Anal. Chem. 1984, 56,1953-1956 Romano, S. J.; Renner, J. A. Am. Ind. Hyg. Assoc. J . 1979, 40, 742-745. Pilny, R. J.; Coyne, L. B., presented at the American Industrial Hygiene Conference, Chicago, IL, 1979. Kring, E. V.; Damreil, D. J.; Basilllo, A. N.; McGibney, P. D., Presented at the American Chemical society National Meeting, Washington, DC, 1983. Mullins, H. E.; Anders, L. W., Presented at the Pittsburgh Conference, Atlantic Cky, NJ, 1983. Occupational Exposure to Ethylene Oxide Fed. Reg/st. 1983, 48 (No.
1953
78), April 21.
RECEIVED for review January 6,1984. Accepted May 7,1984. The opinions or assertions contained herein are the private views of the authors and are not to be construed as reflecting the view of the Department of the Army or the Department of Defense.
Purge and Trap Chromatographic Method for the Determination of Acrylonitrile, Chlorobenzene, 1,2=Dichloroethane,and Ethylbenzene in Aqueous Samples J. Michael Warner* and Ronald K. Beasley Research Department, Monsanto Agricultural Products Company, 800 North Lindbergh Boulevard, St. Louis, Missouri 63167
A purge and trap method has been developed for the determination of four priority pollutants in aqueous samples. Water samples are purged at 50 O C with helium and the analytes are trapped on Tenax GC. The trap is thermally desorbed directly into a gas chromatograph equipped with a flame ionization detector. Instrument calibration is performed by direct Injectlon of standard solutlons in methanol through a modified purge vessel. The method was laboratory validated for the range of 20-500 ppb for each anaiyte using a 5-g aqueous sample. Average recoveries (reiatlve standard deviation) were 62% (0.08), 100% (0.07), 99% (0.07), and 92% (0.07), respectlveiy, for acrylonitrile, chlorobenzene, 1,2-dichloroethane, and ethylbenzene.
The purge and trap (PT) technique for the determination of volatile organics in aqueous samples is widely used (1-3). The advantages of the method include high sensitivity, simplicity of use, and relatively low cost. A disadvantage is the difficulty in directly calibrating the detector response after installation of the PT sampler. Further, water-soluble analytes often show low recoveries with this technique. Purge and trap samplers usually bypass the gas chromatograph injection port. Instrument calibration is generally performed one of two ways. In one, standards are injected and a response curve is constructed before sampler installation. Alternatively, aqueous standards are exhaustively purged and analyzed. The first procedure carries the assumption that peaks produced by direct injection have the same retention times and shapes as do those produced using the PT sampler. This calibration technique requires that the PT desorption process and passage of the analytes through the sampler have the Same efficiency as does sample vaporization in the injection port during direct injection. The second procedure carries the assumption that analyte recovery from any aqueous matrix is quantitative and reproducible. For a number of compounds, including many which are water soluble, this is not the case (4). Absolute recovery is difficult to determine and relative recoveries of analytes and internal standards from various real-world matrices may differ markedly from those of standard solutions prepared in deionized water. This work describes a method of instrument calibration using a PT sampler. It also describes a simple system for sample heating which improves analyte recovery. Finally, 0003-2700/84/0358-1953$01.50/0
these procedures are applied in the development and validation of a method for determining acrylonitrile (AN), 1,2dichloroethane (DCE), ethylbenzene (EB), and chlorobenzene (CB) in water. The method was developed as part of Monsanto Company's continuing efforts to monitor, document, and control priority pollutant levels in wastestreams.
EXPERIMENTAL SECTION Reagents. Methanol was Matheson, Coleman and Bell distilled in glass OmniSolv. Charcoal-filtereddeionized water was obtained with a Continental water system (St. Louis, MO). The priority pollutants AN, CB, DCE,and EB were obtained in 99+% purity from Supelco, Inc. Fisher Certified bromoform was used as an internal standard. Apparatus. A Hewlett-Packard 5840A gas chromatograph equipped with a flame ionization detector and a Hewlett-Packard 7675A purge and trap sampler were used for all analyses. Separations were performed on a Supelco, Inc., 6 f t X l/a in. 0.d. stainless steel column packed with Carbowax 1500 on SO/lOO Carbopack C using a 20 mL min-' flow of helium carrier gas. The column was conditioned by heating for 3 days at 170 "C with a 20 mL min-* flow of helium. The sampler trap was packed with 60/80 Tenax GC and was conditioned by heating 3 days at 250 "C under a 20 mL m i d flow of helium. A standard 1 5 - d needle purge vessel was used for analyses of all samples except instrument calibration standards. Instrument Calibration. The modified purge vessel shown in Figure 1 was used for instrument calibration. Instrument calibration standards were prepared in methanol over the range 5.0-250 pg mL-l of each analyte. These solutions also contained 202 pg d - l of bromoform as an internal standard. For instrument calibration, a modified purge vessel was fitted with a 5-mL size serum stopper over the side arm. The vessel was attached to the purge and trap sampler. Concomitant with the initiation of the purge cycle, 10 pL of a calibration standard was injected through the serum stopper into the empty (dry) vessel. The sample was vaporized with a heat gun and the analysis cycle was continued to completion. Method Validation. The method was validated by analyzing standard aqueous solutions of AN, DCE, EB, and CB, each at five levels from 20 to 500 ppb (20-500 ng mL-'). These solutions were prepared by first making primary standards in methanol and diluting aliquots of these to known volume with deionized water. A five-gram aqueous sample was accurately weighed on a balance accurate to five decimal places in grams into a standard 15-mL needle purge vessel. To this weighed sample was added 1.00 mL of a 2.0 pg mL-l standard solution of bromoform in water as internal standard. The vessel was attached to the sampler and purged with helium at 20 mL min-' for 15 min. The purge vessel 0 1984 American Chemical Society
1954
ANALYTICAL CHEMISTRY, VOL. 56, NO. 11, SEPTEMBER 1984
i-
DUCT TAPE
GLASS INSULATIONCLOTH
HEATING TAPE TWO PRONG MALE OUTLET 25 X 150 mm TEST TUBE
Flgure 1. Schematic diagram of modified purge vessel used In dry vessel instrument calibration.
heating assembly shown in Figure 2 had been preheated to 50 "C and was placed over the sample tube at the start of the purge cycle. It was immediately removed at the end of the purge. The Tenax trap desorption temperature was 200 O C . A 28 min run program was used for each analysis. The initial temperature was 55 O C for 2 min. The program rate was 8.0 O C min-l and the final temperature was 160 "C. The helium carrier flow was 20 mL min-l and the chart speed was 0.5 cm min-I. After the initial purge and analysis, one additional milliliter of bromoform internal standard solution was added and the analysis procedure was repeated. Results from the two analyses were combined in calculations of sample levels. Quantification was based on area ratios, analyte:bromoform, measured by an electronic integrator and using a calibration curve from the linear regression of the area ratios vs. micrograms of analyte. For each analyte, method validation was performed for five concentration levels (20,50,100,250,and 500 ppb analyk in water) over the validation range. At each level at least five independent determinations were performed. Field Samples. A test of the method was performed on field samples collected at an industrial site at which AN, CB, DCE, and EB were either in use as raw materials or potentially present as side products of various chemical processes in operation.
RESULTS AND DISCUSSION Instrument Calibration and Sample Recovery. Currently accepted procedures for calibration of a GC equipped with a purge and trap (PT) sampler suffer from difficulties with a mode of standard sample introduction. If the detector is calibrated before installation of the PT sampler, the assumption must be made that standard injection and aqueous sample purge produce peaks having similar shape and retention time characteristics. Further, that calibration mode makes it difficult to construct updated response curves. The dry vessel calibration procedure described here circumvents these problems by allowing the volatiles of the calibration solution to follow the same path as would those from an aqueous sample. Instrument calibration by exhaustive purge of aqueous standard solutions carries the assumption that complete and/or reproducible analyte and internal standard recovery occurs from all matrices. In some cases this is not true and absolute recovery cannot be determined. With the dry vessel procedure the anal* and internal standard are quantitatively vaporized. Detector response is thus more representative of the absolute amount of analyte in the calibration sample.
Flgure 2. Schematlc diagram of purge vessel heating tube assembly.
A simple approach to the recovery problem would conclude that the dry vessel calibration procedure is unnecessary since the internal standard method theoretically provides overall response ratios which should automatically take into account less-than-quantitative purge recoveries. However, this reasoning does not take into account real-world matrix effects. If a given matrix allows the ratio of the recoveries of analyte and internal standard to remain constant relative to that from deionized water, the analysis remains unbiased. If, however, the analyte:internal standard recovery ratio from a real-world matrix is not similar to that from deionized water the analysis becomes biased. Use of the dry vessel calibration procedure described in this work allows the comparison of sample results to an unbiased set of detector responses for the diagnosis of recovery ratio problems. Governmental agencies often require data on analyte recoveries for a proposed method. Thus a second advantage of the dry vessel calibration procedure is that it allows the direct determination of this information. An example of a chromatogram used in the instrument Calibration (1.25 pg each of AN, CB, DCE, and EB; 2.02 pg of bromoform) is shown in Figure 3. The average recovery from a 5-g aqueous sample containing 100 ng mL-' AN was 28% when using a 15 min room temperature purge of 20 mL mi& helium. By use of the purge vessel heating assembly for a 50 "C,15-min purge, recovery was increased to 44%. A second purge on the sample recovered an additional 18% for a total of 62%. Average AN recovery over the validation range was 62% for the two-purge method. Recoveries of CB, DCE,and EB were essentially quantitative on the first purge. When a single 30-min heated purge was used, the resulting chromatogram showed large interferences in its latter half. While the precise source of thii interference was not determined, it could be circumvented by using two 15-min purges. Method Validation. Validation data were analyzed by NIOSH-type statistical testing (6-8). Linear regression analysis of concentrations of analytes found in water by the PT method vs. their known concentrations showed excellent correlation (R22 0.996). Least-squares fit parameters are shown in Table I. The relative standard deviations for the four compounds ranged from 0.069 to 0.079. Examples of
ANALYTICAL CHEMISTRY, VOL. 56, NO. 11, SEPTEMBER 1984
1955
P
I
,F
I
/Ii
Flgure 3. Gas chromatographic dry vessel calibration chromatogram from the injection of a 10-pL aliquot of a standard solution containing 125 pg/mL each of AN, CB, DCE, and EB in methanol. The injection was made by use of the modified purge vessel of Figure 1.
Table I. Results for Validation of AN, CB,DCE,and EB in Aqueous Samples analyte AN
std concna 20 50 100 250
500
detd % concnavb recovery 11.7 31.6 61.6 153 323
59 63 62 61 65
RSDc
s
pooled RSD
5 ti
0.052 0.083 0.040 0.12 0.042
I!
0.079
CB
20
50 100 250
500
18.8 49.3 101 266 488
94 99 101 106 98
0.11 0.040 0.064 0.060 0.073 0.073
DCE
20
50 100 250
500
19.4 34.7 98.5 268 497
98 91 99 107 99
0.091 0.046 0.072 0.063 0.068
Table 11. Aqueous Field Sample Storage Study
0.069
EB
20 50 100 250 500
16.1 45.4 89.4 262 468
81 91 89 105 94
0.096 0.049
0.058 0.059 0.078
Flgure 4. Method validation chromatograms resulting from the analysis of a 5% aqueous sample containing 250 ppb each of AN, CB, DCE, and EB: (top) first purge; (bottom) second purge.
0.069
'Units in ppb. bAverage of at least five determinations. Relative standard deviation. chromatogramsfrom the f i t and second purges of a standard aqueous solution (250 ppb each of AN, CB, DCE, and EB) used in method validation are shown in Figure 4. In order to test the applicability of this method to the analysis of samples containing concentrations of AN, CB, DCE, or EB in excess of 500 ppb, a standard solution was prepared containing 5000 ppb of each of these compounds. One-half-gram samples of this solution were weighed on a balance accurate to five decimal places in grams, diluted to approximately 5 g with charcoal-filtered deionized water, and analyzed. The following average percent recoveries were realized AN, 72%; CB, 84%; DCE, 93%; and EB, 76%. The recoveries of CB, DCE, and EB were lower than those observed in the 20-500 ppb range. One explanation may be that the 5000 ppb level represented such a high concentration in water of these hydrophobic materials that volatilization be-
concn, ppb day 0 day 5
analyte
sample
AN CB DCE
1
3000
2 3
840 1300
500 ppb) are encountered. AN recovery did not decrease presumably because it is water soluble. Field Samples. Testing of this method on field samples was performed by using samples collected from various sources at an industrial site at which the four analytes were present either as raw materials or as byproducts. The following ranges of concentrations were found (figures given in ppb): AN < 20--14000; CB < 20-840; DCE < 20--15000; EB < 20-660. The site final effluent contained only CB a t levels of 4 5 and 1 4 3 ppb on two different days. The higher levels observed were from samples collected prior to waste treatment. Three field samples were selected for storage stability studies. The samples were stored 5 days at 10 "C in vials with no headspace. The results are summarized in Table 11. The sharp decrease in concentrations with time points to the very short shelf life of samples and emphasizes the need for immediate analysis upon collection. Other authors (I) indicate
1956
Anal. Chem. 1984, 56, 1956-1959
that aqueous solutions of volatiles are stable for up to 2 weeks under the storage conditions used in this study. However, work on rates of biodegradation (9) shows that many priority pollutants are rapidly lost in the presence of yeast cultures (100% biodegradation within 7 days for AN) or microorganisms commonly found in activated sludge. The short shelf life observed for the samples in this work shows the importance of immediate analysis in the absence of data proving that storage of a given field sample matrix is an acceptable practice.
ACKNOWLEDGMENT Helpful discussions with D. F. Tomkins on purge and trap methodology are gratefully acknowledged.
LITERATURE CITED (1) "Guidelines EstablishingTest Procedures for the Analysis of Pollutants" Fed. Reglst. 1979, 44, 69464-69575. (2) Ramstad, T.; Nicholson, L. W. Anal. Chem. 1982, 5 4 , 1191-1196. (3) Grob, R. L. "Modern Practice of Gas Chromatography"; Wiiey: New York, 1977; Chapter 3. (4) Gargus, A. G.; et ai. Am. Lab. (FalrfleM,Conn.) 1981, 114-120. (5) Beliar, T. A.; Lichtenberg, J. J. J . Am. Waste Water Assoc. 1974, 66, 739-744. (6) Hiieman, D. 0.; et ai. "Documentation of the NIOSH Validation Tests"; DHEW (NIOSH) Publication No. 77-185, Cincinnati, OH, 1977. (7) Grubbs, F. E.; Beck, G. Technometrlcs. 1972, 14, 647-854. (8) Bethea. R.; et ai. "Statistical Methods for Engineers and Scientists", Marcel Dekker: New York, 1975; pp 247-251. (9) Tabak, H. H.; et ai. J. Water Pollut. Control Fed. 1981, 53, 1503-1518.
RECEIVED for review January 17,1984. Accepted May 1,1984.
Concentration and Separation of Trace Metals from Seawater Using a Single Anion Exchange Bead Minoru Koide,* Dong So0 Lee,l and Martha 0. Stallard Scripps Institution of Oceanography, A-020, La Jolla, California 92093
A technique has been developed for the quantitative adsorption of trace metals onto a single anion exchange bead. The application to the assay of trace metals in seawater was explored with the following radionuclides: Io9Cd,loBPd,lS2Ir, '''Au, ='PU, and 9 s T c . The major ions, Na', K+, Mg2+, and Ca2+ exlst primarily as positively charged species in seawater under nearly ail conditions and did not Interfere in the adsorption of anlonk forms of trace metals onto the single bead. Three types of applications of the technique were investigated (A) determinatlon of metals in seawater by the direct adsorption onto a single bead without prlor concentratlon, with or without a subsequent desorption from the bead (e.g., Cd, Zn); (B) determination of metals in seawater by the adsorption onto a slngie bead afler a preconcentration step from several liters of seawater (e.g., W, Au, Ir), and (C) increasing the yield of Pu and Tc onto a single bead for improved sensitivity in mass spectrometric analyses.
The use of the individual ion exchange bead was first proposed by Freeman ( I ) as a micro calibration standard for sodium. However, the useful application of the technique had its origin in the field of mass spectroscopy. Since, in general, only isotopic ratios are measured, quantitative adsorption onto the bead is not critical. Plutonium and uranium (2,3),in the anionic form, and zirconium (4), in the cationic form, were measured at nanogram levels by mass spectroscopy after partial adsorption onto ion exchange beads. For example, Walker et al. (3) equilibrated five to six beads in 8 M "03 for 24-48 h which contained 0.1-100 ng of plutonium. The adsorption yields were 27-58%. In subsequent work (5),q c was determined at the picogram levels utilizing this bead technique. The bead uptake of geTcranged from 37 to 85%. However, the present work demonstrates that essentially quantitative adsorption of plutonium and technetium onto 'Present address: Massachusetb Institute of Technology, Carleton Street 42, Cambridge, MA 02139. 0003-2700/64/0356-1956$01.50/0
a single bead can be achieved by using the optimal conditions outlined herein. We have explored the application of the single bead techniqu as an analytical isolation step for the determination of trace metals in seawater. A broad range of different trace metals will be discussed. More comprehensive data for each element along with seawater profiles will be published elsewhere [e.g., Pd (811. Use of this technique reduces contamination problems due to manipulation of the sample during conventional extraction or precipitation procedures (9). The anion exchange system graphs established by Kraus and Nelson (6) and Buchanan and Faris (7) for trace metals in the chloride and nitrate forms indicate those elements for which the single bead technique would be most applicable. Seawater is an ideal medium for the application of the bead technique since a large number of trace metals in seawater exist partially or totally in anionic forms which might be adsorbed onto an anion exchange bead, thus allowing separation from the major cations. In practice an additional complexing agent (CN- or SCN- in addition to C1-) is added to maximize metal adsorption onto the bead.
EXPERIMENTAL SECTION Standards and Reagents. All solutions were prepared with doubly distilled water. Redistilled sulfuric, nitric, and hydrochloric acids were obtained from G. Frederick Smith Chemical Co. (Columbus,OH). Ammonium hydroxide was isothermally distilled in this laboratory. The 4% KCN and 4% KSCN solutions were purified by passing the solutionsthrough a Bio-Rad AG 1x2 anion exchange resin column. The resin had been precleaned by washing with hot 14 M "OB, rinsing with double distilled water, and converting to the chloride form. The resin column was 0.6 cm i.d., 1.5 cm long. The Ce(S04)2in 1.2 N H2S04was purified by passing it through a sulfate form AG 1x2 resin column. Amberlite IRA-900 (Sigma Chemical Co., St. Louis, MO) and Bio-Rad AG-1 anion exchange resins (16-50 mesh) were cleaned by soaking in 3 M "OB and 3 M HC1 repeatedly. Glass beads (16-20 mesh Glasperlen B.Braun Melsungen) were cleaned by washing with ethyl alcohol, 4 M "OB, and distilled water. Stable and radioactive working standards were prepared by successive dilution of 1000 pg/mL atomic absorption standards (VWR Scientific Co., San Diego, CA) and of 10 fiCi radioactive 0 1984 American Chemical Society
