Mass Spectrometric Determination of the Isomer Distribution of Carbon,, to Carbon,, Phenyl-n-alkanes A. BRUCE KING and M. R. BASILA Gulf Research and Development Co., Pittsburgh, Pa.
b From consideration of previously established pattern relationships, isomer matrices were synthesized on a per cent of total ionization basis for each isomer in the carbon number range of c16 to CZ2. The relative total ionization sensitivities of several of the isomers were found to differ slightly from unity, but were constant over the entire carbon number range. Carbon number distributions were calculated on the isotope corrected parent peaks with a constant molar sensitivity. A composite isomer matrix was synthesized using the carbon number distribution and assuming statistical isomer distribution. Typical analyses are presented and compared with gas chromatographic results.
R
in biodegradable detergents has led to shift from detergents derived from highly branched alkylbenzenes made from propylene polymers to those derived from phenyln-alkanes. Because of possible variations of both the detergency and biodegradability of the detergents with the isomer distribution of the alkylate used, the need for a quick, reliable analysis of the isomer distribution of phenyl-nalkanes was obvious. Mass spectrometry is frequently used in the determination of carbon number distribution (3) as well as compound group types (4, 5 ) in similar alkylbenzenes. The incorporation of an isomer distribution determination would, therefore, provide a natural adjunct to the use of the mass spectrometric technique in the analysis of these materials. A mass spectrometric method for determination of isomer distribution from routine spectra is the subject of this paper. It is apparent from examination of available spectra of phenyl-n-alkanes ( I ) , that major fragment ions in the spectra involve the loss of either of the alkyl groups on the alpha carbon atomECENT INTEREST
i.e..
the
+
ions
+
C,H2,+1CHC6H5 and
C,H2,+ lCHC6Hs from CttHzn+i CHCJLm+i C6H5 (where m > n and m n
+ + 1 = 5,the
734
ANALYTICAL CHEMISTRY
number of carbon atoms in the alkane
+
chain). Thus, m/e 105 (CH3CHC6Hs) is characteristic of all 2-phenll compounds, m/e 119 for all 3-phenyl conipounds, etc. Similarly, the 1-phenyl alkanes are characterized by a higher m/e 92 in comparison with the internally substituted isomers and this ion is, therefore, appropriate for calculation of the 1-phenyl isomer. On the basis of available spectra for the phenyl dodecZnes, an analytical method has recently been reported ( 8 ) which, however, only considers the l-phenyl through 6-phenyl isomers. This procedure is clearly inadequate for carbon numbers greater than C1, because it cannot determine 7-phenyl, etc. In addition, spectral differences for the same isomer a t other carbon numbers will cause an analysis based on a single carbon number matrix to be in error for a sample with a wide carbon number range. In order to formulate a n analytical method for full-range material (C16 through Czz is considered here), it is necessary to overcome two problems: to provide a means of reliably estimating spectra of the isomers involved, most of which will not be available in pure form; and to provide a procedure for synthesizing an average matrix, or average spectra for each isomer, from the spectra for each carbon number that can adequately give an analysis for a multi-carbon number mixture. The solution to the former problem is based on our recently completed detailed study of the mass spectral fragmentation mechanisms of the phenyl-n-alkanes (6). To avoid duplication of information given in that paper, we will not attempt to give any more details here than is absolutely necessary to logically develop a method. For further experimental details pertaining to the foundations of some of the general relationships given here between spectra and structure, the reader is referred to that paper. EXPERIMENTAL
The experimental techniques are described in the forthconiing publication ( 6 ) . I Consolidated 21-103C mass
spectrometer was used that was equipped with an all-glass heated inlet system and a CEC hIascot and IBXI card punch to provide spectra that could be transferred to a computer. A computer program to calculate the isomer distribution was used in which standard patterns and sensitivities could be easily varied. In addition to the spectra of pure compounds previously cited ( 6 ) , a number of isomer mixtures were available that had been synthesized from single carbon number n-paraffins or terminal olefins. Thus, not only were single carbon number isomer mixtures available (CI6 through Czz),but variations of the isomer distribution were also represented. A number of these isomer mixtures were also analyzed with a Perkin-Elmer Model 226 gas chromatograph equipped with 150 foot, 0.01-inch i d . , Xpiezon L coated capillary to provide calibration and checks on the relative mass spectral sensitivities of the different isomers. SINGLE CARBON NUMBER MATRICES
Before a matrix applicable to a full carbon number range can be synthesized, it is necessary to obtain reliable single carbon number matrices. On the basis of data and relationships established in the earlier study ( 6 ) , these matrices were synthesized. The following discussion relates to estiniation of the per cent of total ionization for each of the ions used in the matrices for every isomer in the C16-C22 range. The data in the previous study were on a monoisotopic basis and, thus, the interpolated spectra initially derived are monoisotopic and they are then corrected back for isotope contributions so that they may be used directly on raw polyisotopic spectral data. I t was shown ( 6 ) that the fragmentation mechanisms for the l-phenyl and %phenyl conipounds were distinctly different from those substituted two or more carbon atoms from the end of the alkane chain. The latter group of compounds was shown to give an identical curve for the ratio of intensity of
+
+
CnH*,+iCHC6Hs to C , H P ~ +1CHC6Hs when plotted against the relative position of the phenyl group to the position of 1) the center of the chain-i.e., 2(n
+
0
i m c I? 15
contributions, with two exceptions : m/e 105 and 119. It had been previously suggested (6) that these two ions did not fit the scheme for the other ions because of significant retardation of subsequent decomposition for these smaller alkylbenzyl ions. It was found that better agreement was obtained with the pure compound spectra, if m/e 119 was increased 10% above that estimated by the above procedure and if nile 105 was evaluated independently as follows. 2.5% for the 3-phenyl and 4-phenyl isomers (where it is formed
\
-
z 0
:
-
10 9 0 8 4
i
7 6 -
*
E s-
:4 0 +o 3 -
+
Figure 1 .
0
X
40-0'
CIB CI9 c26
+
Absolute intensity
(70total ionization) of C,H2,
+
+ 1-
+ICHCC,H~ and C,H2,
CHC6H5vs. relative position of phenyl group on alkane chain
+
(A' 1). Ident,ical relationships were found to hold equally well for the c 1 6 c 2 2 range as for the pure c 2 6 compounds studied. The absolute intensities of these two primary fragment ions are shown in a similar graph (Figure 1) where the abscissa has been exl)/(-V 1). panded to include 2(m These data were obtainable only from pure conilwund spectra which were primarily phenyl cicosanes. However, the good agreement of the plot of the ratio of the intenqities for different carbon numbers as well at; the continuity of the plot when expanded to
+
+
of these ions rises linearily from a zero value for i E m - 2 to i % 7 where a greater rate of rise of intensity with decreasing value of i is found. Since for the CIB-c22 range, only the isomer range through &phenyl is considered, i 6 7) are m/e 105 through 189 ( I the only alkylbenzyl ions used in the calculation and the behavior of the intermediate ions in this range (i 7) is all that need be considered. Examination of several of the applicable pure compounds suggests that the total contribution of the intermediate carbon numbers for i 7 formed from the
2
0, for i
(5)
where a N ( i , j ) = 0, for i
> N+l 2 , and
where A(i,j) is the combined matrix for the i-th isomer and j-th ion of the list
Comparison of Analytical Results with Calculated Distribution Based on Blends of Single Carbon Number Isomer Mixtures
Cla voi.yo 2 0 . 8 100.0 ClS ... C,,_. C18 CZO CZl
k [ref. ( S ) ]
UNO
r and
Comparison of P-1 Corrections and Sensitivities of Parent Peaks
Carbon
...
...
...
... ... 0 0 8 1 1 0
21 23 25 30
CIS 23.8 1.5 ...
97.3 ... 1.2 ... ...
0 0
17 7 19 1 18 2 23 I 21 9
0 0
Blend I CZz 25.9 29.5 CZO
1.8 ...
95.9 ... 2.3 0 1 15 4 15 6 14 7
18 1
17 9 175 0 7
Blend (calc.) 21.2 0.0
... ...
2.0
...
98.0 0 9 15 8 13 6 12 4 14 6 14 5 142 140
23.6
0.0
25.7
0.0
29.5 0 3 17 4 17 4 17 0 20 8 14 1 8 7 4 3
M.S. (obs.)
20.4 0.2 23.2 0.3 25.9 0.0
30.0 0 0
16 17 16 20 14 9 4
9 2 4 4
8
6 7
CI~ 39.0 100.0 I
.
.
... ...
... ...
... 0.0
21.8 23.1 25.1 30.0
Blend I1 Czz Blend 61.0 (talc.) ... 39.0 ...
0.0 0.0 0.0
...
0.0
, . .
... 2.0 98.0 0.9 15.8 13.6 12.4 14.6 14.5 14.2 14.0
1.2
59.8 0.6
18.1
17.3 17.4 20.6 8.8 8.7 8.5
M.S.
(obs.1
Cls 43.1
37.1 0.3 0.5 0.1 1.3
... 98.9 ...
1.1
0.0
60.7
0.0
17.4 17.1 16.9 20.0 10.0 9.6 9.0
Blend 111 CZZ Blend 5 6 . 9 (calc.) ... 0.5 ... 0.0 ... 42.6 ... 0.0 2 0 11 98 0
0 0
37 20 14 15 12
3 4
5 1
7
0 9
15 8 13 6 12 4 14 6 14 5 14.2 14.0
0 0
55 8 0.5
25 16 13 14 13
i
5 3 8 7
8.1
8.0
VOL. 37, NO. 6, M A Y 1965
M.S.
(obs.)
0.6 0.2 43.1 0.2 1 6 0 0
54 3 n_ 2_
24 4 16 7 14 0 15 5 14 9 7.1 7.2 737
Table V.
0.1 1.0 0.4 0.2 0.2
Cll 0.3 5.2 3.5 3.0 2.8 1.3
1.9 1.4
16.1 14.7
ClS
1-P
2- P 3- P 4- P 5- P 6- ‘P 7-4
Sum M.S.
Comparison of Gas Chromatographic and Mass Spectrometric Results for Several Typical Samples
Sample A (vol. To) G-C results CI8 CIS Czo 0.5 5.0 4.5 3.4 3.7 3.7 20.8 20.6
1.0 11.5 6.8 6.0 6.5 5.0 2.5 39.3 44.0
(Table I), u , ~ ( i , j )the corresponding element of the A‘-th single carbon number matrix, and S, the relative sensitivity of the i-th isomer (Table 11). The denominator corrects for the fact that the higher isomers can only arise from the higher carbon numbers in the mixture. After inversion of the matrix A ( i , j ) , the volume per cent isomer distribution (9,) is determined in the standard manner by matrix multiplication using in the inverse matrix [ B ( j , i ) ] 100
c B(j,i)l(j)
The carbon number and isomer distribution calculated by the above procedures are shown for several test mixtures in Tables IV and V. I n the former, blends of single carbon number mixtures were analyzed and compared with calculated distributions based on analysis of the blend components with single carbon number matrices. These blends thus serve to test the internal consistency of the method and the procedure for obtaining the carbon number distribution and the composite matrix. Because several of these blends cover the entire carbon number range with a highly distorted distribution, these blends represent an extreme test of the method. Some error is noted in these blends for the higher isomers and these should represent the greatest errors expected for complete range material. Far less error will be encountered in the narrower carbon number distributions normally studied. In Table V, typical mixtures are examined and compared with gas chromatographic results. These data are thus a test of the entire procedure on a typical isomer and carbon number distribution as well as a test of the agree-
738
e
ANALYTICAL CHEMISTRY
0.0 5.8 4.6 3.8 4.5 1.6 1.6 21.9 19.3
Sum
11,s. results
1.9 28.5 19.8 16.4 17.7 11.6 4.1
2.3 29.4 18.4 17.4 15.0 12.3 5.2
C17 0.2 0.5 0.1 0 1 0.1
1.0 1.0
Clg 1.4 10.5 4.6 2.9 2.5 2.3 24.2 23.7
ment with the gas chromatographic method. In both cases, the agreement is quite satisfactory. The statistical factors (Equation 5 ) were determined by assuming a statistical isomer distribution. One might expect that samples with markedly different isomer distributions might require an iterative correction procedure to redetermine new uN.,from the initially calculated total isomer distribution. However, the use of b y , , in Equation 6 is only to determine the weighting factor for the same isomer a t different carbon numbers. This procedure therefore reflects the effect of inclusion of the higher isomers a t higher values of 9on the isomer fraction of the lower isomers. It is apparent that even with an isomer distribution significantly different from a statistical one, the calculated relative value of u ~ .us. , A’ (at constant i) will be very similar to that given by Equation 5. Secondly, errors arising from this incorrect estimation of by,, would be expected to have their greatest effect ‘on mixtures with the uidest carbon number range. This conclusion was born out by applying a correction procedure to determine u ~ , from , the isomer distribution expected a t each carbon number using the total isomer distribution initially calculated as above. This procedure was then applied to the samples given in Tables IV and V and the isomer distributions recalculated with the new set of uN,,’s. The samples in Table V, which exemplify typical range encountered, showed maximum changes in the isomer distribution of less than 0.2YG. Although the extreme range blends such as those given in Table IV showed larger changes in the calculated results (as much as 2% change in the 5-phenyl value which is the most sensitive to carbon number change), it is felt that this procedure is
Sample B (vol. 70) GC results CIS Cz0 Sum 1.8 0.3 3.7 5.7 13.9 30.6 10.4 3.7 18.8 8.4 2.6 14.0 3.1 14 1 8.4 7.5 2.7 12.5 3 6 2.7 6 3 54.0 20.8 55 8 19.5
M.S.
results 2.7 33.5 18.7 14.0 13.9 12.0 5.2
generally not necessary for most analyses. That is, Equation 5 is quite adequate for the estimation of the relative weighting factors for a particular isomer at each carbon number for the usual isomer distribution and carbon number ranges encountered. Although the method described in this paper is concerned only with the C:6-C22 range, the procedures for extending the carbon number range are obvious from the above discussion. ACKNOWLEDGMENT
The authors thank R. W. Rosenthal, D. J. Hurley, and W. K. Porter, Jr., for supplying mixtures; the mass Petrocelli analyses.
the single carbon number
P. W. Mazak for obtaining spectral data; and J. A. for the gas chromotographic LITERATURE CITED
( 1 ) Amer. Petrol. Inst. Research Project 44, “Mass Spectral Data,” Carnegie
Institute of Technology, Pittsburgh, 1960, spectra 1742-7. (2) Beynon, J. H., “Mass Spectrometry and Its Application to Organic Chemistrv,” p. 555, Elsevier, Amsterdam, 1960. . (3) Boyer, E. W., Hamming, M. C., Ford, H. T., ANAL.CHEM.35, 1168 (1963). (4) Brown, R. A,, Skahan, D. J., Cirillo, V. A,, Melpolder, F. W., Ibid., 31, 1531 (1959). ( 5 ) Galbraith, F. J., Paper No. 82, Twelfth Annual Conference on Mass Spectrometry and Allied Topics, Montreal, June 7-12, 1964. (6) King, A. B., J . Chem. Phys. 42, in press. ( 7 ) XZlcAdams, K. l)., “Isotope Correction Factors for Mass Spectra of Petroleum Fractions,” Esso Research Laboratories, Baton Rouge, La., 1957. (8) Hiibinfeld, J., Emery, E. M., Cross, H. I)., 111, Spring Meeting, American
Oil Chemists Society, ?Jew Orleans,
April 19-22, 1964.
RECEIVED for review September 16, 1964. Accepted December 21, 1964.
