Anal. Chem. 1086, 58, 2041-2048
2041
Determination of Selected Neutral Priority Organic Pollutants in Marine Sediment, Tissue, and Reference Materials Utilizing Bonded-Phase Sorbents Robert J. Ozretich* and William P. Schroeder
US.Environmental Protection Agency, Environmental Research Laboratory-Narragansett, Pacific Division-Marine Science Center, Newport, Oregon 97365
An extraction procedure utiiirlng sonkation with acetonitrlie and cleanup wlnq aminopropyl and/or C-18 bonded-phase columns was developed to prepare marine sediments and animal t l w e s for priority organic pollutant analysis. Recoveries from standard referewe and interlaboratory comparison sediments and tissue preparationscompared favorably to puMbhed mean valuer. The recovery efficlencles of the procedwer were determined by opwdng marine sediments and a marine gnimal-tissue homogenate. Mean recoveries of 22 prlorlty organic pollutants from the sediments ranged from 0% to 84% wtth a median recovery of 71% and an average percent relative standard deviation (%RSD) of 9%. Mean recoveries of 13 priorky organlc pollutants from the tissue homgmate ranged from 7% to 76% with a median recovery of 64 % and an average % RSD of 5 % Effects of sediment type and slorage method on the spike recoverles are discwsed.
.
Quantifying synthetic organic compounds in complex natural matrices has been an ongoing challenge for environmental chemists. The range in complexity of published methods atteats to the difficult task of obtaining accurate and precise measurements of contaminant concentrations. Recent attempts to simplify and standardize the extraction of contaminantsfrom fish tissue (1,2)and sediments (2-4) have been published but time- and solvent-consuming procedures with mixed-solvent elutions from a variety of cleanup steps are still the rule. Thia study presents extraction and cleanup procedures that use a single solvent and prepackaged bonded-phase silicas for the recovery of neutral priority organic compounds from marine sediments, animal tissue, and reference materials.
EXPERIMENTAL SECTION Tissue Extraction. Tissue was thoroughly mascerated in a blender with distilled water added (if necessary) to obtain a homogeneous slurry. Five grams of slurry and 25 g of anhydrous NafiOl were ground in a tared glazed alumina mortar until dry and powdery. The powder was transferred to a 50-mL beaker by stainless steel spatula and Pastuer pipet with 2-10 mL acetonitrile rinses of the mortar and pestle. The contents of the beaker were initially sonified for 3 min at maximum power (Model 300 with macro tip, Fisher Scientific Co., Pittsburgh, PA). The solids in the beaker were allowed to settle briefly, and the supernatant was decanted into a 100-mL culture tube and 15 mL of acetonitrile was added to the beaker and the contents were sonified for 1 min and decanted. The diluting, sonifying, and decantingsteps were repeated a third time. The culture tube with approximately 50 mL of solution was capped with an aluminum-foil-lined lid and centrifuged at 2500 rpm (1400g)for 15 min. The supernatant was decanted into a 100-mL volumetric flask, brought to volume with acetonitrile, and thoroughly shaken. The flask WBB refrigerated for at least 2 h (usually overnight) to reduce the lipid concentrationby precipitationand then allowed to warm
to room temperature before 50 mL was transferred by volumetric pipet to a 100-mL round-bottom evaporating flask. The volume was reduced to approximately2 mL by use of a rotary evaporator with a vacuum of 635 mmHg and a temperature of 40 OC. The extract and four 1-mL rinses of the flask were transferred to a conical vial using a Pasteur pipet. Extract Cleanup. C-18 and aminopropyl bonded-phase, disposable, 500-mg columns with polyethylene frits (Bond Elut, M y t i c h e m International, Harbor City, CA) were used in tandem to clean the extracts. The C-18 column was used to retain lipids, and long-chained hydrocarbons while the aminopropyl column was used to retain more polar interferences such as amines and organic acids. Using these columns in tandem (C-18 atop aminopropyl) effectivelyremoves many polar and nonpolar interfering materials in a one-step cleanup. The effect of column order on cleanup was not evaluated, although the ion exchange capacity of the aminopropyl column could be compromised by column loading from nonpolar material if it preceded the C-18 column. The columns were conditioned by passing 10 mL of acetonitrile through precleaned, C-18 columns atop aminopropyl columns and into an evacuated box (380 mmHg) (Vac Elut, Analytichem International). The tissue extract in the conical vial was transferred by Pasteur pipet with two 1-mL acetonitrile rinses to the (2-18 column and was drawn through both columns into a 10-mL volumetric flask within the box at a vacuum of 50-100 mmHg. Two 1-mL acetonitrile rinses of both columns preceded opening the box and bringing the volume to 10 mL with a methyl stearate (MS) internal standard solution and acetonitrile (MS final concentration, 1.0 mg/L). The extract solutions were thoroughly mixed and "es were dispensed by Pastuer pipet to crimptop gas chromatography (GC) vials. Some of the extracts to be quantified by GC/MS were exchanged to issoctane after the last column rinse by reducing the volume to 0.2 mL with nitrogen gas, adding distilled water (0.2 mL) and isooctane (0.55 mL), and mixing vigorously. Reducing the volume of the extracts is recommended if lower detection limits are desired as larger sample sizes could result in breakthrough of interfering biogeneous material. We have measured no losses of 16 polynuclear aromatic hydrocarbons (SRM 1647, National Bureau of Standards) during volume reduction with nitrogen gas at room temperature. The isooctane was drawn-off with a pipet and added to a GC vial where phenanthrene-& was added as the internal standard (1.0 mg/L) for on-column injection. The reason for using only half the extract supernatant and a 500-mg column was that column breakthrough by the retained interfering materials was observed when all the supernatant or a 200-mg column was used. Column Cleanup. Initially, the C-18 columns were found to contain contaminants after following the manufacturer’s recommended “preparation” step of passing methanol through the columns. These unidentified contaminants were eliminated by drawing 10 mL of acidic methanol (0.5%, v/v, concentrated HCI, pH 2-3) and air for 1min at 380 mmHg. The acidic rinse of the C-18 columns used for tissue cleanup was followed by 2 mL of unacidified methanol before attaching them atop the aminopropyl columns because the acidic solution displaced from the dead volume of the C-18 column would have reduced the ion exchange capacity of the aminopropyl columns by protonating the amine groups. Sediment Extraction. Wet sediment samples were thoroughly mixed with a stainless steel spatula. From this point on the
Thls article not subject to U S . Copyright. Published 1988 by the Amerlwn Chemical Society
2042
ANALYTICAL CHEMISTRY, VOL. 58, NO. 9, AUGUST 1986
Table I. Characteristics of the Sediments Used in Experiments 1,3, and 4 % composition
location
sediment type"
sandb 0.063-2 mm
shipping channel (SC) Kings Slough (KS) deep disposal (DD)
well-sorted fine-grained sand poorly sorted coarse silt very poorly sorted sandy mud
96 35
a
12
% total volatile solids
siltb 0.004-0.063 mm
clayb 0.05). Analysis of variance (ANOVA) was used to detect differences among the various treatments on single compounds. When significant differences among treatment means were detected by ANOVA, multiple comparisons of means were made using Tukey's significant difference method (6) to determine where the differences occurred. Linear regression analysis was used to determine whether mean recoveries obtained by our method were dependent upon contaminant concentration. Student's t tests were used to compare mean concentrations obtained by our method with those obtained by other methods used in various laboratories. Interlaboratory Comparison Samples. We obtained sediment (NOAA, NAFC) and mussel homogenate (EPA, ERL-N) samples that had been characterized in the interlaboratory comparison studies of MacLeod et al. (3) and Galloway et al. (7), respectively. The sediments and homogenate were received frozen
ANALYTICAL CHEMISTRY, VOL. 58, NO. 9, AUGUST 1986
2043
Table 11. Analysis of Intercalibration Samples from the Study of MacLeod et al. (3)
compound naphthalene fluorene phenanthrene anthracene fluoranthene pyrene benz[a]anthracene chrysene
Duwamish I sample mean concn, mg/kg (dry) X (% RSD) this study tumbleP Soxhlet4 n=3 n = 11 n=4 NDb ND 0.19 (36) ND 0.52 (21) 0.46 (15) 0.187 (25) 0.28 (24)
ND 0.030' (30) 0.330' (28) 0.057' (26) 0.57 (23) 0.76O (21) 0.44e (23) 0.270 (20)
this study n=3
ND
ND
0.050 (50)
0.050 [0.014]
0.61 (44) 0.12 (50) 0.84 (40) 1.10 (38)f 0.9 (71)' 0.53 (48)
0.477 [0.019] 0.203 [0.005] 1.48 [0.103] 1.24 [0.079] 0.516 [0.021] 0.82 [0.043]
Duwamish I1 sample mean concn, mg/kg (dry) x [Sf] tumbler 5c 0.049 [0.017] 4d 0.108 [0.004] 5 0.69 [0.132] 4 0.28 [0.031] 2 1.7 [0.32] 1 1.4 [0.37] 3 0.8 [0.35] 2 0.96 [0.058] 2
Soxhlet 2c
0.063 [0.001] 2 0.120 [0.028] 2 0.78 [0.042] 2 0.30 [0.085] 2 1.85 [0.212] 1 1.50 [0.141] 1 1.0 [0.30] 2 1.5 [0.35] 2
From Brown, et al. (9). ND, not detected. Number of included labs. Number of labs whose means from triplicate samples differed from this study (P< 0.05). 'Mean of this study was significantlydifferent from that of Brown, et al. P < 0.05. f%RSDsignificantly greater than this study (P< 0.05). and subsequentlywere thawed with three, 5.0-g subsamples taken for extraction and water-content determination. The sediment extracts were quantified by GC/FID, and the mussel extracts were exchanged to isooctane for on-column injection and GC/MS quantification. Reference Samples. Well-characterized water pollution quality control samples were obtained from the U.S. EPA, EMSL-C. These freeze-dried samples included PCB's in Fish, PCB's in Sediment, and Pesticides in Fish. Triplicate 3.0-g subsamples of the PCBs in Sediment and Pesticides in Fish were processed without solvent exchange. Triplicate 0.75-g subsamples of the PCBs in Fish sample were extracted with solvent exchange. A freeze-dried riverine sediment was obtained from the National Bureau of Standards. It was made available as a trace metal reference material (SRM 1645)after multilaboratory analysis. Subsequently,the concentrations of a number of priority organic pollutants have been reported (4). Triplicate 3.0-g subsamples of the riverine sediment were processed with solvent exchange and copper reduction of elemental sulfur. All reference sample extracts were quantified by GC/MS. Experiment 1. We evaluated the effects of sediments with different grain sizes and water content on recovery and the day-to-day variability in recoveries obtained using our extraction and analysis procedures (questions A and B). On three separate days (days 1-3) each of the three sediments (SC, KS, and DD) was spiked at 2.5 mg/kg (wet) with seven compounds. On day 4 test sediments were spiked with an additional 15 compounds. The 22 compounds were chosen as representatives of the classes of compounds that have been reported in the waterways of Tacoma Harbor, Commencement Bay, WA (8). Days 1 4 sediments were extracted the same day they were spiked. Two sediments, KS and SC, were spiked one day with the seven compounds, stored at 4 "C, and extracted the next day (day 5) (question C). Sediment DD was inadvertently not spiked and left overnight. Several extracts were split into three GC vials and analyzed at different times during a GC run to determine precision at the instrument level. Experiment 2. We spiked a homogenate of English sole at 2.5 mg/kg (wet) with 13 neutral priority organic pollutants to determine if there were differences in recovery between tissue and sediment matrices when using our method (question A). Experiment 3. We evaluated the effecta of storage temperature and duration on the recovery efficiency of our method (question C) by spiking 100 g of SC sediment at 2.5 mg/kg (wet) with 13 neutral priority organic pollutants. Twenty grams of spiked sediment was placed into each of five alurrrinum-foil-cappedjars. Three jars were refrigerated (4 "C) and two were frozen (-20 "C). Subsamples from jars at 4 "C were extracted after 3 h and from jars at both temperatures after 3 and 7 days to evaluate recovery over longer storage times than tested in experiment 1. Experiment 4. We determined if recovery was dependent on concentration with this method (question D) by spiking SC sediment with 13 neutral priority organic pollutants at concentrations representing severely to moderately contaminated sediment [2.5, 1.0, 0.5, and 0.25 mg/kg (wet)].
RESULTS A N D DISCUSSION I n t e r l a b o r a t o r y Comparisons w i t h Contaminated Sediments. We obtained two different sediments (Duwamish I and 11) which were used in an intercalibration study described by MacLeod et al. (3).Of the compounds found by MacLeod et al., only eight polynuclear aromatic hydrocarbon compounds corresponded to.compounds that we routinely analyzed, Table 11. Naphthalene, fluorene, and anthracene were not detected in the Duwamish I sample. Recovery of the remaining five compounds averaged 71% of the recovery by ball-mill tumbler extraction, and individual recoveries were not significantly different (P> 0.05) from the recoveries by Soxhlet extractions. Our percent relative standard deviations (%RSD's) were indistinguishable from those of the tumbler method and significantly less than two of the five possible Soxhlet comparisons. We did not detect naphthalene in the Duwamish I1 sample. MacLeod e t al. compared the variances of the recoveries obtained by eight participating labs and excluded two Soxhlet-using labs from further evaluation of the Duwamish I1 sediment because of their consistently large variances. Our variances were similar to those of the remaining six labs. With the exclusion of naphthalene our average recovery was 77% of the tumbler method and 65% of the Soxhlet methods with the recovery of three of the seven compounds being within one standard error of the grand mean of the two methods and six labs. The effectiveness of the C-18 cleanup is shown in a chromatogram of the Duwamish I1 sediment extract, which has a flat baseline (Figure 1). MacLeod et al. (3) showed a chromatogram of the Duwamish I1 extract resulting from their tumbler extraction and silica gel cleanup procedure (9). Their chromatogram had a "hump" called the unresolved complex mixture (UCM) that contained peaks of unidentified compounds coinciding with their analyte compounds. They rechromatographed their silica-gel-cleanedextract on Sephadex LH-20 and effectively eliminated the UCM, producing a chromatogram that looked very similar to Figure 1. They recommended this combined procedure to facilitate more precise comparisons of hydrocarbon extractions in the future. We believe that our single solvent extraction and one-pass cleanup procedure provides equally interference-free extracts for the determination of aromatic compounds in contaminated urban sediments. Furthermore, the time and money saved in the rapid processing of samples by our method could offset the apparent reductions in recoveries when large survey studies were undertaken which sought to discern spatial or temporal differences in concentrations rather than absolute concentrations.
2044
ANALYTICAL CHEMISTRY, VOL. 58, NO. 9, AUGUST 1986
Table 111. ERL-N Mussel Homogenate I: Intercomparison Results compound
mean concn,O mg/kg (dry), X (%RSD) lab B lab C lab D
lab A
this study
PCB's as Aroclor 1254
0.470 (10) 0.90 (31)' 0.412 (6) 0.51 (27)c 0.559 (7) (12Y (10) (4) (3) (3) naphthalene 0.005 (40) 0.10 (123)' 0.003 (29) 0.036 (32) phenanthrene plus anthracene 0.013 (38) 0.032 (125)' 0.008 (20) 0.0216 (23) 0.056 (32) fluoranthene 0.042 (88)' 0.080 (15) 0.075 (15) 0.046 (28)' pyrene 0.034 (91)' 0.092 (15)' 0.0615 (5) benz[a]anthracene plus chrysene 0.029 (21) 0.028 (114)c 0.047 (13) 0.059 (24) [61b [IO1 [41 [31 "From Galloway et al. (7). *Parentheses denote n for PCB's. Brackets denote n for PAH's. %RSD significantly greater than this study (P < 0.05). Table IV. Recovery of Pesticides and PCB's from Reference Samples compound
reference sample
DDE
pesticides in fish 3
DDD
pesticides in fish 3
DDT
pesticides in fish 3
PCB 1254
PCB's in sediment 2
PCB 1254
PCB's in fish 1
EPA-EMSL-C" this study 18.6 f 4.2 20.0 f 1.0 6.8 f 2.7 8.0 f 0.4 4.2 f 1.1 7.5 f 0.3 2.34 f 1.24 1.62 f 0.24 3.12 f 1.32 2.20 f 0.08
EPA-EMSL-C this study
reference sample pesticides in fish 4
1.72 f 0.55 2.20 f 0.14 0.70 f 0.25 0.89 f 0.02 0.68 f 0.34 0.47 f 0.02 6.48 f 1.12 8.87 f 0.46
pesticides in fish 4 pesticides in fish 4 PCB's in sediment 3
aReported mean concentration (mg/kg), 8 f Sa.
Interlaboratory Comparison with Contaminated Mussel Homogenate. We obtained a mussel homogenate from the US.National Mussel Watch Program for which the concentrations of a number of polynuclear aromatic hydrocarbons (PAH's) and PCB's had been determined in an interlaboratory intercomparison study [Galloway et al. (7)]. PCB's as Aroclor 1254 and seven PAH's (Table 111) were in common with our routine analyses. With the exception of benz[a]anthracene plus chrysene our mean concentrations were within the range of mean values reported by the four laboratories with each using different techniques. In 10 of the 19 comparisons of relative standard deviations, our resuits were significantly less ( P < 0.05) and the remainder were indistinguishable from the other laboratories. EPA Quality Control Reference Comparisons. The fBh and sediment quality control samples were a t two levels of contamination for the pesticides and PCB's. With the exception of DDT in fish sample 3, Table IV, our recoveries of DDE, DDD, and PCB 1254 using 0.75 g or 3 g per sample were not significantly different (P> 0.05) from the reported mean values obtained from the recommended 5-7 g sample sizes.
NBS Standard Reference Material Determination. Concentrations determined in NBS SRM 1645 (river sediment, certified for metals only) by our method and that of LopezAvila et al. (4) are given in Table V. Significant differences in concentrations were found for the majority of compounds. Although the retention times of all our peaks were within 0.1 min of the standards, the extracts were reanalyzed by use of the primary and two secondary electron impact ions of those compounds for which our concentrations exceeded (by a factor of 1.5 or more) those found by Lopez-Avila et al. The ratios of the secondary to primary ions deviated less than 20% from the standard ratios indicating that the quantifying primary ions had not been diluted by different, coeluting compounds. Since our method blank for the samples contained none of these compounds, an explanation for the positive deviation could lie in sample heterogeneity or in a reduced recovery of these compounds by Lopez-Avila et al.
5
10
15
20
TIME (min)
Flgure 1. Capillary gas chromatogram of Duwamish I1 extract.
Although the sample mass that we extracted was smaller than that of Lopez-Avila et al. (3 g vs. 20 g), it is unlikely that sample heterogeneity could explain the differences in concentrations as the bulk sample is sufficiently fine and homogeneous to permit the NBS-recommended sample size of 0.1-2 g for metals. The complexity of the method used by Lopez-Avila e t al. is typical of many published methods. It consists of extracting a water/sediment slurry a t two pH values with dichloromethane in a high-speed homogenizer,
ANALYTICAL CHEMISTRY, VOL. 58, NO. 9, AUGUST 1986
Table V. Standard Reference Sediment, SRM 1645 Intercomparison Results mean concn, mg/kg dry, X (%RSD) ref 4 this study compound
n=3
n=3
naphthalene hexachlorobutadiene acenaphthylene acenaphthene fluorene diethyl phthalate hexachlorobenzene phenanthrene anthracene di-n-butyl phthalate fluoranthene PYene n-butyl benzyl phthalate benz[a]anthraceneplus chrysene bis(2-ethylhexyl) phthalate benzo[a]pyrene P~P-DDE p,p-DDD P~P-DDT PCB as Aroclor 1254 benzo[k,h]fluoranthene
0.88 (40)d
0.254 (16)
ND"
ND 0.092 (14) 0.173 (6) 1.93 (4)
1.82 (19) 0.85 (14)d 1.58 (8)
ND ND
ND ND
3.77 (8)d 3.06 (3) 2.09 (3) 0.80 (9) 0.58 (11) 0.66 (14) 14.8 (19)d 27.5 (5) 20.0 (23)d 58.7 (5) 0.92 0.51 (13) 27 (27)d*e 104 (6) 0.58 (15) 9.2 (25) 23 (27)d 75 (7) 0.34 (29)d 0.447 (7) 0.12 0.83 (11) 0.840 (2) 3.24 (2) 1.23 (34) NRb N$' 31.6 (8)
a Not detected. Not reported. Not quantified. %RSD significantly greater than this study (PC 0.05). eChrysene only.
followed by extract cleanup by silica gel chromatography utilizing four solvent mixes with varying percentages of hexane, dichloromethane, and acetone. In demonstrating the distribution of compounds in the various silica gel elutions and their method's overall recovery, Lopez-Avila et al. also reported the results of spiking 42 base/neutral compounds onto their silica gel cleanup columns and into slurries of SRM 1645 and water. The best recovery and precision that could be attained by a method would be in the absence of extraneous, interfering compounds. For the Lopez-Avila et al. method this would be represented by the recovery of compounds from solutions placed directly on their silica gel columns. The reported recoveries for the compounds in Table V (from silica gel columns and in the absence of sample-originating interferences) ranged from 51% to 142% with %RSD's from 7% to 64%. The lowest recoveries and highest relative standard deviations tended to be associated with those compounds that we found in greater concentrations, e.g., fluoranthene and benzo[a]pyrene with percent recoveries and
2045
(%RSD's) of 51% (39) and 70% (49), respectively. LopezAvila et al. found higher concentrations of the lower boiling compounds for which we had severely reduced recoveries, e.g., naphthalene 9% to 21% recovery (see experiments 1-4). However, it is likely that the true concentrations of many of the higher boiling neutral priority pollutants in SRM 1645 are closer to those we report than those reported in LopezAvila et al. because of the inconsistent recovery and the generally higher uncertainty of their method. Experiment 1. The %RSD from triplicate injections of extracts ranged from 0.3% to 8.5%,Tables VI and VII. The KS sediment on day 3 was apparently spiked twice with the spiking solution containing hexachlorobenzene and anthracene because recoveries averaging 143% were found. These recoveries were not used in comparisons involving the KS sediment. The day-to-day recovery from sediment spiked and extracted on four different days (days 1-4) was evaluated for each of the seven compounds in each sediment using one-way ANOVA. Significant differences (P < 0.05) were found for most of the seven compounds (Table VI). Multiple comparisons tests revealed that day 1 of the series had significantly lower recoveries than the others. The order of extraction on day 1 of experiment 1 was SC, KS, and DD and the recoveries improved during the day as is evident in the increasing recoveries as the day progressed when compared to the mean of days 2-5 (Table VI). This is a likely result of starting the first experiment with techniques that were not optimal but that improved through the day and remained good throughout the remainder of the experiment. We feel that some of the recoveries on day 1 of this experiment are not representative of the expected range in recoveries which are obtainable in the routine application of our method. These low recoveries are examples of outliers in a quality control sense and as such all of the data from day 1 were not included in subsequent comparisons. The results from day 1 illustrate the importance of making sure the procedures for any method are thoroughly practiced and result in consistent recoveries from spiked or reference materials before irretrievable samples are analyzed. The day-to-day (days 2-5) and sediment effects on the recovery of the seven compounds in Table VI were tested using two-way ANOVA. There were no significant differences ( P > 0.05) for any of the compounds among the four treatment days, which included the samples left overnight (day 5). Significant differences (P < 0.05) in recovery were found among the sediment types for the three most volatile compounds with low recoveries of naphthalene and fluorene from SC sediment and isophorone from DD. For anthracene and
Table VI. Recovery of Compounds Spiked at 2.5 mg/kg (Wet) from Different Sediments and a Tissue Homogenate
compound
isophorone
instrument precision, % RSD
sc
0.4
51" (25)b 1
,e
I L
naphthalene fluorene
hexachlorobenzene anthracene PPene
chrysene
e
2.0 1.0 1.0 1.0
0.9 2.3
9 (4) 19 62 (35) 3 70 (40) 4 71 (40) 4 85 (49) 5 85 (51) 6
mean % recovery KS 51 (33) 17 20 (9) 24 72 (57) 8 71 (60) 11 66 (59) 10 84 (69) 8 85 (72) 9
DD 41 (41) 12 19 (16) 16 72 (65) 4 73 (66) 7 71 (66) 7 83 (81) 6 83 (76) 5
mean sediment, % RSD
mean % recovery sediment E. sole homogenate 48e f 6d
13 f 8d 15e f 6
20 h 9 6ge f 5 6f6 71
f
4
7f6 69 f 5 7f6 84 f 5 6f5 84 f 4 6f6
61 13 7 19 65 8 62 8 64 2 7 If 1 76 3
" Days 2-5 except DD with days 2-4. Day 1. e Mean %RSD from a given matrix. Mean (not including day 1) and its standard error. Significant (0.05 > P > 0.025) differences among means. f Significant difference between mean sediment and homogenate recoveries.
2046
ANALYTICAL CHEMISTRY, VOL. 58, NO. 9, AUGUST 1986
Table VII. Recovery of Compounds Spiked at 2.5 mg/kg (Wet) from Different Sediments and a Tissue Homogenate
compound
instrument precision, % RSD
hexachlorobutadiene dimethyl phthalate
2.0
diethyl phthalate
2.0
phenanthrene di-n-butyl phthalate
0.4 0.9
mean % recovery sc KS DD 0
0
0
65 46 61 4 76 3 68
59 5 56 5 77 4 53 5 76 4 83 4 79 4 71 8 78 4 770 3 64 5 88
55 6 52 5 79
1
P,P-DDE P~P-DDD P~P-DDT n-butyl benzyl phthalate benz[a]anthracene bis(2-ethylhexyl) phthalate
0.3 2.0 1.7 2.6 2.0
0.7
78 3 84 3 81 3 63 3 80 3 83" 2
di-n-octyl phthalate
7.6
Aroclor 1242
2.6
Aroclor 1248 Aroclor 1254
3.6 8.5
65 3 80 20 80 20 85 15
mean % recovery sediment E. sole homogenate 0
5f
4f2
12
6f5
83
f
5gd f 8
64 2
77 f 1
f f
83 f 1 6f5
12
73
78 f 4 6f5
12
68 3 76 12 77" 4 63
6f5
2
3f2
24
77 f 2
5 f l
56 5 78
25
56d f 5
60 6 62 8
le
6f5
89 21 80
f
Sod f 5?
12
66 23 62 18 65
22
mean sediment, % RSD
67 f 4
70
5f3
1
78 f 2
f
79d f 3
58e
64 f 1
67 3
78 f 11
f
72 f 14
f
77 f 10
f
3fl
1
22 f 2
20 f 2 21 f 6
"Bis(2-ethylhexyl) phthalate was spiked at 5.0 mg/kg (wet). %RSD from a given matrix. @Meanand ita standard error. dSignificant (0.05 > P > 0.025) differences among means. eSignificant difference between mean sediment and homogenate recoveries. fCompound not spiked into tissue homogenate. fluorene significant inexplicable interactions between the day of extraction and sediment type were found. One-way ANOVA of the recoveries from among the three sediment types for the remaining 15 compounds (Table VII) indicate that significant differences in recoveries OcCulTed with four of the six phthalate esters. Multiple comparisons testa revealed that the mean recovery from KS and DD sediments were indistinguishablefor the four eaters. The mean recoveries for all four esters from the DD sediment were significantly less than from the SC sediment. On comparison of the KS and SC sediments only the recovery of di-n-butyl phthalate was less in the KS sediment. Recoveries were not significantly different among sediments for the remaining nine compounds. Hexachlorobutadiene was not recovered from any of the spiked sediment. If the low recoveries of isophorone (48%), hexachlorobutadiene (O%), and naphthalene (15%) were due to evaporative losses during the volume reduction and/or sonication steps, then the losses would be proportional to their vapor pressures, which at 20 "C are 0.35 torr (lo), 0.15 torr (ll),and 0.05 torr (lo), respectively. The recovery of isophorone did not follow this relationship as a likely result of its polar carbonyl carbon, which would tend to interact with the polar solvent, acetonitrile, and reduce its volatility. The presence of a carbonyl carbon in isophorone and two less polar carbonyl carbons in the phthalate esters may contribute to the lower recoveries of these compounds from the clay-particle-rich DD sediment. Preferential and strong adsorption of many organic compounds to the finest sediment fraction have been reported (12). The presence of polar carbonyl groups may retard the solvation of those compounds that have chemical interactions through polar groups with the solid surfaces compared to nonpolar compounds' predominant van der Waals interactions (13).
Table VIII. Effect of Time and Temperature on the Recovery of Compounds from Sediment Spiked at 2.5 mg/kg (Wet).
compound isophorone
mean % recovery, n = 3 4 "C -20 OC overall % 0.13b 3b 7b 3b 7b recovery
di-n-butyl phthalate
38 3c 22 6 61 6 48 6 63 7 62 4 55 8 68 8
1
pyrene
62
n-butyl benzyl phthalate
9 74
65 1 65 5 64 3 62 5
naphthalene dimethyl phthalate fluorene diethyl phthalate hexachlorobenzene anthracene
11
chrysene bis(2-ethylhexyl) phthalate di-n-octyl phthalate
61 8 69 6 102 11
42 5 13' 1
64 1 50 2 67 2
63 3 54 1 69
112
14
41 41 f 2d 10 lle llf 131 13' 14 f 4 45 7 11 14 63 62 62 62 f 1 7 8 4 5 54 50 50 50 f 1 1 9 7 1 7 68 64 62 65 f 3 8 1 0 6 7 65 6 6 6 6 62 f 2 4 3 4 4 73' 57 67f 61 f 8 6 6 7 6 65 78 70 70 f 5 10 31 9 12 68 67 67 66 f 2 7 8 8 7 lod 68 93f 80 f 16 4 8 1 2 8 71' 64 66' 65 f 4 7 1 8 5 103f 70 94' 80 f 18 5 2 0 1 1 9 137 77 109 107 f 23 14 2 18 12 42
32
41 5
Except bis(2-ethylhexyl) phthalate which was at 5.0 mg/ kg (wet). bDays exposed. %RSD. dMean percent recovery and its standard error. e Mean %RSD. f Significant differences between treatment and 0.13 day control.
ANALYTICAL CHEMISTRY, VOL. 58, NO. 9, AUGUST 1986
d
the highest variability typically coming from the lowest spiked concentrations (Table IX). Linear regression analysis with t tests of the resulting regression coefficients showed a significant dependency (P < 0.05) of recovery on concentration for three of the six phthalate esters and chrysene. Problems with di-n-octyl phthalate contamination were encountered and the bis(Bethylhexy1) phthalate found in the unspiked sediment (0.16 mg/kg) was subtracted from the recovered concentrations. The large %RSD in the recovery of 0.25 mg/kg naphthalene and hexachlorobenzene was a result of two of the subsamples being undetected a t this level because of the low fecovery of naphthalene and the FID’s low sensitivity to hexachlorobenzene. It is likely, but not tested, that these reductions in recovery could have been mitigated by using larger sample sizes or concentrating the extracts to smaller final volumes.
d
CONCLUSIONS
Table IX. Recoveries from SC Sediment Spiked at Several Concentrations
compound isophorone naphthalene dimethyl phthalate fluorene
mean % recovery (n = 3) for the analysis following spike concnsa of regres2.5 1.0 0.5 0.25 variance sion coeff 41 54 42 32 lgb 5 21 42 11 21 11 6 38 6 60 173 64 67 56 44 3 3 6 8
anthracene
54 65 53 49 8 2 3 1 0 66 73 59 43 2 5 1 9 54 59 45 12 8 1 8 173 61 65 56 59
di-n-butyl phthalate
4 75
diethyl phthalate hexachlorobenzene
7
d
e
d
1
5
61 45 6 8 10 pyrene 73 75 69 2 5 6 1 butyl benzyl phthalate 69 58 49 3 11 12 chrysene 72 69 65 4 914 bis(2-ethylhexyl) 73 77 82 phthalate 3 9 6 1 di-n-octyl phthalate 81 84 c
44 62
d
57
d
6 43
d
e
d
e
4
4
e
5
59
2047
4
89 2 c
21
Double these concentrations for bis(2-ethylhexyl) phthalate, * %RSD. Samples contaminated. dSignificantdifferences in recoveries, P < 0.05. e Significantly different from zero, P < 0.05. mg/kg (wet).
Experiment 2. Significant differences between the mean recoveries from the spiked tissue homogenate and the mean recoveries from the three sediments were found for 2 of the 13 compounds. Pyrene and bis(2-ethylhexyl) phthalate recoveries from the fish homogenates were lower than from the sediment (Tables VI and VII). Experiment 3. Two-way ANOVA revealed that storage time and/or temperature had significant (P < 0.05) effects on the recovery of 6 of the 13 compounds from spiked sediment (Table VIII). The recoveries of anthracene, n-butylbenzyl phthalate, chrysene, and bis(2-ethylhexyl) phthalate were significantly greater at 7 days but not after 3 days of storage at both temperatures. Significant reductions were found for naphthalene at both temperatures and times, and for hexachlorobenzene, only the frozen samples were lower. Although evaporative losses of naphthalene and the high fugacity of hexachlorobenzene in the salt-saturated environment produced by the extruded salt from freezing could explain the lower recoveries of these compounds, we have no explanation for the enhanced recoveries of the other compounds. The high RSD’s and greater than 100% recoveries for di-n-octyl phthalate indicated that the extracts of these samples had been contaminated with a compound often found in the method’s blanks in these experiments. Experiment 4. An ideal extraction and detection procedure would yield consistent recoveries over any concentration range. However, apparent reduction in recovery with increased variability could be expected as the Concentration of contaminant in a fixed final volume (10 mL) approached the instrument detection limit. One-way ANOVA among the spiked concentrations revealed significant differences (P< 0.05) in the percent recovery for 8 of the 13 compounds with the lowest mean recoveries and
The methods described here represent significant reductions in the time, solvent volume, and complexity of procedures typically used to extract, clean, and fractionate sediment and tissue samples for the analysis of low volatility, neutral priority organic pollutants. Our method has been shown to be precise enough to detect differences in recovery due to sediment sample storage techniques, differences in sediment type, and compound class and concentration. The recoveries obtained by our sediment and tissue methods from reference and interlaboratory comparison materials were often statistically indistinguishable from the results obtained from other techniques. The reduction in procedural steps contributes to the high precision of our methods and would make them suitable for monitoring and survey programs where spatial gradients and temporal changes in concentrations were sought and rapid turn around times were necessary.
ACKNOWLEDGMENT We express our appreciation to M. Daves, D. Killian, F. Roberts, and R. Stuart for their technical assistance and to R. Barrick, D. J. Baumgartner, S. Ferraro, J. Lichtenberg, and D. Schults for their comprehensive reviews. &&#try NO.p,pDDE, 72-55-9;p,p-DDD, 72-54-8;p,p-DDT, N29-3; Aroclor 1242,53469-21-9;Aroclor 1248,12672-29-6;Aroclor 1254, 11097-69-1; isophorone, 78-59-1; naphthalene, 91-20-3; fluorene, 86-73-7; hexachlorobenzene, 118-74-1; anthracene, 120-12-7; pyrene, 129-00-0; chrysene, 218-01-9; hexachlorobutadiene, 87-68-3; dimethyl phthalate, 131-11-3; diethyl phthalate, 84-66-2; phenanthrene, 85-01-8; di-n-butyl phthalate, 84-74-2; n-butyl benzyl phthalate, 85-68-7; benz[a]anthracene, 56-55-3; bis(2-ethylhexyl)phthalate, 117-81-7;di-n-octyl phthalate, 117-84-0.
LITERATURE CITED Erney, D. R. J . Assoc. Off. Anal. Chem. 1983, 4 , 969-973. Interim Methods for the Sampllng and Analysis of Rloriv Pollutants in Sediments and fish Tissue; USEPA, EMSL: Cincinnati, OH; EPA 600/4-81-055. Macled, W. D.; Prohaska. P. G.; Gennero, D. D.; Brown, D. W. Anal. Chem. 1982, 5 4 , 386-392. Lopez-Avlla, V.; Northcutt, R.; Onstot, J.; Wickhem, M. Anal. Chem, 1983, 55. 881-889. Folk, R. J . Geol. 1954, 62, 344-359. Sokal, R. R.; Rohlf, F. J. Biometry, 2nd ed.; W. H. Freeman: San Franclsco, CA, 1981; Chapter 13. Galloway, W. 6.; Lake, J. L.; Phelps, D.K.; Rcgerson, P. F.; Bowen, V. T.; Farrlngton, J. W.; Goldberg, E. D.; Laseter, J. L.; Lawler, 0. C.; Martin. J. H.; Rlsebrough, R. W. Environ. Toxicol. Chem. 1983, 2 , 395-410. Malins, D. C.; McCain, B. 6.; Brown, D. W.; Chan, Sin-Lam; Myers, M. S.; Landahl, J. T.; Prohaska. P. G.; Friedman, A. J.; Rhodes, L. 0.; Burrows, D. G.; Gronlund, W. D.; Hodglns, H. 0. Environ. Scl. Techno/. 1984, It?, 705-713. Brown. D. W.; Ramos, L. S.; Uyeda, M. Y.; Friedman, A. J.; MacLeod, W. D. A&. Chem. Ser. 1980, No. 185, Chapter 14. Weast. R. C. Handbook of Chemistry and Physics, 53rd ed.; Chemical Rubber Co.: Cleveland, 1972. Pearson, C. R.; McConnel, G. R o c . R . SOC.London, B 1975, 189, 305-332.
2048
Anal. Chem. 1988, 58,2048-2052
(12) Theng, 6 . K. G. Formetion and prspertles of Cky-Polvmer Complex8s; Elsevler: New York, 1979 p 362. (13) Stumm, W.; Morgan, J. J. Aquatic Chemistty; Wley-Interscience: New York, 1970; p 583.
RECEIVED for review January 21,1986. Accepted April 7,1986.
This is contribution no. ERLN-NO19 of the Pacific Division (Newport, OR) Of the Research Laboratory a t Narragansett, RI. Mention of trade names or commercial products does not constitute endorsement or recommendation for use.
Evaluation of Graphitized Carbon Black Cartridges for Rapid Organic Trace Enrichment from Water: Application to Priority Pollutant Phenols Claudio Borra,' Antonio Di Corcia,* Marcello Marchetti, and Roberto Samperi Dipartimento di Chimica, Universitd "La Sapienza" di Roma, Piazza Aldo Moro, 00158 Roma, Italy
Trace enrlchment of the 11 prlorky pollutant phenM on a graphltked carbon black (We) cartrldge has been studled In order to evaluate the feadbUlty of tMs swbent In retalnlng very polar compotMdr from aqueous sokltlon. Optlmlzatlon studies of the eluant have been performed In order to recover quantltatlvely all phenols conskkd from the GCB surface. Recovery of phenols was unaffected by the sampllng rate even at a flow rate of 32 mL/mln. The extraction efficiency of the sodwnt wasassemed bysamplhg up to 4 L of dWiUed water. At such water volume sampled, only phenol and In part o-chkrophenol were lost. The mstrbc effect was studled by extractslg phenols rplked in water samples from dlfferent sources. The effklency of the Carbopack cartrdge In trapping phenols was compared wlth that of a C,@bomled phase column. When coupled wlth a HPLC method, samplng of water by the Carbopack cartrklge allows determlnation of phenoleto beperfomed wlth a Itm# ofdetectknof 4-40 parts per trllllon (pptr), except for phenol, wMch can be detected at the level of 250 pptr In surface water.
Up until some years ago, sample pretreatment for isolation and/or enrichment of organic compounds from aqueous solution was invariably done by liquid-liquid extraction. The recent introduction of adsorbents having elevated chemical inertness and reproducible chemicophysical characteristics has prompted searchers to adopt solid-phase extraction of organics from an aqueous matrix instead of liquid-liquid partitioning, especially when organic trace enrichment is necessary. In this case, the solvent extraction method requires large volumes of expensive, toxic, and flammable organic solvents. In addition, the solvent evaporation time is timeconsuming and bias may be introduced in the analysis due to solvent impurities or evaporative losses of volatile sample constituents. Among these adsorbing media of relatively recent introduction, chemically bonded phases for both off-line and on-line extraction (1-4) have received particular attention. Although promising results have been obtained, this class of sorbents has two drawbacks. One is that very polar compounds have small breakthrough volumes. The second limiting factor is 'Present address: Department of Chemistry, Indiana University, Bloomington, IN 47405.
that a relatively low sampling rate has to be used in order to avoid losses of analytes. Vulcan is a well-known example of graphitized carbon black (GCB)having a nonporous surface and being essentially free of chemical heterogeneities. This adsorbent has already been employed for enriching some organic compounds from water (5, 6) as well as for isolating analytes from biological fluids ( 7-9). The object of this work has been that of evaluating the ability of an experimental kit consisting of fine particles of Vulcan packed in plastic tubes in rapid and reliable enrichment from water of very polar compounds. For this purpose, we selected the 11priority pollutant phenols, as they comprise a large range of polarity and they are important environmental pollutants.
EXPERIMENTAL SECTION Apparatus. A 6 cm x 1cm i.d. cylindrical polypropylene tube was one-sixth full with 250 mg of Vulcan. The surface area of this adsorbing material was reported to be about 100 m2/g (10). The particle size range of the sorbent was between 100 and 60 pm. Polyethylene frits were located above and below the GCB bed to hold minute particles in place and keep the chromatographic column intact. All the materials cited above were kindly supplied by Supelco, Bellefonte, PA. The sample reservoir had a narrow opening at the bottom that fitted into the cartridge down to 3 cm of the top of the GCB bed. The extraction cartridge fitted directly into the vacuum manifold below. Vacuum was done by a water pump. No care was taken to ensure reproducible and constant vacuum applied to each water sample. Also, no unwelcome effect was observed if the Vulcan bed was casually allowed to go dry during an experiment. Standards. Analytical standards were from various commercial sources. Organic solvents were of analytical grade (Carlo Erba, Milano, Italy) and were used as supplied. The standard solution used to spike water samples was prepared by dissolving the 11 phenols in methanol at the individual concentration of 0.1 mg/mL. Tetramethylammonium hydroxide (TMAOH)was purchased from Fluka (Bucks, Switzerland). Procedure. The GCB cartridge was cleaned by passing sequentially 3 mL of methylene chloride/methanol (70/30, by volume), 2 mL of methanol, and 5 mL of water acidified with HC1 (pH 2). This acidic pretreatment was necessary to eliminate some chemical heterogeneities on the GCB surface capable to give a pH value of water equal to 10.5 (IO).In this situation, low recovery of some phenols was observed. The water samples spiked with phenols were acidified with HCl (pH 3) and filtered when necessary. After the samples had passed through, 400 pL of methanol was percolated through the extraction column to eliminate water.
0003-2700/86/0358-2048$01.50/00 1986 American Chemical Society
