Supercritical Fluid Extraction of Energetic Nitroaromatic Compounds

This paper explores the use of supercritical fluid extraction (SFE), in combination with various analyte collection strategies, for extracting energet...
0 downloads 0 Views 98KB Size
Download PDF
Anal. Chem. 2005, 77, 4241-4247

Supercritical Fluid Extraction of Energetic Nitroaromatic Compounds and Their Degradation Products in Soil Samples R. Batlle,*,† C. Nerı´n,† C. Crescenzi,‡ and H. Carlsson‡

Department of Analytical Chemistry, Arago´ n Institute of Engineering Research i3A, CPS-University of Zaragoza, 50018 Zaragoza, Spain, and Department of Analytical Chemistry, Stockholm University, 106 91 Stockholm, Sweden

This paper explores the use of supercritical fluid extraction (SFE), in combination with various analyte collection strategies, for extracting energetic nitroaromatic compounds and their degradation products from soil samples. The required selectivity has been achieved by a combination of an SFE program and active trapping. Several different collection strategies were tested, using a selection of liquids (methanol, toluene, methyl tert-butyl ether, acetonitrile), inert and solid-phase extraction materials (Nexus, Oasis, LiChrolut), and 1-cm liquid chromatography precolumns (porous graphitic carbon, PGC). The best results were obtained using SFE in combination with a PGC precolumn. This setup allows on-line cleanup of the extract, and comparable results were obtained using either GC-ECD or GC-chemical ionization-MS for confirmatory analysis. The time required for a complete analysis was less than 60 min, and only 1 mL of toluene was needed for a 0.5-g representative sample. In contrast, the EPA standard method 8330 required 18-h sonication and 20 mL of acetonitrile for a 4.0-g sample and further time for sample cleanup and HPLC analysis. The method presented here provides method detection limits in the low-nanogram range, with relative standard deviations lower than 7%. The optimized method has been compared and validated with EPA method 8330 in terms of efficiency parameters such as robustness, accuracy (trueness and precision), and capability of detection. The validation demonstrated that the two analytical methodologies give comparable performance for the determination of nitroaromatic compounds, but SFE is superior for analyzing amine degradation products. In activities such as decommissioning military bases, characterizing hazardous sites or detecting landmines and unexploded ordnance, and forensic investigations, samples often need to be collected and analyzed for traces of energetic nitroaromatic materials. Explosives are unstable compounds that can be degraded or transformed by sunlight, soil microorganisms, or plants in the environment, generally by sequential reduction of nitro groups to amino groups. Thus, there is a need for analytical * Corresponding author. E-mail: [email protected]. Fax: + 34 976 762388. † CPS-University of Zaragoza. ‡ Stockholm University. 10.1021/ac050339+ CCC: $30.25 Published on Web 05/14/2005

© 2005 American Chemical Society

tools that are capable of measuring the concentrations of these chemicals and their derivatives to identify and monitor explosivescontaminated sites.1-5 A range of analytical techniques has been developed to determine energetic materials in soil samples, most of which focus on determining trinitrotoluene.6-11 Supercritical fluid extraction (SFE) has also been applied,12-15 but no detailed studies concerning SFE optimization, selective trapping strategies, or validation have been performed. SFE is a rapid, nonenvironmentally hazardous extraction technique that has gained acceptance as an alternative to conventional extraction techniques, which are mainly based on the utilization of organic solvents. However, its application for extracting moderately polar to polar analytes has been limited due to the nonpolarity of the most commonly used extraction fluid, CO2. Alternative fluids with greater solvating strength have a limited applicability due to their extreme critical parameters. This limitation can been partially overcome by introducing small amounts of organic cosolvents, called modifiers, into the SF stream.16-18 (1) Yinon, J.; Zitrin, S. Modern Methods and Applications in Analysis of Explosives; Wiley: New York, 1993. (2) Kaplan, D. L.; Kaplan, A. M. Appl. Environ. Microbiol. 1982, 44, 757-760. (3) Sheremata, T. W.; Thiboutot, S.; Ampleman, G.; Paquet, L.; Halasz, A.; Hawari, J. Environ. Sci. Technol. 1999, 33, 4002-4008. (4) Hawari, J. In Biodegradation of Nitroaromatic Compounds and Explosives; Spain, J. C., Hughes, J. B., Knackmuss, H., Eds.; CRC Press: Boca Raton, FL, 2000; pp 277-300. (5) Pennington, J. C.; Brannon, J. M. Thermochim. Acta 2002, 384, 163-172. (6) Jenkins, T. F.; Walsh, M. E.; Schumacher, P. W.; Miyares P. H.; Bauer, C. F.; Grant, C. L. J. Assoc. Off. Anal. Chem. 1989, 72, 890-899. (7) Test Methods for Evaluating Solid Wastes (SW 846), Method 8330: Nitroaromatics and Nitramines by HPLC; U.S. Environmental Protection Agency, Office of Solid Waste and Emergency Response; Washington DC, 1994. (8) Echols, R. T.; Christensen, M. M.; Krisko, R. M.; Aldstadt, J. H., III. Anal. Chem. 1999, 71, 2739-2744. (9) Walsh, M. E. Talanta 2001, 54, 427-438. (10) Felt, D. R.; Larson, S. L.; Valente, E. J. Chemosphere 2002, 49, 287-295. (11) Weiss, J. W.; McKay, A. J.; Derito, C.; Watanabe, C.; Thorn, K. A.; Madsen, E. L. Environ. Sci. Technol. 2004, 38, 2167-2174. (12) Deuster, R.; Lubahn, N.; Friedrich, C.; Kleibo ¨hmer, W. J. Chromatogr., A 1997, 785, 227-238. (13) Po ¨rschmann, J.; Blasberg, L.; Mackenzie, K.; Harting, P. J. Chromatogr., A 1998, 816, 221-232. (14) Radcliffe, C.; Maguire, K.; Lockwood, B. J. Biochem. Biophys. Methods 2000, 43, 261-272. (15) Halasz, A.; Groom, C.; Zhou, E.; Paquet, L.; Beaulieu, C.; Deschamps, S.; Corriveau, A.; Thiboutot, S.; Ampleman, G.; Dubois, C.; Hawari, J. J. Cromatogr., A 2002, 963, 411-418. (16) Snyder, J. L.; Grob, R. L.; McNally, M. E.; Oostdyk, T. S. Anal. Chem. 1992, 64, 1940-1946.

Analytical Chemistry, Vol. 77, No. 13, July 1, 2005 4241

A potential advantage of SFE in comparison to other extraction techniques is the possibility it offers of combining different extraction conditions (e.g., modifier addition and variations in pressure and temperature) in a single run, allowing sequential extraction of analytes of differing polarity. Obviously, this strategy implies the use of a trapping system that can effectively retain the analytes of interest during the analysis. Several different trapping systems for collecting analytes and isolating them from the SF stream have been reported.19-21 They can be roughly divided into liquid or solid trapping techniques, the latter including inert systems (such as cooled sylanized glass beads) or functionalized sorbents, especially solid-phase extraction (SPE) cartridges. A recently introduced alternative is to use liquid -chromatography columns or precolumns, which provide greater durability and reproducibility than SPE traps.22,23 The study presented here had several aims. The main goal was to develop and optimize an extraction method based on SFE in order to analyze nitroaromatic and related compounds in different soil environments. A further goal was to compare different trapping systems for SFE-extracted analytes (solvent, inert solid, SPE, PGC precolumn) in terms of extraction efficiency, repeatability, total extraction time, robustness, and scope for sample post-treatment. A final goal was to compare the developed extraction methodology with EPA method 8330 in terms of efficiency parameters, especially for amino degradation products. Safety Considerations. TNT and other explosive compounds can cause headaches, weakness, anemia, and liver injury and the vapor of such chemicals is very dangerous. Explosive solutions must be prepared and handled in a fume hood and stored in closed glass containers, at a safe distance from any reducing agents. Disposable latex gloves must be worn to avoid any contact or exposure while working with these compounds because they can be absorbed through the skin. Since TNT can penetrate gloves, they have to be replaced when any contact with TNT material is suspected. The explosives are also toxic and some are mutagenic, so special care must be taken when disposing of waste solutions. EXPERIMENTAL SECTION Chemicals and Materials. The nitroaromatic substances selected for the tests were those reported to be the most prevalent in soil. 1,2-, 1,3-, and 1,4-dinitrobenzene (DNBs; CAS identification numbers, 528-29-0, 99-65-0, and 100-25-4, respectively) were purchased from Dr. Ehrenstorfer (Augsburg, Germany). 2,3-, 2,4-, 2,6-, and 3,4-dinitrotoluene (DNTs; CAS 602-01-7, 121-14-2, 606-20-2, and 610-39-9), 2,4,6-trinitrotoluene (TNT; CAS 11896-7), and 2,6-diamino-4-nitrotoluene (DANT; CAS 59229-753) were obtained as a gift from the Department of Analytical Chemistry, Stockholm University. Trinitrobenzene (TNB; CAS 99-35-4), 2-amino-4,6-dinitrotoluene (2-ADNT; CAS 35572-78(17) Levy, J. M.; Dolata, L.; Ravey, R. M.; Storozynsky, E.; Hollowczak, K. A. J. High Resolut. Chromatogr. 1993, 16, 368-372. (18) Jeong, M. L.; Chesney, D. J. J. Supercrit. Fluids 1999, 16, 33-42. (19) Hu ¨ sers, N.; Kleibo ¨hmer, W. J. Chromatogr., A 1995, 697, 107-114. (20) Yang, Y.; Hawthorne, S. B.; Miller, D. J. J. Chromatogr., A 1995, 699, 265276. (21) Mannila, M.; Koistinen, J.; Vartiainen, T. J. Chromatogr., A 2002, 975, 189198. (22) Hyo ¨tyla¨inen, T.; Riekkola, M.-L. Anal. Bioanal. Chem. 2004, 378, 19621981. (23) Po´l, J.; Hyo ¨tyla¨inen, T.; Ranta-Aho, O.; Riekkola, M.-L. J. Chromatogr., A 2004, 1052, 25-31.

4242

Analytical Chemistry, Vol. 77, No. 13, July 1, 2005

2), and 4-amino-2,6-dinitrotoluene (4-ADNT; CAS 19406-51-0) were purchased as 1000 µg/mL acetonitrile solutions from Restek (Bellefonte, PA). Dinitroanisol (DNA, injection standard; CAS 119-27-7) was also obtained from the Department of Analytical Chemistry, Stockholm University. Stock solutions of the analytes were prepared in acetonitrile. Spiking solutions were prepared by mixing appropriate amounts of stock solution with acetone to ensure solvent evaporation during the aging procedure. Except for water, which was provided by a Milli-Q system (Millipore Ibe´rica, Madrid, Spain), all solvents used were HPLC grade or better and were supplied by Scharlab (Barcelona, Spain). Soil Material, Sample Preparation, and Spiking Procedure. Two types of soil samples were selected for method evaluation. Siliceous earth, purified and calcined (Scharlab), was used as the reference inert-soil material. Native soil was collected from a location outside the laboratory at different depths (maximum 1.5 m) and mixed together to a final bulk sample. This soil has the following average composition: 1.3% organic matter, 22% silt, 15% clay, and 61.7% sand. Prior to spiking, samples of native soil were air-dried to constant weight in a fume hood. Then, they were manually ground using a mortar and pestle to pass through a 30-mesh (0.6-mm) sieve to remove stones and facilitate mixing. Spiking was performed according to the guidelines provided by Doick et al.24 Briefly, 100 g of soil was placed in a round-bottomed mixing vessel; the spike solution (500 µL) was added and blended for 1 min using a bench drill fitted with a modified stainless steel paddle. Then another 150 g of soil was added and blended for 3 × 1 min intervals. Spiking was performed at three concentration levels (approximately 2000, 2, and 0.02 µg/g of soil) representing heavily, moderately, and lightly contaminated soil, respectively. For blank analysis, the same volume of acetone containing the appropriate amount of acetonitrile was added. Following spiking, the soils (total mass 250 g) were placed in sealed, amber glass containers and stored for at least 60 days prior to analysis. 2-g subsamples of each soil were chosen for analysis, according to the work by Walsh et al.25 Each subsample was further divided into four final analysis units, each weighting 0.5 g. Gas Chromatography-Electron Capture Detection (GCECD) and GC-Chemical Ionization-Mass Spectrometry (GCCI-MS). GC-ECD measurements were done with a Varian Star 3400 CX gas chromatograph equipped with a 63Ni ECD system with on-column injection and a DB-1701 (14% cyanopropylphenyl)methylpolysiloxane column (30 m × 0.25 mm i.d., film thickness 0.15 µm) supplied by Agilent Technologies (Madrid, Spain). The temperatures were as follows: injector temperature (on-column injection), 200 °C; detector temperature, 325 °C; initial oven temperature 105 °C, held for 2 min, linear temperature gradient 30 °C/min to 170 °C, second ramp 10 °C/min to 280 °C, and held 5 min. The carrier gas was hydrogen (C-50, Carburos Meta´licos, Barcelona, Spain) at a flow rate of 1.0 mL/min. Confirmatory analyses were performed using a Varian CP-3800 gas chromatograph hyphenated to a Saturn 2000 ion trap mass spectrometer operated by the MS software version 6.30 (Varian). (24) Doick, K. J.; Lee, P. H.; Semple, K. T. Environ. Pollut. 2003, 126, 399406. (25) Walsh, M. E.; Ramsey, C. A.; Jenkins. T. F. Chemosphere 2002, 49, 12671273.

Figure 1. Supercritical fluid extraction setup. V1-V4, three-port airactuated switching valves (Valco).

Injection was carried out in the splitless mode (3 µL, PTV injection, splitless time 1.5 min), and the injector temperature program was as follows: initial temperature 100 °C held for 0.5 min and then 200 °C/min to 250 °C hold for the remaining run time. Analyses were performed using a DB-200 (35% trifluoropropyl)methylpolysiloxane column (30 m × 0.25 mm i.d., film thickness 0.25 µm, Agilent). The carrier gas was helium (C-50, Carburos Meta´licos) at a flow rate of 1.0 mL/min. MS conditions were as follows: transfer line temperature, 270 °C; trap temperature, 200 °C; 0.27 s/scan; acquisition start time, 7.5 min. The reactant gas for chemical ionization measurements was methanol. Scan-mode acquisition was used for the non-amino congeners and method detection limit (MRM) acquisition mode for the amino derivatives. Parent ions chosen for fragmentation were 198 m/z for 2- and 4-ADNT and 168 m/z for DANT. Supercritical Fluid Extraction Setup. Supercritical fluid extractions were performed using a 1-mL stainless steel extraction vessel and an Autoprep 44 stand-alone SFE system (Suprex, Pittsburgh, PA). The system also included a MPA-1 solvent pump (Varian, Walnut Creek, CA), for dynamic modifier delivery, and a variable restrictor able to provide supercritical flow rates in the range 0.5-7.0 mL/min. High-purity carbon dioxide (C-60 quality) under a helium headspace (pressure, 13.8 MPa) was purchased from Carburos Meta´licos. The complete setup, depicted in Figure 1, also included two LC pumps (isocratic model T 414 and a binary model 322, both provided by Kontron Instruments, Milan, Italy) used to supply desorption, cleaning, and reconditioning solvents and four threeport air-actuated pneumatic switching valves (V1-V4, Valco, Houston, TX). The complete procedure was as follows. In the first SFE step, supercritical fluid (with or without modifier) was passed through the extraction vessel and the heated restrictor, where it was depressurized and transferred to a liquid or solid trap via valve V3. When the solid trap was used, SF was passed to the trap and directed to waste by means of valve V4. When extraction was finished, sequential elution (V2, isocratic pump), washing (V2, binary pump channel A), drying (N2, V1),

and reconditioning (SPE trap, V2, binary pump channel B) steps were performed. Trapping materials evaluated included the following: cooled solvents (toluene, methyl tert-butyl ether, methanol, acetonitrile); cryogenically cooled (-10 °C, CO2), inert, sylanized glass beads (80/100 mesh, Supelco); solid-phase extraction columns (LiChrolut EN, Oasis HLB); and PGC liquid chromatography (LC) precolumns. The setup also included an additional oven (Kontron model 480) to maintain the temperature of the solid trap (35 °C) and a two-way valve to prevent supercritical fluid diffusion in the step, as shown in Figure 1. SPE trapping columns (5 cm × 2.1 mm i.d.) were packed with LiChrolut EN (Merck, Darmstadt, Germany), Nexus (Varian), or Oasis HLB (Waters, Milford, MA), to a length of ∼3.5 cm, and a stainless steel frit with 10-µm pores was connected to each column’s inlet and outlet. A 1-cm porous graphitic carbon (PGC, Hypercarb, Thermo Quest, Cheshire, U.K.) precolumn was used as the analyte retaining LC precolumn. EPA Method 8330. Reference extraction was performed according to the procedure developed by Jenkins et al.,6 which was adopted by the American EPA as Method SW-846 83307 for determining explosives in soil. In this standard analytical method, a 4.0-g soil sample is extracted by sonication for 18 h with 20 mL of acetonitrile in a cooled sonic bath (maximum temperature 38 °C). Each extract is filtered through a 0.45-µm disk filter and concentrated to a final volume of ∼0.5 mL under a nitrogen stream (40 °C). Volume reduction is not a step of the official method, but it is included to achieve sensitivity required to work at the 0.02 µg/g spiking level. To check that no degradation of the analytes occurs, the same protocol was applied in triplicate to a standard acetonitrile solution containing the analytes under study. No adverse effects must be reported. Then, extracts are split into two fractions and the final analysis is carried out by HPLC-UV and confirmed by GC-CI-MS. Liquid chromatographic determination was performed according to a previously developed protocol.26 Briefly, 20 µL of the extract was injected onto a 100 × 4.6 mm (5 µm) Hypercarb analytical column (Thermo Quest). The chromatographic conditions for the analysis were as follows: solvents A and B were mixtures of water-acetonitrile-methanol (50:40:10 v/v) and methanol-acetonitrile-toluene (73:25:2 v/v), respectively. The initial percentage of solvent B was 40%, flow rate 1.0 mL/min, held for 3 min. The concentration of B was then raised to 100% at a flow rate of 2.0 mL/min over 22 min and held for a further 10 min. The column was then washed with 100% B at 2.5 mL/min for 2 min, after which the mobile phase was returned to its initial composition over 1 min, and the system was allowed to equilibrate with this mixture for 10 min. Detection was performed at 254 nm. Confirmatory GC-CI-MS was performed according to the procedure described in the previous paragraph. RESULTS AND DISCUSSION Desorption and Trapping. In a preliminary study, the desorption performance of the solid traps was assessed. An injection valve (with an injection loop; see Figure 1) was used to inject 10 µL of the reference analyte mixture, containing 20 ng of (26) Sa´nchez, C.; Carlsson, H.; Colmsjo ¨, A.; Crescenzi, C.; Batlle, R. Anal. Chem. 2003, 75, 4639-4645.

Analytical Chemistry, Vol. 77, No. 13, July 1, 2005

4243

Table 1. Average Collection Efficiencies Obtained Using Liquid, Solid, and Chromatographic Trapping Materialsa solid trapping liquid trapping

SPE trap

analyte

toluene

MtBE

methanol

acetonitrile

inert

Oasis

Nexus

LiChrolut

PGC

1,2-DNB 1,3-DNB 1,4-DNB 2,3-DNT 2,4-DNT 2,6-DNT 3,4-DNT TNB TNT 2 A 4,6-DNT 4 A 2,6-DNT 2,6-DANT

86 (4) 87 (5) 110 (8) 123 (7) 98 (6) 101 (2) 77 (10) 78 (7) 82 (5) 34 (12) 30 (23) 24 (17)

76 (7) 77 (7) 71 (10) 82 (13) 78 (7) 91 (4) 87 (7) 88 (6) 79 (5) 35 (17) 30 (22) 34 (7)

45 (6) 47 (5) 43 (4) 62 (9) 67 (8) 72 (12) 67 (9) 54 (7) 69 (18) n.d. n.d. 6 (18)

32 (6) 27 (8) 32 (10) 45 (6) 54 (7) 53 (18) 50 (7) 45 (7) 67 (9) n.d. n.d. n.d.

56 (7) 45 (8) 56 (9) 44 (8) 27 (12) 23 (5) 42 (8) 13 (6) 12 (7) n.d. n.d. n.d.

71 (9) 76 (8) 77 (7) 89 (9) 90 (8) 91 (9) 86 (8) 90 (9) 90 (8) 41 (8) 50 (9) 11 (9)

88 (7) 82 (6) 80 (9) 112 (7) 101 (8) 102 (7) 93 (10) 105 (7) 102 (8) 40 (5) 52 (7) 21 (6)

78 (8) 67 (9) 60 (8) 103 (19) 81 (8) 82 (10) 90 (11) 83 (8) 90 (8) 28 (5) 41 (7) 17 (7)

86 (7) 88 (9) 87 (7) 102 (8) 107 (3) 99 (6) 95 (7) 87 (9) 92 (12) 45 (18) 53 (12) 27 (8)

a

n ) 3; RSD in parentheses.

Table 2. Supercritical Fluid Extraction Optimization Parameters optimum SFE conditions variable (units)

range

screeninga

RSMb

dinitro

trinitro

monoamino

diamino

SF pressure (atm) SF temperature (°C) static extraction time (min) dynamic extraction mass (g) restrictor temperature (°C) SF flow (mL/min) fictitious variable (-) static modifier (DCM; % v/v) dynamic modifier (methanol; % v/v)

150-450 40-150 0-10 1-40 40-100 0.5-5.0 0-1 0-10 0-20

yes yes yes yes yes yes yes no no

yes yes yes yes no no no yes yes

150 40 0 15 1.0 0 0

300 40 2 15 2.0 2 3

450 40 7 30 2.0 10 12

450 40 10 40 2.5 10 20

a

combination 1 250 40 5 15 80 1.0 0 0

2 440 40 5 15 80 2.0 5 10

Plackett-Burman experimental design. bResponse surface modeling (RSM; CCF design).

each analyte. Analytes were transferred to the top of the different traps using 100 µL of Milli-Q water and a range of elution solvents (dichloromethane, methanol, acetonitrile, acetone, methyl tertbutyl ether, toluene) were tested. The best mean results, in terms of high recovery and low solvent requirements, were obtained when toluene (1 mL) was used as the elution solvent, so toluene was selected for further work. In a second preliminary study, the ability of the different trapping systems to retain selected analytes was compared. For each system, three inert glass fiber filters (used to avoid any matrix interference) were spiked with the same amount of reference standard solution, with analyte concentrations in the range 1.0-3.5 µg. The filters were then extracted using previously published SFE conditions (pressure 325 atm; temperature 100 °C; static time 5 min; dynamic mass 20 g; flow rate 1.5 mL/min; 2% acetonitrile added as both static and dynamic modifier).27 Table 1 shows the results obtained. For liquid trapping, toluene gave the best results, while Nexus and PGC gave the best (and very similar) recoveries and precision of the solid trapping options. Therefore, these traps were selected for further study. SFE conditions. The SFE was optimized by evaluating the efficiency of extracting commercially supplied, inert soil spiked with the analyte mixture (spiking level, 2.0 µg/g) using a (27) Batlle, R.; Carlsson, H.; Holmgren, E.; Colmsjo ¨, A.; Crescenzi, C. J. Chromatogr., A 2002, 963, 73-82.

4244 Analytical Chemistry, Vol. 77, No. 13, July 1, 2005

Plackett-Burman experimental design for 7 variables at 2 levels (see Table 2) including 12 runs plus 3 replicates at the central point, in duplicate, and liquid trapping to minimize the time required for the evaluation. A fictitious variable (i.e., a virtual factor representing no procedural changes) was included to evaluate the robustness of the method (Table 2). This factor showed no relevance, indicating that the data were realistically described. Four significant factors (temperature, pressure, static extraction time, mass of supercritical fluid in the dynamic extraction step) were found to be relevant and thus selected for further evaluation. The next step in the SFE method development was to study the role of potential modifiers. Five different modifiers (acetonitrile, dichloromethane, methanol, MtBE, toluene) were selected and tested in both static (10% v/v) and dynamic (15% v/v) addition systems in extractions of inert soil spiked analytes. Table 3 shows the average recoveries obtained relative to nonmodified SFE. In addition, the ratio of unidentified to identified peak counts (designated “nonassigned”) is presented in Table 3, as an indication of the selectivity of the different modifiers. The highest recoveries were obtained when using dichloromethane and methanol as the static and dynamic modifiers, respectively. The applicability of these results to real, untreated soil was assessed by repeating the experiments with native soil. The effect of the static modifier addition was much more significant, especially for the amino compounds, whereas the effects of the dynamic

Table 3. Comparison of Mean Ratios of Recoveries of Analytes Obtained from Aged Inert Soil (Standard Font) and Aged Native Soil (Italics), with and without Modifiers, Using Both Static (St) and Dynamic (Dy) Addition Systems modifiers (% yhrv/v) methanol

MtBE

toluene

DCM

ACN

analyte

St

Dy

St

Dy

St

Dy

St

Dy

St

Dy

dinitrobenzenes dinitrotoluenes trinitrotoluene trinitrobenzene monoamino diamino nonassigned

0.9 1.0 0.9 0.9 1.1 1.0 1.1

1.2/1.0 1.1/0.9 0.8/1.1 1.2/1.2 1.8/2.5 2.1/2.6 2.0/3.5

0.9 0.8 0.9 1.0 0.8 0.7 0.6

0.9 2.9 2.3 1.7 1.2 1.1 1.0

1.0 0.7 0.8 0.4 1.1 1.3 1.0

0.8 1.1 0.9 0.8 1.4 1.2 1.1

1.0/1.1 1.1/1.9 0.8/2.0 1.1/3.1 1.9/4.5 2.3/4.7 1.1/2.9

0.9 0.3 0.8 1.1 1.2 1.3 1.7

0.7 0.9 0.8 1.0 1.2 1.1 1.0

0.6 0.7 1.0 1.0 1.4 1.3 1.4

Table 4. Recovery Optimization Results, Aged Native Soil (n ) 3, % RSD)a SFE conditionsb

a

analyte

dinitro

trinitro

amino

combination

combination/ cleanup

1,2-DNB 1,3-DNB 1,4-DNB 2,3-DNT 2,4-DNT 2,6-DNT 3,4-DNT TNB TNT 2-ADNT 4-ADNT DANT

85 (8) 118/88 (19) 125/96 (17) 87 (9) 80 (11) 86 (3) 78 (9) 76 (10) 65 (12) 50 (18) 52 (9) 35 (9)

80 (7) 126/87 (17) 112/89 (11) 67 (9) 77 (12) 82 (4) 75 (6) 90 (11) 87 (8) 40 (8) 50 (10) 44 (8)

65 (7) 61 (12) 76 (21) 54 (18) 63 (15) 52 (7) 57 (11) 65 (6) 77 (7) 156/61 (12) 144/52 (21) 174/65 (23)

83 (9) 114/84 (11) 119/82 (13) 122/97 (7) 77 (19) 80 (19) 79 (8) 80 (11) 72 (8) 134/77 (17) 157/75 (29) 166/61 (38)

81/87 (3) 91/85 (7) 92/87 (8) 98/100 (8) 82/81 (9) 84/86 (5) 82/83 (6) 89/85 (4) 83/82(11) 81/86(8) 81/85 (9) 69/66 (6)

Numbers in italics represent recovery percentages obtained by GC-CI-MS (confirmatory results). b SFE conditions, please refer to Table 2.

modifier were comparable for both types of soil. As can be seen in Table 3, addition of most of the tested modifiers reduced selectivity (i.e., gave a higher proportion of nonassigned counts than nonmodified SFE). Nevertheless, the increase in the recovery compensates for this drawback. We believe that this is a very interesting finding. The nitroaromatic compounds have high solubility in methanol but only moderate in dichloromethane, the least polar modifier with the smallest effect on the SF polarity. This supports the hypothesis that the effect of a static modifier is mainly related to matrix characteristics, whereas the effects of dynamic modifiers are related to the solubility of the analytes. Thus, when selecting a static modifier, the interactions between the matrix and the analytes are matters of paramount importance. The static modifier had the stronger effect on the extraction of the amino derivatives. Substitution of a nitro group with an amino group increases the soil-water partition coefficient (Kd) of the compounds dramatically. If a second amino group is substituted, giving diamino derivatives, Kd will be further increased, leading to a greater capacity for adsorption to soil matrixes. Reduced forms of TNT are capable of binding to soil by several mechanisms including hydrogen bonding, van der Waals forces, and hydrophobic interactions.28,29 Therefore, the main effect of the static modifier is to overcome this binding, thus enabling extraction of these contaminants. The final choice of trapping strategy was made at this developmental stage. Liquid trapping was discarded since dichlo(28) Thorn, K. A.; Kennedy, K. R. Environ. Sci. Technol. 2002, 36, 3787-3796. (29) Burns, D.; Knicker, H.; Drzyzga, O.; Butehorn, U.; Steinbach, K.; Gemsa, D.; von Low, E. Environ. Sci. Technol. 2000, 34, 1549-1556.

romethane was the static modifier and its presence in the final extract prevented the use of the GC-ECD detection system. For the solid trapping, PGC was selected for practical reasons; no additional washing step was needed for the PGC precolumn after elution, whereas SPE traps need regeneration with 2.0 mL of acetonitrile/toluene followed by a drying step with nitrogen. Another very important feature is the potential for reusing the solid trap for repeated analyses over the course of several days. Only 2 PGC precolumns were used during the course of the work presented here (∼350 complete analyses), whereas the average working life of SPE traps was ∼20 runs with the inclusion of the regenerative washing step. The final optimization was performed by using a response surface design (CCF) including the four relevant factors plus the two modifier-associated variables (Table 2). The design included 44 experiments plus 3 replicates at the central point. The optimum values were found to depend on the analytes. The highest recoveries of the dinitro derivatives (DNBs and DNTs), trinitro derivatives (TNB and TNT), and amino congeners (mono- and diamino) were obtained with moderate pressures and temperatures with no modifier, moderate pressures with moderate amounts of modifier, and the highest pressures and modifier concentrations, respectively. SF mass and static extraction time had little influence on the recovery of the first and second of these but had a clear positive effect for the third. Temperature had a negative effect, related to analyte degradation, especially for the more labile trinitro compounds. Once an effective trapping technique had been developed, a combination of two sets of SFE conditions (in single runs) was Analytical Chemistry, Vol. 77, No. 13, July 1, 2005

4245

Table 5. Method Comparison: SFE vs EPA Method 8330, Aged Native Soil dinitro

trinitro

monoamino

figure of merit

SFE

US

SFE

US

SFE

robustness recovery (trueness %) 0.02 µg/g spiking level 2.0 µg/g spiking level 2000 µg/g spiking level repeatability (%) 0.02 µg/g spiking level 2.0 µg/g spiking level 2000 µg/g spiking level intermediate precision (%) 0.02 µg/g spiking level 2.0 µg/g spiking level 2000 µg/g spiking level capability of detection (ng, R ) 0.05, β ) 0.05, K ) 1, x0 ) 0.2 µg g-1)

yes

yes

yes

yes

yes

98 97 88

101 102 101

95 96 89

99 100 102

8 6 7

5 4 5

9 7 8

8 7 8 1.6

6 5 5 20

8 8 9 5.8

Figure 2. Comparison of chromatograms obtained following SFE extraction (combination conditions; see Table 4) of aged native soil with (middle panel) and without (upper panel) cleanup steps and of inert soil (lower panel). Peak identification: (1) 1,4-DNB; (2) 2,6-DNT; (3) 1,3-DNB; (4) 1,2-DNB, (5) 2,4-DNT; (6) 2,3-DNT; (7) 3,4-DNT; (8) TNT; (9) TNB; (IS) DNA surrogate standard; (10) 2-ADNT; (11) 4-ADNT; (12) 2,6-DANT.

tested that was designed to include optimal, or near-optimal, conditions for extracting all three groups of compounds, as shown in Table 2. The first step had the following parameters: pressure, 250 atm; temperature, 40 °C; 5-min static extraction; and 15 g of dynamic extraction fluid (1.0 mL/min) with no modifier. The parameters of the second step were as follows: pressure, 440 atm; temperature, 40 °C; 5-min static extraction (5% dichloromethane 4246 Analytical Chemistry, Vol. 77, No. 13, July 1, 2005

diamino US

SFE

US

no

yes

No

99 97 81

72 71 79

64 57 68

48 49 37

6 5 6

9 8 11

11 13 17

8 7 10

15 16 21

7 8 10 22

9 10 12 2.7

12 12 16 97

10 8 11 3.4

14 18 20 156

modifier); and 15 g of dynamic extraction fluid (2.0 mL/min, 10% methanol as dynamic modifier). After the two extraction steps, the extracts were cleaned up and eluted. Table 4 compares average recoveries obtained using combined conditions with those obtained using single-step SFE. The best overall results were obtained with the combination approach, although the recovery was lower for some of the compounds. Figure 2 shows typical chromatograms obtained with the combination of conditions for SFE-PGC extractions of both inert (lower panel) and native soil (upper panel). As can be seen, native soil extraction yielded chromatograms that include some nonidentified peaks that have a detrimental effect on DNT and amino quantification, as confirmed by the GC-CI-MS results (italic values, Table 4). A potentially useful approach for overcoming this problem would be selective trapping after the SFE. Distilled water (2.0 mL at 4.0 mL/min) was pumped through the solid trap to waste to eliminate any water-soluble interferences. Then, nitrogen was used to dry the solid trap in preparation for the organic solvent elution step. As shown by the data in Table 4, the rinsing step caused no reduction in analyte recovery, even for the more water soluble diamino congeners, and it improved the analytical performance for amino products, as confirmed by GC-CI-IT-MS. As shown in Figure 2, the cleanup steps reduced the number of unidentified peaks in SFE-PGC chromatograms obtained following extractions of native soil (middle and upper panels, respectively), resulting in chromatograms that are very similar to those obtained following inert soil extractions (lower panel). This method was then selected as the optimum one and validated by comparison with the modified EPA method 8330. Method Performance. The developed method was validated according to the guidelines and recommendations provided by IUPAC and the International Standard Organization.30,31 Explored parameters were robustness (whether the method is sensitive to minor changes (5% of the optimum values in procedural variables), trueness (the difference between results obtained from spiked and nonspiked samples, with the theoretical value added considered the reference, n ) 12), repetitivity (n ) 8, three spiking (30) Currie, L. A. Pure Appl. Chem. 1995, 67, 1699-1723. (31) Garcı´a, I.; Ortiz, M. C.; Sarabia, L.; Vilches, C.; Gredilla, E. J. Chromatogr., A 2003, 992, 11-27.

levels), intermediate precision (n ) 5, calculated for sets of four experiments performed on five consecutive days with different extraction vessels and PGC traps at the three spiking levels), and capability of detection (concentration of the analyte leading, with probability β, to the conclusion that the concentration of analyte in the sample is different from that in the blank material with a probability R of false positive). Table 5 presents a summary of parameters validated for the determination of nitroaromatic congeners in native soil for both the developed SFE-PGC method and modified EPA method 8330. Results represent average values for each group of analytes. As can be seen, comparable results were obtained for the nitro congeners, whereas the developed SFE-PGC method gave better results regarding the amino congeners, especially in terms of robustness. These compounds were very sensitive to minor changes ((5 °C) in the sonication bath temperature, and detection capability of the SFE-PGC method was 12.5, 3.8, 36, and 46 times higher, for the dinitro, trinitro, monoamino, and diamino congeners, respectively. CONCLUSIONS The utility of supercritical fluid extraction for soil analysis has been demonstrated and validated for a set of energetic nitroaromatic materials and nitroamine degradation products. The developed method consists of a SFE program that includes a combination of two sets of experimental conditions, retention on a PGC

precolumn, cleaning, drying, and desorption using toluene. The time required for a complete analysis of the test soils was 60 min, and 1 mL of toluene was used for a 0.5-g representative sample. The EPA method 8330 required 18 h for the extraction step, additional time for the chromatographic analysis, and 20 mL of acetonitrile for a 4.0-g sample. The combination of selective SFEPGC allows the use of GC-ECD detection, as confirmed with highly selective chemical-ionization-MS detection. Further work will focus on expanding the application range of the technique to different soil matrixes and other energetic materials, such as 1,3,5-trinitro-1,3,5-triazacyclohexane (RDX), 1,3,5,7-tetranitro-1,3,5,7-tetrazacyclooctane (HMX), and nitroamine explosives. ACKNOWLEDGMENT The authors thank the Department of Analytical Chemistry at Stockholm University for providing the Supercritical Fluid Extraction equipment. R.B. gratefully acknowledges the former Spanish Ministry of Science and Technology for personal funding through the Ramo´n y Cajal program.

Received for review February 24, 2005. Accepted April 18, 2005. AC050339+

Analytical Chemistry, Vol. 77, No. 13, July 1, 2005

4247