ores, the Compton scatter method provides matrix compensation within + 2 0 x for silver concentrations greater than 10 oz/ton. The slope of the Ag Ka-to-Compton scatter ratio curves was equal for the both sets of silver ores. The conclusion is reached that silver can be determined in ores with a single calibration curve using the Compton scatter method. The internal scatter technique could not accurately be evaluated in this work for compensation of heavier matrix ores because of uncertainties of 15 to 20 % in the intensity measurements. More quantitative work at higher concentration is needed to determine the accuracy of the Compton scatter method for heavier matrix ores. However, it was found that serious under-compensation did not occur for the silver ores as others have reported for heavier matrix ores. In my opinion the under-compensation was absent because measurements of discrete Compton intensities separated from the coherent scatter could be made. Laboratory results are favorable for nondiffractive X-ray analysis of silver ores. Other high-Z ores could similarly be excited with radioactive sources and a t least two other electron-capture isotopes are available with high energy X-rays near those from I 2 5 I . Tin ores could be as efficiently excited with the Cs K a X-rays from 133Baas Te K a excites silver. The Sn K a X-rays would have about equal energy separation
from the Compton peak as Ag K a excited by 1251.Cadmium109 emits Ag K a X-rays and could be used for Sr, Y , Zr, Nb, and M o in ores with results equal to that of silver analysis in light matrix ores provided strong X-ray interferences d o not occur and matrix elements are not present which have absorption edges between the desired X-ray and the Compton peak energy. I n choosing the excitation energy particularly for trace analysis, one should select E, sufficiently high that the desired X-ray is well separated from the Compton peak to improve the peak-to-background ratio. ACKNOWLEDGMENT
I a m indebted to C. Mastromonico and W. J. Campbell of the College Park Metallurgy Research Center for their assistance in the computer processing of the data and technical help in preparing this publication, respectively. G. Potter of the Salt Lake City Research Center supplied the variable matrix silver ores. The Westcliffe District ore samples were obtained by C. Roach of the Denver Mining Research Center. RECEIVED for review July 7, 1970. Accepted October 5 , 1970. Reference to specific models of equipment is made for identification only and does not imply endorsement by the Bureau of Mines.
Application of Field Ionization to Gas-Liquid Chromatography-Mass Spectrometry (GLC-MS) Studies J. N. Damico and R . P. Barron Dicision of Food Chemistry and Technology, Food and Drug Administration, U . S. Department of Health, Education, and Welfare, Washington, D . C. 20204
A GLC-MS field ionization system incorporating a permeable membrane interface has been used to obtain spectra of GLC effluents consisting of components that yield relatively weak molecular ions by electron impact. These field ionization spectra complement the electron impact spectra. Field ionization sensitivity studies of the molecular ion (mV/pg) for a variety of carbonyl compounds with relative intensities by electron impact between 0.1 and 6.6% suggest that some types of compounds are significantly more sensitive to field ionization than others. The percentage of sample which permeates the membrane ranged between 30 and 72% for these carbonyl compounds. COMBINED GLC-ELECTRON impact mass spectrometry is widely used for the analysis of G L C effluents, as evidenced by the excellent review articles and the references therein (I,2). One of the most important types of information that can be obtained from mass spectral data is the molecular weight of a compound. In electron impact ionization the relative intensity of the molecular ion is very small for many compounds and comprises a small percentage of the total ion yield. Since components of interest in natural products are often present in the parts per million range or lower, the total ion yield will be small and the molecular ion may be of low in-
(1) W. s. Updegrove and P. Haug, Amer. Lab., pp 8-30 (Feb. (1970). (2) F. E. Saalfeld, Znd. Res., 11, 58 (1969).
tensity and difficult to distinguish from column bleed and/or background. Beckey (3) was the first to show that field ionization (FI) is effective in enhancing the relative intensity of molecular ions. Other examples and additional information on FI are provided in Table I and in the publications (4-9) and references therein which resulted from a symposium o n FI mass spectrometry (10). Field ionization is not so convenient a technique as electron impact ionization. First, most FI anodes must be activated before they attain maximum sensitivity, although Chait (11) has reported that commercially available chromium-plated razor blades do not require activation to reach full sensitivity. Second, it is not known a priori whether activation of a given (3) H. D. Beckey, in “Mass Spectrometry,” R. I. Reed, Ed., Academic Press, London and New York, 1965, pp 93-127. (4) H. D. Beckey, Int. J. Muss Spectrom. Ion Phys., 2 , 101 (1969). (5) D. F. Barofsky and E. W. Muller, ibid., p 125. (6) E. M. Chait, W. 0. Perry, G. E. Van Lear, and F. W. McLafferty, ibid., p 141. (7) M. Barber, R. M. Elliott, and T. R. Kemp, ibid., p 157. (8) J. N. Damico, R. P. Barron. and J. A. Sphon, ibid., p 161. (9) P. Schulze, B. R. Simoneit, and A. L. Burlingame, ibid., p 183. (10) Sixteenth Annual Conference on Mass Spectrometry and Allied Topics, ASTM Committee E-14, Pittsburgh, Pa., May 12-17, 1968. Papers numbered 9-12, 21-24. (11) E. M. Chait and F. Kitson, Org. Muss Spectrom., 3, 533 ( 1970).
ANALYTICAL CHEMISTRY, VOL. 43, NO. 1, JANUARY 1971
17
CARRIER G A S VfNT
LINE OF SIGHT INLET EMITTER W!RE /
\
,3000/3000v
r 1I
SULATING PIPE
'I
4 ,
Q
195v
-450/250 v EXCHANGEABL
Figure 1. Detailed cut-away drawing of EFO-4B source (ion optic potentials underlined are for E1 mode), pressure measuring source (GLC)detector, and GLC interface used in FI studies (not drawn to scale) anode will ensure sensitivity; in addition, the attainable sensitivity varies for different anodes. Tou et a/. (12), using acetone as the reference material, encountered two Wollaston wires (2.5 k m in diameter) which had a field ionization sensitivity ratio of 70 after activation. I n their study of the surfaces of the two wires by electron microscopy, they presented some interesting conclusions to account for the striking sensitivity ratio observed. A third disadvantage of FI is its relatively low sensitivity compared to EI. To our knowledge, no work has been published in which FI was utilized for the analysis of G L C effluents. (Since this manuscript was prepared, we learned that Professor Beckey's group has carried out GLC-FI M S studies of multicomponent mixtures containing several hundred unresolved components.) The object of this report is to demonstrate that FI does have utility in combined GLC-mass spectrometry, despite the disadvantages cited. EXPERIMENTAL
All studies were made with a n Atlas CH-4B mass spectrometer equipped with a dual electron impact/field ionization source (EFO-4B) as supplied with the spectrometer. An exploded drawing of the GLC-MS system is shown in Figure 1. Instrumental Parameters. Source exit and collector slits were 0.1 m m and 0.3 mm, respectively. The gain of the electron multiplier was usually lo6. Wires of 2.5-pm diameter, obtained commercially, were used as anodes; they were activated as described previously (8) except that maximum sensitivity could usually be approached in only 4 t o 5 hours. Three such wire anodes ( A , B, and C) were used to obtain the data reported here; wire B was preconditioned with benzonitrile. The G L C detector was a pressure measuring source (PMS), a n electron impact total ion collector. The electron energy was 20 eV and the trap current was 100 p A . (The total ion monitor, widely used as the G L C detector for EI, is inadequate for FI because the total ion current is very small. F o r example, the maximum total ion current observed on the total ion monitor from FI of 2 pg of CloA lactone injected on the G L C column was about 10-14 A, whereas the maximum response for the same amount of sample using the PMS was about 10-lo A,) The column bath for the G L C was a BarberColman Model 5360 equipped with temperature programmer (12) J. C. Tou, L. B. Westover, and E. J. Sutton, Int. J . Mass Spectrom. Ion Phys., 3, 377 (1969). 18
and controller. The bath also contained a flame ionization detector (FID); the effluent could be split to the FID and t o the mass spectrometer. This made it possible to simultaneously monitor the effluent with the F I D and the PMS. The column was a glass coiled column 6 ft X in. o.d., packed with 10% diethylene glycol succinate o n G a s Chrom Z, 100/120 mesh. The G L C was interfaced to the mass spectrometer with a permeable membrane separator (Varian 5620). The membrane is a methyl silicone polymer layer 0.001-in. thick and has a surface area of 0.4 cm2exposed to the column effluent. [Black et a/. (13) investigated single-stage and double-stage membrane molecular separators in conjunction with columns 0.02 and 0.03 in. i.d. but found that only the single-stage system was satisfactory; severe band broadening was observed for the higher molecular weight compounds (maximum molecular weight ca. 250) which was minimized by temperature programming the separator.] Since transmission of the sample through the membrane is decreased at high temperatures, the temperature of the membrane should be maintained as low as possible. Green and Littlejohn (14) reported that the membrane could be operated a t 50 OC below the column temperature without loss of column efficiency, and this temperature was usually employed in this work. The connecting lines were maintained a t 200 O C . The entire interface assembly was mounted line-of-sight to the ionization chamber and was heated by insulated heating wire. The carrier gas vent (see Figure 1) for the effluent, which does not permeate the membrane, was sometimes monitored with a FID. Procedure. The sensitivity determinations listed in Table I were usually obtained by injecting 30-50 pg on the G L C column. When the component enters the ionization chamber as detected by the PMS, the maximum response of the molecular ion is determined by scanning the mass region of interest at the top of the G L C peak. Although it is sometimes feasible to obtain both electron impact and field ionization spectra on the same sample charge with a dual source, this is not possible when obtaining spectra of G L C peaks because switching between modes of operation requires at least several minutes. Accordingly, the spectra were recorded on separate GC runs. The recorder response of the molecular ion is converted from chart divisions to millivolts from prior calibration data of the recorder. The column temperatures were selected so that the base-line peak widths were between 1'/z and 2 min. All the sensitivity data listed in Table I were obtained with wire anode C. Prior to each sensitivity determination, the anode was reactivated until the collector current (Farady Cage) for mie 58 of acetone was lo-" A at a n acetone pressure of 5 X 10-6 m m of H g as measured on the penning gauge of the analyzer tube. [Reactivation is necessary because sample and/or column bleed condenses on the surface of the wire and renders it insensitive. This problem might be alleviated if the wire were heated, preferably by current flow. Beckey et al. (15) have designed a vacuum lock whereby a wire anode mounted on a sliding rod can be readily interchanged, This design also makes it easy t o heat the anode by passing a few milliamperes of current through the wire, a procedure that is not feasible with the system described in this report except by extensive modification.] The transmission of sample through the membrane was determined indirectly. The peak area response was obtained for a given sample injection with the column connected only to the FID in the column bath (CFID),and again with the column (13) D. R. Black, R. A. Flath, and R. Teranishi, J. Chromatogr. Sci., 7, 284 (1969). (14) D. E. Green and D. P. Littlejohn, Pacific Conference on Chemistry and Applied Spectroscopy, Anaheim, Calif., Oct. 6-10, 199. (15) H. D. Beckey, A. Heindrichs, and H. V. Winkler, Int. J . Mass Spectrom. Ion Phys., 3, App. 9-11, (1970).
ANALYTICAL CHEMISTRY, VOL. 43, NO. 1, JANUARY 1971
Table I. Field Ionization Sensitivity for the Molecular Ion of Some Carbonyl Compounds Relative intensity Amount Maximum molecular of molecular ion, injected, ion response, FI E1 Compound I.Lg mV/I.Lg 100 6.6 C8-A Lactone 50 45 100 0.3 30 100 C10-y Lactone 100 3.5 30 250 ClO-A Lactone 100 0 .l a 30 88 CIz-y Lactone 100 4.3 30 110 Cln-ALactone 100