Determination of Submicrogram Quantities of Mercury in Lake Waters Yiu-Kee Chau and Hirohumi Saitoh Canada Centre for Inland Waters, Burlington, Ont., Canada
Extraction of' Mercury A simple and sensitive method for the determination of submicrogram amounts of mercury in lake water was developed by combination of concentration by dithizone extraction and gas-phase atomic absorption. The mercury was first extracted by dithizone, back extracted by hydrochloric acid, and then converted to vapor by a reduction-aeration reaction. Hg, Hg(I), Hg(Il), and some organomercuric compounds were extracted by the present procedure. A sensitivity of 0.008 pg. per liter (0.008 p.p.b.) was achieved for water analysis. Standard deviations of 0.0087 and 0.0042 were found for lake water samples containing 0.478 and 0.048 pg. per liter of Hg, respectively.
M
ercury as a highly toxic trace element to humans and animals is well known. Its compounds are widely used in agriculture as fungicides and preservatives for seeds of wheat, oats, and barley. In industries, mercury compounds find applications in pulp and paper mills, and mercury metal is used extensively in electrolytic processes in the manufacture of chlorine and caustic soda. According to a survey by Fimreite (1969), some 200,000 pounds of mercury are required annually as make-up in Canadian chlorine-alkali plants, and most of the lost mercury probably finds its way to rivers and lakes. Certain species of fish concentrate mercury from the surrounding water. and pike (Esox lucius L.) has been used as an indicator of mercury contamination in the environment (Johnels, Westermark, et al., 1967). Seed-eating birds such as pheasant (Phasianus colchicus L.) have also been found to contain significant amounts of mercury derived from uncovered seed (Tejning, 1967). The mercury pollution in our environment is alarming and has necessitated the development of a sensitive technique for the determination of submicrogram amounts of mercury in lake water and biomaterials. Mercury exists in several forms in natural waters; metallic mercury, mercurous (Hg?*+), and mercuric (Hg*+) compounds are possible. Both ionic species of mercury form a number of complexes with organic and inorganic ligands; mercuric complexes are very stable. There are several methods available for the determination of microgram quantities of mercury (Sandell, 1965), the dithizone method is well known. A very sensitive physical method based on the atomic absorption of mercury vapor has recently been adopted in mercury analysis. The present study describes the combination of concentration by dithizone extraction and the vapor-phase flameless atomic absorption technique to provide a simple and sensitive method for mercury determination in lake waters.
Since the concentration of mercury in natural waters is too low for direct determination, a large volume of sample has to be used and a concentration process is required. Both coprecipitation with copper sulfide (Stock and Cucuel, 1934) and extraction with dithizone (Hamaguchi, Kuroda, et ai., 1961) have been used to concentrate mercury from seawater. The dithizone method has been widely used in the spectrophotometric determination of mercury in a variety of natural materials (Sandell, 1965); its capability of extracting microgram amounts of mercury(I1) from a large volume of solution (1OOX factor) provides a simple technique for the concentration of mercury from water. Dithizone has also been found to react with a number of organomercury(I1) salts to form complexes extractable into the organic phase (Irving and Cox, 1963). Complete extraction has been reported at pH 1 to 5 for the following compounds: methylmercury(I1) iodide, benzylmercury(I1) chloride, phenylmercury(I1) nitrate, ptolylmercury(I1) chloride, pchlorophenylmercury(I1) acetate, and ethylmercury(I1) chloride. The technique of dithizone extraction, as described by Hamaguchi, Kuroda, et al. (1961), was investigated for its efficiency in concentrating mercury from lake water. Aliquots (500 ml.) of membrane (0.5 p ) filtered lake water spiked with *a3HgC1?containing 1 pg. of Hg as a carrier were made 0.1N in sulfuric acid. After addition of 1 ml. of EDTA and 2 ml. of hydroxylamine hydrochloride, the solution was equilibrated with 25 ml. of dithizone in chloroform for 10 min. in a mechanical shaker. The chloroform extract was carefully separated into a 25-ml. volumetric flask, made up to the mark by chloroform, and the percentage of mercury extraction was assayed by radiometry. Quantitative recovery was found in acidity as low as 0.05N. According to Sandell (1965), mercury can also be extracted from as high as 12N sulfuric acid medium with dithizone in carbon tetrachloride. The extraction of mercury metal by dithizone under the same conditions as for Hg(I1) was studied by spiking a 500-ml. aliquot of lake water with a solution of mercury containing ca. 0.1 fig. of Hg (metallic) a n 1 carrying through the entire procedure. The absorption signal was compared with that produced by the same amount of mercury but without the extraction. The yield of this experiment (100 + 1) indicated that mercury metal was also quantitatively extracted. The metallic mercury solution was prepared by shaking a few drops of mercury with 100 ml. of distilled water; water saturated with mercury contains ca. 60 pg. of Hg per liter at 25" C. (Moser and Voigt, 1957). Similarly, the extraction of organomercuric compound was tested by adding a solution of phenylmercuric chloride (0.3 pg. of Hg) into 500 ml. of lake water. The spiked water was left for two days before the extraction was carried out. The control was prepared by adding the same amount of Volume 4, Nu nber 10, October 1970
839
Figure 1. Relationship of acidity and back extraction of mercury from dithizone chloroform solution phenylmercuric chloride to 40 ml. of 2.5N hydrochloric acid and following the reduction-aeration procedure. The mercury was quantitatively recovered. Mercurous compounds undergo a disproportionation reaction to mercuric and mercury. Since the mercuric ion forms more stable complexes with most ligands in water, the existence of the mercurous ion is restricted. In the extraction process, where the mercuric ions and mercury are removed by dithizone, the tendency for mercurous ions to disproportionate is increased, and as a result, mercurous species will eventually be removed from solution. The present extraction technique thus removes mercury, mercurous ion, and mercuric ion, as well as some organomercuric compounds.
Back Extraction of Mercury from Ditliizone
To determine the mercury by atomic absorption, the mercury has to be converted to vapor phase by a reductionaeration process carried out in aqueous medium. Mercury can be transferred from the chloroform phase back to the aqueous szdution using bromide o r iodide as complex formers in an acidic solution (Laug and Nelson, 1942). The presence of bromide, however, was found to prevent the aeration of mercury vapor, probably because of the stability of the bromo-complex formed. Experiments were carried out to investigate the use of hydrochloric acid for back extraction of mercury. The dithizone extract (25 ml.) containing the *03Hg was shaken with 10-ml. aliquots of hydrochloric acid of various strengths. The acid extract was separated and its activity was measured radiometrically. The results of these experiments (Figure 1) showed that back extraction of mercury from dithizonechloroform solution is quantitative with 4 to 8 N hydrochloric acid but falls off rapidly below 3N. Sulfuric acid (6N) was not able to back extract mercury; nitric acid (6N) removed only 94% of mercury from dithizone chloroform solution.
absorption tube of smaller diameter was used to increase the absorption sensitivity. A peristaltic pump was found most satisfactory in avoiding amalgamation or corrosion caused by the mercury vapor. As hydrochloric acid was used for back extraction of mercury, the stannous chloride reduction was carried out in hydrochloric acid medium instead of sulfuric acid. The optimum concentration of hydrochloric acid for the reduction-aeration of mercury was studied. Hydrochloric acid (20 ml.) of different normality was spiked with 0.2 pg. of Hg. After addition of 1 ml. of stannous chloride, the solution was aerated and the absorption intensity was measured. The results in Figure 2 indicated that the optimum acidity was from 1 to 4N. Acid stronger than 4 N would probably cause the formation of chloro-complexes, which hinder the aeration of mercury. As mercury(I1) has a rather high standard reduction potential (0.854 V), it is readily reduced to metal by stannous chloride. The amount of stannous chloride used was found sufficient for the purpose; further increase of the amount of reducing agent did not increase or speed up the aeration. The absorption signals were reproducible and Beer’s law was obeyed up to at least 0.5 pg. of mercury. For higher concentrations, a shorter absorption tube was used or the sample size was reduced.
Interferences The possible interferences of the method are metallic ions extracted by dithizone, back extracted by hydrochloric acid into the aqueous phase, reduced to metal with tin(I1) ions, and are alloyed with the mercury. In the presence of EDTA and hydroxylamine, some of the metallic ions were prevented from extraction with dithizone. No interference was produced in the determination of 0.3 pg. of Hg in the presence of at least 100 pg. of each of the following ions spiked in 500-ml. aliquots of lake water: Cd, Cu(II), Co(II), Ni(II), Pb, Zn, Fe(II), Fe(III), Pt(IV), Mo(VI), and W(V1). Gold(II1) caused interferences at the 100-pg. level. When it was reduced to 50 p g . , no interference was observed.
Determination of Mercury in Lake Waters Place 500 ml. of the membrane (0.5 p ) filtered and acidified sample (see sample storage) in a cylindrical separatory funnel, add 2 ml. of hydroxylamine hydrochloride and 1 ml. of EDTA. After addition of 25 ml. of dithizone in chloroform, shake the mixture for 10 min. Let the mixture stand for 20 min., then separate the chloroform layer carefully into a 60-ml. separa-
z
50. 40
-
Q
0.
-
5
m 30
Determination of Mercury by Atomic Absorption The extremely sensitive method of mercury determination by flameless atomic absorption of its vapor has been reported by several workers (Dill, 1967; Hatch and Ott, 1968; Ling, 1968). As the mercury in the sample is totally converted to vapor in an absorption tube, the sensitivity attained is far beyond that of the conventional flame technique where only a small fraction of the atomized sample is present in the optical path. The closed-system aeration technique of Hatch and Ott (1968) was modified to increase the sensitivity. A longer 840 Environmental Science & Technology
8 20
-
10
I 1
2
3
4 5 NOFMALITY
6
7
8
9
9
OF h C I
Figure 2. Relationship of reduction-aeration of mercury with acid strength
tory funnel. Back extract the mercury by shaking the chloroform solution with 20 ml. of 5N hydrochloric acid for 10 rnin. Transfer the acid extract to the aeration chamber and rinse the funnel with 20 ml. of distilled water. After addition of 1 ml. of stannous chloride, connect the aeration to the absorption tube and to the pump to form a closed circuit. Turn on the pump to start the aeration; until the signal has reached a plateau of maximum, open the system by letting off the mercury vapor into an exhaust hood. Continue the aeration for a while until the mercury is driven off as indicated by the return of signal to its minimum point. Run a reagent blank with distilled water. As both the extraction and back extraction are quantitative, calibrate the method by simply adding 0.1 pg. and 0.3 pg., respectively, to the aeration chamber: add 40 ml. of 2.5N hydrochloric acid, 1 ml. of stannous chloride. and turn o n the pump to obtain the corresponding signals to produce a calibration curve.
Experimental
Apparatus. A Jarrell-Ash atomic absorption spectrophotometer (Model 820-528) was used for the investigation. Any other atomic absorption instrument with an open space betueen the source and the monochromator for mounting the absorption tube may be used. The radiant source (2536.5 A) was a high-intensity mercury hollow cathode lamp by Westinghouse. The signal was presented on a ColemanHitachi 165 recorder without the use of scale expansion. Absorption tube, 22-mm., 0.d. X 24 cm., made from borosilicate tubing with quartz windows and inlet and outlet ports. Tubing pump, peristaltic type (Cole-Parmer Instrument Co., Model 7206-11) with 1/8-inchi.d. X 3/s-inch 0.d. Tygon tubing. Circulation rate is 1.5 liters per min. Aeration chamber, made from an ordinary gas washing bottle with a fritted cylinder (Fisher catalog no. 3040). The bottle was shortened to ca. 85-ml. capacity. The aeration was a closed system connected by Tygon tubing. A magnesium perchlorate drying tube was inserted in the line before the absorption tube to remove the moisture. A Nuclear-Chicago gamma counting system (Model 4454) with a thallium activated NaI well (21/32-inchdiameter X I ‘/?-inch depth) was used to measure the y-activity of 203Hg. Reagents. Glassware was rinsed with nitric acid and water before use. All the usual precautions for working with dithizone were observed. Dithizone. Dissolve 6 mg. of clithizone in 1 liter of distilled chloroform. Mercury contamination of this reagent can be checked by extracting an aliquoi of this solution with 20 ml. of 5N hydrochloric acid and determining its mercury content. Generally A.R. grade reagent can be used without purification. It can be conveniently purified by extracting the dithizone chloroform solution with 4N hydrochloric acid before use. Stannous chloride, 2 0 z in 6N hydrochloric acid. Hydrochloric acid, approximately 6 N acid was prepared by constant boiling distillation. EDTA, 0.1 M solution of disodium salt of ethylenediaminetetraacetic acid. Hydroxylamine hydrochloride, 50 in water. Standard mercury solution. Dissolve 0.134 g. of mercuric chloride in 100 ml. 6 N hydrochloric acid (1000 pg. per ml.). A substandard (100 pg. per ml.) is prepared by dilution with distilled water. The dilute standard is stable for a period of at least one month. Daily working standard (0.1 pg. per ml.) is prepared by dilution with 0.2N hydrochloric acid. “13Hg isotope. obtained from the Radiochemical Centre,
Table I. Recovery of Mercury from Lake Water CaliReSample Hg added Absorb- Hg found bration covery (500 ml.) (pg.) ance“ (pg.) curve (2) Distilled water 0 0 0 Lake water 0 0.027 0.027 Lake water 0.05 0.075 0.074 0.048 94 Lakewater 0.10 0.135 0.130 0,093 103 0.219 0.201 96 Lakewater 0.20 0.221 Lake water 0.30 0,322 0.318 0.305 97 Lake water 0.40 0.407 0.403 0.409 94 Lake water 0.50 0.523 0.517 0.509 98 Net absorbance after subtraction of blank; absorbance of blank 0.013.
Amersham, England. A working solution containing 0.1 p C. per ml. with 0.1 pg. H g carrier was prepared by dilution with 0.2N HCI.
Results and Discussion
Accuracy and Precision. The accuracy of the method was tested by analyzing several lake water samples that had been spiked with known amounts of mercury. The results of these analyses (Table I) showed that the accuracy of the described method was satisfactory. The precision of the method was evaluated by replicate analyses ( 5 ) of a spiked lake water sample and a lake water sample taken from Lake Ontario near Hamilton Bay. The average mercury contents were 0.478 and 0.018 pg. of H g per liter, with standard deviations of 0.0087 and 0.0042, respectively, at these levels. The sensitivity of the method for lake water analysis computed from the calibration (Table I) was 0.0084 pg. of Hg per liter for a signal of 2 absorption (0.038s absorbance). Storage of Samples. Radiometric investigations indicated that no appreciable loss of mercury activity occurred in water samples stored for a period of at least two weeks in borosilicate glass, polyethylene, and polypropylene containers if the samples were membrane (0.5 p ) filtered and acidified to 0.1N (pH 1 ) with sulfuric acid. Losses of 35, 82, and 65 respectively, were observed after one week in containers of borosilicate glass. polyethylene, and polypropylene when the samples were not acidified.
z
z
Literature Cited Dill, M. S., Report No. Y-1572, Union Carbide Corp., Oak Ridge, Tenn., 1967. Fimreite, N., “Mercury uses in Canada and their possible hazards as sources of mercury contamination,” Canadian Wildlife Service, Rept. No. 17 (1969). Hamaguchi, H., Kurada, R., Hosohara, K . , J . Chem. Soc., Jap. 82, 374 (1961). Hatch, W. R., Ott, W. L., Anal. Chem. 40,2085 (1968). Irving, H., Cox, J. J., J. Chem. SOC.(London) 466 (1963). Johnels, A. G . , Westermark, T., Berg, W., Person, P. I., Sjostrand, B., Oikos 18, 323 (1967). Laug, E. P., Nelssn, K. W., J . Ass. Ojjicial Agr. Chem. 35, 399 (1912). Ling, C., Anal. Chem. 40, 1877 (1968). Moser, H. C., Voigt, A. F., J . Atner. Cheni. Soc. 79, 1837 (1957). Sandell, E. B., “Colorimetric determination of traces of metals,” Interscience, New York (1965), p. 621-639. Stock, A., Cucuel, F., Naturwissenschafren 22, 390 (1934). Tejning, S., Oikos 18, 334 (1967). Receiced for reciew January 9, 1970. Accepred M u y 12, 1970. Volume 4, Number 10, October 1970 841
