See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

Panorama General Motors Proving Ground

Downloaded via RMIT UNIV on April 21, 2024 at 02:28:24 (UTC).

of

Knocking Characteristics of Aromatic

Hydrocarbons
Wheeler G. Lovell, John M. Campbell, Frank K. Signaigo, and T. A. Boyd
General Motors Research Laboratories, Detroit, Mich.

publications The relative tendencies to knock in an engine given in terms of the octane
from this laboratory have have been measured for fifty-nine hydrocarbons
numbers of the 20 per cent
PREVIOUSdescribed the knocking
embracing the aromatic series. These measure-
blends. Further data in terms of
characteristics of over one hun- compression ratio have been re-
ments were made not on the hydrocarbons alone, ported by Hofmann and others
dred paraffin, olefin, and naph-
thene hydrocarbons of varied but in admixture with gasoline; the results have (7, 8) for benzene, toluene,
structure. This paper relates to been expressed, as in previous work, using the xylenes, ethylbenzene, styrene,
a continuation of the previous antiknock effect of aniline as the standard of phenyl acetylene, cyclopenta-
work and is concerned in par- diene, and dicyclopen tadiene.
comparison. The data given in the present
ticular with the behavior of
aromatic hydrocarbons. Upon this basis there appear great differences paper include all but one of the
During the past decade, data among the knocking properties of these com- hydrocarbons previously re-
on the relative knocking char- pounds. Certain relationships between molecular ported upon and constitute in
acteristics of some of the more structure and the knocking properties of these general a confirmation of the
common aromatic hydrocar- published data as well as a con-
compounds are consistent with similar relation- siderable extention of these data.
bons have gradually been ac-
In 1921 Ricardo ships previously observed among paraffins, olefins,
cumulating. Evaluation of Knocking
(16) reported the relatively high
and naphthenes.
antiknock properties of ben- Properties
zene, toluene, and xylene with respect to certain naphthenes As in previous work from these laboratories (11, 12, 13, 17)
and paraffins. In 1922 Midgley and Boyd (14) reported upon the tendencies of these hydrocarbons to knock have been com-
benzene, toluene, and xylene in solution with kerosene. They pared in a single-cylinder, variable-compression engine which
found that the antiknock effect of these three hydrocarbons was equipped with an evaporative cooling system and fitted
increased in the order of increasing molecular weight. In with a bouncing-pin indicator for matching the fuels in re-
1929 Birch and Stansfield (2) reported data on benzene and spect to antiknock quality. This equipment and the general
1,3-cyclohexadiene. Since that time Howes and Nash (9) method of operating it have been described in previously pub-
have published data on twelve aromatic hydrocarbons in 20 lished work (4).
per cent solution with gasoline and have expressed their re- The relative knocking tendencies under these conditions
sults in terms of the concentration of tetraethyllead required have been evaluated in terms of aniline (a knock suppressor),
in the reference gasoline to match the 20 per cent blends, and the unit of this evaluation has been called the “aniline
More recently Garner and Evans (5) have reported upon some equivalent.” A positive aniline equivalent indicates that
of the aromatics in 20 per cent solution with gasoline and the compound knocks less than the reference gasoline and rep-
under different sets of engine conditions. Their data are resents the amount of aniline, expressed as the number of
475
 476 INDUSTRIAL AND ENGINEERING HE I S TRY Vol. 26, No. 5

basis furnishes a measure of the partial knock

effect of each compound in gasoline at concen-
trations which are comparable with those in
which such a compound might actually be
present in commercial fuels.
In order to convey some conception of the
magnitude of this scale of knock rating in units
which may be more generally familiar than
aniline equivalents, a compound having an
aniline equivalent of 20 added in one-molar
concentration to the reference gasoline hav-
ing an octane number of 55 would produce a
mixture having an octane number of 65.
This numerical ratio appears to be valid over
the range of positive aniline equivalents used,
but a distinction should be clearly recognized
between aniline equivalents obtained at one-
molar concentration in gasoline and the be-
havior of compounds by themselves or not in
admixture with gasoline.
At concentrations below about 40 per cent
by volume the aniline equivalent computed
from concentrations other than one mole per
liter has been found to be substantially inde-
pendent of concentration for a number of
compounds investigated. But the aniline
Figure 1. Relative Knocking Characteristics of Aromatic equivalent as computed at concentrations
Hydrocarbons below 40 per cent by volume has been found
to be not necessarily an index to the behavior
centigram-moles per liter, which must be added to the refer- of the compound in greater concentrations with gasoline, or
ence fuel to produce a fuel that is equivalent in tendency to even in the pure state, and so extrapolation of values ob-
knock to a one-molar solution of the compound in the refer- tained at low concentrations to 100 per cent is not, as a
ence gasoline—that is, to a solution containing one gram-mole general rule, recommended.
of the compound made up to a volume of 1000 cc. with the The numerical values of the aniline equivalents reported in
reference gasoline. A negative aniline equivalent indicates this investigation depend somewhat upon experimental con-
that the compound knocks more than the reference gasoline ditions and, of course, upon the reference fuel. It is also rec-
and represents the amount of aniline, expressed again as cen- ognized that the relative values of different aromatic hydro-
tigram-moles per liter which must be added to the molar solu- carbons may change with engine conditions, and some data
tion of the compound in the reference gasoline to make it in this connection are presented.
equivalent in tendency to knock to the reference gasoline.
For example, the aniline equivalent of benzene (molecular
weight, 78) is 10. This means that 78 grams of benzene made
up to a liter with gasoline (about a 9 per cent solution of the
hydrocarbon by volume) is equivalent in knock to a liter solu-
tion of aniline (molecular weight, 93) in gasoline containing
10/100 X 93 grams of aniline (about 0.9 per cent solution by
volume). Similarly, the aniline equivalent of —3 for n-
heptylbenzene (molecular weight, 176) means that 176 grams
of n-heptylbenzene, and 3/100 X 93 grams of aniline made
up to a liter with gasoline knocks the same as the reference
gasoline. This makes the solution contain about 20 per cent
n-heptylbenzene and 0.3 per cent aniline by volume.
The precision of the measurements obtained in this way
has been found to be usually within one unit of aniline equiva-
lent for compounds upon which two or more determinations Figure 2. Effect of Unsaturation in the
have been made. Those compounds upon which check de- Side Chain
terminations have been made with different samples are so
indicated in the tables. Most of these hydrocarbons have The use of a molecular basis for evaluating the knocking
been available in limited quantities, so that only one determi- properties of the compounds is convenient because of the con-
nation of the knocking tendency was possible, and, although sistent relationships between molecular structure and knock
one determination represents the average of four or five sepa-
rating which then become apparent. It is the main purpose
rate bouncing-pin readings, the precision is possibly less in of this paper to present and to discuss some of these relation-
some cases than the figure given, owing to unavoidable varia-
ships.
tions in engine conditions during the period in which the meas-
urements were being made. Sources of Hydrocarbons
The aniline-equivalent method of expressing knock ratings Most of the hydrocarbons used in this investigation were
provides a convenient and uniform basis for comparing com- synthesized in this laboratory. A few were obtained from the
pounds representing a wide range in tendency to knock. This Eastman Kodak Company and elsewhere, and those com-
May, 1934 INDUSTRIAL AND ENGINEERING CHEMISTRY 477

The three acetylene hydrocarbons were prepared by the

pounds which weresuspected to be impure were purified by methods of Bourguel (S), and the trimethylphenylallene was
fractional distillation. In general, methods of synthesis were
prepared from bromobenzene and mesityl oxide (10).
chosen which were known to give products free from isomers
and which would not yield any by-products that could not be The data are tabulated in Table I. The molecular weights
removed by chemical means. and densities are also given in order to facilitate conversion
The methylalkylbenzenes, with the exception of the di- of these values to other concentrations. Figure 1 is a graphi-
methyl-, trimethyl-, and diethylbenzenes and p-cymene, were cal presentation of the alkyl benzenes in which aniline equiva-
prepared by the alcohol-olefin-paraffin method which has lents are plotted approximately against molecular weights.
previously been described From this chart, relation-
for the preparation of alkyl- ships between molecular
cyclohexanes (17). This structure and knocking
preparation can be outlined tendency may be more
as follows: easily visualized than from
the tabulation. Figure 2
An aldehyde or ketone is shows the relationships be-
added to the Grignard re- tween several saturated and
agent prepared from a bromo- unsaturated alkyl benzenes,
toluene. The addition prod- and Figure 3 a comparison
uct on hydrolysis yields a
secondary or a tertiary alco- of some five- and six-car-
hol which, on heating with bon atom compounds in dif-
iodine, is dehydrated to an ferent states of hydrogena-
olefin. This olefin is hydro- tion.
genated to the methylalkyl-
benzene with the use of a
platinum oxide catalyst at Discussion of Results
room temperature and under
a pressure of 2 to 3 atmos-
The data as shown,
Section of General Motors Proving Ground especially in Figure 1, illus-
pheres of hydrogen using
ethyl alcohol as the solvent, trate a number of interest-
under which conditions the benzene ring is not hydrogenated. ing relationships between knocking tendency and molecular
The o-, m-, and p-bromotoluenes were prepared by the Eastman structure.
Kodak Company from the corresponding toluidines by use of
the diazonium reaction. Addition of Straight Side Chains to Benzene. Up to
sec-Butylbenzene was prepared by this method from bromo- n-propylbenzene there is a progressive increase in aniline
benzene and methyl ethyl ketone, and also cyclohexylbenzene
from bromobenzene and cyclohexanone. equivalent with increasing length of the side chain. Further
The olefins listed in Table I, with the exception of phenyl- increase in the length of the side chain results in progressive
ethylene, were prepared by the first two steps in the above lowering of aniline equivalent as shown by the heavy line in
synthesis. Figure 1. ro-Heptylbenzene was a slight knock inducer in
Cyclopentadiene was prepared by depolymerization of di- the reference gasoline. Thus it appears as if, after the alkyl
cyclopentadiene. The two fulvene hydrocarbons were obtained side chain has reached a certain length, its lower antiknock
from cyclopentadiene by condensing it with acetone and aceto-
phenone (18). character begins to weigh against the antiknock effect of the
1,3- and 1,4-cyclohexadiene were prepared by removal of benzene nucleus, and further increases in length of the alkyl
hydrobromic acid from 1,2- and 1,4-dibromocyclohexane, re- side chain finally overcome the antiknock effect of the benzene.
spectively ( , 6). 1-Phenyl-1,3-butadiene was prepared from Addition of Methyl Groups. Successive addition of
methyl iodide and cinnamic aldehyde as prescribed by Muskat
(15). methyl groups to benzene results in a progressive increase in

Table I. Aniline Equivalents

Mol. Aniline Mol. Aniline
Compound Wt. ¿20 Equivalent Compound Wt. d20 Equivalent
FIVE-CARBON ATOM RING SIX-CARBON ATOM RING (cont’d)
Cyclopentadiene 66 0.803 34 n-Butylbenzene 134 0.862 20°
Dicyclopentadiene 132 0.978 65 Isobutylbenzene 134 0.855 20
Dimethylful vene 106 0.885 61 sec-Butylbenzene 134 0.860 21°
Methylphenylfulvene 168 1.021 36 tert-Butylbenzene 134 0.867 25°
SIX-CARBON ATOM RING
1,2-Methyl-n-propylbenzene 134 0.880 19
Benzene 78 0.878 10° 1,3-MethyI-n-propylbenzene 134 0.864 25
1,3-Cyclohexadiene 80 0.828 39 l,4-Methyl-7i-propylbenzene 134 0.861 25
1,4-Cyclohexadiene 80 0.842 29 1,2-Methylisopropylbenzene 134 0.885 20
Toluene 92 0.866 15° 1,3-MethyIisopropylbenzene 134 0.865 26
1,4-Methylisopropylbenzene 134 0.857 27
Phenylacetylene 102 0.930 5°
Phenylethylene 104 0.903 20 1,3-Diethylbenzene 134 0.860 30
Ethylbenzene 106 0.868 19 1,4-Diethylbenzene 134 0.865 34
o-Xylene 106 0.879 17° 1-Phenyl-l, 3-butadiene 140 0.928 33
wi-Xylene 106 0.865 23° n-Amylbenzene 148 0.862 17
p-Xylene 106 0.861 26° ferZ-Amylbenzene 148 0.866 23°
3-Phenylpropyne 116 0.939 8 1,2-Methyl-n-butylbenzene 148 0.874 17
1-Phenylpropyne 116 0.939 11 1,3-Methyl-n-butylbenzene 148 0.869 21
Indene 116 1.006 21a 1,4-Methyl-n-butylbenzene 148 0.860 25
p-Tolylethylene 118 0.896 30 Trimethylphenylallene 158 0.922 36
n-Propylbenzene 120 0.864 24
Isopropylbenzene 120 0.865 22° 1,2-Methyl-n-amyIbenzene 162 0.874 14
1,3-MethyI-n-amyIbenzene 162 0.861 14
1,2-Methylethylbenzene 120 0.883 15 1,4-Methyl-n-amylbenzene 162 0.857 18
1,3-MethylethyIbenzene 120 0.868 22 1,2-Methylisoamylbenzene 162 0.869 16
1,4-MethyIethyIbenzene 120 0.863 27 1,3-Methylisoamylbenzene 162 0.858 20
Mesitylene 120 0.863 31° 1,4-Methylisoamylbenzene 162 0.858 26
Pseudocumene 120 0.870 20
Phenylcyclohexene 158 0.986 26
2-o-Tolylpropene 132 0.892 25 Phenylcyclohexane 10 0.948 -7
2-m-ToIylpropene 132 0.899 37 Diphenylmethane 168 1.006 26
2-p-T olylpropene 132 0.898 37 n-Heptylbenzene 176 0.867 -3
1-p-Tolylpropene 132 0.900 29
Tetrahydronaphthalene 132 0.975 19 °
Two or more complete determinations.
478 INDUSTRIAL AND ENGINEERING
aniline equivalent. Thus toluene is higher in antiknock
quality (molecule for molecule) than benzene, the xylenes are
higher than toluene, and mesitylene is higher than the xylenes.
The effect of the addition of methyl groups to an alkyl ben-
zene, in general, is not so simple. Thus the addition of methyl
to n-propylbenzene may result in a slight increase in the ani-
line equivalent, in a decrease, or in substantially no change,
depending upon the position of the added methyl group. A
somewhat similar situation also prevails for the introduction
of a methyl group to butyl- and amylbenzenes.
Position Isom-
erism. The effect
of position isomer-
ism is illustrated by
the xylenes. Here
p-xylene is higher
than m-xylene, and
m-xylene is higher
than o-xylene. This
relationship be-
tween the ortho,
meta, and para
positions holds
in general for the
higher alkyl methyl-
benzenes, as is in-
dicated by the
Figure 3. Comparison of Five- spread between the
and Six-Carbon Atom Ring dotted lines in Fig-
Compounds ure it appears
1;
also that the effect
of position isomerism on aniline equivalent may be very
great.
Branching Side Chains. Branching of the side chains
containing more than three carbon atoms resulted in an ap-
preciable increase in aniline equivalent. It may be that the
effect of branching is greater when the alkyl group contains a
greater number of carbon atoms; and such an effect would be
consistent with the conception previously expressed, that the
antiknock effect of an alkyl benzene appears to be a balance
between the effect of the ring and the effect of the side chain.
The differences between aniline equivalents of the isomeric
butylbenzenes is considerable, and a relatively small amount
of branching of the compounds containing the amyl groups
appears to have a great effect on antiknock quality. The
general effect of branching appears consistent with the pre-
viously observed behavior of the paraffin and naphthene hy-
drocarbons.
Distribution of Carbon Atoms among Side Chains.
Among compounds of given molecular weight, the aniline
equivalents vary with the distribution of the carbon atoms in
the side chains. It appears as if, in general, the more sym-
metrical and centralized the molecule, the higher the aniline
equivalent, although this cannot be estimated quantitatively.
Thus, among the compounds having a molecular weight of
134 the aniline equiyalents varied between 20 and 34. In
terms of octane numbers this means that the octane numbers
of the blends, in the concentrations at which the aniline equiva-
lents were determined, which was 1 gram mole per liter, or
about 15 per cent by volume for these compounds having a
molecular weight of 134, varied between 65 and 72; that is,
the increase in octane number accompanying these additions
varied between 10 and 17.
Unsaturation in Side Chains. Figure 2 illustrates the
effect of unsaturation in the side chain. For these few com-
pounds a double bond in the side chain adds slightly to the
aniline equivalent, and a triple bond in the side chain causes
a decided reduction in aniline equivalent with respect to the
corresponding saturated alkyl benzene.
CHEMISTRY Vol. 26, No. 5
Relation to Other Types of Hydrocarbons
Compared with data previously reported in these investi-
gations {11, 12, 18, 17), the aromatic hydrocarbons as a class
are higher in antiknock quality than either the paraffins, the
olefins, or the naphthenes, although there are individual com-
pounds in the latter groups which are closely comparable to
the aromatics. Comparing individual aromatic compounds
with the corresponding naphthenes, it is clear that hydrogena-
tion of the benzene nucleus materially lowers the aniline
equivalent. This decrease in aniline equivalent with hydro-
genation of the alkyl benzenes increases with molecular weight
so that there is a much greater difference between mesitylene
and the corresponding naphthene, for example, than between
benzene and cyclohexane.
In Figure 3 a comparison is made between the five- and the
six-carbon ring compounds in various degrees of hydrogena-
tion and some of the corresponding six-carbon open-chain
compounds. This figure shows the comparatively powerful
antiknock effect that is sometimes exerted by a conjugated
double bond as it is found in cyclopentadiene, 1, 3-cycIohexa-
diene and 2,4-hexadiene. This antiknock effect seems to be
characteristic for certain types of conjugation and was most
strikingly shown by dimethylfulvene, which has an antiknock
value 0.6 of that of aniline. This was the highest aniline
equivalent observed among 180 hydrocarbons, with the excep-
tion of dicyclopentadiene which may be considered a polymer
of cyclopentadiene.
Susceptibility to Engine Conditions
Changes in
The aniline equivalents which have been reported thus far
were obtained under conditions of engine operation equivalent
to those specified in the method of knock rating now gener-
ally known as the Research method {19). Aniline equiva-
lents determined by the Motor method {1,19) now in general
use sometimes vary appreciably from those made by the Re-
search method of knock rating. The essential difference be-
tween the Research and the Motor methods is that the latter
procedure is carried out at a higher speed and at a higher mix-
ture temperature. The aniline equivalents of a few typical
hydrocarbons representing aromatics, naphthenes, olefins,
increase: in octane number/gm. mol./liter
Figure 4. Comparison of Aniline Equiva-
lents with Other Published Data
and paraffins, as determined by the two methods of test, are
compared in Table II. These data indicate that the aromatic
hydrocarbons as a class are more susceptible to changes in
engine conditions than the other classes of hydrocarbons.
This is in agreement with the conclusions reached by Garner
and Evans (5) who have made similar comparisons in an Ethyl
Gasoline Corporation Series 30 engine at jacket temperatures
of 212° and 300° F. (100° and 148.9° C.), respectively. The
data in Table II also present some evidence which indicates
that the relative antiknock properties of two hydrocarbons,
even of the same class, may vary somewhat with widely vary-
May, 1934 INDUSTRIAL AND ENGINEERING CHEMISTRY 479

ing engine conditions. Thus 1,3-diethylbenzene was approxi- From these octane numbers the increase in octane number
mately equivalent to mesitylene by the Research method, per gram mole per liter has been calculated and plotted
but by the Motor method the diethylbenzene was definitely against the corresponding aniline equivalent in Figure 4.
higher in antiknock value than mesitylene. In connection Similarly the data of Garner and Evans (5) obtained in 20
with this discussion of aniline equivalents as determined by volume per cent solution have been converted to a molecular
the two methods of test, the octane number calibration for basis and plotted in Figure 4. For the majority of the com-
aniline in the reference gasoline is not changed between one pounds, agreement is well within the limits of experimental
method and the other. error. Lack of good agreement among these different groups
of data in the relative antiknock effects of these hydrocarbons
Table II. Comparison of Aniline Equivalents Determined is shown by the following compounds: isopropylbenzene, n-
under Different Test Conditions butylbenzene, see-butylbenzene, ieri-amylbenzene, mesitylene,
.-Aniline Equivalent-. and p-cymene. After this comparison was made, the aniline
Research Motor
Compound method method Difference equivalents of freshly distilled samples of these hydrocarbons,
AROMATICS as procured from the Eastman Kodak Company, were rede-
Benzene 10 6 4 termined; the results checked previous determinations.
Toluene 15 8 7
o-Xylene 17 11 6
Acknowledgment
m-Xylene 23 13 10
p-Xylene 26 13 13
Ethylbenzene 19 11 8 The authors wish to thank P. L. Cramer of this laboratory
Mesitylene 31 16 15
for helpful suggestions in connection with the preparation of
1,3-Diethylbenzene 30 24 6
Indene
1-Phenyl-l,3-butadiene
21
33
16
30
5
3
many of the hydrocarbons listed in this work.
NAPHTHENES
Literature Cited
Cyclopentene 16 14 2
Cyclopentane 14 12 (1; Am. Sue. Testing Materials, Proceedings, 33 (I), 746, Specifica-
Cyelonexene 10 9 1
Cyclohexane 7 6 1
tion D 357-33T (1933),
Methylcyclohexane 5 5 0 (1A) Baeyer, Ber., 25, 1840 (1892),
1,2-Dimethylcyclohexane 6 4 (2) Birch and Stansfield, Nature, 123, 490 (1929).
1,2-Methylethylcyclohexane 0 0 0
(3) Bourguel, Ann. chim., [10] 3, 379 (1925).
1,2-Methyl-n-propylcyclohexaue -5 -6 1
1,2-Methyl-n-butylcyclohexane -16 -13 -3 (4) Campbell, Lovell, and Boyd, Ind. Eng. Chem., 20, 1045 (1928)
OLEFINS (5) Garner and Evans, J. Inst. Petroleum Tech., 18, 751 (1932).
2-Pentene 16 13 3 (6) Harries, Ber., 45, 811 (1912).
2-Methyl-2-butene 23 15 8 (7) Hofmann, Lang, Berlin, and Schmidt, Brcnnstoff-Ckem., 13, 161
Diisobutylene 31 27 4
(1932) .

PARAFFINS
(8) Ibid., 14, 103 (1933).
n-Hexane -6 -6 0 (9) Howes and Nash, J. Soc. Chem. hid., 49, 16 (1930)
n-Heptane -14 -13 -1
2,2,4-Trimethylpentane 16 13 3 (10) Klages, Ber., 37, 2305 (1904).
(11) Lovell, Campbell, and Boyd, Ind. Eng. Chem., 23, 26 (1931).
(12) Ibid., 23, 555 (1931).
Comparison with Other Data (13) Ibid., 25, 1107 (1933).
(14) Midgley and Boyd, Ibid., 14, 589 (1922).
In Figure 4 comparison is made between the aniline equiva- (15) Muskat and Herrman, /. Am. Chem. Soc., 53, 25S (1931).
(16) Ricardo, Automobile Engineer, 11, 92 (1921).
lents presented here and some related data previously pub- (17) Signaigo and Cramer, J. Am. Chem. Soc., 55, 3326 (1933),
lished by other investigators. The data of Howes and Nash (18) Thiele, Ber., 33, 671 (1900).
(9) originally expressed in centimeters of tetraethyllead per (19) Veal, Best, Campbell, and Holadav. S. . E. Journal, 32, 105
gallon were converted to equivalent octane number by Garner (1933) .

and Evans (-5). Received January 24, 1934,

Steep Grade at General Motors Proving Ground