Mercury Contamination of Sea Water Samples Stored in Polyethylene Containers Michael H. Bothner' Department of Oceanography, University of Washington, Seattle, Wash
D. E. Robertson Battelle-Northwest Laboratory, Richland, Wash.
If it is not possible to begin analysis of water samples for mercury immediately after collection, considerable care must be taken to avoid changes in mercury concentrations because of storage conditions. Several studies in the literature have shown that acidification of water samples is necessary to prevent rapid adsorption onto container walls or volatilization and subsequent loss from solution (1-4). If seawater samples are not acidified, both ionic mercury and methylmercury chloride are completely adsorbed o n t o polyethylene container surfaces after several weeks ( 5 ) . Some investigators have recommended that strong oxidizing agents should also be added (6). In this study, polyethylene bottles, commonly used in oceanographic and other sampling because of their durability and relatively low trace element impurity, were compared to Pyrex (borosilicate) glass bottles for suitability as storage containers for solutions analyzed for mercury. Acidified samples of seawater and distilled water with no mercury added were stored in both containers. The samples were analyzed for total mercury periodically over a few months, a storage period which is commonly necessary for deep-sea samples when analysis on board ship is not possible. In contrast to studies in the literature showing losses of mercury from dilute solutions stored in polyethylene under similar conditions (2, 3, 6), this study showed that natural water samples could, under certain conditions, increase in mercury with time because of contamination. No appreciable changes in mercury levels were observed in identical solutions stored in Pyrex glass.
EXPERIMENTAL The polyethylene bottles (with polyethylene screw caps) used for the storage experiment were precleaned by rinsing with either 6N "03 or 6N HC1 for about two minutes and subsequently rinsing three times with distilled water known to be low in mercury. The glass-stoppered Pyrex flasks used in parallel storage experiments and all other glassware were precleaned by filling with distilled water and employing the oxidizing procedure described below for digesting water samples. The flasks were finally rinsed three times with distilled water. This precleaning step is particularly important with new glassware. Polyethylene bottles could not be cleaned in the same rigorous manner as the glassware without seriously altering the polyethylene surfaces. Seawater samples (salinity 30%0) for the storage experiments were collected from Puget Sound with polyvinyl chloride Niskin type sampling bottles. These sampling bottles were found neither to add mercury to unspiked distilled water nor to subtract mercury from water spiked to about 0.05 ,ug(Hg2+)/1,in tests lasting 45 minutes, a maximum exposure time for samples collected in shallow water areas. A large sample of seawater was acidified to pH 1.5 with HCl, well mixed, and then transferred to polyethylene bottles and Pyrex flasks for the storage experiments. Water was analyzed for mercury by flameless atomic adsorption on a Mercury Monitor manufactured by Laboratory Data Control, Riviera Beach, Fla. The wet chemical oxidation of water samples Present address, Office of Marine Geology, U S . Geological Survey, Woods Hole, Mass. 02543. 592
ANALYTICAL CHEMISTRY, VOL. 47, NO. 3, MARCH 1975
was slightly modified from that of the U S . Environmental Protection Agency (7). One hundred ml of water was added from polyethylene and Pyrex storage containers to 250-ml flat-bottom Pyrex reaction flasks with ground glass stoppers. The following oxidizing 2 ml of 5% by agents were added: 2.5 ml concentrated " 0 3 , weight KMn04, and 2 ml of 5% KzSzOs (all Baker reagent grade). As much air as possible was removed from the sample flask by placing the flask under hot (50-60 "C) water and allowing the contained air to expand. The bottles were then lightly stoppered with ground glass stoppers and placed in a water bath for 2 hours at 80 "C. The preheating technique minimizes the occurrence of stoppers being blown off the sample bottles during heating. After heating, the samples were cooled to room temperature. Carefully avoiding the formation of bubbles, two ml of 12% hydroxylamine hydrochloride was added to reduce dissolved Clp and MnOz particulates in the solution. The Clz gas formed during the oxidation of seawater strongly absorbs at the 253.7-nm mercury wavelength. Chlorine present in the air above the solution was removed by flushing the air out of the flask with nitrogen or air without generating bubbles in the sample solution. Finally, 10 ml of 10% SnC12 in 3.6N HpS04 was added to reduce the mercury to elemental form and the sample flask was immediately connected to the aeration apparatus. The sample was then swirled vigorously and purged with nitrogen bubbles generated with a gas dispersion tube at a flow rate of 700 ml per minute. The concentration of mercury was determined by comparing the peak height of samples with peak heights obtained from samples containing known additions of mercury. Blank corrections were made by measuring the increment of additional mercury from double and triple amounts of reagents. Reagents contributed 0.2 ng mercury (0.002 ppb) to the sample. T o give an indication of C12 and other interferences, a mercury trap consisting of granulated silver in Tygon tubing (0.6-cm diameter X 5 cm long) was inserted in the gas line between the sample flask and the optical cell. This trap removed mercury vapor from the gas leaving the sample solution. Since the silver did not remove Clz, the trap was useful in analysis of test seawater samples to determine if Clp was still present in the samples. This trap is also useful for determining the presence of other interfering substances, such as organic vapors, provided they are not also removed from the carrier gas by the silver metal. No other interferences were identified in this manner in analysis of seawater containing effluent from both a sewage treatment plant and a pulp mill, or samples of interstitial water from sediments. Coefficients of variation, based on 6 replicate measurements of two seawater samples, were 1.5 and 14%,respectively, a t mercury concentrations corresponding to 0.13 and 0.009 pg/l.
Table I . FWQA Reference Solution Mercury-1970, Analysis by Flameless Atomic Absorption, October 1971 FbCA
\
alue,
Sample 20.
Chemical ionnula
ug l a
Thia s c u d \ , u q 1
1 2 3 4
HgC1, HgC1, HgC12,CHsHgCl CH3HgC1
0.34 4.2 6.3 4.2
0.33 i: 0.02* 4.2 0.1 6.5 i 0.1 4.4 0.1
*
a FWQA standards were diluted to the range of 0 02-0 10 pg/l for analysis Error is one standard deviation based on triplicate analyses
O14r
f
Table 11. Comparison of Mercury Concentrations in Sea Water by Different Analytical Techniques Values in ppb Sample No.
1 2
Thionalid extractionneutron activatlon
0.05 i 0.004"
*
Flameless AA
0.048
i
0.005b
0.12 i 0.01 Error based on counting statistics. * One standard deviation based on replicate analyses. These concentrations are several times higher than usually found in Puget Sound water because of storage in polyethylene bottles.
0.13
0.01
Q
-
Checks on the accuracy of the method were made by analyzing EPA reference solutions and by analyzing seawater solutions by two independent methods. Results on the reference solutions (Table I) indicate agreement within analytical uncertainty between suggested and measured values. Two seawater samples were analyzed by neutron activation in addition to flameless atomic absorption. The technique (8, 9 ) involves coprecipitation of mercury in seawater with thionalid a t pH 3, irradiating the precipitate and measuring the mercury-203 photopeak. Based on radiotracer experiments, the thionalid removed 100 f 3% of the mercury in solution. No interferences from '5Se or a39Np were observed. The results of the two methods are within experimental uncertainty (Table 11).These values are higher than typical values of Puget Sound water (0.01-0.02 pg/l.) (IO), because the samples were contaminated during storage.
JL - 0
c
-0
4c 50 80 STORAGE T I M E DAYS
CO
245
Flgure 1. Change in mercury concentration during storage of seawater at room temperature, acidified to pH x.5 with HCI
Each point represents mean value of 3 bottles, maximum standard deviation about the mean was 8 % . ( 0 )Bel-Art polyethylene bottles, 3 I. ( 0 )Pyrex wlth ground-glass stoppers 250 ml
RESULTS AND DISCUSSION The temporal changes in mercury concentrations in seawater stored in polyethylene and Pyrex glass bottles is seen in Figure 1. In the case of Bel-Art (Pequannock, N.J.) polyethylene bottles, an increase from 0.01 to 0.14 wg/l. was observed in 245 days for bottles stored on an open bench top, while no change was observed in Pyrex glass during the same period. In a later experiment, a similar increase was observed in Monsanto Co. (St. Louis, Mo.) polyethylene bottles compared to Pyrex flasks (Figure 2). Samples of acidified, distilled water in Monsanto polyethylene bottles under otherwise identical conditions also increased in mercury concentration from 0.007 to 0.044 gg/l. in 60 days (not illustrated). Based on periodic tests with the silver powder trap discussed above, the increased signal from solutions with time is due to increases in mercury and not due to chlorine or organic interferences. Less rigorous experiments using Monsanto polyethylene bottles with seawater acidified with HNO3 to pH 0.5 at room temperature gave a similar increase in mercury with time as observed with HCI a t room temperature indicating that the kind of acid used is probably not important. These results differ from a recent study ( 6 ) which showed an 80% decrease of mercury in 10 days from distilled water solutions spiked to 0.2 pg/l. mercury, acidified with "03 (1%v/v), and stored in borosilicate glass and polyethylene bottles. Substantial losses were also observed under stronger acidification (5% v/v). Part of the difference could be due to the different starting solutions used in the two studies. The results of the present study of natural mercury in seawater with increased chloride concentration from added HCl may suggest that a chloro-mercury complex (HgC14-2) or a mercury complex with organics in seawater inhibits sorption by container walls. A difference in the brand of containers used in the earlier study may also account for the different results. The increase of mercury in polyethylene observed in this study could be a result of slow leaching of mercury from the container walls by the acid solution. This explanation is suggested although not confirmed by the finding that samples digested for 2 hours in Monsanto polyethylene bottles,
0
I0
20 30 40 STORAGE T I M E D I Y S
50
6C
Figure 2. Change in mercury concentration during storage of seawater at room temperature, acidified to pH 1.5 with HCI
( 0 )Monsanto polyethylene bottles, 1 I. (0)Pyrex bottles with ground-glass stoppers, 250'ml
using the hot oxidizing solution described above, increased in mercury concentration to levels higher than the observed increase during the long-term storage experiment. Another mechanism which can account for contamination of acidified seawater samples stored in polyethylene is diffusion of mercury vapor from the ambient air through polyethylene container walls resulting in an accumulation of mercury in solution. This contamination mechanism was suspected following an interlaboratory storage experiment, in which three acid-rinsed Bel-Art bottles were filled at each laboratory with pH 1.6 seawater of known, low mercury concentrations. The mercury concentrations in the bottles were monitored as a function of time a t each laboratory using identical analytical techniques. It was observed that the seawater in the 3 bottles placed on a laboratory bench at the University of Washington gradually increased in mercury concentrations as shown in Figure 3. However, the seawater in the 3 bottles placed on the floor in a corner of a laboratory at Battelle-Northwest increased in mercury concentrations at a 40-fold faster rate than the samples a t the University of Washington (see Figure 3). Since the bottles a t each laboratory were from the same manufactured batch and were identically precleaned, these results suggested that perhaps the floor in the laboratory a t Battelle-Northwest was contaminated with mercury, and that mercury vapor was emanating from the floor and diffusing through the Bel-Art bottles stored on the floor. The floor was subsequently checked for mercury contamination using a Bacharach Instrument Company Model MV-2 Mercury Sniffer, which has a sensitivity of 0.01 mg/ m3 of air. Floor-level air concentrations were less than 0.01 ANALYTICAL CHEMISTRY, VOL. 47, NO. 3, MARCH 1975
593
/
/
r
1
I
tI
/ /
/
r
/
4
/
/ / /
/
I
U
a
0.00I STORAGE TIME DAYS
Increase in mercury concentrations in seawater stored in identical Bel-Art polyethylene bottles placed on an open bench top (-) and on a laboratory floor (- - -) Figure 3.
Error bars represent one standard deviation for analyses of four bottles
mg/m3, but when the floor was smeared in the corner near the baseboards with tissue paper, detectable mercury vapor was found to emanate from the tissue, and a reading of approximately 0.08 mg/m3 was obtained when the tissue was placed near the air intake of the Mercury Sniffer. The laboratory floor is surfaced with 12-inch square vinyl asbestos tile, and apparently some mercury, spilled during an earlier occupation of the laboratory, had accumulated in the tile cracks and under the baseboard. Mercury vapor from this source although undetectable by the Sniffer is thought to account for the rapid rate of mercury increase in seawater samples stored on the floor. The contaminated floor, however, did not interfere with ultra low concentrations of mercury in seawater analyzed a t bench top levels in the same laboratory. T o prove that mercury vapor can rapidly diffuse through the walls of polyethylene bottles containing acidified seawater, the following experiment was performed. A 70-mg drop of metallic mercury was placed in an open vial, and the vial was placed on the bottom of a 20-liter, heavy walled polyethylene bottle. A 2-liter, precleaned Bel-Art polyethylene bottle was filled with p H 1.6 seawater having a mercury concentration of 0.002 pg/l. The Bel-Art bottle was tightly capped and placed on a small stand a t the bottom of the 20-liter bottle approximately 15 cm away from the vial containing the mercury droplet. The 20-liter bottle was tightly closed to seal in mercury vapor released from the droplet of metallic mercury. If complete equilibrium was established, the air concentration of mercury within the 20-liter bottle would be about 20 mg/m3, or about 0.4 mg of mercury in 20 liters of air. After 5 days, the Bel-Art bottle containing the acidified seawater was removed from the sealed 20-liter bottle. The outside surface of the BelArt bottle was washed with distilled water and wiped dry to avoid cross-contamination from surface-adsorbed mercury. The mercury content of the seawater was then measured and found to be greater than 2 pg/l., an increase of over 1000-fold. Thus, diffusion of airborne mercury vapor through polyethylene container walls can create a serious contamination problem if the circulating ambient air contains significant quantities of mercury. 594
0.001
0
10
20 STORAGE
30
40
50
T I M E DAYS
Changes in concentration of mercury in seawater stored in identical Bel-Art polyethylene bottles placed in a sealed plywood box (-) and on a laboratory floor (- - -) Figure 4.
T o determine if Bel-Art bottles could be used for storing p H 1.6 seawater if they were isolated from mercury-contaminated air, the following test was performed. Eight identical acid-rinsed Bel-Art bottles were filled with p H 1.6 seawater containing 0.004 pg Hghiter. Four of the bottles were encased in a sealed plywood box and four were again placed on the contaminated laboratory floor. After storage periods of 18 and 40 days, the mercury concentrations in the bottles were determined (see Figure 4). The mercury levels in seawater in the bottles resting on the floor again increased a t about the same rate as previously observed and, after 40 days, the mercury concentrations rose to levels between 0.19 and 0.24 pg/l. However, the seawater in the bottles stored in the plywood box did not change from a mercury concentration of 0.004 pg/l. after storage for 40 days. Thus, in the case of Bel-Art bottles, leaching of mercury from the bottle walls does not appear to be a contaminating mechanism, and the bottles appear to be suitable containers for storing acidified seawater if the bottles are isolated from large volumes of circulating, mercury-contaminated air. Such protection can be afforded by encasing the bottles in sealed plywood boxes or in other suitable secondary storage containers or cabinets.
CONCLUSIONS Seawater and distilled water samples acidified to pH 1.5 with HCl and stored a t room temperature in po1yethyler.e bottles have been shown to increase in mercury concentration with time under certain conditions. Identical samples contained in ground glass-stoppered Pyrex flasks had unchanging or slightly decreasing mercury levels during the same period. The contamination of samples in polyethylene containers may come from either leaching of mercury from container surfaces or from passage of mercury vapor
ANALYTICAL CHEMISTRY, VOL. 47, NO. 3, M A R C H 1975
from the ambient air through the container wall into the solution or from both sources. It should be pointed out that polyethylene made by other manufacturers or from different lots from the same manufacturers may not contaminate water samples as observed in this study. Other procedures such as enclosure of polyethylene bottles in tightly constructed plywood boxes can eliminate contamination in some cases. The present data suggest, however, that checks which can determine the degree of sorption, leaching, and diffusion reactions under proposed storage conditions should be made on all containers used to store solutions for mercury analysis. One of the major advantages of the flameless atomic absorption technique described is that it can be readily used on board oceanographic research vessels, thus facilitating analysis of water samples as they are collected and avoiding the problems encountered during storage. However, if analysis of natural water samples for mercury cannot be carried out immediately after collection, it appears that storage of acidified samples in sealed Pyrex glass bottles or flasks, carefully precleaned and tested as described, is the most desirable storage technique.
LITERATURE CITED (1)E. D. S. Corner and F. H. Rigier, J. Mar. Biol. Ass. U.K., 36, 449 (1957). (2)M. R. Greenwood and T. W. Clarkson, Amer. lnd. Hyg. Ass. J., 31, 250 (1970). (3)L. C. Bate, Radiochem. Radioanal. Len., 6, 139 (1971). (4) D. E. Robertson, Anal. Chim. Acta, 42,533 (1968). (5)D. E. Robertson, Batelle Pacific Northwest Laboratories, Richland, Washington, Unpublished Work, 1974. (6)C. Feldman. Anal. Chem.,46, 99 (1974). (7) U.S. Environmental Protection Agency, "Methods for Chemical Analysis of Water and Wastes," National Environmental Research Center, Cincinnati, Ohio 16020, 121 (1972). (8)M. G.Lai and H. V. Weiss. Anal. Chem.,34, 1012 (1962). (9)D. E. Robertson and R. Carpenter, "Neutron Activation Techniques for the Measurement of Trace Metals in Environmental Samples." National Academy of Sciences, National Research Council, NAS-NS-9114, USAEC, January 1974. (10)M. H. Bothner, "Mercury: Some Aspects of Its Marine Geochemistry in Puget Sound, Washington," Ph.D. Thesis, University of Washington, Seattle, Wash.. 1973.
RECEIVEDfor review May 10, 1974. Accepted November 25, 1974. Contribution No. 791 from the Department of Oceanography, University of Washington. This research was supported by National Science Foundation Grant Number GI 33325X.
Determination of Chromium(V1) in Industrial Atmospheres by a Catalytic Method B. M. Kneebone and H. Freiser Department of Chemistry, University of Arizona, Tucson, Ariz. 8572 1
There is need for sensitive, selective methods for Cr(V1) analysis to be able to determine compliance with the Occupational Safety and Health Act (OSHA) in such areas as chrome plating plants and in the paint and dye industry. The present OSHA ceiling limit is 0.1 mg CrOJm3 of air which corresponds to 0.04 wg/ml when 40 1. of air are collected in 10.0 ml of solution ( I ) . Current methods of analysis usually utilize the specific reaction of Cr(V1) with 1,5-diphenylcarbohydrazide (2) after the samples have been collected in NaOH or water ( 3 ) . Dean and Beverly ( 4 ) found that vanadium(V) and iron(II1) interfere with this method, so they modified it by extracting chromium(V1) from HC1 into MIBK. Other methods include the hematoxylin reaction (3) and the oxidation of triiodide and subsequent titration of iodine with thiosulfate ( 3 ) . The sensitivities of these methods vary from about 0.06-3.0 wg/ml. In view of the present trend to tighten pollution controls, the detection limit could be lowered so that it would fall outside the working range of these methods which are already at their threshold sensitivity. Such levels are ideally suited to analysis by kinetic methods using catalytic reactions. Yatsimirskii ( 5 ) summarized the essentials of such catalytic methods and listed more than 50 of them, including four for chromium, none of which had the required specificity and sensitivity for use in air analysis. Hadjioannou (6) described the determination of submicrogram quantitites of Cr(V1) by monitoring its catalytic effect on the oxidation of iodide by hydrogen peroxide. As a starting point in this study, the report by Dolmanova et al. ( 7 ) of the microdetermination of Cr(V1) by means of the catalytic oxidation of 0-dianisidine by hydrogen peroxide was investigated. A significant improvement was
made in the precision of the method by changing reagent concentrations, and some interesting disparities were found between this work and the published results. A procedure was developed and successfully applied to filter samples furnished by NIOSH. Chromium(V1) was determined on filters which contained between 0-30 wg with an accuracy of 3~6%.The sensitivity of this method is 0.001 wg Cr(V1) and the precision ranges from f10% for the blank to f l - 3 % for Cr(V1) above 0.10 wg/ml and within 6% for the lower concentrations.
EXPERIMENTAL Apparatus. A Gilford 2400 spectrophotometer, equipped with a Honeywell recorder and a circulating constant temperature water bath to thermostat the cell compartment, was used for all measurements. Cells had a 1.0-cm path length. Reagents. 3,3-Dimethoxyhenzidine ( 0 - dianisidine), practical grade, was obtained from Eastman Kodak (caution: toxic, suspected carcinogen). It was not purified before use. Stock solution (3.60 X 10-2M): 0.879 g dissolved in 100 ml of absolute ethanol. This solution is stable for one week when protected from light. Dowex 50W-X8 was obtained from J. T. Baker. Primary standard KZCr20; was obtained from Matheson Coleman and Bell. Stock solution containing 100 pg/ml: 0.283 g/l. All water used was deionized and redistilled in Pyrex glassware. All other reagents were ACS Analytical Reagent grade. Glassware was thoroughly cleaned with concd HC1, then rinsed several times with redistilled water and oven-dried. Study of Reaction Rate Parameters. Dependence on [Cr(V I ) ] .In a 10.0-ml volumetric flask were mixed 4.0 ml of absolute ethanol, 1.0 ml of 2.4 X 10-2M o-dianisidine, 1.0 ml of 3.8M H202, amounts of Cr(V1) as KzCr2O.i ranging from 0 to 0.4 pg and 10-3M HC1 to the mark. T h e solutions were thoroughly mixed, and the reaction rate was monitored by following the change in absorbance at 450 nrn as a function of time for 15 min. The cell compartment was thermo-
ANALYTICAL CHEMISTRY, VOL. 47, NO.
3, MARCH 1975
595