Environ. Sci. Technol. 2003, 37, 1625-1632
Nondestructive, Minimal-Disturbance, Direct-Burial Solid-Phase Microextraction Fiber Technique for Measuring TNT in Sediment J A S O N M . C O N D E R , * ,† THOMAS W. LA POINT,† GUILHERME R. LOTUFO,‡ AND JEFFERY A. STEEVENS‡ Department of Biological Sciences, Institute of Applied Sciences, University of North Texas, P.O. Box 310559, Denton, Texas 76203, and Engineer Research and Development Center, U.S. Army Corps of Engineers, CEERD-EP-R, 3909 Halls Ferry Road, Vicksburg, Mississippi 39180
We explored a novel technique to deploy solid-phase microextraction (SPME) fibers to nondestructively measure the explosive compound 2,4,6-trinitrotoluene (TNT) and its nitroaromatic (NA) degradation products in laboratory sediment toxicity tests and field sediments in situ. SPME fibers within steel mesh envelopes were exposed statically via direct burial within sediment. Six fiber types (polymer coatings) were tested. Polyacrylate (PA) SPME fiber was sufficiently durable for this application, yielded the lowest detection limits, and exhibited a linear uptake relationship across toxicologically relevant sediment NA concentrations (100-2000 nmol/g dw (20-500 µg/g dw)). Temperature greatly influenced SPME absorption kinetics. Via evaluation of absorption at different temperatures, recommended sampling times needed to achieve steady-state equilibrium were 48 h for room temperatures (23-25 °C) and up to 7 d for cold (5 °C) temperatures. Although a comparison of TNT residues by SPMEs and TNT bioavailability and toxicity in sediments has not been completed, differences in SPME availability of TNT and its degradation products were found between two different TNT-spiked sediments. Our disposable SPME technique was slightly less expensive and as precise as the conventional extraction for total NAs and may prove to be a powerful exposure evaluation tool for assessing the ecological risk of these compounds.
Introduction Solid-phase microextraction fibers (SPMEs) and semipermeable membrane devices (SPMDs) have received considerable attention as tools for measuring organic compounds in complex sample matrixes. The devices are dissimilar in design and rely on different substances to “extract” organic compounds from aqueous solutions or gases: SPMDs contain triolein lipid housed in layflat, thin-walled, low-density polyethylene bags; SPMEs have a microlayer of organic * Corresponding author e-mail:
[email protected]; phone: (940)565-2178; fax: (940)565-4297. † University of North Texas. ‡ U.S. Army Corps of Engineers. 10.1021/es0260770 CCC: $25.00 Published on Web 03/06/2003
2003 American Chemical Society
polymer coating on a thin silica fiber mounted on a syringe applicator. Whereas the use of SPMEs has revolutionized modern analytical chemistry (1), their use in environmental toxicology and risk assessment for assessing the bioavailability of environmental contaminants is still incipient. In applications within these fields, SPMDs and SPMEs are commonly referred to as biomimetic devices since they can be used as surrogates for organisms to estimate the bioavailability of organic contaminants (2-4). With regard to contaminant uptake, biomimetic devices share two features with organisms: they absorb only potentially available compounds and can concentrate them to higher concentrations than the surrounding matrix. In contrast to vigorous strong-solvent extractions, which attempt to remove 100% of the compound from the matrix, biomimetic devices sample only what is dissolved in solution or easily dissociated from other matter (2). SPME- or SPMD-available compounds, in contrast to solvent-extractable compounds, may more accurately represent potentially bioavailable compounds. Although there are several biomimetic devices and techniques, including C18 disks (5-7), weak solvent extractions (8, 9), and Tenax beads (10), SPMEs and SPMDs may be the most promising (11). With regard to measuring most organic contaminants, the SPME technique is simpler, less expensive, and faster than the SPMD method, primarily because of the long extraction times, large samples, and complex purification methods needed for SPMD sampling (12, 13). However, SPMDs continue to be favored because they are more rugged than SPME fibers and can be buried in soils and sediments for intimate contact with matrix porewater and solids (14). Currently, SPMDs are routinely used for measuring sedimentand effluent-associated contaminants in situ (12, 15); SPMEs are largely restricted to the laboratory benchtop. While Supelco’s (Bellefonte, PA; http://www.sigmaaldrich.com) standard syringe-mounted SPME design is extremely beneficial to the analytical chemist (1), its applicator design and unprotected fragile fiber restricts its use as an in situ biomimetic device. Recently, it has been possible to obtain raw lengths of SPME fiber (not mounted for use in the SPME applicator syringe). Use of disposable SPME fiber pieces (extracted in solvent or manually inserted into a GC (3) after exposure) may enable biomimetic SPME sampling in a variety of scenarios including in situ sediments and sediment toxicity bioassays. Although the syringe-mounted SPME could be used to measure potential contaminant availability in sediments (using conventional destructive sampling followed by SPME exposure), deployment of the fiber within the intact microenvironment with minimal sediment disturbance is extremely advantageous for measuring sediment-associated contaminants. SPMEs deployed in exactly the same conditions and during the same time period that organisms are exposed to contaminated sediments may provide a more relevant measure of sedimentassociated compounds, especially for compounds that may be modified or transformed during conventional sampling (due to photosensitivity, oxidizability, etc.). Objectives. The main objective of our study was to develop and demonstrate a technique for using raw SPME fiber to measure organic contaminants directly within whole sediment. We chose to investigate the explosive 2,4,6-trinitrotoluene (TNT) and its nitroaromatic (NA) degradation products, contaminants present in sediments at concentrations up to percentage levels at many military sites (16-18). As bioavailability of these compounds in soil has been found to differ by soil composition (19), the biomimetic SPME VOL. 37, NO. 8, 2003 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
1625
technique may be a powerful tool for use in ecological risk assessment. Although we do not attempt to assess the relationship between SPME residues and bioavailability in this study, this relationship must be established before using SPMEs in a biomimetic manner to assess the risk of TNT and its degradation products in sediment. Before this work began, three goals were established to facilitate development of a biomimetic SPME method for NAs in sediment: (i) Development of a reproducible, precise method to expose SPME fiber to intact whole sediments contaminated at toxicologically relevant concentrations. (ii) Selection of the best fiber coating from the several that are commercially available for measuring nitroaromatics. (iii) Assessment of the effects of temperature, which cannot be controlled during in situ deployment, on the uptake rates and fiber-solution partition coefficients (Kfs).
Experimental Section SPME Fiber Sampling Procedure. Fifty-centimeter lengths of SPME fiber were purchased from Supelco. Six fiber types were obtained (fiber thickness, fiber coating): 85-µm polyacrylate (PA), 50-µm carbowax-templated resin (CW-TR), 70µm carboxen-divinylbenzene (CX-DVB), 70-µm carbowaxdivinylbenzene (CW-DVB), 65-µm poly(dimethylsiloxane)divinylbenzene (PDMS-DVB), and 7-µm poly(dimethylsiloxane) (PDMS). Fibers were cut into 1.00-cm pieces using a double-bladed, stainless steel razor blade apparatus. Fiber pieces were placed in small (2 cm by 2 cm) 60-µm stainless steel mesh envelopes, rinsed with 50:50 HPLC-grade acetonitrile-Ultrapure water (Milli-Q purification system), rinsed with ultrapure water, and allowed to dry at room temperature. The mesh envelope provided a means to safely handle the fragile fibers when deploying into and retrieving from sediment. The 60-µm openings were large enough to allow free passage of porewater and fine sediment particles for intimate contact with the fiber, but were small enough to retain the small diameter (200-300 µm) fibers and keep toxicity test organisms from being inadvertently removed from the sediment upon SPME envelope removal. To expose SPMEs to TNT and its degradation products, SPME envelopes were immersed in TNT-spiked sediments or water (see below). As TNT and its degradation products are photosensitive, all work was conducted under non-UVemitting gold fluorescent lighting. In all experiments, SPMEs absorbed less than 5% (molar basis) of the compounds present. This minimal sampling prevents the SPME absorption process from becoming an exhaustive extraction, thus meeting requirements for its use as a “negligible depletion” biomimetic device (2, 3, 7, 20, 21). After exposing the fiber, the envelope was removed from the matrix, and the fiber was carefully removed from the envelope and placed into an HPLC autosampler vial containing 400 µL of 50:50 HPLCgrade acetonitrile-ultrapure water for 10 min to desorb compounds from the fiber. This desorption time is 10 times longer than the desorption time used in the standard Supelco SPME-HPLC desorption interface (22). We then analyzed the extract via HPLC. Experimental Matrixes and Exposure Conditions. All but one of the sediment experiments were conducted using uncontaminated sediment (“Denton”) collected from the top 15 cm of control ponds at the University of North Texas Water Research Field Station, Denton, TX. Denton sediment was dried for 24 h (105 °C), ground, and sieved (98% purity, purchased from R. Spanggord, SRI International, Menlo Park, CA) were dissolved in pure HPLC-grade acetonitrile and spiked into ultrapure water. In addition to
2,6-diamino-4-nitrotoluene (2,6DANT), which could not be obtained, these compounds are the predominant compounds found in TNT-contaminated sediments and soils (16, 2528). Since solvent concentrations above 1 vol % can affect Kfs (1), spiking levels for these experiments were such that solvent concentrations were 1%. Each replicate consisted of an envelope-protected SPME immersed into spiked water (15 mL) in a capped glass vial. Temperature can affect both steady-state equilibrium exposure times and Kfs (1). In addition to the previous experiments in sediment and water (23 °C), we also investigated absorption at temperatures representing the extremes of temperatures expected to be encountered in most temperate field sediments (5 and 30 °C). These experiments were conducted with ultrapure waters spiked with TNT and TNT degradation products as described previously. Chemical Analysis. SPME and sediment extracts were analyzed via HPLC using a Waters Nova Pak C-18 column with an 82:18 2-propanol-water isocratic mobile phase (fixed UV detector, 254 nm) or Supelco Discovery RP-Amide C-16 column with a 55:45 methanol-water isocratic mobile phase (photodiode array UV, variable wavelengths). Sediment samples (1-5 g, ww) were extracted with 10 mL of HPLCgrade acetonitrile following U.S. EPA method 8330A for analysis of explosives (29). To minimize possible compound degradation, samples were not air-dried before extraction. Sediment sample dry weights were determined via gravimetric moisture determination (dried at 105 °C for 24 h) of a separate sediment sample or the post-extraction sediment. Quality assurance/quality control measures included spiked and method blank analyses for sediment and SPMEs. For spiked sediments and SPMEs, average percent recoveries for the analytes measured ranged from 99 to 109% and from 88 to 99%, respectively. For NA compounds, approximate method detection limits were 0.5-2.5 nmol/g, dw, for sediment and 0.03-0.12 nmol for 1-cm SPME fibers. Data Analysis. The amounts of compounds absorbed by SPMEs from spiked ultrapure waters were plotted against time to generate one-compartment, first-order kinetics (1CFOK) curves using nonlinear regression techniques (21, 30, 31). The following modified 1CFOK equation was used:
SPMEt ) SPMEss(1 - e- ket)
(1)
where SPMEt ) SPME residue (nmol) at time t (h), SPMEss ) SPME residue at steady-state equilibrium (nmol), and ke ) elimination rate constant (h-1). A term such as SPMEss is usually calculated by dividing ke by a model-derived uptake rate constant (ku). As we were more interested in the steadystate residue value rather than a calculation of ku per se, we replaced (ke/ku) with SPMEss. This does not affect the model and is only a minor change for simplicity. Kfs values for the NA compounds were calculated using the following equation (1):
Kfs )
SPMEss C0 V f
(2)
where SPMEss ) SPME residue at steady-state equilibrium (nmol), C0 ) the concentration of compound in solution (nmol/mL), and Vf ) the volume of fiber coating (mL). For the remainder of the data, we used either standard linear regression, ANOVA with a posteriori Fisher’s protected least significant difference (LSD) comparisons, or t-tests (R ) 0.05). SAS/LAB (SAS, version 8.02, Cary, NC) was used to test parametric assumptions and identify outliers. Data were transformed using square root or log10 operations, as suggested by SAS/LAB, to meet the requirements for parametric analysis. In our analyses, transformations were needed because sample variances were positively correlated with
FIGURE 1. (a) SPME TNT, (b) 4ADNT, and (c) 2ADNT residues over time (mean ( SD, n ) 2) for three different SPME fiber types buried in TNT-spiked sediment (440 nmol/g dw, aged 24 h). Data points with the same letter (at 24 h) are not significantly different (Fisher’s protected LSD, P > 0.05). the mean in some untransformed data, making them inappropriate for parametric analyses (32, 33). In particular, the square root transformation is useful for reducing heteroscedasticity in Poisson-distributed data, and the log10 is useful for reducing highly heteroscedastic data (33).
Results and Discussion SPME Selection. Three SPME fiber types (coatings) were excluded from experimentation after preliminary experiments in TNT-spiked sediments. CW-DVB fibers absorbed TNT and its NA degradation products well, but this fiber coating was extremely brittle and flaked off the silica core with handling. The PDMS and CX-DVB coatings absorbed very low amounts of NAs:
