Anal. Chem. 1996, 68, 4507-4511
Comparison of Supercritical CHF3 and CO2 for Extraction of Sulfonamides from Various Food Matrices M. T. Combs, M. Ashraf-Khorassani, and L. T. Taylor*
Department of Chemistry, Virginia Polytechnic Institute and State University, Blacksburg, Virginia 24061
Supercritical CHF3 and methanol-modified CHF3 were compared with supercritical CO2 and methanol-modified CO2 for the extraction of sulfonamides from a spiked sand sample. In addition, a fortified nonfat milk powder, fortified egg yolk, and fortified beef liver were studied. The results showed CHF3 has higher solvating power and selectivity for extraction of two of the three sulfonamides than CO2. Extraction efficiencies of sulfamethazine and sulfadimethoxine using pure CHF3 were more than 45% higher from spiked sand samples and over 200% higher from fortified beef liver samples than pure CO2. Other supercritical fluids such as ammonia, water, nitrous oxide, pentane, and sulfur hexafluoride are not ideally suited for routine analytical use owing to their unfavorable (a) physical properties, (b) reactivity, (c) solvating power, or (d) incompatibility with modern instrumentation. However, applications of these nonideal supercritical fluids do exist. Smith et al.1-3 employed supercritical ammonia coupled with mass spectrometry for the direct extraction and analysis of diesel fuel. Additionally, Hawthorne et al.4 used sub- and supercritical water for the extraction of a variety organic pollutants from environmental solids. AshrafKhorassani et al.5 compared CO2 and nitrous oxide for extraction of primary, secondary, and tertiary amines using on-line SFE/ GC. McNally et al.6 used CO2, methanol-modified CO2, N2O, and amine-modified N2O for extraction of different aromatic amines. Onuska and Terry7 and Levy et al.8 compared N2O, CO2, and mixtures of these fluids with different percentages of methanol for extraction of 2,3,7,8-tetrachlorodibenzo-p-dioxin and polynuclear aromatic hydrocarbons (PAHs) from aquatic sediment and soil. Onuska et al. also compared sulfur hexafluoride as an extraction fluid with CO2 and N2O. The use of N2O as an extraction solvent has been refuted by Raynie.9 In his research, an extraction vessel filled with N2O and coffee resulted in an explosion and destruction of the extraction vessel. (1) Smith, R. D.; Udseth, H. R.; Hazlett, R. N. Fuel 1985, 64, 810. (2) Smith, R. D.; Udseth, H. R. Fuel 1983, 62, 466. (3) Smith, R. D.; Udseth, H. R.; Wright, B. W. In Supercritical Fluid Technology; Penninger, J. M., Radosz, M., McHugh, M. A., Krukonis, V. J., Eds.; Elsevier Science: New York, 1985; p 191. (4) Hawthorne, S. B.; Yang, Y.; Miller, D. J. Anal. Chem. 1994, 66, 2912. (5) Ashraf-Khorassani, M.; Taylor, L. T.; Zimmerman, P. Anal. Chem. 1990, 62, 1177. (6) Oostdyk, T. S.; Grob, R. L.; Snyder, J. L.; McNally, M. E. Anal. Chem. 1993, 65, 596. (7) Onuska, F. I.; Terry, K. A. J. High Resolut. Chromatogr. 1989, 12, 357. (8) Levy, J. M.; Storozynsky, E.; Ravey, R. M. J. High Resolut. Chromatogr. 1991, 14, 661. (9) Raynie, D. E. Anal. Chem. 1993, 65, 3127. (10) Stahl, E.; Willing, E. Mikrochim. Acta. 1981, 465. S0003-2700(96)00643-9 CCC: $12.00
© 1996 American Chemical Society
Literature regarding the use of trifluoromethane as an extraction fluid is limited. In 1980, Stahl and Willing10 determined solubilities of different alkaloids in N2O, CO2, and CHF3. Their results showed alkaloids are more soluble in CHF3 compared to either CO2 or N2O. Howard et al.11 studied the SFE of environmental analytes using CHF3. In their work, CHF3 was a better extractant than CO2 or CHClF2 for sulfonylurea herbicides, but was less effective for the extraction of three- and four-ring PAHs, with CHF3 exhibiting the lowest recovery of the three fluids investigated. The polar nature of CHF3 was believed to assist the extraction of the polar sulfonyl urea herbicides, whereas no polar effects occur for PAHs. Ashraf-Khorassani et al.12 also used CHF3 and methanol-modified CHF3 to extract sulfonamides from chicken liver tissue. Near-quantitative recovery was obtained for sulfamethazine and sulfadimethoxine with 10% methanol-modified CHF3. However, only 30% of a third sulfa drug (sulfaquinoxaline) was obtained under the same conditions. This paper considers the use of supercritical CO2, methanolmodified CO2, CHF3, and methanol-modified CHF3 as solvent for the extraction of several sulfonamides from various additional matrices including, nonfat dry milk, egg yolk, and beef liver. Currently, liquid/solid extraction is employed with several cleanup steps for isolating sulfa drugs from food matrices. In the first part of our study, the effect of pressure and addition of modifier upon extraction efficiency of sulfamethazine (SMZ), sulfadimethoxine (SDM), and sulfaquinoxaline (SQX) from an inert matrix using both CHF3 and CO2 is investigated. Structures of the target analytes are shown in Figure 1. In the second part of our study, the selectivity of CHF3 and CO2 is compared for the extraction of sulfa drugs from nonfat dry milk, egg yolk, and beef liver. EXPERIMENTAL SECTION A Suprex (Pittsburgh, PA) Prepmaster equipped with an Accutrap automatic variable restrictor system was used in a nonautomated mode for extracting sulfonamides from each food matrix. A Suprex Autoprep 44 automated SFE also equipped with an Accutrap automatic variable restrictor was used for all extractions in the pressure study. Both systems were outfitted with an in-line SSI microbore reciprocating pump (Alltech Associates, Deerfield, IL). The HPLC grade methanol and acetonitrile were purchased from EM Science (Gibbstown, NJ). HPLC grade water was purchased from Mallinckrodt (Paris, KY). CO2 and CHF3 were both obtained from Air Products and Chemicals, Inc. (11) Howard, A. L.; Yoo, W. J.; Taylor, L. T.; Schweighardt, F. K.; Emery, A. P.; Chesler, S. N.; MacCrehan, W. A. J. Chromatogr. Sci. 1993, 31, 401. (12) Ashraf-Khorassani, M.; Taylor, L. T.; Schweighardt, F. K. AOAC Int. 1996, 79, 1043.
Analytical Chemistry, Vol. 68, No. 24, December 15, 1996 4507
Figure 1. Structures of target analytes, sulfamethazine (SMZ), sulfaquinoxaline (SQX), and sulfadimethoxine (SDM).
(Allentown, PA). Both were pressurized with 2000 psi helium to minimize pump cavitation. The extraction vessels used for the pressure study were 1 mL and for the matrix extractions were 3 mL in volume (Keystone Scientific, Bellefonte, PA). All of the extractions involved a single 60 min dynamic step at the desired pressure. All matrix extractions were accomplished at 490 atm using CO2 and methanolmodified CO2 or 400 atm using CHF3 or methanol-modified CHF3 at 40 °C. A flow of 1.5 mL/min for liquid solvent measured at the pump was used for all extractions. Trapping for matrix extractions consisted of a tandem solid sorbent trap of octadecyl silica (Applied Separations, Allentown, PA) at 10 °C and a liquid trap of 5 mL of methanol using pure fluids. During modifiedfluid extractions, the solid phase trap temperature was increased to 80 °C. No tandem trap was needed for extractions in the pressure study, and the solid phase trap was cryostated at either 10 or 80 °C with the use of modifier. After completion of each extraction, the solid phase trap was rinsed at 40 °C with 5 mL of the HPLC mobile phase (85% 8 mM ammonium acetate (NH4OAc)/15% acetonitrile solution) into a 5 mL volumetric flask. The liquid tandem trap was also diluted to 5 mL with 85% 8 mM NH4OAc/15% acetonitrile. Pressure study samples were prepared by spiking 10 µL of drug standard (0.6 µg/µL each of SMZ, SQX, and SDM in methanol) directly onto Ottawa sand standard (Fischer Scientific, Fair Lawn, NJ). The solvent was then allowed to evaporate prior to extraction to avoid modification of the fluid. The nonfat dry milk powder was purchased at a local grocery store. Samples were prepared by spiking 10 µL of the drug standard directly onto 0.2 g of the milk matrix (30 ppm). The solvent was then allowed to evaporate prior to extraction. The vessel was then sealed and extracted under specified conditions. The egg yolk sample matrix was prepared by first separating the egg white and yolk. Next, the egg yolk was mixed 1:3 with Celite 545 (Supelco, Bellefonte, PA) and spiked to 6 ppm of each sulfonamide per 1 g of egg/Celite mixture. The mixture was allowed to incubate at -10 °C for a minimum of 30 min. Beef liver samples were prepared by spiking 0.5 g of liver tissue with 10 µL of the drug mixture (12 ppm). The liver was provided by the USDA/ARS located in Philadelphia, PA, and was ground prior to receipt. The spiked matrix was allowed to incubate for a minimum of 30 min at -10 °C. Following frozen storage, 1 g of Celite was added to the liver and mixed thoroughly. The entire contents were added directly to the extraction vessel and extracted under specified conditions. 4508 Analytical Chemistry, Vol. 68, No. 24, December 15, 1996
Figure 2. Extraction behavior of sulfonamides from sand using pure CO2.
A Hewlett-Packard (Little Falls, DE) series 1050 HPLC equipped with a variable-wavelength UV detector was used to assay all sample extracts. A 250 × 4.6 mm (5 µm) Deltabond ODS (Keystone Scientific, Bellefonte, PA) column was used throughout the study. The mobile phase employed was 85% 8 mM NH4OAc/ 15% acetonitrile operated at a flow of 1 mL/min. All three sulfonamides were detected at 266 nm. RESULTS AND DISCUSSION Inert Matrix. The objective of the first part of our study was to investigate the effect of fluid pressure and methanol modifier addition upon the extraction efficiency of SMZ, SQX, and SDM using both CHF3 and CO2. Figure 2 shows the effect of raising the pressure from 360 to 490 atm upon the extraction efficiency of SMZ, SQX, and SDM from a sand matrix for pure CO2. The pressures investigated (360, 400, 440, and 490 atm) exhibited little effect on recoveries of the three sulfonamides. At 360 atm recoveries were 38%, 28%, and 40% for SMZ, SQX, and SDM, respectively. Increasing the pressure to 490 atm shows insignificant increases in recovery to 43% SMZ, 34% SQX, and 44% SDM. Similar recoveries were obtained at intermediate pressures. With the addition of 10% methanol to the CO2, quantitative recoveries could be obtained from sand for all three sulfonamides due to the much greater solvating power of modified fluid. At 490 atm, 104%, 104%, and 105% recoveries were observed for SMZ, SQX, and SDM, respectively. Again, altering the pressure showed a minimal effect upon extraction efficiency. Decreasing the pressure to 360 atm only decreased recoveries to 92% for SMZ, 90% for SQX, and 94% for SDM. At 400 and 440 atm, the recoveries lie between that obtained at 490 and 360 atm. There is a distinct difference in percent recovery obtained for SMZ and SDM for CHF3 relative to CO2. At 360 atm, 69% SMZ and 61% SDM were extracted with CHF3, whereas with CO2 only 38% and 40% were extracted for SMZ and SDM, respectively. This corresponds to an 82% increase in extractability for SMZ and a 53% increase for SDM on going from CO2 to CHF3. However, SQX showed no enhanced extractability for CHF3 relative to CO2. Trifluoromethane is believed to show selective extraction characteristics due to the chemical nature of each analyte (Figure 1). We measured pH values for the three sulfonamides in methanol
Figure 3. Amphoteric behavior of sulfaquinoxaline.
Table 1. Percent Recoveries of Sulfamethazine (SMZ), Sulfaquinoxaline (SQX), and Sulfadimethoxine (SDM) from Nonfat Milk Powdera
SMZ SQX SDM a
Figure 4. Extraction behavior of sulfonamides from sand using 10% methanol modified CHF3.
(SQX exhibits very poor water solubility) to determine whether the trends in extractability were related to base strength. CHF3 has an acidic proton and a dipole moment of 1.6 D;13 therefore, a basic analyte should be able to bond with the fluid, making the extraction more efficient. SMZ had a pH of 6.7, SDM a pH of 6.3, but SQX a pH of 11.2. It appeared that SQX was the strongest base in methanol and should hydrogen bond better with trifluoromethane. But the opposite results were obtained. The spike standard solution was prepared in methanol and since SMZ and SDM have pH near 7 they should be neutral, but SQX at pH 11.2 would be ionic (Figure 3).14 SFs are known to poorly solvate ionic analytes, which may explain the results obtained. Although the trend in base strength was not confirmed, the polar character and ability of CHF3 to hydrogen bond is believed to increase the extractability of SMZ and SDM with CHF3 relative to CO2. For the remaining three pressures investigated, 400, 440, and 490 atm, the same trend was observed. At each pressure, the extraction efficiency of SMZ was at least 80% higher than the corresponding CO2 extraction, in addition SDM was at least 45% higher. The extraction efficiency for SQX was similar regardless of the fluid used for these three pressures. With the addition of 10% methanol modifier, strikingly different behavior could be observed (Figure 4). At 360 atm, recoveries obtained for SMZ and SDM were similar to those obtained for pure CHF3 (e.g., 78% and 65%, respectively with pure CHF3, and 69% and 72% with 10% methanol-modified CHF3), but the recovery of SQX was increased to 73% compared to 41% with pure CHF3. At both 400 or 440 atm, quantitative recovery could be obtained for all sulfonamides with methanol modifier. The behavior of CHF3 at 490 atm was the most puzzling. Recoveries for each of the three sulfonamides were surprisingly lower than that obtained at lower pressures. Solvating strength is usually believed to increase as the density of the fluid is increases15 (e.g., increased (13) Reid, R. C.; Prausnitz, J. M.; Sherwwood, T. K. The Properties of Gases and Liquids, 3rd. ed.; McGraw-Hill: New York, 1977. (14) Broglie, H. Chem. Ind. 1982, 385.
100% CO2
90% CO2, 10% MeOH
100% CHF3
90% CHF3, 10% MeOH
72(7) 63(9) 69(7)
95(4) 100(5) 99(6)
90(3) 64(9) 92(7)
102(3) 103(1) 107(4)
Numbers in parentheses are relative standard deviation.
pressure at constant temperature). However, recovery was found to decrease at pressures above 400 atm. It is also important to note that similar recoveries were obtained at 490 atm whether modifier was absent or present. The addition of modifier should aid the extraction by (1) increasing the solvating strength of the fluid and/or (2) disrupting matrix/analyte interactions. However, the presence of a polar modifier with a polar fluid may produce different behavior. Johnston et al.16 found that as the pressure of SF6 increased the degree of hydrogen bonding between tertiary perfluorobutanol and dimethyl ether decreased. Studies using CHF3 by Kim and Johnston17 found the hydrogen bond donor strength of CHF3 to be comparable to that of CHCl3. The degree of hydrogen bonding between analytes or analyte and fluid may be envisioned to decrease due to increased intermolecular interaction between the fluid molecules, thereby decreasing the tendency of the fluid to hydrogen bond with the analyte. A similar phenomenon could occur between the polar fluid and a polar modifier. Strong hydrogen bonding can be envisioned to occur between F3CH‚‚‚O(H)CH3 (methanol-trifluoromethane). If this fluid/fluid interaction occurs to a greater extent at higher pressures, a decrease in recovery at 490 atm would be expected. Since 400 atm was found to yield optimum recovery, it was used throughout for all CHF3 extractions of food matrices. Food Matrices. The second part of our study was concerned with comparing CO2, methanol-modified CO2, CHF3, and methanolmodified CHF3 for the extraction of SMZ, SQX, and SDM from nonfat milk powder, egg yolk, and beef liver. The first matrix investigated was a commercially available nonfat milk powder sample. The highest recovery obtained using pure CO2 was 72% for SMZ and 70% for SDM, whereas greater than 90% of both were obtained using pure CHF3 (Table 1). It is important to note that the recovery of SQX is the same regardless of the fluid obtained (63% with CO2 and 64% with CHF3), which is similar to results obtained on sand. Upon the addition of 10% methanol modifier, quantitative recovery could be achieved with either CO2 or CHF3. Nonfat milk powder appears to be a relatively inert matrix that can be easily extracted with the addition of modifier. A matrix consisting of egg yolk and Celite (to immobilize coextractable water) was investigated next. The egg yolk was separated from the egg white prior to spiking the matrix. Very poor recoveries (Table 2) from egg yolk were obtained when nonmodified CO2 was used. Only 10% of the SMZ and less than 2% of the SDM and SQX were extracted under the experimental (15) Patthy, M. J. Chromatogr. 1983, 275, 115. (16) Kazarian, S. G.; Gupta, R. B.; Clarke, M. J.; Johnston, K. P.; Poliakoff, M. J. Am. Chem. Soc. 1993, 115, 11099. (17) Kim, S.; Johnston, K. P. AIChE J. 1987, 37, 1603.
Analytical Chemistry, Vol. 68, No. 24, December 15, 1996
4509
Table 2. Percent Recoveries of Sulfamethazine (SMZ), Sulfaquinoxaline (SQX), and Sulfadimethoxine (SDM) from Egg Yolk/Celite Mixa
SMZ SQX SDM a
100% CO2
90% CO2, 10% MeOH
100% CHF3
90% CHF3, 10% MeOH
10(11)