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
1314
The work reported here suggests that it would be of interest to study the effect of externally heated recovery sections on the stability limits of hydrocarbon flames. Further studies should be made of blowoff at very high air rates and also of the possibility of using larger air lines to permit higher rateE of air flow.
Vol. 46, No. 6
Patterson Air Force Base, Ohio. The author wishes to express his thanks to R. E. Poling for his able and faithful assistance in carrying out the experimental work and to J. F. Foster, division chief a t Battelle, for his helpful interest and consideration. LITERATURE CITED
ACKNOWLEDGMENT
This work was carried out under the sponsorship of the Flight Research Laboratory, Wright Air Development Center, Wright-
(1) Kurz, P. F., IND. ENG.CHEXf., 45, 2072 (1953). (2) Kurz, P. F., Rev. Sci. Instr., in press. RECEIVED for review October 23, 1953.
ACCBPTED March 8, 1954.
Combustion of Hydrogen as to Knock PARALLEL BEHAVIOR OF HYDROGEN AND PARAFFINIC FUELS W. F. ANZILOTTI, J. D. ROGERS, G . W. SCOTT, AND V. J. TOYISIC Organic Chemicals Department, E. I . d u Pont de AVemours & Co., Inc., Wilmington 99, Del.
H
YDROGEN and oxygen atoms and their inteiniediate reaction products invariably appear in postulated hydrocarbon combustion mechanisms reported in the literature ( 7 , 8). Geib and Harteck (4)have shown that the introduction of hydrogen atoms causes a very rapid reaction in methane-oxygen mixtures a t temperatures as low as -183" C., although ordinarily such mixtures react slowly even a t 300" C. That the presence and mobility of hydrogen and oxygen atoms and hydrovyl radicals are important for flame propagation is shown by Tanford and Pease ( 1 4 ) , who calculated the equilibrium concentrations of these components in moist carbon monoxide flames and found a striking correlation between burning velocities and calculated hydrogen atom concentration. The part played by hydrogen atoms in hydrocarbon thermal decomposition is emphasized by Rice (18, 13).
simple hydrogen-oxygen sj stem could be used with some confidence in interpreting knock reactions in engines. I n order to evaluate this postulation it was necessary fiist to show that there was a marked similarity between the behavior of hydrogen and hydrocarbons in an engine and then, by observing the effect of additives on hydrogen combustion both in the engine and laboratory reaction tubes, to interpret these data in the light of hydrocarbon combustion. The experimental information summarized in this paper shows that hydrogen knocks in engines, has a good tetraethyllead response, and responds very much like paraffins to variation? in engine fuel-air ratio. Laboratory tube evperiments carried out in conjunction with the enginp investigations throw additional light on the hydrogen-oxygen reaction and the influence thereon of tetraethyllead, Finally, reaction mechanisms are suggeated to explain the antiknock activity of tetraethyllead in hydrogen combustion. EYGINE EXPERIMEh-TS
U
HZ CYLINDER
Figure 1.
ROTAMETER
ROTAMETER
TEL BUBBLER
Hydrogen Engine Fuel System
The possibility that the relatively simple hydrogen-oxygen interaction may play a key role in hydrocarbon preknock reactions suggests that a study of hydrogen combustion might shed light on the changes occurring prior t o hydrocarbon fuel knock. This suggestion is strengthened considerably by the recent finding of Downs, Walsh, and Wheeler ( 2 ) that hydrogen knocks in an engine and has a tetraethyllead response. If it could be demonstrated that hydrogen behaves in engines like hydrocarbons in other respects, further evidence would be obtained that the hydrogen-oxygen interaction may be a, common denominator in engine knock. I n that event, laboratory tube studies of the relatively
The knock resistance of hydrogen a as studied in a single CJ linder Cooperative Fuel Research (CFR) variable compression ratio engine modified so that hydrogen could be metered directly into the carburetor Venturi 'u here it vas mixed with air the humidity of which was controlled a t 35 to 50 grains per pound. A schematic diagram of the engine fuel system is presented in Figure 1. Tetraethyllead vapor was added by conducting a portion of the metered hydrogen stream through a sintcied-glass bubbler immersed in liquid tetraethyllead. This tetraethyllead-saturated stream joined the main hydrogen-air mixture a t the intake manifold, There intimate mixing took place because of the combined carburetor T-enturi effect and swirling action of a shrouded intake valve. The cngine operating conditions used in this investigation are summarized in Table I.
TABLE I.
E S G I N E OPERSTING CONDITIOKB
Speed r.p.m. Jacket temp., F. Inlet misture temp., Spark advance Fuel-sir ratio Compression ratio
900 212
F.
100
Varied Varied Varied
1315
INDUSTRIAL AND ENGINEERING CHEMISTRY
June 1954
*
5 I\
i
eo
0 14t
!
i i
z
g
2
i
60-
a
s
i i
c
5
40-
V
i
W c
*
,
Figure 2.
l4
0
2
O
/
L
0
I
80 120 160 200 MIXTURE STRENGTH, % STOICHIOMETRIC
ISOOCTANE
240
TIME-CRANK ANGLE DEGREES
Figure 3.
Knock Characteristics of Hydrogen, n-Heptane, and Iso-octane
Hydrogen and Iso-octane Combustion Rates
I MIXTURE
% STOICH.
13
a I
,*
- - -- - - - - - - ---
203
0
g
IO
i Y
B
9 350 0
0.005
0.010
M O L E PER CENT T E L IN H e
Figure 5.
400
450 500 TEMPERATURE ,'C
5 50
6 00
Hydrogen and Oxygen Explosion Limits
Figure 4. Tetraethyllead Response of Hydrogen
Stoichiometric mixture
Knock was detected by means of a Phillips Model D-1 internal pickup mounted in the engine cyclinder. The signal from the pickup vias fed to a Model 208-3 Dumont oscilloscope. Audible knock usually occurred a t the same time knock was first observed on the oscilloscope screen. HYDROGEX KNOCK. The resistance of hydrogen to knock was studied by measuring the knock-limited compression ratio, which is defined as that engine compression ratio which gives the first trace of knock. The antiknock values found for hydrogen, isooctane, and n-heptane throughout the mixture range are illustrated in Figure 2. The three fuels were knock tested under identical conditions except for spark advance, which was adjusted to approximately that for maximum power with each fuel. A spark advance of 30 before top dead center (BTDC) was used with the paraffin fuels, but with hydrogen the spark was a t top dead center because of it3 much faster flame speed. Hydrogen had a tendency to backfire a t mixture strengths near stoichiometric. These data show that hydrogen knocks more readily than iso-octane and responds to variations in mixture strength in much the same manner as does iso-octane. Calculations of the time necessary for complete combustion of iso-octane and hydrogen were made from indicator card data ( 1 1 ) . These showed that hydrogen combustion was completed in only 29% of the time necessary for the iso-octane to burn completely (Figure 3). The effect of the much faster flame speed of hydrogen is beneficial because the residence time of the last, knocking portion of the fuel-air mixture is much shorter and
therefore higher compression ratios can be tolerated without knock. If it were not for this shorter end gas residence time, hydrogen probably would knock far more readily than does isooctane or possibly even n-heptane. Hydrogen resembles the knocking combustion of paraffinic hydrocarbons in another respect-it responds well to tetraethyllead. The marked increase in knock-limited compression ratio made possible by the addition of small amounts of tetraethyllead is illustrated by Figure 4. The noticeable increase in tetraethyllead response as the mixture becomes richer is believed to result from the increase in the ratio of tetraethyllead to active knocking species. Inasmuch as it is based on total hydrogen, the amount of tetraethyllead present increases as the mixture becomes richer, while the number of knock-producing components may remain constant or be reduced as the air concentration is lowered. The foiegoing discussion has shown that hydrogen knocks readily in engines, has a good tetraethyllead response, and responds much like iso-octane to variations in mixture strength. This information, indicating that hydrogen has several of the engine performance characteristics of paraffinic hydrocarbons, is interpreted as indicating that the hydrogen-oxygen reactions may be, a t least in part, controlling in the preknock reactions of the more complex hydrocarbons T o assess this concept more fully, additional information was obtained in laboratory hydrogen combustion tube experiments in which better control of temperature and contact surface was possible.
INDUSTRIAL AND ENGINEERING CHEMISTRY
1316
Vol, 46, No. 6
LABORATORY COMBUSTION TUBE EXPERIMENTS
Many investigations of the thermal reaction between hydrogen and oxygen have shown the existence of the three definite explosion limits (8) illustrated in Figure 5, Conditions just inside and outside the third explosion limit a t a pressure of 1 atmosphere were selected for these studies. A pressure of 1 atmosphere is sufficient to hold to a reasonable minimum the effects of diffusivity of reacting atoms and radicals and chain breaking at the wall, which play a major part in the reaction kinetics at lower pressure.
TABLE 11. REPRODUCIBILITY OF HYDROGES-OXYOES REACTIOP
.
[Tube Lemperature, 543' i 3' C total gas flow, 40 cc./min.; mole % 111, 66.7 (stoichiometrio)] Gas ReaiHz02 HzO Rate of HPO dence Collected, Collected, Formation Tube Time, Sec. hlg. g hIg./Min.' Quartz 0 31 44-46 1.5-1.5 0
50-93
1.7-3.1
98-137 114-135
8.3-4.6 3.8-4.6 1.8-2.2 1.5-1.9
Borosilicate
glass
32 38
0.4-0,s 0.46-0.53 0.20-0.24 0.24-0.25
54-6B
44-57
41
a
1.3-1.4 44-48 1.5-1.6 Range of values given for each separate set of data represents degree of
vanation found in experiments run on single day. TUBE AND FURNACE MIXER
I
A
HZ
I
:
FLOW
U
FLASH ARRESTORS
MANOMETERS
RECEIVER AND DRYER
METERS
Figure 6. Combustion Train
The rate of the hydrogen-oxygen reaction was measured in a flowing system either by the method of Pease (IO), in tvhich the amount of water and hydrogen peroxide formed was determined, or by observing the temperature a t which a known hydrogenoxygen mixture exploded. A schematic flow diagram of the glass combustion train is shown in Figure 6. Hydrogen (99 5% pure, obtained from Paschall) and oxygen (99.5% pure, obtained from Air Reduction) were reduced to 2 to 3 pounds per square inch gage and each was passed through a separate, identical series of dry ice traps in order to reduce the likelihood
1
SINTERED
1
COMEUSTIOV TUBE
Hz --+
ThERFIOCObPLE GAS MIXER
I1
HYDROGES-OXYGEN REACTIOSBELOW EXPLOSION TEMPERAA st'udy of the rate of water and hydrogen peroxide format'ion employing a stoichiometric hydrogen-oxygen mixture revealed that very little reaction took place below 530' C. and that the mixture became explosive a t 556" C. Accordingly, an intermediate temperature range of 540'' to 645" C. was selected for initial study of n-ater and hydrogen peroxide formation. Reproducibility and the effects of tube type and residence time are demonstrated in Table 11. The literat'ure (10) attributes this poor reproducibility, in part, to the extreme sensitivity of the reaction to the combustion tube surface. However, it is improbable that better reproducibility would be attainable, since the values which others have reported (b)! even with great precautions to ensure reproducible surface conditions, indicate variations of the same order as those found in this work. The data of Table I1 indicated that hydrogen peroxide n.as not isolable from experimrnta conducted with quartz tu1 jes; this has been reported by ot'hers (10). For this reason, borosilicate glass reactors were used in place of quartz in the experiments reported in the remainder of this paper. Meaeurenients were also made of the effect of temperature on the rates of formation of water and hydrogen peroxide from a stoichiometric hydrogen-oxygen mixture in clean and lead oxide-coated tubes. The lead oxide coating was likely an equilibrium mixture of lead oxide and lead dioxide at the time of the experiment. The results, plotted in Figure 8, shox that in a clean tube water formation increased steadily and appreciably up to the temperature of explosion, whereas hydrogen pcroxide Iyas present in only trace amounts even just below the explosion temperature. The coating increased the temperature a t which wat,er vapor formed and eliminated hydrogen peroxide. TURE.
RECEIVER
Figure 7.
Combustion Tube Assembly
of explosion flash back and to eliminate Jvater. The diied gases were brought together in a mixing chamber and then fed into the heated combustion tube, the temperature of which was regulated and maintained within 11' C. by means of a recorder controller. Flow rates were usually such that the residence time of the mixture in the reaction chamber was within the range 30 to 40 seconds. The findings of others (IO)that the surface of the combustion tube is critical were confirmed and therefore before use each tube was waehed two t o four times with hot nitric acid and then at least ten times with distilled water. A typical combustion tube assembly is shown in Figure 7 . The reaction chamber i s 11 em. in length and 1.8 em. in diameter. Tetraethyllead was added to the combustion tube by passing the hydrogen stream through a sintered-glass tetraethyllead bubbler, maintained a t constant temperature, before introduction into the mixing chamber.
HzOz 0
520
E40
--
._*I-
560 580 TEMPERATURE ."C
600
Figure 8. Pre-explosion Reactions Stoichiometric hydrogen-oxygen mixture
620
INDUSTRIAL AND ENGINEERING CHEMISTRY
June 1954
650 0
t
1317
1
2a 6103
t
-
E 570 530
-
STOICHIOMETRIC I
I
I
0.01
0.02
0.03
0.04
MOLE PER CENT TEL IN Hp
Figure 10.
The similarity of the curves of rate of water formation versus temperature indicates that the lead oxide may have retarded rather than altered the oxidation mechanism. The failure to isolate hydrogen peroxide is not unexpected because it is known t o decompose readily on contact with a lead oxide surface (3). EXPLOSION TEMPERATURES OF HYDROGEN-OXYGEN MIXTURES. I n contrast with the measurements of products obtained a t intermediate temperatures, the data obtained by determining the temperature a t which a hydrogen-oxygen mixture exploded were reproducible, generally within &2' C. Inasmuch as most of the variables investigated caused far greater changes in explosion temperatures than &2" C., this reproducibility was satisfactory. Measurements of the influence of mixture strength on explosion temperature, presented in Figure 9, showed that in mixtures leaner than 25 mole % or richer than 80 mole % hydrogen no explosion occurred when the temperature was raised as high as 640' C., which approaches the softening point of borosilicate glass. Best checks were obtained with slightly lean mixtures-i.e., approximately 50 mole % hydrogen. As is shown in Figure 10, the addition of very small quantities of tetraethyllead caused a substantial elevation of the explosion temperature of a stoichiometric hydrogen-oxygen mixture. This marked effect of tetraethyllead correlates with the observation that this compound reduces the knock of hydrogen in engines It v a s of particular interest to find that lead oxide reaction tube coatings increase the explosion temperature while metallic lead coatings have no effect, A lead oxide deposit prepared by decomposing tetraethyllead in an oxygen stream a t 350' to 500' C. raised the explosion temperature of hydrogen-oxygen mixtures as shown in Figure 11. When this deposit was reduced to metallic lead b y heating in the presence of excess hydrogen, this effect was destroyed completely. It is apparent in these experiments that lead oxide rather than metallic lead is the active species causing the elevation of the hydrogen-oxygen explosion temperature. This is in accordance with the present theory that the active antiknock component of tetraethyllead in hydrocarbon fuels is lead oxide (1). DISCUSSION
The information obtained indicates that hydrogen knocks and has a tetraethyllead response and that lead oxide rather than metallic lead is probably the effective antiknock agent. The antiknock effect of tetraethyllead in engines and in combustion tubes arises from the ability of lead oxide to nullify in some way the effects of chain initiators and carriers which otherwise would cause a branched-chain explosive reaction. An analysis of the differences between steady state and explosive reactions in hydrogen-oxygen systems permits speculation re-
Effect of Tetraethyllead on Explosion Temperature
Stoichiometric hydrogen-oxygen mixture O'tV
W
I
I I I
620 I I
a
+a
2 600
I I
EXPLOSION
B
I I
z
2 580 fn
4
REGION OF NO EXPLOSION
W x
560
540
' 0
Figure 11.
I IO0 200 300 MIXTURE COMPOSITION PER CENT OF STOICHIOMETRIC
-
400
Effect of Lead and Lead Oxide on Explosion Temperature
garding the action of lead oxide in raising the conditions for explosion to higher temperatures. STEADYSTATE AND EXPLOSIVE REACTIOKS.The principal reactions between the second and third explosion limits (Figure 5 ) can be summarized in the following equations: Steady state Chain-branching explosive
+ + +
+
H 0 2 M +HO; M HO; Hz +Hz02 H H Oz--+HO' 0
+
+
+ 0 + Hz +HO. + H
HO'
+ Hz +HzO + H
(1)
(2) (3)
(4)
(5) According to this mechanism, which is advanced by Lewis and Van Elbe (8), the reaction between H and oxygen in the presence of a third body, M , forms the postulated and logical HOi radical (9) which in turn reacts with hydrogen to form hydrogen peroxide and H. Thus, in the nonexplosive region between the second and third explosion limits a steady state exists as shown in Equations 1 and 2. Reaction 2 becomes progressively dominant as the temperature and pressure or both are increased, until a t the third limit the chain-branching reactions 3, 4,and 5 come into play, giving rise to an explosion. ACTIONOF LEADOXIDE. It is plausible that H, 0, OH. and HOi radicals all exist a t various stages in the hydrogenoxygen explosive reaction. The effects of lead oxide on each must be considered in explaining the observed antiknock effect.
INDUSTRIAL AND ENGINEERING CHEMISTRY
1318
As is evident in the explosive sequence shown in the previous equations, a most effective point of attack for lead oxide is the HOi radical. If this could he degraded to nonchain-branching products, it n-ould serve as a method of removing indirectly the chain carrier H. I n addition to the method proposed by Chamberlain and Walsh ( I ) , lead oxide might accomplish this in the following xwy:
+ PbO + PbO! + OH. + Hz + HPO + H H + PbOp +OH' + PbO
HO,
OH.
+ PbO +PbO? H? H + PbOs +PbO + OH. +H20 + H 0
9 and 10 depend upon ready oxidation and reduction of the metal oxide, which was long ago observed to be a characteristic of metals having antiknock activitv. However, these particular reactions are merely suggested mechanisms; proof of their validity must await further work. LITERATURE CITED (1)
(6) (8)
(9) (10)
It is lcnoivn that lead oxide can react n-ith 0. Evidence is provided by the finding of Hoare and Walsli ( 6 ) that a lead monoxide coating raises the explosion limit of a dry carbon monoxide-oxygen mixture. It will he noted that Equations
Chamberlain, G. R. S . , and Walsh. A . D., Proc. Roy. SOC.
( L o n d o n ) , A215, 175 (1952). (2) Downs. D., Walsh, A. D., and Wheeler, R. W., T r a n s . Roy. SOC. (London),243, KO.870,463-521 (July 19, 1951). (3) Egerton, 1. C., and Jain, B. D., Fuel, 31, 62-74 (1952). (4) Geih. K. H., and Harteck, P., Z. p h y s i k . Ciiena.. A170, 1-19 11934). (5) Giguere, P. A , , Can. J . Research, 25B, 135 (1947).
(7)
According to this mechanism, the net influence of lead oxide is to Cause HO; to degrade but one radical rather than to give rise to several resulting in chain branching. The possibility remains that lead oxide may suppress hydrogen knock by reducing the number of chain carrier hydrogen atoms a,nd oxygen atoms or hydroxyl radicals, which n-ould suppress chain branching. Mechanisms by which this might occur are:
Vol. 46, No. 6
(6) Roare, D. E., and Walsh, A. D., Proc. R o y . Soc. (London), A215, 464-66 (1952). (7) Jost. W,, "Explosion and Combustion Processes in Gases," h-en- York, hIcGraw-Hill Book C h . , 1946. (8) Lexyi?, B., and Von Elbe, G., "Combustion, Flames and Explosions." Sew York, Academic Pres?, Inc.. 1951. (9) LIinkoff, G . J., Discussions Faruday Soc., KO.2 , 151-8 (1947). (101 Pease. R. K,, J . Am. C h e m . Soc.. 52. 5106 11930). ( l l j Rassweiler, G. AI., and Withrow,' L., s'.A.E: Q u a r f . Trans., 42, SO. 5, 185-204 (1938). (12) Rice, K. K., J . Am. Chem. Soc.. 53, 1959 (1931). (13) Rice, F. O., and Rice, K. K., "The Alinhntic Free Radicals." Baltimore, Lid., Johns Hopkins Pres., i935. (14) Tanford, C.. and Pease, R. 3 . .J . Chem. Phys., 15, 431 (1947).
RECEIVED for review Nol-ember 7 . 1953.
ACCEPTED
hIaroh 10, 1954.
Presented before the Division of Petroleum Chemistry a t t h e 124th llreting o i the AMERICAX CHELIICAL SOCIETY, Chicago, Ill,, 1953.
tion of Sulfur ilms in F. A. M. BUCK Shell Oil Co., P.O. Box 711, Martinez;, Calif.
A
CIDIC substances resulting from the oxidation of fuel are know1 to be important contributors to the coriosive wear of cylinder walls and piston rings of internal-combustion engines (3, 4). The principal products from the oxidation reactions are water and carbon dioxide from the hydrocarbon constituents of the fuel. Invariably, hoir ever, commercial Diesel fuels contain small amounts of substances such as sulfur and nitiogen which are oxidized to their respective acid anhydrides The average sulfur content of A4STRI Grade 213 Diesel fuels sold in the Cnited States during 1952 v a s 0.34 weight % ( 2 ) . The activity of strong acid anhydrides under the temperature and pressure conditions in the combustion chamber is, however, so great that such small concentrations are of practical concern. This is particularly true under operating conditions of l o ~ vcrankcase and cylinder jacket temperatures, which are knoivn to be conducive to high rates of corrosive cylinder wear. It is known, for instance, that in at least some commercial Diesel engines under these conditions, the rate of wear of cylinder liners is adversely affected by increasing the concentration of sulfur in the fuel (3). The corrosion of the cylinder n-alls takes place on surfaces 1% hich are continuously covered ~ ~ i at film h of crankcase lubricating oil, I n ordcr to react with the metal surfaces, the sulfur oxides must first be absorbed from the gas phase into the film of lubricating oil and then diffuse through the film to the metal-oil interface. There, prohably acting with water molecules which have also reached the metal surface by diffusion through the oil film, they form corrosion products which are swept away in the
turbulent oil film, leaving the surface open to fresh att,ack. (The term "absorption" as used throughout this paper is to be interpretcd in the general sense of being a dissolution of material from the gas phase int,o the liquid oil films. Some workers contend that the absorption of sulfur oxides is preceded by the condensation of water and sulfur anhydrides as droplets of acid, This point is not involved here.) It is the object of this paper to present some information on the absorption of the sulfur oxides in the oil film, in the hope that a better understanding of the mechanism of this process will contribute to developments which will reduce the corrosive !rear of Diesel engines by fuel oxidation products. The experiments to be described here were designed to measure the rate a t LThich sulfur from the fuel accumulates in the erankcase lubricating oil of a typical 4-cycle Diesel engine. The principle is simple. An organic molecule in the Diesel fuel boiling range was synt,hesized with radioactive sulfur-35. 811 suifur atoms in the fuel are assumed t o oxidize completely to sulfur dioxide during combustion. [Sulfur dioxide is the only oxidation product of sulfur considered in this paper, although the arguinent,s apply to sulfur trioxide as well. However, recent work with a similar engine has sho..rn that by far the greater amount of fuel sulfur is actually oxidized to sulfur dioxide ( 6 ) . ] Therefore, the reactions of the fuel sulfur, after the combustion stcp, are assumed to be identical to the reactions iiivolvjng the sulfur-35. The accumulation of compounds containing sulfur-35 in the crankcase oil was determined by measuring the specific activity of the crankcase oil at frequcnt intervals. From this accumulation
