Relation of Octane Number to Cool Flame Formation h1. R. BARUSCH AND J. Q . PAYNE California Research Corp., Richmond, Calif. T h i s work was undertaken to determine the possibility of stabilizing a cool flame in a straight tube and to investigate the relationship between octane numbers and the tendency of fuels to produce cool flames. Apparatus was developed in which a stationary cool flame could be obtained. The equilibrium position of the stabilized flame was a function of temperature, flow rate, and composition of the fuel. In all experiments pressure and fuel-air ratio were maintained at a constant value. As the temperature was decreased or the flow rate was increased, the cool flame moved away from the inlet of the tube. Blends of iso-octane and n-heptane were employed to establish a correlation between octane number and cool flame position. The tendencies of other hydrocarbons to produce cool flames and the effect of antiknock additives and a prodetonant were correlated with octane number. The resistance of a fuel to knock appears to be controlled principally by its resistance to cool flame formation.
ARIOUS hypotheses have been advanced (4, 6) to explain the phenomenon of knock in gasoline engines. One of the more popular theories assumes that in the unburned fuel-air mixture ahead of the flame front, preflame reactions occur which set the stage for a rapid, more violent ignition, which is knock. I t has further been suggested that the preflame reaction causing knock is the same as that which causes cool flames ( 5 ) . As an aid in deciding where to accent research on gasoline engine combustion, the present work was carried out to establish firmly the relationship between fuel performance and the cool flame reaction. By passing a fuel-air mixture through a straight, heated glass tube, under carefully controlled conditions, it was possible to obtain stationary cool flames. The induction period of cool flame formation as measured by position of the flame in the tube correlates with octane number.
FUEL ASPIRATOR STAINLESS STEEL CAPILLARY TURING
WAX S E A L
> Figure 1. Cool Flame Tube
In all experiments except those relating flame position to flow rate the linear gas velocity was 10 cm. per second. This is the cool flame speed which several investigators ( 1 , 7 ) have reported. Under this condition the flame appeared as a plane, the top of which sloped toward the tube inlet at an angle of about 30 degrees from the vertical. Flame positions were determined by sighting the top of the flame through two parallel hairlines. The term “flame position” refers to the distance between the top of the cool flame and the sintered plate-Le., the shortest distance to the flame. This is illustrated by the distance, d, in Figure 1. After each determination of flame position the gas flow was interrupted to allow all parts of the tube to return to equilibrium temperature. RELATION OF FLAME POSITION TO FLOW RATE
The position of the cool flame was dependent upon the flow rate. Increasing the flow rate caused the flame to migrate downstream to a new equilibrium position. Figure 2 shows the
EXPERIMENTAL METHOD
A diagram of the borosilicate glass combustion tube used in this work is shown in Figure 1. The length of the tube was 2.2 meters and the inner diameter 25 mm. The tube in a horizontal position was maintained a t constant temperature in an air bath controlled by a thermostat. The longitudinal temperature gradient measured when no gases were flowing through the tube was less than 2” C. but when the tube was in operation, the cool incoming vapors depressed the temperature of the inlet end about 7 ” C. The temperature of the tube was measured by a thermocouple located a t a point 1.4 meters from the inlet. A sintered-glass plate was sealed a t the inlet end of the tube to prevent turbulence in the gas stream. The fuel was metered by means of a motor-driven syringe to the injection device diagrammed in Figure 1. This injector consisted of a stainless steel capillary tube through which the fuel was introduced. The steel capillary was surrounded by a glass tube drawn down to give a very small clearance. The air was introduced through this outer glass tube and its high velocity caused dispersion of the fuel into a fine mist.
REGION OF DIFFUSE U C
0’ 5c
8
REGION ff COOL FLAME PUL%ATIOND
CJ
5
Figure 2.
LI;QEAR FLOW
- C‘ijs~c.
Eo
Cool Flame Position vs. Flow Rate
60 octane reference fuel, temperature 296’ C.
2329
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
2330
relationship between flame (position and gas velocity a t 296" C. for a mixture of 60% iso-octane (2,2,4-trimethylpentane) in nheptane. The fuel-air ratio was stoichiometric for conversion to carbon dioxide and water. This fuel-air ratio was used in all experiments described. I n order to maintain this ratio in thr flow experiments the mixture was obtained by carburetion a t 0' C. When the flow rate became sufficiently low, a steady flame was not obtained and puldations of the flame front occurred. As the flow waa further decreased, the amplitude of these pulsations increased until finally the cool flame traveled to the inlet and disappeared. Following disappearance of the flame there was a period during which no luminescence was observed. The cool flame then reappeared in the tube and again swept t o the inlet and vanished.
Vol. 43, No. 10
This fact was established by frequently reproducing the flame positions obtained with reference fuels. Limited data indicated that changes in the surface of a given tube resulted in curves of identical slopes when cool flame position was plotted against temperature or composition. RELATION O F OCTANE NUMBER 1W COOL FLAME POSITIOY
Table I summarizes the data which show the relationship between cool flame position and octane number for primary reference fuels. Minor fluctuations in temperature occurred when these data were obtained. Therefore the flame positions were corrected to a constant temperature of 291.7' C. using Figure 3. The corrected cool flame positions are plotted as a function of octane number in Figure 4. A smooth curve resulta and establishes an excellent correlation between flame position and octane number for primary reference fuels. Under the conditions used, 100% iso-octane would exhibit no appreciable reaction, so it might be argued that iso-octane was acting a8 an inert diluent. The work of Beatty and Edgar ( 3 ) refutes this conclusion, however, as they established that under similar conditions a large proportion of isodctane admixed with nheptane undergoes induced oxidation
TABLE1. RELATION OF COOLFLANEPOSITION TO OCTANE NUMBER FOR PRIMARY REFERENCE FUELS Primary Reference Fuel, Octane No. 0
'r:mz.
& Figure 3.
&
re5
290
zb5
300
;05
3;O
d
0
TEMPERATURE -T
Effect of Temperature on Cool Flame Position
80
Primon. reference fuel
Under conditions of stable fl-lnnies, as the linear gas velocity was increased the length of the flame increased-Le., the flame became more nearly horizontal. When the flow rate was increased beyond the limiting conditions a sharp flame front was not obtained. Under such conditions a zone of diffuse glon wa8 observed. RELATION O F COOL FLAME POSITION TO TEMPERAl'URE
Figure 3 shows the relationship of cool flame position to temperature for seven blends of iso-octane with n-heptane. As the temperature was increased the flame moved upstream. The flame position n-as extremely sensitive to temperature. Thus, careful temperature control was a necessity for the accurate measurement of flame position.
::
40 51 83
67 79
70 104
110 154
154
Table I1 presents similar data for primary reference fuels conlaining both proknock and antiknock additives and also blends containing an olefin and an aromatic compound. The data were corrected t o constant temperatwe making the assumption that reference fuel containing an additive behaved the same way as a primary reference fuel of the sitine octane number. Figure 5 illustrates the relationship of cool flame pobition to F-1 octane number for these fuels. The solid line I U the graph is the reference fuel curve. The experimental points for the reference fuels plus additives fall close to this curve. Tetraethyllead was not included in the above experiment*, a8 it was desired to avoid the surface difficulties caused by this
EFFECT O F SURFACE ON COOL FLAME POSITION
The position of the cool flame was greatly influenced by the surface condition of the tube. To obtain reproducible results in a new tube it was necessary to pass a mixture of hydrocarbon and air through the equipment for several hours, using conditions that gave rise t o a cool flame. Apparently this pretreatnieut caused the formation of an invisible coating on the tube which reached an equilibrium state. Subsequent to this treLttment reproducible flame positions were readily obtained. As a new tube was gradually brought to its equilibrium surface condition, the position of the cool flame from a given fuel tended to move upstream. After establishment of equilibrium, the surface remained unaltered for as long as several weeks even though operation was intermittent. Interruption of operation for long periods caused a surface change which led to new flame positions. The data reported herein were all obtained in a tube whose surface condition remained unaltered during the determinations
Position, Cm. 16 33
292 2 290.6 290.6 201.1 291 1 291.7 290 6 291.7
20
30 40 50 60 70
Cool Flame Position at 291.7O C . 17 30
Cool Flame I
r50
z.125 U I
z 0 E
100
P
w '5
I
5
Y
d
50
U 0
0
1
0
2
0
3
0
4
0
0
0
W
?
9
B
o
OCTANE NO.
Figure 4.
Cool Flame Position
US.
Ootane Number
Primary reference fuel, temperature 291.7'
C.
~ 0
INDUSTRIAL A N D ENGINEERING CHEMISTRY
October 1951
2331
TABLE 11. RELATIONOF COOLFLAME POSITION TO F-1 OCTANE NUMBERFOR PRIMARY REFERENCEFUELSPLUS ADDITIVES
Fuel n-Heptane and 10% toluene 60-Octane reference fuel and 0.5% di-tert-butyl peroxide 48-Octane reference fuel and 10% diisobutylene 60-Octane referenoe fuel and 2% ethyl iodide 48-Octane reference fuel and 1% aniline n-Heptane and 40% diisobutylene
I
I
I
290.6
14
32
290.6
40
64
52
291.7
64
83
83
z
30
291.7
67
90
90
291.7 289.4
67 71
93 131
93 119
N-PENTANE0
0
d,
io
io
io
9-METMTL HLYANL
4o
d~
so
70
8o
90
I im
F-l OCTANE NO,
Figure 6. Cool Flame Position us. F-1 O c t a n e N u m b e r Pure hydrocarbons, temperature 291.7' C.
data on olefins may explain the heptene discrepancy. It is interesting to note, however, that when the cool flame data are plotted against F-1blending octane numbers as is done in Figure 7, the point for 1-heptene as well as the other points falls much closer to the reference curve. DISCUSSION
. 0
o
IO
20
30
40
60
eo
70
f-1 OCTANE NUMBER
Figure 5 .
Cool Flame Position
us.
80
w
1
io0
F-1 O c t a n e N u m b e r
The position of a stabilized cool flame from the various mixtures tested correlated well with octane number. As the octane number of the mixture increased beyond a critical value, the curve rapidly flattened out. This effect could be compensated for to a degree by increafiing the temperature. The apparatus. used in this work was limited t o a maximum temperature of
Primary reference fuels plus additive, temperature 291.7' C.
compound. Primary reference fuel containing tetraethyllead was run in a similar piece of equipment. The determination on tetraethyllead was difficult t o reproduce. When a fuel containing lead was run, the surface of the flow system was presumably altered by deposition of a film of lead oxide. If the flow of the leaded fuel was allowed to continue, the flame gradually moved down the tube and finally came t o a new equilibrium position. After leaded fuels were introduced into a flow tube, it was extremely difficult to obtain reproducible results with other materials. A clear fuel run immediately after a leaded fuel tended to behave as if the uncompounded fuel were leaded. In a tube not used in the other experiments reported herein, a 48-octane number-primary reference fuel containing 1 ml. of tetraethyllead was introduced. This fuel gave an F-1 octane number of 60. The cool flame from this fuel occurred initially within 1 cm. of the flame position occupied by 60% iso-octane in n-heptane. Because of changes of the surface this cool flame began to migrate gradually down the tube. Because the initial flame position coincided with that predicted by its octane number, it is believed that the cool flame positions obtained with fuels containing tetraethyllead correlate with F-1 octane number. Table I11 summarizes the data relating cool flame position for pure hydrocarbons t o F-1 octane number. Except for the 1heptene, the octane numbers were obtained from the Eleventh Annual Report of the American Petroleum Institute Research Project 45 (E). Assuming that these hydrocarbons behaved in a manner identical to primary reference fuels of the same octane number, the flame positions were corrected t o constant temperature. Figure 6 presents a plot of cool flame position us. F-1octane number for these data. Except for 1-heptene, the compounds tested fall in the proximity of the reference fuel curve. Further
TABLE 111. RELATIONOF COOLFLAME POSITION TO F-1 OCTANB NUMBERFOR PUREHYDROCARBONS
Fuel n-Heptane 2-Methylheptane 2-Methylbexane n-Hexane 1-Octene n-Pentane 1-Heptene 2-Methylpentane
Tzrnd.,
F-1 Octane
292.2 288.3 296.1 292.2 292.2 291.7 291.7 295.0
0 21.7 42.4 24.8 28.7 61.7 42 73.4
No.
Cool Flame Position, Cm. 16 23 31 40 66 103 107 112
Cool Flame Position at 291.7O C. 17 18 40 41 69 103 107 128
175>
&'Z-MET"IL PLNT4HL
OJ
0
10
20
30
40
SO
W
70
80
$0
D
OCTANE NUMBER .
Figure 7. Cool Flame Position us. F-1 Blending O c t a n e Number Pure hydroaarbons, temperature 291.7O C.
2332
INDUSTRIAL A N D ENGINEERING CHEMISTRY
about 316" C. At this temperature the curve of octane uuniber vs! flame position had an inflection point a t an octane number of about 80. Consequently, a correlation for higher oct,ane fuels than this value was not obtained. Cool flames xere observed under somewhat different conditions with fuels having octane nymbers as high as 95. A survey of the literature indicated that it inlight not be possible to produce cool flames for hydrocarbons of q u c h higher octane numbers a t atmospheric pressure. ,The position of the cool flame as measured in these erperin(ents is evidently related to, but not directly proportional to, the induction period that precedes cool flame formation. \Vheneder the fuel-air mixture was introduced the cool flame appeared in, a portion of the tube downstream from its equilibrium position. Start,ing immediately the cool flame Trould slowly migrate upstream to the equilibriuni point. Thus the time required for the gases to travel to the stabilized flame front was always less than the induction Deriod would have been a t conditions uninfluenced by the presence of the flame. This behavior was possibly due to diffusion effect's and the localized heating caused b!. tmhe Cool flame reaction.
Vol. 43, No. 10
LITERATURE CITED
(1) Aivezov, B. V., and Neiman, M. B., J . Phys. Chent. ( U . S . S . R . ) ,8, 88 (1936). (2) American Petroleum Inst. Research Project 45, Eleventh Annual Report, (July 1, 1948, to .June 30, 1949) Ohio State University Research Foundation. (3) Beatty, H. A,, and Edgar, G., J . -4m.Chem. Soc., 5 6 , 107 (1934). (4) Jost, W. "Explosion and Combustion Processes in Gases," KeTzYork and London, McGraw-Hill Book Co., 1946, translated by H. 0.Croft. (5) Jost, W., "Physico-Chemical Investigations of the Combustion Process in an Engine." Technical Oil Missions Reel 52, Frames 509-549, Library of Congress, Photoduplication Service, Publication Board Project, Washington, D. C. (Dee. 16, 1937), translation FL-45, Consultants Bureau, New York (1948). ( 6 ) Lewis, B., and Yon Elbe, G. "Combustion, Flames and Explosions of Gases," London, Cambridge University Press, 1938. (7) Townend, D. T. A., and Chamberlain, E. A. C., Pror. Roy. SOC. Lon,dois, 158,415 (1937). R E C E I V December ~~D 2, 1950. Portions of this paper were preaented before the r)irision of petroieuin chemistry at the 118th hfeeting of the AYERICAU CHE\IIC%L SOCIETY, Chicago, TI!
Some Properties of
Perfluorocarboxylic Acids d
E. A. IUUCK AKD A. R. DIESSLIN Minnesota Mining & Manufacturing Co., S t . Paul, Minn. During an investigation of the electrocheniical fluorination of organic compounds dissolved in anhydrous h : drogen fluoride, i t was discovered that fully fluorinated acjl fluorides Mere produced which hydrolyze readily to the corresponding perfluoro acids. Boiling points and liquid densities are presented for the following acids: perfluoroacetic acid, perfluoropropionic acid, per0tiorobutyric acid, perfluoroisobutyric acid, perfluorovaleric acid, perfluorocaproic acid, perfluoroheptanoic acid, perfluorocapr?lic acid, perfluorocapric acid, perfluoromyristic acid, perfluorocyclohexanecarboxylic acid, and perfluorocyclohexaneacetic acid. With the exception of the first three compounds listed, these substances are reported for the first time. Data are presented on vapor pressure, iiscositj, pH, and equivalent conductance of perfluoroacetic and perfluorobutyric acids. A feasible method of preparing a new series of fluorinated acids has been developed and these fluorinated acids are being made available to the chemical public. A wide variety of useful derivatives can be made from these acids. Interest in the acids and their derivatives is expected because of their unusual chemical and physical properties.
T
HE electrochemical fluorination of organic rompounds in liquid hydrogen fluoride was first accomplished by Simons ( I S , 1.4) by electrolysis of a solution of organic material in liquid hydrogen fluoride at an electrolyzing potential which is insufficient to generate free fluorine. Fluorination is accomplished in one step to obtain fully fluorinated end products Jyhich, depending upon their boiling points, either leave the cell as gases or settle t o the cell bottom and are drained out. Trifluoroacetio acid, CF,COOH, was first prepared by Swarts in 1922 (15). Its preparation since has been described by others ( 1 , 2, 6). Perfluoropropionic and perfluorobutyric acids have
recently been prepared ( 4 ) . It has been found that a great variety of fully fluorinated acid fluorides can be readily produced by the electrolysis of the corresponding hydrocarbon acids or anhydrides in hydrogen fluoride. The acid fluorides thus formed are easily hydrolyzed to the coi ponding perfluoro acids. EXPERIMENTAL PROCEDURE
An electrolytic cell was constructed from &inch standard iron pipe. Suspended from the cover were nine nickel anodes and nine iron cathodes, each l/,e X 3.5 X 6.5 inches spaced 0.125 inch apart. Figure 1 is a photograph of the cell disniantled and assembled to show the electrode pack. The cell was charged with 2000 grams of a 4% solution of acetic anhydride in anhydrous hydrogen fluoride. The electrolysis was conducted a t 50 amperes, 5.2 volts direct current with an electrolyt,e t,eInperature of 20" C. a t atmospheric pressure. The gaseous products were passed through a reflux condenser maintained a t about -30" C. to reniove t,he bulk of the accompanying hydrogen fluoride and then through a dry sodium fluoride tube to remove the final traces of hydrogen fluoride. h typical reaction may be illustrated as follows: (CH3CO)ZO t l O H F 4 2 C F s C O F OF2 8H2
+
+
I n actual cell pract,ice, some decarboxylation and fragmcntation occur, producing fluorinated material with fewer carbon atoms than the starting material. Trifluoroacetyl fluoride is readily absorbed by water, hydrolyzing to trifluoroacetic acid and hydrogen fluoride. These acids are separated in various TTays-for example, by ion exchange or by the extraction of an alkali salt with a solvent ( 7 ) . The trifluoroacetic acid produced in one run was found by titration to be 99.9% pure. The electrolysis is readily operated continuously. Acetic anhydride and hydrogen fluoride are charged continuously to the cell to replace that consumed by the electrolysis and t'he reaction products are withdrawn continuously. This electrolysis has been conducted in a 2000-ampere pilot plant cell for several months