Rapid Determination of Selenium in Soils and Sediments Using Slurry Sampling- Electrothermal Atomic Absorption Spectrometry Journal of Analytical Atomic Spectrometry IGN ACIO LO PEZ-G A R C ~ A M ATEO s ANCHEZ-M ER LO s AND MANUEL H ERN ANDEZ-c ORDOB A* Department of Analytical Chemistry Faculty of Chemistry Unimmity of Murcin E-30071 -Murcia Spain An electrothermal atomic absorption spectrometric procedure for the rapid determination of selenium in soil and sediment samples is presented. The samples were suspended in concentrated (40% m/v) hydrofluoric acid containing 1% m/v nickel nitrate and then injected into the electrothermal atomizer. A fast heating programme with no conventional ashing step was used. Platform atomization at 2300°C and Zeeman-effect background correction were used.Calibration was performed directly by using aqueous standards prepared in a 10% v/v hydrofluoric acid solution. The determination limit was 0.1 pg g-' of selenium when using the maximum recommended slurry concentration of 10% m/v. The results obtained for five certified reference materials show the reliability of the procedures. Keywords Electrothermal atomic absorption spectrometry; fast heating programme; slurry sampling; selenium; chemical modijication; soil; sediment The dual nature of selenium as both an essential and potcntially toxic element has led to growing interest in the development of rapid and reliable methods for its determination. The high sensitivity of ETAAS makes it a suitable technique for such a purpose. However the determination involves a number of problems.Most importantly volatile selenium compounds may be lost during the dissolution stage or during tlie pre- atomization heating steps inside the atomizer' and in addition severe spectral and chemical interferences may hinder the determination.'-I5 It is therefore not surprising that in spite of the great interest focused on slurry sampling-ETAAS methods,I6 there are few references to their use in selenium determination. To the best of our knowledge following the work of Ebdon and Parry on coal analy~is,'~ the slurry sampling ETAAS approach has only been used for determining selenium in fly ash,'* rnilk,lg flour2' and seafood.21 The amount of selenium in soils is highly variable the levels mainly reflecting the weathering of parent materials all hough anthropogenic contributions may also be of importance.The mean level for a variety of soils has recently been reported22 to be about 0.5 pg g-'. This is a level within the range of ETAAS a technique which is widely used nowadays in labora- tories dealing with environmental analysis. Based on the fact that slurry sampling-ETAAS methodology is particularly suitable for trace element analysis in soils and a study was made to develop rapid procedures for determining selenium in this type of sample. The results reported here show that good analytical results can be otrtained with a considerable saving of time provided that Zeeman- effect background correction is used and that the samples are previously slurried in a concentrated hydrofluoric acid schhon.* To whom correspondence should be addressed. EXPERIMENTAL Instrumentation Measurements were made with an ATI-Unicam (Cambridge UK) 939 QZ atomic absorption spectrometer equipped with a GF90 electrothermal atomizer. The spectrometer was pro- vided with both a Zeeman-based and a deuterium-arc back- ground corrector. The source of radiation was a selenium hollow cathode lamp (Photron Victoria Australia) operated at 10 mA. Measurements were performed at 196.0 nm using a spectral bandwidth of 1.0 nm. For comparative purposes measurements were obtained with a Perkin-Elmer (Norwalk CT USA) Model 1 lOOB atomic absorption spectrometer equipped with a deuterium-arc background corrector and an HGA-400 electrothermal atomizer the source of radiation in this instance being an electrodeless discharge lamp operated at 200 mA from an external power supply (Perkin-Elmer System 2).Pyrolytic graphite coated graphite tubes with platforms obtained from ATI-Unicam (reference 9423 393 95 19 1 ) and Perkin-Elmer (reference B050 5057) were used. Argon was used as the purge gas the flow rate being 300 ml min-' except during atomization when the flow was stopped. For preparation of the sample a Fristch Pulverisette (Idar-Oberstein Germany) agate ball-mill of 80 ml capacity was used. Reagents A selenium (IV) standard solution (1000 pg ml-') was obtained from Panreac (Barcelona Spain) and diluted as necessary to obtain working standards. Concentrated hydrofluoric acid (40% m/v) and all other chemicals used were obtained from FIuka (Buchs Switzerland).High-quality water obtained using a Milli-Q water purification system (Millipore; Milford MA USA) was used exclusively. Plasticware [poly(propylene)] of the type commonly used for clinical purposes was used for storing and handling the solutions or suspensions containing hydrofluoric acid. Procedure Five soil and sediment samples with certified selenium contents were used throughout this work. The samples were first ground for 15 min using the ball-mill and the resulting powders were kept in tightly closed plastic containers until analysis. No sieving was carried out. The suspensions were prepared by weighing the samples directly into Eppendorf tubes and then adding 1000 pl of the concentrated hydrofluoric acid solution containing 1 YO Ni(N03)2.6H20.The suspensions were manu- ally shaken (caution concentrated hydrofluoric acid solutions cause severe burns. Eyes and skin must be protected) and then left for 15 min after which the suspensions were again shaken and immediately 10 p1 aliquots were taken and manually injected into the electrothermal atomizer. The experimental Journal of Analytical Atomic Spectrometry October 1996 Vol. 11 (1003-1006) 1003Table 1 Instrumental and furnace parameters Wavelengt h/nm Bandpass/nm Background correction Atomizer type Injection volume/pl Calibration range/pg 1-' Characteristic mass/pg Slurry volume/ml Slurry concentration (YO m/v) Chemical modifier Hydrofluoric acid concentration (Yo v/v) 196.0 1 .o Zeeman-effect Pyrolytic graphite coated 10 5-200 25 1 0-10 1 % m/v Ni( N03)2*6H20 100 graphite tube + L'vov platform Step TemperaturePC RampPC s-l Hold/s 1 400 20 5 2* 2300 0 4 3 2650 0 4 * The flow of argon was stopped during the atomization step.conditions used for the preparation of the suspensions as well as the instrumental settings of the spectrometer are summarized in Table 1. The fast heating programme recommended is also given in Table 1. Calibration was performed using aqueous standards prepared in a solution containing 10% v/v concen- trated hydrofluoric acid and 1% m/v nickel nitrate. RESULTS AND DISCUSSION Halls26 demonstrated several years ago that the drying and ashing stages normally included in conventional furnace pro- grammes can be replaced by a modified drying stage thus shortening and simplifying the heating cycle.This fast heating m e t h ~ d o l o g y ~ ~ . ~ ~ was used throughout the present work. The temperature and the holding time for the modified drying stage are dependent on the characteristics of the electrothermal atomizer used. A 400°C drying temperature with a heating rate of 20°C s-l and a hold time of 5 s was used in all the experiment^.^^ In spite of the relatively low temperature inside the atomizer at the end of the modified drying stage prelimi- nary experiments showed that a chemical modifier was neces- sary because of pre-atomization loss of analyte. Consequently nickel nitrate was incorporated in the suspension medium. In addition it is known that when suspensions prepared from samples with a high silica content are atomized the graphite atomizer severely deteriorate^.^'*^' Bendicho and de Loos- V ~ l l e b r e g t ~ ' .~ ~ were the first to overcome this difficulty by adding hydrofluoric acid to the suspension medium. This simple approach which has subsequently been ~ s e d ~ ~ > ~ ~ * ~ ~ for the analysis of other samples with a high silica content was followed here. For preliminary experiments the samples were suspended in a medium containing 5% v/v concentrated hydrofluoric acid in addition to the nickel modifier and deuterium-arc background correction was used. Matrix Effects Preliminary experiments showed severe matrix effects. The atomization profile of aqueous selenium standards was unaffected by the presence of hydrofluoric acid but the analyt- ical signals obtained from suspensions prepared from sediment samples containing similar amounts of selenium were very low.The matrix effect was so severe that the signal obtained from a 120 pg 1-l selenium solution virtually disappeared when solid sediment was added to the solution to make a 0.5% m/v suspension (Fig. 1). In an attempt to clarify the exact nature of this depressing effect a number of experiments were per- formed by preparing suspensions from pure A1203 Fe203 O.* It J . . . 1 0.0 0.2 .................. . . . . . :: 0.0 L L :::It A I (c;l . . . . . . . . ................ g 0.2 5 0.0 0 1.5 1 .o 0.5 - . . . . 0.OIl 4 .... + I ..... (el 2.0 1.5 . . . . . . 1.0 1 _ . . !.A I - . . 0.5 ..... 0.0 0 1 2 3 4 Time/s Fig. 1 Absorbance uersus time curves for selenium (solid line) and background (dashed line) for (a) a 120 pg 1-' selenium solution in a 5% v/v hydrofluoric acid medium containing the nickel modifier; (b) (c) ( d ) and (e) as above in the presence of 0.5 0.4 0.6 and 3% m/v of sample BCSS- 1 Fe203 A1203 and SiO respectively.Deuterium-arc background correction. Si02 and soil samples always using an aqueous selenium solution containing 5% hydrofluoric acid and the nickel modi- fier as the suspension medium. The concentration of the suspensions prepared from oxides was chosen in such a way that a typical 5% m/v soil suspension was simulated. The selenium atomization profiles some of which are shown in Fig. 1 suggested that the severe depressing effect was mainly due to the presence of solid SO2. In addition the presence of iron was shown to lead to a significant baseline drift as a consequence of its spectral interference which cannot be corrected by the deuterium device.In order to avoid this baseline drift Zeeman-effect background correction was used for further experiments. Since the above results suggested that silica was mainly responsible for the matrix effect the possible benefit of increas- ing the hydrofluoric acid concentration in the suspension medium was considered. As expected the matrix depressing effect considerably decreased when the amount of hydrofluoric acid was increased. Fig. 2 shows the analytical signal obtained from 3% m/v suspensions prepared from NIST SRM 2711 Montana Soil sample when solutions containing different concentrations of hydrofluoric acid were used as the suspension medium. It is clear that a very high percentage of hydrofluoric acid was needed to obtain the maximum and constant signal.The use of concentrated hydrofluoric acid solutions which contain about 40% by mass of the pure acid can be hazardous for the operator and several attempts were made to avoid its use. However because the level of selenium in soils and sediments is very low suspensions containing a high percentage of solid matter must be used which necessitates the direct use of concentrated acid solutions. This can be seen in Fig. 3(a) where results for some of these experiments are shown. A number of suspensions were prepared from the Montana Soil sample with different amounts of solid matter using 500 p1 of concentrated hydrofluoric acid. After 15 min of contact between the sample and the acid 500 pl of pure water were 1004 Journal of Analytical Atomic Spectrometry October 1996 Vol.110.30 . 0.00 1 1 0 20 40 60 80 100 Hydrofluori acid concentration (?h v/v) Fig. 2 Effect of the percentage of concentrated hydrofluoric ;acid in the suspension medium on the selenium peak area. A and B mr:asure- ments obtained 15 min and 24 h after the suspension prephxation respectively added to reduce the acid concentration and measurements were taken from the whole slurry (curve A) and aftx the suspensions had been filtered through chromatographic mem- brane filters from the supernatant solution (curve B). The results showed that interference could only be overcoine for slurry concentrations below 5 %. The signal obtained from the supernatant solutions was as expected unaffected by the matrix effect which appeared to be solely due to the slurried solid matter.On the other hand when the samples were directly slurried in concentrated hydrofluoric acid there was a linear relationship [Fig. 3(b)] between the percentage of solid matter in the slurry and the analytical signal obtained from the whole suspension (15 min after its preparation). Other experiments performed using dilute hydrofluoric acid solutions as the suspension medium and mild heating treatments with a heating block or a microwave oven produced the same result. In order to avoid the matrix effect completely and relialdy for high slurry concentrations (up to 10% m/v) it is necessary for the samples to be slurried directly in concentrated hydrofuoric acid. This does not involve the total dissolution of the sample and in fact selenium was only partially transferred to the liquid phase by the action of hydrofluoric acid.By filtering the solid residues through chromatographic membrane filter and measuring selenium in the supernatant solutions it was shown that only about 70-75% of the total selenium present was extracted. This value increased to 80-85% when the suspen- sions were filtered 24 h after preparation. On the other hand one important practical point must be 0.3 (a) I I I I I I 1 o.2; 0.1 00 0 2 4 6 8 1 0 Slurry concentration (./o) Fig.3 Effect of the slurry concentration on selenium peak area. (a) Using a 50% v/v hydrofluoric acid solution as the suspension medium. A and B signals obtained from the whole suspensions and from the supernatant solutions respectively; (b) using concentrated hydrofluoric acid.A B and C signals from suspensions prepared from samples SRM 2711 PACS-1 and BCSS-1 respectively noted since the use of concentrated hydrofluoric acid solutions may at first sight appear to be harmful to the electrothermal atomizer. In order to obtain the results summarized here nearly 1000 heating cycles were performed and the quartz windows of the atomization device were periodically checked. No signs of deterioration were noted nor did the pyrolytic graphite platforms show any indication of damage or prema- ture ageing. The presence of 1% nickel nitrate in the suspension medium proved suitable for avoiding premature loss of the analyte.On the other hand the atomization temperature was varied in the range 1900-2600 "C maximum analytical signals being obtained at 2300°C. Calibration and Results In order to confirm that concentrated hydrofluoric acid com- pletely overcame the matrix effect suspensions were prepared from five soil and sediment samples and standard additions calibration graphs were obtained. The slopes of these graphs (Table 2) did not show significant differences (95% confidence level) from those obtained with aqueous standards which validates the direct and more simple calibration. It is important to point out that working selenium standards prepared in concentrated hydrofluoric acid proved unstable. For this reason aqueous standards were prepared in a dilute (10% v/v) hydrofluoric acid solution.Using the conditions recommended a characteristic mass of 25 pg Se was obtained. The detection limit was calculated using the criterion based on three times the standard deviation. For this purpose the standard deviation obtained from a plot of the analytical signal against the slurry concentration for a representative sample was used instead of the standard devi- ation of the blank.35 Using this approach the detection limit for a 10% m/v suspension was calculated to be 0.03 pg g-' Se which corresponds to a quantification limit of 0.1 pg 8-l. Table 3 summarizes the final results obtained for selenium determination in five certified soil and sediment samples using both the standard additions method and the direct calibration against aqueous standards prepared in 10% v/v hydrofluoric acid solution. There were no significant differences (95% confidence level) between the results obtained and the certi- fied values.Once the final experimental conditions had been optimized and taking into account that the deuterium-arc corrector is still the most commonly used correction device in most routine analysis laboratories the need to use Zeeman-effect back- ground correction was reconsidered. This was facilitated by the fact that the spectrometer used was provided with both correction systems which allowed a more realistic comparison than that based on two different spectrometers. Fig.4 shows selenium atomization profiles obtained from a 5% m/v suspen- sion prepared from the PACS-1 sample. Although the slopes of the calibration graphs obtained from aqueous standards were similar when either correction system was used (1.78 x s pg-' 1 for Zeeman and deu- terium-arc devices respectively) it is evident that Zeeman- and 1.89 x Table 2 Slopes of the standard additions calibration graphs Sample Slope* + S / I O - ~ s pg-' I Aqueous solution 1.78 & 0.02 NIST SRM 2709 (San Joaquin Soil) 1.75 f 0.04 NIST SRM 2711 (Montana Soil) 1.81 k0.03 NIST SRM 2704 (River Sediment) 1.78 0.03 1.74 & 0.04 NRCC PACS-1 (Marine Sediment) NRCC BCSS-1 (Marine Sediment) 1.82 f 0.03 * Each graph was constructed with four points. Three measurerncnts were obtained from each point.Journal of Analytical Atomic Spectrometry October 1996 Vol. 11 1005Table 3 Results for the determination of Se in certified reference materials Content* +s/pg g-' Sample Description Certified Direct calibration Standard additions 0.43 f 0.06 0.44 0.02 0.43 -t 0.02 NRCC BCSS-1 Marine Sediment NRCC PACS-1 Marine Sediment 1.09f0.11 1.03 f 0.07 1.05 f 0.06 NIST SRM 2704 River Sediment 1.12 f 0.05 1.1 3 f 0.05 1.10 f 0.05 NIST SRM 2709 San Joaquin Soil 1.57k0.06 1.60 f 0.04 1.57 k 0.05 NIST SRM 2711 Montana Soil 1.52 k0.14 1.49 +_ 0.03 1.51 k0.04 * Mean k standard deviation for five suspensions... . . . . . . _ U - 0.20 1 0.8 5 0.15 0.6 0.10 0.4 I fbi I k- AT-/ . . . ... '. . . . . . . .. . . . 0 1 2 3 Time/s Fig. 4 Absorbance versus time curves for selenium (solid line) and background (dashed line). A 5% m/v PACS-1 Marine Sediment slurry prepared in concentrated hydrofluoric acid containing 1 YO nickel nitrate was used.(a) Deuterium-arc background correction system; (b) Zeeman-effect background correction system effect background correction is mandatory. This was confirmed by using another spectrometer (Perkin-Elmer Model 1100B) equipped with a deuterium-arc background corrector and a selenium electrodeless discharge lamp. The slope of the cali- bration graph was similar (1.80 x lop3 s pg-' 1) but the atomiz- ation profiles of selenium obtained from suspensions proved unsuitable for quantification. The foreseeable improvement in performance due to the electrodeless discharge lamp could not be checked on the ATI-Unicam instrument since at present this radiation source is not available. CONCLUSION The direct suspension of soil and sediment samples in concen- trated (40% m/v) hydrofluoric acid permits the rapid and reliable determination of selenium by ETAAS without the need for a dissolution stage.The acid acts as a true chemical modifier by simplifying the matrix during the heating cycle. Suspensions as concentrated as 10% m/v can be introduced into the electrothermal atomizer with no damage to the graphite material or to the quartz windows of the atomizer. However personal safety precautions when handling concen- trated hydrofluoric acid solutions must be taken into account. Zeeman-effect background correction is mandatory. The results confirm that slurry methodology is particularly suitable when dealing with trace analysis in soil and sediment samples since background levels are not high and the time-consuming and hazardous dissolution stage is avoided altogether.This work was financially supported by the Spanish DGICYT (Project PB93-1138). M. Sanchez-Merlos acknowledges a fel- lowship from DGICYT. REFERENCES 1 Radziuk B. and Thomassen Y. J. Anal. At. Spectrom. 1992 7 397. 2 Saeed K. and Thomassen Y. Anal. Chim. Acta 1981 130 281. 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 Saeed K. and Thomassen Y. Anal. Chim. Acta 1982 143 223. Carnrick G. R. Manning D. C. and Slavin W. Analyst 1983 108 1297. Bauslaugh J. Radziuk B. Saeed IS. and Thomassen Y. Anal. Chim. Acta 1984 165 149. Welz B. Schlemmer G. and Mudakavi J. R. J . Anal. At. Spectrom. 1988 3 93. Nham T. T. and Brodie K. G. J. Anal. At. Spectrom.1989,4,697. DoEekalova H. DoEekal B. Komarek J. and Novotny I. J. Anal. At. Spectrom. 1991 6 661. Aller A. J. and Garcia-Olalla C. J. Anal. At. Spectrom.. 1992 7 753. Garcia-Olalla C. and Aller A. J. Anal. Chim. Acta 1992,259,295. Kumpulainen J. and Saarela K.-E. J . Anal. At. Spectrom. 1992 7 165. Welz B. Bozsai G. Sperling M. and Radziuk B. J. Anal. At. Spectrom. 1992 7 505. Johannessen J. K. Gammelgaard B. Jerns O. and Hansen S. H. J. Anal. At. Spectrom. 1993 8 999. Laborda F. Viiiuales J. Mir J. M. and Castillo J. R. J. Anal. .4t. Spectrom. 1993 8 737. Volynsky A. B. and Krivan V. J . Anal. At. Spectrom. 1996 11 159. Bendicho C. and de Loos-Vollebregt M. T. C. J. Anal. At. Spectrom. 1991 6 353. Ebdon L. and Parry H. G. M. J. Anal. At. Spectrom. 1988,3 131.Bradshaw D. and Slavin W. Spectrochim. Acta Part B 1989 44 1245. Wagley D. Schmiedel G. Mainka E. and Ache H. J. At. Spectrosc. 1989 10 106. Bendicho C. and Sancho A. At. Spectrosc. 1993 14 187. Lopez-Garcia I. Viiias P. Campillo N. and Hernandez- Cordoba M. J. Agric. Food Chem. 1996 44 836. Neal R. H. in Heavy Metals in Soils ed. Alloway B. J. Chapman and Hall London 2nd edn. 1995 ch. 12. pp. 260-283. Hinds M. W. Latimer K. E. and Jackson K. W. J. Anal. At. Spectrom. 1991 6 473. Bermejo-Barrera P. Barciela-Alonso C. Aboal-Somoza M. and Bermejo-Barrera A. J. Anal. At. Spectrom. 1994 9 469. Lopez-Garcia I. Sanchez-Merlos M. and Hernandez- Cordoba M. Anal. Chim. Acta 1996 328 19. Halls D. J. Analyst 1984 109 1081. Hoenig M. and Cilissen A. Spectrochim. Acta Part B 1993 48 1003. Halls D. J. J. Anal. At. Spectrorn. 1995 10 169. Eames J. C. and Matousek J. P. Anal. Chem. 1980 52 1248. Miiller-Vogt G. and Wendl W. Anal. Chem. 1981 53 651. Bendicho C. and de Loos-Vollebregt M. T. C. Spectrochim. Acta Part B 1990 45 679. Bendicho C. and de Loos-Vollebregt M. T. C. Spectrochim. Acta Part B 1990 45 695. Lopez Garcia I. Arroyo Cortez. J. and Hernandez-Cordoba M. J. Anal. At. Spectrom. 1993 8 103. Lopez-Garcia I. Arroyo Cortez J. and Hernandez-Cordoba M.. Anal. Chim. Acta 1993 283 167. Miller J. C. and Miller J. N. Statistics for Analytical Chemistry Ellis Horwood Chichester 3rd edn. 1993 p. 11 5. Puper 6/03660K Received Muy 28 1996 Accepted July 10 1996 1006 Journal of Analytical Atomic Spectrometry October 1996 Vol. 1 1