1160
Anal. Chem. 1085, 57, 1160-1163
in proportion to the concentration after 0.2 M of potassium chloride is added to the sample solution. However these sample solutions often contain a large amount of dissolved organic matter: therefore sensitivity should be corrected by the standard addition method. The detailed electrode reaction mechanism in this system and the oceanographicdata of iodine species will be mentioned in forthcoming reports. Registry No. Ag3SI, 12003-00-8; I-, 20461-54-5; iodate, 15454-31-6;water, 7732-18-5.
Table I. Results of Seawater Analysis
location
depth, m
M
W[total iodine], M
24' N, 128' E
0 50 150 320 1200 1800 0 0 0 0
1.1 1.2 0.25 0.10 0.14 0.14 1.0 0.85 0.51 1.3
3.9 4.1 4.0 4.1 4.6 4.5 4.0 4.0 3.9 4.1
lo'[ iodide],
30' N, 137' E 20' N, 1 3 7 O E 10' N, 137' E 0' N, 137' E
LITERATURE CITED (1) Truesdale, V. M. Mar. Chem. 1978, 6, 253-273. (2) Nakata, R.; Okazaki, S.; Hori, T.; Fujinaga, T. Anal. Chlm. Acta 1983, 149, 67-75. (3) Fuglnaga, T.; Kimoto, T. Talanta 1984, 31, 720-722. (4) Lyman, J.; Fleming, R. H. J . Mar. Res. 1940, 3 , 134-146. (5) Truesdale, V. W. Mar. Chem. 1978, 6 , 1-13.
centration range. The sensitivity of approximately 0.4-0.5 pA/pg of I- is given by these curves. Since total concentration of iodine species in seawater is ca. 50 pg/L (4.0 X lo-' M) sufficient sensitivity is given by 7.5 mL of sample solution for total analysis (in this case the increase of sensitivity with preconcentration is 4 to 5 times). Although the concentration of iodide in seawater varies significantly with location and depth, it is around 1 pg/L (8 X M) in the lowest case according to existing data (5). The iodide is detectable with 40 mL of sample solution even in such a case (this preconcentration gives a -20 times increase of sensitivity). Table I shows results of the seawater analysis. These values indicate good agreement with existing data (5) for total iodine and are more accurate and precise for iodide because values of iodide less than M are included within the error of iodate (or total iodine) analysis in the existing data. In addition, iodine species in freshwater samples such as lake water, river water, and rainwater can be determined similarly by applying an adequate amount of sample solution
'
Present address: Kimoto Electric Co., Ltd., 3-1, Funahashl-Cho, Tennoji-Ki, Osaka 543, Japan. *Present address: Department of Chemistry, Faculty of Science, Kyoto University, Kyoto 606, Japan.
Eiichiro Nakayama* Takashi Kimoto' Satoshi Okazaki2 Research Center for Instrumental Analysis Faculty of Science Kyoto University Kyoto 606, Japan RECEIVED for review October 26,1984. Resubmitted January 7,1985. Accepted January 23, 1985. The research was supported by a grant from the Ministry of Education, Culture and Science, Japan.
Abundances of Molecular Ion Species Desorbed by Fast Atom Bombardment: Observation of (M 2H)'. and (M 3H)'
+
Sir: Fast atom bombardment mass spectrometry (FAB) (2,2) has seen ever increasing use in the characterization of ionic, polar, and/or labile samples. In particular, many studies on compounds with molecular ion species of over mass 1000 ( 3 , 4 )and up to nearly 10 000 (5) have been reported. With the commercial availability of magnetic sector instruments capable of upper mass limits of at least 10000 daltons at full accelerating potential, the applications of FAB as an ionization technique for the study of biomolecules surely will continue to increase. Accurate mass measurement has been an important tool in the confirmation of molecular weight and empirical formulas in the study of electron ionized compounds up to mass 1000. Full mass scans at mass resolutions of 10000 to 30 000 can yield mass values with 1-2 ppm accuracies. Accurate masses can be determined by using FAB as well, by employing either calibrated full scan methods of more commonly peak matching. However, as one goes to masses greater than 1000 amu, for example, even a resolution of loo00 will mean a peak width or Am of 0.1 amu or greater. The number of possible elemental compositions contained within a 0.1 amu window increases dramatically with an increase in mass. High mass measurement accuracy (sub part per million) will be necessary so that the candidate formulas are restricted to a tractable number. 0003-2700/65/0357-1160$01.50/0
+
Another problem associated with accurate mass measurements of higher molecular weight compounds is that peaks are no longer singlets, but rather complex combinations of isotopic species. This has been theoretically demonstrated by Yergey et al. for the m / z 5776.6 ion of porcine insulin (6). A resolution of 5000000 is required to achieve complete separation. This is well beyond present instrumental capabilities. Any accurate mass determination becomes an "average mass" determination and loses some of its significance. In situations where there is a low abundance of ions produced, either because of very small sample amounts or because of poor ionization efficiency, accurate mass measurements requiring high resolution may be impossible. One recourse for validating elemental compositions may be the use of the isotopic patterns obtainable a t low resolution (7, 8). This strategy would be particularly useful for halogen or isotopically rich metal-containing species which often have rather unique isotopic patterns. For example, aglucovancomycin,C53H52N8Ol7Cl2,mol w t 1142, has a pattern distinctly different from that of the more than 1500 peptides containing the standard amino acids and having the same nominal mass. A peptide with a large number of cystine and methionine residues will be rich in sulfur, and its isotopic cluster should be distinguishable from those of other peptides. 0 1985 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 57, NO. 6, MAY 1985
1161
Table I. Comparison of Precision and Accuracy of E1 and FAB for Isotopic Ratios of ClsF3zN3+from Tris(perfluorohepty1)-s-triazine theoretical abundance
mass
%
866 867 868
80.99 17.27 1.74
E1
FAB %
abundance 80.61 17.34 2.05
S
abs error
80.54 17.68 1.78
0.86 0.49 0.54
0.36 0.07 0.31
0.63 av
0.25 av
The interferences commonly observed in the mass region of the (M H)+ are (M - 2H)+-,(M - H)’, M+., and background ions originating from the liquid matrix (9,lO). Contributions from background ions can be minimized by carefully selecting the appropriate liquid matrix and adjusting the level of the sample. However, the origin of (M - 2H)+-,(M - H)+, and M+. is not well understood. In this paper, we address the problem of the accuracy of isotopic abundance ratios obtainable for FAB-produced ions under conditions in which the problems mentioned above have been minimized. Therefore, isotopic patterns in the molecular ion region for several classes of compounds were determined in both the positive and negative ion mode. In the course of this evaluation, desorption of significant amounts of (M + 2H)+. apd (M 3H)+was found.
+
% abundance
RESULTS AND DISCUSSION To compare the ability of FAB and ET to provide correct abundances of isotopic cluster patterns, the C18F32N3+ fragment ions from tris(perfluorohepty1)-s-triazinewere investigated by using both methods. Strong signals were observed for each ionization mode. Ten acquisitions were made using each technique to obtain average values and standard deviations (see Table I). The precision of these measurements is better than 1% (relative standard deviation), and the FAB results compare favorably with those obtained by using EI. The results obtained for C1&’32N3+ desorbed using FAJ3 did not involve the use of a matrix. T o determine the accuracy of isotopic ratios by using FAB and a matrix, preformed ions from a salt, rubidium iodide, were selected. Chemical interaction between matrix and sample is limited to glycerol adduct formation. Rb21+ and Rb312+were studied in the positive ion mode and Rb12- and Rb213-in the negative ion mode (see Table 11). As can be seen, the accuracy of the determined ratios is also within the experimental precision of the measurement for these benchmark experiments. A comparison of experimental and theoretical isotopic ratio values for the organometallic complex [R~(2,2’-bpy)~(CO)(CH2C6H5)]+desorbed using FAB is shown in Figure 1. The average absolute error for all the masses (0.40%) is well within the experimental precision range which indicates random rather than systematic error. In this case the isotopic pattern
0.64 0.32 0.35
0.45 0.41 0.04
0.44 av
0.30 av
Table 11. Comparison of Theoretical and Experimental Isotopic Ratios of Rubidium Iodide Clusters Desorbed by FAB species
m/z
% abundance exptl theory
RbzI+
297 299 301
52.43 40.18 7.38
52.06 40.19 7.76
0.37 0.01 0.38 0.25 av
Rb&+
509 511 513 515
38.04 43.28 16.03 2.65
37.56 43.49 16.79 2.16
0.48 0.21 0.76 0.49 0.48 av
RbIy
339 341
72.74 27.26
72.15 27.85
0.59 0.59 0.59 av
Rb&
551 553 555
52.55 39.94 7.51
52.06 40.19 7.76
0.49 0.25 0.25 0.33 av
+
EXPERIMENTAL SECTION The E1 data were obtained with a Kratos MS50 mass spectrometer and the FAB data by using a Kratos MS60-TA instrument. All data were acquired into a Nicolet 1170 signal averager taking 256 1-s scans over a mass range of 10-15 amu. Resolution was maintained at approximately 3000. The FAB ion source was of the standard Kratos design utilizing an Ion Tech atom gun. The samples were dissolved in a glycerol matrix, and the solution was bombarded with 8-keV xenon atoms. An accelerating voltage of 8 keV was used. The mass region of interest was checked to make certain it was free from background matrix interference before each sample run. Electron ionization was with 70-eV electrons and the ions were accelerated at 8 keV. Samples of the nucleosides were purchased from Sigma Chemical. All others were purchased from Aldrich Chemical. The [R~(2,2’-bpy)~(CO) (CH2C,H,)]+(PF6)-complex was a generous gift from B. P. Sullivan and T. J. Meyer.
abs error
S
error
THEORY
H EXP.
20
-
x W
0
z < n z
2
m
