Determination of Selenium in Water and Industrial Effluents by Flameless Atomic Absorption Earl L. Henn Calgon Corporation, Pittsburgh, Pa. 15230
A procedure for determining selenium in water and industrial effluents is described. The method utilizes flameless atomic absorption preceded by treatment with a cation exchange resin to eliminate interference from metallic cations and the addition of molybdenum to enhance sensitivity and suppress interference from inorganic anions. Advantages and limitations of the method are discussed, and analytical results on real samples are compared with those obtained using the colorimetric diaminobenzidlne method. The range of the test is 1.0 to 50 pg/l. Se.
Selenium is a cumulative toxic substance that can be a serious health hazard when present in food or water supplies. Because of this, the findings of even minute amounts of this element in water is of serious concern to environmentalists. Present regulations set forth by the Environmental Protection Agency (EPA) require that the selenium concentration in drinking water never exceed 0.01 mg/l. Se ( I ). The selenium concentration permitted in industrial effluents being discharged into rivers and streams has not as yet been defined but it is expected that the value will be close to the drinking water standard. These regulations mandate the need for a sensitive and reliable method of analyzing for selenium. In the case of industrial effluents which vary greatly in composition and often contain high concentrations of dissolved solids, the method used must also be highly specific and essentially free from interferences. The test most commonly used for selenium analysis is a colorimetric method whereby 3,3’-diaminobenzidine is reacted with selenate ion to form an intensely colored piazselenol which can be concentrated by extraction into toluene ( 2 ) . The major difficulty with this test is that diaminobenzidine is a carcinogen and stringent regulations must be adhered to in order to ensure its safe use. In addition, the method is very time consuming and requires a large volume of sample. Experience in our laboratory has shown very poor recovery of added selenium on samples containing high concentrations of dissolved organics when this method is used. Selenium can also be determined by X-ray fluorescence ( 3 ) and fluorometric techniques (4, 5 ) . However, these methods are mainly applicable to “clean” water and are generally lengthy and tedious. Selenium can be analyzed by conventional flame atomic absorption; however, the method suffers from a number of drawbacks. First of all, the primary resonance line of selenium is below 200 nm, resulting in a large portion of the energy from the source being absorbed by the flame. This results in a very poor signal-to-noise ratio for the test. In addition, the sensitivity of the flame method is poor, with a concentration of about 2.0 mg/l. Se being required to give 1%absorption. The combination of these two factors results in a very poor detection limit for selenium (about 0.5 mg/l. Se) using the flame technique (6). Improvement in the detection limit for selenium by use 428
of an indirect atomic absorption method has been reported ( 7 ) . Unfortunately, the claimed detection limit (0.018 mg/l. Se) is still not low enough to be used for water quality analysis. Some success has been reported by using a hydride generation technique whereby selenium is converted to the corresponding hydride which is then swept into ‘an argonhydrogen flame where the selenium is measured by atomic absorption (8). However, the method necessitates alteration of the nebulizer system and a changeover to a fueloxidant combination which is not generally used in atomic absorption measurements. In addition, the method suffers from poor precision and is mainly applicable to potable waters which do not contain large amounts of impurities (9). In searching for a method which would be sufficiently rapid and sensitive, flameless atomic absorption appeared to be the most promising. When this technique is used, the atomic vapor created during the atomization stage has a much longer residence time in the light beam from the source as compared to the conventional flame method, resulting in greatly increased sensitivity. In the flameless method, the problem of absorption of source light by a flame is eliminated. In addition, the signal-to-noise ratio for selenium analysis can be further improved by the use of a selenium electrodeless discharge lamp (EDL) which has recently become commercially available. The EDL provides a considerably more intense light source than the corresponding hollow cathode lamp, enabling the instrument to be run at a lower gain setting. All of these factors combine to provide a much better detection limit by the flameless method as compared to the flame technique. The application of flameless atomic absorption to the analysis of selenium in domestic sewage has been reported (IO). However, no provision is made for eliminating the extremely severe matrix interferences caused by inorganic ions in the flameless selenium determination. The method also requires a tenfold concentration of the sample in order to measure selenium concentrations in the low ppb range. This would tend to aggravate the problem of matrix interferences. One of the difficulties encountered in the flameless atomic absorption analysis of selenium is that high concentrations of inorganic salts cause severe scattering effects in the ,180-220 nm wavelength region. The use of ion exchange in eliminating this problem in arsenic analysis has been reported ( 1 1 ) .The present work shows that the use of ion exchange can also be used in eliminating background scatter in selenium analysis. The method described herein provides sufficient sensitivity to measure selenium concentrations as low as 1 wg/l. Se without concentration of the sample and is sufficiently free from interferences so as to permit accurate analysis of samples of widely varying composition containing high concentrations of dissolved inorganic ions. The method utilizes flameless atomic absorption preceded by treatment of the sample with a cation exchange resin to eliminate interference from metallic cations. Molybdenum is added to samples and standards in the concentration of 100 mgA. Mo in
ANALYTICAL CHEMISTRY, VOL. 47, NO 3, MARCH 1975
order to 1) Enhance the sensitivity of the test, 2) permit the use of a high charring temperature to selectively volatilize inorganic acids, and 3) minimize interference from high concentrations of inorganic anions. The range of the test is from 1to 50 kg/l. Se. EXPERIMENTAL Apparatus. A Perkin-Elmer Model 305B atomic absorption spectrophotometer equipped with an HGA-2000 heated graphite atomizer, deuterium background corrector, and strip chart recorder was used for all measurements. Eppendorf 100-microliter pipets were used for injecting samples into the graphite furnace. Instrument Settings. The 196.0-nm resonance line with a spectral bandpass of 2.0 nm was used for all measurements. The line source was a selenium e1ec;rodeless discharge lamp (EDL) operated a t a power setting of 6 watts. Background correction was used for all measurements. Controls on the Model HGA-2000 were experimentally optimized and set to provide the following drying, charring, and atomization times and temperatures: dry, 125 "C for 60 sec; char, 1000 "C for 35 sec; atomize, 2500 "C for 15 sec. Nitrogen was used as purge gas. The purge gas flow was interrupted during the atomization stage to increase the residence time for metal atoms in the graphite tube and provide improved sensitivity. A 1 0 0 - ~ sample 1 size was used throughout. Reagents. The 1000 mg/l. Se solution used for preparing selenium standards and the 1000 mg/l. Mo solution used in fortifying samples and standards with molybdenum were obtained from Fisher Scientific Co., Pittsburgh, Pa. All acids used in the test were reagent grade and used as received from J. T. Baker Chemical Co. The cation exchange resin used was Rohm and Haas IR-124 which was converted from the sodium form to the hydrogen form by treatment with hydrochloric acid. Sample Preparation. All samples were digested using the EPA digestion procedure for total metals analysis (12) after the addition of hydrogen peroxide to oxidize any organic selenium compounds to the selenite anion. After taking 50 ml of sample to dryness as required in the nitric acid digestion, the residue was taken up in 10 ml of 1:3 HCl. This solution was then passed through a 25-cm cation exchange column prepared from Rohm and Haas IR124 cation exchange resin in the hydrogen form. The column was rinsed with deionized water, and the rinsings were collected to restore the sample to its original 50-ml volume. The final solution matrix of this digestion procedure is 5% (v/v) HC1. Standard solutions were also passed through the cation exchange column and had the same concentration of molybdenum added as the samples. Sample Analysis. The graphite tube had to be conditioned with molybdenum by injecting 1 0 0 - ~ aliquots 1 of the 1000 mg/l. Mo solution into the furnace and running through the programmed drying, charring, and atomization cycles a few times before reproducible results could be obtained. The calibration curve was approximately linear up to 50 wg/l. Se using a 1OO-pl sample size and 3X recorder scale expansion. I
RESULTS AND DISCUSSION Interference Effects. In initial attempts at applying flameless atomic absorption to selenium analysis, tests were run to determine the degree of interference caused by various metallic ions on the method. Solutions of 50 kg/l. Se in 5% HC1 were each fortified with 10 mg/l. of a different metal to determine the effect of the various metal ions on selenium sensitivity. Results are listed in Table I. Note that extremely severe enhancement and depressant effects are caused by most metals, with only five of the 31 metals tested showing less than 10% interference. The enhancement effects were not caused by molecular absorption since background correction was used for all measurements. However, it was found that when the concentration of most metals was increased to 100 mg/l. or more, severe background absorption which exceeded the compensating capabilities of the deuterium background corrector was encountered. This was apparently caused by minute spattering which resulted from sudden volatilization of various inor-
Table I. Interference of Various Metals in the Flameless Atomic Absorption Analysis of Seleniuma Interfering substance added, 10 m g / l .
A1 As B
Ba
Be Bi
Ca Cd co Cr cu Fe Ge Hg K Li
Mg Mn Mo Na Ni Sb Sn Sr
Te Ti
T1 V W Zn Zr
Enhancing or depressing e f f e c t
on selenium absorption, %
+15 0 -19 -4 -46 +27
4 -1 2 +19 +196 +lo8
+I92 +69
+I77 +3 5 -9 +30 s13 +243 +3 5 +86 +96
-61 +26
+7a +74 -4 +278
+70 +39 +217
Se concentration, 0.050 mg/l. Se in 5% HC1.
ganic salts during the atomization cycle. This effect is not generally observed at higher wavelengths but is very pronounced a t the extremely low wavelength at tvhich selenium is measured. The most promising method of eliminating the interferences of metallic cations and inorganic salts appeared to be removal by ion exchange. The amphoteric nature of selenium causes it to revert to the anionic selenate or selenite form under the strong oxidizing conditions of the digestion, thereby permitting removal of cations without loss of selenium. A column containing 35 ml of Rohm and Haas IR124 cation exchange resin in the hydrogen form was prepared and found to be satisfactory for this purpose. In all cases studied, the replacement of metallic cations with hydrogen ions has reduced background absorption to a manageable level. Comparison of absorption signals obtained from standards treated with ion exchange and those untreated show that no selenium is lost on the cation exchange column. Good background correction is still required in the test for two reasons: 1) some background absorption is caused by volatilization of high concentrations of acids during the atomization cycle and 2) a spurious broad-band absorption signal from the graphite tube at the selenium wavelength must be corrected for in order to achieve a detection limit of 1 pgh. Se. In addition to the interference caused by metallic cations, the high acid concentration of the final sample matrix (5% HC1) and high concentrations of various anions were found to severely depress selenium sensitivity. A search ANALYTICAL CHEMISTRY, VOL. 47, NO. 3, MARCH 1975
429
50
9 30
L
' 0
I
I
200
400
1 1000
I
600
800
1
I
1200
1400
1
10
CHARRING TEMPERATURE, 'C
Figure 2. Loss of selenium at various charring temperatures with and without added molybdenum (A) No molybdenum added. ( 6 )100 mg/l. Mo added lot
0
1
20
I 40
I
I
60
80
I 100
I I20
I 140
160
MOLYBDENUM ADDED,mq/l
Figure 1. Effect of varying concentrations of molybdenum on absorption of 25 pgA Se in 5 % (v/v) HCI
was begun for a reagent which could be added to samples and standards to overcome the depressant effect of the high acid concentration of the sample matrix and the high anion concentrations anticipated in some samples. Numerous metals are known to react with selenium to form relatively refractory selenides. The use of metal additives was investigated to determine their effectiveness in eliminating interference effects from high concentrations of anions. The only metal which showed promise in this regard was molybdenum. Molybdenum has a very significant enhancing effect on selenium sensitivity in a 5% HC1 matrix; however, this effect is concentration dependent. This is illustrated in Figure 1 where the concentration of added molybdenum is plotted us. the peak height obtained with a 25 pg/l. Se standard at the various molybdenum concentrations. Note that maximum enhancement is achieved at a concentration of 30 mg/l. Mo and that above 50 mg/l. Mo, the enhancing effect drops off rather sharply. Another phenomenon associated with molybdenum addition to samples is that a much higher charring temperature can be tolerated without loss of selenium when molybdenum is present than when it is not present. In Figure 2, the peak height obtained for selenium standards is plotted us. charring temperature for standards containing: 1) No molybdenum and 2) 100 mg/l. Mo. Note that, when no molybdenum is present, loss of selenium during the charring stage begins occurring a t about 900 "C, and the apparent sensitivity drops off gradually from that point. When molybdenum is present, however, loss of selenium does not begin to occur until a t least 1400 "C and, above this temperature, the apparent sensitivity drops off very sharply. In addition to the possibility of selenide formation, another explanation for this behavior may be that selenium and molybdenum react to form a heteropolymolybdate anion in which a central selenium atom is surrounded by molybdate ions. In such a chemically bound state, the selenium would probably not begin to volatilize until a temperature was reached where the much more refractory molybdate ion began decomposing. At such a temperature, the selenium would be released and rapidly volatilized. The effect of added molybdenum on minimizing interferences from sulfate, nitrate, phosphate, and chromate is illustrated graphically in Figure 3. In all cases, the anions were added in the acid form. In these figures, the per cent change observed in selenium sensitivity both with and without added molybdenum in the presence of different concentrations of anions is plotted us. increasing concen430
Table 11. Comparison of Selenium Values Obtained by Atomic Absorption with Those Obtained by the Colorimetric Diaminobenzidine Method Selenium found, u g t l . Lab. Y o .
Atomic absorption
Colorimetric
18 8
G3564" G3565b G3567" G3 8 03 G3858 G3862 G3940 G4007 G5374 G5376
25 10 10 4 4 1 2 11 7 90
9 5 2 2 1 9 5 105
10 pg/l. Se 0 20 pg/l. Se standard submitted as control. standard submitted as control. 9.5 pg/l. Se standard submitted as control.
Table 111. Recovery of Known Amounts of Selenium Added to Samples
ANALYTICAL CHEMISTRY, VOL. 47, NO. 3, MARCH 1975
R L C O V Kof ~Y S ~ . l e n l u mrecovered,
G3567 G3696 G3801 G3802 G3803 G3851 G3858 G3859 G3860 G3 8 62 G3881 G3940 G3979 G4007 G4031 G4033 G4034 G4052
added selenium,
o
Idded
Added
Addid
hLfore
alter
before
alter
uil 1
digestion
digestion
digestion
digestion
25 25 25 25 25 25 25 25 25 25 25 25 25 25 25 25 25 25
21.5 20.0 24.5 21.5 23.5 15.5 18.5 17.5 20.5 17.5 18.0 22.0 12.5 17.0 24.5 15.5 24.5 13.5
25.0
Sr Addid Lab. \ o .
UCJ 1
...
... 26.5 25.5 19.5 19.5
...
22.5 20.0
... ... 22.5 20.0
... 25.0
...
22.5
86 80 98 86 94 62 74 70 81 70 72 88 50 68 98 63 98 54
Added
100
...
... 106 102 78 78
... 90 80
... ... 90 80
... 100
... 90
+20 r
- 100 L
t.0
1
1 10
I
1
IW
l.000
I 10,WO
SO,,mqll
W
3 Y
-10
z
5
-30
u Y
w -50
I 0
I
I
100
1,000
I
10,OOC
NO,.mqll
0
100
1.0
CrO*,mp/l
Figure 3. Effect of varying concentrations of common anions on selenium sensitivity in a 5 % (v/v) HCI matrix with and without added molybdenum ,(A) 100 mg/l. Mo added, 1000 O C char. (E)No Mo added, 700
O C
char
tration of the concomitant. Note that some interference from high concentrations of these ions is still encountered even when molybdenum is added; however, the addition of 100 mg/l. Mo considerably minimizes the error involved. Chloride ion does not interfere a t all, even a t the extremely high concentration of 10,000 mg/l. C1. Comparative Studies. Numerous samples, most of which were ash slurries from an industrial boiler and high in dissolved solids content, were analyzed using the atomic absorption procedure and the results compared with those obtained using the colorimetric diaminobenzidine method. Results are listed in Table 11. With the exception of the specially prepared standards submitted as controls, most of the selenium concentrations were low. However, the data do demonstrate good agreement between the two methods. Recovery Studies. Recovery studies were run by fortifying numerous samples with a known amount of selenium and taking the fortified sample through the sample preparation and atomic absorption procedures. The per cent recovery of added selenium was then calculated. Results are listed in Table 111. Also listed for some samples are the recovery values obtained when the selenium was added after digestion of the sample. Note that the per cent recovery values for the vast majority of samples are reasonably good, with all but three in the range of 68 to 98% recovery. In order to determine whether any selenium was being lost in the digestion step, a known amount of selenium was added to some of the samples after digestion. Note that, in allcases, the per cent recovery of selenium added after diges-
30
I
500
600
I 7M)
I
I
I
I
1
800
900
1000
1100
I200
CHARRING TEMPERATURE.
1300
C'
Figure 4. Effect of charring temperature on recovery of added selenium from an ash slurry sample at two different levels of molybdenum concentration (A) 100 mg/l. Mo added. (B) 50 mgll. Mo added
tion was higher than that obtained when the selenium was added before digestion. This indicates that some selenium is being lost in the digestion step, and this is the major cause of low recovery values in the atomic absorption procedure. In initial attempts a t applying the flameless atomic absorption method to real samples, a 700 "C charring temperature and 50 mg/l. Mo concentration were used. Under these conditions, recovery of added selenium was very poor.
ANALYTICAL CHEMISTRY, VOL. 47, NO. 3, MARCH 1975
431
Subsquent experiments showed that the charring temperature as well as the molybdenum concentration had a significant bearing on the per cent recovery of added selenium obtained. A sample (Lab No. G3860) for which recovery had been low was analyzed a t various charring temperatures using two different concentrations of molybdenum. In Figure 4, t)e per cent recovery of added selenium obtained under these varying conditions is plotted us. the charring temperature. Note that, in a11 cases, a significant improvement in per cent recovery is obtained when a concentration of 100 mg/l. Mo is used as compared to the recovery obtained when 50 mg/l. Mo is used. Also, the per cent recovery obtained is definitely a function of charring temperature, with the recovery increasing as the charring temperature increases until a temperature of 1000 “C is reached. Above this temperature, the recovery drops off slightly. The main reason why per cent recovery is a function of charring temperature is probably because the higher charring temperature causes volatilization of interfering anions which results in increased recovery values. However, this would not explain why the recovery drops off a t temperatures above 1000 “C. Another possible explanation may be that the increased molybdenum concentration, coupled with an increased charring temperature, results in the formation of a higher molecular weight heteropolymolybdate anion which is more successful a t isolating the selenium from interfering substances. Precision Studies. The precision of the atomic absorption measurement at the 0.010 mg/l. Se level was determined. The standard deviation of ten replicate analyses of 10 wg/l. Se standard in 5% HC1 was found to be f0.75 gg/l.
Se, giving a relative standard deviation of 7.5%. This precision test did not include the digestion procedure. The precision of the entire procedure, including the digestion step, would probably be poorer.
CONCLUSIONS A rapid and highly specific method for determining selenium in water and industrial effluents has been developed. Unfortunately, the digestion method required by the Environmental Protection Agency for “total” metals analysis results in loss of selenium on some samples. Improvements in the digestion procedure will be necessary before the full value of the method can be utilized in monitoring environmental waters for selenium content. LITERATURE CITED (1) Bureau of National Affairs, Washington, D.C., Environ. Rep., 4, 666 (1973). (2) American Public Health Association, “Standard Methods for the Exam!nation of Water and Wastewater,” 13th ed.,1971, p 295 (3) F. J. Marcie. Environ. Sci. Techno/., 1, 164(1967). (4) J. H. Wiersma and F. G. Lee, Environ. Sci. Techno/.,5, 1203 (1971). (5) J. M. Rankin, Environ. Sci. Techno/.,7 , 823 (1973). (6) C. S Rann and A. N. Hambly, Anal. Chin?,Acta, 32,346 (1965). (7) H. K. Y. Lau.and P. F. Lott, Talanfa, 18, 303 (1971). (8) D. C. Manning, At. Absorption Newslett., 10, 123 (1971). (9) J. S. Caldwell, R. J. Lishka, and E. F. McFarren, J. Amer. Wafer Works Ass., 85, 731 (1973). (10) R. B. Baird, S. Pourian, and S. M. Gabrielian. Anal. Chem., 44, 1887 (19721. (11) k. 6. Baird, S. Pourian and S. M. Gabrielian, Preprints, 166th National Meeting of the American Chemical Society, Aug. 1973, 13, Paper 15. (12) Environmental Protection Agency, Cincinnati, Ohio, “Methods for Chernical Analysis of Water and Wastes,” 1971, p 88.
RECEIVEDfor review July 31, 1974. Accepted November 18, 1974.
Determination of Lithium in Microliter Samples of Blood Serum Using Flame Atomic Emission Spectrometry with a Tantalum Filament Vaporizer J.
K. Grime and T. J. Vickers
Department of Chemistry, Florida State University, Tallahassee, Fla. 32306
A method is described for the determination of lithium in microliter samples. A conventional flame atomic emission system has been adapted such that the atomization and vaporization processes are Isolated; the former occurrlng in either an air/hydrogen or air/acetylene flame, the latter from a heated tantalum filament. The disadvantages of using electrothermal atomization alone have thus been minimized. The technique has been employed in the determination of lithium in three matrices, viz,, water, artificial serum, and reconstituted human serum. The limits of detection (signalhoise = 2) in aqueous samples, for the air/H2 and air/ C2H2 flames are 0.009 nanogram (0.0018 ppm/5pI) and 0.003 nanogram (0.0006 ppm/5@) respectively. Lithium is determined at normal levels in human sera (0.01 ppm), using only a 5-microliter sample. It is recommended that serum samples be determined by either calibration with artificial serum standards or by the method of standard additions. 432
ANALYTICAL CHEMISTRY, VOC. 47,
NO. 3, MARCH
It was determined in 1949 that lithium salts could be used effectively in the treatment of manic-depressive psychosis ( I ) . The normal level of lithium in blood serum is approximately 0.01 gg ml-I (2);during treatment, however, the level is usually maintained at 7 wg ml-l (3). Toxic symptoms have been reported for concentrations as low as 11.2 wg ml-l (4),and, accordingly, the serum lithium level must be carefully monitored in order that this level is not exceeded. The need for an accurate and reproducible method of analysis requiring a minimal amount of sample is therefore apparent. Conventional flame atomic absorption and atomic emission spectrometry have been successfully used to determine abnormal levels of lithium in blood serum (3-6). It has been suggested that there is no significant difference in sensitivity between the two techniques (4, 6, 7 ) , although atomic emission is generally accepted as the more convenient. The normal level of lithium in sera is, however, below the
1975
A procedure for determining selenium in water and industrial effluents is described. The method utilizes flameless atomic absorption preceded by treatment with a cation exchange resin to eliminate interference from metallic cations and the addition of molybdenum to enhance sensitivity and suppress interference from inorganic anions. Advantages and limitations of the method are discussed, and analytical results on real samples are compared with those obtained using the colorimetric diaminobenzidlne method. The range of the test is 1.0 to 50 pg/l. Se.
Selenium is a cumulative toxic substance that can be a serious health hazard when present in food or water supplies. Because of this, the findings of even minute amounts of this element in water is of serious concern to environmentalists. Present regulations set forth by the Environmental Protection Agency (EPA) require that the selenium concentration in drinking water never exceed 0.01 mg/l. Se ( I ). The selenium concentration permitted in industrial effluents being discharged into rivers and streams has not as yet been defined but it is expected that the value will be close to the drinking water standard. These regulations mandate the need for a sensitive and reliable method of analyzing for selenium. In the case of industrial effluents which vary greatly in composition and often contain high concentrations of dissolved solids, the method used must also be highly specific and essentially free from interferences. The test most commonly used for selenium analysis is a colorimetric method whereby 3,3’-diaminobenzidine is reacted with selenate ion to form an intensely colored piazselenol which can be concentrated by extraction into toluene ( 2 ) . The major difficulty with this test is that diaminobenzidine is a carcinogen and stringent regulations must be adhered to in order to ensure its safe use. In addition, the method is very time consuming and requires a large volume of sample. Experience in our laboratory has shown very poor recovery of added selenium on samples containing high concentrations of dissolved organics when this method is used. Selenium can also be determined by X-ray fluorescence ( 3 ) and fluorometric techniques (4, 5 ) . However, these methods are mainly applicable to “clean” water and are generally lengthy and tedious. Selenium can be analyzed by conventional flame atomic absorption; however, the method suffers from a number of drawbacks. First of all, the primary resonance line of selenium is below 200 nm, resulting in a large portion of the energy from the source being absorbed by the flame. This results in a very poor signal-to-noise ratio for the test. In addition, the sensitivity of the flame method is poor, with a concentration of about 2.0 mg/l. Se being required to give 1%absorption. The combination of these two factors results in a very poor detection limit for selenium (about 0.5 mg/l. Se) using the flame technique (6). Improvement in the detection limit for selenium by use 428
of an indirect atomic absorption method has been reported ( 7 ) . Unfortunately, the claimed detection limit (0.018 mg/l. Se) is still not low enough to be used for water quality analysis. Some success has been reported by using a hydride generation technique whereby selenium is converted to the corresponding hydride which is then swept into ‘an argonhydrogen flame where the selenium is measured by atomic absorption (8). However, the method necessitates alteration of the nebulizer system and a changeover to a fueloxidant combination which is not generally used in atomic absorption measurements. In addition, the method suffers from poor precision and is mainly applicable to potable waters which do not contain large amounts of impurities (9). In searching for a method which would be sufficiently rapid and sensitive, flameless atomic absorption appeared to be the most promising. When this technique is used, the atomic vapor created during the atomization stage has a much longer residence time in the light beam from the source as compared to the conventional flame method, resulting in greatly increased sensitivity. In the flameless method, the problem of absorption of source light by a flame is eliminated. In addition, the signal-to-noise ratio for selenium analysis can be further improved by the use of a selenium electrodeless discharge lamp (EDL) which has recently become commercially available. The EDL provides a considerably more intense light source than the corresponding hollow cathode lamp, enabling the instrument to be run at a lower gain setting. All of these factors combine to provide a much better detection limit by the flameless method as compared to the flame technique. The application of flameless atomic absorption to the analysis of selenium in domestic sewage has been reported (IO). However, no provision is made for eliminating the extremely severe matrix interferences caused by inorganic ions in the flameless selenium determination. The method also requires a tenfold concentration of the sample in order to measure selenium concentrations in the low ppb range. This would tend to aggravate the problem of matrix interferences. One of the difficulties encountered in the flameless atomic absorption analysis of selenium is that high concentrations of inorganic salts cause severe scattering effects in the ,180-220 nm wavelength region. The use of ion exchange in eliminating this problem in arsenic analysis has been reported ( 1 1 ) .The present work shows that the use of ion exchange can also be used in eliminating background scatter in selenium analysis. The method described herein provides sufficient sensitivity to measure selenium concentrations as low as 1 wg/l. Se without concentration of the sample and is sufficiently free from interferences so as to permit accurate analysis of samples of widely varying composition containing high concentrations of dissolved inorganic ions. The method utilizes flameless atomic absorption preceded by treatment of the sample with a cation exchange resin to eliminate interference from metallic cations. Molybdenum is added to samples and standards in the concentration of 100 mgA. Mo in
ANALYTICAL CHEMISTRY, VOL. 47, NO 3, MARCH 1975
order to 1) Enhance the sensitivity of the test, 2) permit the use of a high charring temperature to selectively volatilize inorganic acids, and 3) minimize interference from high concentrations of inorganic anions. The range of the test is from 1to 50 kg/l. Se. EXPERIMENTAL Apparatus. A Perkin-Elmer Model 305B atomic absorption spectrophotometer equipped with an HGA-2000 heated graphite atomizer, deuterium background corrector, and strip chart recorder was used for all measurements. Eppendorf 100-microliter pipets were used for injecting samples into the graphite furnace. Instrument Settings. The 196.0-nm resonance line with a spectral bandpass of 2.0 nm was used for all measurements. The line source was a selenium e1ec;rodeless discharge lamp (EDL) operated a t a power setting of 6 watts. Background correction was used for all measurements. Controls on the Model HGA-2000 were experimentally optimized and set to provide the following drying, charring, and atomization times and temperatures: dry, 125 "C for 60 sec; char, 1000 "C for 35 sec; atomize, 2500 "C for 15 sec. Nitrogen was used as purge gas. The purge gas flow was interrupted during the atomization stage to increase the residence time for metal atoms in the graphite tube and provide improved sensitivity. A 1 0 0 - ~ sample 1 size was used throughout. Reagents. The 1000 mg/l. Se solution used for preparing selenium standards and the 1000 mg/l. Mo solution used in fortifying samples and standards with molybdenum were obtained from Fisher Scientific Co., Pittsburgh, Pa. All acids used in the test were reagent grade and used as received from J. T. Baker Chemical Co. The cation exchange resin used was Rohm and Haas IR-124 which was converted from the sodium form to the hydrogen form by treatment with hydrochloric acid. Sample Preparation. All samples were digested using the EPA digestion procedure for total metals analysis (12) after the addition of hydrogen peroxide to oxidize any organic selenium compounds to the selenite anion. After taking 50 ml of sample to dryness as required in the nitric acid digestion, the residue was taken up in 10 ml of 1:3 HCl. This solution was then passed through a 25-cm cation exchange column prepared from Rohm and Haas IR124 cation exchange resin in the hydrogen form. The column was rinsed with deionized water, and the rinsings were collected to restore the sample to its original 50-ml volume. The final solution matrix of this digestion procedure is 5% (v/v) HC1. Standard solutions were also passed through the cation exchange column and had the same concentration of molybdenum added as the samples. Sample Analysis. The graphite tube had to be conditioned with molybdenum by injecting 1 0 0 - ~ aliquots 1 of the 1000 mg/l. Mo solution into the furnace and running through the programmed drying, charring, and atomization cycles a few times before reproducible results could be obtained. The calibration curve was approximately linear up to 50 wg/l. Se using a 1OO-pl sample size and 3X recorder scale expansion. I
RESULTS AND DISCUSSION Interference Effects. In initial attempts at applying flameless atomic absorption to selenium analysis, tests were run to determine the degree of interference caused by various metallic ions on the method. Solutions of 50 kg/l. Se in 5% HC1 were each fortified with 10 mg/l. of a different metal to determine the effect of the various metal ions on selenium sensitivity. Results are listed in Table I. Note that extremely severe enhancement and depressant effects are caused by most metals, with only five of the 31 metals tested showing less than 10% interference. The enhancement effects were not caused by molecular absorption since background correction was used for all measurements. However, it was found that when the concentration of most metals was increased to 100 mg/l. or more, severe background absorption which exceeded the compensating capabilities of the deuterium background corrector was encountered. This was apparently caused by minute spattering which resulted from sudden volatilization of various inor-
Table I. Interference of Various Metals in the Flameless Atomic Absorption Analysis of Seleniuma Interfering substance added, 10 m g / l .
A1 As B
Ba
Be Bi
Ca Cd co Cr cu Fe Ge Hg K Li
Mg Mn Mo Na Ni Sb Sn Sr
Te Ti
T1 V W Zn Zr
Enhancing or depressing e f f e c t
on selenium absorption, %
+15 0 -19 -4 -46 +27
4 -1 2 +19 +196 +lo8
+I92 +69
+I77 +3 5 -9 +30 s13 +243 +3 5 +86 +96
-61 +26
+7a +74 -4 +278
+70 +39 +217
Se concentration, 0.050 mg/l. Se in 5% HC1.
ganic salts during the atomization cycle. This effect is not generally observed at higher wavelengths but is very pronounced a t the extremely low wavelength at tvhich selenium is measured. The most promising method of eliminating the interferences of metallic cations and inorganic salts appeared to be removal by ion exchange. The amphoteric nature of selenium causes it to revert to the anionic selenate or selenite form under the strong oxidizing conditions of the digestion, thereby permitting removal of cations without loss of selenium. A column containing 35 ml of Rohm and Haas IR124 cation exchange resin in the hydrogen form was prepared and found to be satisfactory for this purpose. In all cases studied, the replacement of metallic cations with hydrogen ions has reduced background absorption to a manageable level. Comparison of absorption signals obtained from standards treated with ion exchange and those untreated show that no selenium is lost on the cation exchange column. Good background correction is still required in the test for two reasons: 1) some background absorption is caused by volatilization of high concentrations of acids during the atomization cycle and 2) a spurious broad-band absorption signal from the graphite tube at the selenium wavelength must be corrected for in order to achieve a detection limit of 1 pgh. Se. In addition to the interference caused by metallic cations, the high acid concentration of the final sample matrix (5% HC1) and high concentrations of various anions were found to severely depress selenium sensitivity. A search ANALYTICAL CHEMISTRY, VOL. 47, NO. 3, MARCH 1975
429
50
9 30
L
' 0
I
I
200
400
1 1000
I
600
800
1
I
1200
1400
1
10
CHARRING TEMPERATURE, 'C
Figure 2. Loss of selenium at various charring temperatures with and without added molybdenum (A) No molybdenum added. ( 6 )100 mg/l. Mo added lot
0
1
20
I 40
I
I
60
80
I 100
I I20
I 140
160
MOLYBDENUM ADDED,mq/l
Figure 1. Effect of varying concentrations of molybdenum on absorption of 25 pgA Se in 5 % (v/v) HCI
was begun for a reagent which could be added to samples and standards to overcome the depressant effect of the high acid concentration of the sample matrix and the high anion concentrations anticipated in some samples. Numerous metals are known to react with selenium to form relatively refractory selenides. The use of metal additives was investigated to determine their effectiveness in eliminating interference effects from high concentrations of anions. The only metal which showed promise in this regard was molybdenum. Molybdenum has a very significant enhancing effect on selenium sensitivity in a 5% HC1 matrix; however, this effect is concentration dependent. This is illustrated in Figure 1 where the concentration of added molybdenum is plotted us. the peak height obtained with a 25 pg/l. Se standard at the various molybdenum concentrations. Note that maximum enhancement is achieved at a concentration of 30 mg/l. Mo and that above 50 mg/l. Mo, the enhancing effect drops off rather sharply. Another phenomenon associated with molybdenum addition to samples is that a much higher charring temperature can be tolerated without loss of selenium when molybdenum is present than when it is not present. In Figure 2, the peak height obtained for selenium standards is plotted us. charring temperature for standards containing: 1) No molybdenum and 2) 100 mg/l. Mo. Note that, when no molybdenum is present, loss of selenium during the charring stage begins occurring a t about 900 "C, and the apparent sensitivity drops off gradually from that point. When molybdenum is present, however, loss of selenium does not begin to occur until a t least 1400 "C and, above this temperature, the apparent sensitivity drops off very sharply. In addition to the possibility of selenide formation, another explanation for this behavior may be that selenium and molybdenum react to form a heteropolymolybdate anion in which a central selenium atom is surrounded by molybdate ions. In such a chemically bound state, the selenium would probably not begin to volatilize until a temperature was reached where the much more refractory molybdate ion began decomposing. At such a temperature, the selenium would be released and rapidly volatilized. The effect of added molybdenum on minimizing interferences from sulfate, nitrate, phosphate, and chromate is illustrated graphically in Figure 3. In all cases, the anions were added in the acid form. In these figures, the per cent change observed in selenium sensitivity both with and without added molybdenum in the presence of different concentrations of anions is plotted us. increasing concen430
Table 11. Comparison of Selenium Values Obtained by Atomic Absorption with Those Obtained by the Colorimetric Diaminobenzidine Method Selenium found, u g t l . Lab. Y o .
Atomic absorption
Colorimetric
18 8
G3564" G3565b G3567" G3 8 03 G3858 G3862 G3940 G4007 G5374 G5376
25 10 10 4 4 1 2 11 7 90
9 5 2 2 1 9 5 105
10 pg/l. Se 0 20 pg/l. Se standard submitted as control. standard submitted as control. 9.5 pg/l. Se standard submitted as control.
Table 111. Recovery of Known Amounts of Selenium Added to Samples
ANALYTICAL CHEMISTRY, VOL. 47, NO. 3, MARCH 1975
R L C O V Kof ~Y S ~ . l e n l u mrecovered,
G3567 G3696 G3801 G3802 G3803 G3851 G3858 G3859 G3860 G3 8 62 G3881 G3940 G3979 G4007 G4031 G4033 G4034 G4052
added selenium,
o
Idded
Added
Addid
hLfore
alter
before
alter
uil 1
digestion
digestion
digestion
digestion
25 25 25 25 25 25 25 25 25 25 25 25 25 25 25 25 25 25
21.5 20.0 24.5 21.5 23.5 15.5 18.5 17.5 20.5 17.5 18.0 22.0 12.5 17.0 24.5 15.5 24.5 13.5
25.0
Sr Addid Lab. \ o .
UCJ 1
...
... 26.5 25.5 19.5 19.5
...
22.5 20.0
... ... 22.5 20.0
... 25.0
...
22.5
86 80 98 86 94 62 74 70 81 70 72 88 50 68 98 63 98 54
Added
100
...
... 106 102 78 78
... 90 80
... ... 90 80
... 100
... 90
+20 r
- 100 L
t.0
1
1 10
I
1
IW
l.000
I 10,WO
SO,,mqll
W
3 Y
-10
z
5
-30
u Y
w -50
I 0
I
I
100
1,000
I
10,OOC
NO,.mqll
0
100
1.0
CrO*,mp/l
Figure 3. Effect of varying concentrations of common anions on selenium sensitivity in a 5 % (v/v) HCI matrix with and without added molybdenum ,(A) 100 mg/l. Mo added, 1000 O C char. (E)No Mo added, 700
O C
char
tration of the concomitant. Note that some interference from high concentrations of these ions is still encountered even when molybdenum is added; however, the addition of 100 mg/l. Mo considerably minimizes the error involved. Chloride ion does not interfere a t all, even a t the extremely high concentration of 10,000 mg/l. C1. Comparative Studies. Numerous samples, most of which were ash slurries from an industrial boiler and high in dissolved solids content, were analyzed using the atomic absorption procedure and the results compared with those obtained using the colorimetric diaminobenzidine method. Results are listed in Table 11. With the exception of the specially prepared standards submitted as controls, most of the selenium concentrations were low. However, the data do demonstrate good agreement between the two methods. Recovery Studies. Recovery studies were run by fortifying numerous samples with a known amount of selenium and taking the fortified sample through the sample preparation and atomic absorption procedures. The per cent recovery of added selenium was then calculated. Results are listed in Table 111. Also listed for some samples are the recovery values obtained when the selenium was added after digestion of the sample. Note that the per cent recovery values for the vast majority of samples are reasonably good, with all but three in the range of 68 to 98% recovery. In order to determine whether any selenium was being lost in the digestion step, a known amount of selenium was added to some of the samples after digestion. Note that, in allcases, the per cent recovery of selenium added after diges-
30
I
500
600
I 7M)
I
I
I
I
1
800
900
1000
1100
I200
CHARRING TEMPERATURE.
1300
C'
Figure 4. Effect of charring temperature on recovery of added selenium from an ash slurry sample at two different levels of molybdenum concentration (A) 100 mg/l. Mo added. (B) 50 mgll. Mo added
tion was higher than that obtained when the selenium was added before digestion. This indicates that some selenium is being lost in the digestion step, and this is the major cause of low recovery values in the atomic absorption procedure. In initial attempts a t applying the flameless atomic absorption method to real samples, a 700 "C charring temperature and 50 mg/l. Mo concentration were used. Under these conditions, recovery of added selenium was very poor.
ANALYTICAL CHEMISTRY, VOL. 47, NO. 3, MARCH 1975
431
Subsquent experiments showed that the charring temperature as well as the molybdenum concentration had a significant bearing on the per cent recovery of added selenium obtained. A sample (Lab No. G3860) for which recovery had been low was analyzed a t various charring temperatures using two different concentrations of molybdenum. In Figure 4, t)e per cent recovery of added selenium obtained under these varying conditions is plotted us. the charring temperature. Note that, in a11 cases, a significant improvement in per cent recovery is obtained when a concentration of 100 mg/l. Mo is used as compared to the recovery obtained when 50 mg/l. Mo is used. Also, the per cent recovery obtained is definitely a function of charring temperature, with the recovery increasing as the charring temperature increases until a temperature of 1000 “C is reached. Above this temperature, the recovery drops off slightly. The main reason why per cent recovery is a function of charring temperature is probably because the higher charring temperature causes volatilization of interfering anions which results in increased recovery values. However, this would not explain why the recovery drops off a t temperatures above 1000 “C. Another possible explanation may be that the increased molybdenum concentration, coupled with an increased charring temperature, results in the formation of a higher molecular weight heteropolymolybdate anion which is more successful a t isolating the selenium from interfering substances. Precision Studies. The precision of the atomic absorption measurement at the 0.010 mg/l. Se level was determined. The standard deviation of ten replicate analyses of 10 wg/l. Se standard in 5% HC1 was found to be f0.75 gg/l.
Se, giving a relative standard deviation of 7.5%. This precision test did not include the digestion procedure. The precision of the entire procedure, including the digestion step, would probably be poorer.
CONCLUSIONS A rapid and highly specific method for determining selenium in water and industrial effluents has been developed. Unfortunately, the digestion method required by the Environmental Protection Agency for “total” metals analysis results in loss of selenium on some samples. Improvements in the digestion procedure will be necessary before the full value of the method can be utilized in monitoring environmental waters for selenium content. LITERATURE CITED (1) Bureau of National Affairs, Washington, D.C., Environ. Rep., 4, 666 (1973). (2) American Public Health Association, “Standard Methods for the Exam!nation of Water and Wastewater,” 13th ed.,1971, p 295 (3) F. J. Marcie. Environ. Sci. Techno/., 1, 164(1967). (4) J. H. Wiersma and F. G. Lee, Environ. Sci. Techno/.,5, 1203 (1971). (5) J. M. Rankin, Environ. Sci. Techno/.,7 , 823 (1973). (6) C. S Rann and A. N. Hambly, Anal. Chin?,Acta, 32,346 (1965). (7) H. K. Y. Lau.and P. F. Lott, Talanfa, 18, 303 (1971). (8) D. C. Manning, At. Absorption Newslett., 10, 123 (1971). (9) J. S. Caldwell, R. J. Lishka, and E. F. McFarren, J. Amer. Wafer Works Ass., 85, 731 (1973). (10) R. B. Baird, S. Pourian, and S. M. Gabrielian. Anal. Chem., 44, 1887 (19721. (11) k. 6. Baird, S. Pourian and S. M. Gabrielian, Preprints, 166th National Meeting of the American Chemical Society, Aug. 1973, 13, Paper 15. (12) Environmental Protection Agency, Cincinnati, Ohio, “Methods for Chernical Analysis of Water and Wastes,” 1971, p 88.
RECEIVEDfor review July 31, 1974. Accepted November 18, 1974.
Determination of Lithium in Microliter Samples of Blood Serum Using Flame Atomic Emission Spectrometry with a Tantalum Filament Vaporizer J.
K. Grime and T. J. Vickers
Department of Chemistry, Florida State University, Tallahassee, Fla. 32306
A method is described for the determination of lithium in microliter samples. A conventional flame atomic emission system has been adapted such that the atomization and vaporization processes are Isolated; the former occurrlng in either an air/hydrogen or air/acetylene flame, the latter from a heated tantalum filament. The disadvantages of using electrothermal atomization alone have thus been minimized. The technique has been employed in the determination of lithium in three matrices, viz,, water, artificial serum, and reconstituted human serum. The limits of detection (signalhoise = 2) in aqueous samples, for the air/H2 and air/ C2H2 flames are 0.009 nanogram (0.0018 ppm/5pI) and 0.003 nanogram (0.0006 ppm/5@) respectively. Lithium is determined at normal levels in human sera (0.01 ppm), using only a 5-microliter sample. It is recommended that serum samples be determined by either calibration with artificial serum standards or by the method of standard additions. 432
ANALYTICAL CHEMISTRY, VOC. 47,
NO. 3, MARCH
It was determined in 1949 that lithium salts could be used effectively in the treatment of manic-depressive psychosis ( I ) . The normal level of lithium in blood serum is approximately 0.01 gg ml-I (2);during treatment, however, the level is usually maintained at 7 wg ml-l (3). Toxic symptoms have been reported for concentrations as low as 11.2 wg ml-l (4),and, accordingly, the serum lithium level must be carefully monitored in order that this level is not exceeded. The need for an accurate and reproducible method of analysis requiring a minimal amount of sample is therefore apparent. Conventional flame atomic absorption and atomic emission spectrometry have been successfully used to determine abnormal levels of lithium in blood serum (3-6). It has been suggested that there is no significant difference in sensitivity between the two techniques (4, 6, 7 ) , although atomic emission is generally accepted as the more convenient. The normal level of lithium in sera is, however, below the
1975
