Structural Analysis of Dinitriles by Chemical Ionization Mass Spectrometry Philip Price, H. S. Swofford, Jr.,' and S. E. Buttrill, Jr. Department of Chemistry, University of Minnesota, Minneapolis, Minn. 55455
For a series of dicyanoalkanes ranging from dicyanomethane to 1,6-dicyanohexane, the relative intensities of the MH+(H20), ions observed in chemical ionization mass spectrometry with water as a reagent gas at 110 O C and 0.22 Torr H20 change smoothly with increasing numbers of methylene groups between the cyano groups. This fact is used to determine the number of carbon atoms between cyano groups in other dinitriles. The water CI spectra of ortho-, mefa-, and para-dicyanobenzene follow the same qualitative pattern as the dinitriles and can be used to identify these three isomers.
Chemical ionization mass spectrometry (CIMS) has been shown to be useful for the analysis of organic compounds ( 1 - 4 ) . CI spectra result from reactions between sample molecules and reagent gas ions. The most commonly used reagent gases are methane (3) and isobutane ( 5 ) . However, water also has many advantages as a reagent gas. It is readily available in a highly pure form, easy to handle, and inexpensive. Furthermore, dilute aqueous solutions of organic compounds can be directly injected into a CI mass spectrometer with no prior sample preparation (6). The principal reagent gas ions formed in water CI are H+(H20), where n = 0-6 (7, 8). The formation of these water cluster ions, and the solvation of other ions, has been studied extensively by Kebarle (9-11). Water-CIMS spectra often demonstrate a similar type of clustering. In addition to the formation of MH+, peaks corresponding to MH+(H20), may be observed (where n = 1-3). The formation of proton-bound dimers of sample molecules and other higher clusters has been observed by other workers (12, 13). Often this may be prevented by maintaining the solute concentration below several thousand parts per million. We observe that difunctional compounds tend to accentuate the formation of MH+(HzO), clusters. In a previous publication ( 6 ) , it was noted that variations in cluster ion peak intensities might yield structural and conformational information. An important property of difunctional compounds is the distance between functional groups. A series of difunctional straight chain aliphatic nitriles was used as a simple model to study NC(CH2), CN where n = 1-6. In our investigations, nitriles were chosen because of their aprotic nature which would tend to minimize any complications due to hydrogen bonding. In addition, these compounds are highly soluble in water, have reasonable volatility, and give intense water CIMS spectra. Changes in clustering patterns were investigated as a function of the distance between nitrile groups. Clustering was also studied for the three isomers of dicyanobenzene. T o test this model for predictive validity, the spectra of four non-straight chain aliphatic dinitriles were determined and compared to those of the six model compounds.
EXPERIMENTAL Reagents. All samples were purchased commercially, and were used as received, Necessary purity was confirmed by electron impact and CIMS. T h e water used for t h e reagent gas was purified by 494
ANALYTICAL CHEMISTRY, VOL. 48, NO. 3, MARCH 1976
passing through three mixed-bed resin ion exchange columns, and one charcoal column (Pierce Chemical Company). T h e final conductance of the water used was 2 X ohm-' cm-' or less. Apparatus. A D u P o n t 21-490B mass spectrometer modified in house for chemical ionization was used for this study (6,14) Conditions. All experiments were preformed a t a source and inlet temperature of 110 f 5 "C. This temperature was high enough to adequately volatilize t h e samples and prevent background problems. Higher temperatures decrease the formation of cluster ions. Figure 1 shows the variations in water reagent gas cluster ion intensities with source pressure, as observed with our instrument. T h e source pressure must be above approximately 0.2 Torr before clusters containing three or four water molecules form an appreciable fraction of the total ion current (TIC). Using the thermochemical data of Kebarle (11), one can calculate t h e equilibrium distribution of water cluster ions expected under our conditions. T h e relative intensities of mle 73 and 5 5 , shown in Figure 1, are in fair agreement with the equilibrium values, but the relative intensities of m/e 19 and 37 are much too large. We believe t h a t these lighter ions represent a fraction of t h e ion population with source residence times too short to allow the achievement of equilibrium ( 1 5 ) Thus, the relative ion intensities detected are quite different from the equilibrium values, even though a fraction of the ions in the source may have been completely thermolyzed. A similar calculation using d a t a taken by Field (7) with his equipment operated in the continuous mode shows behavior in closer agreement with this work. Table I shows a comparison of all three sets of data. Figure 2 is a comparison between the water reagent gas T I C and sample T I C as the source pressure is increased. T h e maximum sample sensitivity occurs below 0.1 Torr. However, most of the signal a t this pressure is due to M H + ions and not clusters. Since the sensitivity falls off with increasing pressure, it was advantageous t o work a t the lowest source pressures still conducive t o clustering. For this reason, all of the studies were performed a t source pressures of 0.22 t o 0.23 Torr to eliminate this source of cluster variation. Clustering between sample and water reagent gas was also studied as a function of sample concentration (sample-to-reagent gas ratio). Until the sample concentrations reach approximately 1000 t o 3000 parts per million, little variation in clustering is observed. All spectra determined by injection were, therefore, run on 100 t o 200 ppm aqueous solutions. Procedure. With the present apparatus, CIMS spectra of compounds having sufficient solubility in water may be taken by directly injecting solutions into the heated batch inlet (6). Samples of low volatility or solubility may be placed on t h e solids probe and inserted directly into the source. Reagent gas is supplied a t sufficient pressure by injecting 10 to 30 ~1 of water into t h e batch inlet. Dicyanomethane, 1,2-dicyanoethane, l,3-dicyanopropane, 1,4dicyanobutane, l,j-dicyanopentane, and 1,6-dicyanohexane were prepared as 100-200 ppm aqueous solutions. Aliquots of 27 ~1 of these solutions were injected into the batch inlet to obtain the spectra of these dicyanoalkanes. Because of their low solubility in water, it was necessary t o run the series of three dicyanobenzene isomers on t h e solids probe. As a check t o ascertain t h a t the clustering spectra would not markedly differ from those run on injected solutions, several of the dicyanoalkanes were also run on t h e solids probe. So long as the probe was not heated (Le., the sample pressure was low), spectra comparable t o those previously run by injection were observed. I t was necessary to heat the probe slightly (40 t o 50 "C) to obtain good spectra of t h e dicyanobenzenes. Caution must be exercised when running samples on the solids probe. If their pressure becomes too high, clustering of the form MH'M may occur. Further increases of sample pressure may also lead to extensive and complex sample-sample ion molecule reactions.
70-
60 -
1
5001
2 50-
\
.4 0 -
301 20
-
10-
0
0"
02 T3QR
03
04
Figure 1. Ions observed in water reagent gas vs. source pressure
Curve i represents H'(H20); curve II, H'(H20)2; curve 111, H+(H20)& and curve IV, H'(H20h
I 4
03
32
31
0
TOQR
Figure 2. Sample and reagent gas intensity vs. source pressure for dicyanomethane
Table I. Comparison of Water Cluster Ion Intensities 76 T I C Cluster t y p e
This work
who)+
14 15 36 38
H(H20)*+ H(H*O),+ H(H,O), +
Field
173
Kebarle
3 9
0 0
39 51
38 61
1137
m 5c
s Finally, four additional compounds were run for comparison with the six dicyanoalkanes previously run. Trans- 1,2-dicyanoethylene, 2,4-dicyanobutene-l, and 1,4-dicyano-trans-butene-2 were run as 100-200 p p m aqueous solutions injected into t h e batch inlet. It was necessary to run 1,4-dicyano-2,2,3,3-tetramethylethane on the solid probe, because of its low solubility in water.
Figure 3. Typical water chemical ionization mass spectrum of 1,4dicyanoethane obtained at 110 OC and 0.22 Torr
RESULTS AND DISCUSSION The total ion current curves shown in Figure 2 decrease rather markedly as the pressure increases. This effect is probably caused by inadequate penetration of the electron beam into the source a t these pressures. Further modifications to the electron beam production system are currently being made to remedy this situation. Figure 3 is a typical spectrum of one of the six aliphatic dinitriles studied (1,2-dicyanoethane, molecular weight 80). The first four peaks in the spectrum, 19, 37, 55, and 73, are water reagent gas cluster ions. The next series of peaks are characteristic of the sample. Molecular weight 1, MH+ is given by the peak a t 81. The peaks a t 99, 117, and 135 represent successive sample-water cluster ions of the form MH+(H20), where n = 1-3. Information relating t o the sample structure can be determined from the relative intensities of these four sample peaks. The spectra of the six model compounds are plotted in Figure 4 as per cent of total sample ion current, and the points for each type of cluster are connected by solid lines. Thus, each solid curve represents the relative intensity of a particular type of sample-water cluster as the sample is varied from dicyanomethane to 1,6-dicyanohexane. The curves reveal that, based solely on relative clustering intensities, one can differentiate between various configurations of dinitriles. As the distance (number of methylene groups) between cyano groups is increased, 1) the molecular ion (MH+) falls rapidly to a steady value, 2) the MH+(H20) ion increases, 3 ) the MH+(H20)2 ion passes through a maximum a t a separation of 4 methylenes, and 4) the MH+(H20)3ion passes through a maximum a t a separation of 2 methylenes and then decreases to zero a t five methylenes.
+
1
2
3
5
4
METHYLENES BETWEEh NI'RILE
6
GROUPS
Figure 4. Cluster ion intensities vs. separation between, nitrile groups for model compounds Solid curves represent the various cluster ions for the dicyanomethane to 1.6-dicyanohexane
six
model compounds
As a test of the validity of this method for determining the distance between cyano groups, the spectra of four compounds different from the six model compounds were determined. Trans- 1,2-dicyanoethylene (I) has a separation of two carbon atoms between nitriles; 2,4-dicyanobutene-1 (11) has a separation of three; 1,4-dicyano-trun.sbutene-2 (111) has a separation of four; and 1,4-dicyano2,2,3,3-tetramethylethane(IV) has a separation of two. ANALYTICAL CHEMISTRY, VOL. 48, NO. 3, MARCH 1976
495
Table 11. Comparison of Test Compounds to Model Cluster ion, 70 TIC
Compound
MH
NCCH:CHCN
I NCC( :CH,)CH,CH,CN
I1 NCCH,CH:CHCH,CN
I11 NCC(Me, )C(Me, )CN
IV
Pred. Obsd. Pred. Obsd. Pred. Obsd. Pred. Obsd.
12 15 10
9 6 6 12 5
MH(H,O)
9 23 25
29 25 21 9 15
MH(H,O),
52 39
61 60 65 66 62 72
intramolecular hydrogen bond. If such structures were formed in the protonated dicyanoalkanes, one would expect a sharp reduction in the intensities of MH+(HzO), ions since all of the hydrogen bonding sites on the ion would be involved in forming the intramolecular hydrogen bond. Thus, with the possible exception of CHz(CN)2, the data in Figure 4 rule out the formation of these proton bridges. However, the data suggest that cyclic species may be formed in which the two cyano groups are connected by either an H30+ ( n = 5 , 6) or H502+ (n = 2-4) ion. We are in the process of modifying our instrument to enable us to make accurate measurements of AHoand A S o for the reaction: MH+(H20), iSOMER
Figure 5. Cluster ion intensities for the three isomers of dicyanoben-
zene
Table I1 compares predicted vs. observed values of percent TIC for each cluster ion of the four test compounds. The predicted values are read from the curves in Figure 4, assuming that the presence of a double bond decreases the distance between cyano groups by 0.2 A. There is generally good agreement between predicted and observed values, even though certain values (e.g., MH+(H20) and MH+(H20)2 for compound I) do differ. In determining the distance between cyano groups on an unknown compound, it would be necessary to find a t what point along the four curves of the model, the cluster ion intensities match those of the unknown. Reading down, one would then have the separation value between groups. In practice, for accurate determinations, it would be necessary to check the unknown vs. a set of standards run on the same instrument under identical conditions. Figure 5 shows the relative cluster ion intensities for ortho-, meta-, and para-dicyanobenzene. One can clearly differentiate between these three isomers, even though they have the same molecular weight, 128. I t is interesting to note that, qualitatively, this Figure is similar to Figure 4. As the cyano groups are separated on the benzene ring, the molecular ion (MH+) is fairly constant; the MH+(H20) ion increases; the MH+(H20)2 ion decreases, and the MH+(H20)3ion falls to nearly zero. Under approximately the same conditions, the water CI spectra of cyanoalkanes are qualitatively different from those of the dinitrile compounds discussed above. With the exception of acetonitrile, the cyanoalkanes show few clusters of the form MH+(H20), where n = 2 or 3 ( 1 6 ) . Other workers have shown that both diamino- (17) and dimethoxy- (12) alkanes, when protonated, form cyclic species in which the two functional groups are joined by an 496
ANALYTICAL CHEMISTRY, VOL. 48, NO. 3, MARCH 1976
+ H20 z MH+(H20),+1
AGon,n+l
using the procedures established by Kebarle ( 1 1 ) . These measurements will determine whether the MH+(HzO) and MH+(H20)2 ions have cyclic or open structures and may well provide a more accurate "ruler" for determining the distance between functional groups on a molecule. The fragmentation observed in CI spectra gives valuable structural information about the sample. This study demonstrates that clustering between protonated sample molecules and neutral reagent gas molecules provides a second source of structural information from CI spectra. The interactions between other functional groups and various reagent gases are under study. LITERATURE C I T E D F. H. Field, M. S. B. Munson, and D. A. Becker in "Ion Molecule Reactions in the Gas Phase", P. Ausloos. Ed., American Chemical Society, Washington, D.C., 1966, pp 167-192. F. H. Field, Acc. Chem. Res., 1, 42 (1968). M. S. B. Munson and F. H. Field, J. Am. Chem. SOC.,88, 2621 (1966). H. M. Fales, G. W. A. Milne, and T. Axenrod, Anal. Chem., 42, 1432 (1970). F. H. Field, J. Am. Chem. SOC., 91, 2827 (1969). P. Price, D. P. Martinsen, R. A. Upham, H. S. Swofford, Jr., and S. E. Buttrill, Jr., Anal. Chem., 47, 190 (1975). D. Beggs and F. H. Field, J. Am. Chem. SOC., 93, 1567 (1971). D. F. Hunt, C. N. McEwen, and R. A. Upham, Anal Chem., 44, 1292 (1972). P. Kebarle, in "Mass Spectrometry in Inorganic Chemistry", J. Margrave, Ed.. American Chemical Society, Washington, D.C., 1968, pp 24-47. P. Kebarle, in "Ion-Molecule Reactions", J. Franklin, Ed.. Plenum Press, New York, 1972, Vol. 1, pp 315-362. P. Kebarle, in "Modern Aspects of Electrochemistry", G. Conway and J. O'M. Bockris, Ed., Plenum Press, New York, 1974, Vol. 9, pp 1-46. T. A. Morton and J. L. Beauchamp, J. Am. Chem. SOC.,94, 3671 ( 1972). P. A. Leclercq and D. M. Desiderio, Org. Mass Spectrom., 7, 515 (1973). I. C. Wang, H. S. Swofford, Jr., P. C. Price, D. P. Martinsen, and S. E. Buttrill. Jr., Anal. Chem., 48, 491 (1976). C. Chang, G. J. Srokia, and G. G. Meisels, lnt. J. Mass Spectrom. /on Phys., 11, 367 (1973). P. Price and H. S. Swofford, Jr., unpublished data. R . Yamadagni and P. Kebarle, J. Am. Chem. SOC., 95, 3504 (1973).
RECEIVEDfor review July 28, 1975. Accepted December 11, 1975. This work was supported by NSF Grant GP-38764X.