674
Anal. Chem. 1986, 58,674-679
Analysis of Complex Mixtures of Ethoxylated Alcohols by Probe Distillation/ChemicaI Ionization Mass Spectrometry Patrick Rudewicz' and Burnaby Munson*
Department of Chemistry, University of Delaware, Newark, Delaware 19716
The distrlbutlon of oligomers in complex mixtures of ethoxylated alcohols are determined by probe distillatlon/chemical loniratlon mass spectrometry using NH, and a 1.1% NH, In CH4 mlxture as the reagent gases. With these reagent gases each oligomer in a mixture gives essentially a one-peak spectrum, the (M NH4)+adduct ion. The area percentages of the (M NH4)+ Ion profiles are used to determine fhe distribution of the ethoxylated alcohols in each mixture. Good agreement is shown between probe distillation/chemicaI loniration mass spectrometric and gas chromatographic analyses on a less complex mixture.
+
+
Ethoxylated alcohols are nonionic surfactants that are widely used as detergents, wetting agents, and emulsifiers. They are prepared by a reaction of ethylene oxide with aliphatic alcohols that results in a distribution of oligomers having different numbers of ethoxy groups in the product mixture. Since the physicochemical and end-use properties of these surfactants are closely affected by this distribution, the rapid and reproducible determination of this distribution is important. Several high-performance liquid chromatographic (HPLC) procedures for the separation and quantitation of ethylene oxide oligomers have been reported (1-3). A major drawback in the HPLC analysis is the lack of an adequate detector. Since ethoxylated alcohols have no significant absorption in the near UV, they must be derivatized before HPLC analysis with a UV detector (1,2). Underivatized ethoxylated alcohols have been analyzed by HPLC using a flame ionization detector (3), but this is not a convenient detector for HPLC and good reproducibility is difficult. Refractive index detectors are not suitable for quantitation of ethoxylated alcohols by HPLC because they are difficult to use with gradient elution ( 4 ) . Other chromatographic methods have been employed for the quantitation of ethoxylated alcohol mixtures (5, 6). A thin-layer chromatographic separation has been described using silica gel and an acetone/THF solvent system using pinakryptol for quantitation under UV light ( 5 ) . Gas chromatography has also been used for the quantitation of ethoxylated alcohols; however, analysis by GC is limited to oligomers having a low degree of ethoxylation (6). The analysis of surfactants, including ethoxylated alcohols, has been reviewed by Neubecker and Llenado (7). Another approach to the determination of the oligomer distribution in a mixture of nonionic surfactants is the distillation of the mixture directly into the source of a mass spectrometer. The separation of the components during the distillation is relatively poor, but the mass spectrometer separates the compounds according to mass. Probe distillations of high-boiling petroleum distillates using electron ionization have been reported (8). The electron ionization spectra of a mixture of high molecular weight polar compounds Present address: Smith, Kline and French Laboratories, 620 Allendale Rd., King of Prussia, PA 19406.
frequently contain potentially interfering ions and may not contain abundant ions corresponding to the molecular weights of the compounds. Chemical ionization (CI) mass spectra, however, with an appropriate choice of reagent ions, may contain only one or a few ions per compound; consequently, the analysis of a complex mixture becomes less difficult. Chemical ionization mass spectrometry using a probe for sample introduction has been used previously for the quantitative analysis of individual components in mixtures without prior chromatographic separation. Usually, an isotopically labeled analogue of the compound of interest or a reference compound of similar structure is added to the mixture prior to any extraction steps and mass spectrometric analysis. Probe introduction of mixtures in conjunction with chemical ionization mass spectrometric analysis has been used for the determination of methadone and its metabolites (9, l o ) , lidocaine and its metabolites ( I I ) , reaction products of an anticancer agent (12),and triglycerides (13).
EXPERIMENTAL SECTION For these experiments, a small sample (0.05 pL) of a mixture of ethoxylated alcohols was placed directly in the well of a heatable Pyrex probe constructed in our laboratory (14). The probe was then introduced into the source of a Du Pont 21-492B mass spectrometer and heated at either 20 OC/min or 30 "C/min to 350 "C. As the samples distilled from the probe, mass spectra were recorded in the mass range from 50 to 830 every 9.6 s using a Hewlett-Packard 21 MX computer and a Du Pont data system. The ion source employed for these experiments was a dual EI/CI source used in the CI mode (zero repeller voltage). The two reagent gases used in these experiments were anhydrous NH3 (Matheson, 99.99%) and a 1.11%NH3 in CH4 mixture (MG Scientific Gases, North Branch, NJ). The source pressure was measured with a MKS Baratron capacitance manometer (MKS Instruments, Burlington, MA). The reagent gas pressure used for these experiments was approximately 0.5 torr. The source temperature was changed from 130 to 250 "C at approximately 15 "C/min for each probe distillation. The electron energy was 75 eV, and the emission current was 250 PA. The accelerating voltage was approximately 1750 V. The NH4(NH3)+ion, m / z 35, was the most abundant ion in pure NH3 under these conditions, but significant amounts of NH4(NH3)2f,m/z 52, and NH4+,m/z 18, ions are also observed. The ratio of ionic abundances, NH4(NH&+/NH4+,decreases markedly with increasing temperature. With the mixture of 1% NH3in CH, as the reagent gas, approximately 85% of the reagent ions were NH4+ions, and CHS+and C2H,+each constituted about 5% of the reagent ions. The gas chromatographic data for the mixture of C14 and C16 ethoxylates were obtained from Alan Ullman, Procter and Gamble, Cincinnati, OH. These data were obtained by use of a packed 3% SP-2100 column in temperature-programmed experiments. The mixtures of ethoxylated isomeric Cll-Cl5 alcohols were obtained from Exxon Research and Engineering Co. RESULTS AND DISCUSSION Ethoxylated alcohols have the general formula CxH2x+l(OCH2CH2),0H. For each of the mixtures analyzed in these experiments, x , the number of carbon atoms, ranged from 11 to 15 and n, the number of ethoxy groups, ranged from 0 to 11. During a probe distillation with NH3 as the reagent gas each ethoxylated alcohol in the mixture reacts with the am-
0003-2700/86/0358-0674$0 1.50/0 0 1986 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 58, NO. 4, APRIL 1986
MW=162
(M+NH4)
+ g
1
675
2b
'
SCAN NUMBER
'0
20
30
40
50
BO
70
PROBETEMP. ( C )
22
42
73
116
148
184
215
Figure 2. (a) Total ion current (m/ z 50-830) for a probe distillation of the ii = 1 sample of ethoxylated alcohols. (b) C,, ethoxylates: reagent gas, 0.5 torr NH,; probe heating, 20 OC/min.
0
50
150
100
200
Figure 1. (A) NH, CI spectra of the mono- and diethoxylates of 1-butanol: pressure, 0.5 torr NH,; source temperature, 180 O C . (B) Electron Ionization mass spectra of the mono- and diethoxylates of
1-butanol. monium ion, NH,+, to form an adduct ion without fragmentation:
+
-
C,H2x+l(OCH&H2)nOH "4' CXH,,+,(OCH2CH2)0H.NH4+ (1) x = 11-15 n = 0-11 The formation of the adduct ions is probably a collisionally stabilized process at these high pressures, although no experiments were done to establish the third-order nature of the reactions. Proton transfer to form (M + H)+ ions might be expected, since the proton affinities of cyclic and acyclic polyethers have been determined to be larger than the proton affinity of ammonia (15,16); however, alkylammonium ions and HaO+ have been shown to form adduct ions with polyethers under high-pressure mass spectrometric conditions (17, 18). Figure 1A shows the NH3CI spectra of two model compounds, the mono- and diethoxylates of 1-butanol. Both of these compounds give essentially one-species spectra, predominantly the (M + NH4)+ adduct ion with a very low abundance peak corresponding to the adduct of the sample with the solvated ammonium ion, (M + NH4+.NH3). An analysis of a mixture of ethylene oxide oligomers of alcohols, therefore, would be simplified by the use of NH, as a reagent gas, since each component would give only one ion and no interferences would occur for different carbon numbers in the alcohol.
In contrast to these NH3 CI spectra, the electron ionization (EI) mass spectra of these compounds, shown in Figure lB, contain no ions indicative of their molecular weights. The E1 spectra of both compounds are dominated by low-mass ions at m / z 41,45 and 57 corresponding to C3H5+,C2H50+,and C4H9+.Clearly, the analysis of a mixture of ethoxylated alcohols would be very difficult using E1 spectra because of peak overlap. The NH3 and 1%NH3 in CHI CI spectra of C8Hl7(OC2H4)30Hwere also essentially one-species spectra, (M + NH4)+. Only small amounts of (M H)+ ions were observed (52%) and even lower levels of fragment ions. In the mixture analyses, temperature profiles of the (M + NH4)+ions from the lower oligomers showed essentially no high-temperature tailing; consequently, there is no evidence that the (M NH4)+ ions decompose by the loss of ethylene oxide units. The NH3 CI spectra of ethoxylated alcohols obtained in these experiments show much less fragmentation than those reported recently (19). Samples of ethoxylated alcohols, each having a different number-average degree of polymerization (designated as a), were analyzed. The three samples had ii values of 1,2, and 3, respectively, based on the method of synthesis. Figure 2a shows the total ion current ( m / z 50-830) for a probe distillation of the ii = l sample using NH3 as the reagent gas. The probe was introduced into the source between scans 9 and 10. The change in base line between the initial and final points results from a change in pressure. The mixture is too complex and the distillation to inefficient to show any separation of individual components or even of ethylene oxide oligomers. However, when the (M + NH4)+ions of a particular carbon series are plotted, the individual compounds (n = 0, 1,2,3,...) are partially resolved in time, as shown in Figure 2b for the CI3series, and easily separated according to mass. The ion currents of the unethoxylated C13alcohol, n = 0, are quite low
+
+
ANALYTICAL CHEMISTRY, VOL. 58, NO. 4, APRIL 1986
676
Table I. Analysis of Ethoxylated Alcohol Mixture" area
C13 Seriea
0
ii=1
%b A ii=3
1 2 3 4 5 6
I 8 9 10 11
26.1 25.0 19.5 13.0 1.5 4.6 2.4 1.3 0.34
29.0 24.3 19.6 13.6 8.2 5.0 2.5 1.2 0.47 0.11 0.015
29.5 (37.2)' 23.9 (25.5) 18.3 (16.9) 12.1 (9.9) 7.3 (5.3) 4.3 (2.9) 2.4 (1.4) 1.0 (0.55) 0.40 (0.21) 0.10 (0.05) 0.039 (0.02)
"P(NH3) = 0.5 torr, t = 130-200 O C , mixture with ii = 2. bArea % calculated from (M + NH,') ions for all oligomer species of each carbon number summed over distillation. Alcohol, n = 0, not included in analysis. 'Calculated from weighted areas, see text.
even though C13H270H is a major component, in agreement with previous reports on the low sensitivities for saturated alcohols with NH3 as the CI reagent gas (20,21). Plots very similar to those of Figure 2b were obtained for the CI1, C12, C14, and C15 ethoxylated alcohols in this mixture. There are no mass interferences among any of these species: each chemical species has a distinct mass. The (M+ NH4)+profiles can be integrated for each oligomer and the mixture composition calculated as area percent for each series of alcohols, n = 11,12,...,etc. The complexity of these samples is indicated by Table I, which lists the area percents of each component of another mixture, ii = 2. In the absence of unusual steric effects or very different structures of the homologous alcohols, the rate of addition of ethylene oxide to the alcohols to form the ethoxylated alcohols should be independent of the number of carbons in the alkyl chain over this range, and the distribution of oligomers for each series, Clz, C13,..., etc., should be similar for a particular reaction mixture. Indeed, Table I shows that the distributions of ethylene oxide oligomers are essentially the same in this sample for ali five of the alcohols. The differences noted in the table probably reflect the precision of the measurements. The absence of the higher oligomers for the Cll, CIE, and C14 alcohols reflects the lower concentrations of these species: the ion currents for (CllH23(OC2H4)110H + NH4+),etc., were too small to be detected. The ratios of oligomers of the homologous series can be estimated from ratios of total areas: C11/C13 = 0.06; C12/C13 = 0.50; C14/C13 = 0.20; C15/C13 = 0.05. The three types of samples give different ethoxylate distributions illustrated in Figure 3 for the major components, C13. These values are reported as area percents for the C13 species, excluding the alcohol, n = 0. For the ii = 2 and ii = 3 samples, compounds having 11 ethoxy groups were detected. With NH3 as a reagent gas, very little of the unethoxylated alcohols were detected in these mixtures. However, from GC data and from probe distillations using methane as a reagent gas it was known that between 20% and 30% of each mixture was unreacted alcohol. We attribute the low sensitivity of the ammonium ion for the alcohol, in part, to a switching reaction in which neutral NH3 displaces the ammonium ion from the alcohol adduct, forming neutral alcohol and the solvated ammonium ion (22) ROH NH4+ F? (ROH.NH4)' (24
+ (ROH-NH,)' + NH3
-*
ROH
I \
30.6 . 30.6 22.9 24.5 18.2 19.2 11.8 11.2 6.6 6.9 4.2 4.4 2.2 2.5 1.4 0.70 0.29 0.03
+ (NH3sNH4)'
(2b)
Consequently, experiments were done with a 1.1% NH3 in CH, reagent gas mixture. This mixture shows NH4+as the major reactant ion, but it has a much lower concentration of neutral ammonia available for the switching reaction. Am-
n
1
"3
2
3
4
5
6
7
8
9 1 0 1 1 1 2
Number of Ethoxy Groups
Figure 3. Distrlbutlons for C,, ethoxylates from ii = 1, 2, and 3 samples: reagent gas, 0.5 torr NH,; probe heating, 20 'C/min. e-a
n= 1
P '\
!\
SCAN NWER PROBE TEMP. ( Cl
10 45
""
20 87
30 140
40 194
Figure 4. (a) CI3 ethoxylates. (b) Homologous serles for CnH2n+1(OC2H,),0H for Cll-C,5: ii = 1 sample: reagent gas, 1.1 % NH,/CH,, 0.5 torr: probe heating, 30 'C/min. monialmethane reagent gas mixtures have been used previously in the analysis of compounds where pure NH3 as a reagent gas gave unsatisfactory results (23, 24). With 1.1%NH, in CHI as the reagent gas, much larger amounts of the unethoxylated alcohols were detected (as (M + NH,)+ ions) for each of the samples. Typical data are shown in Figure 4A for a PD/CIMS experiment. The curve for the alcohol, n = 0, with this reagent gas can be compared with a similar plot in Figure 2b using pure NH3 as the reagent gas: approximately 22% by area from the data of Figure 4a vs. 1.4%by area from the data of Figure 2b. The exact shapes of the probe distillation curves depend on the heating rate, the source temperature, and to some extent the sample size and total source pressure and not on the nature of the reagent
ANALYTICAL CHEMISTRY, VOL. 58, NO. 4, APRIL 1986
I
s
C 1 3 Series
u.
677
1
1.1% NH31 E 3 0 t I
I
I
0
0
1
2
3
I
l
4
A
4
5
8
7
Number of Ethoxy Groups
Flgure 5. Distribution of C,3 oligomers for ii = 3 mixture obtained from duplicate analyses done 6 months apart: reagent gas, 1.1% NH3/CH4, 0.5 torr.
40
i I
0 1.1% NH3in CHq A
NH3
C13 Series, ii=3
+
Flgure 7. Sample ionization as (M NH,)' ions for an equimolar mixture of 1-butanol and the mono- and diethoxylates of 1-butanol introduced from the Gc: reagent gas, 1.1% NH3/CH,, 0.3 torr; source temperature, 180 OC.
experiments in this laboratory show that the relative NH3 CI sensitivities of polyfunctional compounds like the ethoxylated alcohols do not change with NH3 concentration, but the sensitivities of simple alcohols or ethers decrease rapidly with increasing NH3 concentration (22). If the (M NH4)+ions are formed according to reaction 1 and little or no decomposition of these ions occurs, then
+
~ M + i s=
k t ~ m i
(3)
in which IM+18 is the ion current for the ammonium ions, (M NH,)+, of the sample, 118 is the ion current for the NH4+ ion, t is the ionic residence time for NH4+, [MI is the molar n O 1 2 3 4 5 6 7 8 9 1 0 concentration of the sample, and k is the rate constant for the ion/molecule reaction (25). Since t and Ils are fixed for Number of Ethoxy Groups any experiment, the area percentages of ion currents, lOOIi/ZZi, Figure 6. Comparison of distributions obtained for CI3 ethoxylates will equal mole percentages if the rate constants for the (excluding ROH) in the ii = 3 mixture using 1.1 % NH3 in CH, and NH, ion/molecule reactions are equal. Alternatively, ktIls is the as reagent gases. sensitivity or slope of the calibration curve of 1 M f L 8 vs. moles of M, and the sensitivities of the compounds are the same if gas. The shapes of the curves will, of course, be strongly dependent on the probe geometry, but all of these experiments the rate constants of the ion/molecule reactions are the same. were done with a single probe. Figure 4b shows the lack of No data are available on rate constants of any ion with any separation of the homologous series for CnH2n+l(OC2H4)30H of the ethoxylated alcohols, nor are the individual compounds for Cll-CI5. It is essential in these probe distillations that there available (or readily separable from this mixture). be no accidental mass interference from the different species A few lower molecular weight model compounds were and that the spectra of the higher homologues contain no available for study, and Figure 7 shows a GC/CIMS analysis fragment ions of the same mass as the characteristic (M + of an equimolar mixture of n-C,HgOH, n-C4HgOC2H40H,and X)+ ions used in the analysis, hence the need for a selective n-C4H9(OC2H4)20H, using 1.1% NH3 in CH, as the reagent CI reagent gas. gas. The ratios of areas in this experiment are 0.05/0.83/1.00 The use of areas to estimate compositions of very complex and the mole ratios are l/l/l. These experiments were done mixtures like these is a convenience because the individual with temperature programming on the oven of the gas chrocomponents are not available for calibration curves. Although matograph without a flow controller; consequently, the one cannot directly relate area percent to mole or weight pressure in the source of the mass spectrometer changed percent, one may certainly use the area percentages for difsomewhat during the course of the experiment. In addition, ferentiation among mixtures. Figure 5 shows an estimate of there is a jet separator between the gas chromatograph and the long-term precision of the analysis with two distribution the mass spectrometer that may cause differential losses of curves for the oligomers of the major component, Cl3HZ70H, the three components. The ratios of area, then, are not exactly using 1% NH3 in CHI as the reagent gas. The data plotted the ratios of rate constants of NH4+with the three species. in this figure represent duplicate analyses under nearly However, one can say with confidence that the rate constants identical experimental conditions taken 6 months apart. The or sensitivities for the two ethoxylated alcohols are similar marked difference between the distribution indicated in Figure and the rate constant or sensitivity for the alcohol is much 5 and that shown in Figure 3 for the ii = 1 mixture is the lower than the rate constants or sensitivities of the ethoxylated exclusion of the alcohol peak for the data obtained with pure alcohols. The switching reaction 2 mentioned previously to NH, as the reagent gas shown in Figure 3. explain the low sensitivities of alcohols with NH3 as the CI Both CI reagent gases, NH3 and 1.1%NH3 in CH,, give the reagent gas is indicated by the increase in NH4+.NH3as butyl same distribution of oligomers, as shown in Figure 6. The alcohol elutes through the source of the mass spectrometer. circles indicate duplicate experiments done on the same day Sensitivity studies in NH3 CIMS will be discussed in detail using 1.1%NH,, and the triangles represent an experiment in a subsequent paper. on the same mixture using "3, done 8 months previously. Rate constants for ion/molecule collisions can be calculated (The alcohol has been excluded from the calculations.) Other if the polarizabilities and dipole moments of the molecules
+
678
ANALYTICAL CHEMISTRY, VOL. 58, NO. 4, APRIL 1986
A = Probe
30 1
0 : FID
Q
1.1% NH3
Q A
0
a lot
I
a
Number of Ethoxy Groups
Figure 8. Comparison of distributions of C,4 ethoxyiates obtained by GCIFID PDICIMS (1.1% NHB/CH4).
are known (26). Values of the dipole moments of these ethoxylated alcohols, however, are not available, but one may estimate the effect of increasing molecular weight (increasing polarizability) on relative rate constants from the simple theory, which does not include dipole moment,
(4)
I
in which a is the polarizability of the molecule and p is the reduced mass of the ion/molecule pair. For large molecular weights, the reduced mass is a slowly varying function that approaches a limit of 18 for NH4+,at high molecular weight. Since the mean polarizability of a molecule, a, is equal to the sum of the individual bond polarizabilities, the polarizabilities of ethoxylated alcohol oligomers can be estimated from tables of individual bond polarizabilities (27). For C4H90H,CY is estimated as 87 X cm3, and the polarizability increases by 43 X cm3 for each ethylene oxide group. For the diand monoethoxylated butanols, the simple polarizability theory predicts an increase of approximately 13% in the collision rate constant and sensitivity, and the data from Figure 7 show an increase of 17%. As indicated by these observations, the rate constants for reaction of NH4+with the ethoxylated alcohols and, therefore, the molar sensitivities needed for the determination of the oligomeric distribution are not independent of, but increase significantly with, increasing molecular weight. Consequently, the mole percentage distributions calculated as area percentages will be erroneously high for the heavier compounds, even though they are useful for comparing different mixtures. For the higher molecular weight compounds (greater than about 300), the reduced mass, p , is essentially constant; consequently, the rate constant (or relative sensitivity) increases with d2. The polarizabilities of these compounds increase linearly with increasing molecule weight (27). Therefore, a somewhat better approximation to the actual mole fractions of the individual species would be calculated by (Ai/ MWi1I2)/ZAiMWi'I2. Table I shows a comparison of the distribution of the C13oligomers calculated by this equation in contrast to the distribution calculated as area percentage. These mixtures were too complex for a good gas chromatographic analysis, since the parent alcohols from which these mixtures were prepared (C13H2,0H, etc.) were themselves mixtures of isomers. Less complex mixtures can be analyzed by gas chromatography. Another sample consisting of oligomers of ethoxylated CI4and C16 alcohols was analyzed by probe distillation/chemical ionization mass spectrometry and by gas chromatography with a flame ionization detector (FID). The comparison of these two analyses is shown in Figure 8. The agreement is perhaps surprisingly good between the two techniques, calculated in both cases as uncorrected area percentages.
The agreement of the two techniques can be explained by similar increases in sensitivity per ethoxy group caused by increasing carbon number for the FID and by increasing Langevin rate constants for the ion/molecule reactions. Studies on the quantitative response of gas chromatographic flame ionization detectors have shown that for hydrocarbons the relative FID response increases linearly with increasing carbon numer (28). Ether oxygens cause a decrease in the molar FID response of molecules, which is the equivalent of reducing the number of carbon atoms by one (28). For ethoxylated alcohols, therefore, the effective carbon number increases by one with the addition of each ethoxy group, corresponds to about a 5% increase in relative FID response with the addition of each ethoxy unit for the C14ethoxylates. With only polarization theory, the increase in rate constant or CI sensitivity is about 4% /ethylene oxide unit. Recently, trimethylsilyl derivatives of formaldehyde oligomers in methanol-water solutions have been characterized by capillary gas chromatography/ammonia chemical ionization mass spectrometry (29). Quantitation of derivatized oligomers on formalin was achieved by calibrating the FID with a set of closely related compounds consisting of (CH3)3Siderivatives of simple alcohols and glycols. The molar FID response of these compounds was found to increase with increasing effective carbon number as discussed above. In conclusion, probe distillation/chemical ionization mass spectrometry is a useful technique for the analysis of complex mixtures of oligomers. When care is taken to ensure similar experimental conditions, both short- and long-term reproducibility are good. For the quantitation of ethoxylated alcohols, probe distillation data correlate well with gas chromatographic data. Although a CIMS system is much more expensive than a GC alone, probe distillation/CIMS has two advantages over gas chromatographic analysis. First, higher molecular weight ethoxylated alcohols that are not amenable to gas chromatographic analysis can be analyzed by using probe distillation. Secondly, probe distillation has the advantage of shorter analysis time than chromatographic analysis. The separation of the C14mixture shown in Figure 8 took 60 min, whereas the probe distillation experiment took only 12 min.
LITERATURE CITED (1) Nozawa, A,; Ohnuma, T. J. Chromatogr. 1980, 187, 261-270. (2) Allen, M. C.; Linder, D. E. J. Am. Oil. Chem. SOC. 1981, 58, 950-957. (3) McClure, J. D. J. Am. 011Chem. SOC.1982, 59, 364-373. (4) Snyder, L. R.; Kirkland, J. J. "Introduction to Modern Liquid Chromatography"; Wiley: New York, 1979. (5) Koehler, M.; Chalupka, B. Fette Seifen Anstrichm. 1982, 8 4 , 208; Chem. Abstr. 1982, 97, 2 5 5 3 0 ~ . (6) Nichikova, P. R.; Rud, A. N.; Getmanskaya, 2 . I.; Ivanov, V. N. Neffeoererab, Beffejgen (Moscow) 1982, 36, 5; Chem. Abstr. 1982, 97, 57505s. (7) Llenado, R. A,; Neubecker, T. A. Anal. Chem. 1983, 5 5 , 93R-102R. (8) Schronk, L. R.; Grigsby, R. D.; Scheppele, S. E. Anal. Chem. 1982, 54, 748-755. (9) Kreek, M. J.; Bencsath, A,; Field, F, H. Biomed. Mass Spectrom. 1980, 7,385-395. (10) Kreek, M. J.; Bencsath, F. A.; Fanizza, A,; Field, F. H. Biomed. Mass Spectrom. 1983, 70, 544-549. (11) Nelson, S. D.; Garland, W. A.; Breck, G. D.; Trager, W. F. J. fharm. Scl. 1977, 66, 1180. (12) Weinkam, R. J.; Huey-shin, L. Anal. Chem. 1979, 5 1 , 972-975. (13) Murata, T.; Takahashl, S. Anal. Chem. 1977, 49, 728-731. (14) Spreen, R. Ph.D. Thesis, University of Delaware, Newark, DE, 1983. (15) Meot-Ner (Mautner), M. J. Am. Chem. SOC. 1983, 105, 4906-4911. (16) Sharma, R. B.; Blades, A. T.; Kebarle, P. J. Am. Chem. SOc. 1984, 106, 510-516. (17) Meot-Ner (Mautner), M. J . Am. Chem. SOC.1983, 105, 4912-4915. 118) Sharrna. R. B.: Kebarle. P. J. Am. Chem. SOC. 1984, f U 6 , 39 13-39 16. (19) Stephanou, E. Org. Mass Spectrom. 1984, 19, 510-513. (20) Hunt, D. F. frog. Anal. Chem. 1973, 6 , 359-376. (21) Hunt, D. F. I n "Advances in Mass Spectrometry"; West, A. R., Ed.; Aoolied Science Publishers: London, 1974; VOl. 6. (22) R;bewlcz, P.; Munson, B. Presented at the 32nd Annual Conference on Mass Spectrometry and Allied Topics, San Antonio, TX, May 27June 1, 1984. ~I
Anal. Chem. 1986, 58,679-684 (23) Horning, E. C.; Stillwell, R. N.; Nowlin, J. G.; Carrol, D. 1. Anal. Chem. 1881, 53, 2007-2013. (24) Smith, D. E.; Smith, J. S.; Jerolamon, D.; Weston, A. F.; Richton, D.; Brozowski, E. J. Presented at the 26th Annual Conference on Mass Spectrometry and Allied Topics, St. Louis, MO, May 28-June 2, 1978. (25) Munson. 8. I n “Interactions Between Ions and Molecules”; NATO Advanced Studies Institute, Series B.; Ausloos, P., Ed.; Plenum: New York, 1975; pp 505-525. (26) Ceili, R.; Weddle, G.;Ridge, D. p. J . Chem. PhYS. 1980, 73, 801-812. (27) Denbigh, K. G. Trans. Faraday SOC.1940, 36, 936-947.
679
(28) Sternberg, J. C.; Gallaway, W. S.; Jones, D. T. L. I n “Gas Chromatography”; Brenner, N., Callen, J. E., Weiss, M. D., Eds.; Academic Press: New York, 1962; pp 231-267. (29) Utterback, D. F.; Millington, D. S.; Gold, A. Anal. Chem. 1984, 56, 470-473.
RECEIVED for review July 26, 1985. Accepted November 20, 1985. This work was supported in part by a grantfrom the National Science Foundation, CHE-8312954.
Electron, Chemical, and Thermal Ionization Mass Spectra of Alkali Halides F. Aladar Bencsath* and F. H. Field
The Rockefeller University, New York, New York 10021
We report the electron Ionization (E1 ), various chemical Ionlzations (CI), and thermal lonlzatlon spectra of alkali hallde salts obtained wlth an unmodlfled commercial mass spectrometer. The salts studied were CsI, CsBr, CsCI, CsF, RbI, K I , NaI, LII, LiF, and NaF. The posltlve E1 and C-I spectra consist of alkali ions and alkali Ion clusters. The negative Ion CI spectra conslst of hallde Ions and halide Ion clusters. Molecular Ions and quasi-molecular ions are observed wlth only small Intensities. The C I sensitlvlty Is about an order of magnitude higher than the E1 sensltlvlty. Useful E1 spectra are obtained from 0.5 pg of sample, and a tenth as much can be detected. The thermal lonlzatlon spectra consist solely of the alkali cations. Thermal ionization sensitivity was 5-6 orders of magnitude higher than that of EI.
Because of their low volatility, the metal salts of mineral acids are generally considered to be intractable for routine analytical work with commercial mass spectrometers oriented for the analysis of organic compounds. Measurements of salts are usually made with mass spectrometers incorporating specialized sources such as high-temperature Knudsen cell (1-3), radio frequency spark (4),field desorption (5), and thermal ionization (6). One finds, however, some mention in the literature of the adequacy of more or less conventional commercial mass spectrometers to cope with salts. Several years ago we pointed out (7) that a slightly modified commercial quadrupole mass spectrometer could be used to determine isotopic ratios in lithium. Hunt and co-workers (8) obtained excellent CI spectra with intense molecular ions from potassium benzoate, and White (9) obtained E1 spectra of alkali metal salts of several organic acids, the spectra containing the metal cation and clusters involving the metal cation. Rosenstock and co-workers (IO)using a noncommercial mass spectrometer found in 1955 that copper halide salts placed on an electrically heated platinum filament evaporated sufficiently in the mass spectrometer source to give rise to characteristic spectra. In 1977 Soltmann and co-workers (11) used an activated field emitter wire to volatilize inorganic alkali metal salts in a slightly modified field desorption source, and the resultant E1 spectra did not differ from those obtained by Knudsen cell mass spectrometry (12). Although their source and probe setup was not the kind used in general purpose mass spectrometers, they pointed out that their 0003-2700/86/0358-0679$01.50/0
method, based on the rapid sample volatilizationfrom a heated thin wire, might be used with commercial instruments. Volatilizing samples from a rapidly heated wire is the basis of the desorption chemical ionization (DCI) method (13), which as proved to be useful for the analysis of involatile and/or thermolabile organic compounds. Because of the development of this method in the last half decade, the DCI probe has become available as a standard or optional sample introduction device in commercial mass spectrometers designed for organic samples. We recently had occasion to attempt to obtain with our DCI probe and commercial mass spectrometer the mass spectrum of a biochemical sample that apparently had been badly contaminated with potassium iodide. We observed abundant K+ and I- ions in the positive and negative spectra, respectively. This interesting finding prompted us to undertake the systematic study reported here. The first series of measurements were made with our mass spectrometer focused to collect ions produced by electron ionization and chemical ionization from gaseous salt molecules evaporated from the DCI probe. We think it of interest and little recognized that spectra of alkali halide salts can be obtained from a completely unmodified contemporary commercial mass spectrometer. Furthermore, the spectra themselves are of interest in that little information seems to be available on the E1 spectra of the alkali halide species emitted from a hot wire, and no information at all is available about the positive and negative CI spectra of such species. In the course of making these measurements, we by chance discovered that our mass spectrometer could be refocused to collect ions formed on the DCI probe wire by thermal ionization. A study was then made using this mode of operation of the instrument.
EXPERIMENTAL SECTION Apparatus. Mass spectra were obtained with a VG 70-250 double-focusing magnetic instrument. Spectra collection and data manipulation were accomplished by the data system of the instrument. The VG DCI probe with programmable heater controller was used for sample introduction. Sample Loading and Running. Aqueous stock solutions (2 pg/pL) of CsF, CsC1, CsBr, CsI, RbI, KI, NaI, and LiI were prepared for the E1 and CI experiments. For thermal ionization experiments, another set of stock solutions (100 ng/pL) was prepared of those salts listed above and additionally of NaF and LiF. The salts were obtained from Matheson, Coleman and Bell, Norwood, OH; Mallinckrodt, Inc., St. Louis, MO; and Alfa Products, Beverly, MA, and they were all analytical reagent grade 0 1986 Amerlcan Chemical Society