be very useful in other type analyses. Accuracy and repeatability of the method are believed to be equivalent to or better than other published methods.
in Table V, were supplied by 31. R. Fenske, Pennsylvania State University. H i s contribution is gratefully acknowledged.
ACKNOWLEDGMENT
LITERATURE CITED
The by Physical separationsphysical property techniques, recorded
(1) Am. SOC.Testing Materials, “ASTM Standards on Petroleum Products and Lubricants," p. 1101, 1958.
(2) Brown, R. A., ANAL- CHEM- 23, 430 (1951) (3) Lum‘pkin, E., Thomss, B. Ibid., 23, 1738 (1951). (4) Lumpkin, H. E., Thomas, B. W., Elliott, A., Ibid,, 24, 1389 (1952).
w.,
(51 . , Purdv. K. M.. Harm, R. J., Ibid., 22, 13$7 (1950). ’ RECEIVEDfor review July 24, 1961. Accepted October 19, 1961. ASTM Committee E-14 on Mass Spectrometry, LOS Angeles, Calif., May 1959.
Correlation of Mass Spectra with Structure in Aromatic Oxygenated Compounds Benzoate-Type Esters THOMAS ACZEL and H. E. LUMPKIN Research and Development, Humble Oil and Refining Co., Boytown, Tex.
b A systematic study of the mass spectra of seven methyl benzoates and of nine benzyl benzoate-type esters is presented, and correlations between the spectral features and the structures of the compounds studied are discussed. These correlations are shown to b e useful for analytical applications. The tables presented include data derived from previous publications on the spectra of methyl-substituted aromatic alcohols, acids, aldehydes, and phenols. Several paths of ion formation are comparable among various types of oxygenated compounds. The major spectral features depend primarily on the nature of the oxygenated functional groups and the positions of the methyl groups substituted on the ring. The latter effects are similar in most oxygenated compounds considered.
T
establishment of correlations between mass spectra and chemical structure in a series of compounds has been shown to provide valuable information to analytical mass spectrometry. Studies of this type allow extrapolation of calibration data, a more rational approach to the identification of unknown components in the absence of comparison spectra, and some intuitive insight into the phenomenon of ionization under electron bombardment. In two previous papers (1, 2 ) we discussed the mass spectra of several methyl-substituted aromatic alcohols, acids, aldehydes, and phenols. I n this communication we examine the spectra of several benzyl benzoate-type esters and methyl benzoates. Furthermore, correlations are pointed out which can HE
be applied to all of the oxygenated aromatic compounds studied in this work. EXPERIMENTAL
The experimental conditions employed are described in some detail in our previous communication (1). I n brief, the data reported were obtained on Consolidated Electrodynamics Corp. Models 21-102 and 21-103C mass spectrometers. Ion abundances are expressed in per cent of total ionization. Because some of the spectra n-ere obtained only in the region between m/e 73 and the parent mass and its isotopes, the total ionization could be calculated only as the summation of the peak intensities, based on standard volumes, from m/e 73 to m/e (parent 2) ( 1 ) . The tables include only the spectral features which were deemed to be significant for analytical applications or for our discussion. The complete data are available in the ASTM E-14 Plan of Exchange of Uncertified Spectra. (Chairman of this program is $. H. Struck, Research Division, dmerican Cyanamid Co., 1937 West Main St., Stamford, Conn.) Except for methyl benzoate, purchased from the Eastman Kodak Co., all the compounds were synthesized in these laboratories. Checks made with the mass spectrometer confirmed the purity of these materials to be sufficient for the purposes of the present study.
+
DISCUSSION OF SPECTRA Benzyl Benzoates. Ar-COO-CH2 -Ar. Spectral correlations for several classes of aromatic esters have been discussed by McLafferty and Gohlke
(6) and by Emery ( 3 ) . The breakdown of the molecule under electron bombardment occurs a t the carbonyl group, and the major peaks correspond to fragments of the general type Ar, ArCO, and BrCH2. The particular group we discuss presents many similarities with the spectra reported by these authors; however, the presence of methyl substituents introduces new parameters of interest. The position of these substituents plays an important part in the formation of the major ions and thus it can usually be determined. Table I includes the s ignifjcant spec tral features of nine methyl-substituted benzyl benzoate-type esters, and Figure 1 illustrates some possibiliti es of frag mentation. In esters of the type
Ra’, Ra”, Rb’, Rb” being inethyl groups, the main position effect (1, 6) is related to the proximity of Rb’ and Rb”-Le., of the substituents on the alcoholic moiety of the molecule. If and only if a methyl group is substituted ortho to the esteric linkage is there a strong probability of the formation of a rearrangement ion, CsHa (P-136) in the sylyl toluates or CSHI~ (P-150) in the dimethyl benzyl dimethyl benzoates, as pointed out previously (4). [For brevity, fragments deriving from these dimethyl-substituted benzyl benzoates will be henceforth discussed together with those deriving from the xylyl toluates, and referred to in parenthesis-for example, P-136 (or P-150) for the ion discussed above.] The position of the substituents on the VOL. 34, NO. 1, JANUARY 1962
33
acidic moiety of the ester has no major effect. The presence of this rearrangement ion affects the entire spectral pattern of these compounds, thus allowing analytical differentiation. Thus the parent peak (m/e 240 or m/e 288) is practically absent from compounds with a methyl group ortho t o the esteric linkage, while it is of average intensity in the others. In addition to the rearrangement ions discussed above, abundant and analytically significant ions are formed by
the breakdown of the molecule @ to the benzyl or benzoyl ring, giving origin to ions corresponding, respectively, to P-135 (or P-149) and to P-121 (or P-135). The former is supposed to be alkylbenzenic type; the latter is believed to possess a carbonyl group. Because of the nominal mass equivalence of two methylene groups and one carbonyl, the P-135 (or P-149) ion could be attributed also to a carbonylictype structure deriving from further fragmentation of the P-121 (or P-135) ion (see Figure 1).
Table 1. Spectral Features of Benzyl Benzoate-Type Esters Peak heights expressed in per cent of total ionization [z: ( m l e 73 to vale P 2)] 2-Methyl4-Methyl2,B-Diniethyl2,5-Dimethylbenzyl-2benzyl-4 benzyl-2,5benzy1-?,4Compound methyl Benzoate methyl Benzoate dimethyl Benaoate dimethyl Benzoate hl. w. 240 240 268 268 2 ( m / e 73 to m/eP +2) 15,728 13,607 9 ,658 8,640 P 0.15 7.94 0.20 0.08 P- 1 ... ... ... ... P-15 0.11 0.34 0.04 0.01 P-17 0.14 0.01 0.09 0.01 0.47 P-18 0.74 0.07 0.04 P44 0.10 0.03 0.03 ... P-45 0.18 0.18 0.07 ... P-I05 0.30 0.01 0.10 ... P-121 10.99 28.96 ... 0.04 P-122 0.42 0.14 ... 30.20 P-135 ii :os 23.37 8.72 27.66 P-136 3.87 0.34 0.28 2.22 2.20 P-137 0.03 0.01 P-149 23.92 27.27 7.78 7.42 P-150 0.90 33.09 0.99 30.64 P-163 3.73 3.16 4.37 3.46 P-177 3.76 ... ... 3.62 P-191 4.68 4.45 ... ... P-119 0.04 ... ... 0.26 4.68 4.45 m/e 77 4.37 3.73 3.62 3.76 7.78 7.42 m/e 91 mle 105 3.45 3.16 23.37 30.20 m)e 119 10.99 28.96 27.27 23.92 m/e 133 0.30 0.05 8.72 11.09 mle 147 *.. ... ... ...
+
Table II.
Compound
w.
hI. z(m/e 73 to
m/e P P P-1 P-15 P-17 P-18
P-3 1 P-32 P-44 P-45 P-59 P-60 m/e 77 m/e 91 m/e 104 m/e 105 m / e 119
34
0
+ 2)
Spectral Features of Methyl Benzoates
2,5-Di2,4Di3,5-DiMethyl Methyl- Methyl- methyl methyl methyl benzoopMethyl Methyl Methyl ate toluate toluate Benzoate Benzoate Benzoate 136 164 164 150 150 164 17,338 13.86 0.58 0.07 0.16 0.08 41.99 0.24 0.57 0.78 27.34 2.00 27.34 0.78 0.24 41.99 0.16
17,194 12.41 0.50 2.07 0.23 0.04 26.99 13.71 0.15 0.64 20.12 5.96 2.31 20.12 0.07 0.64 26.99 -
ANALYTICAL CHEMISTRY
15, 805 14.64 0.66
0.26 0.16 0.04 45.62 0.31 0.16 0.62 20.05 2.00 0.67 20.05 0.05 0.62 45.62 __
21,005 10.63 0.36 2.95 0.23 0.03 18.99 14.56 0.09 0.46 12.37 7.73 7.68 1.71 7.73 12.37 0.46
25 ,907 10.50 0.53 1.52 0.18 0.04 27.41 11.09 0.12 0.52 14.00 3.95 7.11 1.46 3.95 14.0 0.52
19 ,838 14.27 0.70 0.77 0.12 0.06
36.29 0.36 0.12 0.66 14.09 1.78 6.49 1.05 1.78 14.09 0.66
?,4,5-Trimethyl Methyl Benzoate 178 18,863 10.83 0.48 2.62 0.19 0.04 21.52 12.62 0.12 0.79 8.54 5.67 3.75 6.47 1.45 0.92 8.54
This point was elucidated recently, a t least for o-xylyl-o-toluate, with a research mass spectrometer built by F. H. Field of the Hunible Laboratories. This instrument is capable of offering the high resolution ( M / A U = 2880) needed for distinguishing bctn-een one carbonyl and two methylene groups. The partial spectrum of 0-xylyl-0toluate run with the afore-mentioned instrument presented a doublet (m/e 105.1038 and m/e 103.0874) at the nominal m/e 105 (P-133) , confirming tliat both alkylbenzenic and carbonylic type structures are present. The alkyl benzenic type ion (m/e 105.1038) is much more abundant than the other (51.2 divisions against 3.3), as expected. Alhyl benzenic type ions (mle 77, nile 91) are formed from more pronounced fragmentation. At this stage of fragmentation obviously the positions of the original methyl groups have little effect, and in fact these peak intensities are nearly the same in all compounds. Methyl Benzoates. Ar-COO-R. Mass spectra of methyl esters of aromatic acids have been published and discussed by the authors mentioned above (3, 5 ) . Although some of the spectra we obtained have been examined in these publications, they are included in this paper to make the following general discussion complete. Table I1 includes the spectra of seven methyl benzoates. In agreement with the data published by AIcLafferty and Gohlke, and also by Emery, the major peak in the spectrum corresponds to the fragment deriving from the loss of a inethosy group [P-31 (OC&)]. The ortho effect is evident for the P-32 [(HOCH,)] peak and the trend is regular also in the dimethyl-substituted compounds. The formation of this ion is more probable in the compound which is substituted in the 2,5 rather than in the 2.4 position and is practically nil in 3,b-diniethyl methyl benzoate. The proximity, or ortho effect, is observed also for the P-60 (H-COOCH3) ion. Methyl esters can be identified in mistures by their parent peaks and the P-31 fragments. The latter are found also in the spectra of the isomeric benzyl alcohols, but together with a strong P-18 peak. GENERAL DISCUSSION
Correlations dependent on the presence of an oxygenated functional group and methyl groups substituted on aromatic rings can be generalized for most compounds discussed in the present work. In particular, when the size of a peak is influenced by the position of a methyl
/’
Figure 1 . Fragmentation possibilities in benzyl benzoatetype esters Peak heights expressed in per cent of total ionization [.Z(rn/e 73 to m/e P 211
group, the effect is the same in most of the compounds examined. Parent peak intensities of 50 oxygenated aromatic compounds are given in Table 111. Except for the benzyl benzoates with methyl groups in positions ortho to the esteric bridge on the alcoholic moiety of the molecule, all parent peaks are intense and therefore useful for analytical purposes. The importance of the parent decreases with increasing carbon number. Phenols possess more stable molecule ions than aldehydes, acids, and alcohols (in this order). This effect possibly can be correlated with resonance probabilities. Scids present more important parent peaks than their methyl esters (5). The abundance of the P-OR (R = H, CH3) peak is related mainly to the nature of the functional group and to the position of the oxygen atom (see Table IV). It decreases from methyl benzoates to acids, alcohols, and phenols, and is obviously absent from aldehydes. The P-OR ion is insignificant, in general, whenever the oxygen is directly substituted on the ring.
+
Table 111. Parent Peaks in Aromatic Oxygenated Compounds Peak heights expressed in per cent of total ionization [ z ( m / e73 to m / e P 2,62,52,42,3o-Me nz-Xe p-hle DiMe DiMe DiMe DiMe Compound Type? 40 12 30 64 34 0 32 43 26 05 25 43 26 74 Phenole 26 13 25 10 24 51 23 97 Benzaldehydes Benzoic acids 24 52 17 51 23 20 19 60 13 89 15 02 15 35 15 96 Methyl benzoates 13 86 12 41 14 64 10 63 10 50 Benzyl alcohols 18 7 2 7 23 14 63 9 57 10 66 Methyl benzyl rnethj 1 benzoatesu 0 15 7 94 Dimethyl benzyl dimethyl benzoatesa 2,4-Di&Ie acid 0 08 0 20 2,5-DihIe acid 0 20 1 13 ?,5-DiMe acid 0 14
+ 2)]
3
3,53,4- 2,4,6DihIe DihIe TriMe 30 42 24 60 24 60 19 18 22 36 18 84 14 58 14 27 14 i 9
2,4,5TriMe 14 28
10.83 9.42
5 27
6 75
Vertical columns refer t o position of niethj 1 groups on ,Llcoholic part of ester.
Table IV.
Coiiipoiind Types Methyl benzoates 41.99 Benzoic acids 31.28 Benzyl alcohols 5.96 Phenols 0.63 Benzaldehydes 0.12
2 0 0 0
08
48 24 55
Ion
+
Peak heights expressed in per cent of total ionization [ Z ( m / e 73 to m/e P 2)] 2,62)s2,P 2,33,s3,4o-Ye m-Me p-Me DiMe DiMe DiMe DiMe DiMe DiMe 26.99 45.62 18.99 27.41 36.29 7.43 15.09 20.62 5.50 4.98 9.00 4.40 8.90 8.01 4.21 3.85 3.81 4.50 2.66 3.44 2.14 1.73 1.32 1.13 1.03 0.96 0.68 0.07 0.12 0.22
Table V.
Compound Types Benzyl alcohols Benzoic acids Methyl benzoates Phenols
Probability of Formation of the P-OR
Probability of Formation of the P-ROH
Ion
2,4,6-
Trihfe 6.79
2,4,5TriMe 21.52 4.88 5.21 1.13
0.10
+
Peak heights expressed in per cent of total ionization [ 2 ( m / e73 to m/e P 2)] 2,62,s 2,42,33,s3,4o-Ne m-&le p-Me DiMe DiMe DiMe Dihfe DiMe DiMe 25 60 2 29 20 56 13 13 3 17 20 36 1 26 0 59 21 07 13 76 17 26 14 47 1 05 3 77 13 71 0 31 14 56 11 09 0 36 7 07 3 05 2 49 2 67 1 91 1 11 0 56 0 42
2,4,6-
Trillle 14 05 16 45
2,4,5-
Trille
0 59
VOL. 34, NO. 1 , JANUARY 1962
9 4.3 12 62
35
Table VI.
Compound Types Benzoic acids
F = -17 (OH)
Probability of Formation of the P-F
-I- 1 ) Ions
+
[Z(m/e73 to m / e P 2)] 2,4 2,33,53?4DiMe DiMe DiMe DiMe
27.34
20.12
14.00
29.14
32.28
15.18
9.53
0.63
2.44
20.05 31.55
12.37
2.46 2.00 1.65 0.88
0.55
1.51 5.96 1.81
0.63 7.07
3.47 1.43 3.05
14.15
22.20
21.43
1.73
2.43 2.00 1.54 0.58 2.49
2,4,6Trihfe
2,4,5TriMe
7.15
13.30 8.54
14.09 13.03
10.59 2.14
12.18
29.71 1.32 P-(F
Benzoic acids Methyl benzoates Benzaldehydes Benzyl alcohols Phenols
and P-(F
Peak heights expressed in per cent of total ionization 2,62,5o-Me m-Me p-Me DiMe DiMe P-F (functional group) 24.10 21.63 35.63 29.58 10.04 15.28
4.70
4.17
3.71
3.87
1.11
1.03
0.96
0.68
0.40
2.83 1.78
3.69
4.99
0.82 0.56
0.42
0.59
+ 1)
12.18
13.82 7.73
9.37 3.95
2.67
0.96 1.91
0.95 1.11
10.70
9.32 5.67 0.64 0.91
Determine the m/e of the parent peak (almost always immediate in a single aromatic compound). Assuming that the parent is m/e 122, it can be given by alcohols, acids, or phenols. The next peak to examine is P-1, indicated by the arrows and by the dashed horizontal line. According to the size of this peak (expressed in per cent of the base peak for convenience), it can be decided whether the compound is acid, alcohol, or phenol. If, for example, the P-1 is larger than 35% of the base peak, the compound is a CS phenol. The sizes of the P-18 and P-15 peaks indicate the possibility of identification between the isomeric CSphenols. (“Check points” are indicated always by the intersection of the arrows and the dashed lines.) According to the size of the P-18, three isomers can be distinguished. In the case of the 3,5-or 3,4dimethylphenols, a check is required also on P-15.
The peak most affected by the positions of the alkyl substituents is the P-ROH (Table V). The very striking ortho effect manifested by this peak has been discussed (1, 2, 6). The table shows the remarkable regularity of this trend. The peak is present also in compounds with no methyl groups. It is assumed, therefore, that the H atom necessary for the formation of the neutral fragment (ROH), or a t least for the formation of a P-18 fragment, can be derived not only from a CHs group but also from the ring. In alcohols, the structure of the positive ion seems t o be analogous t o the rearrangement ion found in benzyl benzoates. Table VI illustrates the probability of formation of ions deriving from the loss of the functional group (F) and of (F 1) fragments. The intensity of
+
36
ANALYTICAL CHEMISTRY
the P-F peak is related to the nature of the functional group itself; the intensity of the P-(F 1) ion, in analogy t o the P-ROH fragment, is related to the positions of the methyl groups. The correlations discussed are only the more important ones and others can be derived from the data reported in the tables. As an example of analytical application, an identification scheme is presented in Table VII. It is similar to that of Meyerson (6) for alkylbenzenes and is applicable to pure compounds and simple mixtures. The table may be amplified using data presented in this report and extrapolated to include compounds not available for calibration. An example will explain the use of Table VII.
+
Cross checks are provided for each single compound, as indicated on the arrow before the compound’s name. The reliability of this scheme is a function of the magnitudes of the spectral differences between isomers. In the above cited example, the identification of the three isomeric Cs alcohols is more certain than that of the five Cs phenols. A final comparison of the spectra of the unknown sample with that of the suspected compound, if available, is always, obviously, the best proof. CONCLUSION
The data discussed in this paper and in the previous communications show that the spectral features of different types of aromatic oxygenated compounds are sufficiently characterized by the various functional groups for identification purposes. Quantitative
compound type analyses can also be developed in favorable cases, as for phenols. Within the same class, spectral differences are due mostly to the positions and number of the substituents. Positional isomers, in particular ortho and para, can be, therefore, distinguished and may be determined quantitatively in simple mixtures. The correlations pointed out are, in most cases, sufficiently linear to permit prediction, by extrapolation, of spectral features of compounds or groups of
compounds. Thus identification and determination of compounds, for which no calibration data are available, are possible. ACKNOWLEDGMENT
We acknowledge the work of B. F. Armendt, G. W. Gr&, R. E. Pennington, and R. H. Perry in synthesizing many of the compounds used, and of Harold Kail, G. R. Taylor, and J. L. Taylor, who obtained and tabulated most of the spectra.
LITERATURE CITED
(1) Aczel, Thomas, Lumpkin, H. E., ANAL. CHEM.32, 1919 (1960). (2) Ibid., 33, 386 (1961). (3) Emery, E. M., Ibid., 32, 1495 (1960). (4) Lumpkin, H. E., Nicholson, D. E., Ibid., 32, 74 (1960). (5) McLafFerty, F. W., Gohlke, R. S., Ibid., 31,2076 (1959). (6) Meyerson, S., A p p l . Spectroscopy 9, 120 (1955).
RECEIVED for review July
20, 1961. Accepted October 27, 1961. Committee E14, ASTM, Atlantic City, N. J., June 27July 1, 1960.
Theoretical Basis for a Continuous, Large-Capacity Gas Chromatographic Apparatus J. CALVIN GlDDlNGS Department o f Chemistry, University of Utah, Salt lake City 72, Utah
b A continuous gas chromatographic apparatus is proposed for certain analytical and preparative scale applications. The relative merits of the system are examined using recent developments in the theory of gas chromatography. The apparatus is potentially capable of better resolution and greater throughput than a large conventional column of similar cross section. Factors affecting the choice of optimum operating conditions are discussed.
T
development of equipment for the continuous analysis of volatile liquids should prove useful in the area of process control. The merits of such a system would be further enhanced if adaptation could be made to preparative scale work. -4theoretical analysis of a system designed for these purposes is presented below with the object of delineating and comparing its potential capabilities with conventional systems. The scale-up of conventional column diameter beyond about 1 inch leads to a loss of resolution. An examination of the reason for this loss leads to the suggestion that an effective column design for preparative scale n-ork would employ the annular space between the malls of two concentric cylinders. This column could be usefully employed in the usual discontinuous manner. With proper consideration given to column packing, the plate height increase typically found in large columns could be kept minimal while the capacity could be made very large. I n addition to this advantage, the column could be made to operate continuously. I n 1949 Martin suggested, without regard to the HE
possible large-capacity, plate-height ratio, that a cylindrical system as described above could be operated continuously b y rotating the cylinders and fixing the points of injection and sample collection (7). I n view of the theoretical analysis presented below, such a system would appear to be highly feasible. The development of the theory for this system proceeds in two parts: First, the problems peculiar to large scale-up are discussed; and second, the equations pertaining to the resolution of the device, used analytically or in preparative scale work, are derived. The theory used for obtaining optimum operating conditions may also be applied to a n analogous system in which a thin paper sheet is formed as a cylinder for continuous paper chromatographic operation (S). RESOLUTION IN LARGE SCALE SYSTEMS
The tendency of large columns to be less efficient for separation is a result of nonequilibrium established laterally across the tube. Lateral nonequilibrium, which has a fundamental role in determining column performance for columns of any dimensions, is a consequence of point to point variations in downstream solute velocity. When the degree of lateral nonequilibrium is controlled by lateral diffusion processes, a contribution to the plate height equal to H= f(a,r/R,)R?/D results, where the tube nonuniformity is expressed in the function, f, which depends on cross-sectional coordinates r (radius) and CY (polar angle). The coefficient for lateral diffusion is shown as D and the tube radius as R,. This equation is based on the assumption that a given nonuniformity persists
unchanged throughout the column length. A case in which the nonuniformity varies from one side of a column to another-Le., f is a function of both CY and r-has been treated theoretically by Golay (6). A case in which f is a function of r only has been treated by Huyten, van Beersum, and Rijnders (6). Without specifying the nature of the nonuniformity, the above equation shows that the plate height increases with the square of the radius for tubes with geometrically similar packing nonuniformity-Le., a fixed degree of variation between any two points which have the same CY and r/Ro coordinate in each of the tubes, or, alternatively, the same function f. While little is known about the variation o f f with tube size, we may conclude that a large tube, by virtue of its size alone, contributes a term to the plate height increasing approximately as the square of radius. This term mill be decreased by the use of any method able to improve the uniformity of the packing. Csing careful packing technique, Carle and Johns (1) found the increase in plate height to be serious for columns larger than 1 inch in diameter. The problem of increasing the uniformity of porous support is closely associated with the physical processes occurring during packing. A theoretical approach has not been made because of the very complicated nature of the physical processes, and the empirical approach has not been entirely successful because of the large number of variables. There is, however, one general theoretical principle which can be used to reduce nonuniformity. This principle employs the symmetry inVOL 34, NO. 1, JANUARY 1962
* 37
