Anal. Chem. 1996, 68, 2713-2716
Microwave Distillation-Solid Phase Adsorbent Trapping Device for the Determination of Off-Flavors, Geosmin and Methylisoborneol, in Catfish Tissue below Their Rejection Levels Eric D. Conte,† Chun-Yi Shen, and Dwight W. Miller*
Divison of Chemistry, National Center for Toxicological Research, Jefferson, Arkansas 72079 Peter W. Perschbacher Aquaculture Experimental Station, University of Arkansas at Pine Bluff, Pine Bluff, Arkansas 71601
Described is a rapid microwave-mediated steam distillation device for determining two predominant off-flavor compounds, geosmin and methylisoborneol, in catfish tissue. A microwave on-time of 10 min is needed to efficiently remove these off-flavor compounds from the sample matrix and trap them on a solid phase adsorbent. A minimal amount of organic solvent is used to elute the trapped compounds. The extract is then analyzed by gas chromatography with ion trap detection in the selective ion storage mode. Detection limits in the sub-parts-perbillion range are obtained with this method.
microwave distillation,5,6 both with cryogenic cold trap collection. These procedures require a great deal of sample manipulation and organic solvent usage to remove the trapped analytes from the cold traps. The large volume of organic solvent required for the rinsing and liquid-liquid extraction of these traps necessitates a concentration step to achieve adequate instrument detection levels. However, concentration may cause a significant loss of these volatile analytes. Microwave-mediated distillation, coupled with solid phase adsorbent trapping, has been described in an earlier paper.7 This technique offered significant improvements in sample preparation time and solvent usage. However, limitations of this design were discovered. There was carryover of geosmin from sample to blank analyses. Also, the detection limit of methylisoborneol was above the rejection level of channel catfish. The rejection levels as suggested by the U.S. Department of Agriculture are 0.8 ppb for methylisoborneol and 8.0 ppb for geosmin in channel catfish.8 The design presented here overcomes these difficulties by utilizing a replaceable C18 solid phase adsorbent cartridge, which both alleviates carryover problems and improves recoveries, thus lowering detection limits below the rejection levels. Finally, selective ion storage (SIS) mode on the ion trap mass spectrometer gives an improved instrumental detectability compared to that of the total ion chromatogram (TIC) mode previously used.
Earthy and/or musty undesirable tastes found in the muscle tissue of the channel catfish have been associated with the presence of two predominant off-flavor compounds, geosmin (GEO) and methylisoborneol (MIB). The phytoplanktonic community in channel catfish culture ponds plays a crucial role in maintaining an economic aquaculture production by providing oxygen and food and removing wastes. Channel catfish are cultured in farm systems with ponds that typically cover 20 acres and contain 5000-7000 catfish per acre.1 Eutrophic conditions arise due to both high stocking density and high feeding rates. One problem resulting from these conditions involves the production of off-flavor compounds typified by geosmin and methylisoborneol that are elaborated as metabolites from certain cyanophycean (blue green) algae, streptomycetes, and/or actinomycetes commonly found in freshwater aquacultural ponds.2 These offflavor compounds, released into the pond water, are bioaccumulated in the fish tissue through the gill membranes or through the alimentary tract in conjunction with food uptake.3 Because of the very low human taste threshold values for these compounds in fish tissue, high detectability methods are required. Popular sample preparation techniques for instrumental analysis of these compounds in catfish tissue involve vacuum4 and
EXPERIMENTAL SECTION Chemicals. MIB (99.5% by GC/MS) was synthesized according to the procedure of Wood and Snoeyink,9 and GEO (99+% by GC/MS) was purchased from Wako Pure Chemicals (Osaka, Japan). The internal standard for GEO, cis-decahydro-1-naphthol (DHN, 98%), was purchased from Aldrich Chemical Co. (Milwaukee, WI) and used as received. The internal standard for MIB, (()-endo-norborneolacetate (NBA), was prepared through the reaction of (()-endo-norborneol and acetic anhydride described previously.7
† Present address: Department of Chemistry, Western Kentucky University, Bowling Green, KY 42101. (1) Armstrong, M. S.; Boyd, C. E.; Lovell, R. T. Prog. Fish-Cult. 1986, 48, 113. (2) Brown, S. W.; Boyd, C. E. Trans. Am. Fish. Soc. 1982, 111, 379. (3) Nijssen, B. Off-Flavors. In Volatile Compounds in Foods and Beverages; Maanse, H., Ed.; Marcel Dekker: New York, 1991; Chapter 19. (4) Lovell, R. T.; Leland, I. Y.; Boyd, C. E.; Armstrong, M. S. Trans. Am. Fish. Soc. 1986, 115, 485.
(5) Martin, J. F.; McCoy, C. P.; Greenleaf, W.; Bennett, L. Can. J. Fish. Aquat. Sci. 1984, 4, 909. (6) Martin, J. F.; Suffet, I. H. Water Sci. Technol. 1992, 25 (2), 73. (7) Conte, E. D.; Shen, C. Y.; Perschbacher, P. W.; Miller, D. W. J. Agric. Food Chem. 1996, 44, 829. (8) Dionigi, C. Water Quality Working Conference, Memphis, TN, personal communication, 1995. (9) Wood, N. F.; Snoeyink, V. L. J. Chromatogr. 1977, 132, 405.
S0003-2700(96)00296-X CCC: $12.00
© 1996 American Chemical Society
Analytical Chemistry, Vol. 68, No. 15, August 1, 1996 2713
Table 1. Percent Recoveries of Geosmin (GEO), Methylisoborneol (MIB), and Their Respective Internal Standards, cis-Decahydro-1-napthol (DHN) and (()-endo-norborneolacetate (NBA) recovery (%)
Figure 1. Microwave distillation-solid phase extraction apparatus.
Apparatus. The microwave distillation-solid phase extraction apparatus used in this work is depicted in Figure 1. Microwave radiation was applied to the fish sample using a 700 W microwave oven (Tappan, Model 56-5472-10/03, Dublin, OH) (A), which was modified as described below. An opening in the microwave oven intended for a meat probe was uncovered. An opening of the same size was drilled through the outside microwave oven covering. This opening was used for placement of a 1/4 in. × 14 in. Teflon gas transfer line (Cole Parmer, Niles, IL) (I). The 1/8 in. × 36 in. Teflon sparge gas tube (Cole Parmer) (E) was inserted through an opening in the bottom of the outer case and then through a small opening in the inner compartment. The sparge gas was regulated by a rotometer (H). Tube E was connected to a check valve (Little Rock Valve and Fitting, Little Rock, AR) (G) to prevent the reverse flow of liquids during and after a distillation. The sparge gas tube (E) was fitted through a Swagelok Teflon male tee (Little Rock Valve and Fitting) (C) and sealed in place with a Teflon reducing ferrule (Alltech, Deerfield, IL) (F). Component C was connected to a 250 mL thick-wall no. 25 Ace-Tred hydrogenation flask (Ace Glass, Vineland, NJ) (B) via a no. 25 Ace-Tred Teflon nut (Ace Glass) (D) equipped with a Teflon O-ring. Argon purged the distillate formed from the microwave radiation through tube I. The flow was directed from I into the 300 mm condenser (No. 5821-05, Ace Glass) (N) via a tee-adapter (No. 5848-46, Ace Glass) (K) and a no. 11 Ace-Tred Teflon union (No. 5841-46, Ace Glass) (M). The tee-adaptor was attached to I and a 1/4 in. Teflon rod (L) by a no. 7 Ace-Tred Teflon bushing (No. 5029-35, Ace Glass) and O-ring (No. 7855-704, Ace Glass) (J). Component L was used as an opening for the addition of the elution solvent. The temperature of the condenser (N) was controlled by a benchtop recirculating cooling unit (Model RB2055AO, FTS Systems, Stone Ridge, NY) (S). On the bottom of the condenser was attached a support disk (No. 5848-07, Ace 2714 Analytical Chemistry, Vol. 68, No. 15, August 1, 1996
spike (ppb)
GEO
5 50 500
91.3 ( 18.0
DHN
MIB
NBA
92.8 ( 8.8 77.6 ( 8.7
78.7 ( 6.1
92.8 ( 8.8 99.6 ( 4.7
Glass) (O) and a no. 11 adapter (No. 5838-72, Ace Glass) (P), which accommodated a 1/8 in. NPT stainless steel union (Little Rock Valve and Fitting) having a male Luer lock fitting epoxy glued to one side (Q). A C18 Sep-Pak (Waters, Milford, MA) (R) was secured to the Luer fitting. Procedure. Prior to its attachment, the Sep-Pak was conditioned by rinsing with 2 mL each of ethyl acetate, methanol, and deionized water in sequence. A 20 g subsample of ground catfish tissue spiked with 50 ppb DHN and NBA was placed in the sample container. The condenser was set at 5 °C, and the argon purge flow was maintained at 40 mL/min. The microwave oven was activated at 40% power for 10 min, during which time distillates were formed and migrated through the condenser. The condenser and Sep-Pak were rinsed with 2 mL of deionized water with a syringe to remove polar interferences through the opening at L. Next, the analytes were eluted by rinsing two times with 1 mL of ethyl acetate through the condenser and Sep-Pak. The collected extract was dried with sodium sulfate, and then 1 µL was injected into the gas chromatograph ion trap mass spectrometer. Chromatographic instrumentation and conditions are the same as described in our previous work.7 The selective ion storage mode of the ion trap mass spectrometer was used to monitor three ions for each analyte during respective retention time windows. For MIB, m/z 95, 135, and 150 were monitored; m/z 93, 94, and 79 for NBA; m/z 97, 126, and 149 for GEO; and m/z 107, 121, and 136 for DHN. Calibration of Response. Calibration solutions were prepared by spiking the internal standards DHN and NBA at 50 ppb and MIB and GEO in a series of samples from 0.5 to 500 ppb into the fish tissue, which was then ground. These samples were prepared according to the described procedure and then injected into the gas chromatograph ion trap detector to calibrate response. Internal standard calibration curves were produced that showed the area ratios of GEO/DHN and MIB/NBA versus the concentration of the analytes. RESULTS AND DISCUSSION Microwave Distillation-Solid Phase Adsorbent Trapping Device. The Sep-Pak design was incorporated to improve percent recoveries of GEO and MIB in the fish tissue over the previous hand-packed solid phase adsorbent design. Table 1 displays the percent recoveries of the analytes at 5 and 500 ppb and the internal standards at 50 ppb. Improved recoveries are obtained relative to the previous hand-packed design.7 The more efficient packing of the Sep-Pak cartridge allows for improved trapping of the analytes due to a decrease in channeling through the adsorbent. Solid phase adsorbents are notorious for carryover problems. This modified design produced no carryover effects above the
Figure 4. Selective ion storage chromatograms from a catfish sample found to contain 72 ppb MIB: (a) ion trace m/z 79 and (b) ion trace m/z 95. Figure 2. Chromatograms of a distillate extract from a 5 ppb methylisoborneol (MIB)-spiked catfish sample: (a) total ion chromatogram, (b) ion trace m/z 95 extracted from the total ion chromatogram, and (c) ion trace m/z 95 extracted from the selective ion storage chromatogram.
Figure 3. Chromatograms of a distillate extract from a 5 ppb geosmin (GEO)-spiked catfish sample: (a) total ion chromatogram, (b) ion trace m/z 112 extracted from the total ion chromatogram, (c) ion trace m/z 126 extracted from the selective ion storage chromatogram, and (d) ion trace m/z 97 extracted from the selective ion storage chromatogram.
detection limit of the instrument at the 500 ppb GEO and MIB level. Because a new Sep-Pak is used for each analysis, incomplete analyte elution will not be a problem as in the earlier version. Gas Chromatography Ion Trap Detection. The ion trap detector was operated in the selective ion storage (SIS) mode as described in the Procedure section, above. Our previous work utilized the total ion chromatogram (TIC) mode. Three representative ions for each compound were chosen on the basis of their intensity relative to instrumental and sample extract matrix background. Figures 2 and 3 illustrate the advantages of the SIS mode versus the TIC mode. Figure 2 shows chromatograms of a 5 ppb methylisoborneol-spiked catfish sample in different modes corresponding to an on-column mass of 46.4 pg. In Figure 2, chromatogram a shows the TIC mode, in which MIB is not observed because of the lower detectability in this mode. MIB is observed in the chromatogram of Figure 2b, which depicts the TIC m/z 95 ion. The chromatogram in Figure 2c is in the SIS mode for the m/z 95 ion, which is the base peak of MIB. The SIS mode offers an improved signal-to-noise ratio (∼1.34 times) compared to the m/z 95 ion TIC mode, which translates into a
lower detection limit. Ion trap selective ion storage does not produce the large magnitude in improvement for monitoring a few ions compared to a full-scan spectrum as does the more widely utilized quadrupole selective ion monitoring (SIM) mode. Numerous ions may be collected in the ion trap compared to a few ions (as in SIS) without any significant decrease in detectivity.10 Figure 3 displays chromatograms of a 5 ppb geosmin-spiked catfish sample corresponding to an on-column mass of 45.7 pg. Again, the analyte is not observed in the TIC mode (Figure 3a). Using the mass chromatogram for the m/z 112 ion, a peak for geosmin is observed (m/z 112 is the base ion of geosmin in the full-scan spectrum). Due to collision-induced dissociation, the m/z 112 ion did not appear in the SIS mode. The chromatograms c and d in Figure 3 depict the mass chromatograms for m/z 126 and 97 ions in the SIS modes, respectively. The greatest signalto-noise ratio is achieved in the SIS mode for m/z 97 (chromatogram d), although baseline resolution from interferences is observed only in chromatograms b and c. Detection Limits and Linearity. Detection limits of the method for GEO and MIB were 0.630 ( 0.109 ppb (m/z 129) and 0.217 ( 0.018 ppb (m/z 95), respectively. These values are 13and 4-fold below the rejection levels of GEO and MIB, respectively. The detection limits are obtained by
DL ) 3sB/S where sB is the standard deviation of the noise and S is the slope or sensitivity of response.11 sB is equal to Np-p/r, where Np-p is the peak-to-peak noise and r is 5, which is an approximation for random Gaussian noise. Linear regression analysis for the calibration of response revealed y ) -0.0388 + 0.0333x, r2 ) 0.9998 for MIB and y ) 0.0067 + 0.0022x, r2 ) 0.9999 for GEO. Applications. A 1.5 lb channel catfish was placed for 6 h in an aquarium that contained 18 gal of water spiked with MIB (10 ppb). The catfish was then removed from the spiked water and placed in an aquarium containing 18 gal of MIB-free water for 7 days. Figure 4 shows chromatograms of the channel catfish tissue microwave extract focusing on the m/z 79 (internal standard, a) and 95 (MIB, b) ions. This catfish tissue was found to contain 72 ppb MIB. (10) Feigel, C. Ultra Trace Analysis of Complex Matrix Samples Using Selective Ion Storage GC/MS; Varian Application Note 21; Varian Chromatography Systems: Walnut Creek, CA; 1994. (11) Foley, J. P.; Dorsey, J. G. Chromatographia 1984, 18, 503.
Analytical Chemistry, Vol. 68, No. 15, August 1, 1996
2715
CONCLUSION This apparatus allows for the determination of methylisoborneol and geosmin in channel catfish tissue below the rejection levels. Compared to more conventional modes of sample preparation, this design allows for faster analysis time with minimal solvent usage and no preconcentration step required. This device has wider applications in the analysis of other semivolatile compounds in solid samples having high aqueous content. Results are consistent with currently used organoleptic analysis, but this method has significant advantages economically and scientifically.
Figure 5. Selective ion storage chromatograms of a mildly off-flavor determined catfish sample containing 9 ppb GEO: (a) ion trace m/z 136, and (b) ion trace m/z 126, and (c) ion trace m/z 97.
Figure 5 depicts chromatograms of an extract from a channel catfish determined to be mildly off-flavored by an organoleptic analyst. The m/z 97 SIS chromatogram that displayed a peak for geosmin was less complex than the corresponding m/z 129 SIS chromatogram. The concentration of geosmin in this tissue was determined to be 9 ppb, which is slightly above the rejection level, a result consistent with the organoleptic categorization.
2716 Analytical Chemistry, Vol. 68, No. 15, August 1, 1996
ACKNOWLEDGMENT This research was supported in part by an appointment for E.D.C. to the Postgraduate Research Program at NCTR, administered through an interagency agreement between the U.S. Department of Energy and the U.S. Food and Drug Administration. Received for review March 26, 1996. Accepted May 23, 1996.X AC960296K X
Abstract published in Advance ACS Abstracts, July 1, 1996.
