2730
Anal. Chern. 1887, 59,2730-2732
with the findings of Brown and Watkinson (7) and Lott et al. (8). Excess HNOBcan be removed from the digest by evaporation, provided that reducing conditions in the digest do not occur. A convenient stage to stop the evaporation of excess HN03 is when the color of the digest changes from a light yellow-brown to purple, or when the volume of the digest reaches 1mL. If complete removal of HN03 is necessary, then formic acid may be substituted for acetic acid. After formic acid is added, the tube should be heated at 130 "C for 'I2 h to remove Hz02and excess HN03 before adding HCl to reduce S e 0 2 - b SeO$-, as described in Reamer and Veillon ( I ) . Our experience indicates that acetic acid tends to provide more reproducible Se values than formic acid, and acetic acid can be added with HCl in the reduction step, rather than adding the two reagents in sequence. Acetic acid heated at 130 "C does not destroy "OB, however. The presence of up to 1/2 mL of HN03 does not interfere with the fluorometric analysis of Se. The NH,OH.HCl in the stabilizing solution helps minimize oxidation of the DAN by HN03 (9).
ACKNOWLEDGMENT The authors thank D. Scherer and A. Jacobson for providing the tomato leaf sample and the value for its Se content obtained by using the HN03/HC104 digestion method.
Registry No. Se, 7782-49-2; HBP04,7664-38-2; "Os,
7697-
37-2; H20z,7722-84-1; Mn, 7439-96-5.
LITERATURE CITED (1) Reamer, D.C.; Veillon, C. Anel. Chem. 1983, 55, 1606-1608. (2) Gorsuch, T. T. Analyst (London) 1959, 84, 135-173. (3) Levesque, M.; Vendette, E. D. Can. J. Sol/ Scl. 1971, 51, 85-93. (4) Koh, T. S.;Benson, T. H. J. Assoc. Off. Anal. Chem. 1983, 66,
918-925. (5) Krivan, V., Petrick, K.; Welz, 8.; Melcher, M. Anal. Chem. 1985, 57, 1703-1706. (6) Creaser, I. I.; Edward, J. 0. Top. Phosphorus Chem. 1972, 7, 379-432. (7) Brown, M. W.; Watkinson, J. H. Anal. Chim. Acta 1977, 89, 29-35. (8)LOR, P. F.; Cukor, P.; Moriber, 0.; Solga, J. Anal. Chem. 1963, 35; 1 159-1 163. (9) Aliaway, W. H.; Cary, E. E. Anal. Chem. 1964, 36, 1359-1362.
Allen Dong V. V. Rendig* R. G . Burau G . S. Besga Department of Land, Air and Water Resources University of California Davis, California 95616
RECEIVED for review April 20, 1987. Accepted July 28, 1987. Funds for this research were provided by the UC Salinity/ Drainage Task Force and the Kearney Foundation of Soil Science.
Existence of Self Chemical Ionization in the Ion Trap Detector Sir: In a recent paper in Analytical Chemistry (I), Olson and Diehl made several references to self chemical ionization (self-CI) while discussing the results of their work with the ion trap m m spectrometer system (2). The concept of self-CI was invoked to explain the observation of intense (M + 1)' ions in the spectra of compounds measured n... at high concentrations of GC eluents ..." with the ion trap detector. Self-CI has been defined (3) as chemical ionization in which the reagent ions are fragment ions from the analyte neutral molecule. The process was observed in a Fourier transform mass spectrometer, which used a trapped ion cell and an appropriate time delay between ion formation and ion detection. Thii method of chemical ionization was most efficient for molecules which produce high levels of low-mass fragment ions, such as fatty acid esters. The abundant low-mass fragment ions serve as proton-transfer reagents during a time delay of 20 ms or longer in the trapped ion cell. Olson and Diehl noted that "the ion trap spectra of the phenol, aniline, alcohols, and nonanal did not show (M + 1)+ ions due to CI effects; however the dicyclohexylamine and esters did. The initial spectra obtained during the elution of the dicyclohexylamine showed a small molecular ion at 181 and no (M + 1)+ion at 182. As the concentration of dicyclohexylamine reached a maximum, self-CI effects resulted in the formation of the 182 ion; however, the rest of the spectrum did not change significantly. The (M + 1)+ions were also found in the spectra of the three methyl esters ...". Unfortunately the authors did not give the quantities of the various substances injected into the chromatograph or the concentrations corresponding to the onset of CI effects. Also, although not stated, we assume that the ions were observed in the bar graph spectra from the ion trap data system and not by direct observation of the mass peak profiles. We have observed, in bar graphs from the ion trap data system, the frequent occurrence of unusually abundant (M
+ 1)+ions in the spectra of many classes of compounds. The occurrence of these intense (M 1)+ions is concentration dependent. Below about 50-ng injected into the gas chromatograph, which is the only method of sample introduction provided with the ion trap, (M + 1)+ ions due to naturally occurring isotopes occur at normal abundances when the M+ ion itself is observed. Above about 50 ng, and this varies somewhat with the compound, when M+ ions are observed, the relative abundances of the (M + l)+ ions are sharply higher than expected from isotopic contributions and often exceed the abundances of the M+ ions. All of our observations were made with helium as the carrier gas, with none of the usual chemical ionization reagent gases present, and with no other compounds in the ion trap concurrent with the substance being measured. The ion trap is designed to operate with a relatively high background pressure of helium, that is 0.001 Torr, which, as described by the designers (2), "... is used as a medium to damp the motion of ions trapped in the device, and has been found to enhance the mass resolution, sensitivity and detection limit". The conditions present in the ion trap during our experiments do not seem likely to promote self-CI, although ion storage times of the order of tens of milliseconds may indeed facilitate processes at higher concentrations similar to those observed in the Fourier transform instrument. An alternative explanation (4) for the elevated abundances of (M + 1)+ions in data system bar graphs at higher concentrations is the algorithm used by the ion trap data system to interpret ion peak profiles and its failure to correctly interpret profiles broadened by saturation and space charging. The ion peak profile in the region of the M+ in the ion trap spectrum of 50 ng of pyrene ( 5 ) is considerably broadened, saturated, and possibly shifted to higher mass. Also, doubly charged molecular ions, M2+, from polycyclic aromatic hydrocarbons (PAH) appear as broadened ion peaks even when
0003-2700/87/0359-2730$01.50/0 0 1987 American Chemical Society
+
ANALYTICAL CHEMISTRY, VOL. 59, NO. 22, NOVEMBER 15, 1987
F
2731
7 t
I I
42 SPEC)
48
68
88
1W
120
140
168
188
200
Flgure 1. Ion trap mass spectrum from 5 ng of anthracene-d,,.
low, 5-ng amounts are injected (5). The ion trap data system interprets these broadened signals incorrectly because it does not find the centroid of the ion peaks, but makes an ion intensity measurement a t each digital-to-analog (DAC) converted step of the mass scale, and allocates ion intensity at each step to an appropriate mass. With broad peaks that span a number of DAC steps, intensity may be assigned to ions at several adjacent masses. Thus the enhanced (M 1)+abundances in the bar graphs of ion trap spectra are a t least partially artifacts of the data acquisition system. Similarly, unusual (M/2 + 1)+(5) ions in the bar graph spectra of PAHs are a t least partially artifacts. The purpose of this communication is to present additional information concerning the sources of the abundant (M + 1)+ and unusual (M/2 + 1)+ions in ion trap spectra. We sought to determine whether self-CI is a factor or if all the unusual ions can be attributed to the acquisition algorithm with broad and shifted peaks. We also considered whether saturation and space charging can account for all the broad peaks or whether other effects are involved, particularly with doubly charged PAH molecular ions.
+
EXPERIMENTAL SECTION Chemicals used in this study were the highest purity materials available from commercial sources and were used without purification. The gas chromatograph (GC) was a Carlo Erba Model 5160 and the Model 700 ion trap detector was purchased from the Finnigan Corp. Additional details about the ion trap and a description of modifications to the system and the transfer line connections are presented elsewhere (5). The data system consisted of an IBM PC/XT interfaced to the internal electronics of the ion trap. All studies were conducted with version 2.0 of the ion trap software supplied by the Finnigan Corp. The ion trap was tuned by using the procedure and standard software supplied by the manufacturer. The analytical chromatographic column was a 30-m X 0.25mm-i.d. fused silica capillary column coated with cross-linked phenylmethylsilicone (Durabond-5, J & W Scientific, Rancho Cordova, CA). All gas chromatographic injections were accomplished with helium carrier gas and the splitless mode with a hold time of 60 s before venting the injector. Helium was obtained from the U.S.Bureau of Mines and was passed through a trap (R & D Separations, Inc., Rancho Cordova, CA, Model OT-3) recommended by the Finnigan Corp. Background water and air were extremely low in the ion trap and were always well within the specificationsgiven by the Finnigan Corp. The transfer line was maintained at 255 OC, the injector at 260 O C , and the column oven was cooled to below 40 "C during the injection. After the 60-5 splitless period, the column oven was heated rapidly to the initial program temperature of 80 OC, and temperature programming to 280 "C at 8 deg/min began. The mass range 40-510 amu was scanned with a 1-s cycle time. RESULTS AND DISCUSSION Figure 1 shows the ion trap mass spectrum from 5 ng of anthracene-d,,. The M+ is a t mass 188, and a slightly-be-
48
SPEC)
54 66 88 94 109
68
le
132 120 1 4
iw
$! 160
1
188
I
201
Flgure 2. Ion trap mass spectrum from 50 ng of anthracenad,,.
SW.
BWC . 142
Figure.3. Ion trap mass spectrum from 5 ng of hexachlorobenzene.
IWL
14
I
L
SW.
BXC , 142
10112e il I/,,
4e
"3)
s'e
lie
171
8 1
150
214
200
250 1%
25e
3Ce
Flgure 4. Ion trap mass spectrum from 50 ng of hexachlorobenzene.
+
low-normal-abundance (M 1)' ion due to natural I3C is a t mass 189. This spectrum also shows a normal M2+at mass 94. Figure 2 shows the ion trap spectrum from 50 ng of the same substance. The (M 1)' ion is the base peak but, because of the large signal at mass 189, most other ions, including the M2+, are relatively weak and cannot be easily observed in the normalized bar graph spectrum. In the 5-ng spectrum (Figure l),the ions at masses 160 and 189 have about the same relative abundances, and the corresponding ions at masses 160 and 190 have about the same relative abundances in the 50-ng spectrum (Figure 2). We conclude that the 50-ng spectrum does not contain evidence for (M + 2)' ions that should be produced from self-CI. Figure 3 shows the ion trap mass spectrum from 5 ng of hexachlorobenzene. The cluster of molecular ions starting a t mass 282 has relative abundances that are a reasonable approximation of a six chlorine isotope distribution pattern, with masses 284 and 286 being the most abundant in the cluster as expected. Figure 4 shows the ion trap mass spectrum from 50 ng of hexachlorobenzene. Mass 285 is now the most abundant ion in the cluster, and other clusters of ions have
+
Anal. Chern. 1987, 59, 2732-2734
2732
64 1 I
i
512
-
2%
-
A
9
e!
"
'
-."
-..-,-
, 1us
----
"
'
,
95 leu 110 Flgure 5. Ion peak profile from the M2+ ion of pyrene: A, 50 ng; B, 5 ng. 91-1
appeared above masses 106, 142, and 250. The concentration dependence of the ion trap spectra observed with the two non-hydrogen-containing compounds is typical of results with numerous other compounds. Clearly the 1mass unit shifts cannot be due to self41 or ordinary CI since no hydrogen was present in these experiments, except perhaps very low picogram to fentogram quantities of background water vapor. These observations support the explanation that the unusually abundant (M + 1)+ ions are caused by the data acquisition algorithm and broad-shifted ion peaks. The small clusters of ions above masses 106, 142, and 250 in Figure 4 probably also result from a faulty interpretation of broad ion peak profiles. If the source of the broad peaks is space charging and saturation, which is a reasonable explanation, it is also necessary to account for an apparent shift of the ion peak to higher mass. The ion trap scans by increasing the radio frequency (rf) amplitude to eject ions from the trap and into the detector. Since space charging would reduce the effective rf potential experienced by an ion, particularly during the long excursion orbits just before elimination (detection), it follows
that the broadened ion peak would be shifted toward higher mass. Broadened M2+ions were observed in the spectra of pyrene and other PAHs with both 5-10 and 50-ng quantities (5). Figure 5A shows the ion peak profile in the region of the M2+ from 50 ng of pyrene, and Figure 5B shows the M2+ profile from 5 ng of pyrene. One possible explanation for these ions at the lower concentration levels, where saturation and space charging are not present, is partial charge stripping of M2+. A sudden increase in mlz, that is lower z , may cause tailing of the ion peak toward higher mass. The phenomenon causing broad M2+peaks a t lower concentrations is clearly different than the space charging and saturation effects at higher concentrations. Recently, a new version of the ion trap controlling software, revision 3.0, was made available by the manufacturer. This software is intended to minimize space charging and saturation by adjusting the ionization time as a function of total ion current. This approach should minimize the occurrence of most of the artifacts discussed in this communication and will be the subject of future studies with the ion trap detector. Registry No. Anthracene-dlo, 1719-06-8;hexachlorobenzene, 118-74-1;pyrene, 129-00-0.
LITERATURE CITED (1) Olson, E. S.;Diehl, J. W. Anal. Chem. 1987, 59, 443-448. (2) Stafford, G. C.; Kelley, P. E.; Syks, J. E. P.; Reynolds, W. E.; Todd, J. F. J. Int. J. Maass Spectrom. Ion Processes 1884, 60, 85-98. (3) Ghaderi, S.; Kulkarni, P. S.; Ledford, E. B.; Wilkins, C. L.; Gross, M. L. Anal. Chem. 1981, 53, 428-437. (4) Campbell, C., prlvate cornmunlcation, Finnigan Corp., Dec 1986. (5) Eichelberger, J. W.; Budde, W. L. Biomed. Environ. Mass Spectrom. 1987, 14, 357-362.
James W. Eichelberger William L. Budde* U.S. Environmental Protection Agency Environmental Monitoring and Support Laboratory Cincinnati, Ohio 45268 Laurence E. Slivon Battelle Columbus Laboratories Columbus, Ohio 43201 RECEIVED for review April 28,1987. Accepted August 4,1987.
Direct Serum Injection with Micellar Liquid Chromatography: Chromatographic Behavior and Recovery of Cephalosporins Sir: For therapeutic drug monitoring or biopharmaceutical and pharmacokinetic studies, it is a matter of vital importance to develop a high-performance liquid chromatographic (HPLC) method for the determination of a drug and its metabolite(s) in serum or plasma. However, the methods so far employed require a tedious, time-consuming extraction procedure for a drug and/or a precipitation step of serum proteins. Recently, HPLC supports have been developed to allow the direct injection of the untreated serum or plasma samples. The supports are the protein-coated octadecylsilyl (ODS) silica column prepared by Yoshida et al. (1, 2 ) and the internal surface reversed-phase silica column by Hagestam and Pinkerton (3). On the other hand, Cline Love et al. (4-6)reported an HPLC method involving the direct injection of the untreated serum samples using a micellar solution of sodium dodecyl
sulfate (SDS) or Brij-35 as an eluent on a reversed-phase column (Le., CIS or CN column). Micellar liquid chromatography, which uses surfactant solutions a t concentrations above a critical micelle concentration as the mobile phase, has been intensively investigated since the first report by Armstrong (7-9).The unique selectivity of such system is due to the solute-micelle association in addition to solute-stationary phase interaction. Cline Love et al. (4-6) stated that the serum proteins were solubilized without their precipitation formation and that the drug bound to protein was displaced by the surfactant monomers and/or micelles, being released for partition to the stationary phase. When we applied the direct serum injection method to the determination of cephalosporins, we found that they were not always observed as single peaks, dependent on the composition of the mobile phase (especially, the eluent pH).
0003-2700/87/0359-2732$01.50/00 1987 American Chemical Society
