Anal. Chem. 1997, 69, 1127-1134
Analyses of Fish Tissue by Vacuum Distillation/Gas Chromatography/Mass Spectrometry Michael H. Hiatt
National Exposure Research Laboratory, Characterization Research Division, U.S. Environmental Protection Agency, P.O. Box 93478, Las Vegas, Nevada 89193-3478
The analyses of fish tissue using VD/GC/MS with surrogate-based matrix corrections is described. Techniques for equilibrating surrogate and analyte spikes with a tissue matrix are presented, and equilibrated spiked samples are used to document method performance. The removal of analytes from the tissue corresponds to Koa and Kwa (octanol-air and water-air partition coefficients, respectively). For a given vacuum distillation, the impact of Koa and Kwa on analyte recovery can be determined by interpreting the recoveries of surrogate compounds and their Koa and Kwa. The use of these surrogates to monitor and correct for both water-air and octanol-air partitioning provides average recoveries of 86% for volatile gases, 97% for volatiles, 90% for neutral semivolatiles, 124% for basic semivolatiles, and 87% for the water-soluble volatiles. The method detection limits are sub-parts-perbillion for most analytes studied. A technique to experimentally determine the octanol-air partition relative volatility is described and the values for 113 compounds are presented. Fish tissue is an important medium for study due to its importance as a food source and as an indicator of the quality of the environment.1 The application of vacuum distillation/gas chromatography/mass spectrometry (VD/GC/MS) for tissue analyses has been reported for volatile organic compounds (VOCs)2 and is being promulgated as a Resource Conservation and Recovery Act (RCRA) method (SW-846 Method 5032).3 It has been shown that surrogate-based matrix corrections can improve method performance by accurately measuring and compensating for matrix effects from water, soil, and oil.4,5 This work was undertaken to apply the surrogate-based matrix corrections to expand the Method 5032 analyte list and to obtain lower method detection limits for analytes in fish tissue. Before the vacuum distillation of analytes from fish tissue could be evaluated, it was necessary to ensure added compounds were in equilibrium with the matrix. Three means of adding anaytes to the fish tissue were investigated. Determination of the interval of time necessary for a spike to equilibrate required an assumption. The assumption was that an analyte was at equilibrium with the matrix when its recovery from the matrix did not decline with (1) U.S. Environmental Protection Agency. National Study of Chemical Residues in Fish; Office of Science and Technology: Washington DC, 1992; Vol. 1. (2) Hiatt, M. H.; Youngman, D. R.; Donnelly, J. R. Anal. Chem. 1994, 66, 905908. (3) U.S. Environmental Protection Agency. Test Methods for Evaluating Solid Waste, SW-846; Office of Solid Waste, GPO: Washington, DC, 1992. (4) Hiatt, M. H.; Farr C. M. Anal. Chem. 1995, 67, 426-433. (5) Hiatt, M. H. Anal. Chem. 1995, 67, 4044-4052.
further equilibration time. The surrogate-based matrix correction data were generated using samples of fish tissue equilibrated with the spiked analytes. The bioaccumulation of organic compounds for fish is closely related to the octanol-water partition coefficients.6 Therefore, a desirable matrix correction treatment should emphasize the accuracy for analytes with higher octanol-water partition coefficients. Conversely, for those analytes that have lower octanolwater partition coefficients, water becomes a better indicator of their concentration in the environment and their presence in fish tissue is not likely due to environmental exposure. The reported surrogate-based matrix correction model requires the solution of an equation relating analyte recovery to two factors, its partitioning between a sample and the vapor phase (waterair partition coefficient) and its condensation in the apparatus (corresponding to boiling point). The values of constants in the equation are determined using the recovery of surrogate compounds. The recovery of analytes from a tissue sample by vacuum distillation was expected to be described using this equation as it had been shown to describe the recovery of analytes from oil.5 A second model with a factor describing the vaporization of analytes from the organic phase became necessary when samples larger than 1 g were analyzed. It was discovered that the vacuum distillation of analytes from an organic matrix corresponds closely to the compound’s Koa and is the primary factor for analytes with higher Kow. Both models required the use of relative volatility (a comparison of water-air partition coefficients for different compounds) as the means of addressing matrix effects relating to the partitioning of analytes between the vapor phase and the water or organic phase. While relative volatility (R) is described as a ratio of two values (Kwa), it is convenient to assign values for relative volatility that correspond to their partition coefficients.5 In this work, the relative volatility of an analyte from the organic phase into the vapor phase (RKoa) is assigned values that correspond to the octanol-air partition coefficient (Koa). The RKoa values in this study were determined using the experimental technique reported for the determination of the RKw.5 EXPERIMENTAL SECTION Vacuum Distillation Apparatus. The vacuum distiller has been previously described.5 In the current study, a Nupro toggle valve (0.172-in. orifice) was used as the sample chamber valve. A pirani vacuum (Edwards Model 1000) gauge was placed at the vacuum pump to monitor the integrity of the apparatus under (6) Neely,W. B.; Branson, D. R.; Blau, G. E. Environ. Sci. Technol. 1974, 8, 1113-1115.
S0003-2700(96)00936-5 This article not subject to U.S. Copyright. Publ. 1997 Am. Chem. Soc.
Analytical Chemistry, Vol. 69, No. 6, March 15, 1997 1127
Table 1. List of Analytes and Their Group Subsets chemical propertiesb conca (ppb)
R Kw
bp (°C)
Kow
dichlorodifluoromethane trichlorofluoromethane vinyl chloride chloroethane chloromethane bromomethane
20 20 20 20 20 20
0.07 0.20 0.48 1.01 1.37 1.82
Volatile Gases -30 24 -13 12 -24 4
144.5 339.0 4.0 26.9 8.1 12.6
20 31 3 8 4 5
10 68 2 27 11 23
3 48 8 60 13 42
2 4 4 9 3 7
1,1-dichloroethene carbon tetrachloride 1,1-dichloropropene 1,1,1-trichloroethane allyl chloride 2,2-dichloropropane tetrachloroethene iodomethane trans-1,2-dichloroethene trichloroethene isopropylbenzene benzene ethylbenzene toluene m,p-xylenes 1,1-dichloroethane n-propylbenzene cis-1,2-dichloroethene o-xylene chlorobenzene chloroform styrene bromobenzene methylene chloride 1,2-dichloropropane 1,1,1,2-tetrachloroethane bromodichloromethane trans-1,3-dichloropropene bromochloromethane 1,2-dichloroethane dibromochloromethane cis-1,3-dichloropropene bromoform dibromomethane 1,3-dichloropropane 1,1,2-trichloroethane 1,2-dibromoethane 1,1,2,2-tetrachloroethane cis-1,4-dichloro-2-butene 1,2,3-trichloropropane trans-1,4-dichloro-2-butene check surrogates fluorobenzene 1,4-difluorobenzene 4-bromofluorobenzene methylene chloride-d6 1,2-dichloropropane-d6 1,1,2-trichloroethane-d3
10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 10
0.63 0.64 0.88 1.31 1.34 1.37 1.43 2.29 2.30 2.34 2.75 3.55 3.60 3.88 3.91 4.12 2.43 5.34 5.54 6.07 6.39 6.87 7.89 10.10 10.90 11.60 12.30 14.10 15.40 18.70 19.20 19.60 17.39 23.90 24.90 26.20 26.70 23.63 33.30 33.60 33.80
Volatiles 37 76.5 104 74 45 69 121 42 48 87 152 80 136 111 138 57 159 60 144 132 62 145 156 40 96 130.5 90 112 68 84 120 104 150 97 120 114 132 146 152 157 156
30.2 537.0 37.2 295.0 26.9 380.2 339.0 30.2 211.4 339.0 4571.0 134.9 1413.0 490.0 1479.1 61.7 4786.0 30.2 1318.3 691.8 93.3 1445.0 977.2 17.8 190.6 1096.0 75.9 25.7 25.7 28.2 123.0 25.7 199.5 11.8 95.5 148.0 39.8 245.0 24.5 234.4 24.5
8 40 9 29 8 33 31 8 24 31 128 19 68 38 70 12 131 8 65 46 16 69 56 6 23 59 14 8 8 8 18 8 23 5 16 20 10 26 8 26 8
19 340 33 390 36 520 480 69 490 790 12600 480 5100 1900 5800 250 12000 160 7300 4200 600 9900 7700 180 2100 13000 930 360 400 530 2400 500 3500 2800 2400 3900 1100 5800 820 7900 830
220 570 320 420 47 200 6100 29 89 2600 9200 460 5000 2100 4500 180 11000 210 8200 5700 570 8500 12000 100 14000 7700 2900 1900 470 810 9300 5600 17000 1800 3400 3500 4900 23000 11000 14000 14000
150 240 80 140 18 30 300 14 14 600 500 80 340 170 200 24 830 10 160 130 45 150 1000 10 370 270 1000 90 90 30 1500 4200 1700 100 400 700 200 2400 460 990 770
5 5 5 5 5 5
3.50 3.83 5.97 11.10 11.00 26.60
85 88.5 152 40 95 112
195 226 1350.0 17.8 190.6 148.0
23 25 66 6 23 20
680 870 8100 200 2100 3900
690 860 13000 98 98 3800
60 70 1400 15 15 20
n-butylbenzene sec-butylbenzene hexachlorobutadiene p-isopropyltoluene tert-butulbenzene 1,3,5-trimethylbenzene 2-chlorotoluene 1,2,4-trimethylbenzene 4-chlorotoluene 1,3-dichlorobenzene 1,4-dichorobenzene 1,2,4-trichlorobenzene 1,2-dichlorobenzene 1,2,3-trichlorobenzene pentachloroethane naphthalene 1,2-dibromo-3-chloropropane check surrogates decafluorobiphenyl naphthalene-d8
10 10 10 10 10 10 10 10 10 10 10 10 10 10 10 20 10
1.88 1.91 2.08 2.50 2.72 3.75 4.04 4.50 4.78 5.72 6.14 7.33 7.86 11.30 13.20 11.03 38.90
Neutral Semivolatiles 183 15849.0 173 17378.0 215 60256.0 183 12589.0 169 12882.0 165 2630.0 159 2399.0 169 4266.0 162 2399.0 173 2399.0 174 2455.0 214 9550.0 180 2399.0 218 13804.0 162 1120.0 218 2344.0 196 426.6
251 264 519 222 225 95 90 123 90 90 91 191 90 233 60 89 35
30000 33000 130000 31000 35000 9900 9700 19000 11000 14000 15000 74000 19000 16000 15000 26000 17000
22000 16000 38000 35000 15000 15000 13000 16000 14000 20000 20000 39000 21000 46000 24000 41000 36000
4500 2300 14000 8900 1800 1700 1500 2200 1500 3500 3500 4500 3500 200 6000 2000 7600
5 10
3.03 18.00
87 89
6800 42000
35000 57000
8900 24000
RKo
1128 Analytical Chemistry, Vol. 69, No. 6, March 15, 1997
206 217
2253.0 2344.0
BCF
Koac
av
dev
Table 1. (Continued) chemical propertiesb RKo
conca (ppb)
R Kw
diethyl ether ethyl methacrylate methyl methacrylate methacrylonitrile 4-methyl-2-pentanone 2-hexanone acrylonitrile 2-butanone propionitrile 1,4-dioxane check surrogates acetophenone-d5 nitromethane-d3 acetone-d6
10 10 10 10 20 10 10 10 10 10 20 5 50
34.90 48.40 71.40 102.90 119.90 131.10 161.00 770.00 1420.00 5750.00 161.00 510.00 600.00
n-nitrosodimethylamine n-nitroso-methyl-ethylamine n-nitrosodi-n-propylamine n-nitrosodiethylamine aniline o-toluidine
67 67 67 67 10 67
2-methylnaphthalene 2-picoline pyridine n-nitrosodibutylamine check surrogates nitrobenzene-d5 hexafluorobenzene pentafluorobenzene benzene-d6 toluene-d8 o-xylene-d10 chlorobenzene-d5 1,2,4-trichlorobenzene-d3 bromobenzene-d5 1,2-dichlorobenzene-d4 1,2-dichloroethane-d4 1,2-dibromoethane-d4 diethyl ether-d10 1-methylnaphthalene-d10 ethyl acetate-C13 tetrahydrofuran-d8 1,4-dioxane-d8 pyridine-d5
5 5 10 5 5 5 5 5 5 5 5 5 20 50 5 50 50
bp (°C)
Kow
Soluble Volatiles 35 117 101 90 117 128 78 80 97 101 202 101 57
BCF
Kaoc
av
dev
6.8 109.7 24.0 8.2 12.3 60.3 1.8 1.8 1.4 0.5 38.0 0.5 0.6
4 17 7 4 5 12 2 2 2 1 10 1 1
240 5300 1700 840 1500 7900 290 1400 2000 2900 6100 260 350
360 2200 1200 400 2000 2300 190 710 680 1700 28000 14 230
310 290 88 53 180 1100 49 700 120 210 1800 23 17
129.00 1900.00 2400.00 4900.00 13700.00 15200.00
Basic Semivolatiles 154 0.7 165 2.3 206 1.2 177 7.9 184 7.9 200 36.1
1 2 1 4 4 9
39000 44000 2900 39000 108000 550000
5300 7000 20000 11000 30000 46000
3700 3900 5100 3000 14000 24000
67 10 10 67
67.00 6800.00 13100.00 21000.00
Marginal Analytes 245 10700 129 21.5 116 4.7 240 239.9
203 7 3 26
720000 150000 62000 5000000
130000 20000 16000 64000
150000 1900 8500 91000
5
87.50
13
6100
38000
6700
14 11 19 38 65 46 191 56 90 8 10 4 203 3 4 1 3
68 78 530 2100 8100 4300 75000 7700 19000 560 1040 220 720000 750 2300 2900 70000
130 350 500 2200 8400 5800 41000 12000 22000 810 5000 105 370000 330 380 1600 13000
16 170 60 110 340 280 6800 1000 4000 70 180 130 490000 50 60 160 9000
210
69.2
Surrogate Correction Compounds 0.86 81.5 79.0 1.51 85 51.9 3.92 79 135.0 4.28 111 490.0 6.14 143 1318.3 6.27 131 692.0 7.88 213 9550.0 7.93 155 977.0 8.03 181 2399.0 20.00 84 28.2 26.00 131 39.8 32.50 35 6.8 67.00 241 10700 150.00 77 5.0 355.00 66 6.6 5800.00 101 0.5 15000.00 115 4.7
a Concentration of analytes used in this study for 10-g samples. The concentration of analytes in the 1-g samples were 10 times more concentrated. The chemical property values were obtained from the following references: relative volatilities (RKw, ref 5), boiling points, (bp10), the octanolwater partition coefficients (Koa) were from ref 10 or calculated using Leo’s fragment constant approach,11 and the bioconcentration factors (BCF) were calculated using the Neely formula.6 c The octanol-air partition coefficients (Koa) were calculated using the relationships of fugacity constants.9 The experimental Koa were obtained by averaging six sets of data. The deviation is 1σ. b
vacuum. The sample chamber six-port valve temperature was maintained at 150 °C (Valcon E rotor). The cryoloop was modified by using nichrome wrapping for heating, which replaced the need for a hot water bath. The vacuum distiller was modified to allow a helium sweep of the condenser column to remove condensate between vacuum distillations. A helium transfer line (5 psi) was connected at the top of the condenser column (Nupro toggle valve), and a helium exit vent (Nupro toggle valve) was attached to the transfer line between the sample chamber valve and the condenser. The condenser column was normally held at 5-10 °C during vacuum distillations and at 60 °C between distillations. Water was
used to replace isopropyl alcohol as the temperature-controlling fluid in the condenser. Vacuum Spike Apparatus. A series of valves (Nupro toggle 4BKT) was connected to a 1/4-in. manifold. The manifold was also connected to a vacuum pump (Edwards E2M-1) and a thermocouple vacuum gauge. The valves and manifold were wrapped with heating tape and heated to 60 °C during use. The valves were connected to 15-mm O-ring fittings made of stainless steel for connection to the glass sample vessels also used for vacuum distillation. GC/MS Apparatus. A Hewlett-Packard mass spectrometer (Model 5972) and gas chromatograph (HP5890 Series II with Analytical Chemistry, Vol. 69, No. 6, March 15, 1997
1129
Model MJSC metal jet separator) with a 60-m × 0.53-mm-i.d., 3.0µm film thickness, Vocol capillary column (Supelco, Bellefonte, PA) was used for the determination of analytes from the vacuum distillation apparatus. Gas chromatograph operating conditions were as follows: 3 min at 10 °C, 50 °C/min ramp to 40 °C; 5 °C/ min ramp to 120 °C; 20 °C/min ramp to 220 °C; and isothermal at 220 °C for 3.4 min, resulting in a total run time of 28 min. The jet separator was held at 210 °C and the transfer line held at 280 °C. The mass spectrometer was operated at 3.1-s scans of 38270 amu. The injector was interfaced to the vacuum distillation apparatus by connecting the carrier inlet gas line to the cryoloop valve and back to the injector. The injection or inlet temperature was 240 °C, and the inlet pressure was 10 psi. Sample Preparation. Samples were prepared by mixing analytes with 1-10-g aliquots of fish tissue. Canned tuna (packed in water) was used as the matrix studied for method development. Seven different fish composites were obtained from EPA Region 9 (San Fransico, CA) and were used to evaluate the method performance for 1-g aliquots. Water (5 mL) was added to the samples prepared for water spike or sonication spike investigations. Water (1 mL) was added to samples to assist mixing after completion of the equilibration studies. Tissue samples were spiked with analytes dissolved in 10 µL of methanol directly in sample vessels (a 100-mL round-bottom flask fitted with a 15-mm O-ring connector) that were also used to contain the sample during vacuum distillation. Analytes and their concentrations used for this study are listed in Table 1. Additional surrogates were added for this study, and their relative volatilities were experimentally determined using the same technique as reported for the other analytes.5 The additional surrogates and their assigned relative volatility values are nitromethane-d3 510 ( 260; tetrahydrofuran-d8, 355 ( 11; nitrobenzene-d5, 87.5 ( 33.5; and diethyl ether-d10, 32.5 ( 3.0. Tetrahydrofuran-d8 was used in this study as a replacement for acetoned6 as acetone-d6 underwent rapid deuterium-hydrogen exchange in the tissue matrix. Vacuum Distillation Procedure. Prior to a vacuum distillation, the condenser column was cooled to 5 °C to condense water being evaporated from the sample. The sample chamber containing the sample (room temperature) was evacuated for 5 min. Most water vapors were collected on the condenser column, and the distillate, containing the analytes to be transferred to the GC/ MS, was collected in the cryoloop immersed in liquid nitrogen (-196 °C). The sample chamber valve was closed at the completion of the vacuum distillation and the cryoloop valve switched to allow the GC carrier gas to sweep the cryoloop. The cryoloop’s liquid nitrogen bath was removed, and the cryoloop temperature was ballistically heated to 120 °C to volatilize the distillate. After the transfer of the distillate to the GC was complete (3 min), the cryoloop was heated to 200 °C and then allowed to cool to room temperature. After the sample was vacuum distilled, the condenser column was heated to 60 °C and the condenser flushed with helium, while the helium/condenser valve and vent valve were opened to remove most of the trapped material. After 3 min the condenser-helium line and vent valves were closed and the system was evacuated with the vacuum pump for an additional 10-min period to remove any condensed water and potential contaminants remaining after the He flush. Spike Procedures. The vacuum spike was performed on 1and 10-g aliquots of tissue. A sample chamber containing the 1130
Analytical Chemistry, Vol. 69, No. 6, March 15, 1997
tissue and analyte spike (no water added) was attached to the vacuum spiking apparatus and emersed in a liquid nitrogen bath (-196 °C). After 5 min in the cryogenic bath, the tissue was completely frozen. The sample chamber was then evacuated by slowly opening the vacuum valve (between the manifold and sample chamber) and leaving it open until the pressure in the chamber was 1 Torr. The sample was then isolated by closing the valve, and the sample was warmed to room temperature (∼3 min when emersed in a warm water bath). After the sample thawed, the equilibration period began. At the completion of the equilibration time, water (1 or 5 mL) was added and the sample was immediately analyzed. The water spikes were performed simply by adding the spike solution to the 1 g of tuna, 5 mL of water matrix; the sample was swirled and allowed to stand. O-Ring connector caps (TG Scientific Glass, Irvine, CA) with Viton O-rings sealed the samples during their equilibration period performed at room temperature (18 to 22 °C). The sonication spike was prepared like the water spike but with the sample sonicated during the equilibration period. A ColeParmer, Model 08855-00 ultrasonic cleaner (Niles, IL) was modified to allow tap water to flow through the tank, maintaining a 15-17 °C bath during continuous operation. The sample chamber was emersed in an ultrasonic water bath to a depth where the sample is just below the surface of the water. RESULTS AND DISCUSSION Before any surrogate correction model could be evaluated, it was necessary to ensure the analytes added to the fish tissue were equilibrated. The absence of reference tissue containing the analytes being investigated made the development of a spiked tissue material necessary. Historically, the dosing of analytes to tissue was performed by spiking the analytes into aquariums containing fish and sacrificing the fish after an exposure period. The aquarium approach was not practical for this study considering the number of samples for analyses and the number of analytes to be determined. Evaluation of the analyte-tissue interaction period was required to assure the analytes are in equilibration with the matrix. It was reasoned that the response relative to a standard (relative response) of an analyte produced by the vacuum distillation of fish tissue was respective to its sorption with the matrix. If the relative response of an analyte continued to decrease as the time of its contact with the matrix was increased, there was not yet a state of equilibration. The analyte was considered at equilibrium with the tissue when its relative response no longer decreased when the equilibration interval was increased. Complicating the determination of analyte recovery from the tissue samples, however, was the enhanced distillation rate of water (almost double that of water without tissue) when 1 g of tissue was added to 5 mL of water. The tissue and water mixture boiled more vigorously than just the 5 mL of water, resulting in the response of some analytes to increase while others decreased. The effects due to the vacuum distillation (not degree of equilibration) were reduced by calibrating the VD/GC/MS system using standards in a fresh tissue and water mixture. The standard with tissue was prepared immediately before analyses to reduce equilibration in the standard. The minimization of analyte response as compared to the fresh standards (relative response) was then assumed to be due to the tissue effects enhanced by extending the equilibration period.
Figure 1. Relative response over equilibration times for naphthalened8 and 1 g of tissue using the different spike techniques.
By lengthening the equilibration period until there is no subsequent drop in analyte relative response, it was assumed the matrix and spike were in equilibration. Replicate analyses were performed to determine average analyte relative responses for various equilibration times. Figure 1 illustrates the relationship of analyte response to equilibration times for naphthalene-d8. Naphthalene-d8 was representative of compounds that required the longest time for equilibration with the tissue. Three types of spike techniques were investigated for this study. These techniques were the water spike (the spiking technique commonly used by the analytical community to add VOCs to a sample), a sonication spike, and a vacuum spike. Initially samples were vacuum distilled using 1-g aliquots of tuna and 5 mL of water. After the initial equilibration studies were conducted, the volume of water was reduced to 1 mL. Vacuum Spike. When tissue was vacuum spiked with analytes, the relative responses for the analytes decreased rapidly with increased equilibration time. The analytes were equilibrated with 1 g of tissue after 2 h as there was little additional decrease in relative response after 64 h of equilibration (Figure 1). Comparing the relative response of naphthalene-d8 to the three different spike techniques for 1 g of tissue, the vacuum spike was seen to be the superior technique as it had a reasonable time requirement and a clear equilibration end point. The water and sonication spike techniques would be evaluated as comparable to the vacuum spike. The effectiveness of the vacuum spike for 10-g samples was also evaluated. This evaluation was done by conducting two separate vacuum spikes to the fish tissue. The first vacuum spike contained analytes for a 2-h equilibration interval. After the equilibration period, a vacuum spike of the containing deuterated analogs of the analytes was conducted. The second vacuum spike had equilibration periods of 125-211 min. It was assumed that when the relative response of the analyte and the relative response of its deuterated analog were equal then the compounds were in equilibrium with the tissue matrix. This study confirmed that most analytes were equilibrated after 2 h, with the relative response of the highest boiling analytes (>200 °C) within 50 and 80% of their labeled analogs. After 3 h of equilibration, the higher boiling analytes’ relative responses were better than 90% compared with their labeled analogs. Evaluating the relationship of equilibration (measured as recovery of the analyte to the relative response of its deuterated analog) to the analytes’ boiling point, molecular weight, Kow, and Koa, it was seen that the boiling point has the greatest influence.
The vacuum spike technique demonstrated an equilibrium of analytes and 1 g of tissue after 2 h. The vacuum spike of 10g of tuna was less equilibrated at 2 h. Extension of the vacuum spike to a 3-h analyte-matrix interaction produced an effective equilibration spike technique. Water and Sonication Spike. The water and sonication spike techniques could not be evaluated using the given assumption (equilibration was reached when relative responses of analytes had reached their minimum) as the relative responses of some analytes continued to drop after 64 h (Figure 1). The sonication spike demonstrated a quicker drop in analyte relative response compared with the water spike equilibration time (Figure 1), but after 20 h, the water and sonication spike techniques resulted in very similar analyte responses. The longer equilibration periods resulted in severe tissue degradation, which was undoubtedly impacting the analyte recoveries. The effectiveness of the sonication spike was then evaluated using two separate spikes (one for the labeled analytes and one for the unlabeled analytes) similar to that done to evaluate the vacuum spike technique. For this study, the unlabeled analytes were first vacuum spiked for 2 h. At the completion of the vacuum spike, 1 mL of water and the deuterated analogs (surrogates) were added to the tissue and the sonication equilibration was performed. By extending the sonication period until the relative response of analyte and the relative response of their labeled analog were equal, the time necessary for spike-tissue equilibration using sonication could be established. An overnight equilibration period (>1000 min) was necessary to bring the differences (between unlabeled and labeled analogs) in relative responses to less than 10%. A similar comparison of the water spike to the vacuum spike was conducted, and 1000-min water spike equilibration was found equivalent to the 1000-min sonication equilibration. The vacuum spike was the most desirable technique. The vacuum spike equilibrated all analytes with tissue in 1000 min) prior to analyses. The analyte concentrations are 10 times those listed in Table 1. b Relative response ratio is the response of analytes from the sample divided by the analyte response from the aqueous standard. The deviation is 1σ. c Recovery is the measured analyte relative response divided by the predicted surrogate-corrected relative response. The diviation is 1σ. d Average predicted surrogate correction precision. e Method detection limits are calculated as 3 times the precision for 21 determinations of analytes at concentrations ∼3 times their estimated MDLs. The average, deviation, and median are of the individual analytes withing the group. f Analyte contained in groupings are identified in Table 1.
An additional function describing the recovery of analytes with losses due to condensation is
Rβ ) b(bp - bp0) + d
(2)
where Rβ is the recovery corresponding to the boiling point, b and d are constants, bp is the analyte’s boiling point, and bp0 is the boiling point where effects were shown to be negligible (80 °C). The recovery of an analyte is then calculated as Rw × Rβ resulting in the equation
recovery ) eaRKw+c(b(bp - bp0) + d)
(3)
The vacuum distillation of a sample that contains four or more surrogates provides the information necessary to solve eq 3. Use of the experimental recoveries of the surrogates (relative responses), their boiling points, and their relative volatilities (RK) provides solutions for the constants. The determination of the constants can be simplified by solving eq 1 or eq 2 separately using the recoveries of surrogates with either boiling points or relative volatilities that are equal. The recovery of any other analyte is then predicted by using its boiling point and relative volatility to solve eq 3. In practice, the surrogates are grouped to solve increments of the range of boiling points and relative volatility values representing the analytes. It was noted that use of more surrogates in a grouping than necessary to solve the equations would provide replicate predicted recoveries for analytes with their variation an estimate of the error associated with the solutions for eq 3. This approach worked very well when 1-g tissue samples were analyzed (Table 2). The method detection limits were lower than those previously reported for fish tissue and improved matrixcorrected recoveries.2 The recoveries reported in Table 2 were a means of gauging the accuracy of the surrogate corrections. Recoveries in Table 2 (and Table 4) were calculated as the experimental relative response of analytes divided by their predicted relative response, 100% recovery being complete agreement. The data presented in Table 2 were generated using analyte responses compared to aqueous standards. Using standards prepared in a tissue matrix had little impact on the results for most analytes. The basic semivolatile analytes were found to have 1132
Analytical Chemistry, Vol. 69, No. 6, March 15, 1997
improved determinations when their standards were prepared in a tuna matrix. Surrogate Correction Model 2. When 10-g samples were analyzed with matrix corrections using model 1, the recoveries (experimental relative response divided by predicted relative response) of some analytes were less desirable. Another surrogate correction model (model 2) was developed to improve the prediction of matrix effects for the larger samples. Model 2 still incorporated the model 1 function, Rw, to describe analyte recovery from a water phase but added a new function to describe the partitioning of the analytes between the organic phase and the vapor phase. The partitioning between an organic phase and the vapor phase during a vacuum distillation is described in the same manner as the water-air partitioning4 using the relative volatility from an organic phase. The relative volatility of analytes from an organic phase made to correspond to the calculated Koa partitioning values is defined in this work as RKo. Before model 2 could be tested, values for Koa and the relative volatilities of the analytes from an organic phase were determined. The fugacity capacity coefficients (Z) for air and water and their relationships to partition coefficients are described as Kow ) Zo/ Zw, Kwa ) Zw/Za, and Koa ) Zo/Za.7 The octanol-water partition coefficients are then calculated as
Koa ) KowKwa
(4)
Using the relative volatility values previously reported (RKw) as Kwa and the Kow values listed in Table 1, values for Koa were calculated and the results listed in Table 1. The calculated Koa were used as a basis to assign values for RKo. The relative volatilities of analytes from an organic phase were experimentally determined similar to the determination of relative volatilities from a water phase.5 In that work, successive vacuum distillations were performed on a sample. The response for an analyte in the first vacuum distillation compared with the sum of analyte responses from the successive distillations (recovery) corresponded to their relative volatilities. Using analytes with known Kwa the relationship of recovery to relative volatility was described in terms of Kwa and analytes were assigned values for their RKw. (7) Mackay, D.; Paterson, S. Environ. Sci. Technol. 1981, 15, 1006-1014.
The experimental RKo values were extrapolated from a relationship of calculated Koa to analyte recovery from an organic phase by vacuum distillation. The recovery of an analyte was determined by comparing the response of an analyte to its response if 100% was detected. The 100% response was determined by performing sequential analyses (between three and six analyses) of 0.2 g of cod liver oil spiked with the analytes. Some analytes were completely removed from the oil phase by the series of vacuum distillations, and the 100% response was simply the sum of their responses. The 100% response for the remaining analytes was calculated as the sum of responses including predictions of responses that could be obtained with additional vacuum distillations (extrapolating the trend of responses). The analyte responses from the first analyses were then divided by the 100% response to determine the recovery of analyte for the first analyses. Using calculated Koa for chloromethane, benzene-d6, toluene-d8, ethylbenzene, o-xylene-d10, naphthalene-d8, and methylnaphthalene-d10 to define the recovery rate-to-relative volatility (organic phase) relationship, experimental relative volatilities from organic phase values for all of the analytes were extrapolated. The resultant RKo values are also listed in Table 1 and these values were used in surrogate correction model 2. The relationship of analyte recovery respective to relative volatility (RKo) is the same general equation as the vapor-water phase partitioning equation used in model 1. Incorporating the organic phase variables, the relationship is described as
Ro ) ebRKo+d
(5)
where Ro is analyte recovery corresponding to its volatilization from the organic phase, b and d are constants, and RKo is the relative volatility of the analyte. Model 2 does not include the effects of condensation as a separate function. The effects corresponding to RKo are much greater and render the condenser boiling point corrections contained in model 1 less important. Treating the condenser boiling point and RKo effects separately becomes even less important considering the close correlation between Koa and boiling point for organic compounds. The determination of analyte recoveries from a two-phase sample required addressing the partitioning of analytes between the water and organic phases. Model 2 was developed with the assumptions that the analytes were at equilibrium between the phases and additional equilibration of analytes between the organic and water phases are insignificant during a vacuum distillation. The partitioning of the analytes between the phases was calculated using the analyte Kow and, for this study, approximating the water and organic content as 95 and 5%, respectively. The following equation is descriptive of analyte recovery calculations used by model 2:
R ) Rw + Ro ) XweaRKw+c + XoebRKo+d
(6)
where R is the recovery of an analyte from tissue and Xw and Xo are the fractions of analyte contained in the water and organic phases, respectively (Xw + Xo ) 1). Solving the equation for an analyte requires solutions for the constants, a-d. In a manner similar to a previous study, these equations were solved for range of values using specific surrogates (Table 3) and their respective
Table 3. Matrix Surrogates and Their Respective Ranges range type RKw
values 200
RK o
10000 bromobenzene-d5 1,2-dichlorobenzene-d4
a The surrogate pair combinations used in model 2 were all combinations of one from the first surrogate column and one from the second surrogate column for each value range.
experimental recoveries (relative response). The solutions for Ro equations were first solved using surrogates that have elevated Kow where losses are almost entirely due to the Koa partitioning. Using the multiple surrogates in each range of Koa values (four solutions), Ro is determined with a confidence interval for all analytes. With the Ro solutions determined for the four ranges (Table 3), Rw relationships are solved for the three ranges of RKw values using their representative surrogate pairs. As with Ro, multiple determinations of Rw provided confidence intervals for each range. Results for the analyses of 1-g tissue samples (sonication spiked with overnight equilibration) and 10-g tuna samples (vacuum spiked with overnight equilibration) are provided in Table 4. There was minimal difference in analytical results for model 1 and model 2 when 1 g of tissue was analyzed, which suggests the models are equivalent for 1-g samples. Model 2 was more effective than model 1 for those analytes that had Koa boiling point relationships dissimilar to that for the bulk of analytes. The major advantage of model 2 is that a larger sample can be analyzed without compromising the accuracy of matrix corrections. Comparison of the analyses of 10-g tissue to standards in water yields the accurate prediction of analyte recoveries (experimental relative response divided by predicted relative response). The low relative responses (less than 10%) for the neutral semivolatile analytes were observed for 10-g samples; however, recoveries of 90 ( 16% illustrate the ability of model 2 to describe and compensate for the matrix effects. The analyses of 10-g samples provide method detection levels (MDLs) that are 1 order in magnitude lower than previously reported. Spectral interferences for the analytes studied were infrequent. The coelution of methanol and water caused periodic chromatography and integration problems for several early-eluting compounds (ethyl ether, acetone, 1,1-dichloroethene, iodomethane, allyl chloride, and methylene chloride), but the surrogates ethyl ether-d10 and methylene chloride-d2 were good monitors for this effect. Several analytes used in this study were found unacceptable as analytes in tissue. These compounds apparently reacted with the tissue matrix and formed adducts that were not detected. Iodomethane and bromomethane disappeared at rates greater than Analytical Chemistry, Vol. 69, No. 6, March 15, 1997
1133
Table 4. Results for Surrogate Correction Model 2a 1-g tissue samples rel
responseb
recoveryc
10-g tissue samples
(%)
groupf
av
dev
av
dev
volatile gases volatiles neutral semivolatiles soluble volatiles basic semivolatiles marginals
0.84 0.52 0.26 0.81 1.45 0.84
0.08 0.16 0.07 0.23 0.61 0.56
103 96 110 101 248 122
14 14 19 28 68 46
rel response Surrd
(%) av
65 24 17 53 182 48
MDLe
recovery (%)
av
dev
av
dev
Surr (%) av
av
dev
median
0.49 0.30 0.07 0.62 0.48 0.29
0.20 0.20 0.02 0.39 0.19 0.21
86 97 90 124 87 100
44 27 16 64 36 32
27 25 24 94 84 93
0.7 0.4 0.5 0.5 13.1 3.9
0.4 0.8 0.8 0.5 16.0 4.9
0.8 0.2 0.3 0.2 3.5 1.6
a Samples were vacuum spiked with an overnight equilibration period (>1000 min) prior to analyses. One-gram samples consisted of seven composites and canned tuna. The 10-g samples were only of canned tuna. Analyte concentrations of 10-g samples are listed in Table 1. Concentrations of 1-g samples are 10 times the values listed in Table 1. b Relative response ratio is the response of analytes from the sample divided by the analyte response from the aqueous standard. The deviation is 1σ. c Recovery is the measured analyte relative response divided by the predicted surrogate-corrected relative response. The diviation is 1σ. d Average predicted surrogate correction precision. e Method detection limits are calculated as 3 times the precision for quadruple determinations of analytes at concentrations ∼3 times their estimated MDLs. The average, deviation, and median are of the individual analytes withing the group. f Analyte contained in groupings are identified in Table 1.
99 and 90 ( 2% per day, respectively, for a 10-g tissue sample. Halomethanes have been reported as being active alkylating agents with hemoglobin, and this explains their disappearance in fish tissue.8 The allyl halides also appeared to react with the tissue (presumably due to nucleophilic substitution and as these compounds are strong alkylating agents). All the allyl halides in this study disappeared quickly at the following rates (percent per day): allyl chloride, 24.1 ( 14.5; trans-1,3-dichloropropene, 72.6 ( 5.6; cis-1,3-dichloropropene, 56.7 ( 4.7; cis-1,4-dichloro-2-butene, 97.3 ( 1.3; and trans-1,4-dichloro-2-butene, 77.9 ( 4.3. Acrylonitrile was found to degrade at a rate of 47 ( 25% per day for a 10-g tissue sample. Acrylonitrile has been found to bind to proteins irreversibly, also a likely source of degradation in fish tissue.9 The concentration of pentachloroethane used in this study was near its MDL and its results were not included in Tables 2 or 4. 1,1,2,2-Tetrachloroethane coeluted with compounds codistilled from the tissue and interfered with determination of its response. This analyte was not used in the summaries presented in Tables 2 and 4. Both analytes are listed in Supporting Information. The basic semivolatiles were generally the poorest performing group of analytes in this study. Surprisingly the performance data for this group were best for 10-g samples (using standards in water) and using model 2 for surrogate correction of matrix effects. The performance data for this group would most likely be improved by using basic semivolatile surrogates (there were none used in this study). The soluble volatiles were the poorest performing analyte group for 10-g samples. The performance of this group would most likely be improved by narrowing the range of RKw values that a given surrogate pair represents (adding surrogate pairs and new ranges for corrections). These two groups of analytes have low BCFs (a lower priority for this study), and additional surrogates were not investigated. CONCLUSION The development of fish tissue samples in which analytes have demonstrated equilibration provides a medium to develop and (8) Xu, D.; Peter, H.; Hallier, E.; Bolt, H. M. Ind. Health 1990, 28, 121-123. (9) Geiger, L. E.; Hogy, L. L.; Guengerich, F. P. Cancer Res. 1983, 43, 30803087. (10) U.S. Environmental Protection Agency. Handbook of RCRA Ground-Water Monitoring Constituents: Chemical & Physical Properties; Office of Solid Waste, GPO: Washington DC, 1992. (11) Lyman, W. J.; Reehl, W. F.; Rosenblatt, D. H. Handbook of Chemical Property Estimation Methods; American Chemical Society: Washington DC, 1990; Chapter 1.
1134 Analytical Chemistry, Vol. 69, No. 6, March 15, 1997
evaluate methods for fish analyses. The analysis of these samples demonstrates VD/GC/MS using surrogate-based matrix corrections can be applied to difficult matrixes with an accuracy and sensitivity comparable to analysis of reagent water. The measurement and correction of the most extreme matrix effects using calibration standards in water verify the correctness of the models. The method is shown to be applicable to both simple and difficult samples without the need for special calibration requirements. The analyst has two surrogate correction models to compensate for matrix effects on analyte response. Model 1 is appropriate for most analytes when 1-g samples are analyzed. Model 2 is recommended for larger samples where lower MDLs are required. The calibration of the VD/GC/MS system is adequately performed with standards in water. ACKNOWLEDGMENT The EPA, through its Office of Research and Development (ORD), funded and performed the research described here. It has been subjected to the Agency’s peer review and has been approved as an EPA publication. Mention of trade names or commercial products does not constitute endorsement or recommendation for use. The U.S. Government has the right to retain a nonexclusive, royalty-free license in and to any copyright covering this article. SUPPORTING INFORMATION AVAILABLE The large amount of data used in the development of this work could not be presented in its entirety due to space restrictions but additional data are available. The data (by analyte) used to prepare Tables 2 and 4 are available as two additional tables. Figure 1 represents the equilibrium profile for only one compound. The corresponding data for all of the analytes are available as three additional tables. The labeled-unlabeled analog data used to evaluate the effectiveness of the sonication spike of 1 g tissue and the vacuum spike of the 10 g tissue samples are available as two additional tables (21 pages). Ordering information is given on any current masthead page.
Received for review September 16, 1996. January 15, 1997.X
Accepted
AC960936J X
Abstract published in Advance ACS Abstracts, February 15, 1997.