I t must be mentioned that the relative extractions of the various metals into TOA-benzene from acid solutions containing uranium, presented above, were computed on the basis of their absorbance with reference to the pure metals similarly extracted from acid media. I t is believed that the matrix effect of small concentration of uranium coextracted with the trace elements into TOA-benzene is primarily the reason for an apparent lower than 100%recovery of Cd, Ag, Hg, Pb, and Bi. Repeated extraction from the aqueous raffinate did not improve the "computed" recovery values. To the extent that the calibration for the elements separated from uranium by TOA-benzene extraction and determined by atomic absorption were reproducibly linear in the entire concentration range, the analytical application of the method is evident. CONCLUSION
The combination of the solvent extraction and atomic absorption determination of a series of trace elements has proved to be a unique approach to the analysis of nuclear grade uranium (ADU and UOs). While several pure elements are well extracted by long-chain amines from hydrochloric acid solution, the presence of appreciable concentration of uranium and chloride ions constitute serious interference, lowering the activity of the metal chloro complexes. Such an interference is so marked for silver, for example, that TOA-benzene does not extract the element from U02ClZ-HC1 medium. Addition of a small concentration of KI surmounted the difficulty and provided practical and reproducible extraction of the various elements in the presence of uranium.
The exact mechanism of the extraction in the presence of iodide ions is not well understood. The possibility exists for the formation of stronger iodide or ever mixed iodo-chloro complexes of the metal ions. I t is also pertinent to indicate that even precipitates like the chlorides and iodides of Hg, T1, Ag, Bi, and P b are dissolved and extracted by TOAbenzene from hydrochloric acid solution ( 1 0 ) . Iron and cobalt were not extracted by TOA-benzene; copper was extracted, but it could be stripped quantitatively from the organic phase by washing with HCl-KI solution. Concentrations of uranium in the aqueous phase have approached up to 300 grams per liter. However, no difficulty in phase separation with TOA-benzene was experienced. Accommodating such high concentration of the matrix uranium necessarily provides for increased analytical sensitivity for the trace elements. The enhanced absorbance obtainable in direct burning of the organic phase is an added advantage of procedure. The technique outlined in this paper is routinely used for the analysis of Cd, Ag, Au, Hg, Pb, and Bi in uranium. Calibration curves obtained with the uranium matrix have realized relative standard deviations of approximately 2 , 4 , 6 , 2, 12, and 3%, respectively, for these elements in the trace concentration of interest. An extension of this work is in progress for the determination of trace metals in nuclear grade thorium (14).
RECEIVED for review October 11, 1973. Accepted April 22, 1974. Abrao and S. de Moraes, lnstituto de Energia Atbmica. Sso Paulo, Brazil, unpublished work, 1972.
(14) A.
Atomic Absorption Determination of Mercury in Fish Using the Coleman MAS-50 Mercury Analyzer Pekka Kivalo, Asko Visapaa, and Runar Backman Chemical Laboratory, Technical Research Centre of Finland, 02 150 Otaniemi, Finland
In this Research Centre, a large number of mercury determinations of fish samples have been performed by means of neutron activation analysis in conjunction with studies concerned with environmental protection. In 1971, the Coleman Division of the Perkin-Elmer Co. introduced a mercury analyzer, called MAS-50 ( I ) , which is based upon cold vapor atomic absorption ( 2 ) and upon the procedure developed by Hatch and Ott ( 3 ) .In view of the time-consuming and costly nature of activation analysis in mercury determinations, it was decided in this laboratory to investigate the suitability of the MAS-50 analyzer for mercury determinations in fish. The literature concerning nonflame methods, up to and partially including 1970, has been reviewed by Manning ( 4 ) . Since 1971, a few references to the MAS-50 analyzer have been made in the literature (5-7). Coleman Instruments. Maywood, Ill., Coleman Mercury Analyzer MAS50, Operating Directions 50-900 (1971). (2) R. G. Smith, "Methods of Analysis for Mercury and Its Compounds: A Review," in "Environmental Mercury Contamination,"R. Hartung and B. D. Dinman, A n n Arbor Science Publishers Inc., Ann Arbor, Mich., 1972. (3) W. R. Hatch and W. L. Ott, Anal. Chem., 40, 2085 (1968). (4) D. C. Manning, At. Absorption Newsleff., 9, 97 (1970).
EXPERIMENTAL Instrumentation. In this instrument, mercury vapor is constantly circulated by a pump within a closed loop that comprises an external BOD (biological oxygen demand) bottle connected by silicone rubber tubes to a glass absorption cell with plastic windows. The mercury vapor in the cell absorbs the 253.7-nm radiation of a mercury lamp a t one end of the cell, and a photoelectric cell with a narrow band-pass filter is placed at the other end. The diminution in intensity is measured. The instrument is equipped with an electronic memory circuit which stops the absorbance reading a t the maximum. In preliminary studies, the original plastic end windows of the absorption cell developed cracks, and rapidly became opaque by reason of the U V radiation of the lamp and the ozone produced by the UV radiation. The original windows were replaced by windows made of quartz. When the instrument was operated with the memory circuit shut off, it was noted that when a sample containing elemental mercury was being aerated, the maximum absorbance was attained
(1)
1814
(5) J. D. Suggs, D. H. Petersen, and J. B. Middlebrook, Water Pollut. Contr. Res. Ser. 1690HTD 03/72, Environmental Protection Agency, Washington, D.C., 1972. (6) J. E. Longbottom and R. C. Dressman, Chromatogr. Newsleff., 2, 17 (1973). (7) S. H. Omang and P. E. Paus, Anal. Chim. Acta, 56, 393 (1971).
A N A L Y T I C A L C H E M I S T R Y , VOL. 46, NO. 12, OCTOBER 1974
within 5 to 6 seconds [and not after one minute, as stated in the instrument manual ( I ) ] .Subsequently, the reading was found to decrease slowly (Figure l).Other researchers ( 8 ) have found a similar phenomenon, which they believe originates in oxidation of the mercury vapor, or adsorption on the walls of the air circulation loop. With the aim of discovering the reason for the gradual disappearance of mercury from the gas phase, the following experiments were made, all of them with a concentration corresponding to a scale reading of 1 kg Hg: 1) Before aeration, a piece of cardboard was placed between the lamp and the absorption cell, to cut off the radiation. Aeration was begun and, in successive experiments, the cardboard was removed after predetermined times, with the memory circuit on. The readings obtained, as a function of timg, were almost the same as those when the radiation was entering the cell continuously with the memory circuit off (Figure 1).Consequently, the diminution in the reading does not result from oxidation of the mercury vapor by the UV radiation or ozone. 2) Since the mercury vapor is not oxidized, its disappearance cannot be caused by absorption into the solution in the BOD bottle. This was confirmed by blowing mercury vapor into an empty and dry BOD bottle as the sample. A similar diminution occurred in the reading after an initial rapid rise, both with the radiation entering the cell, and with the radiation blocked off for given lengths of time, as indicated in 1). 3) When glass wool, with a surface area of about 4000 cm2-uiz., a tenfold increase in the adsorbing surface-was placed in the air circulation loop, no change was observed in the speed of decrease in the reading. When thin silicone rubber slices, about 4000 cm2 in area, were placed in the air circulation loop, the speed of decrease in the reading became slightly faster. Consequently, the diminution in the reading was not attributable to adsorption of the mercury vapor on the glass surfaces of the air circulation loop. Apparently, the mercury vapor is slightly removed with silicone rubber. 4) When the silicone rubber tubes outside the instrument case were replaced by Teflon tubing, definite retardation in the diminution of the reading was observed (Figure 1). This indicated the escape of part of the mercury vapor through the walls of the silicone rubber tubing. T o confirm this, equal amounts of mercury were placed in pieces of tubing, 3 cm in length, made from silicone rubber and Teflon. These pieces were closed by tightly fitting glass rods. Each piece of tubing, containing mercury, was placed in a separate BOD bottle, and the bottles were closed; after these had stood 10 minutes, determinations were made in the usual way, by starting aeration. The results indicated that the amount of mercury vapor diffused through the wall of the silicone rubber tubing was about ten times that diffused through the Teflon tubing wall, although the tube wall thicknesses were 2.2 mm for silicone rubber, and 0.4 mm for Teflon. Thus, a part of the loss of mercury on prolonged aeration arises from diffusion through the walls of the connection tubes. In the pump, mercury vapor comes into contact with surfaces of nylon, rubber, and hard rubber. Hard rubber, in particular, may induce absorption of mercury vapor as a result of its sulfur content. A suitable rotameter must be employed for periodic checks of the air flow. Permanent connection of a rotameter in the air circulation loop is unnecessary, as calibration is performed before each series of determinations. Furthermore, permanent connection of a rotameter could lead to more rapid disappearance of the mercury vapor. However, it must be borne in mind that, despite the causes mentioned for the disappearance of mercury vapor from the gas phase in prolonged aeration, the determination itself is not affected, provided that the tubing material does not allow mercury vapor to escape too rapidly. All the effects (absorption, adsorption, and diffusion) are eliminated in the calibration procedure, and by use of the memory circuit. Precision and Accuracy of the Instrument. The reproducibility and accuracy of the instrument were checked by means of three concentrations (0.50, 1.00, and 1.50 ppm) of mercury as mercuric chloride, without acid digestion, that is, starting with the addition of permanganate, followed by hydroxylamine hydrochloride, and continuing with reduction by stannous chloride and aeration. Table I presents the results of six parallel determinations for each (8) J. F. Uthe, F. A. J. Armstrong, and M. P. 27, 805 (1970).
Stainton, J. Fish. Res. Bd. Can.,
INSTRUMENT READING r g Hg
I
2 4 6 8 10 TIME I N WIN F R O M THE START OF AERATION
Figure 1. Decrease in the reading as a function of time on aeration of a reduced HgClp-solution with the memory circuit off
-- ) Readings obtained with the original silicone rubber tubing. (- - -) Readings obtained with Teflon tubing replacing the silicone rubber tubing outside the instrument case. (X) Readings obtained after blocking the radiation for 0.5, 1. 2, 4, and 5 minutes (
Table I. Parallel Determinations of t h e Mercury Content of Mercuric Chloride Solutions of Known Concentration without Acid Digestion H g added, ppin
0 60
1.00
1 50
Hg found, ppm
0.51 0.52 0.48 0.50 0.49 0.49 0.50 0.00022 0.015 2.95
0.97 1.02 0.99 1.00 1.02 0.98 1.00 0.00043 0.021 2.07
1.52 1.50 1.49 1.52 1.50 1.49 1.50 0,00019 0.014 0.91
Mean value Variance S t d deviation Re1 std dev, %
concentration. It is observable that the degree of accuracy is good at all three concentrations. The values of the relative standard deviation vary between l and 3. Acid Digestion Procedure. For investigation of the acid digestion procedure for fish samples, given in the Applications Data Sheet (91,several determinations were made with known amounts of mercuric chloride. The precision and accuracy were satisfactory. However, when determinations were made with fish samples of which the mercury concentrations had been determined by neutron activation analysis ( l o ) ,the precision was satisfactory in parallel determinations, although the results obtained were only about 30 to 70% of those with activation analysis. Consequently, it was obvious that the acid digestion procedure was unsatisfactory with respect to organomercuric compounds. In biological samples, mercury is often very strongly bound to the organic material. No complete knowledge is possessed of the exact nature of the bonding ( I 1 )~ For biological material, one of the major problems is the separation of mercury from the large mass of organic matter with which it is associated (2). Sample digestion constitutes a critical step in quantitative determination of the total mercury. During digestion, errors can arise as a result of the incomplete extraction of mercury, the nonquantitative conversion of organomercury into mercury(I1) ions, or the loss of mercury vapor ( 1 2 ) . Trials were made with a number of different acid digestion procedures (different acids, mixtures, temperatures, and reaction (9) Perkin-Elmer Co., Coleman Instruments Division, Maywood, Ill., Appiica-
tions Data Sheet MAS-50-6 (1971). (10) E. Hasanen, Suom. Kemistilehti, 438, 251 (1970). (1 1) I. M. Kolthoffand P. J. Elving, "Treatise on Analytical Chemistry," Part 11, Vol. 3, lnterscience Publishers, New York, N.Y..1961, p 253. (12) I. K. Iskandar, J. K . Syers, L. W. Jacobs, D. R. Keaney, and J. T. Gilmore, Analyst(London), 97, 388 (1972).
A N A L Y T I C A L C H E M I S T R Y , VOC. 46, N O . 12, OCTOBER 1974
1815
T a b l e 11. Results in ppm, O b t a i n e d for O r g a n o mercuric C o m p o u n d s b y the M e t h o d Proposed w i t h and w i t h o u t A c i d Digestiona Compound
Without acid digestion
With acid digestion
Methylmercuric chloride Phenylmercuric a c e t a t e Phenylmercuric hydroxide
0.10 0.10 0.05
1.01 1.02 0.98
'I
Table IV. Six Parallel D e t e r m i n a t i o n s of the M e r c u r y C o n t e n t of O n e Pike Samplea
In all cases, the known concentration equals 1 ppm Hg.
a
D e t n No.
Hg. p p m
1 2 3 4 5 6 M e a n value Variance S t d dev Re1 s t d dev, %
0.86 0.88 0.89 0.92 0.94 0.84 0.89 0.0014 0.037 4.18
Result ohtained by activation analysis: 0.90 ppm.
Table 111. Parallel D e t e r m i n a t i o n s of the M e r c u r y C o n t e n t of M e r c u r i c C h l o r i d e Solutions of K n o w n C o n c e n t r a t i o n w i t h Acid D i g e s t i o n Hg added ppiii
Hg found, pprn
M e a n value Variance S t d dev Re1 s t d dev, %
0 50
0.48 0.48 0.50 0.52 0.47 0.49 0.49 0.00320 0.018 3.65
1 00
0.97 0.94 1.00 0.98 1.04 0.95 0.98 0.0013 0.036 3.71
1 50
0.48 1.55 1.53 1.50 1.46 1.45 1.50 0.0016 0.039 2.63
times) (R,9, 13-16); that finally adopted, a modification of the procedure published by Hasanen ( I O ) , was as follows: About 1 gram of homogenized fish tissue is placed carefully in a 50-ml round-bottom flask, provided with a ground glass joint. The exact weight of the sample ( f O . O 1 gram) is determined by weighing the flask before and after addition of the sample. The sample is transferred to the bottom of the flask; 2 ml of fuming nitric acid is added slowly, followed by 1 ml of sulfuric acid ( d = 1.84). The flask is immediately connected to an Allihn reflux condenser, 30 cm in length, which has at its upper end a 15-cm long capillary (0.5-1 mm inner diameter) connected by a ground glass joint to the condenser. The function of the capillary is to prevent mercury and acid losses, and diminish bumping. The sample must always be completely covered by the acid mixture. The mixture is boiled gently for 30 minutes; the clear brown solution is then transferred quantitatively to a BOD bottle, into which the condenser is rinsed with one 10-ml portion, and the flask with two 5-ml portions of distilled water. The volume of the solution in the BOD bottle is adjusted with distilled water to 100 ml. Potassium permanganate crystals are added to the bottle with swirling, until a pink color persists. The determination is then continued according to the Applications Data Sheet instructions (9). One determination takes about 40 minutes. No frothing occurs in the aeration step after this digestion procedure. Recovery with Organomercuric Compounds. For investigation of whether organically-bound mercury reacts completely in the acid digestion procedure proposed, methylmercuric chloride, phenylmercuric acetate, and phenylmercuric hydroxide were used as model substances for organomercuric compounds. The acid digestion procedure gave complete recovery (Table 11). Without acid digestion (beginning with permanganate oxidation), the recovery was not more than 5 to 10%. For methylmercuric chloride, 1 hour of acid digestion time was required for complete recovery whereas a boiling time of 30 minutes was adequate for phenylmercuric acetate and phenylmercuric hydroxide.
(13) S. H. Omang. Anal. Chim. Acta, 63, 247 (1973). (14) R. J. Evans and J. D. Bails, Environ. Sci. Techno/.,6, 901 (1972). (15) W. L. Hoover, J. R. Melton, and P. A. Howard, J. Ass. Offic. Anal. Chern.,54, 860 (1971). (16) V. A . Thorpe, J. Ass. Offic. Anal. Chern., 54, 206 (1971).
Table V. M e r c u r y C o n c e n t r a t i o n s in ppm of D i f f e r e n t Fish Samples, Obtained w i t h MAS-50 A n a l y z e r , A p p l y i n g the Procedure Proposed and N e u t r o n A c t i v a t i o n A n a l y s i s (10) T y p e of fish
MAS-50
Activation analysis
Whitefish Roach Roach Perch Perch Perch Bream Perch Walleye Pike Pike Bream Pike Pike Pike Pike Pike Pike
0.17 0.25 0.26 0.33 0.34 0.40 0.46 0.46 0.48 0.49 0.57 0.58 0.58 0.61 0.62 0.86 0.88 0.92
0.17 0.26 0.28 0.32 0.32 0.44 0.50 0.42 0.50 0.52 0.53 0.61 0.54 0.62 0.62 0.90 0.90 0.90
RESULTS AND DISCUSSION T h e precision a n d accuracy of t h e method proposed (acid digestion, t r e a t m e n t with permanganate a n d hydroxylamine hydrochloride, reduction, and t h e determination proper) were ascertained by six parallel determinations with known a m o u n t s of mercuric chloride. T h e results o b tained a r e listed in T a b l e 111, and indicate that, at levels of 0.50 a n d 1.00 p p m Hg, t h e relative s t a n d a r d deviation is 3.7, a n d t h a t t h e precision is satisfactory a t all t h r e e concentrations used. A t a level of 1.50 p p m (seldom encountered in fish samples), t h e relative s t a n d a r d deviation is 2.6. Consequently, t h e acid digestion induces a slight increase in t h e spread of t h e results (see T a b l e I), b u t does not affect t h e accuracy. T h e mercury content of one pike sample was determined six times t o check t h e reproducibility a n d accuracy in d e termination of t h e mercury in fish (Table IV). Activation analysis showed t h a t t h e mercury concentration in this pike sample was 0.90 p p m . It is discernible from t h i s T a b l e , t h a t t h e m e a n value obtained was 0.89 p p m Hg: t h e relative s t a n d a r d deviation was 4.2; this can be considered satisfactory in analysis of t h i s type. T h e accuracy of t h e method proposed was further checked by determination of t h e mercury content of several fish samples (pike, bream, perch, roach, whitefish, a n d walleye), both by t h e method proposed, a n d by neutron activation analysis (10). T h e activation analyses were con-
1816 * ANALYTICAL CHEMISTRY. VOL. 46, NO. 12. OCTOBER 1974
ducted in the Reactor Laboratory of this Research Centre. Table V presents typical results. The value of correlation coefficient r was 0.992, and its statistical significance t = 30.96; this indicates that the correlation is significant a t the 99% probability level (17). When mercury determinations are being made, the utmost care must be exercised throughout, to ensure that (17) K. Doerffel, "Beurteilung von Anaiysenverfahren und -Ergebnissen,"
Springer-Verlag,Berlin, 1962.
contamination is prevented. Precise analysis is dependent upon avoiding contamination of glassware, reagents, and standards. A reagent blank should be run to confirm the purity of the reagents. When analytically pure reagents are utilized, a typical value found for the reagent blank is 0.01 Pg Hg.
RECEIVEDfor review January 21, 1974. Accepted May 7, 1974.
Analytical Applications of the Graphite Braid Nonflame Atomizer Akbar Montaser' and S. R. Crouch2 Department of Chemistry, Michigan State University, East Lansing, Mich. 48824
In a previous report ( I ) , the graphite braid atomizer (GBA) was introduced as a new filament-type nonflame atomizer for atomic absorption (AA) and atomic fluorescence ( A F ) spectrometry. The GBA has the advantages of a medium power required t o reach high temperatures, a furnace-type environment during the various heating steps, and a low cost (
