Correlation of Mass Spectra with Structure in Aromatic Oxygenated Compounds Aromatic Alcohols and Phenols THOMAS ACZEL and H. E. LUMPKIN Manufacturing Deparfmenf, Research and Developmenf Division, Humble Division, Humble Oil & Refining
b A systematic study of the mass spectra of eight benzyl alcohols, 1 1 monohydroxybenzenes (phenols), and three dihydroxybenzenes is presented. Correlations between spectral features and chemical structure are established. The consequent analytical applications, as qualitative and quantitative analyses of mixtures, and the possibility of prediction of sensitivities of compounds at present unavailable for calibration, are pointed out. The fragmentation pattern of the compounds examined depends on the nature of ihe functional groups as well as the position and number of the alkyl substituents attached to the benzenic ring. The influence of the position of the substituents on the spectra can be used to distinguish isomeric compounds.
T
investigation of the correlations existing between mass spectra and chemical structure in various series of compounds provides a valuable tool for the analyst, in that it permits one to extrapolate calibration data for compounds which may be unobtainable. I n addition, it leads to a more rational approach to identification problems. Structural information on the ions formed under electron impact, as well as on the mechanism involved. can also be deduced, although some further effort, such as the study of labeled compounds, is generally required for this purpose (IO). Information of the nature delineated above, which is extremely valuable in the field of theoretical and analytical mass spectroscopy, has been published for a number of compound types, including aliphatic acids (S), alcohols @), and esters ( I $ ) , as well as alkylbenzenes (6, 9) and aromatic acids and esters (8). This study deals with the spectra of aromatic oxygenated compounds, most of which are present in the oxidation products of alkylbenzenes. The spectra of eight benzyl alcohols, 11 monohydroxybenzenes, and three dihydroxybenzenes are discussed, with particular emphasis on the ions useful for analytical purposes. The spectra of a number sf aromatic acids, aldeHE
hydes, and esters, as well as the correlations which are applicable to most of the aromatic oxygenated compound types examined in the course of this work, will be discussed in subsequent communications. EXPERIMENTAL
The data reported were obtained on Consolidated Electrodynamics Corp. Models 21-102 and 21-103C mass spectrometers, and recorded either on an oscillograph or with the CEC peak digitizer (Mascot). Samples were introduced into the mass spectrometer heated inlet system (250" to 300" C.) in the liquid or solid state, For the latter case a solids inlet system developed in these laboratories will be described in a separate communication. Peak heights are expressed as per cent of the total ionization (4,II), obtained by the summation of the intensities of every peak of the spectra from m/e 73 to m/e parent 2. Partial total ionization summation is used because some spectra were originally obtained only above m/e 73. Neglecting the mass numbers below m/e 73 assumes that aromatic oxygenated compound peaks from m/e 12 to m/e 73 are similar in analogous aromatic compounds. To check this assumption, the total ionization in some phenols was calculated both from m,'e 73 to m/e P 2, and from m/e 12 to m/e P 4-2. The data reported in Table I show that the ratios of z(m/e 12 to P 2) and z(m/e 73 to P 2 ) are practically constant (s = 0.0679 or 3.99%) in
+
+
+
Table I,
+
Co., Baytown,
Tex.
various compounds, which means that peak heights expressed either in per cent of z(m/e 12 to P 2) or of Z(mle 73 to P 2) are equivalent within the same series. The compounds used were purchased, when available; otherwise they were synthesized in these laboratories. The mass spectra indicate the purities of all the compounds discussed to be above 99%.
+
+
DISCUSSION OF SPECTRA
Spectral features of the compounds examined can be correlated to the stability and fragmentation pattern of the benzenic ring, as well as to the nature, position and number of the "functional" oxygenated groups and of the alkyl groups substituted on the ring. I n our case only methyl substitution mas investigated, but the data obtained can be extrapolated to other types of alkyl substitutions with reasonable certainty. Apart from the influence of the functional groups, the position of the alkyl substituents with respect to the former seems to have a very definite part in leading to different paths of fragmentation. This is particularly true in cases in which the ions are formed by the contemporary loss of fragments from both groups, as in the P - 18 (HOH) peak. The abundance of the ions formed in this manner is in direct relation to the respective proximity of the group. This effect was observed also by McLaiYerty and
Comparison of Total and Partial Summation of Peak Heights of Several Phenols
Compound o-Cresol m-Cresol 2,6-Dimethylphenol 2,5-Dimethylphenol 2,4-Dimethylphenol 3,gDimethylphenol 3,4-Dimethylphenol
z(m/e 12 to P 3. 2) 27,990 27,890 23,385 24,728 28,716 24.377 25,589
z(m/e 73 to p 2) 15.827 15.660 14.465 15.850 16.641 15.091 14.880
+
%12+) 2(73+) 1.768 1.781 1.618 1.660 1.726 1.615 1.720
VOL. 32, NO. 13, DECEMBER 1960
0
1819
Table 11. Spectral Features of Aromatic Alcohols [Peak heights expreesed in per cent of total ionization z(m/e 73 to m / e P
+ 2)]
z(m/e
73 to
Compound Benzyl alcohol 0-Xvlyl alcohol p-Xklyl alcohol 2,5-Dimethyl benzyI alcohol 2,PDimethyl benzyl alcohol 3,5-Dimethyl benzyl alcohol 2,4,5-Trimethyl benzyl alcohol 1,4-Benzendicarbinol
mle
+
M.W. P 2) Parent P-1 P-15 108 15,459 18.72 15.56 0.85 122 16,016 7.23 1.39 8.64 122 '16,795 14.63 3.52 17.98 136
18,433
136
17,730 10.66
e
0.50
15.18 5.96 23.68 10.15 9.53 10.51 11.07 10.99 8.79
7.14
7.10 10.56
2.35
1.23 10.66 4.50 13.13
5.12
3.71 0.95
7.28
6.97 10.76
2.40
3.17
8.66
3.87 0.82 10.10
7.14 11.32
2.58
6.11 5.21 14.05 0.07 0.86 2.34
4.34
7.40 0.07
16,897 14.79
1.89 13.44 2.66 0.63
ANALYTICAL CHEMISTRY
m/e 77 m / e 91 m/e 79
4.17 0.96
17,095 12,629
,.12
P-43 2.14 0.63 10.51 0.57 8.79
4.09
136
!.42
P-32
0.88
5.16 3.81 20.56
150 138
Gohlke (8) and by Lumpkin and Nicholson (6). Aromatic Alcohols. Spectral features of eight aromatic alcohols are reported in Table 11. Only ions which are believed to be significant for identification purposes are included and are discussed separately in the following paragraphs. (Complete spectra of all the compounds discussed in t h e present paper are available through t h e ASTM E-14 plan of exchange of uncertified mass spectra. Chairman of this program is A. H. Struck, Research Division, American Cyanamid Co., 1937 Kest Main St., Stamford, Colin.) The molecule ion is very abundant in aromatic alcohols, in contrast to the corresponding aliphatic series in which the parent ion is insignificant or absent (W). This difference is attributed to the stability of the benzenic ring, the effect being general in most aromatic compounds. The decrease in peak intensity due to the increase of the molecular weight is also evident. I n addition, the importance of the peak is in indirect relation to the proximity of a methyl substituent to the alcohoiic group. This effect is believed to depend on the opposite behavior of some of the most abundant fragment peaks. The ion resulting from the loss of hydrogen (P- 1) is significant only in benzyl alcohol. It is relatively small in the homologous alcohols (including the dicarbinol) and i t is therefore very useful in distinguishing methyl-substituted alcohols from the isomeric phenols, which, as will be discussed later, present a very strong P - 1 peak. The P-15 (CH3) peak is practically absent from benzyl alcohol and the unsubstituted dicarbinol, but is important in the other alcohols. Its abundance decreases as methyl substituents are moved closer to the functional group, and with the number of methyl substituents present on the 1820
9.57
P-17 P-18 P-29 P-31 5.96 2.08 23.68 15.18 4.21 23.60 5.31 9.53 3.85 2.29 8.50 10.59
1.08
4.70 0.91 0.08 15.78 0.94
ring. This second trend is contrary to that shown by other types of oxygenated compounds. A very characteristic peak in alcohols is the P-18 (HOH) peak which is the base in several compounds, Its abundance increases very sharply in the presence of a methyl group ortho to the alcoholic functional group. Isomeric alcohols can be identified by considering the abundance of this peak, since the trend is very regular, as evident from Table 11. The P-31 (CHzOH) ion originates from an a cleavage of the functional group. Its importance decreases with molecular weight. If a P-18 peak is also present, the occurrence of this peak is a good criterion for the identification of alcohols. Although of very small size, the P-32 (CH30H) peak, deriving from a process analogous to that of the P HOH ion, seems t o be subject to the same tvpe of influence, dependent on the position of the substituent methyl groups, as the P-HOH. The base peak in benzyl alcohol is the P-29 (COH)but it is much less important in its substituted homologs. The m/e 79 ion is present also in the methyl-substituted alcohols. The abundance of this ion could be explained by the stability of the CaH7+ product ion. Blkylbenzene type ions (m/e 77, 91 . . .) may be derived either from direct a: cleavages (P-31) or from n ( a cleavages) n-1 (H rearrangements). The abundance of the m/e 91 ion in methyl-substituted compounds might be explained also with the tropyliuni ion hypothesis of Meyerson and Rylander ( I O ) . Phenols. Table 111 shows the spectral features of 10 monohydroxyand three dihydroxybenzenes. Again, only the most characteristic peaks are reported and discussed. Monohydroxybenzenes. T h e great stability of the molecule ion may be attributed to the resonance structure
+
4.32 8 . 0 7 2.15 11.48 11.68 19.42
involving the unshared electrons on the oxygen atom. The abundance of t h e peak decreases slightly with added CH, substitution (more if the substituents are other than methyl, particularly if the side chain is branched, as in 2,6-di-tert-buty1-4-methylphenol). The position of the methyl groups has little significance. Isomers present about equal sensitivities for this peak, and the parent peaks of higher molecular weight compounds do not give large interferences on the parent peaks of lower molecular weight compounds of the series. A breakdown in carbon number types and determination of the average molecular weight of a mixture of phenols is therefore possible. The procedure is analogous to that devised by Lumpkin and Thomas (?) for alkjlbenzenes. Table IV illustrates a calibration matrix of this type employed by the Humble Laboratories in a determination of the type of phenolic inhibitors in spent caustic material extracted by CS2. Phenols present a very important P-1 (H) peak, derived from a p cleavage of the H of one of the alkyl groups. This peak is a good identification peak with respect to acids and alcohols. It is the base in p-cresol. The peak decreases rapidly n-ith increasing molecular weight, but is absent from phenol itself. The P-15 (CHI) peak is practically absent from phenol and crecols, but grows in abundance in the higher homologous compounds. P-15 is the base peak in the dimethylphenols and in 2,6-di-tertbutyl-4-methylphenol.The absence of the peak in the cresols may be explained again by the tropylium theory ( I O ) . Based on the P-1 and P-15 peak sensitivities ( m / e 107, 121, 135.. .), a good compound-type analysis for phenols in the presence of alkylbenzenes can be developed. Interferences with the 277 peaks are constant between isomers. The peak resulting from the loss of an OH fragment is not very abundant
in phenols. The fact is probably due to the vicinity and interaction of the "electron-sharing" oxygen to the aromatic ring. The peak intensity decreases with increasing molecular weight. There is also a slight preference for the formation of this fragment ion when a methyl group is ortho to the functional group. The P - 18 (HOH) peak is derived from a simultaneous a and cleavage. Because of the above-mentioned resistance to the CI cleavage (of the OH group), the peak is less important in phenols than in other oxygenated compounds. Nevertheless, its intensity is a remarkably regular function of the proximity of the methyl substituents to the functional group. P - 28 (CO) is a very important and characteristic peak in phenol, but not in its homologs. The peak is probably due to the loss of CO, as pointed out by Beynon ( I ) . The loss of a COH ( P - 29) fragment is probably due to a process similar to the formation of the above-mentioned P - 28 ion. Although no experimental support is available, i t is believed that both ions may be present in a fivemembered ring type structure. P 29 is more abundant than P - 28 in the substituted phenols, while the contrary is true for phenol. The P - 29 ion obviously can also derive from the loss of an alkyl-type fragment, I n phenol, P - 55 corresponds to m/e 39 - CaH3, and is, therefore, important. I n cresols it may correspond to the ion formed by breaking the molecule in two about-equal parts; precisely, to the fragment with the methyl-substituted half (CsH2-CH3-m/e 53). The peak is slightly bigger when the methyl group is in para position to the OH group and is almost absent from the other homologs. Furthermore, it is larger than the peak correspondent to the ion formed from the other half of the molecule, P - 53 (CBHS - 013). Apparently, in this type of cleavage the positive charge remains preferentially on the CsHz CHB fragment. Both P - 55 and P - 53 ions are obviously almost absent in the higher homologs. The alkylbenzenic type ions are derived from simple cleavage of the functional group, or by contemporary cleavage of an OH and one or more CH3 groups, with rearrangement of one or more H atoms to the ring. Contrary to most other classes, the most abundant alkylbenzenic type ion is not always the one correspondent to the simple cleavage :
*% E d :. .: .: .: .: .: .: .: :' Czr i,
g::: '
'
'
Cresols. m/e 77 > 91 > 105 Dimethylphenols. 77 > 91 > 105 > 119 Trimethylphenols. 77 < 91 > 105 > 119 > 133 2,6-Di-tert-butyl-4-methylphenol.77 < 91 > 105 > 119 > 133 > 147 VOL. 32, NO. 13, DECEMBER 1960
a
1821
277-147 is almost constant in isomers and the contribution of different isomers t o any single alkylbenzenic type peak i s very similar. Dihydroxybenzenes. Only three compounds of this group-hydroquinone, resorcinol, and catechol-have been examined. This structure is even more stable than the phenols, a s evidenced b y very large parent peaks. Indeed. no fragment peak is greater than about 25% of the parent. As indicated above, the parent is always the base peak and is slightly more abundant in hydroquinone (para) than in catechol (ortho). Methyl-substituted homologs should also give a very stable parent peak. Thus determination of this type of compound a t the respective parent peaks is feasible. -4s in phenol, the P - 1 (H) peak is low in abundance. The P - 18 (HOH) peak is present in catechol, and decreases in the other two isomers, which do not possess functional groups ortho to each other. A comparison of the analogous peak in the correspondent cresols shows that the formation of this ion is more probable in the presence of a CH, and an OH group than in the presence of two OH groups. P - 55 (CsH2-OH) is the ion resulting from the breaking of the molecule into two equal parts whose abundances decrease as a function of the vicinity of the two functional groups to each
Characteristic rihydroxyazo eagent for
Table IV.
Matrix for Determination of Molecular Weight Distribution of Phenols
Phenol
Cresols
Dimethylphenols
Trimethylphenols
5285"
4.9 5182
116.5 297.2 4099
64 71 354 3770
m/e 94 m / e 108 m/e 122 m/e 136 Q
...
...
... ...
...
...
Divisions/pipet full.
other. The structure of this could be the following types:
OH I
C
/\
HC-CH+
OH
or
'CH-CH
=
/
C
The presence of an even stronger P 57 peak (with analogous trends in the three isomers) could indicate the possibility of a quinonic-type structure, as :
8
HCC" Both peaks can be employed successfully in distinguishing among the three isomeric dihydroxybenzenes. ACKNOWLEDGMENT
The authors thank Harold Kail, G. R. Taylor, and J. L. Taylor for their valuable contributions in obtaining and tabulating the spectra.
stunts of
LITERATURE CITED
(1) Beynon, J. H., Lester, G. R., Williams, A. E., J . Phys. Chem. 63, 1861 (1959). (2) Friedel, R. A., Sharkey, A. G., Jr., Shultz, J. L., ANAL. CHEM. 28, 926 (1956). (3) Happ, G. P., Steward, D. W., J . Am. Chem. Soc. 74, 4404 (1952). (4) Hood, A,, ANAL. CHEM. 30, 1218 (1958). (5)' Kinney, J. W., Jr., Cook, G. L., Ibid., 24, 1391 (1952). (6) Lumpkin, H. E., Xicholson, D. E., Ibid., 32, 74 (1960). (7) Lumpkin, H. E., Thomas, B. W., Ibid., 23,1739 (1951). (8) McLafferty, F. W., Gohlke, R. S., Ibzd., 31, 2076 (1959). (9) Meyerson, S., Appl. Spectroscopy 9, No. 3, 120 (1955). (10) Meyerson, S., Rylander, P. M., J.Phys. Chem. 6 2 , 2 1 (1958). (11) Otvos, J. W., Stevenson, D. P., J.Am. Chem. SOC.78,546 (1956). (12) Sharkey, A. G., Jr., Shulta, J. L., Friedel, R. rl., ANAL. CHEM.31, 87 (1959).
RECEIVEDfor review May 19, 1960. Accepted September 9, 1960. E 1 4 Committee on Maes Spectrometry, ASTM, Atlantic City, N. J.
2, ulfonic Acid, tornetric Analysis
MARY H. FLETCHER
U. S. Geological Survey, Washington, D. C. The dye 2,2',4'-trihydroxyazobennene-5-sulfonic acid, has shown promise as a reagent for the determination of zirconium. As the literature contains very little information about this dye, basic data pertinent to its use as a reagent were determined. The sulfonic acid group and all three of the hydroxy groups show acidic characteristics. Apparent dissociation constants were determined for the three more labile protons and the approximate order of magnitude for the fourth constant was estimated. Absorption spectra for the different ionization species are given. A curve
1822 *
ANALYTICAL CHEMISTRY
is also included which shows the fraction of dye in the different ionization forms a t acidities from 10.35M hydrochloric acid to p H 11.9. A sixth dye species was found in 1.0 to 8.4M potassium hydroxide solutions, but its nature is unknown.
paper reports a study of an o,o'-dihydroxyazo dye, 2,2',4'-trihydroxyazobenzene-5-sulfonic acid, that has shown potentialities as a reagent for the determination of zirconium. Although several references relate to applicationa of the dye, the literature HIS
yielded very meager basic information concerning the dye. I t s sodium salt was discovered by Erdmann and Borgman in 1893 (1). Drew and Dunton (4)used it to prepare a vanadyl complex and were able to isolate crystalline disodium and diammonium salts of the complex. Powell and Saylor (9) used the dye as a reagent for an indirect fluorometric determination of fluorine in solutions at p H 4.8. Their method is based on the decreased fluorescence given by an aluminum-dye complex when fluorine is present. Schubert (11) developed a fluorometric method for the determina-
