Environ. Sci. Technol. 1984, 18, 127-130
Automated Rain Sampler for Trace Organic Substances William M. J. Strachan* and Henri Huneault Environmental Contaminants Division, National Water Research Institute, Burlington, Ontario, L7R 4A6 Canada
Introduction
centrated in order to permit their analytical determination. In this laboratory, we have previously employed large (2 and 0.36 m2) stainless steel collectors but have often observed fine particulate matter adhering to the metal surface. This is of considerable concern since preliminary observations (16) indicated that much of the organic wetfall loading of these chemicals was via the particulate matter. It is also desirable that wetfall samplers be capable of being left unattended in the field for considerable periods of time without affecting the integrity of the sample. Codistillation plus biological and abiotic transformations makes samples stored in the aqueous phase suspect. These factors, and the need to distinguish wetfall from dryfall, led to the development of the sampler described in this report.
Atmospheric deposition of persistent organic chemicals has been identified as a major pathway for the entry of such chemicals to the Great Lakes ecosystem (1,2). Such compounds have also been reported in precipitation elsewhere although attempts to calculate the significance of this route have seldom been undertaken. Locations for earlier studies of such deposition included Hawaii (3), Norway (4), Tokyo (5), and the Caribbean (6) among others and are illustrative of the global aspects of this depositional pathway. Concentrations in various samples of biota from sites not expected to be subject to inputs from other mechanisms also emphasize the widespread nature of this problem (7-10). Compounds such as PCBs, hexachlorobenzenes (HCB), DDT, and isomers of benzene hexachloride (BHC) have all been noted, with none having vapor pressures obviously indicative of their likely presence in atmosphere. Further, PCB’s and DDT have had restricted use or even been banned from North American usage since 1971. These facts would suggest that they should no longer be present in the atmosphere 10 years later since sorption to soil and sediment might be expected to effectively remove them from circulation. That they are still found in significant amounts (10-12) indicates a need to precisely determine their levels and those of other such chemicals in order to provide information necessary to locate sources and evaluate trends. “Precipitation” can be loosely categorized as wetfall and dryfall. Theoretical models are available for vapor-phase transfer using equilibrium constants and transfer coefficients (13,14). For particulate fluxes, only crude estimates have been made (I),and the concentrations of persistent organic substances in or on this material are equally unknown. The nature of dryfall samplers is a major difficulty since most designs to obtain the quantities needed depend upon a flow of air across a filter which may result in loss of sorbed contaminant from the particulates. While this type of sampling is of considerable interest, the topic of this report is wetfall sampling. Wetfall deposition of chemicals, which may be either adsorbed on particulates or dissolved in the rainfall, has been reviewed for metals and inorganic substances (15). For organic compounds, however, it is only in the last few years that concentrations in rainfall have been determined in the Great Lakes region (10-12). Even then it is seldom that substances other than the PCB’s are investigated. Large samples must be collected and the substances con-
Description of Sampler In the Great Lakes region, and in other areas of the northern temperate zone, monthly average rainfalls of 5-10 cm may be expected, and individual events of 1 cm are not infrequent. It was desired to develop a sampler which could be used for sampling both types of situations (monthly and event), and a surface area of approximately 0.2 m2 was therefore indicated in order to acquire samples of at least 2 L. This volume is necessary in order to reach the quantification limits of 0.5-5.0 ng of compound/sample given the concentrations observed in the area (16). The material chosen for the sampler was Teflon-coated stainless steel. Uncovered metals are potentially problematic because charged particles and bare metal surfaces have been observed to result in strong retention of particulates under dryfall conditions. If this were to occur during wetfall, it would underestimate actual concentrations. Teflon is an inert material and can be readily coated on metal surfaces; for these reasons it was the material of choice for both the funnel surface and the column. The sampler was constructed by M.I.C. Co. (Thornhill, Ontario, Canada) to a design and specifications prepared by the authors. The final prototype is shown in Figure 1. It consists of a square funnel (0.46 m X 0.46 m) and a lid, both of which are constructed of stainless steel. A funnel lip of 10 cm is used, and the funnel surface is sloped 20° toward the center in order to provide rapid drainage to a column attached below. All surfaces in contact with the sample are coated with 0.076 cm (30 mil) of Teflon. The legs and chassis are constructed of 0.318-cm (1/8 in.) cast aluminum and are detachable for transport. Rain splash from horizontal surfaces into the sampler is reduced to unobservable levels by the use of stainless steel screening (0.014 gauge, 24 X 24 mesh). Total weight of the sampler is 47 kg. The overall dimensions of the closed sampler are 0.5 m (width) X 1 m (length) X 1.2 (height); when open the lid adds an additional 0.5 m to the length. The lid is automated to be open only during rainfall conditions. The rain sensor is set on an arm slightly above and 0.5 m away from any part of the sampler. It has two sensing faces set at 20° from the horizontal which are slightly heated to ensure response only during rainfall. The control electronics are modified from the Sangamo inorganic rain sampler and operate a 1/50-hpfractional gearmotor at 120 VAC. A small heater in the central electrical
w An automated rain sampler was designed, built, and evaluated as a collecting device for persistent organic chemicals in rain. It consists of a large Teflon-coated funnel (0.209 m2) with an automated lid and a Teflon column containing XAD-2 or -7 resin. The sampler was tested for recoveries and field precision with a number of organochlorinesubstances and PCB’s frequently found at the nanogram per liter level in rain. Recoveries averaged 86%,and the mean coefficient of variation for a-benzene hexachloride, lindane, and polychlorinated biphenyl was 33% under field conditions where samples of 0.4-22 L of rain were collected.
0013-936X184/0918-0127$01.50/0
0 1984 American Chemical Society
Environ. Sci. Technol., Vol. 18, No. 2, 1984
127
Table I. Recoveries with XAD-2 and XAD-I R e a i ~ coefficients test concn, mean, VEX, rg/L rdL % reeoverY,% XAD-2 Resin
I
a-BHC
heptachlor aldrin
?.-chlordan a-chlordan dieldrin endrin P,P’-DDT
0.29 0.42 0.22 0.55 0.55 0.22 0.65 0.41
XAD-7 0.12 lindane(7-BHC) 0.25 heptachlor 0.12 aldrin 0.067 ychlordan 0.71 a-chlordan 0.48 endosulfan 0.42 p,p’-DDE 0.54 p,p’-TDE 0.41 p,p‘-DDT 0.36 methoxychlor 0.51
HCB
noUr 1. W n i c auhmnted rain sllmplsr (lid haw opar).
chassis avoids condensation which otherwise occurred. The outlet of the funnel is threaded to receive an allTeflon column (2 cm i.d. X 25 em) fitted with a support disk a t the bottom. The outlet of the column is equipped with a glass U-tube to ensure that the resin remains wet. Under normal use, a head of at least 10 em is maintained. Flow through the column under these conditions is 2&50 mL/min. The XAD resins used in the column were successively Soxhlet extracted for 3 h with each of acetone and methanol; storage was in methanol. The resin was added to the column as a methanol slurry. Fired glass wool plugs are required above and below the 30-40 mL resin bed which was washed with 5-10 bed volumes of organic-free water obtained by passing glass-redistilled water through a freshly prepared and washed XAD-2 resin column. While it was not done for this study, the columns can be detached and capped, thus providing a suitable conveyance for mailing the resin, they can also be provided to field personnel in a prewashed condition. Experimental Section k v e r i e s . A stock solution of approximately0.14.5 m g / d ench of a number of organochloriueswa8 made up in acetone. Due to the unavailability of some of the subStanceS a t the two recovery test times, the solutions for evaluating XAD-2 and XAD-7 were somewhat different. The two recovery teats, however, both used a comparahle polarity range of compounds, and the results are considered comparable. Spikes (2.0 pL) of these solutions were injected into 2.0-L samples of organic-free water. After 15 min of stirring, three of these samples were slowly poured over the surface of the funnel over a period of 1 h and then through separate columns containing either XAD-2 or XAD-7 resin. The flow rates were approximately 20 mL/min. The resins were removed from the columns and extracted with 50-75 mL of ether, 1 mL of isooctane keeper was added, and the individual extracts were concentrated to 1 mL according to procedures described elsewhere (17). The concentrates were analyzed 118 Envfon. Sci. T M . . Vd. 18. No. 2,
le&(
0.25 0.32 0.16 0.46 0.54 0.20 0.63 0.35
8 8 9 15 17 13 13 25
87 76 78 83 98 88 98 85 av: 86
8 7 90 16 5
63 42
Resin 0.077 0.11
0.029 0.045 0.74 0.34 0.34 0.29 0.29 0.21 0.29
6 6 9 4
7 12
67 103 72 82 53 80 57 58 av: 68
Value not included in average. by chromatography (25-m OV-1 capillary column, programmed from 90 to 280 OC a t 4 OC/min, electron capdetector). Results of these recovery experimenta are presented in Table I for both XAD-2 and XAD-7 resins. Field. Precision for environmental samples was 888e88ed for normal field conditiona Three samplers were employed to collect samples from a field site at 50 Mile Point near Hamilton, Ontaric-a location 10 km downwind from a “heavily industrialized” area. Samples were collected during April and early May of 1981. The samplers were subsequently moved to the Turkey Lakes acid-rain watershed site near Sault Ste.Marie, Ontaricr-a “remote” area 50 km downwind from that city. Collectionsthere d during July and Aug, 1981. A t both locations, a tipping bucket rain gauge (0.2-mm incrementa) was employed to measure millimeters of rainfall, and collected volumes were calculated. Efforta were made at both sitea to ensure that the three samplers and the rainfall gauge were placed in open areas at least 3 m apart and well away from any objecta of equivalent or greater height. Field samples were analyzed as described under Recoveries hut with an additional cleanup step according to established procedures of the Inland Waters Directorate (17). Analytical resulta are given in Table 11. Only data for those substances and occasions when observed levels for all replicates of the sampled event exceeded the analytical q w t i t a t i o n limita (0.5-5 ng/sample) are reported. Additional substances which were looked for included (quantification limits in nanograms per sample) hexachlorobenzene (0.5). heptachlor epoxide (1.0). a-chlordan (1.0). p,p‘-DDE (l.O), p,p’-TDE (4.0). endosulfan (1.51, mirex (2.0), and methoxychlor (5.0). Some of these were observed at trace levela (i.e., as low as 20% of the quantification limits). Results and Discussion In the laboratory recovery experiments reported in Table I, the mean coefficientof variation for the teat compounds was 13% (range 3-25%) for the XAD-2 resin; the corresponding average for XAD-7 resin was 8% (range 4-16%) when heptachlor is excluded. These values are comparahle
Table 11. Field Sampling Results sample
rainfall volume, L 0.42 2.1 2.1 2.2 1.0
50-1104 50-1404 50-2404 50-2804 50-0505 TL-0206 18.0 TL-0406 3.2 TL-0906 4.2 TL-1806 22.0 TL-1007 15.0 av coeff of var (%) quantitation limit, ng
a-BHC 35 (49) 38 (15) 29 (29) 10 (31) 5 (3) 28 (31) 48 (41) 19 (39) 32 ( 3 1 ) (30) 0.5
concentration (percent coefficient of variation), ng/L lindane 7-chlordan dieldrin p,p’-DDT endosulfan endrin 7 (20) 11 (15) 10 (28) 5 (35) l(4) 7 (601
2 (25)
0.9 (11)
0.1 (41)
0.4 (12)
(26)
0.6 (12) (12j 1.0
0.7 (69)
0.2 (49)
0.4 (38) (3oj 0.5
1.0
to the precision obtained for the gas chromatographic calibration standards (mean 9%, range 5 1 2 % ) and are better than the average 25% reported by IWD (17). This improvement possibly results from the cleanup step which was not included in the laboratory recovery tests but which was included in the analyses of the field samples which were done in the IWD laboratory. The recoveries for the substances tested using XAD-2 and XAD-7 (Table I) averaged 86% (range 73-98%) and 68% (range 42-103%), respectively; heptachlor on XAD-7 is not included because of the high coefficient of variation observed in that situation. If the comparison is limited to the four compounds common to both recovery tests, the XAD-2 and XAD-7 results become 85% (range 73-96%) and 75% (57-103%); only y-chlordan was recovered in higher yields on XAD-7. The XAD-2 results compare favorably with those obtained in the IWD laboratories where 80% was the average using the identical procedures and the same compounds but excluding a cleanup step using Porasil columns. The XAD-2 recoveries are slightly better than those using XAD-7, a result to be expected considering the nonpolar nature of all of the test substances for which XAD-2 is designed. The XAD-2 resin was employed in all field sampling, but for collecting more polar substances that may be present in rainfall but which were not investigated here, XAD-7 may offer some advantages. A mixed resin bed may also be worth considering for general sampling of wetfall organics. It can be seen from Table I1 that no systematic relationship exists between the coefficients of variation and the concentration levels of a-BHC, lindane, or PCB’s observed in the field samples; no such relationship was found with the quantities absorbed either. A somewhat higher but statistically insignificant frequency of detection of other compounds at the trace levels was noted for samples from the Turkey Lakes site probably due to the higher volume of samples collected. In addition to those compounds reported in Table 11, endosulfan, methoxychlor, heptachlor epoxide, and p,p’-DDE were occasionally found at trace levels. Since only a-BHC, lindane (7-BHC), and PCB’s were observed with any degree of frequency and quantitative significance, only these are discussed. The coefficients of variation for a-BHC, lindane, and PCBs in the field samples in Table I1 averaged 27,30, and 42%, respectively. While there is considerable variation in these precision levels, sampling of environmental matrices is often highly variable. In the case of rainfall, much of the contaminated precipitation is deposited at the beginning of an “event” (16),and identical sampling is difficult. The corresponding values for the same compounds in water samples (17)averaged 25% for replicate analyses on single samples. It would appear, therefore, that the sampling procedures reported for our sampler have a
3.0
(38)
1.5
(47)
2.0
(49)
comparable level of precision to that obtained by welltested water sampling methods, and this indicates the suitability of both the sampler and the use of XAD-2 resin columns in the field. The slightly higher variation observed may be the result of localized variability either in the rain intensity or in the concentrations of the chemicals in rain collected by the different samplers. Since most of these substances appear to be washed out of the atmosphere in the early phase of a given rainfall event (16), it is likely the latter which is the more important factor, but this is by no means certain. Conclusion The sampler described in this paper is designed to collect and concentrate presistent organic substances that occur in rainfall in either particulate or dissolved forms. It offers an average recovery of 86% for organochlorines with an average coefficient of variation of 33% on replicate samples collected under environmental conditions. Few quality control studies have ever been attempted for this type of sampler, but it is believed that these results are at least as good as those from any other technique. The sampler also offers the advantage of unattended operation for extended periods, and it provides a simple method for field handling of samples since the columns can be readily exchanged and capped for delivery. Acknowledgments
Analysis of field samples in the laboratories of the Ontario laboratories of the Ontario Region of IWD (R. Thompson) is gratefully acknowledged. The engineering assistance and advice of A. Pashley, National Water Research Institute, were much appreciated. Literature Cited (1) Eisenreich, S. J.; Looney, B. B.; Thornton, J. D. “Assessment
(2)
(3) (4) (5) (6) (7)
of Airborne Organic Contaminants in the Great Lakes Ecosystem”; International Joint Commission: Windsor, 1980; Appendix A to the Science Advisory Board Report, pp 1-150. Science Advisory Board “A Perspective on the Problem of Hazardous Substances in the Great Lakes Basin Ecosystem”; International Joint Commission: Windsor, 1980; pp 1-70. Benvenue, A.; Ogata, J. N.; Hylin, J. W. Bull. Environ. Contam. Toxicol. 1972,8, 238-241. Lunde, G.; Gether, J.; Gjos, N.; Lande, M.-B. S. Atmos. Environ. 1977, 11, 1007-1014. Masahiro, 0.; Takahisa, H. Environ. Pollut. 1975, 9, 283-288. Bidleman, T. F.; Olney, C. E. Science (Washington,D.C.) 1974,183, 516-518. Clausen, J.; Braestrup, L.; Berg, 0. Bull. Environ. Contam. Toxicol. 1974, 12, 529-534. Envlron. Scl. Technol., Vol. 18, No. 2, 1984
129
Environ. Sc/. Technol. 1984, 18, 130-132
Peterle, T. J. Nature (London) 1969,224, 620. Sodergren, A.; Svensson, B.; Ulfstrand, S. Environ. Pollut. 1972,3,25-37. Swain, W. R. J. Great Lakes Res. 1978, 4 , 398-407. Murphy, T. J.; Rzeszutko, C. P. J. Great Lakes Res. 1977, 3,305-312. Strachan, W. M. J.; Huneault, H. J. Great Lakes Res. 1979, 5, 61-68. Smith, J. H. H.; Bomberger, D. C., Jr.; Haynes, D. L. Chemosphere 1981, 10, 281-290. Wolff, C. J. M.; van der Heijde, H. B. Chemosphere 1982, 11, 103-117. Allen, H. E.; Halley, M. A. "Assessment of Airborne Inorganic Contaminants in the Great Lakes"; International Joint Commission: Windsor, 1980; Appendix B to the
Science Advisory Board Report, pp 1-160. (16) Strachan, W. M. J.; Huneault, H.; Schertzer, W. M.; Elder, F. C. In "Hydrocarbons and Halogenated Hydrocarbons in the Aquatic Environment"; Afghan, B. K.; Mackay, D., Eds.; Plenum Press: New York, 1980; pp 387-396. (17) Inland Waters Directorate "Analytical Methods Manual (1979)"; Inland Waters Directorate: Department of the Environment, Ottawa, 1979. Received for review March 28, 1983. Accepted September 14, 1983. The development of the sampler and acquisition of three prototypes was funded by the Department of Environment, Environmental Contaminants Contract Fund, under Grant 9904-EMS-IWD.
Rapid Method for the Digestion of Sewage and Sludge for Metal Analyses Jeppe S. Nielsen" and Steve E. Hrudey
Department of Civil Engineering, University of Alberta, Edmonton, Alberta, Canada T6G 2G7
A steam digestion method for metal determinations in sewage and sludge is described and compared to methods using nitric acid digestion and high-speed homogenization. The steam digestion method employs a domestic pressure cooker, provides quantitative metal recoveries, is not prone to metal contamination, and requires less reagents and supervision than the methods with which it was compared.
Introduction Measurement of total metal concentrations in sewage and sludge generally involves initial sample digestion to destroy the organic matrix and to solubilize the metals. However, there are disadvantages associated with the digestion methods commonly used. Dry-ashingin a muffle furnace at high temperatures (550 "C), followed by dissolution in dilute acid, gives incomplete recovery of volatile metals such as cadmium and lead (2-3). On the other hand, low-temperature ashing methods have been reported to give incomplete metals recovery (4). Wet digestion methods using nitric ("OB), hydrochloric (HCl), sulfuric (H2S04),hydrofluoric (HF), and perchloric (HC104) acids and hydrogen peroxide (H202), singly or in combination, can be hazardous and generally require considerable sample handling ( I ) . Digestion at atmospheric pressure (open digestion) with HN03 is one of the methods recommended for wastewaters by "Standard Methods" (5) and the US. Environmental Protection Agency (6). However, open HN03 digestion has been found to give low results for some metals in sludges (3,4). Mixtures of HNO:, with HC1 or H202have generally given better results (2,4, 7). As well, low recoveries of lead have been found with the use of HN03/H2S04mixtures (2,8) because of the formation of insoluble lead sulfate. Generally, the most efficient metal recoveries can be obtained by acid digestion in high-pressure vessels ( I ) . However, these methods are expensive and can be limited by the small weight of sample that can be contained in the pressure vessel (1). Recently, a digestion method utilizing maceration and ultrasonic energy generated by a high-speed homogenizer has been described (2,9,10). This method yielded results *Present address: Department of Civil Engineering,New South Wales Institute of Technology, New South Wales 2007, Australia. 130
Environ. Sci. Technol., Vol. 18, No. 2, 1984
for a wide range of metals which were not significantly different from those obtained by using high pressure or open digestion with HN03/H202and HN03/H2S04mixtures. The major advantage of high-speed homogenization was that it only required 5 min per sample. In this paper, we present a rapid and simple sludge digestion method in which acidified samples are steam digested under low pressure. The results from this method are compared with those when open acid digestion and high-speed homogenization are used. Sewage and sludge samples from Edmonton as well as a U.S. Environmental Protection Agency reference municipal sludge sample were used. The methods were compared for cadmium, chromium, copper, nickel, and zinc.
Experimental Section Apparatus. Steam digestion was carried out in an aluminum Model 411 domestic pressure cooker of 11-L capacity (Presto, Scarborough, Ontario). Homogenization was performed with a Polytron Model PT 10-35 homogenizer (Brinkmann Instruments Ltd., Rexdale, Ontario) by using a specially constructed titanium probe of the same design as the stainless steel Polytron PT 10 ST probe. Metals were determined by flameless atomic absorption spectroscopy by using a Model 5000 spectrophotometer (Perkin-Elmer, Norwalk, CT) with deuterium background correction, an HGA 2200 graphite furnace, and an AS-1 autosampler with 20-pL sample injection volumes. Reagents. High-purity deionized water from a R015Q2 Laboratory Water system (Millipore Corp., Bedford, MA) and Aristar nitric acid, "OB (British Drug Houses), were used throughout. Certified atomic absorption standard solutions (Cd metal, C U ( N O ~ Ni(N03)2, )~, ZnO in "OB, and KzCrZO7in water; Fisher Scientific Co.) were used to prepare the atomic absorption calibration standards. Samples. Grab samples of raw sewage, activated sludge, mixed liquor, and primary sludge from the City of Edmonton Gold Bar Wastewater Treatment Plant and a reference quality control sample of dried municipal digested sludge from the U.S.Environmental Protection Agency (sample no. 2792, WP976) were used. The raw sewage was undiluted whereas the mixed liquor and primary sludge samples were diluted 1:lO (v/v) with high purity water prior to digestion. The reference sludge was
0013-936X/84/0918-0130$01.50/0
0 1984 American Chemical Society