Derivatization in trace organic analysis: use of an all-glass conical

Samy Abdel-Baky/ Kariman Allam, and Roger W. Giese*. Department of Pharmaceutical Sciences in the Bouve College of Pharmacy and Health Sciences and ...
0 downloads 0 Views 305KB Size
2882

Anal. Chem. 1992, 64, 2882-2884

Derlvatization in Trace Organic Analysis: Use of an All-Glass Conical Reaction Vial Samy Abdel-Baky,? Kariman Allam, and Roger W. Giese' Department of Pharmaceutical Sciences in the Bouue College of Pharmacy and Health Sciences and Barnett Institute of Chemical Analysis and Materials Science, Northeastern University, Boston, Massachusetts 02115 As we have reported,1*2we are developing an analytical procedure for diol epoxide-polyaromatic hydrocarbon (DEPAH) DNA adducts in which the isolated adducts are subjected to a sequence of three chemical steps prior to detection by gas chromatography-electron capture negative ion mass spectrometry (GC-ECNI-MS). The three steps are as follows: (1)acid hydrolysis to release the DE-PAH as a tetrol from the DNA, (2) oxidation of this tetrolwith potassium superoxide to a corresponding dicarboxy-PAH, and (3) esterification of the latter compound with pentafluorobenzyl bromide. During the course of progressively applying steps 2 and 3 of the method to lower amounts of an authentic tetrol (namely benzo[a]pyrene-7,8,9,10-tetrahydrotetrol),we encountered severe irreproducibility at the 100-pg level in terms of both losses and interferences. Here we report that this problem was due to microscopic, recalcitrant brown/black specks in the commercial vials employed for this purpose and the presence of PTFE liner/plastic caps. Reproducible results were achieved by employing an all-glass conical reaction vial fabricated in-house. EXPERIMENTAL SECTION General Information. The reagents, instrumentation,commercial vials, and reactions were described All solvents used for the following cleaning procedures were HPLCgrade. All-Glass Reaction Vial. This was fabricated in-housefrom a 14/20 hollow, ground-glassstopper and a size 14/20,7585 joint (outer member, medium length, 12-mm0.d.) obtained from ACE Glass (Vineland, NJ). Vial Cleaning. Scrub each vial (commerical or home-made) with a test tube brush (Cat. No. 17030-006, VWR Scientific, Philadelphia, PA) in water:Micro (No. 6731, International Products Corp., Burlington, NJ), 90/10v/v. Heat the vial at 60 O C for 2 h in a covered (Pyrex watch glass) beaker filled with 10% Micro cleaning solution, and leave at room temperature overnight. Wash the vials individually lox with distilled water and keep in water for 2 2 h. Heat the vials in chromic acid3for 2 h at 60 OC, and leave at room temperature overnight. Wash the vials as before with water, and stack them upright in a Petri dish. Fill each vial with methanol, and 0.5 h later remove the methanol with a pipet attached to a vacuum source. Loosely cover the dish with aluminum foil,remove the residual methanol on a hot plate, and sealthe vials after cooling to room temperature with a polytetrafluoroethylene (PTFE) linear/plastic cap or ground glass stopper, cleaned as described below, for storage until use. Cap Cleaning. Rinse the caps briefly with water and keep in 10% Micro at room temperature for 224 h. Wash the caps individually with water and keep in water for 1 2 4 h. Heat the caps in methanol at 60 "C four times for 2 h each. Partly air-dry the caps upside down on a paper towel, transfer to and wrap

* Address reprint requests to this author. t Current address: BASF Corp., Research Triangle Park, NC 27709-

9.528.

(1)Li, W.; Sotiriou, C.; Abdel-Baky, S.; Fisher, D. H.; Giese, R. W. J. Chromatogr. 1991,588, 273-280. (2) Allam, K.; Abdel-Baky, S.; Giese, R. W. Anal. Chem. 1992, 64, 238-239. (3) Vogel, A. Textbook of Practical Organic Chemistry; John Wiley: New York, 1987; p 3.

loosely in aluminum foil, place in a 70 "C oven overnight, and store wrapped tightly until use. PTFE-Liner Cleaning. Three-Step Method. The method is the same as above for the caps. Six-Step Method. After 10% Micro and a water wash (see above), stir the liners in chromic acid at 50 O C for 1 h. Stir in 60 O C water 6X, 70 O C toluene 3X, and 50 O C methanol 3X (including 10 min of boiling in the last methanol), with 20 min for each step. Air-dry in the covered (loose aluminum foil) beaker, place in a 70 "C oven overnight, and store until use wrapped in aluminum foil.

RESULTS AND DISCUSSION We decided to first examine whether the vials themselves were contributing to the irreproducibility that we had encountered at the 100-pglevel of tetrol analyte, even though the vials were routinely cleaned with detergent, 3 M nitric acid (both at 60 OC for 8 h), and methanol. While the vials were always clean by ordinary visual inspection, closer examination with a hand-held magnifying lens (X5) revealed the presence of tiny brown and black specks, stuck to the inner wall, that varied vial-to-vial. New vials, examined as received from two other suppliers, were found to contain the specks. As implied above, the specks were recalcitrant to our existing cleaning procedure. A second problem became apparent as well from use of the magnifying lens. When the plastic cap fitted with a PTFE liner was screwed down tightly (as necessary) onto a vial to seal the contents for the subsequent chemical reaction, white particles (having the shape of "shavings") were dislodged into the vial, obviously from abrasion of the PTFE-liner against the lid of the vial. This could be seen by placing some methanol in the vial, tightening down the PTFE-lined cap, inverting the vial to contact the liner with the solvent, and observing the sinking white particles with the magnifying lens. The presence of the brown/black specks (at least a t the magnifying lens level) was overcome by a different cleaning procedure (including the rigorous use of a micro test tube brush). The white particles were overcomeby fabricating an all-glass vial consisting of a ground-glass stopper attached to a short, conical tube (see Figure 1). Apparently, a vial of this type and dimensions is not available commercially at the present time. To compare the performance of the old (PTFE-lined plastic cap) and new (glassstopper) vials, both cleaned free of brown/ black specks (the specks were never encountered in the glassstoppered vials), we conducted three experiments in which only blank vials (no added tetrol) were tested. Experiment I: reactions 2 and 3 (see Introduction or Table I) were conducted sequentially in vials fitted with plastic caps and PTFE-liners, where these caps and liners had been separately cleaned with detergent, water, and hot methanol (three-step cleaning) prior to assembly. Experiment II: same as I except the liners were cleaned stepwisewith detergent, water, chromic acid, water, hot toluene, and hot methanol (six-step cleaning). Experiment III: the caps and liners were cleaned as in 11,but only reaction 3 was conducted. At the end of reaction 3 in each case, sample cleanup was performed as described? followed by injection into the GC-ECNI-MS.

0003-2700/92/0364-28S2$03.00/0 0 1992 Amerlcan Chemlcal Society

ANALYTICAL CHEMISTRY, VOL. 64, NO. 22, NOVEMBER 15, 1992

Figure 1. AiLglass conical reaction vial that was fabricated in-house.

Table 1. Interference Level by GC-ECNI-MS When Blank Reactions (NoAnalyte Added) Are Conducted in a Vial. Sealed with P, a PTFE-Lined Plastic Cap, or G, a Glass Stopper experiment (reacn 2 = KO2 oxidation; reacn 3 = pentafluorobenzylation) I. reacns 2 and 3 P (3-step c1eaning)a

G 11. reacns 2 and 3 P (6-step cleaningp

G 111. reacn 3 P (6-step cleaning)"

G

contamination level specific general (PPIvia! (abundance of a coelutmg, x 10-3 unknown compound)b a t 10 min)c 23 f 35 2.3 f 1.3

144 f 21 58 f 29

8.1 f 8.4 4.2f 3.2

59 f 14 30 f 22

0.2 f 0.2

18f1 16 f 2

0.0

All vials were cleaned free of microscopically-visiblebrown/black specks prior to this experiment. *An amount is assigned to the unknown compound by assuming that it is contaminating analyte. It is meaningful to compare the general abundance values because all of the MS measurements were made on the same day and the baseline for the instrument was constant throughout this day.

The results of these experiments are summarized in Table

I. Both the magnitude of a specific interference peak (a contaminating peak that always coelutes with authentic analyte) and the abundance of the general background signal a t an arbitrary retention time (but where major interfering peaks are absent) are presented as two measures of the contamination level. Most importantly, the specific interference was fully overcome only in the all-glass vial in experiment 111-G (G = glass) where only reaction 3 was conducted. This interference shows up consistently once reaction 2 is done as well (2.3 f 1.3 pg/vial in experiment I-G and 4.2 f 3.2 pg/vial in experiment 11-G). (An amount is assigned to the unknown interfering peak by assuming that it is contaminating analyte; see below.) The level of this interference is highest and most variable (23 f 35 pg/vial) when reactions 2 and 3 are conducted sequentially in a commercial vial capped with a plastic cap and PTFE liner, where the last two components have been cleaned only with the three-step procedure (experiment I-P; P = plastic cap + PTFE liner). More thorough cleaning of the PTFE liner (experiment 11-P)reduces the interference to 8.1 f 8.4 pg/

2889

Flgurr 2. Representative GC-ECNI-MS selected ion (mlz 469) chromatogamsforbiankvlalsfrom(A)expefimentI 1 and(B)experlment I11 (see Table I). For each pair of chromatograms, trace P Is from a vial sealed with a PTFE-lined plastic cap and trace G is from a vial sealed with a glass stopper. The unknown designated peak eluting near 12 min in each chromatogram (except trace G in (B), where the peak is absent) is consbred a spectfic interference, as indicated In Table I, since it always coelutes with authentic product, 2,3-bis(pentafiuorobenryi)pyrenedicarboxylate. (Largely becauseof variation In the vacuum of the MS, the retentbn time of the apparent analyte peak is not constant, but the coelution of this peak with true analyte was always confirmed by coinjecting authentic analyte.)

vial. Thus, both reaction 2 and the PTFE liner/plastic cap are origins of the specific interference. We postulate, for two reasons, that the specific interference is due to the analyte or an analyte analog arising as a contaminant in the procedure. First, this interference shifts in its retention time exactly the same as the analyte when each is derivatized (after superoxide oxidation) with p-tetrafluorobenzyl bromide instead of pentafluorobenzyl bromide (data not shown). Second, benzo[alpyrene, an ultimate parent compound for the analyte, is ubiquitous in the environment, and the superoxide step is known to convert a variety of oxidation products of benzo[olpyrene to our intermediate product 2,3-pyrenedicarboxylicacid.4 Returning to Table I, we see that the general contamination level also is always lower, more or less, when the all-glass vial is used. To more directly show the type of data leading to Table I, selected ion chromatograms are shown in Figure 2 from representative, corresponding P and G vials utilized in experiment I1 (Figure 2A) and in experiment I11 (Figure 2B). A possible limitation of the glass-stoppered conical vial for some reactions is that the seal is not as tight as that provided by a PTFE-lined plastic cap. The volume reduction was 25 20 pL for reaction 2 (conducted a t room temperature for 18 h) and 50 21 p L (50 O C for 5 h) for reaction 3 in the all-glass vial. In this case, however, comparable volume reduction also takes place in the plastic cap/PTFE liner vial: 25 20 p L for reaction 2 and 50 35 p L for reaction 3. For the latter vial, the solvent loss apparently occurs through the spaces at the top of the capped vial, by absorption of solvent vapor into the PTFE liner and plastic cap, or by both. As a less convenient alternative to the all-glass vial presented here to conduct a chemical reaction, a glass ampule (available commercially) or small tube can be flame-sealed under vacuum for this purpose (e.g. ref 5). Relative to a conventional vial, cleaning the all-glass vial is easier since the former involves three different components (vial, plastic cap, and PTFE liner). The improved reproducibility for the blanks extends to our tetrol samples. When eight, 100-pgsamples (asduplicates in four separate experiments) of tetrol were subjected to

-

-

-

-

(4) Sotiriou,C.; Li, W.; Giese, R. W. J.Org. Chem. 1990,55,2159-2164. ( 5 ) Crain, P. F.; McCloskey, J. A. Anal. Biochem. 1983,132,124-131.

2664

ANALYTICAL CHEMISTRY, VOL. 64, NO. 22, NOVEMBER 15, 1992

reactions 2 and 3, the overall absolute yield of product by GC-ECNI-MS was 36 i8% ( x f SD). For the last duplicate tested, an isotopic internal standard1 was included, which gave peak ratios (analyte/IS) of 1.81 and 1.80. Thus, the all-glass vial is useful both in terms of ita convenience and the reproducibility of the procedure that we

HealthEffects Institute (HEI),an organization jointly funded by the U.S.Environmental Protection Agency (EPA) (Assistance Agreement X-812059) and automative manufacturers. The specific grant was HE1 Research Agreement 86-82. The contents of this paper do not necessarily reflect the views of the HEI, nor do they necessarily reflect the policies of the EPA or automotive manufacturers. Contribution No. 550 from the Barnett Institute.

ACKNOWLEDGMENT This work was funded by Grant OH02792 from the Centers for Disease Control and in part under a contract from the

RECEIVED for review May 15, 1992. Accepted August 17, 1992.