(Products of Hydrogenation of Carbon Monoxide) AROMATIC HYDROCARBONS W LLIAM E. CrlDY, PHILIP J. LAUNER, AND A. W. WEITKAMP Research Department, Standard Oil Co. (Indiana), Whiting, Ind. Aromatics constitute about 670 by weight of the hydrocarbons obtained in the hydrocarbon synthesis process. They were separated by adsorption on silica gel and by distillation, and w-ere analyzed by ultraviolet and infrared spectroscopy. Benzene is present i n traces, but successive carbon-number fractions contain increasing concentrations of aromatics. Monocyclic, dicyclic, and tricyclic aromatics occur i n proportions that decrease with the number of aromatic rings. Some of the aromatic hydrocarbons contain a nonaromatic ring fused with the nucleus. Olefinic unsaturation is an important feature of the high molecular weight aromatics. Some unsaturation may be in the rings of compounds of the indene or the dihydronaphthalene series; the remainder is side-chain unsaturation. Aromatic rings occur to a small extent in the structures of the oxygenated compounds. I n the benzene series, monoalkyl and ortho-dialkyl isomers predominate. A regular pattern of alliylbenxene isomer distribution suggests a way i n which the rings may have been formed.
A
ROMATIC hydrocarbons comprise a minor proportion of the hydrocarbons produced by the hydrogenation of carbon monoxide. Neveitheless, a knowledge of their structures contributes to interpretation of the fragmentary information on the structures of the closely related alicyclic components. As early as 1929, Fischer and Koch (6) found O.lY0 benzene and 0.4 to 0.5Y0 toluene in a naphtha cut. However, Eldus and ZelinskiI ( 5 ) found no evidence of benzene or toluene in a product from a nickel-cobalt catalyst. The present study is concerned with products from iron catalysts in fluidized bed reactors. Analyses for aromatic hydrocarbons in similar products were reported by Rruner ( 1 ) and Clark ( 2 ) . I n analyzing the aromatic hydrocarbons consideration wa? given t o distribution by carbon number, t o the eight-carbon and nine-carbon isomer distributions, to structures of the aromatics in the CIS to C30 range, and to aromatic structures in CI6to C30 oxygenated compounds. The analytical results have been interpreted in terms of a relation between the structures of aromatic and aliphatic hydrocarbons, and a schemehas been developed for predicting the distribution of alkylbenzene isomers.
DISTRIBUTION BY CARBON NUMBER. The presence of benzene and higher homologs is demonstrated in Figure 1 by the succession of maxima in the plot of the refractive indices of successive cuts from distillation of liquid hydrocarbon product. The proportions of the t'hree principal classes of hydrocarbonsparaffins, olefins, and aromatics-were determined by a micropercolation technique (3)using fluorescent dyes to mark paraffinolefin, olefin-aromatic, and aromatic-alcohol interfaces in columns packed with silica gel. Above C14the micropercolation technique failed t o develop an olefin-aromatic interface. The CIS to C30 aromatics were determined on a larger scale by adsorption on silica gel. After elution of the nonaromatic hydrocarbons with pentane, the aromatics were desorbed and weighed. The yields and compositions of the individual carbon-number fractions are presented in Table I together with aromatic contents from the literature. From these results it wasestimated that about 6c7, by weight of the total hydrocarbon product was aromatic. ISOMER DISTRIBUTION I N EIGHT-CARBOK AND NINE-CARBON FRACTIUZS. The three xylene isomers and ethylbenzene were determined in the 124' to 145OC. distillate by ultraviolet absorption. The analysis presented in Table I1 shows t'hat ethylbenzene predominates. The nine-carbon alkylbenzenes were separated from the 145' to 178" C . distillate by adsorption on silica gel, and then were fractionally distilled. The minor comp0nent.s in each cut were determined by infrared, according t o the base-line method of \$:right ( 1 0 ) ; the major component was determined by difference. The proportions of the eight, isomers are shown in Table TI. n-Propylbenzene is the most abundant isomer. AROMATICSTRUCTURES I N HIGHER HYDROCARBON FRACTIONS. Distributions of ten-carbon and higher alkylbenzene isomers cannot be det,ermined by available methods. Molecules containing aromatic rings condensed with nonaromatic rings are particularly intractable, especially if olefinic unsaturation is present. However, complex mixtures can be simplified by means of silica gel percolation and fractional distillation, and functional groups in complex molecules can be identified spectroscopically. O Application of these techniques to the CIS to C ~ aromat'ics, before and after hydrogenat,ion, provided clues to the types of structures present. Silica gel percolation of the aromatics, before hydrogenation, gave a series of cuts whose refractive indices and mole percent-
ANALYSIS
Liquid hydrocarbon product was fractionally distilled to segregat'e aromatic hydrocarbons differing by one carbon atom. The aromatic hydrocarbons were separated from nonaromatic components-and sometimes determined quantitatively-by adsorption on silica gel. Individual isomers were determined by use of the Beckman Model DU ultraviolet spectrometer and the Beckman IR2 infrared spectrometer. Functional groups in the more complex molecules were analyzed spectroscopically. Total unsaturation was determined bromometrically ( 4 ) .
TABLEI.
HYDROCARBON DISTRIBUTIOX BY CARBONNUMBER A N D CLASS IN L%ROMATIC RANGE
Carbon Number 6 7 8 9 10
Yield, Wt. yo Hydrocarbons 8.6 6.6 5.0 3.9 3.0 2.4 1.8
11 12 13 14
Residue
1.4 1.1 8.2
Sample B-1 Distribution, Volume % ' Paraffins Olefins Aromatics 12.8 86.9 0.3 12.4 83.9 3.7 13.2 81.4 5.4 15.0 77.5 7.5 15.1 72.7 12.2 15.4 73.6 11.0 16.2 68.2 15.6 15.4 62.9 21.7 16.5 63.6 19.9 .. .. 34u
Aromatics Clark (21, Brunei- ( I ) . 701. 70 wt. % 0.2 2 2.1 5 6.0 6.2 7.2 5.8 3.6
...
..
...
42.0 a
Calculated from 39 wt.
350
determined gravimetrically in
R
THE
similar material ((3-4).
INDUSTRIAL AND ENGINEERING CHEMISTRY
February 1953
TABLE11. COMPOSITION OF EIGHT-AND NINE-CARBON AROMATIC FRACTIONS
Component Eight-carbon isomers, mole % Ethylbenzene o-Xylene rn-Xylene p-Xylene Nine-carbon isomers, mole % n-Propylbenzene Isopropylbenzene I-Methyl-2-ethylbenzene 1-Methyl-3-ethylbenzene 1-Methyl-4-ethylbenzene 1 2 3-Trimethylbenzene 1:2:4-Trimethylbenaene 1,3,5-Trimethylbenzene Identification not positive.
Experimental
Thermodynamic Equilibrium at 600' K. ( 8 )
46 29 18 7
21.6 50.1 22.4
39 2 19 23 9 la 6
1
5.9
indicated the presence of diolefins. Unsaturation varied with molecular weight, and confirmatory evidence of diolefins resulted from examination of carbon-number fractions obtained by distillation. Unsaturation ranged from 0.8 double bond per molecule in the C16 fraction to about 1.3 in CZZand higher fractions. Thus, some molecules map contain both unsaturated rings and unsaturated side chains.
1.2 0.6 4.0 12.3 8.2 10.1 45.9 17.7
ages of unsaturation are pletted in Figure 2. Unsaturation, as CISmono-olefins, was calculated from bromine numbers which were determined by an electrometric titration procedure ( 4 ) designed t o give a minimum of side reactions. Infrared analyses for olefin types were made on selected cuts. Absorption bands at 10.0 and 11.0 microns were used to estimate nontertiary terminal (RCH=CH2) double bonds, and the band at 10.35 microns was used to estimate trans-internal (RCH=CHR') double bonds. Hydrogenation of the recombined aromatics to eliminate olefinic unsaturation was carried out over Raney nickel at 70" C. and 70 atmospheres. Infrared analysis showed that all terminal double bonds and about 7oy0 of the trans-internal double bonds were hydrogenated. Percolation through silica gel gave the refractive index curve plotted in Figure 2. Except for a slight lowering of the maximum from 1.63 to 1.62, the curve is very similar to that obtained before hydrogenation. Accordingly, aromatic structures were little affected by hydrogenation. The concentrations of the monoalkylbenzenes in selected cuts of the hydrogenated sample were estimated from the intensity of the 14.3-micron band in theinfrared spectraandare plotted in Figure2. Although resolution of the aromatic types was incomplete, the analytical results plotted in Figure 2 and additional data on selected cuts indicated the types of structures present. The specific types which tended to concentrate a t various index ranges are discussed in the following paragraphs : 1. Refractive Index 1.45 to 1.48. Unsaturation in excew of 100% reflected the presence of aliphatic diolefins (9),which were separated, with difficulty, from alkylbenzenes. The paraffins resulting from hydrogenation of the olefins were easily separated by percolation. 2. Refractive Index 1.48 to 1.52. Unsaturation passed through a minimum and the concentration of monoalkylbenzenes exceeded 15y0, Absence of an absorption band a t 13.15 microns indicated that the side chain on the monosubstituted benzene ring was not branched adjacent to the ring. Bands indicative of ortho, meta, para, and other classes of polyalkyl substitution were observed. Some of the side chains may have been unsaturated. 3. Refractive Index 1.52 to 1.54. A pronounced weakening of 14.3-micron absorption demonstrated the separation of certain benzene derivatives from the monoalkylbenzenes. These molecules had higher indexes than known alkylbenzenes and must have contained a five- or six-membered ring fused with the benzene nucleus. Unsaturation rose toward one double bond per molecule. Infrared determination of olefin types accounted for one fourth of the unsaturation as nontertiary terminal (RCH=CH2) and one tenth as trans-irrternal (RCH=CHR'). Even after liberal allowances for cis-internal and tertiary double bonds, it appeared that 25 to 50% of the unsaturation was in rings. 4. Refractive Index 1.53to 1.57. The fused-ring derivatives of benzene became increasingly unsaturated in this portion of the chromatogram. The maximum unsaturation of 113%
351
210
I .44
I70
1.42 N c
0
0'
w
0
k
5 0 a
-
c
X
c3
1.40
130
z w L
GU
(9
z
K
2
LL
0 90
1.38
50
I.36
m
20
Figure 1.
40
60
VOLUME % DISTILLED Distillation of Liquid Hydrocarbons
Numbers on curve indicate carbon number
5 . Refractive Index 1.57 to 1.61. Naphthalenes were identified by their characteristic ultraviolet absorption maximum at 3230 A. The presence of dicyclic aromatics containing two benzene rings, one or both of which were sometimes monosubstituted, was suggested by infrared absorption at 14.3 microns. Unsaturation diminished in this range, but the relative proportion of terminal and trans-internal types was about the same as in the 1.52 to 1.54 range. 6. Refractive Index 1.62 to 1.63. Small amounts of tricyclic aromatics were present. Ultraviolet absorption suggested that some molecules had phenanthrene ring systems. Anthracene ring systems could not be detected by ultraviolet absorption in the presence of phenanthrene derivatives. Absorption a t 14.3 microns persisted in the tricyclic range. The alternative that 14.3-micron absorption in the polycyclic aromatics might be due to structures other than phenyl groups was considered. Hydrogenation of selected cuts eliminated the 14.3-micron band as well as other strong bands at 11 to 15 microns characteristic of aromatic structures. Absorption at 11.22 microns, which was observed after hydrogenation, is characteristic of cyclohexyl groups. The proof that these cyclohexyl groups were formed by hydrogenation of phenyl groups was not rigorous, because the relatively weak 11.22-micron band may have been present in the spectra of unhydrogenated samples but may have been obscured by interference from strong neighboring bands due to aromatic structures. AROMATIC STRUCTURES IN OXYGENATED COMPOUNDS. Some of the carbon skeletons of hydroxyl, carbonyl, and carboxyl derivatives contained aromatic nuclei. The presence of aromatic structures in the hydroxyl and carbonyl compounds was suggested by a small refractive index maximum at the end of a silica gel
352
INDUSTRIAL AND ENGINEERING CHEMISTRY
Vol. 45, No. 2
t o the growing end of the chain with a probability of 0.9 and to the carbon adjacent to the end with aprobability of 0.1, restricted only by the nonoccurrence of quaternary structures. Rings of five or six carbons are formed when, instead of a single entering carbon, a carbon already present in the chain is joined to the growing end of the chain. Hoog ( 7 )showed that the configuration of the chain adjacent to the reacting carbon greatly affects its reactivity. Thus, terminal (primary) carbons react less readily than nonterminal (secondary) carbons. Tertiary carbons are unreactive but enhance the reactivity of adjacent primary or secondary carbons. These pifects of structure may explain the increasing yields of aromatics with inWEIGHT X WEIGHT % creasing molecular weight, observed in BEFORE H Y D R O G E N A T I O N AFTER H Y D R O G E N A T I O N the present work. Thus, the formation of benzene requires the difficult Figure 2. Separations of CISt o C30 Aromatics on Silica G e l end-carbon closure involving only the unbranched six-carbon chains, which comprise about 73% of the six-carbon aliphatic structures. The product (9) and was conchromatogram of acid-free C16to 27% with branched structures cannot form six-membered rings. firmed by infrared absorption. Toluene may be formed from all seven-carbon structures, Analysis for aromatic structures in the carboxyl derivat,ives except those which are dimethyl branched (3%). The unbranched included exhaustive extraction of nonacid constituents and fractional distillation of methyl esters of the purified acids. seven-carbon chain (66%) may be cyclized by two routes; Refractive indices of successive carbon-number fractions were the four monomethyl-branched structures (31%) may be cyclized increasingly higher than those of analogous aliphatic esters only by closure between the end carbons. with equivalent unsaturation. Infrared spectra of selected To predict isomer distribution in the eight-carbon and higher fractions contained bands characteristic of monosubstituted, alkylbenzenes requires that the probabilities, 0.9 and 0.1, for the terminal and adjacent-to-end growth reactions be merged disubstituted, and more highly substituted benzene rings. The extent and character of monosubstitution mere further examined in a 16-carbon ester fraction, fig, 1.4531. The intensity of the 14.3-micron band indicated that' about 1.5% of the molecules T ~ 111. ~ SLO U R ~ C ~ O~F E ~ ~ ~ LknowAT~c ~ ~ -I c ~~ contained a phenyl group. The absence of an absorption band Aliphatic Precursors Isoiner Resulting from CloEure at 13.15 microns indicated that the side chain on the monosubstiAdjaoentCalculated Terminal to-end distribution, tuted benzene ring was not branched adjacent t o the ring. mole % Structure closure closure INTERPRETATION
The concentrations of aromatic hydrocarbons in successive carbon-number fractions increase almost linearly with carbon number; the rate of increase averages about 2.5YG per carbon atom in the 6- to 14-carbon range. The accompanying decrease in olefin contents may not be entirely a mathematical consequence inasmuch as the increasing aromatic to olefin ratios are paralleled by a similar but less pronounced increase in paraffin to olefin ratios. This trend suggests that aromatics, like some paraffins (9), may be fornied a t the expense of olefins. The observed distributions of the C8 and C Q alkylbenxene isomers presented in Table I1 differ widely from thermodynamic equilibria. In these fractions as well as in higher molecular weight fractions, the thermodynamically unfavored monoalkylbenzenes greatly predominate. Absence of branching adjacent to the monosubstituted ring excludes alkylation of benzene with olefins. Wide divergence from equilibrium isomer distribution suggests that alkyl groups have not migrated about the ring. High yields of monoalkylbenzenes would not be expected from catalytic cyclization of olefins, in view of Hoog's finding ( 7 ) that chromia catalysts yielded ortho-substituted aromatics. PREDICTION OF ISOMER DISTRIBUTION. Aromatic-isomer distribution can be predicted with fair accuracy by assuming that aliphatic olefins and aromatic structures are derived from common precursors. A scheme has been proposed (9) for predicting distributions of aliphatic isomers, wherein an entering carbon atom is added
C-7-6-C-C-C--2---1
Ethylbeneene
6.6
C-6-C-C-C-C-1
m-Xylene
...
7.3
7-6-C-C-c-2-1
p-Xylene
o-Xylene
m-Xylene
m-Xylene
XI. 0
I
+Xylene (I
c C
7.3
?-6-C-C-C--2--1
7.3
7-6-C-C-C--2-1
cI
o-Xylene, ethylbenzene
I
6
7.3 0.8
7-C-c-c---c-2-c I 7 6-C-C-C-C-1
l
C 0.8
6-C-C-C-C-1
I
e-c-c-c-c--1 I
6 0.9
6-C-C-C-C-1
I
i
1
o-Xylene
o-Xylene
. ..
m-Xylene
...
p-Xylene
...
o-Xylene
...
m-Xylene
...
C
I
C
0.3
...
C I C
c c 0.9
6-C-C-C-C-1
0.9
6I C 6-C-C-C-C-1
I
6
0 1 I
1oo.o
I
1
C
c-c-c-c-c cI cl cl
p-Xylene
o-Xylene
...
Q
. ..
5 Closures which would require alteration of the carbon skeleton t o yield aromatics a m excluded.
~~ ~
I N D U S T R I A L A N D E N G I N E E R I N G CHEMISTRY
February 1953
TABLE IV.
DISTRIBUTIONS OF EIQHFAND NINE-CARBON ALKYLBENZENES Predicted Adjacentto-end closure
Terminal closure
Component Eight-carbon isomers, mole % Ethylbenzene 67.7 o-Xylene 6.7 nz-Xylene 16.8 p-Xylene 8. 8 Sine-carbon isomers, mole % n-Propylbenzene 61.9 Isopropylbennene 8.1 1-Methyl-2-ethylbenzene 3.6 1-Methyl-3-ethylbeneene 13 7 1-Methyl-4-ethylbenzene 7 2 1 2 3-Trimethylbenzene 1.1 1'2'4-Trimethylbenzene 3.6 1:3:5-Trirnethylbeneene , 0.8 a Identification not positive.
0 83.4 8.3 8.3
