Anal. Chem. 1984, 56, 2687-2692
2687-
Gas Chromatographic Determination of Chemical Residues in Food Using a Rugged High-Resolution Mixed-Bed Column James
L. D a f t
US.Department of Health and Human Services, Food and Drug Administration, Kansas City, Missouri 64106
A 1.8m mlxed-bed column of 3% OV-225/5% OV-101 (1:l) Is used to determlne chlorlnatedorganophosphoruspestlclde resldues In food. The test column Is compatlble wlth ECD, FPD, MCD, and HECD detectlon. The precisely controlled 1:l bed mlxture glves the best separatlon to detected residues Isothermally at 200 and 150 OC wlth mlnlmal phase "bleed". The test column Is compared to three currently used GC Screening columns of (1) 5 % OV-101, (2) 5 % OV-105, and (3) 7'/*% OV-210/5% OV-101 (1:l). The test column s e p arates nlne detected resldue comblnatlons which are not adequately resolved In 20-mln sample scans wlth the other three columns. I n addltlon, the test column mlnlmlres lnterference by sample coextractlve which sometlmes obscures the detection of Incurred resldues wlth the current-screening columns. The analytical capaclty achleved by alternately uslng all four of these columns Improves quallty control for about 80 regularly determlned resldues In faster analysls tlme.
Heavily researched 20 years ago to improve column selectivity (1-24) and, later, to eliminate "redundant" stationary phases (25-29), gas chromatographic mixed-type columns are most useful in determining residual chemicals in complex food samples. Through a random sampling of 110 residue specialists in research institutes, universities, state, EPA, and FDA laboratories in the United States (30),I have found that 89% of 64 respondents to a questionnaire say they use variations of two mixed columns in the analysis of water, soil, and food. The columns cited are binary mixtures of OV-210 (trifluoropropyl methyl silicone) and OV-101 (methyl silicone) (6) or OV-17 (phenyl methyl silicone) (7, 8, 31). Nevertheless, this current application of mixed columns does not give complete selectivity to food-incurred chemical residues. For example, residues of pentachlorophenol or acephate must be determined, respectively, with phenolic or polyester-type packings. In terms of separation, early eluting residues ( t R < 1-4 min) such as trichloro- or tetrachlorobenzenes may coelute on commonly used methyl silicone packings. Since column sensitivity diminishes as residue t R increases, so-called late eluters ( t R = 15-30 min) including p,p'-methoxychlor and endosulfan sulfate may go undetected at 10-ng residual levels on polar single-bed packings such as OV-225 (phenyl cyanopropyl methyl silicone). In addition, well-known residue combinations such as HCB/pentachloroanisole, octachlor/heptachlor epoxides, and degradation products of p,p'-DDT are not adequately separated on many general purpose column packings. Furthermore, if a complex food sample yields detector response (sample effect) which obscures the detection of residues, as many as 10 different GC columns, some at uncommon instrumental conditions, may be needed to isolate and confirm all chemical contamination. The superior separating capability of currently used mixed columns minimizes this amount of GC work. However, excessive OV-210 phase "bleed" is msociated with these columns. Consequently, these columns are used more for identity confirmation of detected residues than for initial screening
analysis. Generally, these columns are neither used in automated instrumentation where constant retention factors are needed or where programmed temperatures up to 250 "C might be encountered nor used with HECD detectors which might be damaged by the effluent bleed. Although mixed columns are used in this restricted sense, they appear to be used extensively. According to my survey of residue specialists, nearly 85% of the respondents say their mixed columns yield satisfactory retention data and yield desired analytical results most of the time. Eighty-four percent say an improved mixed column would support the technology. This sampling trend suggests that the selectivity gained through these columns is more important than bleed tendencies. This trend also may have been augmented by the fact that newer OV-210 liquid phases are more thermally stable than those of the past. In any case, packed columns, including the mixed-bed type, are the mainstay for the GC analysis of complex real-world samples such as food. Although capillary GC is assuming part of this function, a unique improved mixed-bed column is needed in this analysis. In this report, a rugged mixed-bed column of 3% OV225/5% OV-101 (l:l),referred to as the "225-mixed" or the test column, is evaluated for the GC screening determination of chlorinated-organophosphorus (COP) residues in food. Evaluations include (1) suitability of the column for use with four GC detectors (ECD, FPD, MCD, and HECD), (2) its durability with food sample injection, (3) its selectivity as a function of isolating and separating food-incurred residues, and (4) its ability both to outperform and to supplement currently used screening analysis columns in specific determinations.
EXPERIMENTAL SECTION Gas Chromatographs. The instruments used were Tracor 560s equipped with nickel-63 electron capture (ECD), flame photometric-phosphorus/sulfur mode (FPD-P/S), microcoulometric-chloride mode (MCD-Cl), and Hall electroconductivity-chloride mode (HECD-C1) detection. The glass columns used were 1.8 m X 2 or 4 mm i.d. Operating conditions were as follows: injector port, 225 "C; column, 200-210 "C and 130-150 "C, isothermal; ECD, 350 "C; FPD-P/S, 190 "C; MCD-Cl, transfer line, 225 "C, valve block, 210 "C, pyrolysis tube, 800 "C, oxygen reaction gas flow, 60-80 mL/min, nitrogen make-up gas flow, about 100 mL/min, and manual vent time, 1min; HECD-C1, reactor base, 250 "C, reactor, 900 "C, conductivity solvent flow (n-propyl alcohol),0.4 mL/min, hydrogen reaction gas flow, 60-100 mL/min, and automated vent time, 0.50-0.75 min. Carrier gas for each detector was 5% argon/methane for ECD, nitrogen for FPD-P/S or MCD-C1, and helium (ultrapure grade filtered through Oxiclear and activated charcoal) for HECD-C1. Carrier gas flow was about 40 mL/min for 2 mm i.d. columns or 90 mL/min for 4 mm i.d. columns. At 200 "C, sensitivity was 50% full-scale (fs) deflection for 1 ng of heptachlor epoxide (tR = 4-5 min) with ECD and HECD-C1 or 2 ng of heptachlor epoxide with MCD-C1 and 50% fs for 2 ng of parathion ( t =~4-5 min) with FPD-P/S. At 150 "C, sensitivity was 50% fs for 0.6 ng of a-BHC ( t =~7-8 min) with ECD.and HECD-C1. Recorder attenuation ranged from 2x to 1OOx; chart speed was 8-12.7 mm/min. Sampling, Cleanup, and Reference Standards. The food samples used in this study came from the Food and Drug Ad-
This article not subject to U.S. Copyright. Published 1984 by the American Chemical Society
2688
ANALYTICAL CHEMISTRY, VOL. 56, NO. 14, DECEMBER 1984
Table I. Reference Standard Solutions Used at 200 OC: Retention Time Ratios Relative to Chlorpyrifos for each Column Noted
RR,
compound
pg/mL
CY-BHC T-BHC heptachlor aldrin heptachlor epoxide cis-chlordane dieldrin endrin
0.06 0.10
225, mixed
OV-101 OV-105
210, mixed
1
0
5 minuter
IO
15
0.20
0.43 0.57 0.69 0.83 1.25
0.40 0.48 0.83 1.05 1.29
0.42 0.52 0.76 0.93 1.27
0.39 0.50 0.69 0.86 1.26
0.20
1.61
1.66
1.62
1.48
0.30 0.30
1.90 2.19
1.91 2.13
1.87 2.10
1.90 2.18
0.10 0.10
2
same as 1 but at half concn
Figure 1. Chromatogram of 17 pesticide compounds from the 3 % OV-225/5% OV-101 (1:l) test mixed column at 150 OC; ECD. (a) 0.15 ng of 1,3,5-trichlorobenzene, (b) 0.05 ng of hexachlorobutadiene,(c) 0.15 ng of 1,2,3,5-tetrachlorobenzene, (d) 0.15 ng of 1,2,3,4-tetrachlorobenzene, (e) 0.15 ng of hexachioronorbornadiene, (f) 0.10 ng of pentachlorobenzene,(9) 0.20 ng of tecnazene (TCNB), (h) 0.25 ng of 2,3,5,6-tetrachloroaniline, (i) 0.10 ng of hexachlorobenzene (HCB), (j)0.15 ng of pentachloroanisole, (k) 0.10 ng of 2,3,5,6-tetrachlorothioanisole, (I) 0.25 ng of a-BHC, (m) 0.25 ng of quintozene (PCNB), (n) 0.40 ng of 7-BHC, (0)0.25 ng of 2,3,5,6-tetrachloroanisidine, (p) 0.40 ng of pentachloroaniline (PCA), (9) 0.20 ng of pentachlorothioanisole (PCTA).
3
ministration’s Total Diet Study. Samples were made in tableready form by peeling, cooking, etc., before compositing by item (32) or by food category (33) with blending or grinding. Composited samples were extracted for chlorinated-organophosphorus (COP) residues by procedures in the Pesticides Analytical Manual (34). Lipids were removed from fatty food extracts by gel permeation chromatography (35).Final cleanup of fatty and nonfatty extracts before analysis by GC was done with Florisil chromatography (34, 36). The 58 reference standards used in this study were made in isooctane. Each solution contained from one to nine compounds (see Tables I and 11). Multiple-component compounds such as toxaphene, technical chlordane, and Aroclors 1248,1254, and 1260 were used at 2-4 wg/mL concentrations. Preparation of the 225-Mixed Column. Separate phase beds of 3% OV-225 and 5% OV-101 (w/w) were prepared on Chromosorb W (HP),80-100 mesh, and were conditioned in 10 mm diameter tubes for 72 h at 240 OC with a nitrogen purge. The final packing was made by mixing portions of these two phase beds at a 1:l ratio. Columns packed with this mixture for use with HECD were conditioned an additional 48 h at 220 “C. Three chromatographic control points were established to ensure precise maintenance or duplication of packing mixture Characteristics. These control points comprised the exact chromatographic placement of two pesticide combinations and the retention time ratio of p,p’-DDT relative to chlorpyrifos (RR,), When these points were used as the reference, the column’s elution characteristics were “controlled by adding one of the phase beds to the packing mixture as needed. The first control point used was a pesticide combination of tetrachloroaniline, hexachlorobenzene (HCB), and pentachloroanisole at 150 OC (see Figure 1). These three reference standards eluted as shown, giving a clean separation to HCB and pentachloroanisole. Adding the OV-225 phase bed to the packing mixture caused tetrachloroaniline and HCB to merge; adding the OV-101 phase bed caused HCB and pentachloroanisole to merge. The pesticides of the second control point were y B H C and heptachlor at 200 “C (see Figure 2). These two reference standards were almost separated with HECD-C1 or were cleanly separated with ECD. Adding the OV-225 phase bed to the packing mixture caused these compounds to merge; adding the OV-101 phase bed caused them to separate up to 1 min.
4
hexachlorobenzene (HCB) p,p’-dicofol trans-chlordane trans-nonachlor PJJ’DDT p,p’-methoxychlor pentachlorobenzene pentachloroanisole quintozene (PCNB) pentachloroaniline pentachlorothioanisole octachlor epoxide p,p’-DDE p,p’-TDE
0.03
0.36
0.45
0.39
0.36
0.05
1.12
1.49
0.99 1.49
1.06 1.53
1.11
0.10
0.10
1.63
1.75
1.73
1.45
0.30 1.00
3.43 6.05
3.13 4.70
3.19 4.83
4.10
0.02
0.19
0.24
0.21
0.21
0.02
0.38
0.46
0.41
0.39
0.04
0.48
0.51
0.50
0.52
0.05
0.71
0.67
0.68
0.60
0.04
0.83
0.94
0.85
0.80
0.10
1.15
1.33
1.24
1.18
0.15 0.20
1.79 3.07
1.88 2.41
1.80
2.62
1.61 2.31
0.03
0.27 0.64 1.09
0.29 0.42 1.06
0.29 0.50 1.08
0.32 0.60 1.30
0.20
1.52 2.96
1.63 2.19
1.55 2.49
1.62 2.50
0.20
5.35
2.83
3.63
4.60
0.10 0.10
0.95 1.01
0.43 0.50
0.65 0.79
0.51 0.58
1.34
2.80
5
tecnazene dicloran DCPA (Dacthal) endosulfan I endosulfan I1 endosulfan sulfate
0.05 0.05 0.10
6 0-BHC
6-RHC 7
0.47 0.52 0.51 0.48 diazinon 0.10 0.52 0.52 0.53 0.56 fonofos 0.10 0.82 0.81 0.82 0.83 ronnel 0.20 1.00 1.00 1.00 1.00 chlorpyrifos 0.20 1.01 1.27 0.91 1.19 malathion 0.30 1.39 0.98 0.85 1.30 fenitrothion 0.40 1.61 1.16 0.98 1.42 parathion 0.20 1.99 1.59 1.43 2.25 methida0.60 thion 2.82 2.76 3.25 2.56 ethion 0.30 The third control point, the RR, of p,p’-DDT at 200 OC, ranged from 3.35 to 3.50 for a freshly made packing. Adding the OV-225
ANALYTICAL CHEMISTRY, VOL. 56, NO. 14, DECEMBER 1984
Table 11. Reference Standard Solutions Used at 130 and 150 "C: Retention Time Ratios Relative each Column Noted
RR,, 130 "C OV-101 OV-105
to
Pentachlorobenzene for
210, mixed
RR,, 150 "C 225, mixed
0.16 0.22 1.00 1.92 2.67 3.83 4.73 6.82
0.18 0.24 1.00 1.69 2.50 3.44 5.58 6.53
0.25 0.41 0.54 0.84
0.27 0.43 0.56 0.87 2.02 3.05
pg/mL
225, mixed
1,3,5-trichlorobenzene hexachlorobutadiene pentachlorobenzene tecnazene (TCNB) pentachloroanisole quintozene (PCNB) pentachloroaniline (PCA) pentachlorothioanisole (PCTA)
0.03 0.01 0.02 0.04 0.03 0.05 0.08 0.04
0.14 0.20
0.15 0.24
1.00
1.00
1.00
1.78
7.72
1.37 2.54 2.98 4.32 6.78
1.60 2.64 3.37 5.22 7.06
1,2,3-trichlorobenzene
0.03 0.03 0.03 0.03 0.05 0.05 0.08
0.24 0.40 0.52 0.86 2.14 3.38 5.08
0.22 0.40 0.50 0.85 1.55 2.05 2.76
0.29 0.44 0.57 0.93 1.89 3.73
2.57 3.50
0.19 0.40 1.76
1.77
0.21 0.42 1.65
0.23 0.44 1.70
2.42
0.18 0.40 1.68 2.40
0.22 0.42
heptachloronorbornene hexachlorobenzene (HCB)
0.04 0.03 0.03 0.02
2.32
2.27
2.27
2,3,5,6-tetrachloroanisole 2,3,5,6-tetrachlorothioanisole 2,3,5,6-tetrachloronitroanisole 2,3,5,64etrachloroanisidine
0.02 0.02 0.03 0.05
1.07 3.04 4.84 5.50
1.01 3.01 3.57 3.79
1.08 2.89 4.40 4.61
1.10
2.88 5.15 4.46
1.08 2.82 4.43 5.01
compound 1
2.74
3.86 6.53
0.18 0.25
2
1,2,3,5-tetrachlorobenzene 1,2,3,4-tetrachlorobenzene
hexachloronorbornadiene 2,3,5,6-tetrachloroaniline
(u-BHC Y-BHC 3 1,2,4-trichlorobenzene 1,2,4,5-tetrachlorobenzene
4
O
W
5
I
mnuter
I5
20
Figure 2. Chromatogram of 16 pesticide compounds from the test mixed column at 200 OC; HECD-GI. (a) 0.30 ng of tecnazene, (b) 0.30 ng of hexachlorobenzene,(c) 0.30 ng of CY-EHC,(d) 0.50 ng of y-BHC, (e) 0.50 ng of heptachlor, (f) 0.50 ng of aldrin, (9) 0.50 ng of DCPA (Dacthal), (h) 0.50 ng of o,p'-DDE, (I) 0.75 ng of p,p'-DDE, (j) 0.35 ng of o,p'-TDE, (k) 1.00 ng of o,p'-DDT, (I) 1.OO ng of p ,p'-TDE, (m) 1.50 ng of p,p'-DDT, (n) 3.00 ng of o,p'-methoxychlor, ( 0 ) 1.50 ng of endosulfan sulfate, (p) 5.00 ng of p ,p'-methoxychlor.
phase bed to the packing mixture caused this RR measurement to increase; adding the OV-101 phase bed caused it to decrease. Control points 1and 2 provided routine visual checks of column characteristics. Control point 3 gave exact measurement to mixture proportions. Adjustment or maintenance of the bed mixture with these control point settings yielded the best separation of detected COP residues. R E S U L T S A N D DISCUSSION Suitability Tests. The 225-mixed column was compatible with ECD, FPD-P/S, MCD-C1, and HECD-C1, and it will work with potassium chloride thermionic detection. The column was not compatible with nitrogen/phosphorus detection (NPD), yielding inconstant base-line and poor detector response.
2.74
2689
1.71
4.42
The average amount of phase loading for the test column was kept at 4% to achieve a bed-mixing ratio of 1:l which usually needed no adjustment. This loading percentage also produced optimal column performance with repeated injection of food sample (31,37,38). Peak shape and peak symmetry of detected residues were satisfactory as shown in the sample chromatograms of this report. Column efficiency was measured with the theoretical plate formula, n = 16(t/W2,at 200 "C. Efficiency was also measured simultaneously on two separate columns of 5% OV-101 and 71/2% OV-210/5% OV-101 (1:1),called "210-mixedn, using the following ten reference standards a t 50% full scale: y-BHC, aldrin, heptachlor epoxide, trans-chlordane, cis-chlordane, trans-nonachlor, dieldrin, endrin, p,p'-DDT, and p,p'-methoxychlor. The average (10) n for the test column was 2200 compared to 1700 for the OV-101 column and 2100 for the 210-mixed column. Peak resolution (R)was observed with the test column by increasing carrier gas flows by 30-50%. Peak pairs which normally separated at the prescribed flows, including those with tR of