 150 - Table 1 Comparison of results obtained with different analytical methods - - - Sample Seronorm Human blood - A B C Sheep blood - A B C D Range expected/pg I - ' 93 +4 M 100 x 200 M 400 x 300 ET-LEAFS,* FIAS,? found/pg 1 - found/pg I - 92+3 95f 10 83+2 92+_11 105 f 9 88+2 96+7 104 + 2 105_+3 98+4 258 & 5 288 2 16 334$5 331 & 13 300+4 321 + 14 * Present group. 7 Schafer et al.FU Berlin.39 Unfortunately the reactor only started operation fairly recently after service work so the first results are not yet available. For these measurements the whole blood standard three human whole blood samples and four blood samples of sheep who had been fed on different diets and had been given different selenium additives in their food were selected. On the whole good agreement of the results for these samples was obtained with only some larger differences for two sheep samples. This can be explained by dilution uncertainties during sample preparation with the FIAS system.Aqueous Arsenic Samples The second trace element investigated with the system was arsenic. This is a more complicated situation because of the short excitation wavelength and more complex energy-level scheme (cf. Fig. 9). Two possibilities exist for excitation from the ground state. The most frequent transition is that at 193.7nm followed by two cross transitions at 243.7 and 245.7 nm for non-resonant observation. The other transition is at 197.3 nm with a cross transition at 249.3 nm. Owing to the high temperature within the graphite furnace during sample atomization some collision-induced transitions into upper energy levels also exist with corresponding fluorescence lines. This can be seen in the spectrally resolved fluorescence spectrum of arsenic obtained with the 0-SMA system used (cf.Fig. 10). After excitation at 193.7 nm only the resonance fluorescence and the two cross transitions at 243.7 and 245.7 nm are expected. Nevertheless the other lines mentioned also appear in particular the anti-Stokes line at 189.0 nm. The calibration curve for aqueous arsenic samples with excitation at 193.7 nm and observation at 245.7 nm shows a similar behaviour to the one obtained for selenium. The I 189.0 nm 1 I I I I l193.7 nm 197.3 nm I I I I r Fig. 9 Simplified energy-level scheme of arsenic 300 0 180 190 200 210 220 230 240 250 260 Wavelengthln m Fig 10 Spectrally resolved fluorescence of an aqueous arsenic sample (400 pg I-'); excitation at 193.7 nm detection limit was 5.4 ng 1-' which is slightly worse than the one for selenium but in comparison with other methods it is still an excellent result.Again from the detection limit a linear dynamic range over nearly six orders of magnitude is obtained. Different combinations of wavelengths for excitation and observation were tested for arsenic. The results for the two best combinations are as follows. The first was excitation at 197.3 nm in combination with the unusual observation on the anti-Stokes line at 193.7 nm. Even in this case a fairly good detection limit of 500 ng 1-' was found which corresponds to an absolute detected mass of 5 pg. The RSD of a single measurement was found to be 5.9% and the precision was 1.0%. Owing to the special type of transition a high saturation energy of 15 pJ was obtained which means a saturation intensity of 1.5 MW cm-2.The best results were achieved with excitation at 193.7 nm followed by observation at 245.7 nm. As already mentioned the detection limit was determined to be 5.4 ng l-l which corresponds to an absolute detected mass of only 54fg with an RSD of 6.9% and a precision of 1.2% for the mean. The saturation energy was found to be 0.8 pJ which implies a saturation intensity of 80 kW cm-2. Conclusions The ET-LEAFS technique was applied to the determination of the trace elements selenium and arsenic in aqueous solution and in samples of human whole blood. The precision accuracy and sensitivity of the method are good and are comparable with or better than those of conventional methods. Because of the double selectivity of the method only simple sample preparation even in the case of whole blood samples is required and no supplementary compensation for the back- ground is necessary.Also the system operates under the full control of a computer enabling easy wavelength scanning for example between selenium and arsenic. The application of the system within an epidemiological study was shown to operate without any problems and promises good results in the future. To our knowledge this is the first time ET-LEAFS has been applied to the detection of the trace elements selenium and arsenic. One reason for this could be the experimental diffi- culties associated with achieving tunable laser radiation in the VUV. Furthermore the best detection limits reported to date for both elements were obtained. With some alterations to the experimentation further improvements are expected.For example lower detection limits should be obtainable by a reduction of the pulse fluctuations of the Nd:YAG laser and also by improved optics for fluorescence detection. An extension of the investigations to other trace elements or the simultaneous detection of elements using an optical442 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1993 VOL. 9 multichannel analyser in combination with several laser sources is intended. Owing to the complexity and the fairly high price of the system it should primarily be applied to the detection of several elements such as antimony and chromium where other systems have problems at low concentrations or with spectral interferences. Therefore ET-LEAFS should be seen as a comp- lementary technique to conventional methods.We thank Dr. K. Schafer and Mrs. K. Meyer for their helpful cooperation and comparative measurements with the FIAS system. This work was financially supported by the Deutsche Forschungsgemeinschaft (DFG). 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 References Tolg G. in Trace Element Analytical Chemistry in Medicine and Biology eds. Bratter P. and Schramel P. Walter de Gruyter Berlin 1988 vol. 5 p. 1. Fishbein L. Int. J. Environ. Anal. Chem. 1984 17 113. Holcombe J. and Hassell D. Anal. Chem. 1990 62 169R. Bozsai G. Schlemmer G. and Grobenski Z. Talanta 1990 37 545. Pyen G. and Browner R. Appl. Spectrosc. 1988 42 262. Vien S. and Fry R. Anal. Chem. 1988 60 465. DUlivo A. FUOCO R. and Papoff P.Talanta 1985 32 103. Blekastad V. Jonsen J. Steinnes E. and Helgeland K. Acta Med. Scand. 1984 216 25. Irons R. Schenk E. and Giauque R. Clin. Chem. 1976,22,2018. Parsons M. Major S. and Forster A. Appl. Spectrosc. 1983 37 411. Saeed K. and Thomassen Y. Anal. Chim. Acta 1981 130 281. Welz B. and Schubert-Jacobs M. J. Anal. At. Spectrom. 1986 1 23. Aller A. J. and Garcia-Olalla C. J. Anal. At. Spectrom. 1992 7 753. Bolshov M. Zybin A. and Smirenkina I. Spectrochim. Acta Part B 1981,36 1143. Falk H. Prog. Anal. At. Spectrosc. 1980 5 181. Sjostrom S. and Mauchien P. Spectrochim. Acta Rev. 1993 15 153. Omenetto N. Spectrochim. Acta Part B 1989 44 131. Goforth D. and Winefordner J. Talanta 1987 34 290. 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 Sjostrom S.Spectrochim. Acta Rev. 1990 13 407. Smith B. Glick M. Spears K. and Winefordner J. Appl. Spectrosc. 1989 43 376. Smith B. Farnsworth P. Cavalli P. and Omenetto N. Spectrochim. Acta Part B 1990 45 1369. Cohen H. and Avissar N. Prog. Clin. Biol. Res. 1993 380 191. Ringstad J. Jacobsen B. and Thomassen Y. J. Trace Elem. Electrolytes Health Dis. 1987 1 27. Oster O. Prellwitz W. Luley C. Meinertz T. Geibel A. and Kasper W. in Trace Element Analytical Chemistry in Medicine and Biology eds. Brgtter P. and Schramel P. Walter de Gruyter Berlin 1987 vol. 4 p. 579. Keshan Disease Research Group of The Academy of Medical Sciences Chin. Med. J. (Beijing Engl. Ed.) 1979 92 471. Yang G. Chen J. and Wen Z. Adv. Nutr. Rex 1984 6 203. Lange J. Toxicol. Ind. Health 1991 7 319. Schrauzer G. in Trace Element Analytical Chemistry in Medicine and Biology eds. Bratter P. and Schramel P. Walter de Gruyter Berlin 1987 vol. 4 p. 403. Shamberger R. in Trace Element Analytical Chemistry in Medicine and Biology eds. Bratter P. and Schramel P. Walter de Gruyter Berlin 1987 vol. 4 p. 181. Smith A. Hopenhayn-Rich C. Bates M. Goeden H. Hertz- Picciotto I. Duggan H. Wood R. Kosnett M. and Smith M. Enuiron. Health Perspect. 1992 97 259. Chen C. J. Chen C. W. Wu M. and Kuo T. Br. J. Cancer 1992 66 888. Calmus Y. Poupon R. Brijnen M. Renaudeau C. and Darnis F. Gastroenterol. Clin. Biol. 1982 6 910. Ricci G. Shepard L. Colovos G. and Hester N. Anal. Chem. 1981 53 610. Leong M. D’Silva A. and Fassel V. Anal. Chem. 1986,58 2594. Schomburg H. Dobele H. and Riickle B. Appl. Phys. B 1983 30 131. Heitmann U. Kotteritzsch M. Heitz S. and Hese A. Appl. Phys. B 1992 55 419. Welz B. Schlemmer G. and Mudakavi J. R. J. Anal. At. Spectrom. 1988 3 695. Schoknecht G. SozEp. Hefte 1985 1 1. Schafer K. personal communication. Paper 3 J04639G Received August I 1993 Accepted September 28 1993