926
A N A L Y T I C A L CHEMISTRY, VOL. 51, NO. 7, JUNE 1979
(17) Maeda, K.: Kashiwabara, T.: Tokuyama, M. Bull. Chem. SOCJpn. 1977, 5 0 , 473. (18) McCapra, F.: Hann, R . A. Chem. Commun. 1969, 442 (19) Legg. K D.: Hercules, D. M. J . A m . Chem. SOC. 1969, 9 7 , 1902. (20) McCapra, F. in "Progress in Organic Chemistrf', Carruthers, W.; Svtherhnd, J. K.. Ed., John Wiley & Sons: New York, 1973; Vol. 8, pp 231-77. (21) Bowen, E. J. Pure Appl. Chem. 1964, 9 . 473. (22) McCapra, F Quart. Rev. 1966, 485. (23) Gundermann, K. D. Top. Curr. Chem. 1974, 4 6 , 61. (24) Erdey. L. Ind. Chem. 1957, 33, 459, 523. (25) Dubovenko, L. I . ; Beloshitskii, N. V. J . Anal. Chem. USSR 1974; 2 9 . 85. (26) Babko. A. K.; Dubovenko, L. I . : Terletskaya, A. V. Sov. Prog. Chem. 1966, 3 2 , 1326. (27) Bognar, J.; Sipos, L. Mikrochm Ichnoanal. Acta 1963, 3 , 442. (28) Dubovenko, L. I . ; Tovmasyan, A. P. J , Anal. Chem. USSR 1970, 2 5 , 812. (29) Dubovenko, L. I . ; Tovmasyan, A. P. Sov. Prog. Chem. 1971, 3 7 , 83. (30) Babko, A. K.; Terletskaya, A. V.; Dubovenko. L. I . J . Anal. Chem. USSR, 1968, 2 3 . 809. (31) Babko, A. K.; Dubovenko, L. I . ; Terletskaya. A. V Sov. Prog. Chem. 1969, 3 5 , 64. (32) Brin, A. Ya; Kashllnskaya, S. E.; Strel'nikova, N. P.; Petrovichera, V. I. Anal. Teknol. Bbgorcd. Met., 1971, 86-90. From Ref. Zh. Khim.. 1972, Abstract No. 6G100: Chem. Abstr. 1973, 78 66575h. (33) Dubovenko, L. 1.: Tovmasyan, A. P. Ukr. Khim. Zh. 1971, 37, 845; C k m . Abstr. 1971, 75, 126070q. (34) Dubovenko. L I ; Guz, L. D. Sov. Prog. Chem. 1970, 3 6 , 61. (35) Dubovenko, L. I . ; Tovmasyan, A. P. Sov. Prog. Chem. 1973, 3 9 , 85. (36) Dubovenko, L. I . ; Tovmasyan, A. P. Sov. Prog. Chem. 1973, 3 9 , 68. (37) Dubovenko, L. I . ; Khotinets. E. Ya. J . Anal. Chem. USSR, 1971, 2 6 , 683. (38) Dubovenko, L. I.; Drokov, V. G. Izv. Vyssh. Ucheb. Zavcd., Khim. Khim. Teknol. 1973, 16, 1010: Chem. Abstr. 1973, 7 9 , 142551k. (39) Dubovenko. L. I.; Tananaiko, M. M.; Drokov, V. G. Ukr. Khim. Zh. 1974, 40. 758; Chem. Abstr. 1974, 8 1 , 98896s. (40) Dubovenko. L. I.: Guta, A. M. Izv. Vyssh. Ucheb. Zavod., Khim. Khim. Teknol. 1975, 18, 1211; Chem. Abstr. 1976, 8 4 , 38257m. (41) Dubovenko, L. I.; Nazavenko, A. Yu. Ukr. Khim. Zh. 1975, 4 1 , 1205; Chem. Abstr. 1976, 8 5 , 133061. (42) Terletskaya. A. V. Ukr. Khim. Zh. 1969, 3 5 , 1065. (43) Tovmasyan, A. P. Mater. Kanf. Molodykh. Uch. Spets., Akad. Nauk. Arm. SSR 1972, 188; Chem Abstr. 1975, 8 3 , 187824
(44) Erdey. L.; Buzas, I . Anal. Chim. Acta 1959, 2 2 , 524. (45) Lurie. Ju. "Handbook of Analytical Chemistry", Mir Publishers: Moscow, 1975. (46) Erdey. L.; Tackacs, J.: Buzas, I . Acta. Chim. Acad. Sci. Hung. 1964, 39, 295: Chem. Abstr. 1964, 6 0 , 9878h. (47) . . Sarudi. I. fresenius' Z . Anal. Chem. 1972, 260. 114: Chem Abstr. 1972, 77, 1347791. (48) Soli, G. U S . Patent 3564388, 1971; Chem. Abstr. 1971, 75, P1488444s. (49) Totter, J. R.; Medina. V . J.; Scoseria, J. L. J . Biol. Chem. 1960, 235, 238. (50) Greenlee, L.: Fridovlch, I.: Handler, P. Biochem. 1962, 7 , 779. 151) Oleniacz, W. S.: Pisano, M. A.: Insolera, R . V. Photochem. Photobiol. 1967, 6 . 613. (52) Dubovenko, L I.: Tanar.aiko, M. M.: Drokov, V . G. Izv. Vyssh. Uchen Zavod., Khim. Khlm. Teknol. 1975. 18. 1068; Chem. Abstr. 1975, 83, 174969b. (53) Veazey, R. L.; Nieman, T. A. "Chemiluminescence Determination of Reductants in Biological Fluids", 29th Pittsburg Conference on Analytical Chemistry and Applied Spectroscopy, Cleveland, Ohio, Feb. 27-Mar. 3, 1'378 _.
(54) Dubovenko, L. I . ; Korotun, L. M.: Kostyshina, A. P. Ukr. Khim. Zh. 1976, 42. 1194: Chem. Abstr.. 1977. 86. 6 1 2 0 8 ~ . (55) Hoyt, S. D.; Ingle, J. D . , J r . ~ A n a l .Chim. Acta. 1976, 8 7 , 163. (56) Parsons, M. L.; Smith, B. W.: Bentley, G. E. "Handbook of Flame Spectroscopy", Plenum Press; New York, 1975. (57) Montano, L. A.; Ingle, J. D., Jr. Anal. Chem., 1979, 5 1 , following paper in this issue. (58) Ryan, M. A.; Miller, R . J.; Ingle. J. D., Jr. Anal. Chem. 1978, 50, 1772.
RECEIVED for review October 20, 1978. Accepted February 12. 1979. Acknowledgement is made t o the N S F (Grant =CHE-76-16711)for partial support of this research and the University Corporation for Atmospheric Research for fellowship support of one of us (L.A.M). Presented in part a t the 1977 Northwest ACS meeting, Portland, Ore., and a t the 29th Pittsburgh Conference on Analytical Chemistry and Applied Spectroscopy, Cleveland, Ohio.
Determination of Cobalt by Lucigenin Chemiluminescence Larry A. Montano' and J. D. Ingle, Jr." Department of Chemistry, Oregon State University, Corvallis, Oregon 9 733 1
The application of lucigenin chemiluminescence for the determination of trace concentrations of Co(I1) is described. This method is based on the enhancing effect that Co(I1) has on the reaction of lucigenin and hydrogen peroxide in a basic solution. The chemiluminescence signal from a Co( 11) enhanced lucigenin reaction, which is in excess of the signal in the absence of added metal activators, is proportional to the Co(1I) concentration. The technique is used for the determination of sub-ppb Co concentrations in solution from dilution of tap water and digestion of NBS orchard leaves. A solvent extraction scheme was developed to isolate Co from the most troublesome interferences, Fe and Mg.
In a recent paper, the chemiluminescence (CL) reaction of Lucigenin (Lc) in alkaline H202solutions was studied and the literature relating to Lc CL was reviewed. It was shown Co(I1) is the most efficient metal activator of the Lc CL reaction and a linear log-log calibration plot was obtained from a detection limit of 20 pptr up to 100 ppm. T h e reagent concentrations were optimized for Co(I1) determinations and an extensive interference study was carried out under the optimized conditions. In this paper, the most troublesome interferences are further studied and the Lc CL reaction is applied to the determination of Co(I1) in NBS orchard leaves and tap water samples.
EXPERIMENTAL The modified discrete sampling CL photometer, analysis procedure, and solutions previously described were used for all Present
address: Gulf Oil Chemicals Company, Orange. Texas. 0003-2700/79/0351-0926$01 .OO/O
measurements ( 1 ) . Exactly 1.0 mL of blank or analyte solution, 0.5 mL of 0.1 M Hz02,and 0.5 mL of 6.8 X M Lc were placed in the sample cell. The CL reaction was initiated by automatic injection of 0.5 mL of 5 M KOH. It was felt that Co(I1) determinations at sub-ppb levels would be most useful by Lc CL since few analytical techniques can be used directly at these ultratrace levels. Over a large range the Co(I1) analytical signal is approximately proportional t o the [Co(II)]1/2. However over the 0 to 1.0 ppb Co(I1) range, an approximately linear calibration curve with some negative deviation can be constructed and was used for all analyses. RESULTS AND DISCUSSION Further Interference Studies. T h e comprehensive interference study ( I ) of 72 elements and 11 anions and complexing agents indicated that the majority of species enhanced or inhibited the Lc CL reaction a t some level. However, most of these interferences would not actually be a problem in analysis of many real samples because they would be present a t a concentration below their respective detection limits by Lc CL. Other than Co(I1) with a detection limit of 20 ppt, only Os(VII1) and Ru(II1) have sub-ppb detection limits and neither of these elements would be expected a t interfering concentrations in environmental samples. An examination of the certified and provisional concentrations of elements in NBS (2) orchard leaves (OL) showed t h a t after digestion of a 200-mg sample and dilution t o 100 mL, only three elements would be present a t concentrations greater than their detection limit. T h e elements are Co, Fe, and Mg and their final concentrations would be about 0.5 ppb, 0.75 ppm, and 15 ppm, respectively. Because of the extremely 1979 American Chemical Society
A N A L Y T I C A L CHEMISTRY, VOL. 51, NO. 7, JUNE 1979
50t O t
-
5 0
0
~
1
1
02 04 [Fe] ( w m )
1
t , 0
1
t
, t
t, IO
[ Mg]
20
0
(ppm)
Figure 1. I n t e r f e r e n c e plots for Fe and M g
low detection limit for Co, all other potential interferents (except Fe and Mg) can be essentially diluted away. For many other biological samples and for direct analysis of river water, it appears t h a t only Fe and Mg would be present a t a concentration large enough to interfere with a Co determination. Direct analysis of seawater (typically 0.3 ppb Co) would not be as simple because five elements would normally be present at a concentration large enough to interfere. These are Fe, Mg, Ag, Ca, and C1. Similarly Co analysis of most rocks would be possible if three interferents (Fe, Mg, and Al) could be controlled. Naturally, the concentration of every element in these samples can vary considerably and perhaps an element which was presumed to be a noninterferent would actually interfere. However, this first approximation did point out that the major interferents in the samples most likely to be analyzed in this study would be Fe and Mg. Consequently, the focus of the remainder of the interference study was to determine how to handle these two interferents. F e and Co enhance the Lc reaction so it is impossible to determine the contribution of each metal to the total analytical signal. I t was anticipated, however. that the Fe concentration in real samples could be determined via atomic absorption spectrometry so the problem remaining was to discover if the Co concentration in an Fe + Co mixture could be estimated if t h e Fe concentration were known. T h e experiment proceeded by measuring the analytical signals of 0.2 ppb Co and five F e solutions whose concentration ranged from 24 to 500 ppb. Analytical signals were also determined for mixtures of 0.2 ppb Co with each of the five Fe solutions. Subtraction of the analytical signal of Fe alone from the analytical signal of an Fe Co mixture, and subsequent estimation of the Co concentration from a Co calibration curve results in a Co concentration that is at least 10% low. However, a satisfactory determination of the Co concentration in a F e + Co mixture can be obtained by using a Co calibration curve t o convert the analytical signal of the mixture (and Fe solution alone) to an "apparent" Co concentration. Subtraction of the apparent Co concentration of the Fe solution from the apparent concentration observed for a Fe + Co mixture results in the Co concentration that was present in the mixture. These data should be compatible with the conditions encountered in a real sample because the ratio of F e to Co in this experiment varied from 120 to 2500. T h e expected ratio for NBS OL is 1500. Figure 1 is a n interference plot which shows how the analytical signal of Fe and Co mixtures varies with F e concentration. The Co concentration is constant at 0.20 ppb. The plot indicates that in the presence of 20 ppb Fe or less, a 0.2
+
927
p p b Co analytical signal can be determined without any interference. An attempt was made to find a suitable masking agent for Fe(II1) such that the analytical signal observed in a Co + Fe mixture would be that of Co alone. EDTA, acetylacetonate, and P20i'- in 100-fold molar excess to 1 ppm Fe(II1) only partially decreased the enhancement of Lc CI, by Fe(II1). Similarily citrate, acetylacetonate, F-, and EDTA did not effectively mask Fe(I1). An attempt was made to find a masking agent for Co because this would permit the evaluation of the Fe analytical signal in mixtures of F e and Co. A 100-fold molar excess of tributylphosphate, EDTA, F-, and CN- with 1.0 ppb Co did not decrease the Co analytical signal by more than 2 0 4 so they were judged not to be useful masking agents for Co. T h e data indicate that formation of complexes by Fe and Co does not greatly affect the ability of these metals to enhance the Lc reaction. This may result because the complexes are destroyed at the high pH the reaction is run at, permitting the metals to react as they would when no complexing agents are present. T h e initial approach in handling the Mg interference problem was based on the finding that the inhibition of the blank CL signal is independent of Mg concentration from 10 to 1000 ppm. By adding Mg to the sample to ensure that its concentration would fall in this range (e.g., 100 ppm), it should be possible to establish a new base-line signal which would be constant in spite of changes in Mg concentration from sample to sample. If the Co analytical signal could be referenced to this new base line, Co determination in the presence of Mg could be accomplished quite simply. This approach did not work, however, because Mg is such an effective inhibitor of the Lc reaction. T h e analytical signals of 1 and 10 ppb Co solutions (above the blank which consists of 100 ppm Mg) in 100 ppm Mg were essentially zero. T h e next approach was to prepare interference plots for mixtures of Co and Mg. Figure 1 shows such a plot in which the Co concentration in various Mg and Co mixtures is constant at 1.0 ppb. T h e data indicate that in order to determine the analytical signal of a dilute Co solution accurately, the concentration of Mg must be about 0.06 ppm or less. This Mg concentration is lower than the detection limit for Mg (0.35 ppm) which may indicate that in the presence of Co, Mg is a more effective inhibitor. It was concluded after these studies that the Fe interference can be easily handled as long as the Fe concentration can be determined by another analytical method. In such cases, the analytical signal of a mixture of Fe and Co can be resolved to yield the concentration of Co in the solution. However, it appears that unless the Mg concentration is less than 0.06 ppm in the final solution, some prior separation of Mg from Co is needed. NBS Orchard Leaves. NBS orchard leaves (OL) were chosen for analysis because the matrix is well defined (2) and the Co concentration is low (0.2 ppm based on dry weight of material). Comparison of the certified values of constituent elements to the known interferences for this method indicated that only Fe and Mg (300 and 6200 ppm, respectively, by dry weight) would be interferents in a Co determination. Since these two remaining interferences could not be masked or diluted to the point where they would not interfere (while keeping the Co concentration above its detection limit), an appropriate separation scheme was needed. The development of an effective method to separate Co from its matrix would have the added advantage of ensuring that many other potential interferents would not present any problems in analysis. This would broaden the applicability of Co analysis via Lc CL to include all liquid and solid samples. The separation technique used in the analysis was solvent
928
A N A L Y T I C A L CHEMISTRY,
VOL. 51, NO. 7 , JUNE 1979
extraction with the chelating agent being 1-nitroso-2-naphthol (1N2N). The procedure was based on the colorimetric determination of Co via its 1N2N complex (Amu 530 nm) taken from Stary (3). Co forms a 1:3 (Co(III):lN2N)complex which is stable in the presence of relatively concentrated acids and bases. 1N2N is capable of oxidizing Co(I1) to Co(III), but usually H 2 0 2is added to speed up the process. The application of this extraction procedure for alloys, ores, and biological samples has been reported (4-6). Extraction of Co via its acetylacetonato complex proved to be unacceptable because of the low extraction efficiencies. In the previous studies, the main interferent was also Fe, but i t could be masked by addition of citrate. Subsequent extraction (usually with CHC1,) leaves the Fe in the aqueous solution and transfers Co (and other metal-lN2N complexes) to the organic phase. Back extraction with 2.0 M HC1 strips interfering ions back into the aqueous phase without dislodging the co-(1N2N), complex from the organic phase. Another back extraction with 2.0 M KOH can be used to strip the excess 1N2N into the aqueous layer. The combination of these two back extractions permits the selective separation of Co from its matrix. It had been reported ( 3 )that Mg is not extracted by 1N2N, so Fe and Mg were not expected to interfere with Co CL analysis of any sample. Before digestion, the OL were thoroughly dried at 120 "C for 1.5 h. When cool, three 250-mg samples were weighed out and placed into 50-mL Erlenmeyer flasks. The digestion and extraction procedure is outlined below. (1)The samples and blanks (empty flasks) were ashed in a muffle furnace a t 450 "C for 18 h. (2) The ash was moistened with a few drops of Millipore water and then dissolved by 5.0 mL of 50% HC1 (v/v). The HCl was evaporated to dryness and the residue taken up with 2.0 mL of 0.10 M HC1. The two reagent blanks were also subjected to the 50% HC1, evaporation, and 0.10 M HC1. A t this point, the reagent blank solutions were clear and the samples varied in color from pale yellow to brown. (3) T o clarify the sample solutions, they were filtered through 1.2 bm Millipore membrane filters. T h e receiving vessels were 10-mL volumetric flasks and 0.10 M HC1 was used to quantitatively transfer from the Erlenmeyer flask to the funnels. It was also used to rinse the filter papers and dilute each 10-mL volumetric flask to the mark. Two reagent blanks were filtered in the same manner as the samples. This completed the digestion, with the Co concentration expected to be about 5 p p b in each sample. (4) The extraction was begun by adding 1.0 mL of 1870 citrate solution, 1.2 mL of 0.50 M KOH, 1.0 m L of 8.8 X M 1N2N, and 1.0 mL of 0.10 M H 2 0 2to each of the three samples and two blanks. The p H of the samples and blanks was then checked and, if necessary, it was adjusted to 3.5 with 0.50 M KOH or 0.10 M HC1. All solutions were then allowed to stand a t room temperature for 1 h. The 18% citrate solution was prepared by dissolving 50 g citric acid (Mallinckrodt) in 15 mL H 2 0 and then adding 50 mL concentrated N H 4 0 H (Mallinckrodt). This solution was diluted to 250 mL after adjusting its p H to 3.4 with 6 M HC1. T h e 1N2N was prepared by dissolving 0.0382 g 1N2N (Baker) in 25 mL of 0.10 M KOH. The solution must be filtered as the 1N2N is not completely soluble. The 0.50 M KOH was prepared by dilution of the stock 5.0 M KOH used in CL determinations and the 0.10 M H202is the stock HzOz used in CL determinations. ( 5 ) To check the efficiency of the extraction, under the exact conditions that were used for the samples, three 10.0-mL Co standards were also extracted. Their concentrations were 4.0, 5.0, and 6.0 ppb, and they were prepared by diluting appropriate volumes of 100 ppb Co stock solution with 0.10 M
Table I . Methods for Co Determination detection limit, ppb
method atomic absorption (flame) atomic absorption
sample volume
7
5
1
ref
5 PL
8
(carbon r o d )
atomic emission (inductively coupled plasma) colorimetry (nitroso-R salt) X-ray fluorescence (energy dispersive, k
cy
9
2
5mL
10
1.06ng
-
11
0.5
10mL
12
0.005a
5 mL
13
)
neutron activation analysis (S9Co(n,Y ) ~ O " C O ) biological assay (Rhizobium meliloti) anodic stripping voltametry ( pre-electrolysis time, 1 0 min) a
0.1
3
14
Based o n incubation period of 4 7 h .
HC1. These extraction standards were treated in the same manner as the samples and reagent blanks. (6) The actual extraction involved quantitative transfer of the sample to a 125-mL separatory funnel which held 25 mL of reagent grade CHCI,. After shaking for 3 min, the layers were allowed to separate and the organic layer was transferred to another 125-mL separatory funnel which held 20 mL of 2.0 M HC1. The second separatory funnel was agitated for 30 s and the layers were allowed to separate. The organic layer was transferred to a third separatory funnel which contained 20 mL of 2.0 M KOH. After shaking for 30 s, the organic layer was transferred to a 100-mL beaker and the process repeated two more times with fresh 25-mL portions of CHC13. When this had been completed for the samples, blanks, and three extraction standards, the respective beakers were placed on a hotplate and the CHC13 was evaporated to dryness. ( 7 ) Then, 0.5 mL of concentrated H2S0, and "210, was added to each 100-mL beaker, the beaker was covered with a watchglass, and then the solution was refluxed at high heat for 1 h. The watchglasses were removed and the HC10, and H 2 S 0 4(in that order) were evaporated away. This rigorous treatment was required to break up the Co (lNZN), complex. (8)T o the residue in each beaker, 10 mL of 0.01 M "0, was added and then heated gently to achieve full dissolution. T h e contents were quantitatively transferred to a 100-mL volumetric flask and diluted to the mark with 0.01 M HN03. The p H of the samples, blanks, and extracted Co standards was now the same as the Co standards used to prepare the calibration curve. The expected concentration of Co in the samples was now about 0.5 p p b a n d t h e extracted Co standards were about 0.4, 0.5, and 0.5 ppb. The analyses of the samples were carried out by preparing a Co calibration curve for Co concentrations from 0.1 to 1.0 ppb, and then determining the analytical signals of the samples, blanks, and extraction standards. The average percent recovery of the extraction standards was used to determine the percent recovery of the samples. T o verify that the extraction procedure was effective in separating Co from its two principal interferents, Fe and Mg, extractions were performed on 5 ppb Co solutions containing 10 ppm Fe(II1) and 200 ppm Mg. These concentrations are about what is expected for Fe after digestion of OL (0.75ppm) and in 13-fold excess for the expected concentration of Mg (16 ppm). After extraction (steps 4 and 6 above), atomic absorption analysis (Varian AA-6) showed that the concentration of the two interferents had remained unchanged in the aqueous phase.
A N A L Y T I C A L CHEMISTRY, VOL. 51, NO. 7, JUNE 1979
929
Table 11. CL Methods for Co Determination conditions
system Lc
+
H,O,
+
KOH
luminol + H,O, KOH + Co
+
gallic acid + H,O, NaOH + Co a
t
+
Co
Lc, 1 x
M ; H,O,, 5 x
M; final pH,
detection limit, ppb’
known interferences
ref.
1.4
T1,Fe, Ni, Ag, Bi, C h ,
16
13.5; discrete sampling system luminol, 2 x lo-‘M ; H,O,, 1 x M; OH-, 1 x lo-,M; flow system with
0.01
Cr, Os, Ce, Mn, PI) Al, Sn, Pb, Bi, Cr, Fe, Cu, Zn, Ag, Mn, Ce
microporous membrane gallic acid, 2 x l o - ’ M ; H,O,, 3 x final pH, 1 2 . 2 ; flow system
0.4
Pb, Cu, Ag, M n
M;
23
24
Lc, H,O,, and Co concentrations are before introduction of the solution into the sample cell
The efficiency of the extraction varied from 61 to 1 0 5 7 ~for 5 ppb Co which is why extraction standards were used in the OL analysis. Contamination from glassware or reagents was judged t o be minimal because the analytical signal of extraction blanks was consistently small. The cause of the variations in extraction efficiency could not be determined, although it was discovered that freshly prepared 1N2K solution gave a higher extraction efficiency than solutions that were 2 or more days old. The estimated Co concentration in the OL (dry weight basis) based on five determinations of OL is 0.12 ppm with a 18% relative standard deviation where the greatest source of imprecision is due to the imprecision in the extraction efficiency. T h e listed Co concentration is 0.2 ppm but this is not a certified value. T a p Water. The other sample picked for analysis was tap water. T h e expected Cu ccncentration in such a sample is typically 0.1 to 1 ppb which means that direct analysis (no extraction) of the sample should be possible with this technique. Atomic absorption analysis of the sample indicated that the Fe and Mg concentrations were 0.11 and 3.17 ppm, respectively. Dilution by a factor of five will be sufficient to eliminate the Fe interference, but a 50-fold dilution is required for the Mg interference. A factor of 50 dilution would certainly reduce the Co concentration to a value below its detection limit for this technique, so direct analysis of tap water is not possible. When direct analysis was tried on the tap water sample (diluted 1% to eliminate Fe interference), the standard additions plot clearly indicated the presence of a negative interferent. No estimation of the Co concentration in this diluted tap water sample was possible because the analytical signal for this solution was negative. T o demonstrate t h a t analysis without prior separation of Co is possible, the tap water was spiked with a small volume of a concentrated Co solution (0.10 mL 1.0 ppm Co added to 20.0 mL tap water). The Co concentration in the tap water was 5 ppb and the other species in the matrix were within 1% of their original value. T h e Co determination was carried out by first diluting the spiked tap water sample by 50, and then performing the CL analysis. By the standard additions method and the Co calibration curve, the Co concentration in the diluted sample was 0.095 and 0.098 ppb, respectively. The expected Co concentration was 0.10 ppb so the analysis was accurate t o about 3.5%. The accuracy of the Co determination using the Co calibration curve indicates that interferences were no problem in the diluted sample. CONCLUSIONS The results indicate that ultratrace determinations of Co(I1) are possible via Lc CL. This method provides a lower detection limit for Co(I1) than that reported for any other instrumental technique except CL. Co detection limits by other instrumental methods are listed in Table I. The table indicates that determination of Co below 10 ppb and particularly below 1ppb which is needed for many environmental samples, is still a difficult analytical problem. Although many species affect the CL signal, most do not a t environmental or natural levels. Direct analysis of water
samples via Lc CL is possible if the concentration of Co is large enough t o permit the elimination of the Fe and Mg interferences by dilution. Analysis of rock, soil, biological samples, or water if the Mg and Fe levels are too high is possible if an appropriate separation scheme can be worked out. The 1N2N solvent extraction procedure appears to be effective but work still needs to be done t o improve the reproducibility of the extraction. This extraction should be useful for any Co determination method in which interferences such as Fe and Mg are a problem. I t appears that Mg and Fe are the most important interferences. The ratio of Fe or Mg to Co concentrations can be about lo4 because of the extremely low detection limit for Co. However, because the natural abundance of Fe and Mg compared to Co in many samples is much greater than 10‘. a problem still exists. U’e have noted that in other CL systems we have studied (e.g. lophine, pyrogallol) that again Fe and Mg cause the same types of problems. Clearly the development of a fast and reliable technique to separate Mg and Fe from the rest of the matrix would be helpful in many CL systems for Co determinations. I t appears that Co(I1) is usually among the best, if not the best, metal activator in many CL reactions in (alkaline H 2 0 2 solutions. This supports the suggestion that Co(I1) effect is related to some degree on its ability to interact with H 2 0 2(15). Co has previously been determined by solution CL with lucigenin (16-18), luminol (19-23), and gallic acid (24) and these data are summarized in Table 11. The detection achieved with Lc here is about two orders of magnitude better than achieved previously with Lc ( 1 6 ) , over an order of magnitude better than with gallic acid (241, and about equivalent to that obtained with luminol (23). ‘There is some confusion in the literature about the best detection limit for Co with luminol. Seitz and Hercules (20) reported the detection limit for Co as 0.6 pptr (parts per trillion) in a flow system. They pointed out, however, that this value was based on extrapolation of data obtained a t much higher concentrations. This extrapolated detection limit has been perpetuated in the literature even though extrapolated detection limits, particularly below 1ppb, must be viewed with caution (25). Recently Nau and Nieman reported a detection limit of 10 pptr Co(I1) in the luminol system using a flow system with introduction of reagents through a microporous membrane (23). This is the lowest achieved detection limit for Co(I1) with luminol. This is the only article in the literature which actually reports experimental measurements of Co below 1 ppb. In terms of interferences, luminol CL is activated by several other transition metals a t sub-ppb concentrations such as Cu, Fe, Cr, Ni, and Mn. T h e gallic acid system appears to be more specific for Co since it has fewer interferences. I t has the disadvantage that no suitable buffer was found so that the p H of samples and reagents would have t o be adjusted carefully. Also undetermined suppression effects were noted in river water samples. LITERATURE CITED (1) Montano, L. A , ; Ingle, J. D.,Jr., Anal. Chem., 1979, 51. preceding paper in this issue.
A N A L Y T I C A L CHEMISTRY, VOL. 51, NO. 7, JUNE 1979
930
National Bureau of Standards, Standard Reference Material No. 157 1, Orchard Leaves, 1971 (Revised 1976). Stary. J. "Solvent Extraction of Metal Chelates," Macmillan: New York,
1964 Lundquist, R.; Markle, G. E.; Boltz, D. F . Anal. Chem. 1955, 2 7 , 1731. Saltzman, B. E. Anal. Chem. 1955, 2 7 , 284. Cogan, E. Anal. Chem. 1960, 32, 973. Fassel. V. A.; Kniseley, 8.N. Anal. Chem. 1974, 46, lllOA. Varian Techtron Publication, #85-100044-00, Palo Alto, Calif., 1975. Olsen, K. W.; Haas, W. J.; Fassel, V. A. Anal. Chem. 1977, 49, 632. Sandell, E. B. "Colorimetric Determination of Traces of Metals", 3rd ed; Interscience: New York, 1959. Aiginger. H.; Wobrauschek, P.; Brauner, C. Meas., Detect. Controlhnviron. Pollution Proc. Int. Symp. 1976. 197;Chem. Abstr. 1977, 87, 11081Ot. Burrel, D. C. "Atomic Spectrometric Analysis of Heavy-Metal Pollutants in Water", Ann Arbor Science: Ann Arbor, Mich.. 1975. Evans, H. J.: Kliewer, M.; Lowe. R.; Mayeux, P. A. Plant Soil. 1964, 21.
153. Hovsepian, B. K.; Shain, I. J . Electroanal. Chem. 1966, 12, 397. Burdo. T . G.;Seitz, W . R . Anal. Chem. 1975, 47, 1639. Dubovenko, ._ L. I.; Beloshitskii, N. V. J . Anal. Chem. USSR. 1974, 29. 85
Babko, A. K.; Dubovenko, L. I.; Terletskaya, A. V. 1966, 32, 1326.
Sov. Prog. Chem.
(18) Bognar. J.; Sipos, L. Mlkrochim. Ichnoanal. Acta 1963, 3 , 442. (19) Babko. A. K.: Lukovskava. N. M. Zavod. Lab. 1963. 2 9 . 404: Chem. Abstr.'1963, 5 9 , 3301d. (20) Seitz, W. R.; Hercules, D. M.in "Chemiluminescenceand Bioluminescence," Cormier, M. J.; Hercules, D. M.; Lee, J., Ed., Plenum Press: New York, 1973;pp. 427-449. (21) Hartkopf, A.; Deiumyea, R. Anal. Letf. 1974, 7 , 79. (22) Sheehan, T. L.; Hercules, D. M. Anal. Chem. 1977, 49, 446. (23) Nau, V.: Nieman. T. A. Anal. Chem. 1979, 51, 424. (24) Stieg, S.;Nieman, T. A. Anal. Chem. 1977, 49, 1322. (25) Ingle, J. D., Jr.; Wilson, R. L. Anal. Chem. 1976, 48, 1691.
RECEIVED for review October 20, 1978. Accepted February 12, 1979. ~ ~ k is made ~ to the ~NSF (grant ~ =CHE-76-16711)for partial support of this research and to the University Corporation for Atmospheric Research for fellowship support for one of us (L.A.M.). Presented in part a t the 1977 Northwest ACS meeting, Portland, Ore., and a t and the 29th Pittsburgh Conference On Applied Spectroscopy, Cleveland, Ohio.
Intracavity Absorption with External Fluorescence Measurement for Detection of Radioiodine Isotopes J. P. Hohimer" and P. J. Hargis, Jr. Sandia Laboratories, Albuquerque, New Mexico 87 185
We report the use of intracavity absorption in a CW dye laser as a sensitive analytical method for the detection of iodine-129. With an external fluorescence detection scheme employing photon counting, quantitative isotope specific measurements of '"I2 and 2I'" have been made at concentrations in the range of 5 X 10" to 7 X ioi5 mo~ecu~es/cm~. Scaling estimates indicate that a lower level may be achieved if desirable. I n addition, this method permits the simultaneous detection of a number of iodine isotopes.
Radioanalytical methods such as liquid scintillation counting and neutron activation analysis are presently used to measure trace concentrations of the long-lived radioisotope iodine-129. However, these methods are time-consuming and cannot be used for real-time measurements. New methods are needed which are capable of measuring iodine-129 in real time at or below the maximum permissible concentration of 2.0 x 10." Ci/cm3 (1). This concentration corresponds to an IBI2 number density of 7.9 X lo8 molecules/cm3. Laser-excited fluorescence spectroscopy is a sensitive method which has been studied for the real-time detection of trace iodine concentrations ( 2 ) . However, the severe quenching of the iodine fluorescence by atmospheric molecules (2-4) may limit the sensitivity of this fluorescence method for airborne radioiodine measurements. An alternative method for the detection of airborne iodine-129 is intracavity absorption spectroscopy. Intracavity absorption spectroscopy has been used to detect very weak absorption lines in a number of atomic and molecular species (5-10). This method involves placing a weakly absorbing species within the cavity of a broadband dye laser and measuring its effect on the spectral output of the laser. T h e wavelength-dependent loss introduced by even a very weak absorber dramatically decreases the intensity of the laser output a t the absorber wavelength. T h e sensitivity of a broadband dye laser to small selective losses has been attributed to multiple passes of the laser beam through the absorbing medium as well as to the strong competition of simultaneously oscillating modes for the available energy in the homogeneously broadened gain medium (7, 8, 1 1 ) . An increase of absorption sensitivity by a factor of lo5,compared to a single-pass experiment, has been observed in an ex0003-2700/79/0351-0930$01 .OO/O
periment with an iodine-127 vapor cell inside the cavity of a CW dye laser (8).
EXPERIMENTAL A schematic diagram of the experimental apparatus used t o measure the sensitivity, linearity, and isotopic selectivity of the intracavity absorption of molecular iodine is shown in Figure 1. The 488-nm output from an argon ion laser (Spectra Physics, model 171-05)was used to pump a CW dye laser (Spectra Physics, model 375) which was amplitude stabilized by means of a feedback loop to the argon laser. Broadband emission was obtained when the dye laser was operated with all tuning elements removed from the laser cavity. This produced a laser output power of 50 mW with a spectral bandwidth of 0.75 nm fwhm (at 1.0-W argon pump power). However, under these conditions, the broadband cavity reflectors allowed the laser wavelength to shift as the wavelength dependent loss was modified by increasing the intracavity iodine vapor pressure. This caused large fluctuations in the experimental data. The reproducibility of the data was improved when the laser was operated with a tuning wedge in the cavity. This reduced the laser bandwidth to about 0.43 nm fwhm at the same output power level but greatly improved the laser wavelength stability. Even though small wavelength shifts were still observed, they had little effect on the reproducibility of the measurements. The intracavity iodine was contained in a quartz vapor cell with Brewster-angle windows and an internal pathlength of 5.0 cm. Because of the sensitivity of the CW dye laser to small wavelength dependent losses, wedged ( 2 30 in.) Brewster windows were required on the intracavity vapor cell to eliminate etalon effects in the dye laser output spectrum. The vapor cells used in these measurements were evacuated and subsequently filled with either iodine-127 or iodine-129. The iodine-129 vapor cells contained an unknown isotopic abundance of fission-produced iodine-127 estimated to be about 25% (12). The intracavity iodine number density could be varied from 5 X 10'l to 7 X 1015 molecules/cm3 by regulating the temperature of a cold finger on the vapor cell (-60.2 to 20.5 "C). External fluorescence detection was used for these intracavity absorption measurements since it gives a quantitative measure of the concentration of the intracavity absorbing species, takes full advantage of the many iodine absorption lines which lie within the laser bandwidth, and is more sensitive than other techniques which directly record the dye laser spectrum with either a high resolution spectrometer or a scanning Fabry-Perot etalon. In addition, the method is insensitive to fluorescence quenching 0 1979 American Chemical Society
