the products from a lead arc d o have
a measurable antiknock effect! even though the particles in the two experiments are apparently in the same size range. There are some plausible explanations for this apparent contradiction. T h e first explanation is based on the possibility that some of the lead particles entering the engine from the lead arc were actually very much smaller than is indicated from the analysis of the material collected on the filter. O n e cannot rule out the possibility that some lead may have been atomically dispersed. T h e existence of lead vapor or in conof agglomerates u p to 500 -4.> centrations of about lOy0 of the total lead, would then easily explain the difference betlveen the two sets of experiments. .4lthough this is very easy to conceive. it is very difficult to verify experimentally. O n the other hand. it may be assumed that the smoke particles on the filter are of a size truly representative of the material entering the engine. Since these particles are of the same size as those obtained in the suspension experiments, the difference in behavior must be attributed to the nature of the particles. I t is possible that the greater age of the suspended particles or their oleic acid coating in some way inactivated the lead surfaces. Another possibility is that the lead smoke particles \yere quickly oxidized to lead oxide in the hot inlet air stream, while the suspended particles were protected from oxidation a t this stage by the oleate coating. If lead oxide is indeed the effective antiknock and lead is not! as previously pro-
Fundamentals
posed (7). this difference in oxidation state would explain the apparent contradiction. T h e major conclusion to be draivn from this !vork is that inorganic lead in some form has antiknock properties. T h e evidence available to date. however. does not allow one to distinguish betneen a mechanism based on a chemical reaction as suggested by Egerton ( 9 ) or a surface inhibition as postulated b) Chamberlain and LValsh (7) T h e observed effectiveness of inorganic lead suggests that the decomposition products of T E L are important in the anriknock mechanism. For this reason a study of the factors Lrhich affect TEL decomposition is being made.
Acknowledgment T h e authors express their appreciation to R. C. Getoor for his help \vith some of the experimental Ivork. They also wish to acknowledge \rith gratitude the electron microscope work. done by C . Lavintal of the Physics Department, Cniversity of Michigan. literature Cited ( 1 ) Badin? E. J.. "'Third Symposium on
Combustion, Flames, and Explosion Phenomena." p. 386. Williams 8r Wilkins. Baltimore, 1949. (2) Ball, G. A,. SAE National Fuels and Lubricants Meeting, Tulsa, Okla., S o v . 5 %1954. (3) Berl. H. E.. Winnacker. %. physik. Chem. A145, 161 (1929). (4) Calingaert, G.. Gambrill. C. M.. IND. ENG. CHEM.. ANAL. ED.. 11,324(1939).
( 5 j Callendar. H . L.. King, R. O., Sims. C. .J., ;\eronautical Research Committee. Reuorts and Memoranda. 1013, 1062: 1925 [Engineering 121, 475 (1926)l. (6) Chamberlain. G. H . K.,Hoare. D. E.. Walsh. A . D.. Discussions Faradav SOC.S o . 14. 89. 1953. ( 7 ) Chamberlain. G. H. Walsh. .A. D., Proc. Roy. Soc. (London) A215, 175 (1952). ( 8 ) Downs, D.: Walsh. A . D.: IYheeler. R . W., Trans. Roy. SOC.(London) 243,463(1951). (9) Egerton, A , . Gates. S. F., .J. Inst. Petrolfum Technoi. 13. 250 (1927 I. f l 0 i Z b i d . . ~ .281 ( 1 1 ) Egerton, A . C.. Rudrakanchana, S., Fuel 33,274 ( 19 54 1. (12) Ethyl Corp.. Detroit. kfich.. Analvtical Method 10-53. (13) Frank, C. E.. Blackham. A . L'.,I s u . EKG.CHEM. 4 4 , 862 (1952). (14) Hoare, D. E.. Walsh. A . D., "Fifth Symposium on Combustion,'' Abstracts. p. 77. Pittsburgh, Pa.. September 1954. (15) King. R . 0.. Can. J . Research F26, 125 (1948). (16) Norrish. R . G. W.. Ptoc. Roy. Snc.. ( L o n d o n ) A-150, 26 (1935). (17) Rifkin, E. B.. Walcutt. C., Betker, G. W..Jr.. SAE Quart. Trons. 6 , 472 (1952). (18) Sims. C. J.. Mardles. E. W..I.. T r a n j . Faraday Soc.. 22, 363 (1926). (19j Sokolik. A . %Yantovskii. S. Acta Physicochim. (C',R,S,S.) 19, 329 (1944). (20) Sortman. C. W.. Beatty. H. A , , Heron, S. D.. I N D . ENG. CHEM.33, 357 (1941 ), (21) Walsh. A. D., "Third Symposium on Combustion. Flames; and Explosion Phenomena," p. 389, Williams and Wilkins. Baltimore. 1949. (22 j Withrow. L.. Rassweiler. G. M., IYD. ENG.CHEM.27, 872 (1935).
- the Arrhenius relationship
These data. unfortunately, are not very useful for investigating the effect of T E L on the combustion process in a n engine for t\vo reasons. First. T E L vapor alone \vas studied, whereas in an engine it might be expected that the fuel and air \vould interact ivith the T E L . Secondly. Leermakers' Lvork covered a temperature range far below that of greatest interest in engines. I t is therefore obvious that additional studies of this nature in a more appropriate temperature range and in a more typical environment \vould be useful. Knock in spark-ignited engines is knoivn to occur in that portion of the fuel-air mixture Jvhich has not as )-et been consumed by the flame front. It is reasonable to expect. then. that the effect of T E L in reducing or retarding knock Ivould be manifested in that portion of the charge which is commonly referred to as the end gas. Recent comprehensive studies (8)of the temperature condition in this body of gas have indicated that in normal operation of a n engine the end gas may be subjected to temperatures of the order 500" to 750' C. for approximately 3 milliseconds before burning. It is obvious that the decomposition of T E L must be studied in this particular region if data of use in the understanding of the mechanism of antiknock action are to be o brained.
Equipment
T h e experiments described here were conducted in a single-cylinder C F R engine. T h e engine {vas coupled to a n electric dynamometer which could be run over a \vide range of speeds. Control was maintained over the rate of fuel flo~v. the temperature. humidity, and the pressure of the air in the inlet manifold. Fired Engine Studies of TEL Decomposition. Two different types of studies of T E L decomposition in a n engine were made. The first set involved the decomposition of T E L in a sparkignited, fixed compression ratio, L-head engine and employed a sampling valve which extracted: repetitively, small samples from the end-gas region of the combustion chamber. The engine conditions under which these samples were taken were as follows: 6.9 compression ratio; 900 r.p.m.; 0.077 fuel-air ratio; 150" F. inlet air; 150" F. coolingjacket; 15' before top center ignition timing. T h e sampling valve unit. shown in
Figure 1. has an electromagnetically operated stainless steel poppet valve. a cooling jacket around the sampling chamber. and a lO-mm.-diameter flexible metal sample outlet tube. T h e valve lift \vas 0.15 mm., and the total time from start of opening to closing was verv nearly 1 millisecond. during which time approximately 1 ml. of gas (standard temperature and pressure) was removed from the combustion chamber. Observation of the opening of the valve as originally designed indicated that it bounced several times on the seat for a period extending over 4 milliseconds. This malfunctioning \vas eliminated by attaching to the stem of the valve a metal cap of about 2-ml. volume and filled with tantalum powder. The valve timing was controlled by a set of breaker points on the camshaft of the engine \vhich discharged a condenser through the coil on the unit. T h e actual time of valve opening \vas calculated from a measurement of the time at \vhich the condenser discharged. The sampling system was maintained at a pressure of about 5 mm. of mercury, and several cold traps kept a t dry ice temperatures served to condense out those portions of the sample of greatest interest. The trap efficiency was determined to be greater than 907c when nvo were used in series. Inasmuch as only very small amounts of condensibles were collected in reasonable periods of operation. it was convenient to use specially designed traps which could be inserted in a large centrifuge in order to collect all the material in a small calibrated pipet a t the bottom of the trap. The condensates consisted of an aqueous phase and a hydrocarbon phase, and determinations of the lead dissolved in each ivere made. It was shown during the course of the work that the fuel-air ratio in the sampling line was csscntially identical lvith that of the gross input to the engine. so that the sampling valve can be regarded as having obtained a representative mixture from the combustion chamber. Motored Engine Studies of TEL Decomposition. I n the second phase of the work, utilizing a motored engine, no sampling valve was used, but 1 BO-liter samples of the gases exhausted from the engine \vere passed through a scrubber and a bubble-cap column in which the extracting medium was a light paraffin oil. Proof that the sampling line \vas withdraxving a representative fraction of the exhaust gases and that the scrubber was essentially 100$, effective \vas obtained by motoring the engine at a very lo\v compression ratio in \vhich case all of the expected lead from T E L was recovered in the organic phase in the scrubber. T h e analysis for undecomposed TEI, was based on the solubility differences betlveen T E L and inorganic lead com-
PlkG SEALS JClNT
c+r
COCLI~~~G
-
00-ING
N
L l h G .ACKET
"?""ET
MLVE
Figure 1. Sampling valve used in fired engine studies
pounds in paraffin hydrocarbons. As regards partial decomposition products, namely di- and trialkyllead compounds. a study of the extent to ivhich they are produced, given in detail below, indicated that it was not necessary to make a specific study of their formation. .4nalytical determinations of total dissolved lead in the paraffin oil were made by the dithizone colorimetric method in t h ? case of low concentrations and by polarographic methods for higher concentrations. Specialized Instrumentation. An important part of the test apparatus was the equipment used to measure the pressure in the combustion chamber. T h e basic measurement of pressure was made by a continuous-type Li strain-gage indicator. Calibration of the indicator was achieved by using a variable capacitance-type balanced pressure unit located in the combustion chamber of the engine adjacent to the strain gage. T h e outputs from both units were amplified electronically and appeared as traces on the face of a dual beam oscilloscope; these were recorded using a camera equipped Fvith an electronic shutter. This permitted obtaining simultaneous pressure-time records and calibration points for individual engine cycles. A detailed description of this equipment has been presented elseXvhere ( 3 ) . .4n improvement in this instrumentation not previously described is the method of establishing a precise time reference mark on the pressure traces. This was done in the following manner. A magnetic pickup was located near the periphery of the flywheel in such a way that the passage of a stud on the flywheel VOl. 48, NO. 9
SEPTEMBER 1956
1533
100
I
I
TEL Decomposition in the Fired Engine
I
I
80 -
a
-
W
v,
0
!$
60-
0 0
45% ISOOCTANE
w
A R R I V A L AT VALVE
IGNITION
-20
-10 TDC +IO t 20 SAMPLE TIMING, CRANKANGLE DEGREES
Figure 2.
TEL decomposition in fired engine
Fuels: Mixtures of iso-octane and n-heptane containing 6 mi. of TEL/gallon and 1 theory of ethylene dibromide
past the magnet changed the reluctance of the magnetic circuit. This generated a signal which activated a stroboscopic light and illuminated the flywheel, thereby determining the effective crankangle position a t Ivhich rhe signal was generated. At the same time this flash actuated a photoelectric cell, the output of \vhich was amplified in a low-lag circuit and appeared on the face of the oscilloscope as a small pip. The lag in the magnetic circuit and its amplifiei
\\'as of no consequence, since the position of the magnetic pickup was manually adjusted until the stroboscopic light flashed, as seen by cyc. at a preselected crank-angle time. T h e onl!- time lag ol' importance was that in the second part of the circuit. The lag in this part of the circuit was minimized, ho\vever, by using an electronic amplifier with a lag of only a few microseconds. which was far shorter than the time intervals of interest in this u.ork.
I "
I
8
9
IO
I1
12
13
14
15
COMPRESSION R A T I O
Figure 3.
1534
Effect of fuel on TEL decomposition in a motored engine
INDUSTRIAL AND ENGINEERING CHEMISTRY
16
I n this phase of the Lvork studies \scrr made using mixtures of iso-octane and n-heptane as fuels. Samples bvere taken every 5 degrees behveen 25' before top center and 20' after top center. Shortly after the latter time the flame front reached the sampling valve, preventing the removal of unburned fuel-TEL-air mixtures a t any later time in the engine cycle. The results of these studies arr shown in Figure 2, from Ivhich several deductions can be made. First, essentially all the T E L is decomposed beforr the sampled mixture is consumed by thr flame front. Secondly, there is a marked effect of fuel on the rate of decomposition of TEL. I n the interpretation of these data, tlvo alternate possibilities were considered. First \vas the obvious one: which implied that the octane level (or some other property) of the fuel mixture in some direct \\,a)- governed the rate of decomposition of T E L . The second possibilit! was a more subtle one. It has been recognized for some time ( 7 ) that fuelair mixtures can undergo extensive exothermic reactions prior to the time at \\,hich they burn or autoignite. Thesc reactions have been known in some cases to raise the temperature of the fuel-air charge by over 100' C. Furthermore, ii has been shoxvn that fuels of lower octane number tend IO undergo these reactions to a yreater degree than fuels of higher octane number, I n relating these facts to the present rcsults. it can t e seen that the second evplanation ivould tie the increased TEI. decomposition observcd ivith the lo\ver octane fuels to the higher temperatures which undoubtedly occurred \\.hen these fuels were tested. O n the basis of these data alone no clearcut distinction bethveen these two possibilities can be made. This is so because i t has not as yet been possible to measure accurately the temperarure of the end gases in a fired engine. I n vie\\o f this difficulty in interpretation oT this ty;x of data. no further work along this line \vas attempted. In order to proceed further. an alternate method had to be found. .I consideration of the usefulness of conventional laboratory apparatus indicated the difficulty of studying railid reactions a t high temperatures ( i ) , Therefore lve chose to carry out most of the present \\.ark in a motored engine. In such a device, a fuel-air mixture can br compressed and rapidly heated. Furthermorr. the repetitive nature of the cycles in a motored engine makes it possible to recover adequate amounts of material for chemical analysis. Also. reliable techniques are available for measuring the state of the gases during the cycle. Finally, because of the essential similarity of the process which
occurs in a motored engine as compared \\ith the fired engine. it would be expected that many of the factors of importance in the latter rvould also sho\v their effect in investigations conducted in the former.
I O@
I
I
'
' I -
'
160% ISOOCTANE IN n - H E P T A N E
'
k\
NO OTHER r ADD ITIV-
~
4
.
ISOOCTANE
TEL Decomposition in the Motored Engine
The second set of studies involved the drcomposition of T E L in a motored, variable-compression ratio: single-cylinder engine. A variable-compression ratio engine was chosen for this ivork in and order to, allo\v the temperature A pressure of the fuel-air mixture to be varied over as \vide a range as possible. T h e conditions under Lvhich these studies \ v e x carried out were generally the same as those used in the sampling valve studies, except that no spark ignition was used. A few tests were also made at 200 r.p.m. I n all, 66 tests were made Lvith the follo\ving fuels: iso-octane, mixtures of iso-octane and n-heptane of 90 and 60 octane number, and diisobutylene (DIB). In addition to T E L , tests were made in the presence of other additives such as ethylene dibromide, tert-butyl bromide, carbon tetrachloride, tert-butylhydroperoxide, and tris-P-chloropropylthionophosphate. Effect of Fuel. Figure 3 shows the results obtained on the tests involving four fuels each containing 6 ml. of TELjgallon and 1 theory of ethylene dibromide. (One theory of ethylene dibromide contains 2 atoms of bromine for each atom of lead.) I n each case the system \vas studied over a range of compression ratios below the autoignition limit of the fuel-air mixture. As \vas expected, for any one fuel, more of the T E L passed through the cycle unchanged when the compression ratio was lo\z~. In addition a very marked effect of fuel type \vas apparent-for example, at a compression ratio of about 10, three quarters of the T E L remained undecomposed \\.hen it \vas blended in a fuel composed of 90 volume % isooctane with n-heptane, but under the same conditions less than one quarter of the T E L passed through undecomposed \vhen the iso-octane concentration \vas decreased to 607,. These results are strong confirmation of the earlier indications from the sampling valve work that the nature of the fuel is an important factor in determining the extent to which T E L will decompose. Effect of Additives. Figure 4 gives the results of additive studies using tivo of the fuels; the addition of various materials tends to increase the extent to which T E L is decomposed. Thus in the fuel composed of 607, iso-octane in nheptane, additions of either tris-chloropropylthionophosphate in concentration
-1
~
,
:
O
10
8
Figure 4.
1
,
,
1
12 13 C O M P R E S S I O N RATIO
,-
L - b
15
I4
16
Effect of additives on TEL decomposition in a motored engine
of 7.87 atoms of phosphorus per atom of lead or six theories of tert-butyl bromide tend to accelerate the rate of T E L decomposition. Similarly, in isooctane the addition of 2.80 moles of tert-butylhydroperoxide per atom of lead promotes the decomposition of TEL. T h e effect of ten theories of carbon tetrachloride is considered marginal. Thus, the second conclusion \vould be that additives can markedly increase the rate of T E L decomposition. S o additive which was studied shoxved an appreciable effect in the reverse direction. Effect of Engine Speed. Figure 5 shows how the change from 900 to 200 r.p.m. affects the decomposition of T E L in iso-octane and diisobutylene. I t is quite clear that speed is another factor of considerable importance in determining the extent to Lvhich T E L decomposes,
with higher speed tending to give i n creased decomposition. T h e T E L decomposition data given in Figures 3 to 5 indicate that variables such as fuel, presence of other additives. engine compression ratio and speed all have important effects on the rate of decomposition. I t is recognized, of course, that specifying the compression ratio of a n engine does not ensure uniform temperature-time relationships in the cycle, as long as the other factors mentioned are varied. Thus a more basic approach to interpretation of the data Jyould involve a correlation Lvith temperature. For this reason peak cycle temperatures \yere calculated for some of the runs a t 900 r.p.m. using methods previously described (7), and the corresponding T E L decomposition data were examined against this variable. The results of this are shown in Figure 6 :
i 401-
c
20
I
1
c
1
-
I
-1
I
,
0
9
Figure
5.
IO
I
,
I2 I3 COMPRESSION RATIO
14
11
,
,
15
16
Effect of engine speed on TEL decomposition in a motored engine VOL. 48, NO. 9
0
SEPTEMBER 1956
1535
it is evident that the greatest variations among fuels have been eliminated in this relationship, although it appears that T E L is somewhat less decomposed in the presence of DIB than in the three other fuels. This relationship would lead one to suspect that temperature may be the one basic variable governing the extent of T E L decomposition in the engine and that the nature of the fuel has only a n indirect influence exerted through the effect of the fuel on the temperature of the charge. I n order to investigate this possibility, it was decided to relate the extent of T E L decomposition not only to the peak cycle temperature but to the entire temperature-time path through which the charge passed. In order to d o this it is clearly necessary to devise some type of integration procedure requiring that the various temperatures encountered by the charge in its passage through the engine be weighted in relation to the extent to which they would cause T E L to decompose. Such integration is necessarily a trial and error procedure, and, before anything can be done, it is necessary to assume a relation between the rate constant of decomposition and the temperature and also to decide what other factors than T E L concentration may be important. As an approach to the problem it was assumed that the only variables of importance ivere the T E L concentration and the temperature; that the relation between the rate constant for decomposition and temperature followed an Arrhenius relationship; and that the reaction was essentially homogeneous and first order. T h e mathematical approach is summarized here : The disappearance of a substance as a result of a first order, thermal decomposition can be described mathematically b) -dn/dt
= kn
9 0 % ISOOCTANE IN n - H E P T A N E -0-
80
ISOOCTANE IN -A
I
0
,
n -HEP
,
730
600
Figure 6.
1
1
I
,
l
600 PEAK CYCLE TEMPERATURE,
,
,
,
,
OK.
TEL decomposition in four fuels related to peak cycle temperature
and determining an averdge value of the rate constant. k. in that interval using Equation 2. If nl be the initial concentration of T E L and n, be the final concentration a t time t , . then n, can be calculated bv summing a large number of terms of the type log n l n?. log n2 n3. log n, n4. log n,-1 n f
Thus it is possible. given an Arrhenius expression for the rate constant of T E L decomposition and a n accurate temperature-time record of the cycle. to make an estimate of the extent to \vhich the T E L will have been decomposed in its passage through the cycle. If thr first trial results in poor agreement betIveen calculated and experimental
(1 1
lvhere n is the concentration a t time 1. and k is the rate constant which can be expected to vary with temperature as follows : k =
-4e-H'HT
(2)
where R is the gas constant, T the absolute temperature, and '4 and B are constants. In attempting to apply this type of analysis to the decomposition of T E L in the engine, we can integrate Expression 1, with the result
where the subscripts 1 and 2 refer to conditions a t the beginning and end of a small time interval. If this be the case. then the integration of the righthand side of the equation can be performed by measuring the temperature in the engine over the short time interval
1536
INDUSTRIAL AND ENGINEERING CHEMISTRY
l
900
% TEL UNDECOMPOSED, C A L C U L A T E D Figure 7.
Prediction of TEL decomposition in a motored engine
values. repeat calculations can be made using different constants in the Xrrhenius expression. Such a study. part of which \\as carried out on a n electronic computer. has been made. and a n Arrhenius expression for the rate constant for T E L decomposition has been obtained v, hich best fits the data from all 66 tests. This expression is log k = -4.124
X 103/T
where k = set.-' and T =
+ 7.617
Table
F!tel Iso-octane
(4)
Iso-octane
TEL If1 /Gal 5 5 5 5 5 5 6 6 5 5
Iso-octane
997 99 99 99 87 87' 01 01 99 991
5 91' 5 5 5 5 5 5 5 5
Precision of Data
In this type test. all the measurements are designed to give the experimenter sufficient information to evaluate two basic variables-temperature of the gases in the engine and the fraction of TEL remaining undecomposed a t the end of the cycle. T h e calculation of temperature involves many measurements, of which the most difficult is the determination of the pressure a t a given time in the cycle. T h e methods of calculation and the precision of the results have been described previously (7). An over-all evaluation of the errors inherent in the calculation of temperature places the precision at about 3yc. Random errors of this magnitude affect the results in an interesting way. From the relation. k = .Ir.-R T: it can be shown that the fractional error d In k is given by B , ' T d In T. ivith d In T set equal to 0.03 in this case. Similar treatment of the effect of this error on the calculated value for undecomposed TEL shows that. if the fractional value of undecomposed T E L is represented by u . the uncertainty in this quantity is given by u In u d In k . In the measure of the actual fraction of TEL undecomposed. a number of measurements must also be made involving chemical analysis as well as r:+tes of fuel and gas flow through the engine and through the sampling system. T h e over-all precision in this set of measurements is estimated to be 5%. T h e appropriateness of the equation as a means of predicting the experimental result for the decomposition of TEL is best evaluated on the basis of Ivhether the calculated values (bvith their uncertainty due to temperature measurement) agree with the measured values (with their uncertainty of determination). Such a correlation is given in Figure 7. which shows that 63 of the 66 tests fall within the prescribed limits of error.
TEL Decomposition in a Motored Engine-900
R.P.M. TEL
K
Table I giies the complete data. showing observed and calculated values for TEL decomposition for each run. T h e equation describes the behavior of the TEL in these tests fairly well. A more quantitative idea of fit can only be obtained in the light of the precision of the data.
I.
51 51' 51/ 51) 91 911 51 51'
5 91
Additices. C'oncn..
C'ompressiora
Ratio
Br(CH?i?Br(1.0)
CCI, (5.05) CClr (4.52) CCl, (3.76) CCl, (4.52) (CHaIsCOOH (2.80)
607' ISOoctane 40% n-Heptane
5 78
Br(CH?)?Br(1.0)
60% Isooctane 40% n-Heptane
6 01
Br(CH2)2Br (1.0) (CHsCHCICH~0)3PS (7,871
6 0 7 ~ISOoctane 40% n-Heptane
6 02
C(CH~)~B (11 I 96)
907' ISOoctane 10% n-Heptane
6 02
Br(CH?),Br (1 . O i
Diiso-
6 07
+
+
+
+ +
butylene
Diisobutylene
Iso-octane
C-tade-
Peak C y d ~ composed. % Tamp.. O K , Ezptl. Calcrl
11.0 11.0 13.0 13.0 13.0 13.0 13.0 13.0 16.0 16.0
730 740 763 773 757 771 780 780 807 819
53.0 67.4 29.6 27.8 34 8 32.9 22.4 26.0 12.6 12.7
58 52 24 27 42 34 20 24 17 10
3 7 1 6 2 6 9
11.5 11.5 12.5 12.5 13.0 13.0 13 0 13.0 13.5 13.5
730 726 767 760 773 765 790 774 766 774
60.7 57.8 38.0 40.1 32.6 30.9 22.0 23,l 28.6 27.5
54 56 37 35 31 36 25 31 37 28
9 9 2 7 1 9 6
11.0 11.0 13.0 13.0 13.19 13.19 14.0
744 760 815 800 796 805 816
43.2 44.1 13.6 10.0 19.7 17.6 17.0
46 42 20 20 21 16 12
9 5
9.0 9.0 9.5 9.5 9.75 9.75
692 692 724 747 751 780
73.2 66.1 45.3 39.0 27.2 23.5
75 1 74 3 60 5 51 9 45 3 31 6
8.5 8.5 8.5 9 3 9.3
704 674 704 766 757
83.6 84.3 61.9 34.1 25.6
68 82 67 29 34
6 7
8.5 8.5 9.15 9.15 9.25
652 674 700 702 718
90.2 86.6 57.3 57.8 41.5
89 82 66 70 52
6 3 7 0 8
9.0 9.0 10 0 11.0 13.0 13 0
678 679 704 725 791 806
78.5 80.5 75.2 32.7 14.5 12.9
78 79 68 56 19 17
9 2 7 0 3 6
11.0 11.0 13.0 13 0 13.0 13.0 14.16
762 739 768 775 747 766 772
63.7 66.3 47.9 45.3 58.8 47.0 35.6
43 57 35 40 47 36 30
9 9 5 8 8 5
8
4 0
0
1 0
6
3 0
9 9
0
9 2
0
6 6 6 6
07 07 00) 00)
10.0 10.0 12.0 12.0
613 61 1 632 627
82.8 83.6 91,l 81.2
82 3 83 0
5 5 5 5 6 6
95) 95 91L 91' 01 01)
10.0
602 600 606 619 726 738
94.5 95.0 93.5 81.2 23.4 22.1
85 86 83 82 12 7
10.0
Br(CHz)?Br (1.0)
12.0 12.0 13.5 13 5
76 4 79 0 8
3 5 3 0
9
VOL. 48, NO. 9
SEPTEMBER 1956
1537
T h e conclusion to be drawn from this is that the equation given above adequately describes the rate of decomposition of T E L in the engine icithin the precision of the measurements regardless of the fuels a n d additives used in this work. A more careful breakdoicn of the data on the basis of each fuel and additive combination shoics that there are trends within the data Lchich might be considered significant. T h u s we may list the data from each fuel and additive combination in order of the average deviation from the mean for the differences between calculated and observed values of per cent T E L decomposed
reasoning, it \could be expected that isooctane! which is of intermediate reactivity, Tvould agree jvith the equation presented here since the equarion \cas derived as the best fit for the entire group of data. X s a consequence of these unresolved thermal gradients in the engine, it is reasonable to find that the calculated values for undecomposed T E L in D I B are somelchat loit-er and the values for the 60 O S fuel are higher than the experimental results. Order of the Reaction
In testing the hypothesis that the Average Deviation. TEI, Decomposed (Observed hfinus Calculated)
careful control over engine conditions indicated substantially identical teniperature-time situations for all tests. I t is clear that a first-order rate constant ( u = 1) \Till result in equal values for u1 and u?. Furthermore, from Equation 6, if u1 is in the neighborhood of SO%, each 0.1 change in a will spread the values of u 1 and up by very nearly -5 percentage points in T E L decomposition. The results presented thus shoiv that the order of the reaction is very nearly 1 \cith possibly 1.3 as the upper limit. Considering the scatter of experimental points, it is concluded that a = 1 and that the hypothesis of a first-order reaction with respect to T E L is justified.
0 Fuel Diisobutylene Iso-octane Dii5obutvlene 60'; Iso-octane n-heptane Iso-octane Iso-octane Iso-octane 90r; Iso-octane n-heptane GOc; Iso-octane n-heptane 60'; Iso-octane n-heptane
+ 40tL0 + 10%. _ + 40% + 4OC0
Br ( CH 2 ) ?Br Br( CH2)ZBr Br( CH2)lBr
900
10.30
200 200
8.65 4 50
(CHaCH CICH?O)?PS Br( CH2)zBr CClr (CH8)aCOOH
__ snn
n.. 69 -.
900 900
-1.21 -1 .41
Br( CH2)zBr
000
-4.23
( C H B)&Br
900
-5.60
Br(CH2)ZBr
900
-10.73
LVe recognize, of course, that even the greatest of deviations could be eliminated by assuming that measured temperatures were in error by 10" or 20" C. However, the consistent trends sho\cn mainly by the DIB and by the 60 octane number (ON) fuel, which are in opposite directions, imply that these deviations may be of a systematic nature. .4relatively simple explanation for this effect is based on the fact that the temperature in the combustion chamber is not uniform because of varying rates of heat loss in different portions of the chamber. I n a fuel that undergoes extensive precombustion reactions, such as the 60 O S fuel used in some of these tests ( 7 ) , these temperature gradients would be accentuated, because the hottest portions of the gases would react more extensively a n d liberate more heat than the cooler portions. I n our measurements of temperature, however, ice can only compute the average temperature in the combustion chamber. Because the decomposition rate of T E L varies in a n exponential \cay with temperature. it \could be expected that the calculated value for decomposed T E L . based o n a n average value of temperature, \could be lower than the measured value. This effect uvould be expected to be large for the 60 ON fuel since it is quite reactive, whereas for DIB, which liberates no heat as a result of precombustion reactions, the effect should be small. As a further extension of this
1538
Speed. K . P. M.
Additive
1 42
900 ~
decomposition reaction is of first order with respect to T E L ? it is convenient to \vrite the general relation dn/dt = -kn'
(5)
\there a is the order of the reaction with respect to T E L . Because the usual kinetic tests for order of the reaction cannot be carried out, \ve can investigate the order by carrying out t\vo experiments under similar temperature conditions using two different initial concentrations of T E L . If ,\-, and .\-? represent these t\vo different initial concentrations, and, if u1 and u? represent the corresponding fractions of undecomposed T E L at the end of the tests. xce may \\rite
From this relation and kno\cing the quantities .VI: .V?, u l > and u ? ! it is clearly possible to calculate a. the order of the reaction with respect to T E L . T ~ c osets of experiments lvere carried out in this way! using D I B with 3.0 and 12.0 ml. of T E L per gallon, respectively. of T E L T h e results, given in terms of 7~ undecomposed. were 3.0 mi. Set I Set I1
61.1 47.5
12.0 ml 46.0 42.3
I n each case. pressure time records and
INDUSTRIAL AND ENGINEERING CHEMISTRY
Decomposition Products of TEL
Although it \vas not the primary purpose of this work to investigate the products of the decomposition reaction. several detailed studies of the fate of the decomposed lead \cere made. I n these tests the engine was run on iso-octane containing 6.0 ml. of TEL/gallon and 1 theory of ethylene dibromide at a compression ratio of 13. Had no decomposition of T E L occurred, 37.5 mg. of lead !could have accompanied the sampled volume of exhaust stream and been extracted by the hydrocarbon oil in the scrubber. Actually, the total lead dissolved in the oil was determined to be 8.4 mg. of jvhich 1.5 mg. was soluble in 15% nitric acid. This latter portion is knotvn to consist entirely of intermediate decomposition products of TEL-namely trialkyl and dialkyl lead salts. I n addition 1.8 mg. of lead \cas found on the \calls of the scrubber in a form insoluble in the scrubbing medium. A further 0.9 mg. of lead was on the exhaust probe and identified by x-ra)- diffraction as lead formate. I n terms of lead input to the engine, 2 2 7 , remained in the hydrocarbon oil in the scrubber. Of this amount 18% was T E L and 4% trialkyl and dialkyl salts. Similar tests made a t a low compression ratio indicated that better than 90% of the expected lead \cas recovered in the scrubber u i t h essentially no dior trialkyl lead salts. These results justified the sampling technique used in this \cork in that essentially all the T E L \cas recovered from the engine operating at low peak temperatures. Furthermore the>- show that the amount of intermediate decomposition products is lo\.\,: both under conditions \There very little of the T E L is decomposed in the engine and under conditions Xvhere a large fraction is decomposed. T h e implication from this is that the breaking of the first leadcarbon bond by whatever mechanism is involved is more difficult than the breaking of subsequent bonds, so that only small amounts of intermediate decomposition products of T E L are involved.
M hether
the over-all extent of decomposition be small or large. Using this reasoning, it \vas decided to base the measurements of T E L decomposition solely on the solubility of lead compounds in the paraffin oil in the scrubber. I t is believed that this represents. within the limits of error set by rhe other measurements involved. the true extent of T E L decomposition during the eng-ine cycle. Discussion of Results
T h e specific constants of the .Irrhenius equation obtained in the motored engine work for T E L decomposition differ very greatly from those obtained by Leermakers (2) who worked a t much loxcer temperatures and in the absence of fuel a n d oxygen. Despite these differences it is surprising to note that in the range of greatest interest in the motored engine ('00 ' t o 800 O K.) the two equations give rate constants of the same order of magnitude. Until a further understanding of the mechanism of the reaction is obrained this must be regarded as mere coincidence. T h e satisfactory agreement bet\veen the measured decomposition of T E L in the engine and that predicted by the equation involves a number of implications. First it appears that the decomposition reaction of T E L in the engine is homogeneous and does not involve interaction with the walls of the combustion chamber, the surfaces of which undoubtedly varied in composition during the course of the tests. I n vieiv of the short time available in the engine cycle and the existence of high pressure \vhich tends to reduce the extent of diffusion to the walls, it is not surprising to observe no effect of surface composition. -4second implication from the equation is that the decomposition proceeds by the unimolecular mechanism. \Virh respect to the variables tested, such as fuels and additives of a varied nature, this can be considered as proved. However, there does exist a possibility that T E L may react by a bimolecular mechanism and still fit the given equation! provided the second participant in the reaction is one whose concentration is relatively large a n d remains unchanged during the decomposition process. Such a reactant might be oxygen, and, since the oxygen concentration in these tests underwent only minor changes, it cannot be definitely stated that no such reaction took place. I t is of specific interest to note that the present work shoivs no effect of tertbutyl hydroperoxide o n the rate of decomposition of T E L in the engine. T h e recent investigations of Egerton and Rudrakanchana ( 7 ) carried out at much lo\ver temperatures in a flow system in the absence of fuel a n d air shobved a definite acceleration of the rate of de-
composition of T E L due to the presence of the hydroperoxide. Although the TEL-peroxide ratio was essentially the same in both experiments. other factors were markedly different, such as temperature, contact time. and concentration of diluents. T h e divergence between these two sets of data can well be attributed to these different conditions. thus confirming the fact that the interpretation of chemical reactions that take place in engines is best made on the basis of data obtained in the temperature-pressure-time ranye of interest in the engine.
TEL as a Thermometric Device Inasmuch as the decomposition of
TEL in the engine has been shown to be related solely to temperature. the possibility arises that the measurement of temperature may be accomplished by a measurement of the decompositon of T E L . T h e method is best presented in mathematical form as folloivs: For a first-order decomposition reaction. the rate o f \vhich obeys the lirrhenius relation over a limited temperature range. one can \vrite from Equations 1 and 2 d In n j d t = - k = - . 4 e c H
1
Solving for T
S o w since - d i n n dt represents the percentage rate of disappearance of T E L per unit time and -1 and B are knoivn constants. then clearly the temperature can be measured if the rate of decomposition under the desired engine conditions is experimentally determined (assuming no discontinuities in the decomposition curve). Such calculations have been made using the T E L decomposition data from rhe fired engine mentioned earlier Peak end-gas temperatures were reached at about 10 degrees after top center and averaged about 900' K. for the three fuels tested. These results appear reasonable in the light of other estimates of end-gas temperatures (5: 8). but no exact comparisons can be made because different engines and engine conditions \rere involved. A s a n actual method for measuring end-gas temperatures this procedure would have limited applicability. T h e basic difficulty, aside from the experimental apparatus and analytical work which is required, is that the method is useful only over a limited temperature range, although this range might be broadened if compounds other than T E L were to be employed. I t must be recognized, of course: that when additives of this nature are used: there
exists the possibility that they may in themselves affect the temperature of the end gas. T h e major purpose of this work was to obtain data which would be prerequisite to a n understanding of the mechanism of T E L action. Further studies of this nature are now in progress and \vi11 be reported at a later date. Summary
T h e use of a highly instrumented motored engine has made it possible to obtain data of kinetic significance for the decomposition reaction of T C L in conventional concentrations in a fuel-air. mixture. T h e reaction was found to bc. first order with respect to T E L and independent of the fuels and additives tested. T h e reaction velocity constant is given by the equation log k = -4.124
x
lO3/T
+ 7.617
\There k = set.-' and T = K. This equation is based on measurements in the range 600" to 800' K. and time intervals of the order of a few milliseconds. This area of temperature and time corresponds very nearly to that of greatesr interest in the end of gas of a fired engine, so that the equation is expected to be of usefulness in the further investigation of the mechanism of the antiknock action of TEL. These data also have a possible application in the measurement of the temperature in the end gas of a fired engine. Hoivever, more development of the method would be necessary in order to make it a tool of practical importance. Acknowledgment
\,Ve acknowledge with thanks thc contribution of others to the work drscribed in this paper-A. Irene Larson helped \vith the calculations and h l . E . Griffing, W. R. O'h7eil15 and their coworkers carried out the large number of analytical determinations required. literature Cited
(1 ) Egerton, A. C., Rudrakanchana, S.. Fuel 33, 274-85 (1954). (2) Leermakers, J. A , J . Am. Chem. SOC. 5 5 , 4508-18 (1933). (3) McCullough, .T. D., S.A.E. @a,(. Trans. 61, 557-67 (1953). (4) Melville, H. W., Discussions Faraday SOC.,SO.17, 9-13 (1954). ( 5 ) Rassweiler, G. M., Withrow, Id., S.A.E. Quart. Trans. 42, 185 (1938). ( 6 ) Ross, AI.,Rifkin, E. B., IND. EKG. CHEM.48, 1528 (1956). (7) Walcutt, C., Rifkin, E. B.. Ibid., 43, 2844-9 (1951). (8) W u , P. C. K., Thesis, h l I T , Cambridge, Mass., August 1953. RECEIVED for review December 7! 1955 ACCEPTEDFebruary 8, 1956 Division of Petroleum Chemistry, 128th Meeting, ACS, Minneapolis, h?inn., September 19 55, VOL. 48, NO. 9
SEPTEMBER 1956
1539
